Hui Ding, Mark Duggan, and Amanda Starc

Publication date: 
2025

We study Medicare’s competitive bidding program for durable medical equipment. We use Medicare claims data to examine the effect on prices and utilization, focusing on continuous positive airway pressure devices for sleep apnea. We find that spending falls by 47.2% after a highly imperfect bidding mechanism is introduced. This is almost entirely driven by a 44.8% price reduction, though quantities also fall by 4.3%. To disentangle supply and demand, we leverage differential cost sharing across Medicare recipients. We measure a demand elasticity of -0.272 and find that quantity reductions are concentrated among less clinically appropriate groups.

Plots of event study coefficients of relative years interacting with group dummies. The observation is at CBSA by year-quarter by subgroup level. Panel A groups sample to duals vs. non duals. Panel B and C define subgroups as those diagnosed with either SA (primary diagnosis recorded as ICD-9 code 327.2 or ICD-10 code G47.3 in an inpatient, outpatient, or physician carrier table) or OSA (primary diagnosis recorded as ICD-9 code 327.23 or ICD-10 code G47.33) in the previous year, restricting to patients who remained enrolled in Medicare for the current and previous three quarters to get full information on their diagnosis (sample period starts from 2009Q4). Outcome variable is the total quantity of product per thousand Medicare FFS recipients, in natural log terms. The gray solid lines show point estimates for the effect of the CBP for duals/No SA/No OSA diagnosis, and the black dashed lines show point estimates for the effect of the CBP for non duals/SA/OSA diagnosed. Vertical lines show the 95% confidence interval for those coefficients, based upon standard errors that are clustered at the CBSA level. All regressions include CBSA fixed effects, year-quarter fixed effects, and their interactions with a dummy variable for each group. All regressions are weighted by Medicare FFS population in 2010.