Using individual-level administrative data, we investigate the spatial patterns of unexplained shares of health care expenditures (HCE) at the municipality level. The focus is on the elderly population in the Italian Region Friuli-Venezia Giulia observed over 2017-2019. The empirical analysis comprises two steps. First, random-effects two-part models are estimated to analyze the effect of age, morbidity, and death on the probability and amount of positive individual total HCE and its components. Second, the unexplained shares of HCE at the municipality level are examined to identify areas with under- or over-spending and substitution among services. Results confirm the existing findings on the determinants of HCE and reveal geographic patterns in the unexplained shares of expenditures. We observe clusters of municipalities with observed HCE higher than predicted for each type of service and clusters with substitution between home care and all other services. These findings are associated with the degree of urbanization of these areas and, consequently, with the ease of access to health care. This is crucial from a policy perspective, as it indicates specific policy targets for public health intervention.
