Explaining Health Care Expenditure Variation: Large-sample Evidence Using Linked Survey and Health Administrative Data

Explaining individual, regional, and provider variation in health care spending is of enormous value to policymakers, but is often hampered by the lack of individual level detail in universal public health systems because budgeted spending is often not attributable to specific individuals. Even rarer is selfreported survey information that helps explain this variation in large samples. In this paper, we exploit the linkage of a cohort-representative survey of 265,468 Australians age 45 and over to several years of hospital, medical and pharmaceutical records. After calculating total health care cost for each survey respondent, we examine health expenditures due to health shocks and those that are intrinsic to an individual. We find that high fixed-effects are positively associated with age, especially older males, poor health, obesity, smoking, cancer, stroke and heart conditions. Hospital admissions are the largest component of fixed effects. High time-varying expenditures are associated with speaking foreign language at home, low income and low education, suggesting greater exposure to adverse health shocks. For these individuals, health expenditure is comprised mainly of out-of-hospital medical services and drugs.