About the
Authors
JAY P. GREENE, is endowed chair and
head of the Department of Education Reform at
the University of Arkansas and a senior fellow
at the Manhattan Institute. He conducts research
and writes about topics such as school choice,
high school graduation rates, accountability,
and special education.
Dr. Greene’s research was cited four times
in the Supreme Court’s opinions in the
landmark Zelman v. Simmons-Harris case
on school vouchers. His articles have appeared
in policy journals such as The Public Interest,
City Journal, and Education Next;
in academic journals such as The Georgetown
Public Policy Review, Education and Urban
Society, and The British Journal of Political
Science; as well as in newspapers such as
the Wall Street Journal and the Washington
Post. He is the author of Education Myths
(Rowman & Littlefield, 2005).
Dr. Greene has been a professor of government
at the University of Texas at Austin and the
University of Houston. He received a B.A. in
history from Tufts University in 1988 and a
Ph.D. from the Department of Government at Harvard
University in 1995. He lives with his wife and
three children in Fayetteville, Arkansas.
MARCUS A. WINTERS is a senior fellow
at the Manhattan Institute. He has performed
several studies on a variety of education policy
issues including high-stakes testing, performance-pay
for teachers, and the effects of vouchers on
the public school system. His research has been
published in the journals Education Finance
and Policy, Teachers College Record,
and Education Next. His op-ed articles
have appeared in numerous newspapers, including
the Wall Street Journal, Washington
Post, and USA Today. He received
his B.A. in political science from Ohio University
in 2002, and an M.A. in economics from the University
of Arkansas in 2006. He is currently a Ph.D.
candidate in the Department of Economics at
the University of Arkansas and is expected to
complete in the spring of 2008. He resides in
Manhattan.
1. Introduction
School-choice policies have played an important
role in the education policy debate over the
last two decades. Currently, 21 school-voucher
programs in 14 states provide taxpayer-funded
scholarships to attend a private school (Enlow
2008). More than half of all states currently
allow students to enroll in government-funded
charter schools that operate outside of many
of the rules and regulations of the public school
system and do not have mandated catchment zones
that determine who may enroll in them.
Over the last few years, voucher programs for
students with disabilities have been among the
fastest-growing choice policies. In 1998, Florida’s
McKay Scholarship Program for Students with
Disabilities (McKay) became the first of its
kind to offer generous taxpayer-funded vouchers
that students identified as disabled can use
to attend a private school or a public school
other than their local one. From the time it
was first implemented statewide, in 2000–01,
to the 2006–07 school year, McKay has grown
from serving 970 students in 100 private schools
to serving 18,273 students in 811 private schools,
making it the largest school-choice program
in the United States.
Other states have recently followed Florida’s
lead by offering voucher programs similar to
McKay for disabled students. Special-education
voucher programs are currently operating in
Ohio, Utah, Georgia, and Arizona. According
to the Alliance for School Choice, bills to
implement or expand a special-education voucher
program have also passed one or both houses
of the Nevada, Wisconsin, and Virginia legislatures.
The substantial growth of school-choice policies
in the United States has created great demand
for research evaluating the impact of these
programs. Though we are interested in the impact
of choice programs on those students who use
them, it may be more important that we understand
the effect of these policies on those students
who remain in the public school system, largely
because zoned public schools will continue to
educate the vast majority of students in the
United States for the foreseeable future.
This paper is not the first to evaluate the
impact of exposure to school-choice policies
on the academic performance of students who
remain in the public school system. However,
it is the first (of which we are aware) to provide
evidence of the impact of a program aimed exclusively
at disabled students on the academic achievement
of those disabled students who remain in local
public schools.
There are two general schools of thought concerning
the impact of school-choice policies on all
types of students who remain in the public school
system. The first school of thought is that
school-choice programs decrease public school
performance by draining public schools of substantial
financial and human resources. As students leave
local public schools for alternatives, they
take with them a large portion of the funding
that the school would have received for their
education. Such losses are likely to harm students’
academic proficiency.
Additionally, it is often assumed that the
best and brightest within any particular pool
of voucher-eligible students are more likely
to take advantage of voucher programs. Several
theoretical models suggest that removing these
students from public schools would tend to stunt
the academic growth of students remaining in
the public schools as the result of a so-called
dilution of peer effects (Epple and Romano 1998,
2002; Nechyba 1999, 2000; Caucutt 2001). That
is, a particular student’s proficiency
is partly determined by the proficiency of his
fellow students—perhaps because such students
provide positive role models, stimulate classroom
discussion, and so on. Thus, as the overall
quality of students within a local public school
deteriorates, so might the academic progress
of any of its individual students.
It could be argued, however, that losing students
to school-choice programs would actually increase
the ability of public schools to elevate student
achievement. If, as many teacher groups and
public school advocates claim in debates unrelated
to school choice, the true cost of educating
a student is greater than the resources that
a school receives to educate him and if funding
is largely allocated on a per-capita basis,
then losing students to school-choice programs
could mean more resources available for the
students who remain. A further benefit could
be that as public school enrollment drops as
the result of school-choice policies, so might
class size.
Such resource arguments apply with special
force to special education. It is frequently
argued that special-education programs are uniformly
underfunded and that the large increase in the
percentage of students who are disabled has
been a substantial burden on local public schools.
If the cost of educating disabled students truly
exceeds the funding provided, then removing
a portion of these students from local public
schools would tend to increase the resources
available on a per-capita basis.
The second school of thought about these so-called
systemic effects holds that school-choice policies
might actually improve the performance of local
public schools even as, and precisely because,
such policies reduce their resources (Nechyba
2003). Many of those in favor of school-choice
policies argue that the current system, in which
students are assigned to schools on the basis
of their address, provides those schools with
a captive clientele it feels little pressure
to truly educate. This weakness is especially
pronounced when students lack the means to move
to another school zone or attend a private alternative.
Special-education students, in particular, might
suffer under this system if private schools
are hesitant to admit them because they are
more difficult to educate. Under this theory,
school-choice policies create a market for educational
services in which local public schools must
begin to compete for their students and the
resources that they draw upon by offering an
educational product of equal quality to the
alternative that the voucher makes available.
Several studies have evaluated the relationship
between exposure to school-choice policies and
public school performance. Utilizing slightly
different methods, Greene and Winters (2004),
Chakrabarti (2005), Figlio and Rouse (2006),
West and Peterson (2006), and Greene (2001)
each found that competition from a voucher program
in Florida (a different program from the one
evaluated in this paper) led to public school
gains on math and reading tests. Hoxby (2001)
found that public schools improved their performance
in response to competition from charter schools.
In a related literature, a growing body of
empirical research measures the impact of greater
exposure to schooling options on public school
students’ academic outcomes. Hoxby (2000),
Bayer and McMillan (2005), and Hanushek and
Rivkin (2003) find evidence that greater competition
among public school districts, often referred
to as Tiebout choice, leads to improved public
school performance, though McHugh (2003) finds
less evidence of this effect. Hoxby (1994) and
Dee (1998) find positive effects from unsubsidized
private school competition, while Sander (1999)
and McMillan (2004) fail to find such an effect.
Unfortunately, to date there is no quantitative
research evaluating the impact on the performance
of public schools of disabled students’
exposure to school-choice programs. This paper
begins to fill this void in the literature.
Focusing on a program directed at disabled
students may be interesting for a variety of
reasons. First, as discussed above, such policies
represent substantial growth in general in U.S.
voucher programs over the last few years.
Focusing on the impact of a special-education
voucher program is also worthwhile because a
frequent criticism of voucher programs has been
that private schools will not accept students
with disabilities because they are difficult
to educate and, if enrolled, would decrease
the average ability level of their student body
and thus their competitive advantage (Epple
and Romano 1998, 2002; Nechyba 1999, 2000; Caucutt
2001; Cullen and Rivkin 2003). This suggests
that not only highly selective private schools
may refrain from accepting disabled students,
but also the more numerous urban private schools
that educate seriously disadvantaged populations.
If private schools are unwilling to accept even
those disabled students whose tuition is paid,
the McKay program should have little, if any,
competitive effect on the performance of public
schools.
A final feature of the McKay program that makes
it of particular interest to study is the size
of both its eligible and participating populations.
An important criticism of previous school-choice
research is that their focus on small programs
may not have produced findings that would remain
valid as these programs grew in scale. For example,
in 2006–07, students in only 21 Florida
public schools were eligible to receive a voucher
from the oft-studied Opportunity Scholarship
Program (Greene and Winters 2004; Chakrabarti
2005; Figlio and Rouse 2006; West and Peterson
2006). In contrast, in 2005–06, about 15
percent of all disabled public school students
in Florida were eligible to receive a McKay
voucher.[1]
This paper utilizes a data set provided by
the Florida Department of Education to study
the impact of increased exposure to the McKay
program on the proficiency of disabled students
who remained in local Florida public schools.
This data set allows us to follow the performance
of each individual student enrolled in grades
three through ten in a Florida public school
from the 2000–01 to the 2004–05 school
years. Our ability to track the performance
of individual students over time (known as panel
data) substantially improves the accuracy of
our estimates by allowing us to directly control
for unobservable factors that play important
roles in a student’s academic progress.
Further, this rich data set enables us to disaggregate
the impact of exposure to McKay on students
with varying degrees of disability. As we will
discuss below, the ability to classify students
on the basis of their particular disability
allows us to better control statistically for
the impact of students’ disability on their
academic proficiency. There is some reason to
expect that students with different degrees
of disability have more or fewer private school
options under the McKay program. If it is the
case that students with minor disabilities are
better able to acquire a seat in a private school,
then we should expect that the McKay program
had a differential impact on the proficiency
of students who remain in the regular public
school system.
Though more research is necessary, the findings
in this paper suggest that greater exposure
to the McKay program increased the ability of
public schools in Florida to produce educational
gains for disabled students. Our results also
suggest that the impact of exposure to McKay
was greatest among students with minor disabilities—in
particular, those in the category Specific Learning
Disability, which is by far the largest special-education
category in Florida and in the United States.
Moreover, in no subgroup within special education
was student proficiency harmed, on average,
by increased exposure to the McKay program.
However, the nature of the analysis here does
not help to discover the exact reasons for such
improvement. These findings could be caused
by reallocations to remaining students of funds
made available by the departures of voucher
recipients; or by schools responding to additional
competition resulting from the advent of a market
for students; or by some other factor not yet
discussed in the literature. Future “inside
the black box” research is necessary to
distinguish the root causes of the improvements
in student proficiency resulting from exposure
to the McKay program. Nonetheless, the findings
of this paper tend to support the use of voucher
programs for disabled students.
The remainder of this paper is divided into
five additional sections. Section 2 briefly
describes the McKay program evaluated here and
its growth over the last several years. Section
3 provides a description of our rich longitudinal
data set. We discuss the empirical approach
of the paper in Section 4, and Section 5 reports
the results of our estimations. Finally, Section
6 summarizes our findings and provides a general
discussion of their implications for the discussion
of school-choice policies.
2. The
McKay Scholarship Program
The John M. McKay Scholarship Program for Students
with Disabilities is a statewide program in
Florida designed to provide disabled students
with the resources to attend either a public
school different from the one that their place
of residence would otherwise dictate or a private
school that accepted them. McKay scholarships
are available to any Florida public school student
who: 1) has been assigned an Individual Education
Plan (IEP)—essentially a judicially enforceable
contract between the school system and every
student diagnosed with a disability that specifies
the services that the school system is to provide;
and 2) was enrolled in the Florida public school
system during the previous year. Once a student
uses a McKay voucher, he remains eligible for
the program for as long as he remains in private
school or until he turns 22 years of age.
For private schools to participate in the program,
they must meet safety requirements and employ
teachers with at least a bachelor’s degree.
Unlike many other school-voucher programs, McKay
does not require private schools to accept the
sum that the voucher represents as full tuition
payment.
Importantly for our purposes, the McKay program
has undergone dramatic growth since it was first
implemented as a small pilot in the 1999–2000
school year. Table 1 reports some basic
statistics about the program for each year of
its existence. Between 2000 and 2001, when the
program was adopted statewide, and 2006 and
2007, the number of students using a McKay scholarship
increased from 970 to 18,273, making it the
largest school-voucher program of any kind in
the nation.[2] The increase
in the number of students is in large part due
to the increase, from 100 to 811, in the number
of private schools in that period willing to
accept the voucher.
McKay is distinguishable from other voucher
programs not only by the size of its eligible
population but also by the generosity of its
grants. Every eligible student is provided with
a voucher that is equivalent in value to the
sum that his original public school would spend
on him if he did not use it or the tuition charged
by the accepting private school, whichever is
smaller. This means that students with more
severe disabilities have the chance of receiving
a larger voucher amount than students with milder
disabilities because the former are more expensive
to educate. According to the Florida Department
of Education, in 2006–07 McKay scholarships
ranged in value from $5,039 to $21,907, with
an average of $7,206.[3]
3. Data
We utilize information from a rich data set
provided by the Florida Department of Education.
This data set contains student-level information
for the universe of public school students who
were enrolled in grades three through ten in
the Florida public school system from the 2000–01
to the 2004–05 school year. For each student-year,
the data set comprises demographic information
and the child’s score on the math and reading
versions of the Florida Comprehensive Assessment
Test (FCAT)—a standardized test administered
to all students in grades three through ten.
If the student is disabled, the data set reports
the student’s disability classification.
Importantly for estimation purposes, the data
set includes an identification number for each
student that allows us to follow his performance
over time.
The
data set includes information on only those
students who were enrolled in public school
during the time period covered. Because private
school students, including those who utilize
McKay scholarships, are not required to take
the state’s standardized exams, we have
no information on their performance after they
leave public school and thus no basis for evaluating
McKay’s impact on them.[4]
Consequently, our expansive data set can be
used to evaluate the impact of public school
exposure to the McKay program on the proficiency
of only disabled students within those public
schools. “Exposure” to the program
refers to the existence and pervasiveness of
schooling alternatives reasonably available
to students assigned to a particular public
school on the basis of their home address. This
definition of exposure does not depend on students’
using a voucher to leave a particular public
school for a private alternative—though
the number of private alternatives nearby and
the number of students leaving public schools
with a McKay voucher are almost certainly related—and
we do not directly control for such attrition.
Instead, we are interested in identifying to
the best of our ability how many feasible options
were available to students enrolled in a particular
public school, since a larger number of alternatives
provides students with more of an opportunity
to take advantage of the voucher program.
We adopt the strategy of some previous studies
of the systemic effect of school-choice policies
and utilize the number of private schools willing
to accept a McKay voucher within a reasonable
geographic radius of a public school as the
measure of a school’s exposure to the program.
Students could theoretically utilize a McKay
voucher at any school in the state willing to
enroll them; but as a practical matter, geography
can limit a student’s ability to attend
private school. It can therefore limit the exposure
to McKay that a public school faces. It should
come as no surprise that public schools with
a number of private schools nearby that are
willing to accept McKay vouchers are disproportionately
likely to lose their students to the program,
while public schools surrounded by few, if any,
such schools are essentially unaffected by the
program. We can thus use the number of private
alternatives within a particular radius of a
public school as a proxy for, or measure of,
the exposure that that school faces from the
program.
For each year in our data set, we used geographical
software to locate every public school and every
private school that had registered with the
state as willing to accept a McKay voucher.
We then counted separately, for each of the
years in the data set, the number of private
schools within five miles of a given public
school accepting McKay vouchers, and then the
number of such schools within ten miles of each
public school.[5] Using
a unique school identifier in the student-level
data set, we then determined the number of private
alternatives available to each student within
these two distances from his local public school
during each year covered by our data. Table
2 reports descriptive statistics by year
on the number of McKay-accepting private schools
within a five- and a ten-mile radius of local
public schools that students in our data set
would, in the absence of alternatives, attend.
An
important feature of our data set is that it
establishes not only the fact of a student’s
disability but also its nature. Federal law
specifies the disability diagnoses meriting
an IEP. Table 3 lists the categories
authorized by the Florida Department of Education
and reports the percentage of all students as
well as all disabled students in the state who
fell within each category during the 1999–2000
school year, the year before the McKay program
was adopted statewide. These categories include
blindness and deafness and range in severity
from the relatively mild Specific Learning Disability
(SLD) to Traumatic Brain Injury (TBI), a severe
disability. In Florida and around the nation,
by far the largest category is SLD, accounting
in Florida for 61.2 percent of disabled students
and 8.5 percent of all students.
One reason that the identification of a student’s
particular disability is useful is that it enables
us to do a better job of statistically controlling
for the severity of each student’s disability
when determining proficiency as measured by
test scores. Previous research, which controlled
only for whether a student had been diagnosed
as disabled, treats identically the impact of
every kind of disability on students’ academic
achievement. That is, noting only whether a
student is identified as disabled without disaggregating
by type of disability ignores the disparate
effects of degrees of disability, from mild
to severe.
Another reason that the ability to identify
a particular disabled student’s diagnosis
is important is that McKay vouchers could have
effects on the academic performance of students
remaining in the public school system according
to their various disability classifications.
In particular, if students with milder disabilities
(SLD, language disability, etc.) are more likely
to find private schools willing to enroll them
than students with more severe disabilities
(blindness, severe mental retardation, traumatic
brain injury, etc.), then their respective public
schools would face different levels of exposure
to McKay.
Among the McKay-eligible, students diagnosed
with SLD, the mildest classification, likely
have the greatest access to private school alternatives.
Some SLD students are not much different from
non-disabled students, and thus pose fewer educational
challenges for private schools. Indeed, there
is some reason to believe that a substantial
portion of students in the SLD category may
not be disabled at all. Singer et al. (1989)
find substantial variation across the states
in the functional abilities of students identified
as having a mild disability, indicating that
not all states follow the same methods for identifying
students. MacMillan and Siperstein (2001) suggest
that public schools use low achievement alone
in classifying students as SLD. Private schools
may be particularly willing to accept students
who suffer from little or no neurological obstacle
to high achievement later on.
In contrast, we might expect that students
with particularly severe or rare disabilities
have fewer private school alternatives, even
though the McKay voucher grant is larger for
students with more severe disabilities. Serving
students with certain disabilities could require
an original fixed-cost investment in facilities
that few private schools are willing to make,
leaving the more severely disabled students
with fewer educational options under McKay than
students with milder disabilities. It is, however,
important to note that previous research (Greene
and Forster 2003) finds that the distribution
of disabilities in the McKay program resembles
the distribution in public schools.
Students’ differential access to private
schooling for the reasons offered above should
also give public schools different levels of
exposure to the McKay program. By distinguishing
each student’s particular disability classification,
our data set allows us to determine whether
the impact on the academic achievement of disabled
students of the addition of a McKay-accepting
private school near their public school varies
according to the nature of their disabilities.
4. Method
We evaluate whether students in schools with
greater exposure to McKay vouchers—measured
by the number of private schools within a certain
radius of their public school that are willing
to accept the vouchers—made larger or smaller
academic gains in math and reading than students
in schools that faced less exposure to the program.
Our analysis also measures the impact of differences
in exposure to McKay on students in each disability
category.
We use the panel data set that includes information
on the universe of public school students in
Florida, whether or not they are disabled. The
dependent variable in the analyses—that
is, the outcome we are evaluating—is the
student’s test score on the state’s
mandated math or reading exam.
We utilize this expansive data set to estimate
student proficiency in math and reading over
time. In essence, through the use of a so-called
fixed-effects regression model, we can control
for factors that influence a student’s
academic proficiency but are unobserved by the
researcher—for example, parental involvement,
nutrition, and wealth. The regression also statistically
controls for observable factors important to
student proficiency, such as the school district
the child attends, the year of the observation,
grade level, and observable student demographic
characteristics that vary over time, such as
whether the student is eligible for a free or
reduced-price lunch or is identified as having
limited English proficiency.
We do more than measure the overall effect
on the performance of the average student (disabled
or not disabled) of exposure to McKay: we also
measure the differential impact on students
within each particular disability classification.
We do so by interacting the child’s disability
classification with the number of private schools
within a given radius of his public school.
These variables are an estimation of the average
effect of increased exposure to the McKay program
on the academic proficiency of students in each
particular disability classification.[6]
We first estimate the effect by using as our
exposure variable the number of private schools
accepting McKay vouchers within a five-mile
radius of a given public school in a particular
year. In order to test for the robustness of
our procedure, we then replace this number with
the number of private schools accepting McKay
vouchers within a ten-mile radius of the same
public school. In order to establish the validity
of a five-mile compass, we look for the impact
of private schools that are farther away. We
would expect that the effect of including more
distant private schools would be smaller than
the impact of the presence of more proximate
private schools on public school performance.
5. Results
The results of our estimation for student proficiency
in math are found in Table 4. As will
be the case throughout, for reasons of space
we report estimates for only the interaction
of the diagnosis and competition variables.[7]
Our variables of interest are the interaction
terms, which evaluate the differential effect
of McKay exposure on students with different
diagnoses. McKay exposure and special-education
students’ performance in light of their
diagnoses. The variables are sorted in the table
according to the size of the student population
in the special-education category, which is
imperfectly related to the severity of the diagnosis.
The estimate on each of the interaction variables
is positive, and most estimates (especially
those for the less severe categories) are significantly
different from (that is,
almost certainly larger than) zero.
The results are similar for both the within-five-miles
and the within-ten-miles specifications. The
primary difference in the analyses appears to
be that the estimates of the within-five-miles
analysis are uniformly larger than the estimates
of the within-ten-miles specification, which
is as we would expect if students are less likely
to enroll in McKay schools farther away from
their homes than the public school in which
they were previously enrolled.
Table 5 reports the results of our estimation
for reading. The results are similar to those
found in math, though the impact of McKay exposure
on the reading scores of most categories of
disabled students appears to be larger than
it was on math scores. We again see a positive
relationship between McKay exposure and academic
proficiency in most of the diagnostic categories,
especially the milder disabilities. As in math,
the relationship uniformly declines as we expand
the exposure variable to include schools within
ten miles.
Table 6 puts our results of the within-five-miles
estimation into a more manageable context. The
table first reports the effect of McKay exposure
on students within each classification in a
school with the average number of nearby private
schools accepting vouchers in 2004–05.
The second column reports the overall effect
on the test scores of a school with average
McKay exposure by incorporating the estimate
for both the overall effect of McKay exposure
and the interaction term for a particular group.
The third column states the overall effect on
students in a school with average McKay exposure
as standard deviation units.
The table shows that the effect on students
in a school with average McKay exposure differed
by disability classification. The effect for
students identified as having a Specific Learning
Disability, who constitute about 8.5 percent
of all students in Florida and 61 percent of
students in special education (see Table
2), was about a 0.05 standard deviation
increase in math and a 0.07 standard deviation
increase in reading.
6. Conclusion
This paper adds to an important and growing
literature evaluating the impact of schoolchoice
policies on the performance of public schools.
Our results from evaluating Florida’s McKay
program provide additional evidence that rather
than being harmed, public schools respond to
the challenge of exposure to school choice by
improving the education they provide. These
findings are consistent with most previous research,
which demonstrates school-choice policies’
positive effect on public school achievement.
More specifically, this paper has provided
the first quantitative evaluation of the impact
of a voucher program on disabled students. Such
research is of growing importance, given the
substantial growth in school-choice programs
aimed at this particular disadvantaged population.
These initial results support the use of special-education
vouchers to improve the educational outcomes
of disabled students within public schools.
Much more quantitative research on these increasingly
important policies is necessary to provide a
fuller understanding of their effects both on
students who remain in public schools and on
students who use such vouchers to attend private
schools.
