Unmasking and rethinking hierarchical inefficiency in healthcare systems

This study introduces a hierarchical Bayesian model that decomposes hospital inefficiency into components inherited from higher administrative tiers and those self-generated at the hospital level. This approach extends traditional efficiency analyses (e.g., stochastic frontier models that assume inefficiency occurs only at the provider level) by capturing how inefficiency cascades from provinces and regions down to individual hospitals. The study applies the model to 942 hospital-year observations (2015–2019) from three Canadian provinces (Alberta, Nova Scotia, Ontario). The framework separates persistent technical inefficiency into three latent factors at the province, region, and hospital levels, quantifying inefficiency inherited from upstream governance and inefficiency generated at each level. Results show that inherited inefficiency from higher tiers accounts for roughly 73–76 % of total persistent inefficiency across these provinces. This challenges the policy orthodoxy of targeting individual hospitals in isolation, as even hospitals with low self-generated inefficiency remain far from the efficiency frontier due to systemic constraints imposed by higher levels. This study offers the first empirical decomposition of inefficiency across a national health system's hierarchy. Although demonstrated in Canada, the approach is generalizable to any healthcare system with multilevel governance. It provides policymakers with a diagnostic tool to benchmark efficiency more accurately and to design system-level reforms, such as aligning funding flows, streamlining administration, and removing bottlenecks to achieve meaningful efficiency gains.