*This is the fourth post on a symposium on Jed Stiglitz’s “The Reasoning State.” For other posts in the series, click here.

“[A]gencies do not have quite the prerogative of obscurantism reserved to legislatures.”  This observation, one of my favorites in the widely taught United States v. Nova Scotia Food Products Corp., describes fundamental features of U.S. public law. Whereas Congress may legislate without explaining what it seeks to do, agencies must expressly justify the reasonableness of their actions, and do so at the time they act.  

Why do agencies not have Congress’s “prerogative of obscurantism”? In Jed Stiglitz’s insightful and compelling book, The Reasoning State, he offers an explanation for the law’s divergent treatment of lawmaking by agencies and Congress, and with it, a general account of administrative law. Stiglitz argues that agencies’ greater burdens of articulation are part and parcel of the reason Congress delegates power to them. In a complex society, Stiglitz’s theory goes, the electorate is not able to assess whether Congress’s actions on substantive policy determinations are in the public’s own interests or simply doing the bidding of special interests. Congress in turn seeks to mitigate distrust in its own substantive choices. It does so by enacting legislation that establishes broad objectives or purposes, which the electorate is capable of evaluating, and otherwise delegating authority to another institution—the administrative agency—to make substantive decisions carrying forward those objectives subject to greater transparency, process, and reasoning requirements. According to Stiglitz, the greater transparency, process, and reasoning constraints that apply to administrative agencies enhance the credibility of agencies in ways that Congress’s own credibility cannot be augmented. In short, administrative law is not simply a response to delegation but part of the reason for it. The Reasoning State then reports the results of intriguing experiments which show that requiring reasons improves decisionmaking, especially when the decider knows his or her reasons will be subject to review, and those very reasons also increase the perceived quality of decisions.  

The Reasoning State thus offers a startingly original theory of the connections between delegation and administrative constraints—and the shape of many familiar doctrines and practices in the administrative state, one complements and provides some counterpoints to Jerry Mashaw’s account of reasoned decisionmaking in Reasoned Administration and Democratic Legitimacy and other writings. My brief contribution here to celebrating the achievements of The Reasoning State focuses on probing two separate points: (1) suggesting that Stiglitz’s theory has stronger implications for the framework of judicial review of agency action on the question of interpretive deference than Stiglitz traces; and (2) urging that an important next step in advancing Stiglitz’s project is to evaluate the constraint reason-giving creates on decisionmakers once a set of reasons has been given.   

1. Purposive Agencies and the Judicial Role 

Suppose that Stiglitz’s delegation theory is roughly correct.  What implications does it hold for how an agency should understand its own role, and how courts should review agencies’ actions? As to the agency’s role, we can draw a clear picture. On Stiglitz’s account, regulatory legislation’s basic function is to establish “some objective” which “the public stands in a good position to evaluate.” (47-48). The role of the agency, then, is to “effectuate” those objectives or goals. (48). As a result, agencies are inherently teleological or purposive—their role is to carry forward the objectives set for it by the legislature, subject to a set of constraints.  I very much agree that agencies are inherently purposive in this way.  Indeed, arbitrariness review reinforces this understanding. To engage in valid means-ends reasoning requires the agency to have identified an aim or objective of its action.  

 The question, then, is how to mediate between an increasingly textualist judiciary and inherently purposivist agencies. They seem to be on course of repeated collisions: agencies must rely upon an understanding of statutory purposes to administer their statutes and yet the judiciary increasingly suggests that ascertaining statutory purposes is not a proper object of judicial inquiry. That is a central conundrum of public law today. In my view, Stiglitz’s theory has stronger implications for how this conundrum needs to be solved than he traces in The Reasoning State. While Stiglitz defends a version of arbitrariness review, he only gestures at a position on judicial deference to agencies on issues of interpretation. But if courts, in the name of textualism, interpret regulatory legislation in ways that disable the authority of agencies to pursue statutory objectives reasonably, Stiglitz’s theory runs into trouble. In that case, the legislature’s solution to its own credibility problem—allow credible agencies to decide—has been overturned by the judiciary. As a result, it seems to me that Stiglitz’s posited solution to legislative credibility problem depends upon courts taking a deferential approach on questions of interpretation—for instance, by asking whether the agency has reasonably advanced statutory objectives—as opposed a non-deferential one—such as by inquiring whether the agency acted on what the court sees as the best interpretation of the statute.  

2. Testing Agency Reason-Giving 

The Reasoning State does not merely posit a new theory of delegation and agency credibility. Stiglitz empirically tests whether requiring reasons has an effect on decisionmakers and whether those reasons improve the public’s assessment of decision quality. In a series of ingeniously devised studies, Stiglitz shows that the requirement of reasons, particularly when the deciders are told their reasons are subject to review, results in decisions that hew more closely to the law’s requirements. Stiglitz’s experiments then reveal that giving reasons, especially when provided under the subject-to-review condition, enhance the public’s perception of the decision’s quality, especially so when the decision appears to be a dubious one. These studies offer provide some comfort to those who view the provision of reasons as a fundamental condition of legitimacy in public law.   

Still I wonder if Stiglitz’s empirical investigations could go further. Part of the promise of reason-giving is that it disciplines the decisionmaker once the reasons are given. Giving a reason, as Fredrick Schauer explains, implies a prima facie commitment to action within the scope of that reason. That is, giving a reason provides some provisional assurance of where the decisionmaker is headed.  Moreover, the members of public may view reasoning-giving favorably in part because they view reason-giving as having some constraining effect on future decisions.  

In The Reasoning State, Stiglitz does not test these dynamics. Those examinations strike me an important next step in advancing Stiglitz’s project. To what extent does giving reasons constrain a decisionmaker with discretion to actions within the scope of those reasons? And does the public perceive agencies’ decisions which reflect that constraint more favorably, other things equal? Empirical examination of those questions could offer another level of confirmation for Stiglitz’s theory. They could show that agencies hold the promise of credible decisionmaking not merely because reasons improve one-off decisions and how the public perceives such one-off decisions but also because they pull upon the public’s deeper understanding of what giving a reason says about what will happen next. Perhaps follow-on research to The Reasoning State will reveal whether agencies’ reason-giving also reinforces their credibility because it provides the public and the regulated some, however provisional, grounds for expectations about agencies’ future actions.  

Kevin M. Stack is the Lee S. & Charles A. Speir Professor of Law at Vanderbilt University Law School.