|
Can Private Companies Contribute To Public Programs’ Outreach Efforts? Evidence From California
We studied an innovative outreach effort in California, which trains and certifies community organizations to help complete Medicaid and State Children’s Health Insurance Program (SCHIP) applications. In this paper we provide a detailed description of participating organizations, the populations they serve, and their success at turning submitted applications into enrollments. We found that insurance brokers and income tax preparers—for-profit groups that are not typically associated with outreach—make important contributions to Medicaid and SCHIP in California. Brokers, in particular, help serve a hard-to-reach population: those on the higher end of the income eligibility thresholds.
THE SHARE OF AMERICANS WITH employer-sponsored health insurance has declined steadily since the late 1970s. Public coverage expansions have largely spared children from the ill effects of this decline. However, the latest figures indicate that the proportion of children without health insurance has started to rise, from 10.8 percent in 2004 to 11.2 percent in 2005.1 If experience is a guide, many of the 8.3 million uninsured children in 2005 were eligible for public insurance. In 2002, for example, about 2.8 million uninsured children age eighteen and under were eligible for the State Children’s Health Insurance Program (SCHIP), and an additional 3.7 million were eligible for Medicaid.2 Together these two groups represented more than 60 percent of uninsured children.3 Devising strategies to increase take-up is essential for reducing the number of uninsured children. Eligible uninsured children are difficult to reach because they are disproportionately children of the "working poor." Their parents often have little experience with means-tested benefit programs, and many do not realize that their children are eligible for public insurance.4 With this in mind, SCHIP, more than past expansions, emphasized and made a financial commitment to outreach. In almost all states, outreach efforts aimed to increase public awareness through radio, television, and newspaper advertisements; brochures and flyers; and toll-free hotlines. Many states also used in-person eligibility workers or volunteers to offer in-depth program information and application assistance.5 Little is known about the impact of these strategies.6 The bulk of the literature on outreach has taken a demand-side perspective, assessing the characteristics of the "eligible uninsured" and recommending efforts that target specific populations.7 Although these studies have been crucial in the design and implementation of current outreach strategies, we remain poorly informed about which efforts were most effective. The best information on the supply side comes from federally mandated SCHIP evaluations.8 This evidence suggests that personalized efforts, such as hotlines and home visits, were more effective strategies than television or print advertisements and that health centers and schools were more effective settings than libraries or senior centers.9 To date, however, independent assessments of the effectiveness of various outreach efforts remain rare. To help fill this gap, we studied an outreach strategy adopted by California as part of its implementation of SCHIP: application assistance. Beginning in 1998, the state worked with a variety of organizations to provide application assistance to families who were potentially eligible for Healthy Families (California SCHIP) or Medi-Cal (California Medicaid). These enrollment entities (EEs) include organizations such as hospitals and clinics that make up the health care safety net as well as schools, faith-based charities, and commercial entities such as insurance brokers and agents and tax preparers. Initially, the EEs were paid $50 for each successful application and $25 for each successful renewal.10
Application assistance addresses several major barriers to enrollment: complexity of the enrollment process, confusion about eligibility, and language difficulties.11 A recent study by Anna Aizer found that proximity to an additional bilingual application assistant increased new monthly Medi-Cal enrollments among Hispanic children by 16–46 percent and, among Asians, by 26–45 percent.12 Also, as shown in Exhibit 1
Despite this apparent success and the fact that application assistance consumed less than 1 percent of the Healthy Families budget, the state suspended the program in 2003 in response to its fiscal crisis. This resulted in a steep decline in the percentage of applications filed with assistance (Exhibit 1  ). Moreover, as suggested by program administrators, this drop in assistance increased the number of applications that were incomplete or completed with error, delaying new enrollments and leading to denials of applications that might have been accepted if they had been properly completed (Exhibit 2 ).
Payments to EEs were reinstated in July 2005 and have since increased to $60 for successful applications and $50 for successful (annual) renewals. The increase since July 2005 in the share of forwarded and eligible applications that were completed with assistance (Exhibit 1 ) indicates the importance of the fees in maintaining entity effort. But the long-term future of this program remains in doubt, as funding is subject to an annual budget and allocation process. Even if funding were continued, questions remain as to how the program should be structured. Should the state continue to partner with so many different types of organizations? Or, as has been suggested, should outreach efforts be targeted to certain entities, such as schools? More generally, as SCHIP comes up for reauthorization in 2007, California’s experience with leveraging outside resources to increase and maintain enrollment could provide important lessons for other states.
We based our analysis on detailed information about all EEs operating in California from the start of Healthy Families in 1998 through June 2002. In addition to location, we know when the EE entered the program, the type of organization it is, how many certified application assistants (CAAs) were associated with it, and, most importantly, how many applications were submitted and how many were deemed eligible for SCHIP or Medi-Cal children. Since EEs differ in how long they were in the program as of June 2002, in some analyses we divided the total number of submissions by that length of time and present the data in terms of the number of applications submitted per 100 days. In fiscal years 2000–01 and 2001–02, forty-two community-based EEs and thirty-one school-based EEs received contracts instead of fee-based payments. The latter were a diverse group, including districts in large urban areas, such as Los Angeles Unified, and in sparsely populated rural counties. Although contract-based EEs are not in our data, they were relatively minor players in the application assistance program, accounting for only 10–11 percent of assisted applications.13
Application assistance by entity type. The number of participating EEs and CAAs varied considerably across types (Exhibit 3 ). For example, insurance brokers represented almost 40 percent of EEs, but because these were small organizations—often just one person—they accounted for a much smaller share of assistants than other types of EEs. In contrast, hospitals, which had many assistants (6.5 on average), accounted for less than 3 percent of EEs but almost 8 percent of all CAAs.
Government-funded organizations—typically, outreach programs run by a county or municipal department of health—had an average of seventeen CAAs per entity, far more than any other organization type. This might reflect government-funded organizations’ efforts to target the neediest cases. Many of their assistants might focus on adults or other groups for whom the state does not pay an application assistance fee.14 Also, many employees at government-funded organizations might be certified to provide application assistance (and thus counted), even if most of their time is spent on other responsibilities.
Application submissions by entity type.
Clinics submitted about a third of all assisted applications (Exhibit 3 According to state administrators, these school-based contractors played an important role in the early days of Healthy Families. The California Department of Health Services attributes 35 percent of all application requests made to its toll-free hotline during the first six months of Healthy Families to school-based efforts.15 However, a rough estimate suggests that for the full period of our analysis, accounting for organizations paid on a contract basis would increase the share of accepted applications submitted by schools by no more than four percentage points.16 Even with this adjustment, schools rank behind most other types of EEs in terms of the number of successful applications.
Successful enrollments.
Overall, about 42 percent of applications submitted led to successful enrollments.17 Although insurance brokers contributed only about 13 percent of submissions, they had the highest acceptance rate (about 44 percent). A more detailed investigation revealed that the high acceptance rate of brokers was driven by a small number of high-volume entities. This relationship between volume and accuracy, which held for other entity types as well, suggests that as EEs submit more applications, they might gain a better understanding of the eligibility rules and requirements concerning documentation. Schools had the lowest acceptance rate—six percentage points below the overall average and nearly eight percentage points below the rate for brokers. Hospitals also had a relatively low acceptance rate. Government-funded organizations, community-based organizations, and tax preparers all performed slightly below average (Exhibit 3 Acceptance rates and incentives. The variation in acceptance rates across groups is broadly consistent with differences in incentives and the nature of the interaction with clients. Applications from hospitals are triggered when an uninsured patient presents for outpatient treatment or admission. Many of these patients arrive at the hospital without the necessary documentation or information for the hospital to make an accurate eligibility assessment. At the same time, submitting an application represents a minimal cost to the hospital, as the application only marginally increases the necessary paperwork. And the benefit to the hospital—the difference between what Medi-Cal or Healthy Families will reimburse for the care and what they can recover from an uninsured patient—is potentially quite large. Thus, hospitals have an incentive to submit applications even when the probability of acceptance might be low. In contrast, because the only benefit that insurance brokers receive from submitting an application is the $50 fee and because they can easily request that a client return with more documentation, they might be reluctant to expend effort on cases where acceptance appears unlikely. Providers and clinics (other than hospitals) are the only other groups that came close to brokers in terms of acceptance rates. Like hospitals, these entities stood to gain reimbursement for services rendered on top of the application fee. Relative to hospitals, however, providers and clinics might have an added incentive to recruit and submit high-quality applications: Obtaining insurance coverage for their clients could increase the likelihood of repeat business. Moreover, like insurance brokers and unlike emergency departments, providers and clinics might be able to request the documentation needed to determine eligibility before rendering services. The reason for schools’ poor performance is less clear. Like brokers, schools’ main incentive is the $50 fee, and school-based EEs can request appropriate documentation before submitting applications. However, enrolling students in health insurance is likely one of a school-based assistant’s many responsibilities. According to state administrators, schools might simply play a different role in the assistance program than other types of entities: Schools help families obtain application forms and direct them to the entities providing assistance.18 Since data were only collected for submissions, this service could not be captured in our analysis. Nonetheless, it might be an important part of the program’s success. Application assistance by entity type and area income. The heterogeneity of the population eligible for public health insurance provides one rationale for partnering with different types of organizations. Previous work suggests the importance of taking account of language differences when designing outreach strategies.19 Income is another important source of heterogeneity. Healthy Families’ income eligibility limit for a family of four is nearly $50,000. Organizations that are effective at reaching very-low-income families might not be as adept at enrolling families at the upper end of the SCHIP income eligibility range, and vice versa. And the geographic distribution of organizations makes them more accessible to different types of communities. For example, during our study period, although the average clinic, provider, school, or hospital-based EE was located in a ZIP code with median family income of below $40,000, the typical tax preparer, government-funded EE, and community-based EE operated in a ZIP code with median family income of just over $40,000. Insurance brokers were located in ZIP codes with an average median family income of almost $50,000. In addition, insurance brokers, in contrast to other EE types, operated in the widest range of ZIP code income, with median family income under $15,000 at the lowest end and just over $119,000 at the highest end.
To shed light on the role of different EEs across the income distribution, we stratified the data by median family income and examined the distribution of submissions and acceptances by income quartiles (Exhibit 4
The first thing to note is that the distribution of submissions and acceptances by entity type varied greatly by area income. The results suggest that clinics and community-based organizations were critical sources of assistance for applicants from very low-income areas but provided fewer services to families living in areas with residents at the upper end of the eligibility range. The opposite pattern held for brokers. Although they submitted only 6 percent of applications in the lowest income quartile, in the highest quartile they accounted for a greater share of applications submitted and deemed eligible than any other type of entity. We see a similar pattern for tax preparers, which accounted for a tiny share of applications in the lowest-income ZIP codes but submitted almost a tenth of all applications in the highest-income areas. Several other differences across income groups and entity types deserve mention. For example, across all but the second income quartile, hospitals had a lower-than-average success rate. It was especially low in the top half of the distribution: In the highest quartile, only a quarter of applications submitted by hospitals were deemed eligible. This pattern supports the idea that hospitals were less able to screen for eligibility and more likely to submit applications irrespective of the probability of acceptance. According to this argument, hospitals located in affluent areas had lower success rates than those in poor areas because they interacted with a population that had a lower probability of eligibility. Insurance brokers performed comparatively well across ZIP codes in all income quartiles. However, those located in ZIP codes in the highest income quartiles performed particularly well. In the two highest income quartiles, where the overall acceptance rates were each just about 38 percent, insurance brokers still turned more than 40 percent of their submissions into successful applications. Thus, in areas where the probability of encountering an eligible client might be lowest, insurance brokers have contributed the most to Medicaid and SCHIP enrollment.20 Schools accounted for 7–8 percent of applications in the middle two quartiles, which is higher than their share in either the poorest or most affluent areas. In each quartile, the acceptance rate for schools was between five and nine percentage points below the mean.
Within-neighborhood comparisons of application assistance.
EEs are not randomly distributed across the state. Exhibit 4 Our regression models also controlled for the number of CAAs working for each entity and how long the entity had been involved in the program. As expected, EEs with more CAAs on staff submitted more applications, although, interestingly, they had a lower success rate. The success rate increased with how long an EE had been involved in the program, which could reflect the importance of learning by doing.
Prior research. Prior research shows that insurance brokers and agents play a critical, if underappreciated, role in the small-group and nongroup health insurance markets.22 Small employers and individuals purchasing nongroup coverage rely on them for information on insurance options and assistance with enrollment. And insurers view brokers as a key component of their distribution channel. In the 1990s, some states that enacted small-group reforms attempted to lower premiums by "cutting out the middleman" and reducing consumers’ reliance on brokers. These efforts were generally unsuccessful. Even where reforms made it easier for small employers to purchase insurance directly from insurers or through a cooperative, a large fraction continued to go through brokers. Our findings. Our results suggest that insurance brokers can also play an important role in helping eligible families enroll in Medicaid and SCHIP. Relative to other types of EEs, brokers have been successful at increasing public health insurance take-up in California. They are located near and thus may have better access to those on the higher end of the income eligibility thresholds for public health insurance. This is important because many of these families have little or no prior experience with the social service system and might not be in contact with the same types of organizations as lower-income families are. Even within neighborhoods, insurance brokers are better than other EEs at turning applications into enrollments. There are many possible reasons for their success. State officials indicate that brokers are experienced at filling out complex forms. Because of their profession, most brokers should already have a system for identifying people who are interested in obtaining health insurance. The financial reward for application assistance is such that brokers have little incentive to submit applications that are unlikely to be eligible. They also might use the programs as alternative options for commercial-product-line customers who have family members who do not qualify for or cannot afford those product lines. Similarly, our results suggest that tax preparers might be an underused resource in outreach efforts. Accounting for the neighborhoods where they operate, tax preparers have higher success rates than clinics and most other organizations operating in the same ZIP codes. Their success may reflect their access to the most important application requirement—income documentation—allowing them to more readily and accurately assess eligibility than other EEs can. As suggested by their low submission rates, however, tax preparers have not actively participated in the program, and when they do, their efforts are seasonal (January through April). Recognizing their potential to be an efficient source of applications and enrollments, state administrators are exploring ways, including partnering with a major tax preparation company, to increase participation rates. Lessons for other states. California’s application assistance program offers lessons for many states and public benefit programs. Only three other states (Illinois, North Carolina, and Virginia) provide incentives to insurance brokers who assist SCHIP enrollments.23 To our knowledge, no other state recruits tax preparers to provide application assistance. Yet our results indicate that means-tested benefit programs may be well served by recruiting a diverse set of entities, including those, like brokers and tax preparers, that are not typically associated with outreach. Insurance brokers and tax preparers may be even more critical in implementing policies aimed at reducing the number of "middle-class" Americans who lack insurance. Need for more-detailed guidance. Although our study provides important new information on organizations involved with Medicaid and SCHIP outreach, more-detailed research would help guide policymakers in this area. How do successful brokers "market" public health insurance to eligible families? What role do employers play? Are there differences across EEs in retention? Finally, enrollment is only an intermediate goal; ultimately, we want to know whether children who are enrolled through alternative channels differ in their access to and use of health care.
Mireille Jacobson (mireille{at}uci.edu) is an assistant professor in the School of Social Ecology, University of California, Irvine. Tom Buchmueller is a professor in the Ross School of Business, University of Michigan, in Ann Arbor. Earlier versions of this study were presented at the 2006 Robert Wood Johnson Foundation (RWJF) Scholars in Health Policy Research Program meeting and the 2006 meeting of the Economic Research Initiative on the Uninsured. Mireille Jacobson received support from the RWJF Scholars in Health Policy Research Program. Both authors received support from the University of Michigan’s National Poverty Center for a research grant. The authors thank Anna Aizer for graciously sharing data and responding to questions, Ernesto Sanchez from the California Managed Risk Medical Insurance Board for explaining many features of California’s outreach effort, and Janette Lopez for providing invaluable feedback. They also thank Alan Cohen, Tom Chang, Sherry Glied, Catherine McLaughlin, Heather Royer, Richard Scheffler, Doug Staiger, and Kathy Swartz for many helpful comments. All mistakes are the authors’ own.
This article has been cited by other articles:
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
Top
Data Source: Enrollment Entities
Study Findings
