The Effect of Medicaid HMOs on Spending and Health Outcomes

In the 1980s and 1990s, there was a dramatic shift in employer-provided health insurance from traditional fee-for-service plans to managed care plans - by 1999, 91% of covered employees were enrolled in a managed care plan. The Medicaid and Medicare programs, which together account for nearly $500 Billion in annual spending and 75 million beneficiaries, have also moved to adopt managed care - currently, over one-third of the beneficiaries of the two programs are enrolled in a managed care plan.

A key motivation for the shift to managed care in these public programs is to reduce the level and growth of expenditures. In "Does Contracting Out Increase the Efficiency of Government Programs? Evidence from Medicaid HMOs" (NBER Working Paper No. 9091), Mark Duggan examines the effect of Medicaid HMOs on spending and health outcomes. As Duggan notes, managed care could lower spending without sacrificing quality - for example, paying providers a flat fee per patient may discourage them from providing services of marginal value and encourage more preventative care. However, managed care could also cause spending to rise or quality to fall - for example, the insurer may charge the government a large mark-up over costs if the bidding process is not competitive or providers may fail to provide key services.

It is difficult to measure the effect of HMO participation on spending and outcomes because enrollment in managed care plans is often voluntary, and those who choose to enroll are likely to differ in unobservable ways from those who do not (for example, to be healthier). Duggan avoids this problem by using mandates in 19 California counties that required half of all Medicaid beneficiaries to join an HMO between 1993 and 1999. In his analysis, HMO participation depends only on whether the county had adopted a mandate.

Using data on 300,000 welfare recipients, Duggan finds that the average effect of the mandate is to increase spending by 12%. This increase may be due to higher payments to providers, higher administrative costs, the inclusion of a normal level of profit for the HMOs, or a mark-up of bids above cost. Duggan also finds that the spending increase depends on the level of pre-mandate HMO penetration, with a 20% increase in counties with no previous penetration and no increase in counties with one-third of beneficiaries voluntarily enrolled in HMOs. The lack of a spending increase in counties with large pre-mandate penetration may reflect that counties were already paying HMOs above costs or that these counties have a more efficient managed care market or a more competitive bidding process.

Looking at health outcomes, Duggan finds that switching to a managed care plan leads to a 30% decrease in avoidable hospitalizations for children, suggesting a possible quality improvement. However, he also finds a 30% decrease in hospitalizations for injuries among children, suggesting that the former result is due to a change in treatment patterns rather than to improvements in health. He also finds that switching to a managed care plan does not reduce infant mortality or the incidence of low birth weight babies.

Duggan concludes that mandates requiring Medicaid beneficiaries to switch to HMOs did not improve the efficiency of the Medicaid program because they led to substantial spending increases with no demonstrable quality improvements.