|
Generosity And Adjusted Premiums In Job-Based Insurance: Hawaii Is Up, Wyoming Is Down
This paper reports national and state findings on the generosity or actuarial value of U.S. employer-based plans and adjusted premiums in 2002. The basis for our calculations is simulated bill paying for a large standardized population. After adjusting for the quality of benefits, we find from regression analysis that adjusted premiums are 18 percent higher in the nation’s smallest firms than in firms with 1,000 or more workers. They are 25 percent higher in indemnity plans and 18 percent higher in preferred provider organizations than in health maintenance organizations. The generosity of coverage increased from 1997 to 2002.
AMERICAN WORKERS WERE POLLED RECENTLY about the attributes of health plans that made for a "good insurance plan." They cited low monthly contributions, rich benefits, freedom to choose their providers, and first-dollar coverage with low copayments.1 However idealized their response, employees raised the question, What is "good insurance," and which employers offer it? A related issue is the price of insurance when adjusted for the financial protection it affords. Thus, when analysts address the failings of the small-employer market, they are generally referring to the accepted belief that small employers pay slightly more for coverage but receive less financial protection. In other words, small employers pay higher adjusted prices. This paper examines variations in the generosity of coverage provided by various employers and insurance plans at both the national and state levels. Major measures of interest are the actuarial value of the plan and its adjusted price. This paper makes four contributions to the discussion. First, we update recent calculations about the generosity of employer-sponsored health insurance and how it varies by type of health plan, employer, and employee characteristics. Second, we update statistics about the cost of health insurance at the state level and, to our knowledge, provide the first state-specific estimates of the generosity of insurance. Third, we calculate and display "adjusted premiums" at the national and state levels. Finally, using regression analysis, we investigate determinants of actuarial value and adjusted premiums. State-based statistics on job-based health insurance are uncommon because of the large sample necessary for each state.2 The Agency for Healthcare Research and Quality (AHRQ) has published information from the 2002 Medical Expenditure Panel Survey-Insurance Component (MEPS-IC) survey on its Web site, by state, for selected characteristics of health insurance.3 There are no state-based data on plans’ actuarial value. Throughout the paper, we use the term "adjusted premium" or "adjusted price." This might more accurately be termed "price adjusted for quality of benefits." Our measure of adjusted price does not consider such factors as the breadth and quality of the provider network, the speed and accuracy of claims administration, the quality of communication with employees, and other factors.
Data sources and approach. This study calculates actuarial value (the percentage of the medical bill that the health plan will pay, on average, for a standardized population) through simulated bill paying.4 For each plan in the sample, we ask: "Given the utilization patterns of a standard population, what percentage of the bill would be paid by the plan, and what expected percentage would that employee pay out of pocket?" For employees with job-based coverage, we present actuarial value for the total population, the top 10 percent of spenders, and the bottom 50 percent of spenders.5 This analysis draws on MEPS-IC data for 2002, MEPS Household Component (MEPS HC) data for 2000, and data from the Henry J. Kaiser Family Foundation/Health Research and Educational Trust (KFF/HRET) Employer Benefit Surveys of 2000–2004. MEPS surveyed approximately 40,000 establishments in 2002. The 76 percent response rate yielded information on about 30,000 establishments, about two-thirds of which offered at least one health plan. The sample produced reliable estimates for forty-six states.6 MEPS HC for 2000 is the source of information for the use and cost of services in the simulated bill paying. Only adults with employer-based insurance were included in the claims database. Our final sample included 9,519 adults, representing approximately 117 million Americans. The MEPS household survey provides data on spending and usage for (1) hospital inpatient care; (2) emergency department (ED) and other outpatient care; (3) office-based visits to providers; (4) prescription drugs; and (5) home health and other medical supplies/equipment.7 The KFF/HRET annual employer surveys are a random sample of public and private employers with three or more workers. The response rate for 2002 was 50 percent. The surveys include certain information on plan features not contained in the MEPS-IC survey.8 Using regression analysis, we imputed information from these surveys to the MEPS-IC database for a number of plan benefits and cost-sharing variables. Spending and use estimates. To develop the distribution of health care spending, we adjusted the medical claims database from the MEPS household surveys with information from the National Health Expenditure Accounts (NHEA) for 2002. We used NHEA estimates to make an initial calibration of per adult spending for the population under age sixty-five in 2002, adjusting MEPS-IC spending in each of ten service categories that were then collapsed to the five used in our simulations.9 To adjust for the different spending levels associated with different types of plans, we used data from Watson Wyatt’s proprietary medical claims database. Watson Wyatt data were also used to adjust for differences in use of out-of-network providers.10 Using data from the Area Resource File (ARF), we made adjustments for local differences in the cost of medical care using area cost factors derived from the Medicare adjusted average per capita cost (AAPCC) index. Simulated payment of claims. We simulated payment of medical claims using the standard adult population described previously. For each health plan in the MEPS-IC sample, we calculated the percentage of the bill paid by the plan and the employee. One notable study limitation is that we did not have access to data on enrollees’ demographics or health status. Calculating quality-adjusted premiums. To calculate adjusted prices, we divided the premium by the actuarial value and then normalized the distribution so that the sample mean for adjusted price equals the sample mean for premiums. MEPS-IC contains insufficient data elements to adequately adjust for differences in health status across establishments. Hence, for individual employers and health plans, some differences in premiums are attributable to differences in covered populations’ health status. For larger populations, presumably some of the differences are smoothed over because of the law of large numbers. Premium equivalents for self-insured firms were treated the same as those for non-self-insured firms. Distribution of claims expenses. We based our estimates of actuarial value on the total medical expenses that a large, standardized population would incur under each plan type. These expenses do not reflect expenses that the median employee would incur. Medical expenses are heavily concentrated among a small percentage of the population.11 Analysis and statistical testing. To understand the distribution of different measures of generosity and price, we conducted descriptive and multivariate analyses. Key dependent variables in the descriptive tables included actuarial value and quality-adjusted price, as well as standard measures of health plans: deductibles, enrollment, premiums, and out-of-pocket maximums. Key independent variables included firm size, industry, earnings of employees at the firm, rural or urban location, and type of plan. In calculating standard errors, using SAS version 8.12, we adjusted for design effects of the MEPS-IC cluster. In multivariate analysis, our dependent variables were actuarial value and adjusted price. We confined our independent variables to factors that were not jointly determined with these variables. Thus, we did not use plan attributes such as deductibles, out-of-pocket maximums, copayments, covered benefits, and benefit limits as explanatory variables. Key sets of independent variables were predetermined factors for employers: firm size, industry, state, rural or urban location, cost of medical inputs in the county, and county per capita income. Plan type, such as health maintenance organization (HMO), preferred provider organization (PPO), and so on, was also included.12
Descriptive findings. Employees in small firms are more likely than their peers in large firms to face deductibles, and when employees in small firms face deductibles, they tend to be greater than those paid by their peers in large firms (Exhibit 1  ). Except for professional services, differences in deductibles across industries are relatively small. Differences across plan types are substantial, though. The percentage of employees with out-of-pocket limits varies little by firm size, as does the average out-of-pocket maximum.
The average actuarial value of an employer-based health plan was 83.2 percent (Exhibit 2 ). For the highest 10 percent of users, the plan pays 86 percent of expenses, and the employee pays 14 percent. For the lowest 50 percent of users, the plan pays 72 percent of the bill. Exhibit 2 shows little variation in actuarial value by firm size, industry, or workers’ wages; however, there is much variation by health plan type. Also, plans in metropolitan-area firms scored about four percentage points higher than in establishments located in rural areas.
Adjusted premiums for single coverage exhibit greater differences by firm size. The average quality-adjusted premium for the smallest U.S. firms (1–9 workers) is approximately 20 percent higher than that for firms with 1,000 or more workers. Differences between HMO and indemnity plans in adjusted prices are also about 20 percent (calculation not shown). In general, there is greater variation in generosity for the bottom 50 percent of users than the top 10 percent. Actuarial value increased approximately eight percentage points from 1997 to 2002. Much of this increase is attributable to two factors. First, networks are broader, so out-of-network use is lower. Out-of-network use drives up out-of-pocket costs both because plan cost-sharing provisions are less generous and because provider charges are likely to exceed fee limits and result in balance billing. Secondly, many PPO and point-of-service (POS) plans have switched from coinsurance to copayments for their in-network provisions. In most cases, employees are still subject to a deductible when they use out-of-network providers. With copayments, members receive plan payments on the first visit, and their out-of-pocket costs are limited to fixed amounts per visit or per prescription filled. The net effect of the shift from coinsurance to copayments was to move most employees to first-dollar coverage.
When we examined the data by state, we found that mean actuarial value ranges from 73.4 percent in Montana to 87.6 percent in Massachusetts (Exhibit 3
Adjusted premiums for single coverage range from $4,001 in Wyoming to $2,717 in Hawaii (Exhibit 3  ). Wyoming, Maine, Wisconsin, and West Virginia, all states with substantial percentages of rural residents, have the highest adjusted prices. Hawaii, California, Alabama, and Arizona have the lowest adjusted premiums; all of these states except Alabama have HMO plans with large market shares.
Multivariate results.
Based on ordinary least squares (OLS) estimation, we found that plan type is the most significant determinant of actuarial value (Exhibit 5
In this paper we have reported a number of surprising findings and documented the magnitude of some conventional wisdom. First, type of plan proved to be a stronger force than the type or size of employer in determining generosity and adjusted premiums. Differences in adjusted premiums between PPOs and HMOs are attributable to HMOs’ lower unadjusted premiums, lower cost sharing, and richer benefits.13 Second, owing to the strong effect of plan type, adjusted premiums tend to be higher in rural states, such as Montana and Wyoming, than in high-cost-of-living, urban states such as Massachusetts, California, and Hawaii. In general, states with strong HMO market shares tend to have lower adjusted prices than rural states, particularly those few states where indemnity insurance still has a strong presence. Third, we have documented the inefficiency of the small-employer market. Adjusted premiums, other factors held constant, are 18 percent higher for the smallest U.S. firms (1–9 workers) than for firms with 1,000 or more workers. Higher administrative costs from marketing, medical underwriting, greater risk, and other diseconomies associated with small size contribute to this. Fourth, actuarial value rose approximately eight percentage points from 1997 to 2002. This increase is attributable largely to two factors: breadth of networks and conversion to copayments. Let us expand on the latter point. In 1997, 54 percent of PPO enrollees belonged to a plan with coinsurance and 38 percent, to a plan with copayments; the remainder faced no cost sharing.14 When employees belong to plans with copayments, deductibles do not apply when they use the services of providers, such as physicians, for whom cost sharing is in the form of copayments. By 2002, 56 percent of PPO enrollees belonged to a plan with copayments; 22 percent, to a plan with coinsurance; and the remainder, to either a free-care or a combination plan.15 Enrollment in POS plans with coinsurance fell from 33 percent to 4 percent during 1997–2002. In our regression analysis, we estimated that actuarial value increases eight percentage points when an employer shifts from a plan with coinsurance to one with copayments. National employee benefit surveys have reported sizable increases in deductibles during the past five years.16 What has been overlooked is that in most cases, these deductibles did not apply to physician visits. This shift to plans with copayments meant that more Americans had first-dollar coverage and thus greater demand for services. The growing number of plans with copayments likely contributed to the rise in health spending since 1998. Although actuarial value of employer-sponsored plans rose from 1997 to 2003, average out-of-pocket expenses also rose, from $539 to $563.17 For low users, out-of-pocket spending declined from $279 to $46, while spending for the top 50 percent of users increased from $779 to $1,049. The decline in spending for low users reflects the shift to copayment plans and broader networks, as discussed previously. The increase in out-of-pocket spending for high users reflects higher health care spending driven by higher prices and utilization. We should note that 2002 was the first year of "buy-downs" as reported by Wall Street analysts.18 These buy-downs take the form of higher deductibles, copayments, coinsurance, and tiered cost sharing for items such as prescription drugs. In an analysis of HRET data, we found that actuarial values had declined slightly since 2002 but were still five points greater than 1997 figures. Given the documented disparities in the size of deductibles, why are the differences in overall actuarial value not greater across firm sizes, low- and high-wage firms, and industries? First, a large portion of total spending is for people who incur high health care costs. In fact, actuaries estimate that the majority of spending in a typical employer-based plan is for people who have exceeded the out-of-pocket maximum. Second, the shift to copayments has negated the impact of rising deductibles among low spenders, who now receive large plan payments for services that would fall below normal deductibles. In an analysis of plans with coinsurance only, we found that differences in actuarial value between small and large firms within each plan type was roughly 10 percent. Of course, with coinsurance, employees must exceed the deductible before the plan pays for any services.19 One limitation to our calculations is that we did not adjust for pre-existing condition clauses. Our measure of "adjusted premiums" pertains to the benefit rate, not to the quality of providers in the network, the efficiency of claims processing, and communication with employees. Moreover, our estimates include spending only for those service categories typically covered by employer-sponsored insurance; they exclude such categories as over-the-counter drugs. Nonetheless, we do not believe that these limitations bias our estimates to any great degree. Many employers are considering adopting high-deductible health plans to dampen consumers’ demand for health care services. Consumer-driven plans with high deductibles aim to make consumers price-sensitive and ensure choice at the point of service, but managed care plans with more modest levels of cost sharing, such as HMOs, will likely continue to represent an important alternative for both employers and employees.
Jon Gabel (jgabel{at}hschange.org) is a vice president of the Center for Studying Health System Change (HSC) in Washington, D.C. Roland McDevitt is director of health research at Watson Wyatt Worldwide in Arlington, Virginia. Laura Gandolfo is a research analyst there. Jeremy Pickreign is a statistician at HSC. Samantha Hawkins is a research associate at the Health Research and Educational Trust in Washington, D.C. Cheryl Fahlman is a health researcher at HSC. The authors thank the Commonwealth Fund for its financial support, Tom Rice and Peter Cunningham for their helpful comments, and the Agency for Healthcare Research and Quality (AHRQ) and the U.S. Bureau of the Census for providing access to study data. Jon Gabel, Jeremy Pickreign, and Cheryl Fahlman were employees of the Health Research and Educational Trust at the time this research was conducted.
This article has been cited by other articles:
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
Top
Study Data And Methods
Study Findings
