MMRR 2012: Volume 2 (2)

Medicare & Medicaid Research Review

2012: Volume 2, Number 2

A publication of the Centers for Medicare & Medicaid Services,

Center for Strategic Planning

Wisconsin’s Experience with Medicaid Auto-

Enrollment: Lessons for Other States
Thomas DeLeire1, Lindsey Leininger2, Laura Dague1, Shannon Mok3, Donna Friedsam1
University of Wisconsin-Madison
1

2
University of Illinois-Chicago
3
Congressional Budget Office

Abstract
The Patient Protection and Affordable Care Act (ACA) relies heavily on the expansion of Medicaid
eligibility to cover uninsured populations. In February 2008, Wisconsin expanded and reformed its
Medicaid/CHIP program and, as part of program implementation, automatically enrolled a set of newly
eligible parents and children. This process of “auto-enrollment” targeted newly eligible parents and older
children whose children/siblings were already enrolled in the state’s Medicaid/CHIP program. Auto-
enrollment brought over 44,000 individuals into the program, representing more than 60% of all enrollees
in the first month of the reformed program. Individuals who were auto-enrolled were modestly more
likely to leave the program relative to other individuals who enrolled in February 2008, unless their
incomes were high enough to be required to pay premiums; these auto-enrollees were much more likely
to exit relative to other enrollees subject to premium payments. The higher exit rates exhibited by non-
premium paying auto-enrollees were likely due to the fact that over 40% of auto-enrollees were covered by
a private insurance policy in the month of their enrollment, compared to approximately 30% for regular
enrollees. A national simulation of an auto-enrollment process similar to Wisconsin’s, including the
expansion of adult Medicaid eligibility to 133% of the federal poverty level under the ACA, suggests that
2.5 million of the 5.6 million newly eligible parents could be auto-enrolled, and approximately 25% of this
population would be privately insured. These results suggest that auto-enrollment may be appropriate for
other states, especially in their efforts to enroll eligible populations who are not subject to premium
requirements.
Keywords: Medicaid; enrollment; parents; auto-enrollment
doi: http://dx.doi.org/10.5600/mmrr.002.02.a02

DeLeire, T., Leininger, L., Dague, L., Mok, S., Friedsam, D. E1

MMRR 2012: Volume 2 (2)

Introduction
Enrolling eligible children and adults into Medicaid and the Children’s Health Insurance
Program (CHIP) has been a challenge, and many states have pursued aggressive strategies to
reach and enroll those eligible for these programs (Wachino & Weiss, 2009). Some estimates
suggest that Medicaid and CHIP reach only 79 percent of eligible children who lack access to
private health insurance, and that as many as three-quarters of uninsured children are eligible,
but not enrolled in Medicaid or CHIP (Dubay, Holahan, & Cook, 2007; Hudson & Selden, 2007;
State Health Access Data Assistance Center & Urban Institute, 2005). Eligible parents can also
be difficult to enroll into the program, despite most states setting parent eligibility for public
coverage well below that of children. Estimates suggest that roughly 28% of uninsured parents
are eligible for Medicaid/CHIP, but are not enrolled (Georgetown University Health Policy
Institute, 2009; Holahan, Cook, & Dubay, 2007).
Wisconsin, in contrast to the experiences of many states, has been very successful in
recent years in enrolling eligible low-income children and parents into BadgerCare Plus (BC+),
its combined Medicaid and CHIP program. Wisconsin expanded eligibility for BC+ to virtually
all children and to low-income parents/caretakers in February 2008. Enrollment increased
rapidly, with net enrollment increasing by 51,000 in the first month alone and by 124,000
between February 2008 and November 2009—representing a 25% increase over the enrollment
numbers for low-income children and parents as of December 2007. In addition to the severe
economic downturn, a number of policy factors have been credited for this dramatic expansion.
Examples include: clear branding, simplified application processes, reductions or eliminations of
deductibles for some low-income children, easing of “anti-crowd-out” provisions, availability of
a user-friendly online application, and extensive community outreach (Leininger et al., 2011).
Beyond these factors, however, an important reason why Wisconsin was so successful in its
enrollment efforts was that it auto-enrolled over 44,000 individuals at the time of program
launch, including over 26,000 parents and 18,000 children. The auto-enrollment involved
electronic database matching, which applied new program eligibility criteria to existing data
already held within state databases, and immediately converted eligible persons to coverage.
The Patient Protection and Affordable Care Act (ACA) is projected to extend health
insurance coverage to an additional 32 million people and relies heavily on an expansion of
Medicaid to do this (Patient Protection and Affordable Care Act, 2010; Health Care and
Education Reconciliation Act, 2010). It creates a new national income eligibility standard at
133% of the federal poverty level (FPL). Medicaid expansion is likely to account for about half—
16 million people—of those who, by 2019, will become newly eligible for health insurance under
the federal health care reform (Iglehart, 2010). Given the considerable discretion with which
states operate their respective Medicaid programs, implementation of the federal health care
reform efforts will rely heavily upon the engagement of states in reaching and enrolling newly
eligible individuals (Holahan & Headen, 2010). We believe that Wisconsin’s experience with

DeLeire, T., Leininger, L., Dague, L., Mok, S., Friedsam, D. E2

MMRR 2012: Volume 2 (2)

auto-enrollment may serve as an instructive exemplar to other states considering similar

processes of enrolling those newly eligible for Medicaid under the ACA.
Auto-enrollment has a number of potential benefits. First, it is a way of quickly
extending coverage to a large number of eligible individuals and decreasing the number of
uninsured individuals. Second, it may help find potentially hard-to-enroll individuals. However,
as implemented in Wisconsin, the auto-enrolled or their family members had some experience
being enrolled in the public insurance program in order to be in the databases, so this paper
cannot speak to the promise of auto-enrollment methods in reaching the traditionally hard-to-
enroll. Finally, it may capture individuals who would eventually enroll, and enroll them sooner
than they would themselves, increasing their time covered and reducing their costs of
enrollment. On the other hand, auto-enrollment has some disadvantages. It may enroll some
individuals who do not want to be enrolled in the program and who may disenroll quickly. This
may occur when they have access to coverage through other mechanisms (such as their own or
their spouses’ employment, COBRA, or individual purchase), or when their income places them
in an eligibility category that requires premium cost sharing.
In this paper, we describe Wisconsin’s experience with automatic enrollment. We use
administrative data to determine whether public insurance was valuable to auto-enrollees
relative to regular enrollees by comparing rates of continuous enrollment, churn rates, and
percentages having private insurance at the time of enrollment. Finally, we simulate the
potential impacts at the national and state levels, should an auto-enrollment process similar to
that used by Wisconsin be adopted along with the expansion of Medicaid to 133% FPL under the
ACA.

Auto-Enrollment: Wisconsin’s Experience

Wisconsin launched BadgerCare Plus (BC+)—a health insurance reform initiative that
expanded the state’s Medicaid and CHIP programs—on February 1, 2008. Immediately prior to
BC+ program launch, the Wisconsin Department of Health Services (DHS) carried out a one-
time auto-enrollment effort that automatically enrolled more than 44,000 previously unenrolled
individuals. Of these, 98 percent were siblings or parents of existing beneficiaries. This process
involved applying new program eligibility criteria to previously ineligible individuals for whom
there was current information in the state’s administrative database of program eligibility. This
included anyone who had at least one family member with an open case (already enrolled in
state health programs in December 2007 or January 2008), or who had had a case closed 30 days
before the BC+ implementation (i.e., in December 2007). Previously ineligible individuals could
become eligible because of any one of a number of policy changes, including the elimination of
income limits for children’s coverage, the elimination of crowd-out restrictions for families with
incomes below 150% FPL, the expansion of eligibility to parents with income levels of 185–
200%, and to caretaker relatives of children with income levels of 45–200%.

DeLeire, T., Leininger, L., Dague, L., Mok, S., Friedsam, D. E3

MMRR 2012: Volume 2 (2)

Automatic Enrollment in Other Settings

Auto–enrollment is a process whereby people who meet income and categorical eligibility
criteria are identified through electronic matching of various existing administrative data sets,
including data culled from Medicaid and other public programs (Dorn, 2007; Dorn, 2009; Dorn
& Kenney, 2006). Eligible people are automatically enrolled into coverage, with the converted
beneficiaries having both the right and responsibility to opt–out of coverage. This approach
contrasts with the traditional model for public benefits enrollment, which relies on the
knowledge, motivation, and initiative of the potential enrollee to seek out and submit
applications, to provide information demonstrating potential eligibility, and to fulfill the
procedural requirements of the administrative agency. The government agency’s responsibility
is limited to offering subsidies or services, educating the public about available assistance, and
processing applications.
Automatic enrollment lifts much of the application burden from potentially eligible
individuals. Studies have consistently shown that opt–out approaches have higher rates of
enrollment than opt-in policies (Remler & Glied, 2003). This finding has been extensively
documented at firms where new workers establish 401(k) accounts. Only 33% of new employees
enroll in a traditional 401(k) plan when they are given the choice to opt-in and are required to
complete application forms, while 90% of new employees enroll when automatically placed in
401(k) accounts without declining participation by completing “opt out” forms (Choi, Laibson,
Madrian, & Metrick, 2004; Choi, Laibson, Madrian, & Metrick, 2005a; Choi, Laibson, Madrian,
& Metrick, 2005b; Holden & VanDerhei, 2005).
Auto-enrollment is already used for dual-eligible Medicare beneficiaries, for whom states
submit their data files to the Centers for Medicare & Medicaid Services for matching. Medicare
Part D subsidies were automatically extended to all Medicare beneficiaries who received
Medicaid or Supplemental Security Income the prior year. The subsidy program reached 74
percent of eligible beneficiaries within six months of implementation. Medicare Part B covers
more than 95 percent of eligible seniors by automatically enrolling them into the program and
deducting premium payments from their Social Security checks, unless the seniors complete
opt-out forms (National Council on Aging, 2006; Federman, Vladeck, & Siu, 2005).
States themselves have considerable experience with data matching for a range of health
care coverage purposes: income and employment verification, coordination of benefits with
other insurers, Medicaid premium-assistance programs, and identification of benefits available
to children from non-custodial parents. Such a process could be extended to target persons who
have already sought public assistance for other social programs and who may be eligible for—
but not enrolled in—Medicaid/SCHIP. For example, Massachusetts successfully employed
database matching when it auto-converted former Uncompensated Care Pool enrollees into
CommCare coverage (Dorn et al., 2009).

DeLeire, T., Leininger, L., Dague, L., Mok, S., Friedsam, D. E4

MMRR 2012: Volume 2 (2)

It has been argued elsewhere that tax information could provide all of the information
needed to determine eligibility for Medicaid and CHIP (Dorn, 2009). Indeed, some states
already use tax information and other databases to verify eligibility of program applicants. More
than six out of seven uninsured individuals (86.3 percent) file federal income tax returns, and
even families that are exempt from paying federal income tax may file state income tax returns
when they can qualify for an earned income tax credit (Dorn, 2009).
While the promise of across-systems database matching is great, the logistics of
implementing such initiatives are often daunting. Negotiating and executing data-sharing
agreements across agencies can take considerable time and, depending on the specific legal
environment, may be impossible. Wisconsin’s experience, however, demonstrates an already
accessible, feasible way for Medicaid agencies to utilize their existing Medicaid enrollment data
to implement auto-enrollment—an approach that does not require participation from other
agencies.

Study Data and Methods

We use data from the Wisconsin CARES eligibility and enrollment systems covering the time
period from February 2008 through November 2009. Our analytic sample includes the 125,418
parents and children who enrolled during the first four months of the expanded program (from
February 2008 through May 2008). We separately consider three groups: those who were
automatically enrolled in February 2008, those who enrolled through regular processes in
February 2008, and those who enrolled between March and May 2008. Auto-enrollment only
occurred in February 2008.
For these three groups, we examine two measures of the continuity of enrollment: the
probability that they remain enrolled for at least 6 months and the probability that they remain
enrolled for at least 12 months. Individuals may disenroll from Medicaid because they do not
want the coverage or because they have qualifying events, most notably changes in income and
household composition that render them ineligible. Wisconsin requires the prompt reporting of
any such changes, within 10 days for household composition changes and by the 10th of the
month following an income change (Wisconsin Department of Health Services, 2008). They are
disenrolled by the state if they fail to pay premiums for two months.
We also examine churning, which we define as the probability of re-enrollment within
six months for those who exit by May 2009. We estimate these probabilities using probit models.
Our models take the form:
yi = f ( X i Β + αFebi + γAutoi + δPi + θ1 Autoi × Pi + θ 2 Febi × Pi )

where yi is one of the following:

an indicator variable for an enrollee remaining continuously enrolled for 6 months;

DeLeire, T., Leininger, L., Dague, L., Mok, S., Friedsam, D. E5

MMRR 2012: Volume 2 (2)

an indicator variable for an enrollee remaining continuously enrolled for 12 months;

or
an indicator variable for an enrollee that exits and re-enrolls within 6 months.
The independent variables in each of these models include:
 Auto, an indicator for the enrollee being auto-enrolled in February 2008;
 P, an indicator that the enrollee is required to pay premiums;
 Feb, an indicator for being a February enrollee; and
 X, a set of demographic variables including whether the enrollee is an adult or
child, includes categories for age of the youngest child in the household (age 0,
age 1–5, age 6–12, age 13–18, or no enrolled children), a female indicator,
number of adults in the household, number of children in the household,
categories for education level (less than high school, high school, or more than
high school), an indicator for whether the county of residence is urban or rural,
categories for income as a percentage of FPL (under 150% FPL, 150–200% FPL,
and more than 200% FPL), and county unemployment rates.
To examine whether enrollees had access to private insurance coverage at the time of
enrollment, we use administrative data from three sources that are merged with the CARES
enrollment data. These include the following:
• UI: Quarterly wage records from required unemployment insurance reporting;
• TPL: Wisconsin’s Third Party Liability database; and
• DOL: a U.S. Department of Labor database of all self-insured firms.
The unit of observation in this dataset is an individual-month beginning in the first month that
individual enrolled in BC+. We link across these different sources by using Social Security
numbers (SSNs) and federal employer identification numbers (FEINs).
TPL is an individual-level database that contains all enrollees in state health insurance
programs who are covered by a private, fully insured health insurance plan. We matched
enrollees to the TPL database using their SSNs. This database, while an excellent resource, is
limited in two ways. First, these data are available for each month in which an individual is
enrolled in BC+, but do not contain information on the health insurance coverage of individuals
in months prior to enrollment or following disenrollment. Second, the database does not
contain individuals who are covered by health insurance provided by a self-funded employer
(whose policies are not subject to state regulation). Thus, those enrollees who do not have
insurance according to the TPL database either do not have private insurance or have health
insurance through a self-funded employer.
To assess whether enrollees may have access to health insurance coverage through a self-
funded employer, we connect BC+ cases to their set of employers by linking CARES through
SSNs to a database of quarterly earnings records from Wisconsin’s unemployment insurance
(UI) system. We use this match to identify workers and to obtain the federal employer

DeLeire, T., Leininger, L., Dague, L., Mok, S., Friedsam, D. E6

MMRR 2012: Volume 2 (2)

identification numbers (FEINs) of their employers. For enrollees who do not have a TPL match,
we use their FEINs to link to data from the U.S. Department of Labor (DOL), in order to see if a
BC+ case member’s employer offers a self-funded plan. We obtained these data through a
Freedom of Information Act request. The data represent the universe of employers within the
United States from 2003 to 2007 that are self-insured for health, life, and disability and related
insurance plans. Because we are unable to observe which members of a case with a DOL match
obtain their health insurance from the self-insured firm, we impute insured status for these
enrollees using data from the Current Population Survey. Details of this data matching and
imputation process are available in Dague et al., (2011).
We then use the American Community Survey (ACS) to assess the potential reach and
targeting of auto-enrollment in other states. Specifically, we simulate the number of parents who
could potentially be auto-enrolled into Medicaid under an expansion to 133% FPL. Newly
eligible parents will represent a significant proportion of newly eligible adults (35%, author
calculations using the American Community Survey). We focus on this population because
childless adults will not have an enrolled child and so may not be in state databases, and also
because low-income parents have been a relatively higher priority adult population for some
state Medicaid/CHIP programs, as substantiated by their preferential treatment in recent state-
level expansions relative to low-income childless adults (Kaiser Commission on Medicaid and
the Uninsured, 2009). Not only is low-income parents’ public insurance eligibility of interest in
its own right, but it is also an important determinant of children’s public insurance take-up
(Dubay & Kenney, 2003).
We use data from the 2008 ACS—the first round of the ACS to include health insurance
variables—for the simulation (Davern, Quinn, Kenney, & Blewett, 2009). The ACS is a
nationally representative survey fielded yearly by the U.S. Census Bureau that collects
information on socio-demographic, economic, and housing characteristics of the non-
institutionalized U.S. population. All estimates are weighted to reflect the complex survey design
of the ACS. There are some limitations to the ACS, including a well-known Medicaid
undercount and the lack of complete information on legal residency (O’Hara, 2009). Our
analysis follows the “official” Census procedure of not adjusting our insurance estimates for
either the Medicaid undercount or for the possibility that legal immigrants and citizen children
of undocumented immigrants may be under-represented in the ACS, or that some respondents
in the ACS may be undocumented immigrants. First, we estimate the total number of parents
who would become newly income eligible for Medicaid under an eligibility expansion to 133%
FPL using information on state-level Medicaid eligibility levels collected by the Kaiser Family
Foundation (Kaiser Commission on Medicaid and the Uninsured, 2009). Some states, where
income eligibility for parents already exceeds 133% of FPL, would have no newly eligible
parents. We then exclude from this number those parents who appear to be newly income
eligible, but report already being on public insurance (Medicaid/CHIP, Medicare, or Veteran’s
Administration insurance). Parents who appear to be income ineligible can report receiving

DeLeire, T., Leininger, L., Dague, L., Mok, S., Friedsam, D. E7

MMRR 2012: Volume 2 (2)

Medicaid for a variety of reasons, including a differential between the relevant time period for
income reported for eligibility purposes and for the survey (monthly vs. annual), measurement
error in either income or reported health insurance, and/or the absence of information in the
survey needed to determine other means of eligibility (e.g., whether the parent is pregnant).
Second, we estimate the number of these newly income-eligible parents who conceivably
would already be present in a state’s Medicaid data system—and potentially a candidate for
auto-enrollment—because their children are already receiving public insurance. Third, we
calculate the “private insurance rate;” that is, the percentage of this population that report
having private health insurance in the survey.

Findings
The Impact of Wisconsin’s Auto-Enrollment
Wisconsin auto-enrolled 44,264 individuals into its BC+ program in February 2008. The auto-
enrollees comprised 63% of the 69,910 new enrollees who entered into the program in February
2008 (Exhibit 1). Of these, 26,062 were parents and 18,202 were children. The vast majority of
the auto-enrolled—almost 98%—had either a child or a sibling already enrolled in the program,
while the remainder had recently exited the program; a few fit both characterizations.

Exhibit 1. Summary Statistics on BC+ Enrollment

All Parents Children
New Enrollees February 2008 69,910 34,770 35,140
Number Auto-Enrolled 44,264 26,062 18,202
Of which:
Premium Paying 9,625 6,389 3,236
Exited in previous 6 months 2,783 1,442 1,341
Sibling/child enrolled in Nov/Dec 07 43,239 25,390 17,849
NOTES. Subcategories are not mutually exclusive; bottom three cells should not add to total number of auto-enrollees.
SOURCE: Authors' calculations from Wisconsin administrative data.

Relative to other new February 2008 enrollees, the auto-enrollees were slightly older and
belonged to larger households (results available from the authors). Almost three-fourths of both
auto-enrollees and other new enrollees had incomes that were less than 150% of the federal
poverty level, and similar proportions in both groups had a required premium based on income
(24% of regular enrollees and 22% of auto-enrollees, results available from the authors).

DeLeire, T., Leininger, L., Dague, L., Mok, S., Friedsam, D. E8

 MMRR 2012: Volume 2 (2)

Exhibit 2. Probability of Continuous Enrollment by Enrollment and Premium Status

All Enrollees Parents Kids
Mar.– Mar– Mar–
May May May
Feb. 2008 2008 Feb. 2008 2008 Feb. 2008 2008
Non- Non- Non- Non- Non- Non-
Auto Auto Auto Auto Auto Auto Auto Auto Auto
At least 6 months
No Premium 81.1% 82.9% 88.1% 78.2% 80.2% 86.0% 83.3% 85.0% 89.7%
Premium 44.1% 55.6% 77.9% 40.2% 51.6% 74.7% 47.5% 58.9% 80.3%
At least 12 months
No Premium 60.4% 66.8% 72.2% 54.5% 61.2% 67.0% 65.1% 71.3% 76.3%
Premium 31.3% 47.0% 62.2% 26.1% 41.0% 56.3% 35.8% 52.0% 66.8%
NOTES. Probabilities are predicted from probit models with covariates other than premium status, time of enrollment, and auto-enrollment status
held at their means. Means for those covariates as well as the coefficients and marginal effects from the probit models are reported in the Appendix.
Sample sizes by cell are reported in Exhibit 3. All pair wise differences between February auto-enrollees and non auto-enrollees are statistically
significant at the p < 0.05 level.
SOURCE: Authors' calculations from Wisconsin administrative data.

Exhibit 2 reports the predicted probabilities of continuous enrollment based on our estimated
probit models described in the previous section. The full results of these models are reported in
the Appendix. Individuals who were auto-enrolled and were not required to pay premiums were
only slightly more likely to exit the program within 6 months than were regular, non-premium
paying February 2008 enrollees—19% vs. 17%, respectively (p < 0.001). There was a modest
difference in the probability of 12 months of continuous enrollment between the auto-enrollees
and regular enrollees who did not pay premiums—60% vs. 67%, respectively (p < 0.001). Similar
differences are evident for parents and for children.
By contrast, individuals who were auto-enrolled and required to pay monthly premiums
exited the program at substantially higher rates than non-auto-enrolled entrants whose income
was high enough to require premiums (both of whom exit at far higher rates than enrollees who
do not pay premiums). Forty-four percent of premium-paying auto-enrollees remained enrolled
at 6 months compared with 56% of premium-paying regular enrollees and the difference is even
greater at 12 months (31% versus 47%, respectively). Premium-paying enrollees, who have
relatively higher incomes, may be unwilling or unable to pay the premium cost sharing, or may
be more likely to have other options for coverage.
We also calculated predicted probabilities for churning (the probability of re-entry
within six months of exiting the program). As with the exit probabilities, the probability of re-
entry conditional upon exit is similar for non-premium paying auto-enrollees and February
2008 regular enrollees (47% vs. 46%, p < .217, results available from the authors). Among the
premium paying enrollees, the re-entry probability is slightly higher among those who were
auto-enrolled (53% vs. 50%, p < 0.004, results available from the authors).

DeLeire, T., Leininger, L., Dague, L., Mok, S., Friedsam, D. E9

MMRR 2012: Volume 2 (2)

Individuals who enrolled in the expanded program in February 2008, whether

automatically enrolled or enrolled through the regular application process, had higher exit rates
than individuals who enrolled at any time in the subsequent three months.

Exhibit 3. Percent Privately Insured in Wisconsin for Auto vs. Non-

Auto Enrollees
Feb. 2008 Mar–May 2008
Auto Non-Auto Non-Auto
All Enrollees
No Premium 43.5% 29.8% 22.1%
34,639 19,518 55,587
Premium 48.9% 37.2% 31.7%
9,625 6,128 7,393
Parents
No Premium 40.3% 29.1% 21.5%
19,673 7,038 23,636

Premium 48.1% 38.2% 28.4%

6,389 1,670 4,350
Children
No Premium 47.7% 30.2% 22.5%
14,966 12,480 33,951
Premium 50.5% *36.9% *36.4%
3,236 4,458 3,043
NOTES. Calculated as the number of individuals who had third party medical insurance liability in the first month of enrollment in
BadgerCare Plus. Details in text. Total number of enrollees per cell (denominators) reported below percentages. Premium status is determined
by income; adults over 150% FPL are required to pay premiums and children over 200% FPL. All within-category (all, parents, children)
differences between columns and rows are statistically significant at the p < 0.05 level or better with the exception of the two starred cells.
SOURCE: Authors' calculations from Wisconsin administrative data.

In Exhibit 3, we provide estimates of the proportion of new enrollees that were covered by
private insurance policies at the time of their enrollment. Individuals who were auto-enrolled
were much more likely to be covered by these private policies—44% of non-premium payers and
49% of premium payers—than those who enrolled through the regular application process (30%
of non-premium payers and 37% of premium payers, respectively). The rates of private
insurance coverage were lower among those who enrolled in March through May 2008 than for
those who enrolled in February 2008.
These findings, overall, suggest that an appreciable proportion of those enrolled into
BC+ via the auto-enrollment process either did not want to be on the program or had a
qualifying event precluding the continuation of BC+ coverage within a year of enrollment, and
that others already had private insurance coverage and therefore may not have needed public
coverage. Both phenomena are especially true among the higher income groups that were
required to pay premiums.

DeLeire, T., Leininger, L., Dague, L., Mok, S., Friedsam, D. E10
MMRR 2012: Volume 2 (2)

Simulation of Parental Auto-Enrollment under the ACA’s Medicaid Expansion

Our estimates suggest that 5.6 million parents would become income eligible for Medicaid
should the eligibility threshold be increased to 133% FPL (Exhibit 4). Many eligible parents—2.5
million or 44% of them—have children who are already enrolled in Medicaid/CHIP. These
parents comprise the group that could most easily be auto-enrolled into public coverage. Of
these almost 1.9 million report being uninsured and the remainder—0.6 million—report being
privately insured. Thus, nationwide, the private insurance rate among those with the potential to
be auto-enrolled is 25%. This evidence suggests that, on average, populations that could
potentially be auto-enrolled are likely to have considerably lower rates of private coverage
relative to Wisconsin auto-enrollees. Of course, these estimates are more dependent upon the
characteristics of the expansion population considered in the policy simulation rather than the
fact that they are targeted for auto-enrollment. A policy to auto-enroll a much lower income
population likely would yield even lower estimates of the private insurance rate.

Exhibit 4. State-Level Estimates of Potential Number of Auto-Enrolled Parents

With
Children
Newly on Percent
Eligible Medicaid/ Privately Privately
State Parents Parents CHIP Uninsured Insured Insured
U.S. 74,700,000 5,580,969 2,476,798 1,865,178 611,620 25%

Alabama 1,067,192 191,525 88,881 61,828 27,053 30%

Alaska 177,717 10,451 * * * *
Arizona 1,553,725 — — — — —
Arkansas 650,261 142,173 89,056 64,786 24,270 27%
California 9,888,105 521,013 202,763 149,100 53,663 26%
Colorado 1,210,316 118,429 40,355 31,163 9,192 23%
Connecticut 870,134 — — — — —
Delaware 202,359 2,581 * * * *
District of Columbia 95,912 — — — — —
Florida 4,033,518 554,196 185,558 142,583 42,975 23%
Georgia 2,412,300 330,353 153,836 114,855 38,981 25%
Hawaii 319,332 — — — — —
Idaho 391,444 70,051 36,791 27,105 9,686 26%
Illinois 3,284,227 — — — — —
Indiana 1,549,137 — — — — —
Iowa 691,626 — — — — —
Kansas 670,648 92,832 40,203 26,230 13,973 35%
Kentucky 1,014,106 145,305 67,306 51,329 15,977 24%
Louisiana 1,031,309 190,001 99,509 76,730 22,779 23%
Maine 289,673 — — — — —
Maryland 1,390,053 23,815 * * * *

DeLeire, T., Leininger, L., Dague, L., Mok, S., Friedsam, D. E11
MMRR 2012: Volume 2 (2)

Exhibit 4 (cont.) With

Children
Newly on Percent
Eligible Medicaid/ Privately Privately
State Parents Parents CHIP Uninsured Insured Insured
Massachusetts 1,583,152 — — — — —
Michigan 2,431,344 265,665 98,054 66,968 31,086 32%
Minnesota 1,312,042 — — — — —
Mississippi 690,934 141,624 67,318 42,115 25,203 37%
Missouri 1,404,865 233,112 111,593 84,138 27,455 25%
Montana 212,975 21,120 * * * *
Nebraska 433,843 40,462 16,781 10,745 6,036 36%
Nevada 638,250 43,868 * * * *
New Hampshire 331,639 18,327 * * * *
New Jersey 2,289,829 — — — — —
New Mexico 466,218 73,728 32,030 22,283 9,747 30%
New York 4,761,878 — — — — —
North Carolina 2,194,468 314,574 149,596 109,994 39,602 26%
North Dakota 150,227 16,021 * * * *
Ohio 2,702,566 174,970 45,992 27,652 18,340 40%
Oklahoma 861,809 141,752 65,736 46,298 19,438 30%
Oregon 888,266 43,852 15,249 10,362 4,887 32%
Pennsylvania 2,878,281 — — — — —
Rhode Island 245,752 — — — — —
South Carolina 1,021,770 85,609 33,780 23,219 10,561 31%
South Dakota 189,363 21,596 * * * *
Tennessee 1,441,600 — — — — —
Texas 6,465,897 1,259,966 665,222 550,151 115,071 17%
Utah 755,150 — — — — —
Vermont 148,555 — — — — —
Virginia 1,917,554 207,342 84,053 61,295 22,758 27%
Washington 1,602,692 — — — — —
West Virginia 386,533 72,306 36,431 31,221 5,210 14%
Wisconsin 1,339,429 — — — — —
Wyoming 128,024 12,350 * * * *
NOTES. An em dash (—) indicates that a state has already expanded eligibility past 133% FPL. An asterisk (*) indicates that although a state
would have potential auto-enrollees, ACS sample sizes are too small for reliable calculations (weighted from <100 respondents). National
estimates are not equal to the sum of state estimates due to the exclusion of individual estimates for the (*) states.
SOURCE: Authors' calculations from 2008 American Community Survey.

Exhibit 4 also reports the state-by-state estimates of the number of parents who would become
newly eligible for Medicaid under the ACA, the number that could be potentially auto-enrolled,
and the private insurance rate. Estimates of the number of newly eligible parents vary widely by
state because of variation in existing eligibility rules. For example, Arizona currently covers
parents up to 200% FPL while Texas covers parents up to 27% FPL (Kaiser Commission on
Medicaid and the Uninsured, 2009). As a result, in Arizona no parents would become newly

DeLeire, T., Leininger, L., Dague, L., Mok, S., Friedsam, D. E12
MMRR 2012: Volume 2 (2)

eligible while all parents between 27% and 133% FPL in Texas (roughly 1.3 million individuals)
would become newly eligible. States also vary somewhat in the number and percentage of newly
eligible parents who potentially could be auto-enrolled, as this number depends on eligibility
levels of children relative to parents and in the private insurance rate among this population.

Discussion and Conclusions

Wisconsin’s auto-enrollment process was an effective route to reaching eligible individuals
already connected to public insurance either through past enrollment or through the enrollment
of a family member. Auto-enrollment reached a large number of previously ineligible siblings
and parents. This group had already demonstrated some proclivity toward participation in
public benefits, and thus may be less likely to opt-out than might persons captured through a
broader match with state income tax or unemployment insurance data, for example. Deploying
the relatively low-cost method of auto-enrollment would allow states to target more resource-
intensive outreach efforts on those groups that may not be well-represented in existing state
databases, who may be eligible for other forms of subsidized coverage, or who may be less
amenable to default coverage.
Our results show a mixed picture as to whether the use of automated database matching
is a reasonably efficient way to enroll uninsured individuals and individuals who would want to
participate in the public program. On the one hand, among those who were not subject to
premium payments, exit rates among those who were auto-enrolled were only modestly higher
than those for individuals who enrolled through the regular application procedure. On the other
hand, a much higher percentage of auto-enrollees were privately insured at the time of
enrollment than were regular enrollees. One important and unambiguous lesson learned from
the Wisconsin experience is that auto-enrollment appears to be an ineffective and poorly
targeted strategy for Medicaid-eligible populations who are subject to premium payments.
The national simulations suggest that up to 2.5 million parents nationwide could
immediately and automatically be enrolled in coverage upon an expansion of Medicaid
eligibility to 133% FPL and that, in a departure from the Wisconsin case, more than 75% of these
potential auto-enrollees are uninsured. These results suggest that the use of auto-enrollment in
other states could be an effective means of enrolling uninsured populations—and as the
Wisconsin case demonstrates, the greatest success will likely be achieved by using auto-
enrollment for eligible populations not subject to premium payments.

DeLeire, T., Leininger, L., Dague, L., Mok, S., Friedsam, D. E13
MMRR 2012: Volume 2 (2)

Disclaimer
The views expressed in this paper are those of the authors and should not be interpreted as those of the
Congressional Budget Office or the Robert Wood Johnson Foundation.

Correspondence
Lindsey Leininger, Ph.D., Assistant Professor, Division of Health Policy and Administration, University of Illinois-
Chicago, 1603 West Taylor Street, Chicago, IL 60612, 773-750-7706, lindseyjeanne@gmail.com

Financial Disclosure
Funding from the State Health Access Reform Evaluation initiative of the Robert Wood Johnson Foundation
(RWJF) and the RWJF Health and Society Scholars Program is gratefully acknowledged. All opinions expressed in
this paper are the authors’ alone and do not necessarily reflect those of RWJF.

DeLeire, T., Leininger, L., Dague, L., Mok, S., Friedsam, D. E14
MMRR 2012: Volume 2 (2)

References

National Council on Aging Access to Benefits Coalition. (2006, June). Pathways to Success:
Meeting the Challenge of Enrolling Medicare Beneficiaries with Limited Incomes. Retrieved
from http://www.ncoa.org/news-ncoa-publications/publications/pathways-to-success-
meeting.html
Choi, J., Laibson, D., Madrian, B., & Metrick, A. (2004). For Better or For Worse: Default Effects
and 401(K) Savings Behavior. In D. Wise (Ed.), Perspectives in the Economics of Aging (pp.
81–121). Chicago, IL: University of Chicago Press.
Choi, J., Laibson, D., Madrian, B., & Metrick, A. (2005a). Saving for Retirement on the Path of
Least Resistance. In E.J. McCaffery & J. Slemrod (Eds.) Behavioral Public Finance: Toward a
New Agenda (pp. 304–351). New York: Russell Sage Foundation Publications.
Choi, J., Laibson, D., Madrian, B., & Metrick, A. (2005b). Passive Decisions and Potent Defaults.
In D. Wise (Ed.), Analyses in the Economics of Aging (pp. 59–78). Chicago, IL: University of
Chicago Press.
Dague, L., DeLeire, T., Friedsam, D., Kuo, D., Leininger, L., Meier, S., & Voskuil, K. (2011, May).
Estimates of Crowd-Out from a Public Health Insurance Expansion Using Administrative
Data (NBER Working Paper No. 17009).
Davern, M., Quinn, B., Kenney, G., & Blewett, L. (2009, April). The American Community
Survey and Health Insurance Coverage Estimates: Possibilities and Challenges for Health
Policy Researchers. Health Services Research, 44(2 Pt. 1), 593–605. PubMed
http://dx.doi.org/10.1111/j.1475-6773.2008.00921.x
Dorn, S. (2007, August). Automatic Enrollment Strategies: Helping State Coverage Expansions
Achieve Their Goals. Robert Wood Johnson Foundation. Retrieved from
http://www.statecoverage.org/files/Automatic%20Enrollment%20Strategies%20-
%20Helping%20State%20Coverage%20Expansions%20Achieve%20Their%20Goals.pdf
Dorn, S. (2009, December). Applying 21st-Century Eligibility and Enrollment Methods to National
Health Care Reform. Urban Institute. Retrieved from
http://www.urban.org/uploadedpdf/411985_national_health_care_reform.pdf
Dorn, S., Hill, I., & Hogan, S. (2009, November). The Secrets of Massachusetts’ Success: Why 97
Percent of State Residents Have Health Coverage. Robert Wood Johnson Foundation.
Retrieved from http://www.rwjf.org/files/research/51368fullreport.pdf
Dorn, S., & Kenney, G. (2006, June).Automatically Enrolling Eligible Children and Families Into
Medicaid and SCHIP: Opportunities, Obstacles, and Options for Federal Policymakers. The
Commonwealth Fund. Retrieved from
http://www.commonwealthfund.org/~/media/Files/Publications/Fund%20Report/2006/Jun
/Automatically%20Enrolling%20Eligible%20Children%20and%20Families%20Into%20Med

DeLeire, T., Leininger, L., Dague, L., Mok, S., Friedsam, D. E15
MMRR 2012: Volume 2 (2)

icaid%20and%20SCHIP%20%20Opportunities%20%20Obsta/Dorn_auto%20enrollingchild
ren_931%20pdf
Dubay, L., Holahan, J., & Cook, A. (2007, January–February). The Uninsured and the
Affordability of Health Insurance Coverage. Health Affairs, 26(1), w22–w30. PubMed
http://dx.doi.org/10.1377/hlthaff.26.1.w22
Dubay, L. & Kenney, G. (2003, October). Expanding Public Health Insurance to Parents: Effects
on Children’s Coverage Under Medicaid. Health Services Research, 38(5), 1283–1302.
PubMed http://dx.doi.org/10.1111/1475-6773.00177
Federman, A.D., Vladeck, B., & Siu, A. (2005, January–February). Avoidance of Health Care
Services Because of Cost: Impact of the Medicare Savings Program. Health Affairs, 24(1),
263–270. PubMed http://dx.doi.org/10.1377/hlthaff.24.1.263
Georgetown University Health Policy Institute Reaching Eligible but Uninsured Children in
Medicaid and CHIP. (2009, March). Retrieved from http://ccf.georgetown.edu/index/cms-
filesystem-action?file=strategy+center%2Feligibleuninsured%2Feligibleuninsuredccf.pdf
Health Care and Education Reconciliation Act, 42 U.S.C. § 1001 et seq. (2010).
Holahan, J., Cook, A., & Dubay, L. (2007, February). Characteristics of the Uninsured: Who is
Eligible for Public Coverage and Who Needs Help Affording Coverage? Kaiser Commission
on Medicaid and the Uninsured. Retrieved from
http://www.kff.org/uninsured/upload/7613.pdf
Holahan, J., & Headen, I. (2010, May). Medicaid Coverage and Spending in Health Reform:
National and State-by-State Results for Adults at or Below 133% FPL. Kaiser Commission on
Medicaid and the Uninsured. Retrieved from http://kff.org/healthreform/upload/Medicaid-
Coverage-and-Spending-in-Health-Reform-National-and-State-By-State-Results-for-
Adults-at-or-Below-133-FPL-Executive-Summary.pdf
Holden, S., & VanDerhei, J. (2005, July). The Influence of Automatic Enrollment, Catch-Up, and
IRA Contributions on 401(K) Accumulations at Retirement. Investment Company Institute.
July. Retrieved from http://www.ici.org/pdf/per11-02.pdf
Hudson, J.L. & Selden, T. (2007). Children’s Eligibility and Coverage: Recent Trends and a Look
Ahead. Health Affairs, 26(5), w618–w629. PubMed
http://dx.doi.org/10.1377/hlthaff.26.5.w618
Iglehart, J. (2010, February). Medicaid Expansion Offers Solutions, Challenges. Health Affairs,
29(2), 230–232. PubMed http://dx.doi.org/10.1377/hlthaff.2010.0010
Kaiser Commission on Medicaid and the Uninsured (2009, December). Where are States Today?
Medicaid and State-Funded Coverage Eligibility Levels for Low-Income Adults. Kaiser Family
Foundation. Retrieved from http://www.kff.org/medicaid/7993.cfm

DeLeire, T., Leininger, L., Dague, L., Mok, S., Friedsam, D. E16
MMRR 2012: Volume 2 (2)

Leininger, L.J., Friedsam, D., Dague, L., Mok, S., Hynes, E., Bergum, A., . . . DeLeire, T. (2011).
Wisconsin’s BadgerCare Plus Reform: Impact on Low-Income Families’ Enrollment and
Retention in Public Coverage. Health Services Research, 46(1 Pt. II), 336–347. PubMed
http://dx.doi.org/10.1111/j.1475-6773.2010.01215.x
O’Hara, B. (2009, August). Is There an Undercount of Medicaid Participants in the ACS Field
Test? Paper presented at the Joint Statistical Meetings, Statistics: From Evidence to Policy,
Washington D.C.
Patient Protection and Affordable Care Act, Act, 42 U.S.C. § 1001 et seq. (2010).
Remler, D.K. & Glied, S. (2003, January). What Other Programs Can Teach Us: Increasing
Participation in Health Insurance Programs. American Journal of Public Health, 93(1), 67–
74. PubMed http://dx.doi.org/10.2105/AJPH.93.1.67
State Health Access Data Assistance Center & the Urban Institute. (2005, August). Going
Without: America’s Uninsured Children. Retrieved from
http://www.rwjf.org/files/newsroom/ckfresearchreportfinal.pdf
Wachino, V., & Weiss, A. M. (2009, February). Maximizing Kids’ Enrollment in Medicaid and
SCHIP: What Works in Reaching, Enrolling, and Retaining, Eligible Children. Robert Wood
Johnson Foundation. Retrieved from http://www.rwjf.org/files/research/3702.pdf
Wisconsin Department of Health Services (2008, July). Medicaid Change Report. Retrieved from
http://www.dhs.wisconsin.gov/forms/F1/F10137.pdf

DeLeire, T., Leininger, L., Dague, L., Mok, S., Friedsam, D. E17
MMRR 2012: Volume 2 (2)

Appendix

Exhibit 1. Probit Models for Results in Exhibit 2

P(12 Months P(6 Months
Continuous Continuous
Enrollment) Enrollment)
Marginal Marginal
Variable Mean Coef. Effect Coef. Effect
0.5305 -0.155 -0.072 -0.229 -0.078
February enrollee
0.0119 0.0040 0.0139 0.0032
0.1740 -0.280 -0.187 -0.412 -0.217
Premium required
0.0218 0.0069 0.0240 0.0068
0.3381 -0.171 -0.080 -0.069 -0.029
Auto-enrollee
0.0124 0.0042 0.0143 0.0034
February enrollee X -0.231 -0.397
premium required 0.0260 0.0274
Auto-enrollee X -0.241 -0.220
premium required 0.0262 0.0264
0.5454 0.277 0.103 0.186 0.049
Child enrollee
0.0091 0.0034 0.0107 0.0029
Youngest HH member 0 0.2014 -0.001 0.000 -0.361 -0.095
years old 0.0249 0.0093 0.0314 0.0082
Youngest HH member 0.4022 0.007 0.003 -0.231 -0.061
1–5 years old 0.0241 0.0090 0.0307 0.0081
Youngest HH member 0.2489 0.069 0.026 -0.202 -0.053
6–12 years old 0.0244 0.0091 0.0310 0.0081
Youngest HH member 0.1184 -0.088 -0.033 -0.224 -0.059
13–18 years old 0.0250 0.0093 0.0317 0.0083
0.5681 0.051 0.019 0.068 0.018
Female
0.0078 0.0029 0.0088 0.0023
Number of adults in 1.6326 -0.018 -0.007 -0.022 -0.006
household 0.0063 0.0023 0.0072 0.0019
Number of children in 2.3287 0.030 0.011 0.025 0.007
household 0.0031 0.0011 0.0035 0.0009

DeLeire, T., Leininger, L., Dague, L., Mok, S., Friedsam, D. E18
MMRR 2012: Volume 2 (2)

Exhibit 1 (cont.)
P(12 Months P(6 Months
Continuous Continuous
Enrollment) Enrollment)
Marginal Marginal
Variable Mean Coef. Effect Coef. Effect
0.5779 -0.012 -0.004 -0.062 -0.016
High school graduate
0.0093 0.0035 0.0108 0.0028
0.1924 -0.035 -0.013 -0.077 -0.020
More than high school
0.0117 0.0044 0.0133 0.0035
0.3642 -0.016 -0.006 0.007 0.002
Rural county
0.0082 0.0031 0.0093 0.0025
HH income under 150% 0.7316 0.333 0.124 0.211 0.056
FPL 0.0228 0.0085 0.0251 0.0066
HH income between 150 0.1814 0.046 0.017 -0.037 -0.010
and 200% FPL 0.0188 0.0070 0.0196 0.0052
County unemployment 5.0826 0.035 0.013 0.034 0.009
rate 0.0039 0.0015 0.0044 0.0012
-0.053 0.987
Constant
0.0388 0.0457
Log likelihood -75947.4 -55920.2
Observations 125418 125418
NOTES. Standard errors in italics. Marginal effects measured as (dydx) at mean; for factor variables, as change from 0 to 1. Covariates
measured at beginning of spell.
SOURCE: Authors' calculations from Wisconsin administrative data.

DeLeire, T., Leininger, L., Dague, L., Mok, S., Friedsam, D. E19
MMRR 2012 Volume 2, Number 2

Medicare & Medicaid Research Review

2012
Volume 2, Number 2

Mission Statement
Medicare & Medicaid Research Review is a peer-reviewed, online journal reporting data and research that informs
current and future directions of the Medicare, Medicaid, and Children’s Health Insurance programs. The journal
seeks to examine and evaluate health care coverage, quality and access to care for beneficiaries, and payment for
health services.
http://www.cms.gov/MMRR/

U.S. Department of Health & Human Services

Kathleen Sebelius
Secretary

Centers for Medicare & Medicaid Services

Marilyn Tavenner
Acting Administrator

Center for Strategic Planning

Anthony D. Rodgers
Deputy Administrator and Director

Editor-in-Chief David M. Bott, Ph.D. Senior Editor Cynthia Riegler, M.A.

Associate Editors
John Hsu, M.D., M.B.A, M.S.C.E. Jennifer Polinski, Sc.D, M.P.H.
Harvard Medical School Brigham & Women's Hospital
James H. Marton, Ph.D Robert Weech-Maldonado, Ph.D.
Georgia State University University of Alabama at Birmingham

Editorial Board
Gerald S. Adler, M.Phil. Melissa A. Evans, Ph.D.
CMS/Center for Strategic Planning CMS/Center for Program Integrity
Andrew Bindman, M.D. Jesse M. Levy, Ph.D.
University of California, San Francisco CMS/ Innovation Center
William J. Buczko, Ph.D. Isidor R. Strauss, F.S.A.
CMS/Innovation Center CMS/Office of the Actuary
Todd Caldis, Ph.D., J.D. Fred G. Thomas, Ph.D., C.P.A.
CMS/Office of the Actuary CMS/ Innovation Center
Craig F. Caplan, M.A.
CMS/ Center for Medicare

Contact: mmrr-editors@cms.hhs.gov
Published by the Centers for Medicare & Medicaid Services
All material in the Medicare & Medicaid Research Review is in the public domain and may be duplicated without permission.
Citation to source is requested.

ISSN: 2159-0354
doi: http://dx.doi.org/10.5600/mmrr.002.02.a02
E20