|
Health Insurance Expansions For Working Families: A Comparison Of Targeting Strategies
We compare three eligibility criteria for targeting health insurance expansions in working families: poverty, hourly wages, and employment in a small firm. Making pairwise comparisons among these, we find that targeting by poverty is the most effective and efficient. A poverty-based method is also the most effective way to target those lacking access to employer-sponsored insurance and those with low take-up of such coverage. When we examine the effectiveness of targeting by family type, we find that marital status and number of workers in the family make little difference once we control for the presence of children and for poverty level.
Over the past several years analysts have suggested a wide range of incremental health insurance expansion options to cover the more than forty million Americans who are uninsured. These options include individual tax credits, public program expansions, alternative purchasing arrangements, and employer subsidies. An often-neglected aspect of these programs’ details is the choice of eligibility critera. This choice is important because the uninsured are a heterogeneous group.1 Some uninsured persons have low wages but not low incomes. Others work in small firms but have higher wages. When an eligibility criterion is chosen, some uninsured persons will be included, and others will not. There are two important gauges of the success of an eligibility criterion: (1) the number of uninsured persons who will be made eligible under this criterion, measured as a proportion (which we refer to as target effectiveness); and (2) the proportion of the population newly eligible under this criterion that is uninsured (which we refer to as target efficiency). If target efficiency is low, crowding out, or the proportion of already insured persons taking up coverage, will generally be high. A good policy should seek to maximize target effectiveness and efficiency, but these goals are often at odds with one another. This paper compares three targeting criteria: family income measured relative to the federal poverty level, individual hourly wages, and employment in a small firm. As other analysts have, we chose these criteria because uninsurance is concentrated among the poor and among those without access to employer coverage. The latter, in turn, are largely employed in low-wage jobs and small firms. In 1999, 63.2 percent of the nonelderly uninsured had family incomes below 200 percent of poverty.2 Similarly, 75.8 percent of the uninsured in working families had no access to employer coverage in 1998.3 Thirty-four percent of uninsured workers in 1999 were employed in firms with fewer than ten employees.4
Poverty. Proposals that target families by poverty limit eligibility to individuals and families whose incomes fall below a specified percentage of the poverty level, a measure that adjusts for family size and number of children. In 1999, for example, a family of three (one child) with income below $13,410 or a family of four (two children) with income below $16,530 both had incomes below 100 percent of poverty. Examples of poverty-based programs include Medicaid and State Children’s Health Insurance Program (SCHIP) expansions, which raise the upper income eligibility limit for existing programs or extend eligibility to new groups. For example, many states have expanded Medicaid to cover persons with family incomes up to 200 percent of poverty (some even higher). Tax-credit proposals often also use poverty-based eligibility criteria.5 Wages. Proposals that are administered through the workplace may target by hourly earnings, since firms know their workers’ wage earnings and can easily determine eligibility. Plans that use wage criteria generally fall into two subgroups: those that encourage employers to offer insurance, and those that induce workers to take up insurance that is offered. Options of the first type may provide a subsidy for the firm’s share of the premium or allow employers to offer coverage through a public program buy-in.6 Programs to improve take-up usually subsidize the employee contribution to a group plan. Small firms. Some proposals target workers in small firms.7 Small firms do not benefit from the economies of scale in administrative costs achieved by large firms and are at a disadvantage in negotiating premium rates with insurance companies. Nearly all establishments with at least 1,000 employees offered health coverage in 1998, compared with 83.8 percent of firms employing twenty-five to ninety-nine workers and 35.9 percent of firms with fewer than ten employees.8
We used data drawn from the 2000 March supplement (Annual Demographic Survey) of the Current Population Survey (CPS). For some analyses we used a match of the 1999 February (Contingent Workers and Alternative Employment Supplement) and 1999 March files.9 We adjusted the data to reflect health insurance purchasing units and assigned each person a primary source of insurance.10 We calculated hourly wages (based on earnings and on weeks and hours worked) and treated respondents with computed hourly wages below $1 as nonworkers. We limited our sample to families with at least one worker and excluded Medicaid recipients and the elderly (age sixty-five and older), most of whom receive Medicare. Our weighted sample represents 196 million persons in working families, 31.5 million of whom are uninsured. We defined three subgroups based on eligibility criteria: poor/near-poor—income below 200 percent of poverty; low wage—families in which at least one worker earned hourly wages of $6.82 or less (other workers in the family may earn more); and small firm—families with at least one worker employed in an establishment with fewer than ten employees (other workers in the family may be employed in larger firms). We present some analyses for employees only, excluding family members and the self-employed. For employees we used a different wage cutoff ($7.50) and firm-size cutoff (under twenty-five) to equalize population size with poor/near-poor employees. The number of uninsured persons covered under any eligibility criteria depends critically on population size. Proposals with universal eligibility would include everyone and have 100 percent target effectiveness. For this reason, we chose the wage cutoff to make poverty/wage comparisons over similar-size populations. Using these cutoffs, there are approximately equal numbers of persons (forty-four million) in poor/near-poor and low-wage families. The CPS reports firm size in intervals (for example, under 10, 10–25), so we could not exactly match the population- and firm-size criteria. We came the closest by specifying small firm size as fewer than ten workers. There are nearly four million more persons in small-firm families than in poor/near-poor families.
Many proposals use higher thresholds than the ones we used here, although proposals that have received the most attention in Congress tend to use thresholds more like ours.11 We compared our target populations to those that would be obtained from defining poor/near-poor, low wage, and small firm more conservatively and more liberally in Exhibit 1
Cost and coverage implications. Exhibit 2  compares the effectiveness, efficiency, and crowding out associated with each targeting criterion. We report crowding out under the assumption that 15 percent of the insured and 50 percent of the uninsured take up insurance.13 In the top panel, the poverty-based criterion out-performs both the wage and firm-size criteria in terms of target effectiveness, target efficiency, and crowding out.14 The wage-based criterion is slightly more effective and efficient at targeting the uninsured than is the small-firm criterion. We obtain the same rank ordering when we focus on employees only in the bottom panel. Again, the poverty-based criterion performs the best on all counts.
Who is excluded? Who is included no matter what? When a choice is made between two policy options, some people are left out. Other people qualify under either option. When there is a great amount of overlap, cost and coverage considerations can take a backseat to other issues such as ease of administration or outreach.
We explore these possibilities here, comparing (separately) a poverty-based criterion with one based on wages or firm size, and then comparing the wage-based criterion to the firm-size criterion. The top panel of Exhibit 3
Estimates of target effectiveness and target efficiency reflect these differences. Nearly 25 percent of the uninsured live in families that are poor/near-poor but not low wage. Only 7.7 percent live in families that are low wage but not poor/near-poor. Among all those in families that are poor/near-poor but not low wage, one-third are uninsured. Among those in families with a low-wage worker who are not poor/near-poor, only 10 percent are uninsured. The middle panel of Exhibit 3  shows that poor/near-poor families are also more effective and efficient targets than are small-firm families that are not poor/near-poor. There is no difference between the effectiveness of low-wage and small-firm targets, however, as shown in the bottom panel of Exhibit 3. A low-wage target is more efficient, though, because families that include a small-firm worker have high incomes.
Firm offers and take-up of insurance.
Proposals seek to expand employer-sponsored insurance by increasing offer rates or employee take-up rates. In Exhibit 4
Poor/near-poor workers have higher individual offer rates than do those earning low wages or employed in small firms. At the family level, though, these results are reversed. Poor/near-poor persons have less access to employer coverage, are less likely to take up this coverage when offered it, are less likely to be covered by employer insurance at all, and, consequently, are less likely to be insured than are those in low-wage and small-firm families. People in low-wage families, which frequently include a higher-wage worker (65 percent), have the greatest access to employer coverage and are therefore most likely to be insured through an employer. Those in small-firm families, who generally have high incomes (78 percent have incomes over 200 percent of poverty), have greater take-up of employer coverage and overall higher insurance coverage rates.
Targeting family types.
Social policies such as Medicaid and cash welfare assistance have historically given priority to children and parents. In considering health insurance expansions, it is useful to know whether it is more effective or efficient to target some family types over others. Exhibit 5
In Exhibit 5 those with a single primary earner in poor/near-poor families are the largest group, representing about 20 percent of the uninsured (in working families). Almost 12 percent of the uninsured are in poor/near-poor families with children in which there is only one married earner. About 9 percent of the uninsured are in poor/near-poor families with children where the primary earner is single. Within groups with children who have similar incomes, we find that target efficiency is relatively unaffected by marital status and the number of workers in the family. Target efficiency is, however, much higher among poor/near-poor family types without children than among those with children. Since more than half of those in this group are now uninsured, and few are eligible for public programs, making this group eligible for new coverage would crowd out relatively little existing coverage.
Our results suggest that poverty is a better targeting criterion for health insurance expansions than either wages or firm size. Since they target the uninsured more efficiently, proposals focusing on poor/near-poor families are likely to be less costly (per newly insured person). Among the poor/near-poor, differences in target efficiency are driven mainly by the presence of children. A large fraction of the poor/near-poor uninsured are in childless families. Targeting (or at least not excluding) childless families, who have historically been excluded from subsidized health insurance programs, would be relatively efficient. These results raise the issue of how to weigh the pros and cons of various targeting strategies. A complete evaluation of policy options would consider more than just target effectiveness and efficiency. As mentioned above, wage-based options would be relatively easy to administer since employers already carry out many of the administrative functions that would be needed. Moreover, employers could inform workers of their eligibility in conjunction with benefits notifications. Policymakers might prefer a less effective or efficient proposal if it had administrative and outreach advantages. This is a trade-off that analysis can identify but cannot resolve without explicit value judgments.
Danielle Ferry is a research assistant in the Department of Health Policy and Management at Columbia University’s Mailman School of Public Health in New York City. Bowen Garrett is senior research associate at the Urban Institute in Washington, D.C. Sherry Glied is associate professor and chair at the Mailman School. Emily Greenman is earning her doctorate in public policy and sociology at the University of Michigan Population Studies Center. Len Nichols is vice-president of the Center for Studying Health System Change in Washington, D.C. This research was supported by a grant from the Commonwealth Fund (Ferry and Glied) and by the W.K. Kellogg Foundation (Garrett, Greenman, and Nichols). The work of Greenman and Nichols was conducted while each was at the Urban Institute. Rafiq Hijazi (Urban Institute) provided excellent programming assistance.
This article has been cited by other articles:
| |||||||||||||||||||||||||||||||||||||||||
Top
Targeting Criteria
Data And Methods
