This Article Abstract Submit a response to this article Services E-mail this article to a friend Alert me to new issues of the journal

D A T A W A T C H
P L A Y - O R - P A Y
W E B  E X C L U S I V E   

13 October 2004  
  

Employers’ Responses To A 
Play-Or-Pay Mandate: An Analysis 
Of California’s Health Insurance Act 
Of 2003

Employers’ Responses To A Play-Or-Pay Mandate: An Analysis
Of California’s Health Insurance Act Of 2003

By Anna D. Sinaiko

ABSTRACT:

California recently enacted a play-or-pay employer mandate (known as SB 2) to expand health insurance coverage to a portion of the state’s working uninsured population. Implementing SB 2 will change the behavior of employers affected by the legislation, with consequences for California’s labor market and uninsured citizens. This paper applies findings from the literature and an economic analysis to California data to estimate the potential reduction in wages, quantify the dispersion of risk across employers, and discuss other employment effects. These employer responses will reduce the number of uninsured people in California who gain eligibility for health coverage under SB 2.

During the past decade, policy making to address uninsurance in the United States has fallen to the states.¹ However, state budget deficits make expanding public programs to cover the uninsured financially and politically difficult.² In this context, employer mandates are an attractive off-budget alternative because they build on the existing system without requiring sizable government spending. Moreover, because 82 percent of the uninsured are in working families, and 70 percent of these families include a full-time worker, employment-related policies might be able to reach a large portion of the uninsured.³

California, with 20 percent of its population uninsured and an unemployment rate of 6.2 percent, recently enacted the Health Insurance Act (known as SB 2), a play-or-pay employer mandate to expand health insurance coverage to a portion of the state’s working uninsured population.⁴ This law requires employers to either “play,” by providing employee health benefits that meet a minimum standard, or “pay” a fee to the state to cover workers under a state-sponsored program. Specific provisions of this program are still being determined. Although the law is scheduled to be phased in starting in 2006, a coalition of business groups is working to overturn it by referendum in November 2004.⁵

The effect of this mandate on California’s employers and uninsured citizens is not clear. Studies have estimated that under current conditions, 1.07–1.56 million people will become eligible for health benefits.⁶ However, implementing SB 2 may change employers’ and workers’ behavior in ways that will alter the number of workers actually affected by the law.

Evidence from other states that can inform an analysis of California is limited. A few states have passed employer mandates for health benefits that use a play-or-pay structure, but only the Hawaii Prepaid Health Care Act (HPHCA) has been implemented.⁷ Previous studies have assessed the impact of the HPHCA, but it is difficult to generalize results from Hawaii to other states.⁸

This paper applies economic analysis and empirical findings from the literature to California data to discuss potential employer responses to SB 2, including wage reductions and changes in workforce composition. Employees’ behavioral responses to SB 2 are also briefly discussed. However, both employer and employee responses will be sensitive to the cost of health benefits under the state program created by SB 2 and to the value that employees place on having health insurance. Because available data on these topics are limited, the discussion remains speculative.

Data And Methods

Data sources. The analyses in this paper use publicly available data, primarily two surveys: the 1997 Robert Wood Johnson Foundation Employer Health Insurance Survey (EHIS) and the 2001 California Health Interview Survey (CHIS).⁹ The EHIS is a national employer-level survey; analysis was conducted on the set of employers from California who had at least twenty employees and would be affected by SB 2 (n = 705).¹⁰ The California employers in the EHIS were more likely than all California firms in the late 1990s to be in the mining/manufacturing or professional services industries and to be larger in size.¹¹ The 2001 CHIS, a statewide survey that asked about public health topics including health insurance coverage, was analyzed for the household income of uninsured workers who would become eligible for health insurance under the new law.

Analyses. An analysis of the potential reduction in wages under SB 2 was conducted using data from the California Employer Health Benefits Survey.¹² Because the most recent premium data for that survey are in 2002 dollars, dollar amounts were adjusted to 2002 dollars for this analysis. All other analyses are in 2003 dollars.

This paper also presents the proportion of employees who are eligible for benefits under SB 2 and who earn the minimum wage. This analysis uses the EHIS data, and a worker’s wage level is assumed to be independent of the number of hours worked each week.¹³ Limitations in the data made it impossible to test this assumption.
Finally, this paper discusses the dispersion of risk across employers and presents relevant data in two ways. First, an analysis was conducted of employers in the EHIS data set based on the distribution of workers across four age categories and the percentage of female workers at each firm, but not workers’ health status.¹⁴ It is assumed that the overall percentage of female workers at an employer is consistent for each age category.¹⁵ The average health risk at each firm was calculated as a weighted average of the risk of each of the age/sex groups using national estimates of the relative cost to insure different demographics.¹⁶ Results were normalized to a mean of 1.00.

A second analysis of risk uses data from the American Medical Association (AMA) and California Medical Association (CMA) on the variability in expected medical expenses across groups, by size of group.¹⁷ The AMA/CMA data report the factors by which expected spending rises over the average for the most costly 5 percent and most costly 25 percent of groups in each size category. Assuming that groups’ medical spending is normally distributed, these data are used to determine the relative factor (in comparison to the mean) of expenditures for each decile along the distribution.

Labor-Market Responses

The cost to employ any worker in terms of total compensation is the sum of cash wages and fringe benefits. Employers will offset increases in labor costs from providing health insurance by shifting the cost to workers through lower wages or to consumers through higher prices, or by changing the composition of their workforces.

Reduced wages. The employer mandate contemplated by SB 2 provides that employers contribute 80 percent of the cost of coverage and employees contribute the remainder; however, economic theory predicts that over the long term, employers will pass the cost of benefits to employees through lower wages.¹⁸ This is possible when workers value health insurance at its cost (or more), so they will accept the trade-off of income for health benefits. If workers value the benefit less than its cost, employers will be able to shift only a portion of the benefit costs to employees, the amount of which will depend on the elasticities of labor supply and demand.

Previous studies report that in general, 83–100 percent of the costs of health insurance are shifted to employees through reduced wages.¹⁹ The average annual premium in 2002 in California was $2,845 for a single subscriber and $7,471 for families.²⁰ Considering single coverage first, the cost to employers after accounting for the employee contribution is $2,276. The average income of workers who would be eligible for insurance under SB 2 was approximately $31,100.²¹ At this wage, an employee would lose $0.363 of every dollar earned to taxes.²² If employers were to pass 100 percent of their health plan costs to their employees, an employee with single coverage who works forty hours per week would see a reduction in net wages of $1,450 ($2,276 x 63.7 percent), or $0.72 an hour. If employers passed along only 83 percent of costs, employees would experience a decrease in wages of $0.60 per hour. For family coverage, the reduction in wages would be much higher, ranging from $1.58 to $1.90 per hour.²³

The reduction in wages can also be calculated as a proportion of income after taxes. After taxes, a person with income of $31,100 would net $22,190. The reduction in wages described above would constitute 7 percent of after-tax income for single subscribers and 17 percent for family coverage. Exhibit 1 summarizes sensitivity analysis of these findings; adjusting the costs of premiums ± 10 percent suggests that there could be a reduction in wages of 5–7 percent for single coverage and 13–19 percent for family coverage.

Although employees who elect family coverage pay a higher premium than employees with single coverage, when it comes to wage levels, antidiscrimination regulations, union rules, and workplace norms will prevent employers from discriminating on the basis of marital or family status and reducing the wages of workers by different amounts.²⁴ It is more likely that wages will fall for all employees by the average amount of increase in their employer’s labor costs, which is higher when the proportion of workers with family coverage is high. California workers at firms that now offer insurance but pay less than 80 percent of premium costs are also likely to experience a reduction in wages, the magnitude of which will depend on the relative generosity of the employer’s previous contribution and benefit package.

The previous analysis is based on actual average 2002 health premiums in California. However, because employees at firms that do not offer insurance are more likely to be young, be male, work fewer than forty hours per week, and have lower earnings (Exhibit 2), and if firms not previously offering insurance offer less generous plans, premiums in these firms may be lower than this published average. Conversely, firms without insurance benefits are more likely to be small, with an average higher cost of coverage and an average lower employee wage. Therefore, the percentage reduction in wages for employees at small firms could be greater.

This analysis of the potential reduction in wages under SB 2 shows that insurance costs, as a percentage of total employee compensation, are higher for low-wage workers and represent a greater percentage reduction in net wages. Other play-or-pay proposals attempt to compensate for this inequity by including subsidies for employers with low-wage workers.²⁵ Although SB 2 does not include such a subsidy, other laws can limit wage reductions below fixed limits.

Increased prices. If a set of firms in the same industry and local market are required to begin providing health insurance under SB 2, and if the demand for their products and services is relatively inelastic, then these firms could raise prices to offset the cost of health coverage. Approximately 90 percent of California firms that do not offer benefits report that their competitors also do not provide benefits.²⁶ This could be true for some service-sector firms, such as restaurants or retail stores, where SB 2 would affect competitors’ cost structures similarly and where there are no effective competitors outside the local market. However, many firms face national competition and cannot use this strategy.

Changes in the workforce. Employers not able to fully shift the cost of health benefits to their employees or unable to raise the prices of their products are likely to attempt other employment strategies.²⁷ Extension of health benefits to part-time workers is more costly relative to their total compensation. The increase in labor costs of single coverage, valued at California’s average premium, for a previously uninsured part-time worker (working twenty-five hours per week) is $1.16 per hour—60 percent higher than that for a full-time worker. Employers can avoid this cost by restructuring part-time jobs to full-time jobs or by using more overtime. Research has shown that as health costs rise, firms substitute hours per worker for the number of workers employed.²⁸

In the presence of minimum wage laws, employers may respond to increasing health insurance costs with layoffs. The following analysis considers low-wage employees at firms that do not offer insurance. On average, 9–11 percent of workers eligible for benefits under SB 2 at firms in the EHIS data earned the minimum wage, and 29.0–34.5 percent earned a wage within $2 of the minimum wage.²⁹ These data can be used to estimate upper and lower bounds for the potential reduction in the number of uninsured workers under SB 2. If all of these workers were to lose their jobs, 38–45 percent of workers at firms not offering health insurance would not gain coverage under SB 2. If none of these workers lost their jobs and there were no other employment effects stemming from the mandate, they all could be offered health coverage; however, whether they actually become insured will depend on how each worker values the health benefit and whether he or she decides to take it up. Minimum-wage workers at firms that offer insurance also may be subject to job loss if they become eligible for benefits under SB 2; given limitations in the EHIS data, it was not possible to complete a similar analysis for these workers.

Employers may also use more workers who are ineligible for benefits. Under SB 2, these are employees who work fewer than 100 hours per month or fewer than three months in a year.³⁰ Farmers or other agriculture entities that rely heavily on seasonal workers could adjust the length of time for which they hire workers, to ensure that they do not meet minimum eligibility criteria. This is particularly relevant because a disproportionate number of California’s uninsured workers are in the agriculture industry.³¹ Moreover, as new firms enter the market, they are more likely to rely on part-time or temporary workers who are ineligible for benefits under SB 2, giving them a competitive advantage over existing firms.

Employers could avoid the costs of SB 2 by reducing their total numbers of employees through downsizing or outsourcing. The law defines categories of employers by their numbers of employees and varies provisions by category. Twelve percent of California firms in the EHIS data could reduce the size of their workforces by 5 percent or less and drop to a smaller size category; quantitatively, this would affect only a few workers.

Finally, employers may simply refuse to comply with the law, which is more likely if penalties for noncompliance and the risk of getting caught are low. In his analysis of the HPHCA, Andrew Dick found substantial employer noncompliance, which reduced the number of eligible workers who gained health insurance under the program.³²

Potential for adverse selection. The outcome of an employer mandate will depend on how the cost of participating in the public pool—the “pay” option—is determined. The demographic mix of employees across employers is highly variable, and the cost to insure certain workers can be half as much as the cost to insure others (such as the difference in insuring a thirty-year-old versus fifty-year-old male).³³ Moreover, additional heterogeneity exists within demographic groups. As a result, costs can vary greatly between firms.

An analysis of the dispersion of risk determined by the demographics of employees in the EHIS data confirms the variation among employers (Exhibit 3). The average health risk (an indicator of premium cost) of employees in the highest decile of California firms is 2.78 times that of employees in the lowest decile, and employers with high-risk employees are more likely to offer insurance. These findings are similar to an earlier research finding that for employer-provided individual insurance, the premium in the top decile is 2.5 times that of the bottom decile.³⁴

The variability of risk across groups can also be analyzed with data on expected medical spending. Relative cost factors published in 2001 by the AMA/CMA can be used to estimate that the variability in use of health services is higher than presented above for small groups but decreases as groups grow larger. For groups with fifty members, expected medical spending for the highest decile is 160 percent of the mean, and expected spending by the highest decile is four times that for the lowest decile (Exhibit 3).
Employers will compare the private market and public pool costs when deciding whether to play or pay. If the fee to participate in the public pool is not adjusted for demographic and health risk so that firms with above-average health risks could realize lower costs in the public pool than in the private market, the public pool may attract groups with higher average health care costs, and the cost of participating in the public program would rise.³⁵

SB 2 has attempted to address the risk of adverse selection by stipulating that the fee to participate in the public program be set according to the actual risk of the group of employees. This is unlike many previous play-or-pay proposals at both the state and federal levels. Depending on its execution, this provision could make the cost of participating in the pool comparable to a firm’s premiums to provide benefits in the private market, reduce opportunities for selection, and make employers more neutral in their decisions of whether to play or pay.

Employee response. The focus of this paper has been employers’ responses to SB 2, but employees’ responses to offers of health insurance and to the mandate will also affect the reduction in the number of uninsured people. The decline in acceptance of employer-sponsored health insurance (which increased from 12 percent to 15 percent during the 1990s) has been shown to be the most important factor behind the decrease in this type of insurance, and cost is the primary reason employees do not accept coverage.³⁶ Employees face an additional benefit cost in the amount of the reduction in their cash wages under SB 2, because a portion of the compensation they receive is in the form of health benefits. It is difficult to assess how workers at firms that do not offer health insurance will respond under the mandate. Workers who do not value health benefits may switch jobs to work for an employer that is exempt under SB 2 to regain their former wage levels, or they may become self-employed. The net effect of this activity is that fewer workers will be insured. Trends in employee take-up of health benefits under a play-or-pay mandate are an important topic for further study.

Discussion And Policy Implications

With the passage of SB 2, California has brought the employer mandate back to the forefront of the health policy reform debate. Although state governments can mandate that employers provide coverage without committing public funds, someone must pay for the newly provided health insurance. In the best-case scenario, workers will value health insurance at more than its cost, and the combination of health insurance and lower wages will make workers better off without any adverse effects on employment or product prices. A more likely scenario is that some of the cost of health insurance will be passed to consumers, some felt by workers as unwelcome wage reductions, and some avoided by firms as they restructure their workforces. In California, workers at firms not offering benefits are more likely to be young males earning less than workers at firms that offer insurance; SB 2 will affect this group most adversely.

As employers restructure their workforces, fewer would-be-eligible workers will be offered health insurance. Although SB 2 will bring eligibility for health coverage to a portion of the uninsured in California, it will be lower than the estimated 1.07–1.56 million that researchers have projected. Moreover, because SB 2 does not include an individual mandate requiring workers to take up the coverage offered by their employers, some employees will decline employer-sponsored coverage, and fewer workers will be insured. This finding is not unique to California; similar results would be expected in other states.

Study limitations. A limitation of this analysis is that the bounded estimates of the potential reduction in the number of uninsured workers at the minimum wage are based on the EHIS data from 1997. The upper bound is likely underestimated because the minimum wage has since increased to $6.75. At this new level, the proportion of workers within $2 of the minimum wage is likely to be greater because increasing the minimum wage toward the mean wage affects a higher proportion of the workforce. However, although a more current data set that reflects recent changes in the labor force and wages would alter the magnitude of the estimates reported here, the behavioral responses are expected to be the same.

Future research. This analysis suggests an important topic for future research. Under pressure from rising labor costs, employers may change the types of benefits they provide to employees. They may raise employees’ cost sharing or offer consumer-directed plans, or they may reduce the scope of benefits (for example, drop drug coverage), retiree health benefits, or other fringe benefits (such as pensions) as strategies to keep fringe benefit costs low.³⁷ Policymakers assessing whether the employer mandate is a good reform model should pay close attention to these potential consequences. If the changes result in more efficient consumption of health care by consumers, this will be a positive outcome. If instead the changes lead to increased adverse selection and gaps in coverage for people with chronic conditions, the failures of the health insurance market will be exacerbated, and populations will be at increased risk of being uninsured.

The author gratefully acknowledges research support from P01 HS10803 from the Agency for Healthcare Research and Quality (AHRQ). She also is grateful to David Cutler, Tom McGuire, Joe Newhouse, Murray Ross, Sara Singer, two anonymous referees, and the editors for helpful comments on earlier versions of this paper.

NOTES

1. N. Kaye, M. Marchev, and T. Riley, Building a Pathway to Universal Coverage: How Do We Get from Here to There? November 2002, www.nashp.org/Files/GNL49_FTF_III.pdf (26 August 2004).
2. Kaiser Commission on Medicaid and the Uninsured, “States Respond to Fiscal Pressure: A Fifty-State Update of State Medicaid Spending Growth and Cost Containment Actions,” January 2004, www .kff.org/medicaid/7001.cfm (28 July 2004).
3. Kaiser Commission on Medicaid and the Uninsured, “The Uninsured and Their Access to Health Care,” January 2003, www.kff.org/uninsured/1420-05.cfm (29 September 2004).
4. California HealthCare Foundation, “Cover the Uninsured Week: May 10–16, 2004,” www.chcf.org/topics/view.cfm?itemID=102217 (18 July 2004); and State of California Employment Development Department, “California Seasonally Adjusted Labor Force and Unemployment Data,” July 2004, www.calmis.ca.gov/file/lfhist/cal$shlf.txt (18 July 2004). For an overview of SB 2, see CHCF, “The Health Insurance Act of 2003: An Overview of SB 2,” November 2003, www.chcf.org/documents/insurance/sb2/overview/SB2FactSheet2.pdf (29 September 2004).
5. For more information, see Californians Against Government Run Healthcare, www.stopthehealthtax.org (28 July 2004).
6. E.R. Brown et al., “SB 2 Will Extend Coverage to One Million Uninsured Workers and Dependents,” September 2003, www.healthpolicy.ucla.edu/pubs/publication.asp?pubID=75 (9 September 2004); and A. Dube, “Impact of SB2 on Health Coverage,” Research Brief, 9 September 2003, www.iir.berkeley.edu/research/healthcoverage.pdf (9 September 2004).
7. T. Oliver, “State Employer Health Insurance Mandates: A Brief History,” March 2004,
www.chcf.org/documents/insurance/sb2/employman/EmployerInsuranceMandates.pdf (9 September 2004).
8. A.W. Dick, “The Impact of Mandated Employer Provision of Health Insurance Benefits: Evidence from Hawaii” (Doctoral dissertation, Stanford University, 1994); and N.K. Thurston, “Labor Market Effects of Hawaii’s Mandatory Employer-Provided Health Insurance,” Industrial and Labor Relations Review 51, no. 1 (1997): 117–135.
9. S.H. Long and M.S. Marquis, “Robert Wood Johnson Foundation Employer Health Insurance Survey,” 1997, webapp.icpsr.umich.edu/cocoon/ICPSR-STUDY/02935.xml (21 September 2004); and California Health Interview Survey, “CHIS 2001 Adult Public Use Data File” (Los Angeles: University of California, Los Angeles, Center for Health Policy Research, April 2004).
10. California employers are located in Riverside–San Bernardino, San Francisco, Los Angeles–Long Beach, Orange County, Santa Rosa, Modesto, and one unidentified area.
11. Qualitatively the survey underrepresents the agriculture industry; because this industry represents only 4 percent of the state’s employers, the effect on the estimates reported here is likely to be small. Results are presented as a range because the data set does not allow analysis of employees who work 25–34 hours per week (and are eligible under SB 2) separately from those working 20–24 hours per week (who are not eligible). For more information on how California employers in the EHIS data compare with the state overall, see Online Supplemental Exhibit 1, content.healthaffairs.org/cgi/content/full/hlthaff.w4.469/DC2.
12. Henry J. Kaiser Family Foundation and Health Research and Educational Trust, “Chartbook: The Health Insurance Act of 2003 (SB2): Updated Findings from the 2002 California Employer Health Benefits Survey,” 5 October 2003, www.kff.org/statepolicy/3376.cfm (9 September 2004).
13. For example, if 10 percent of all employees earn the minimum wage, then 10 percent of employees who work 40 hours a week earn the minimum wage, 10 percent of employees who work 35–39 hours earn the minimum wage, and so on.
14. Age categories are under 30, 30–39, 40–49, and 50 and older. Because of the structure of the data set, all employees who work at least 80 hours per month are included in the analysis. Those who work 80–100 hours per week would not be eligible under SB 2.
15. For example, if an employer reported a 52 percent female workforce, then 52 percent of workers under age 30 were female, 52 percent of workers ages 30–39 were female, and so on.
16. Cost estimates were provided by Roland McDevitt and Watson Wyatt Worldwide.
17. American Medical Association and California Medical Association, Benchmark Capitation Rates: The Physician’s How-to Guide for Calculating Fee-for-Service Equivalents (Chicago: AMA, 2001).
18. L.H. Summers, “Some Simple Economics of Mandated Benefits,” American Economic Review 79, no. 2 (1989): 177–183.
19. J. Gruber, “Health Insurance and the Labor Market,” chap. 12 in Handbook of Health Economics, vol. 1, ed. A.J. Culyer and J.P. Newhouse (Amsterdam: Elsevier Science B.V., 2000), 645–706.
20. Kaiser Family Foundation and HRET, “Chartbook: The Health Insurance Act of 2003 (SB2).”
21. Based on household income for workers at firms with at least fifty employees in 2001 CHIS data. Calculated using the midpoints of the range of each wage category; for the highest wage category the value used was $150,000.
22. This tax rate includes the federal income tax rate of 15 percent, California state income tax rate of 6 percent, and employee-paid payroll tax rate of 7.65 percent. This tax rate also includes the employer-paid portion of the payroll tax (7.65 percent); just as the incidence of health insurance premiums will be borne by employees, employers also shift the burden of their payroll tax contribution to their employees.
23. Because health insurance premiums and cash wages are both deductible from corporate income, there is no adverse effect on employers’ income from substituting one for the other. However, the state and the federal government lose income tax revenue on the portion of employee compensation that was formerly cash wages and is now health insurance premiums; this paper does not formally account for the loss in income tax revenues.
24. Gruber, “Health Insurance and the Labor Market.”
25. Institute of Medicine, Insuring America’s Health (Washington: National Academies Press, 2004).
26. A. Dube and M. Reich, “2003 California Establishment Survey: Preliminary Results on Employer Based Healthcare Reform,” Research Brief, 18 September 2003, www.iir.berkeley.edu/research/ces.pdf (9 September 2004).
27. The decision to work at firms not offering health insurance may demonstrate that some workers contemplated by SB 2 value health benefits less than their cost. If so, the long-term wage reduction at these firms will be lower and the employment effects greater.
28. D.M. Cutler and B.C. Madrian, “Labor Market Responses to Rising Health Insurance Costs: Evidence on Hours Worked,” RAND Journal of Economics 29, no. 3 (1998): 509–530.
29. If firms with 20–49 employees are excluded, 32–36 percent of workers eligible for benefits under SB 2 at these firms earned within $2 of the minimum wage; however, the sample size for this analysis is small (n = 19).
30. A similar effect was observed in Hawaii. See Thurston, “Labor Market Effects.”
31. P. Fronstin, Health Insurance Coverage and the Job Market in California, Special Report 36, September 2000, www.ebri.org/sr36/index.htm (9 September 2004).
32. Dick, “The Impact of Mandated Employer Provision of Health Insurance Benefits.”
33. Cost estimates were provided by Roland McDevitt and Watson Wyatt Worldwide.
34. D.M. Cutler, “Market Failure in Small Group Health Insurance,” NBER Working Paper no. 4879 (Cambridge, Mass.: National Bureau of Economic Research, 1994).
35. SB 2 includes a provision that extends market rules for the small-group market to firms of 51–199 workers, restricting the range of prices an insurer can offer based on the risk profile of an employer group to ± 15 percent.
36. D.M. Cutler, “Employee Costs and the Decline in Health Insurance Coverage,” NBER Working Paper no. 9036 (Cambridge, Mass.: NBER, 2002).
37. J.S. Lee and L. Tollen, “How Low Can You Go? The Impact of Reduced Benefits and Increased Cost Sharing,” Health Affairs, 19 June 2002, content.healthaffairs.org/cgi/content/abstract/hlthaff.w2.229 (28 July 2004); and J. Gabel and T. Rice, “Understanding Consumer-Directed Health Care in California,” August 2003, www.chcf.org/documents/insurance/ConsumerDirectedHealthCare.pdf (9 September 2004).

Anna Sinaiko (sinaiko{at}fas.harvard.edu) is a doctoral candidate in health policy at Harvard University in Cambridge, Massachusetts.

DOI: 10.1377/hlthaff.var.469
©2004 Project HOPE–The People-to-People Health Foundation, Inc.