Why Do HMOs Spend Less? Patient Selection, Physician Price Sensitivity, and Prices

A long line of research has documented that HMOs spend less on healthcare than other insurers, although the causes of this difference remain unclear. I study HMO spending in the context of statins, the leading class of anti-cholesterol drugs. On average in a given year, HMOs spend 19% less than other insurers on statins. Patients in HMOs and others are roughly equally likely to refill their prescriptions, but prescribing patterns are very different: HMO physicians are much more likely than other physicians to prescribe generics. HMOs and non-HMOs face similar prices on average, but HMO prices are more dispersed: generics are cheaper and branded drugs are more expensive.

To better interpret these facts, I develop and estimate a structural model of patient-physician interactions in the statin market. Partially altruistic physicians choose which drug to prescribe their patient. They trade off patient health and utility against an incentive to prescribe drugs with low insurer prices. Physicians in HMOs may be especially sensitive to insurer prices because their compensation is sometimes tied to the spending their patients generate, which creates incentives for physicians to utilize low-price care, and may attract price-sensitive physicians into HMOs. Once the physician writes a prescription, patients decide whether to fill it each month. In their fill decision, they tradeoff the health benefits of refilling against the costs—out-of-pocket payments and side effects. The model allows for three causes of different spending in HMOs: patients with better health (and therefore lower drug demand) may select into them; physicians may be more price-sensitive; and HMOs may face a different set of prices.

Separating these causes requires separately estimating physician and patient drug preferences in HMOs and other insurance plans. The model suggests that we can learn about patients’ drug preferences—and thus whether healthy patients select into HMOs—by looking at patients’ refill decision.  After accounting for patient preferences, we can learn about the strength of physician incentives by looking at their prescribing decisions.  The model estimates show that patients in HMOs and other plans make similar refill decisions: both respond to copays but not to insurer prices. Physicians in HMOs—and not other physicians--are sensitive to insurer prices, suggesting that HMO’s cost control incentives induce physician price sensitivity.

The estimated model matches the key features of the data well. The predicted HMO spending differential matches the actual spending differential, and the model largely replicates the differences in prescribing patterns by HMO status. I use the model to decompose the spending differential into physician price sensitivity, prices, and selection. Giving all non-HMO physicians the price sensitivity of HMO physicians reduces the HMO spending differential by about a third.  The interaction of lower generic prices and higher price sensitivity explains a further third of the spending differential. The remaining difference is due to selection of low spending patients into HMOs. Overall the results suggest that the high price sensitivity of HMO physicians helps restrain spending with little or no harm for patient health or welfare.