ABSTRACTResearch assessing the relationship between government health expenditure and development assistance for health channeled to governments (DAHG) has not considered that this relationship may depend on whether DAHG is increasing or decreasing. We explore this issue using general method of moments estimation and a panel of financial flows data spanning 119 countries and 16 years. Our primary concern is how DAHG affects government health expenditure as source (GHES). We disaggregate the average effect of DAHG and separately identify the effects of increases versus decreases in DAHG. We find that a $1 year‐over‐year increase in DAHG leads to a $0.62 (90% confidence interval (CI): 0.15, 1.09) decrease in GHES, whereas a $1 year‐over‐year decrease in DAHG does not have an effect on GHES that is statistically different from zero (CI: −0.67, 1.17). Simulation shows that the displacement of GHES between 1995 and 2010 reduced total government health expenditure by $152.8 billion (CI: 46.9, 277.6). Moreover, the irregular disbursement of DAHG reduced total government expenditure by $96.9 billion (CI: 0.5, 212.4). Thus, this research shows that health aid is fungible and highlights the cost of displacement and erratic aid disbursement. © 2013 The Authors. Health Economics published by John Wiley & Sons, Ltd.