High frequency evidence on variation in spending growth
*Robert D. Lieberthal, Jefferson School of Population Health 

Keywords: health disparities, cost curve

My objective was to analyze the variability of healthcare spending and healthcare spending growth rates across gender and age groups. I calculated raw moments of daily spending and daily spending growth rates. I then fit spending and spending growth to simple time series models. The population was employed adults below age 65 and their spouses for a large, employer sponsored plan. I found that the standard deviation of spending was higher for females than for males at all age groups except the oldest (55-64). Comparison of higher moments of spending, skew and kurtosis, did not show a consistent pattern across genders. Daily spending growth rates were more predictable for females than for males across all age groups. The differences in healthcare spending between males and females go beyond higher mean spending for females. In some ways, day to day spending for females is also more variable. The healthcare cost curve for females is less variable than the cost curve for males. Policymakers should consider the different needs of males and females when setting policy and work towards reducing the risks that different groups face. Researchers should focus on using high frequency data to help model longer term trends in healthcare spending.