Administrative Persistence and the Problem of Outcome Blindness

The views expressed are those of the author and do not necessarily reflect the views of ASPA as an organization.

By Robert Choi
January 9, 2026

Public administration has long concentrated on designing systems to process people and money, while showing far less consistency in taking responsibility for the downstream consequences when those systems fail. From early administrative theories that treated workers as standardized units to be measured and controlled, to later models that equated “good administration” with rules, compliance and procedural fidelity, the pattern is consistent: government is effective at generating activity, programs, offices, trainings, plans and reports, but weak at terminating activities that do not produce results. This pattern is best explained by incentive structures rather than intent. Bureaucratic organizations are designed for durability, risk avoidance and the appearance of fairness through uniform process; elected officials operate within incentives that reward visible action, stated intentions and appropriations, while dispersing responsibility when outcomes fall short; and administrative systems, once staffed, funded and embedded in networks of contractors, advocates and regulated entities, develop internal and external constituencies with a stake in their continuation. In such an environment, failure is rarely treated as information that drives redesign. More often, it becomes justification for additional resources, expanded mandates or broader definitions of success. The result is a public sector in which systems persist even when measurable outputs are weak, outcomes stagnate and negative side effects are evident.

DEI provides a clear illustration of how process accountability displaces outcome accountability. Many DEI initiatives are framed as moral imperatives and implemented through compliance-oriented infrastructures, trainings, statements, reporting structures, committees and dashboards, rather than as interventions with testable causal mechanisms and explicit termination criteria. The empirical evidence on widely used corporate diversity programs is mixed at best. Large-scale studies of mandatory diversity training consistently find little to no durable effect on representation in management and, in some cases, measurable backlash effects, while interventions that alter managerial incentives, such as accountability tied to hiring and promotion decisions or structured recruiting and mentoring systems, show more promise. None of this suggests that inclusion lacks importance. Rather, it shows that the dominant administrative approach, mandate, formalize, repeat, creates an evidence vacuum in which participation and expenditure become proxies for effectiveness. Programs continue even when their strongest claims of success rest on completion rates or internal sentiment measures, while downstream outcomes, hiring, retention, promotion, skill development, productivity and service quality, remain weak, inconsistent or unmeasured.

Debates over affirmative action expose a related administrative tendency: treating observed disparity as presumptive evidence of injustice while relying on highly visible policy levers that are poorly suited to address root causes. When admissions rules change, the relevant question is not only who gains access to selective institutions, but what those changes do to degree completion, field of study and long-run economic mobility, because those outcomes determine whether policy expands capability or merely reallocates opportunity. A large-scale analysis of California’s Proposition 209, which ended race-based affirmative action in University of California admissions, found that underrepresented minority applicants were, on average, redirected into lower-quality institutions, experienced declines in degree attainment, including in STEM fields, and earned roughly five percent less annually between ages 24 and 34; it also estimated a cumulative decline of at least three percent in the number of early-career underrepresented minority Californians earning over $100,000 by the mid-2010s. These findings are often invoked to argue that affirmative action “worked,” but they also reveal a deeper accountability failure. Disparities observed at the point of selection were treated as proof of injustice requiring redistribution, rather than as indicators of unequal preparation shaped years earlier by K–12 instructional quality, access to advanced coursework, grading standards, teacher assignment patterns and institutional expectations. A substantial body of education research shows that achievement gaps emerge early, widen over time and strongly predict postsecondary readiness long before admissions decisions are made. Admissions policy cannot repair those production failures; it can only offset them temporarily at the point of selection. By repeatedly cycling admissions criteria instead of repairing the systems that produce readiness, the administrative system substitutes symbolic equity for capacity building, then treats the persistence of gaps as justification for pulling the same lever more forcefully.

Public education offers the longest-running illustration of sustained resource investment without proportional outcome improvement, and debates over charter schools show how administrative systems protect incumbency. Traditional public school systems operate under layered accountability structures that include budgets, staffing rules, procurement requirements, collective bargaining agreements, compliance regimes and political oversight. These constraints are often defended as equity safeguards, yet they also make performance differences difficult to act on, turning school improvement into an exercise in procedural reform rather than results. Charter schools are not uniformly superior and outcomes vary by model and context; online charters, for example, perform substantially worse than traditional public schools on average. But when charters are high-performing and oversubscribed, lottery-based studies and large multi-state analyses consistently find meaningful gains, particularly in urban settings and among disadvantaged students. Stanford CREDO’s National Charter School Study III (2014–2019) found that charter students gained the equivalent of approximately 16 additional days of learning in reading and six in math compared with matched traditional public school peers, with larger effects for students in poverty and for Black and Hispanic students. Quasi-experimental lottery evidence from Boston shows large achievement gains and increased four-year college enrollment, while similar analyses in New York City find positive achievement effects for many charter schools over the periods studied. The administrative question is therefore straightforward: if some schools consistently produce stronger learning growth for similar students at comparable per-pupil costs, why is replication constrained? In many jurisdictions, political and bureaucratic incentives favor preserving existing delivery systems over scaling effective alternatives, because institutional structures reward organizational stability, employment protection and negotiated rules rather than measured learning outcomes.

Observed improvements in minority economic outcomes surface a different administrative problem: how public systems assign causality. In complex social environments, outcomes often change for reasons that sit outside any single program’s scope, yet administrative practice routinely treats temporal overlap as evidence of program effectiveness. Public agencies are poorly equipped to distinguish between changes driven by macroeconomic conditions, sectoral dynamics, demographic shifts or private behavior and those plausibly caused by specific interventions. This creates a systematic bias toward credit-claiming rather than causal inference. Disparities are therefore interpreted backward: differences in outcomes are read as proof of injustice at the point of policy action, while improvements are claimed as validation of existing programs, even when neither interpretation is supported by mechanism-level evidence. From an administrative-design perspective, the issue is not whether equity-oriented programs are normatively justified, but whether the system has the analytical capacity to determine what actually changed outcomes, when and why. Without that capacity, clear theories of change, counterfactuals and stopping rules, public administration cannot reliably learn from success or failure and disparity becomes a rhetorical signal rather than a diagnostic indicator of preparation, institutional performance or upstream constraint.

A serious public administration reframing of contemporary equity debates would therefore move away from intent-based arguments and toward institutional design. The central questions are how government defines success, how it measures it, how it incorporates failure into learning and how it ends programs that do not work. When agencies choose to fund DEI initiatives, affirmative-action–adjacent programs or K–12 reforms, they should be required to articulate a falsifiable theory of change, identify a limited set of outcome metrics tied to service delivery and economic mobility, establish comparison baselines involving similar populations without the intervention and specify explicit termination criteria that trigger redesign or defunding. This approach reflects a conclusion long acknowledged in serious administrative theory: people are not interchangeable inputs and systems cannot be managed through process alone. Treating disparity as automatic proof of injustice may satisfy moral signaling, but refusing to diagnose preparation failures and measure outcomes does not advance equity; it preserves the administrative machinery that produces the disparities in the first place.

---

(4 votes, average: 3.50 out of 5)
Loading...