Cancer has been the leading cause of death in Taiwan since 1982, with nearly one out of four deaths caused by malignant neoplasm. The huge amount of money being spent in the acute care setting for terminally ill cancer patients does not increase their wellbeing. In this study, we employ an Instrumental Variable (IV) model to correct for the self-selection problem and use population-based insurance claim data to test two null hypotheses: there is no difference in total expenditures between hospice care and conventional care, and that there is no difference in total expenditures between hospital-based hospice care and home hospice care. Both null hypotheses are rejected.