Through A Statute Darkly: Delegation and Goal Ambiguity in Executive Agencies

• In Event: Discrimination, Fairness, and Algorithmic Bias in Machine Learning Applications

Abstract

The question of how much discretion Congress actually delegates to administrative agencies has returned to the center of both scholarly and doctrinal debate following Loper Bright Enterprises v. Raimondo and the dismantling of Chevron deference. Bureaucratic discretion is commonly measured from the formal properties of statutory text, an approach that treats delegation as an intentional transfer of authority calibrated by Congress. We argue that this framework captures only part of the phenomenon and misses a structurally distinct source of agency latitude that is not reducible to legislative choice. Drawing on the goal ambiguity literature in public administration, we develop a formal model that decomposes total agency discretion into two theoretically separable components. The first is a delegated component, which Congress confers through explicit statutory design and can in principle be adjusted through more precise drafting. The second is a constitutive component, which arises from the interpretive latitude inherent in ambiguous statutory goals and cannot be eliminated through drafting precision because the ambiguity is intrinsic to the policy objective itself. The decomposition is derived from a judicial review mechanism, which makes the magnitude of the constitutive component conditional on the prevailing standard of deference and generates testable predictions about how regime changes in judicial review reshape effective agency authority. We operationalize the model using a large corpus of federal authorizing statutes. An LLM-based classifier measures the delegated component, and new transformer-based NLP architectures measure the four Chun-Rainey dimensions of goal ambiguity (target, timetable, evaluative, and priority ambiguity), allowing us to construct both components from the same textual source. An event-study design using statute-level goal ambiguity as treatment intensity tests whether agencies operating under more ambiguous statutory mandates experienced differential compression of behavioral discretion relative to those operating under more concrete mandates following the Loper Bright decision. Preliminary results indicate that the constitutive component accounts for a substantial and previously unmeasured share of effective agency discretion, and that agencies facing high goal ambiguity experienced the largest post-Chevron contractions in observable policymaking latitude. The findings speak to long-standing debates about the measurability of bureaucratic discretion and to the policy consequences of the post-Chevron doctrinal regime for administrative capacity.

Authors

• Lefteris Jason Anastasopoulos, University of Georgia
  Non-Presenting Co-Author

• William G Resh, Georgia State University
  Presenting Author