Do State Parity Laws Reduce the
with Mental Health Care Needs?
Colleen L. Barry and Susan H. Busch
Objective. To study the financial impact of state parity laws on families of children in
need of mental health services.
Data Source. Privately insured families in the 2000 State and Local Area Integrated
Telephone Survey National Survey of Children with Special Health Care Needs
Study Design. We examine whether state parity laws reduce the financial burden on
families of children with mental health conditions. We use instrumental variable es-
timation controlling for detailed information on a child’s health and functional impair-
care with other CSHCN.
Principle Findings. Multivariateregressionresults indicatethatlivinginaparity state
significantly reduced the financial burden on families of children with mental health
care needs. Specifically, the likelihood of a child’s annual out-of-pocket (OOP) health
care spending exceeding $1,000 was significantly lower among families of children
needing mental health care living in parity states compared with those in nonparity
states. Familieswith childrenneeding mentalhealth care in parity states were also more
likely to view OOP spending as reasonable compared with those in nonparity states.
that a child’s health needs caused financial problems. The likelihood of reports that
additional income was needed to finance a child’s care was also lower among families
with mentally ill children living in parity states. However, we detect no significant
difference among residents of parity and nonparity states in receipt of needed mental
Conclusion. These results indicate that state parity laws are providing important
economic benefits to families of mentally ill children undetected in prior research.
Key Words. Parity, mental health, CSHCN, economic burden
Theintentofparitylawsistoimprove equity inprivateinsurancecoveragefor
mental health care. Health insurers have covered mental health care at a
r Health Research and Educational Trust
significantly lower level than coverage for other conditions for many years
(U.S. Bureau of Labor Statistics 1982; Jensen et al. 1998; Buck et al. 1999;
Barry et al. 2003). Health plans commonly require higher cost sharing and
impose special inpatient day and outpatient visit limits on service utilization
for mental health treatment. By requiring equivalent levels of insurance cov-
erage for mental health and general medical care, parity policies aim to
broaden access to services while decreasing the financial burden associated
with seeking treatment. Parity insurance regulation may particularly benefit
for high treatment expenses.
A seriesof multistateresearch studies find that stateparity laws have had
Sturm 2000; Sturm 2000; Bao and Sturm 2004). These findings have been
cited as evidence that state laws are not fulfilling the aims of policy makers.
This article examines the effects of state parity laws but differs from
previous research in several important ways. Most notably, all previous mul-
tistate studies considered the impact of state parity laws on adults. In contrast,
health care needs. Unlike adults, children with mental health disorders are
likely to be covered under private health insurance irrespective of the severity
of their illness. Also important, prior multistate studies have not evaluated the
impact of state parity laws on the economic burden of seeking mental health
treatment. In this study, we examine how state parity laws affect out-of-pocket
(OOP) health care spending and other measures of the financial burden of
mental health treatment costs on families.
We use data from the national 2000 State and Local Area Integrated
TelephoneSurvey(SLAITS)National SurveyofChildren withSpecialHealth
Care Needs (CSHCN) to study how state parity laws affect children with
mental health disorders. These data are a large nationally representative sam-
ple of children whose parents report that they had more health care needs or
disability than other children, and that the their condition is expected to last
for at least 12 months. This allows us to examine the impact of parity on the
Address correspondence to Colleen L. Barry, Ph.D., Assistant Professor, Department of Epi-
demiology and Public Health, Yale University School of Medicine, Division of Health Policy and
Administration, 60 College Street, New Haven, CT 06520. Susan H. Busch, Ph.D., Associate
Professor, is also with Department of Epidemiology and Public Health, Yale University School of
Medicine, Division of Health Policy and Administration, New Haven, CT.
1062 HSR: Health Services Research 42:3, Part I (June 2007)
child populationmost likelytobenefit——thosewithmore severementalhealth
care needs. We compare those in parity and nonparity states and those need-
ing mental health care with other special needs children.
A difficulty in estimating the impact of state parity laws on health care
outcomes in cross-sectional data lies in the potential confounding effects of
state characteristics that may be associated with both the passage of parity
legislation and the use of mental health care by children and adolescents (also
known as omitted variable bias or endogeneity). Level of stigma is one ex-
ample of a factor that might be heterogeneous across states that could affect
both the likelihood of state enacting a parity policy and the family financial
burden of mental health services use within a state. We use instrumental
functional impairment to address this problem. We chose our instruments on
the basis of political theory on state policy making. Dating back to Daniel
Elazar’s classic study of state political cultures (1966), empirical research has
supported the view that electoral beliefs, legislative institutions, and policy
priorities are uniquely ascribed to a state.1Based on this conceptual frame, we
structure, a state electorate’s political ideology, and the professionalism of a
Our results indicate that living in a parity state significantly reduced the
financial burden on families of children with mental health care needs. Spe-
cifically, the likelihood of a child’s annual OOP health care spending exceed-
ing $1,000 was significantly lower among families of children needing mental
health care living in parity states compared with those in nonparity states.
Families with mentally ill children living in a parity state also had significantly
lower reports of various other measures of financial burden compared with
those in nonparity states. These results indicate that state parity laws are pro-
viding important economic benefits to families of mentally ill children un-
detected in prior research.
Rationale for Parity
Insurers have traditionally limited coverage for mental disorders out of con-
cern that generous benefits could lead to high costs due to long-term psycho-
therapy and lengthy hospital stays. The RAND Health Insurance Experiment
State Parity Laws and the Financial Burden on Families1063
decreased cost sharing for ambulatory mental health care was roughly double
that observed for outpatient medical services under fee-for-service insurance
(Manning, Wells, and Buchanan 1989; Newhouse 1993). Adverse selection
also explains comparatively low levels of mental health coverage. Mental
illnesses are often persistent, and individuals with these disorders tend to use
other health services at higher rates compared with otherwise similar indi-
viduals (Ellis 1988; Frank, Glazer, and McGuire 2000). Various studies have
identified that a history of mental health use significantly affects choice of
health plan (Perneger et al. 1995; Deb et al. 1996; Cao 2006). Because more
generous coverage appears to attract these costlier users, insurers have a fi-
nancial incentive to compete to avoid enrolling them by providing minimal
mental health benefits (Frank and McGuire 2000).
Claims that insurance should not discriminate against persons with
mental disorders and that benefit limits arbitrarily prevent access to effective
treatments form the basis of arguments in support of benefit parity. Under
traditional insurance arrangements, parity regulation can also counteract
selection-related market competition. Finally, parity policies have been
advancedas a mechanism forimproving access to services forwhich there are
high levels of unmet need. Research indicates that a substantial majority of
those with mental disorders do not receive treatment in a given year (Regier
et al. 1993; Kessler et al. 1994; Wang et al. 2005).
Importantly, the demand response noted in the HIE may no longer be a
valid justification for discrepancies in coverage in the era of managed care. If
large increases in service use. Recent successes in expanding mental health
insurance coverage without triggering substantial cost increases provide evi-
dence that traditional moral hazard concerns may be less pressing given the
proliferation of managed mental health care (Goldman, McCulloch, and
Sturm 1998; Ma and McGuire 1998; Rosenbach et al. 2003; Goldman et al.
with reduced cost sharing and the elimination of limits without exacerbating
adverse selection, parity could have the effect of increasing risk spreading and
improving the efficiency of the health insurance market.
History of Mental Health Insurance Regulation
Various forms of benefit regulation have been implemented over the years to
address limits on private insurance coverage for mental health care. In the
1064HSR: Health Services Research 42:3, Part I (June 2007)
1970s and 1980s, state legislatures enacted mandated benefit laws in over two
dozen states. These laws served to establish minimum levels of coverage
(McGuire and Montgomery 1982), and did not address the issue of benefit
equivalence for mental health with general health care. Mental health advo-
cates began pressing for benefit parity at the state and federal levels in the
1990s; at this time the issue was framed explicitly as an antidiscrimination
mental illnesses. This law does not apply to other kinds of benefit limits, such
as special annual day or visit limits and higher cost sharing.2Because of its
limited scope, federal parity has been viewed as a largely symbolic policy
change. Efforts to expand it have repeatedly stalled in Congress.
In the absence of a broader federal parity law, 37 states have passed
parity legislation. These state policies vary substantially in terms of the type of
benefits covered, diagnoses included, population eligible, and direction re-
garding use of managed care. Some policies are quite limited in scope. For
example, South Carolina’s parity law applies to public employees only and
North Carolina’s policy mirrors the federal partial parity law by prohibiting
special dollar limits while continuing to allow other types of mental health
benefit limits. More extensive state laws require equal cost sharing and pro-
hibit the imposition of special inpatient day and outpatient visit limits. State
laws also differ in the conditions covered with some applying to only a subset
of severe or ‘‘biologically based’’ disorders.3
Previous Research on Effects of Parity
Research on effects of state parity laws come from an evaluation of compre-
hensive parity in Vermont and three multistate analyses. The Vermont study
found that consumers paid a smaller share of the total amount spent on MH/
SA services after implementation of parity (Rosenbach et al. 2003). For those
with serious mental health conditions, the decrease in OOP spending follow-
little to no impact of parity. An early analysis by Sturm using Community
Tracking Study (CTS) detected no statistically significant differences in per-
ceptions of perceived insurance generosity or access among those living in
parity and nonparity states (2000). In a subsequent analysis using the Health
Care for Communities (HCC) data, Pacula and Sturm found that state parity
laws appear to have a small positive effect on the level of utilization among
adults in poor mental health but not for other adults (2000). In a more recent
State Parity Laws and the Financial Burden on Families1065
paper using two waves of HCC data, Bao and Sturm found no statistically
significant effects of state parity laws on perceived quality of health insurance
coverage, perceived access to needed health care, and use of mental health
multistate studies directly examined the effect of parity on financial protection.
Studying Effects of State Parity Laws on Children and Families
Parity laws have the potential to provide economic benefit to those with the
most severe disorders by expanding catastrophic insurance protection
through the eliminating annual and lifetime service and dollar limits. Al-
though no one has previously studied the effect on children, children with
severe mental illnesses (SMIs) may be more likely than their adult counter-
parts to have private insurance coverage. Adults obtain privatecoveragemost
often either through the workplace or a spouse. Yet, only about a third of
adults with SMI are employed (Kaye 2002). Adults with SMI are also less
likelytobemarried ( Jayakody,Danziger,and Kessler1998).Incontrast,even
children with the most SMIs may still be covered through a parent’s private
insurance policy. Young people covered under private insurance have sub-
Sturm 2001). In contrast, among those with public coverage, adults have
higher inpatient use than adolescents.
Research examining the cost of eliminating limits also suggests that
children will benefit most from parity legislation. Sturm (1997) compares the
cost per enrollee in plans with no limits to the cost per enrollee in plans that
limit mental health care to 30 inpatient days and 20 outpatient visits. He finds
the increase in cost per child beneficiary is higher than that for adult bene-
ficiaries (23 versus 17 percent).
Although children have low rates of any use of mental health services,
those with behavioral health problems can be very expensive to treat. In the
extreme case, catastrophic treatment costs can result in a family relinquishing
custodial rights to gain access to mental services for a child. According to a
U.S. General Accounting Office (GAO) report, state child welfare officials in
19 states and juvenile justice officials in 30 counties estimated that parents
placed over 12,700 children in welfare or juvenile justice systems to receive
mental health care treatment in 2001 after exhausting savings and health
Most state laws do not distinguish between parity in coverage for adults
and children. However, a number of parity statutes specify criteria for parity
1066HSR: Health Services Research 42:3, Part I (June 2007)
coverage in childhood. For example, the California SMI parity statute iden-
tifies a ‘‘serious emotional disturbance of a child’’ as one or more mental
disorders identified in the DSM (other than a primary substance abuse or
development disorder) that result in behavior inappropriate to child’s age
according to expected developmental norms (Peck 2001). Under the Massa-
chusetts SMI law, insurers are required to provide coverage to children
and adolescents for treatment of nonbiologically based mental, behavioral,
and emotional disorders described in the DSM that ‘‘substantially interfere or
limit functioning and social interactions’’ (Ruthardt 2000). The law requires
documentation by physician or evidence of inability to attend school, hos-
pitalization, or behavior patterns posing serious danger to the child’s self
or others. No state parity laws explicitly exclude coverage of disorders in
SLAITS National Survey of CSHCN data are used to analyze the effects of
state parity laws. The SLAITS National Survey of CSHCN was collected by
Van Dyck et al. 2002, 2004; Newacheck, Inkelas, and Kim 2004; Newacheck
and Kim 2005).6The sampling scheme used was devised specifically with the
goal of allowing researchers to make inferences about state differences in the
experiences of these children. In each of the 50 states approximately 3,500
households with children were screened to yield 750 individual CSHCN by
distinct questions relating to a child’s health care needs and activity limita-
tions.7If a parent responded affirmatively to the screener question, they were
asked whether the condition was expected to last 12 months or longer. If any
of the five questions were answered affirmatively and the condition was ex-
pected to last 12 months, the child was included in the survey. For these
children, the extent of the child’s disability and health care needs were as-
sessed by the respondent (usually a parent). Parents answered detailed ques-
tions about the types of services needed, sought, and received.
We limit our sample to children with private insurance coverage. Chil-
dren with more than one type of coverage (e.g., Medicaid and private) were
omitted from the sample. Three states enacted parity legislation during
the time of study collection. Since we do not know the precise date of the
State Parity Laws and the Financial Burden on Families1067
were obtained through the National Alliance for the Mentally Ill website
(2004) and validated with data collected by other groups.8
We used three data sources to obtain data on state-level political char-
acteristics. Information on gubernatorial party affiliation and state legislative
majorities obtained from the Council of State Governments was used to de-
velop a measure of political party power. State political party identification
scores and state ideology scores were obtained from a data set compiled by
Erikson and colleagues (Erikson, Wright, and McIver 1993; Wright et al.
2001).9These data aggregate 336 national CBS News/New York Times
polls with 400,327 respondents collected over a 23-year period. Legislative
professionalism scores by state were adopted from work by Perevill Squire
Dependent Variables. We study the effects of state parity laws on financial
positively affect a family’s financial burden by lowering the OOP cost of
obtaining mental health care services. We study four measures of financial
protection:(1) whether annual child OOP health spending (not mentalhealth
care specific) exceeded $1,000 (yes/no); (2) whether a family reported that a
child’s health care has caused financial problems (yes/no); (3) whether a
family reported needing additional income for a child’s medical expenses
(yes/no); and (4) whether a family reported that OOP charges for care were
reasonable (never, sometimes, usually, or always). We coded responses of
‘‘never’’ or ‘‘sometimes’’ as 1 and 0 otherwise. OOP spending was collected
within six ranges. To convert these categories to a dollar amount, we took the
log of the midpoint of each range.
Likewise, we might expect these laws to affect a child’s utilization of
specialty mental health services. As noted above, changes in utilization will
To measure use, respondents were first asked if the child needed mental
health care. If yes, the respondent was asked if the child received all needed
mental health care. Unfortunately, respondents were not asked about the
number of visits. Thus, we study only one measure of utilization——whether a
child received all needed mental health care. Because only those respondents
reporting a need for mental health care were asked this question, this sample
is much smaller (N54,823).
1068 HSR: Health Services Research 42:3, Part I (June 2007)
Parity Measure. The primary explanatory variable is whether a child lived in a
state with some form of parity law implemented before January 2001. In
practice, there are some challenges to studying the effects of state parity laws.
As noted above, state parity laws are quite heterogeneous and hard to
characterize. There is substantial variation in the state parity literature about
defining a parity state. States with parity laws that apply only to state
employees, mirror the federal law, or allow insurers to impose special
inpatient day or outpatient visit limits are not considered parity states in this
analysis.10Using the National Survey of CSHCN data, families in 23 states
are considered parity during this time period using these criteria.11
Other Independent Variables. We control for child and family characteristics
that are likely to affect our outcome measures but are unrelated to state parity
laws. Demographic characteristics include a child’s age, age-squared, gender,
race (Hispanic, nonwhite, black), whether the interview was conducted in a
language other than English, whether the mother has only a high school
education or less, eight dummy variables indicating family income, and the
number of adults in the household.
We also control for observable measures of disease severity and disease
characteristics. Variables denoting severity and functional impact of the
child’s disability include five dummy variables representing responses to the
five screener questions. We also include information from four additional
survey questions: parents ranking of the child’s level of disability on a scale of
1–10 (with zero the most mild and 10 the most severe), parents report of the
amount of time the child is affected by the condition (never, sometimes,
usually, always), a description of the child’s health care needs (usually stable,
change only once in awhile, change all the time), and whether the child’s
health conditions affect her ability to do things (a great deal, some, very little).
The latter three variables are coded as categorical in the model.
To overcome the potential endogeneity of state passage of parity with study
outcomes, we use state political characteristics as instruments using an instru-
mental variable estimation approach. We compared this IV approach to OLS
regression results. Our IV models are similar in flavor to a prior study of the
effects of state parity laws (Pacula and Sturm 2000). We opted to develop
alternative instruments because prior work included measures of a state’s
State Parity Laws and the Financial Burden on Families1069
supply of mental health services that we were concerned might be correlated
with our service utilization outcome. The conceptual framework for our IV
estimation approach is based on political theory of state policy making. A rich
empirical literature in political science supports the assertion that political
beliefs, institutional factors and policy priorities are uniquely ascribed to a
state. Therefore, we conceptualize the likelihood of a state enacting a parity
law as a function of state-level political characteristics.
We identified state political characteristics likely to be correlated with
the endogenous regressor (passing a state parity law) but unlikely to be cor-
related with our outcomes, thus orthogonal to the error term. First, we hy-
pothesized that Democrat-leaning states will be more likely to pass parity
legislation. We created an index of state political party power equal to one if
either a state’s governor or the majority of either chamber of a state’s legis-
lature is Democratic, zero if Republican, and 0.5 if Independent or evenly
divided in the year of enactment (for a maximum value of 3).12Second, our
prior is that states with citizens holding a more conservative political ideology
are less likely to pass parity laws due to preferences against government
mandates. We developed a state electorate’s mean political ideology score
aggregating state-level data from 336 national media polls. Respondents in
or conservative. A score of ?100 is assigned to each conservative response, a
score of 0 is assigned to each moderate response, and a score of 1100 is
specific political events. Finally, we expected that the professionalism of a
state’s legislature would be negatively correlated with a state’s parity enact-
ment because less professional state legislatures have been more active in
enacting mandated benefit laws. To develop state legislative professionalism
scores,we used anindex of salaryof state legislators, number of staff members
per legislator and total days per legislative session developed by Squire
estimate the parameters of the models. GMM was compared and chosen over
two-stage least squares and other specifications as a more efficient estimator in
the presence of heteroskedasticity of unknown form (Baum, Schaffer, and
perform nearly as well as the correctly specified maximum likelihood esti-
mator (Angrist 1991, 2001). We predict the passage of parity laws in our first
stage, and then use these predicted values to estimate the effect of state parity
1070 HSR: Health Services Research 42:3, Part I (June 2007)
laws on financial burden and utilization outcomes in equations (1) and (2). We
estimate the following equation for each of our outcomes:
economicburden ¼ a þ Xib1þ b2paritysþ b3needMHcareiþ e
where X is a vector of individual characteristics, parity refers to whether the
reported that thechild needed mental health care in thepastyear. Because we
hypothesize that parity will only impact those in need of mental health care,
we next run the following model which includes a term indicating the inter-
action of parity and need for mental health care:
economicburden ¼a þ Xib1þ b2paritysþ b3needMHcarei
þ b4paritys? needMHiþ e
A priori, we anticipate that the estimated coefficient b2will be insignificant
because we expect parity to only effect economic burden for the subset of
children needing mental health care. We expect b3to be relatively large and
positive. That is, we expect children needing mental health care to have a
greater economic burden compared with other children with special health
the interaction between these two variables. Because we expect the presence
of a state parity laws to reduce the economic burden of health care for those
children with mental illness, we anticipate that this interaction term will be
The coefficient b4can be thought of as a difference-in-difference esti-
mate of the impact of parity on mental health care need. In this difference-in-
difference estimation, we implicitly compare the effect of parity legislation for
thosewithmentalhealthneeds withthosereportingno need formentalhealth
approach for all four economic burden outcomes.
Table 1 reports unadjusted descriptive statistics for the full sample of privately
insured CSHCN (N521,930). About half of children in the sample live in a
parity state and 20.8 percent report needing mental health care. The mean
reported level of disability is 3.3 on a 10-point scale. More than half of the
sample report that their health condition affects their ability to do things and
State Parity Laws and the Financial Burden on Families1071
Table1: Descriptive Statistics, SLAITS National Survey of CSHCN
Lives in parity state (%)
Age (mean years)
Interview conducted in language other than English (%)
Mother has education of high school or less (%)
Only 1 adult in household (%)
Poverty level (%)
Parent reports of child’s disability
Child needed mental health care (%)
Child’s disability level from 0–10 as most severe (mean response)
How often child’s health conditions affect ability to do things (%)
Stability of child’s condition (%)
Change all the time
Change only once in awhile
CSHCN screener questions (%)
Child needs more medical care than peers
Child currently needs prescription medication
Child limited in ability to do things
Child needs physical or speech therapy
Child has emotional, behavioral, or developmental problems
Child OOP spending 4$1,000(%)
Respondent reports that OOP spending on child’s care is never
or rarely reasonable (%)
Respondent reports that child’s health care has caused financial problems (%)
Respondent reports that family needed additional income to care for child (%)
State’s institutional partisanship power structure (mean ranking)
State electorate’s political ideology (mean ranking)
Professionalism of state’s legislature (mean ranking)
Note: Full sample includes CSHCN respondents reporting private insurance (and no other) cov-
erage. We exclude CSHCN living in the District of Columbia and in states where the timing of
enactment of parity makes it ambiguous whether the law was in effect at the time of the survey.
CSHCN, Children with Special Health Care Needs; OOP, out-of-pocket.
1072 HSR: Health Services Research 42:3, Part I (June 2007)
10.5 and 60 percent of the sample are male. Thirteen percent are nonwhite
and 6.4 percent are Hispanic children. Ninety-two percent of the population
reported familyincomesover150 percentofthefederalpovertylevel.Table1
also includestheproportionoffamilies responding affirmativelytoeach ofthe
five CSHCN screener questions.
reported. Fourteen percent of families spent more than $1,000 annually OOP
to treat a child’s special health care needs and 28.2 percent viewed OOP
spending on a child’s care as unreasonable. Likewise, 17.4 percent of re-
spondents reported that a child’s health care treatment had caused financial
problems, and 14.3 percent needed additional income to care for their child.
Table 2 compares unadjusted mean outcomes among families of
CSHCN in parity and nonparity states with and without mental health care
needs. The difference in the proportion of those receiving needed mental
at the .10 level). Among those in need of mental health services, unadjusted
mean financial burden measures differ somewhat. In particular, families in
parity states were slightly less likely to report that a child’s mental health care
needs caused financial problems and required additional income to treat than
to their counterparts in parity states. No significant differences were detected
in other economic burden measures.
In firststage GMM regression results predicting whether a state adopted
a parity law, state-level political characteristics appear to be good instruments
(results not shown). Staiger and Stock (1997) characterize weak instruments as
those where the partial correlation between the instruments and the included
endogenous variables is low (e.g., F statistic o10). The F statistic of joint
significance for these policy variables is 11.21. Coefficients indicate that both
political party power and electoral ideology are positively and significantly
associated with passing parity as expected. States with Democrat-dominated
legislature was significantly and negatively associated with passing parity as
Table 3 presents stage two instrumental variable regression results for
model specificationswithand withoutinteractions.14Results aredescribed for
interaction models only. As anticipated, families of children needing mental
than families of other special needs children across these models. Likewise, as
we expected, the coefficients estimating the parity effect are insignificant in all
State Parity Laws and the Financial Burden on Families1073
Unadjusted Effect of Parity Law on Economic Outcomes
Need for Mental
Need for Mental
Need for Mental
Child OOP spending
OOP spending reasonable
(15never, rarely) (%)
Child’s health care has
caused financial problems (%)
Needed additional income to
care for child (%)
Received all needed mental
health care (%)
1074HSR: Health Services Research 42:3, Part I (June 2007)
burden of only the subset of children needing mental health care. The key
variable of interest in each of these models is the interaction term estimating
the effect of living in a parity state on outcomes among families of children
with mental health care needs compared with other CSHCN families. These
results uniformly indicate that living in a parity state significantly reduced the
financial burden on families of children with mental health care needs. We
detect no significant difference among residents of parity and nonparity states
in receipt of needed mental health care.
Based on these regression results, we calculate predicted effects of living
in a state with a parity law on a child reporting a need for mental health care.
These results are presented in Table 4. We find that the predicted probability
Table3: IV GMM Regression Results
Full Sample with
Interaction of Parity and
Mental Health Care Need
OOP spending 4$1,000
Parity law in effect
Needed mental health care
Parity ? mental health
OOP spending reasonable (15never, rarely)
Parity law in effect
Needed mental health care
Parity ? mental health
Childs health care has caused financial problems
Parity law in effect
Needed mental health care
Parity ? mental health
Needed additional income to care for child
Parity law in effect
Needed mental health care
Parity ? mental health
Received all need mental health care (N53,799)
Parity law in effect
0.011 (0.013) N/A
Note: Authors estimated the same model using OLS. These results were similar in direction to
the IV results presented above but statistically insignificant (results available from authors upon
OOP, out-of-pocket; GMM, generalized method of moments.
State Parity Laws and the Financial Burden on Families 1075
of a child’s annual OOP health care spending exceeding $1,000 was 7 per-
centage points lower among families of children needing mental health care
estimates suggests that 21 percent of children living in a parity state with a
reported need for mental heath care would have annual OOP spending in
excess of $1,000 compared with 28 percent of in nonparity states. Families
with children needing mental health care in parity states had a 11 percentage
points lower probability of viewing these OOP costs as reasonable compared
with those in nonparity states. Likewise, families living in a parity state had a
10 percentage points lower probability of reporting that a child’s health needs
caused financial problems. The likelihood of reports that additional income
families with mentally ill children living in parity states.
In this paper, we examine how state parity laws affect CSHCN reporting a
need for mental health care. Our results indicate that state parity enactment
leads to beneficial outcomes undetected in prior research. We find that fam-
ilies living in a parity state had a significantly lower financial burden due to
caring for children with mental health care needs compared with their coun-
terparts in nonparity states. We detect no significant difference among resi-
dents of parity and nonparity states in receipt of needed mental health care.
These findings are important for several reasons. First, understanding
the policy effects of mental health benefit regulation on children is valuable
from a societal perspective. A unique characteristic of mental health disorders
is that they often emerge in childhood and young adulthood, and can be
Reported Need for Mental Health Care
Parity States Nonparity States
Out of pocket spending on child greater than $1,000 (%)
OOP spending reasonable (15never, rarely) (%)
Child’s health care has caused financial problems (%)
Needed additional income to care for child (%)
Received all needed mental health care (%)
1076 HSR: Health Services Research 42:3, Part I (June 2007)
highly disruptive from an educational and professional standpoint. The Na-
tional Comorbidity Survey Replication found that half of all lifetime cases
DSM-IV disorders start by age 14 (Kessler et al. 2005). If parity policies im-
prove access to high-quality mental health care to individuals at younger ages,
they may have beneficial indirect effects on educational attainment and long-
term earning potential.
We note two other benefits of studying effects on children using the
National Survey of CSHCN data set. First, these recently released data pro-
vide an opportunity to assess policy effects after 23 states passed parity laws.
We view this as an advantage of our study since prior research provides
evidence on the effects of either early-enacting states (Pacula and Sturm 2000;
Sturm 2000) or a subset of states enacting parity during a 2-year period (Bao
and Sturm 2004). Second, a unique attribute of these data is the inclusion of
extensive information on family financial burden along with some (albeit
limited)data on perceivedneed for services.Ourfindings provideevidence of
the beneficial effects of state parity laws on risk spreading with no detectable
differences in receipt of needed care. Interestingly, only about 14 percent of
families needing mental health care in both parity and nonparity states report
not receiving needed mental health care, suggesting a somewhat less severe
access problem than might have been expected within this population of
special needs children.
It is important to note that OOP spending provides information on the
net effect of changes in OOP price and changes in quantity, and we cannot
disentangle these two effects in this study. In theory, a decrease in OOP
spending could reflect a decrease in OOP price (due to parity) alongside an
increase, a decrease or no change in the quantity of use. If consumer cost
sharing decreased but quantity increased dramatically then OOP spending
may increase despite a reduction in the financial burden of seeking treat-
Several other limitations with this analysis are worth noting. First, as
mentioned previously, the CSHCN survey is a single cross section. Making
causal inferences is more difficult with cross-sectional data than with panel
data. Unfortunately, no panel data are available at this time with detailed
health and disability information for children large enough to make state
estimates. Nonetheless, the recent availability of the CSHCN is a dramatic
improvement over previously available data. Second, there are some limita-
tions with our outcome variables. Our measures of financial burden related to
a child’s receipt of all health care paid for OOP. A more ideal outcome would
State Parity Laws and the Financial Burden on Families1077
OOP spending. Also, we measure a child’s service utilization as the receipt of
all needed mental health care. This outcome provides no information on the
clinical nature of a child’s need for mental health care, the level of mental
health care use (e.g., number of outpatient visits/inpatient days) or whether
care was delivered in the primary or specialty mental health care sector.
health sector.) Although we restrict our analysis to families with private health
insurance, we are unable to identify those covered by self-insured plans ex-
empt from state parity laws under the Employee Retirement Income Security
Act (ERISA). Therefore, we are unable to exclude a subset of our sample
state parity laws shares this same data limitation. However, this limitation
would bias our analysis toward finding no effects. Finally, if parity laws lead to
more effective treatment being received, measures of child’s health and func-
tional limitations included as control variables in our model will be endog-
in scope or extending the existing federal parity law might produce greater
financial protection from the costs of treating mental illness. Various states
are considering legislation to expand existing state parity policies. At the fed-
eral level, legislation has stalled for the pass few sessions of Congress that
would provide more comprehensive parity by prohibiting the use of special
day or visit limits and higher cost sharing for mental health care. A political
President George W. Bush publicly stated his support for more equivalent
mental health coverage for persons with mental illness. Most recently, the
President’s New Freedom Commission on Mental Health report reiterated
support for full parity in insurance coverage for mental health and physical
health care (2003).
It is worth noting that our findings are consistent with the two largest
parity evaluations conducted to date. Evaluations of both the comprehensive
mental health and substance abuse parity in the Federal Employees Health
Benefits (FEHB) program and the Vermont parity law found that these pol-
icies lowered OOP spending without significantly increasing total mental
health treatment costs (Rosenbach et al. 2003; Goldman et al. 2006). Collect-
ively, these findings signal that parity policies can produce important eco-
nomic benefits to families without prompting large spending increases. More
research is needed to assess how these economic benefits may differentially
1078 HSR: Health Services Research 42:3, Part I (June 2007)
work is that we detect aggregate state parity effects even though some state
laws are far less comprehensive in scope that either the FEHB or Vermont
This research was supported by a grant (no. 56465) from the Robert Wood
Johnson Foundation through the Changes in Health Care Financing and Or-
ganization (HCFO) initiative. Dr. Barry also received training grant support
from the National Institute of Mental Health (T32 MH19733-08). The authors
are grateful to Richard G. Frank, Haiden A. Huskamp, Thomas G. McGuire,
and three anonymous reviewers for helpful suggestions on an earlier draft of
1. See, for example, Sharkansky (1969), Peters and Welch (1978), Hayes and Stone-
cash (1981), Hanson (1983, 1984), Klingman and Lammers (1984), and Fitzpatrick
and Hero (1988).
2. Companies with fewer than 50 employees and those that offer no mental health
benefit are exempt from the federal parity law. Payers experiencing more than a 1
percent increase in premiums as a result of federal parity can apply for an ex-
3. Mental health conditions typically characterized under state parity laws as severe
or biologically based include schizophrenia, schizoaffective disorder, bipolar dis-
order, major depression and sometimes autism, anorexia/bulimia, obsessive com-
pulsive disorder, and panic disorder.
4. Among individuals spending more than $1,000 annually on MH/SA services,
OOP spending was reduced by more than half. Within the two Vermont health
plans studied, use of outpatient mental health services increased without prompt-
ing substantial spending growth after implementation of parity. For the two largest
health insurers in the state of Vermont, the level of use increased slightly in one
plan and decreased in the other.
5. Nationwide, this estimate is likely to be higher since 32 states, including the five
largest states, and many counties were unable to provide data on the number of
affected children. No formal federal or state tracking of these placements occurs.
6. More detailed information about this data set, including the questionnaire and
codebooks are available at http://www.cdc.gov/nchs/about/major/slaits/cshcn.htm
State Parity Laws and the Financial Burden on Families1079
7. These include questions regarding inordinate service use, use of prescription
medications, activity limitations, participation in special therapy, and emotional/
8. Other data used to validate state parity date were obtained from the American
Psychiatric Association, National Mental Health Association, National Council of
State Legislatures, and published articles (Gitterman 2001; Peck and Scheffler
9. These data are publicly available at http://www.indiana.edu/? iupolsci/bio_
10. State parity policies may differ by the mental health (and sometimes substance
abuse) diagnoses covered and whether a state law mandates coverage or offer,
applies to both individual and group plans, includes a small employee exemption,
andexempts employersexperiencingcost growth. Forthe purposeof this analysis,
differences along these parameters did not affect whether a state was considered a
11. These states are Arkansas, California, Connecticut, Colorado, Delaware, Georgia,
Hawaii, Indiana, Kentucky, Maine, Maryland, Minnesota, Missouri, Montana,
Nebraska, New Hampshire, New Jersey, New Mexico, Oklahoma, Rhode Island,
South Dakota, Vermont, and Virginia.
12. Both houses of the Nebraska unicameral legislature were assigned 0.5 scores.
13. The top four ranking states were New York, Michigan, California, and Massa-
chusettsandthe bottomfour wereNewHampshire,Wyoming,NorthDakota,and
Utah. While these data are over a decade old, levels of legislative professionalism
across states were deemed to have remained stable enough over time to be useful
measures for the purpose of this analysis.
14. We also stratified the sample into two groups: those with and those without a need
for mental health services. This is empirically the same as interacting only the
mental health need variable with parity. The results were qualitatively similar.
15. This interpretation is absence any managed care reaction to parity. Under man-
aged care, the assumption that a decrease in price always leads to an increase in
quantity no longer strictly holds. If there is a particularly strong managed care
reaction to parity, quantity could fall along with price. However, we believe this is
16. Data from the Medical Expenditure Panel Survey——Insurance Component
(MEPS-IC) indicate that 30.7 percent of private-sector establishments that offer
health insurance self-insured at least one plan in 2001. This proportion varies
somewhat by state (e.g., 41 percent of employers self-insured in Alaska while only
21percentself-insured inConnecticutin2001). Amongthoseprivatefirms with50
or more employees, this proportion was substantially higher (58.3 percent).
Econometrics and Epidemiology.’’ NBER Technical Working Paper No. 115.
1080 HSR: Health Services Research 42:3, Part I (June 2007)
—— —— ——. 2001. ‘‘Estimation of Limited-Dependent Variable Models with Dummy En-
dogenous Regressors: Simple Strategies for Empirical Practice.’’ Journal of Busi-
ness and Economic Statistics 19 (1): 2–16.
Bao, Y., and R. Sturm. 2004. ‘‘Effects of State Mental Health Parity Legislation on
Perceived Insurance Coverage, Access to Care, and Mental Health Specialty
Care.’’ Health Services Research 39 (5): 1361–77.
2003. ‘‘Design of Mental Health Benefits: Still Unequal after All These Years.’’
Health Affairs 22 (5): 127–37.
Estimation and Testing.’’ Stata Journal 3 (1): 1–31.
Bethell, C. D., D. Read, J. Neff, S. J. Blumberg, R. E. Stein, V. Sharp, and P. W.
Newacheck. 2002. ‘‘Comparison of the Children with Special Health Care
Needs Screener to the Questionnaire for Identifying Children with Chronic
Conditions——Revised.’’ Ambulatory Pediatrics 2 (1): 49–57.
Buck, J. A., J. L. Teich, B. Umland, and M. Stein. 1999. ‘‘Behavioral Health Benefits in
Employer-Sponsored Health Plans, 1997.’’ Health Affairs 18 (2): 67–78.
Burden, B. C. 2005. ‘‘Institutional Features and Policy Representation in the States.’’
State Politics and Policy Quarterly 5: 373–93.
Cao, Z. 2006. ‘‘Comparing Pre-HMO Enrollment Costs between Stayers and Switch-
ers: Medicare Evidence of Service-Level Selection.’’ Working paper.
Deb, P. J., V. Rubin, Wilcox-Gok, and A. Holmes. 1996. ‘‘Choice of Health Insurance
by Family of the Mentally Ill.’’ Health Economics 5 (1): 61–76.
Elazar, D. 1966. American Federalism: A View from the States. New York: Thomas Y.
Ellis, R. P. 1988. ‘‘The Effect of Prior Year Health Expenditures on Health Coverage
Plan Choice.’’ In Advances in Health Economics and Health Services Research,
edited by R. M. Scheffler and L. F. Rossiter, pp. 149–70. Greenwich, CT: JAI
and Policy in the American States. Cambridge: Cambridge University Press.
Fitzpatrick, J. L., and R. E. Hero. 1988. ‘‘Political Cultures and Political Characteristics
of the American States.’’ Western Political Quarterly 41: 145–53.
Frank, R. G., J. Glazer, and T. M. McGuire. 2000. ‘‘Measuring Adverse Selection in
Managed Health Care.’’ Journal of Health Economics 19: 829–54.
Frank, R. G., and T. G. McGuire. 2000. ‘‘Economics and Mental Health.’’ In Handbook
York: Elsevier Science.
Mental Health Parity Really Matter?’’ Administration and Policy in Mental Health
28 (5): 353–69.
Goldman, H. H., R. G. Frank, M. A. Burnam, H. H. Huskamp, M. S. Ridgely, S. T.
Normand, A. S. Young, C. L. Barry, V. Azzone, A. B. Busch, S. T. Azrin, G.
Moran, C. Lichtenstein, and M. Blasinsky. 2006. ‘‘Evaluating the Impact of
Parity Mental Health and Substance Abuse (MH/SA) Benefits in the Federal
State Parity Laws and the Financial Burden on Families1081
Employees Health Benefit (FEHB) Program.’’ New England Journal of Medicine
354 (13): 1378–86.
Goldman, W. J., J. McCulloch, and R. Sturm. 1998. ‘‘Costs and Use of Mental Health
Services before and after Managed Care.’’ Health Affairs 17 (2): 40–52.
Hanson, R. L. 1983. ‘‘Medicaid and the Politics of Redistribution.’’ American Journal of
Political Science 28: 313–39.
Hayes, S. W., and J. Stonecash. 1981. ‘‘The Source of Public Policy: Welfare Policy in
the American States.’’ Policy Studies Journal 9: 681–98.
Jayakody,R.,S. Danziger,andR.C.Kessler.1998.‘‘Early-OnsetPsychiatric Disorders
and Male Socioeconomic Status.’’ Social Science Research 27 (4): 371–87.
Jensen, G. A., K. Rost, R. P. Burton, and M. Bulycheva. 1998. ‘‘Mental Health In-
surance in the 1990s: Are Employers Offering Less to More?’’ Health Affairs 17
Kaye, H. S. 2002. Employment and Social Participation among People with Mental Disabil-
ities. San Franscisco, CA: National Disability Statistics and Policy Forum.
Kessler, R. C., P. Berglund, O. Demler, R. Jin, K. R. Merikangas, and E. E. Walters.
2005. ‘‘Lifetime Prevalence and Age-of-Onset Distributions of DSM-IV Disor-
ders in the National Comorbidity Survey Replication.’’ Archives of General
Psychiatry 62: 593–602.
Kessler, R. C., K. A. McGonagle, S. Zhao, C. B. Nelson, M. Hughes, H. U. Wittchen,
and K. S. Kendler. 1994. ‘‘Lifetime and 12-Month Prevalence of DSM-II-R
Psychiatric Disorders in the United States. Results from the National Comor-
bidity Survey.’’ Archives of General Psychiatry 51: 8–19.
Klingman, D., and W. W. Lammers. 1984. ‘‘The General Liberalism Factor in Amer-
ican State Politics.’’ American Journal of Political Science 28: 598–610.
Ma, C. A., and T. G. McGuire. 1998. ‘‘Costs and Incentives in a Behavioral Health
Care Carve Out.’’ Health Affairs 17 (2): 53–69.
Manning, W. G., K. Wells, and J. Buchanan. 1989. Effects of Mental Health Insurance:
Evidence from the Health Insurance Experiment. Santa Monica, CA: RAND.
McGuire, T. G., and J. T. Montgomery. 1982. ‘‘Mandated Mental Health Benefits in
Private Health Insurance.’’ Journal of Health Politics, Policy and Law 7 (2): 380–99.
National Alliance for the Mentally Ill. 2004. ‘‘NAMI State Mental Illness Parity
Laws: Chart of State Laws’’ http://www.nami.org/Template.cfm?Section=Issue_
National Conference of State Legislatures. 2001. ‘‘State Laws Mandatingor Regulating
Mental Health Benefits.’’ Available at http://www.ncsl.org/programs/health/
National Mental Health Association. 2002. ‘‘What Have States Done to Ensure
Insurance Parity?’’ http://www.nmha.org/state/parity/state_parity.cfm
Newacheck, P. W., M. Inkelas, and S. E. Kim. 2004. ‘‘Health Services Use and Health
Care Expenditures for Children with Disabilities.’’ Pediatrics 114 (1): 79–85.
Newacheck, P. W.,andS. E. Kim. 2005. ‘‘A National Profile of HealthCareUtilization
and Expenditures for Children with Special Health Care Needs.’’ Archives of
Pediatrics and Adolescent Medicine 159 (1): 10–7.
1082HSR: Health Services Research 42:3, Part I (June 2007)
New Freedom Commission on Mental Health. 2003. ‘‘Achieving the Promise: Trans-
forming Mental Health Care in American——Final Report.’’ DHHS Pub. No.
SMA-03-3832, Rockville, MD.
Newhouse, J. P. 1993. Free for All? Lessons from the RAND Health Insurance Experiment.
Cambridge, MA: Harvard University Press.
Pacula, R. L., and R. Sturm. 2000. ‘‘Mental Health Parity: Much Ado about Nothing?’’
Health Services Research 35 (12): 1601–8.
Peck, M. C. 2001. ‘‘Mental Health Parity for Children in California.’’ Psychiatric Services
52 (6): 743–5, 68.
Peck, M. C., and R. M. Scheffler. 2002. ‘‘An Analysis of Definitions of Mental Illness
Used in State Parity Laws.’’ Psychiatric Services 53 (9): 1089–95.
Perneger, T. V., A. F. Allaz, J. F. Etter, and A. Rougemont. 1995. ‘‘Mental Health and
Choice between Managed Care and Indemnity Health Insurance.’’ American
Journal of Psychiatry 52 (7): 1020–5.
Peters, J. G., and S. Welch. 1978. ‘‘Political Corruption in America: A Search for
Definitions and a Theory.’’ American Political Science Review 72: 974–84.
Regier, D. A., W. Narrow, D. S. Rae, R. W. Manderscheid, B. Z. Locke, and
F. K. Goodwin. 1993. ‘‘The De Facto U.S. Mental and Addictive Disorders
Service System. Epidemiologic Catchment Area Prospective 1-Year Prevalence
Rates of Disorders and Services.’’ Archives of General Psychiatry 50: 85–94.
Rosenbach, M., T. Lake, C. Young, W. Conroy, B. Quinn, J. Ingels, B. Cox, A. Pe-
terson, and L. Crozier. 2003. ‘‘Effects of the Vermont Mental Health and Sub-
stance Abuse Parity Law.’’ DHHS Pub. No. (SMA 03-3822). Rockville, MD:
Center for Mental Health Services, Substance Abuse and Mental Health
Ringel, J., and R. Sturm. 2001. ‘‘National Estimates of Mental Health Utilization and
Expenditures for Children in 1998.’’ Journal of Behavioral Health Services and
Research 28 (3): 319–33.
Ruthardt, L. 2000. ‘‘Massachusetts Department of Insurance Bulletin on Mental
Health Parity2000–06.’’ Available
Sharkansky, I. 1969. ‘‘The Utility of Elazar’s Political Culture.’’ Polity 2: 66–83.
Squire, P. 1992. ‘‘Legislative Professionalism and Membership Diversity in State
Legislatures.’’ Legislative Studies Quarterly 17 (1): 69–79.
Staiger, D., and J. Stock. 1997. ‘‘Instrumental Variables Regression with Weak Instru-
ments.’’ Econometrica 65 (3): 557–86.
Sturm, R. 1997. ‘‘How Expensive Is Unlimited Mental Health Care Coverage
under Managed Care?’’ Journal of the American Medical Association 278 (18):
—— —— ——.2000.‘‘StateParityLegislationandChangesinHealthInsuranceandPerceived
Access to Care among Individuals with Mental Illness: 1996–1998.’’ Journal of
Mental Health Policy and Economics 3 (4): 209–13.
U.S. Bureau of Labor Statistics. 1982. National Compensation Survey: Employee Benefits in
Medium and Large Size Firms, 1981. Bulletin 2140. Washington, DC: U.S. Bureau
of Labor Statistics.
State Parity Laws and the Financial Burden on Families 1083
U.S. General Accounting Office. 2003. ‘‘Federal Agencies Could Play a Stronger Role
in Helping States Reduce the Number of Children Placed Solely to Obtain
Mental Health Services.’’ Pub. GAO-03-397. Available at http://www.gao.gov/
Van Dyck, P., M. D. Kogan, D. Heppel, S. J. Blumberg, M. L. Cynamon, and P. W.
Newacheck. 2004. ‘‘The National Survey of Children’s Health: A New Data
Resource.’’ Maternal and Child Health Journal 8 (3): 183–8.
Van Dyck, P. C., M. McPherson, B. B. Strickland, K. Nesseler, S. J. Blumberg, M. L.
Cynamon, and P. W. Newacheck. 2002. ‘‘The National Survey of Children with
Special Health Care Needs.’’ Ambulatory Pediatrics 2 (1): 29–37.
Wang, P. S., M. Lane, M. Olfson, J. A. Pincus, K. B. Wells, and R. C. Kessler. 2005.
‘‘Twelve-Month Use of Mental Health Services in the United States. Results
from the National Comorbidity Survey Replication.’’ Archives of General Psychi-
atry 62: 629–40.
Wright, G. C., J. P. McIver, R. S. Erikson, and D. B. Holian. 2001. ‘‘Change and
Stability in State Public Opinion: 1977–1999.’’ Unpublished manuscript, Mid-
western Political Science Association.
1084HSR: Health Services Research 42:3, Part I (June 2007)