Which Signals Matter? Performance Profiles and Federal Resource Allocation in the Continuum of Care System

• In Event: Coming Full Circle: Design, Deliberation, Decision, and Delivery of Collaborative Policymaking

Abstract

A central challenge in intergovernmental governance is how higher-level governments distribute limited resources across local systems that differ in need, capacity, and operating context. Governments managing networks of local providers — common in policy domains addressing homelessness, poverty, and climate change — frequently implement performance-based budgeting systems that reward local systems achieving the greatest impact. Yet when local systems pursue multiple, potentially competing goals, no single criterion can fully capture what high performance means across heterogeneous jurisdictions. Within formal performance management systems, indicators provide an administrative language for budgeting and control, but that language matters only insofar as it enters actual allocation decisions. Prior research has distinguished formal adoption from meaningful use, showing that performance information depends on organizational routines, analytic capacity, and external pressures. The problem intensifies in multi-layered performance regimes, where information passes through several institutional levels before shaping consequential decisions. Goal ambiguity is not peripheral to such systems; it is a defining condition. Once performance is recognized as genuinely multidimensional, a simple hierarchy of better and worse local systems becomes difficult to sustain and the question of which signals actually drive resource allocation becomes both theoretically important and empirically underexplored. We examine this question using HUD's Continuum of Care (CoC) program, where standardized System Performance Measures (SPMs) and federal grant allocations are tightly linked within a decentralized homelessness response system. We use panel data covering all 380 CoCs from 2015 to 2023. We conduct a factor analysis on the 38 SPMs consistently available across the study period, followed by latent profile analysis (LPA) to identify recurring performance profiles. We then use a two-way fixed effects estimator with clustered standard errors and controls for urbanicity, bed capacity, participation rate, comparing funding allocations between those changing profile memberships to ones remaining stable. Factor analysis yields five empirically distinct and largely uncorrelated performance dimensions (all |r| < 0.28), confirming that performance is multidimensional and that strength on one dimension does not predict strength on others. LPA identifies five recurring CoC profiles, each representing a distinct combination of strengths and weaknesses. Funding patterns reveal that scale of need carries substantial allocative weight: "High-Need Durable-Exit Systems" receive the greatest average grants despite only average performance on most outcome dimensions, while "Stability-Driven Low-Income Systems" receive the least. These findings contribute to research on performance regimes, goal ambiguity, and intergovernmental resource allocation. They suggest that standardized indicator systems do not impose a unified performance hierarchy, with different signals carry different administrative salience. Need-based signals appear to shape allocation more than outcome-based signals, with implications for how performance-based budgeting functions under conditions of genuine goal multiplicity.

Authors

• David Lee, Sungkyunkwan University
  Presenting Author

• Andrew Alfred Sullivan, University of Central Florida
  Non-Presenting Co-Author