Academic journal article Economic Inquiry

Do HMO Plans Reduce Health Care Expenditure in the Private Sector?

Academic journal article Economic Inquiry

Do HMO Plans Reduce Health Care Expenditure in the Private Sector?

Article excerpt


Since the HMO Act was passed in 1973, health maintenance organizations (HMOs) have become one of the most popular types of managed care plans in the U.S. health delivery system. Faced with a dramatic increase in health care costs, HMOs are designed to implement a tight utilization review and authorization process, along with restricted choice of providers, as a means of managing utilization and quality.

By comparing health care expenditure patterns of HMO members with those of non-HMO members among the privately insured, nonelderly population, this study seeks to understand the impact that HMOs have on health care expenditure in the private insurance sector. Our data come from the 2000 Medical Expenditure Panel Survey (MEPS), a nationally representative survey of the U.S. civilian, noninstitutionalized population.

Despite the dramatic growth in the number and proportion of individuals enrolled in private HMO plans, much of existing research on this issue has focused on the relationship between HMO enrollment and use of health care services (Miller and Luft 1994, 2002), whereas only a limited number of studies have been conducted on the role of prevalent HMO enrollment in saving health care expenditure (Kralewski et al. 2000). Analyses on expenditure performance are mostly carried out on the performance of the Medicaid or Medicare managed care plans (Buchanan et al. 1996; Mello et al. 2002).

This study extends previous research in several ways. First, rather than using localized or insurance, group-wise data (Johnson and Aquilina 1986; Kralewski et al. 2000; McLaughlin et al. 1983; McLaughlin 1987; Robinson 1996) or employer data (Baker and Corts 1996; Feldman et al. 1993), we use data provided directly by HMO members through nationally representative surveys. Second, by using decomposed expenditure data, we are able to ascertain whether the provision of financial incentives to both consumers and providers in HMO plans makes any differences in medical care expenditure by each party, compared with traditional plans. Furthermore, by testing the hypothesis of selectivity in health insurance choice, we try to avoid the biased findings that often find in favor of HMO plans.

Trends and Relevance

During the past two decades, the number of HMO enrollees has proliferated from 9.1 million (4% of the total population) in 1980 to 76.1 million (26.4%) in 2001 (Health, United States 2003). Between 1977 and 2001, the percentage of the U.S. population under age 65 that joined private insurance HMOs increased dramatically from 3.7% to 27.9% (Table 1).

Over the same time period, per capita national health expenditure increased from $1,980 to $4,773 (in 1995 constant dollars), and total national health expenditure as a percentage of gross domestic product rose from 8.8% to 14.1%. Out of total national health expenditure, the share of private expenditure decreased from 57.3% to 54.1%.

Motivated by the prevalence of HMOs in the private insurance market and the insufficiency in understanding its cost containment performance, we investigate the patterns of health care expenditure among HMO members to understand whether HMOs contribute to limiting medical expenditure relative to non-HMO arrangement. There are two main complications to our empirical analysis. The first is selection bias due to endogenous health plan choice decisions. Previous studies (Buchanan and Cretin 1986; Jackson-Beeck and Kleinman 1983) suggest that the utilization reduction effect of HMOs may be a consequence of the selective enrollment of a healthier population in HMO plans, rather than the more efficient medical care delivery management of HMOs. Thus, we first estimate the probit model of health insurance plan choice decisions and test this hypothesis in line with the analysis of Taylor et al. (1995). The second complication is that the dependent variables are nonnegative with a high frequency of zero observations. …

Author Advanced search


An unknown error has occurred. Please click the button below to reload the page. If the problem persists, please try again in a little while.