Charter School Entry in Market Equilibrium
Dennis Epple
Carnegie Mellon University
Maria Marta Ferreyra
Carnegie Mellon University
Brett R. Gordon
Columbia University
November 9, 20101
Preliminary and Incomplete
Please do not cite without permission
Abstract
We develop and estimate a structural model of charter school entry. In this model, charter schools offer an
alternative to neighborhood public schools. In the model’s first stage, the potential entrant chooses his/her
physical location and characteristics based on the perceived opportunities not exploited by conventional
public schools. In the second stage, students choose among the available schools. We estimate the model
using data for the DC school district. We use our parameter estimates to investigate the potential effects
of changes in charter school legislation on charter school entry, and on post-entry outcomes such as
enrollment and achievement.
1
We thank Liran Einav, Justine Hastings, Susanna Loeb, Marc Rysman, and Alan Sorensen for useful
conversations. We also benefitted from comments by session participants at the 2009 LACEA conference. We thank
Gary Livacari, Sivie Naimer, Hon Ming Quek, Jeremy Whyte and especially Nick DeAngelis for their research
assistance. Bill Buckingham from the Applied Population Lab at the University of Wisconsin provided Arc GIS
assistance. All errors are ours.
1
1.
Introduction
The dismal academic performance of many students in central city school districts has been a
source of growing concern in recent decades as academic achievement and high school
completion have become increasingly important for economic success. Fairly or not, poor
student performance is often attributed to ineffective urban public schools. Greater school
competition has been advocated by many economists as a mechanism to foster improved
achievement, with educational vouchers being the most prominent such proposal. While
educational voucher programs have made significant inroads in some states and districts,
voucher initiatives have faced intense opposition from teachers’ organizations and have had
limited success at the ballot box. By contrast, charter schools have received surprisingly
widespread acceptance.
Charter schools are nonsectarian public schools of choice that operate with freedom from
many of the regulations that apply to traditional public schools. They are regulated by state laws,
and authorized by a chartering entity which approves the school’s charter – the contract that
establishes the school's mission, program, goals, students served, assessment methods and
teaching philosophy. In exchange for greater autonomy, charter schools are subject to the same
accountability requirements as ordinary public schools. They receive a per-student stipend yet
they rarely receive funding for facilities. Hence, the financial survival of charter schools is
intimately connected with their ability to attract and retain students. While some charter schools
have thrived and replicated their model across the nation, others have failed.
The charter school concept emerged in the late 1980’s. It was embraced at its inception
by Albert Shanker, then president of the American Federation of Teachers. Beginning with
adoption by state of Minnesota in 1991, charter school laws have been adopted by more than 40
states and the District of Columbia. As of 2009, there were about 4,900 charter schools in the
country serving over 1.5 million students, or approximately 4% of the nation’s primary education
market (National Alliance for Public Charter Schools, 2010).
While charter schools do not enjoy the latitude envisioned by many voucher advocates,
charter schools nonetheless have considerable freedom in designing curricula and in managing
their operations. Charter schools control hiring of their teachers and staff, and they operate
2
independently of the public school districts in which they locate. Indeed, when school districts
report public school enrollment, they typically do not count enrollment of students in the district
who are enrolled in charter schools. In many states, school districts do not possess the authority
to grant a charter to a school. Instead, a separate chartering agency at the state or local level is
charged with deciding whether an application for a charter is granted. In most instances, this
chartering agency also exercises oversight of schools that receive charters and has the power to
rescind a charter.
Charter schools share with public schools the requirement that they be tuition free. Unlike
public schools, which are typically subject to residence requirements and can only enroll children
who live in their attendance zone, charter schools must be open to all students. When
oversubscribed, a charter school is typically required to select by lottery from its applicant pool.
In other words, charters must compete to attract students, but cannot select among them.
In order to open a charter school, a prospective charter operator must formulate and
present a proposal to the chartering agency. This proposal sets forth the grade level(s) to be
served and the thematic focus of the school. The latter includes a statement of subject emphasis
(e.g., arts, language, math and science, core curriculum) and whether a special population is to be
served (e.g., bilingual, disadvantaged, gifted). The proposal also includes a statement of and
whether a particular instructional model is to be implemented. For example, the applicant might
propose to adopt an instructional model that is “franchised” nationwide, or a model that is being
developed by the charter-school applicant. The proposal also includes anticipated enrollment, the
physical facilities to be occupied, and a financial plan. In many ways, then, the charter applicant
confronts a decision similar to that of a private firm contemplating entry to provide a
differentiated product or service. The applicant must choose the form of differentiation of the
planned product or service. In addition, the applicant must forecast demand, identify suitable
facilities, and assess financial viability. If it decides to enter, the applicant must manage the
resulting enterprise and make an ongoing decision about whether to continue operation or exit.
In this paper we develop and estimate an equilibrium model of household school choice
and charter school entry in a large urban school district. We seek to understand what determines
charter entry and choice of location and program offering. This, in turn, allows us to study how
charters affect the school choices available to families, what kind of students are attracted to
3
them in any given location, and what specific elements determine the attractiveness of charters
relative to public schools. Moreover, our approach allows us to explore the consequences of
changes in the regulatory or institutional landscape of public and charter schools, an issue that
seems particularly timely given the current emphasis on charter schools in federal education
policy.2
In our model, prospective charter school entrants choose a geographic location, thematic
focus, and grade levels to serve. Since charters must be able to forecast the demand for their
services in order to assess their viability, we model how charters predict enrollment and
demographic composition of their student body as a function of their geographic location, grades
served, and thematic focus. This forecasting model views parental school choice decisions as an
input to modeling the decision of the prospective charter entrant. We model the prospective
entrant as making the choices that maximize its likelihood of successful entry and subsequent
viability, which in our framework means maximizing expected net revenue. Since the actual
managing of the enterprise may be more involved than expected by the entrant when submitting
its application, we model the entrant as being uncertain about its own quality at the entry stage.
We estimate the model using a unique and detailed data set from Washington D.C. from
2003 to 2007. The data set consists of information for all public and charter schools in the
Washington DC school district, including enrollment by grade, school demographics, focus, and
achievement. Our estimation exploits the variation across neighborhoods and over time in
demographics, number and types of schools, and enrollment and achievement. We treat each
grade as a separate market, which has the benefit of increasing the effective sample size and
generating exogenous variation in choice sets across households depending on the age of their
children. We cast the parental school model as an aggregate market share model and estimate it
using an approach similar to Berry, Levinsohn, and Pakes (1995). In the model, school
characteristics such as the peer composition of the student body and aggregate achievement
The federal “Race to the Top” program favors state with permissive charter legislation. See
http://www2.ed.gov/news/pressreleases/2009/06/06082009a.html for further details.
2
4
depend on households’ choices and their equilibrium sorting across schools. Our estimation
strategy captures this aspect,3 which is critical to our ability to conduct counterfactuals.
In modeling the charter school entry decision, we have chosen to focus on a single large
urban district. This permits us to study the entry decisions of prospective entrants all of whom
confront the same institutional structure, criteria for being granted a charter, reimbursement
(funding) policies, and public school organization. Thus, we hold constant institutional structure
in order to focus on the charter entrant and parental choice decision. We have chosen to focus on
the Washington, D.C. school district, for the following reasons. First, the district has had a
charter school law in place since 1996, and the law is highly permissive compared to charter
school laws in other states (National Alliance for Public Charter Schools 2010). For instance, the
per-student stipend for charter schools is equal to the full per-student spending in the
Washington, D.C. school district, whereas it is equal to only a fraction of the corresponding
district spending in most states. Moreover, in recent years a single entity, the Public Charter
School Board (PCSB) has been responsible for the granting of charters. Second, in 2008 charter
schools achieved a student enrollment share of nearly 30 percent of all students in the public
system (conventional public plus charter schools), one of the largest in the country. Third, the
district is large, providing scope for potential entry of a large number of charter schools as well
as substantial intra-district variation in population demographics across both residential locations
and incumbent public schools. An analysis that includes multiple districts is challenging for a
number of reasons—one being the challenge of obtaining a sufficiently complete data set—and
we leave the pursuit of such a model to future research.
Charter school entry has been explored before by Glomm, Harris, and Lo (2005), who
study charter school location in Michigan and California. They investigate the extent to which
there is greater charter school entry where there is greater population heterogeneity. Using public
school districts as the unit of observation, they find, as predicted, that there is more charter
school entry where there is greater diversity in race and in adult education levels. In earlier work,
Downes and Greenstein (1996) study private school entry across the 910 public school
3
Our endogenous school characteristics are similar to the local spillovers in Bayer and Timmins (2007). See
Ferreyra (2008) for a structural model that endogenizes school demographics and achievement in full-solution
estimation.
5
elementary districts in California. Rincke (2007) estimates a model of charter school diffusion in
California. A theoretical model of charter school entry is developed by Cardon (2003), who
studies strategic quality choice of a charter entrant facing an incumbent public school. We build
on the foundation established in these papers by modeling intra-district charter school entry
decisions, parental choice, and the impact of entrants on public school incumbents. In addition to
research on charter school entry, there is also an extensive body of research on the effects of
charter schools on achievement of charter and public school students.4 Perhaps closest to our
approach is the work of Imberman (2009) who studies a single large urban district.
Our work is also related to research on entry in empirical industrial organization. A
recent review of this literature is provided in Draganska, et. al. (2008). An important difference
between our model and those found in the entry literature is that we do not model prospective
entrants as engaging in a game of strategic behavior. The assumption of non-strategic charter
school entrants is consistent with our understanding of the charter schools’ decision processes.
Draganska, Mazzeo, and Seim (2009) incorporate a model of demand into a product entry model,
permitting them to estimate a fully structural profit function. Our approach is similar in that we
model the demand side of the market rather than specify a reduced-form profit function for
schools. Interestingly, while the entry literature in Industrial Organization often does not
examine post-entry outcomes, our data on achievement and enrollment enables us to explore the
consequences of the re-sorting induced by charter entry. This question is ultimately of great
interest to the policy-maker.
We use our parameter estimates to study the effect of changes in the regulatory,
institutional and demographic environment on charter entry, household sorting across schools
and student achievement. For instance, we explore whether more permissive charter legislation
would spur the creation of more charter schools, where these would locate, which students would
attend them, and how achievement would be affected among the pre-existing schools.
4
See, for instance, Bettinger (2005) Bifulco and Ladd (2006), Booker, Gilpatric, Gronberg, and Jansen (2007,
2008), Buddin and Zimmer (2005a, 2005b), Clark (200), Hanushek, Kain, Rivkin, and Branch (2007), Holmes
(2003), Hoxby (2004), Hoxby and Rockoff (2004a, 2004b), Hoxby and Murarka (2009), Imberman (2009,
forthcoming), Sass (2006), Weiher andTedin (2002), and Zimmer and Budin (2003).
6
The rest of the paper proceeds as follows. Section 2 describes our data sources key
patterns in the data. Section 3 presents our theoretical model. Section 4 describes our estimation
strategy. In Section 5 we describe our intended counterfactuals. Section 6 concludes.
2.
Data
In this section we describe the construction of our dataset and some salient aspects of the data.5
Our data set consists of information on every public and charter school in Washington, D.C.
between 2003 and 2007. We have focused on this time period to maximize the quality and
comparability of the data over time and across schools. In addition, 2007 marked the beginning
of some important changes in DCPS and hence constitutes a good endpoint for our window of
study.6
2.1
Public and Charter Schools in Washington D.C.
In Washington, D.C., public schools fall under the supervision of the District of Columbia Public
Schools (DCPS). As for charters, until 2007 there were two authorizers: the Board of Education
(BOE) and PCSB. Since 2007, the PCSB has been the only authorizing (and supervising) entity.
An overarching institution at the “state” level is the Office of State Superintendent of Education
(OSSE), which, among other roles, collects some data both for public and charter schools.
The unit of observation in our dataset is a campus-year (public schools have one campus
each, but many charters have multiple campuses).7 Note that student-level data are not publicly
available. We have 701 observations for public schools, and 230 for charters. For each school
and year we observe address; enrollment by grade;8 percent of White, Black, Hispanic and other5
Readers interested in further detail are invited to consult Appendix I.
6
In 2007, Michelle Rhee began her tenure as chancellor of DCPS. She implemented a number of reforms, such as
closing and merging schools, offering special programs and changing grade configurations in some schools, etc.
7
A campus is identified by its name and not its geographic location. For instance, if a campus moves but retains its
name, then it is still considered the same campus. In multi-campus charters, each campus has a separate address.
8
We do not include adult or ungraded students, who account for less than 0.6% of the enrollment in this dataset.
7
ethnicity students; percent of low-income students (i.e., students who qualify for free or reduced
lunch); reading proficiency rate (i.e., the percent of students who are proficient in reading based
on DC’s own standards and assessments), and math proficiency rate. Proficiency rates constitute
our measures of achievement.
For public schools we also observe the percent of students who reside outside the
school’s attendance zone (out-of-boundary students). For the few public schools that closed or
merged during the sample period, we observe the year of closing or merger.
In addition to the variables listed above, for charter schools we observe the authorizer,
opening year and mission statement. For the charters that closed or merged in or after 2003, we
observe the year of and reason for closing or merging.
The starting point for our dataset is the list of audited enrollments by school, year and
grade from OSSE. This list includes public and charter schools. From this list we eliminated
alternative and special education schools, early childhood centers, and schools that offered a
residential program.9 Since OSSE also reports proficiency rates for each school and year as
mandated by No Child Left Behind, our achievement data for public and charter schools come
from OSSE. In 03 and 04, proficiency levels were determined according to the Stanford-9
assessment. To be considered proficient, a student was supposed to score at the national 40th
percentile or higher. Since 05, proficiency has been determined according to DC Comprehensive
Assessment System. Before 2005, the grades tested were 3, 5, 8 and 10. Beginning in 2005, the
grades tested have been 3 through 8 and 10. In compliance with NCLB, OSSE also reports the
percent of White, Black, Hispanic and low-income students in the grades tested.
For public schools, the Common Core of Data (CCD) from the National Center for
Education Statistics is the main source for school address, ethnic composition and low-income
status of the student body. If needed, we relied on OSSE’s reporting of ethnic composition and
low-income status. In addition, we drew on the Common Core of Data, web searches and phone
calls to DCPS to identify year of closing or merger. DCPS provided us with the percent of outof-boundary enrollment.
9
An alternative school is a school for children with behavioral problems. Children spend either the full year or part
of it there. Sometimes they can choose to attend those schools, but they are often sent there. Moreover, these schools
often enroll students aged 16-24. We excluded early childhood centers as long as they never had enrollment in
regular grades during the sample period.
8
The fact that some charter schools have multiple campuses posed a major challenge in the
construction of the dataset, as no single data source contains the complete or correct history of
campuses. To re-construct this history, we drew on OSSE audited enrollments, School
Performance Reports (SPR’s) for PCSB-authorized charters, web searches of current websites
and past Internet archives, charter school lists from Friends of Choice in Urban Schools
(FOCUS), phone calls to charters that are still open, and achievement data at the campus level.
The resulting campus-level data reflect our efforts to piece together campus-level information
from these different sources, with the greatest weight being given to the OSSE audited
enrollments, the SPRs and the NCLB achievement data. BOE charters that are no longer open
represented a special challenge, as there is no institutional memory of these schools. Faced with
conflicting information on campuses for these charters, in the absence of better evidence we took
the most recent available addresses and campus configuration data as correct.
For PCSB-authorized charters, ethnic composition and low-income status come mostly
from the School Performance Reports. For BOE-authorized charters, before 2007 this
information comes from OSSE, and for 2007 from the School Performance Reports. If needed
and possible, we supplemented these sources with the CCD.
We identified the year of opening based on SPR’s, FOCUS listings and web searches. We
identified the year of and reason for closing mostly based on reports from the Center for
Education Reform and SPR’s.
According to grades covered, we have classified schools into the following levels:
elementary (if grades covered fall completely within the preschool-6th grade range), middle (if
grades covered are 7th and 8th), high (if grades covered fall completely within the 9th - 12th grade
range), elementary/middle (if grades span both the elementary and middle level), middle/high (if
grades span both the middle and high level), and elementary/middle/high (if grades span the
elementary, middle and high levels). This classification follows DCPS’s criteria for identifying
single-level categories (elementary, middle and high schools), and incorporates mixed-level
categories, which are quite common among charters.
Each charter school has a statement that describes its mission or program focus. Mission
statements come from the schools’ web sites, SPR’s, and FOCUS. Based on them we have
classified thematic foci into the following six categories: arts, civics, core curriculum
9
(henceforth, core), language, math and science, and vocational. Appendix II lists the charter
schools in our sample, their statements and the focus we assigned to each.
2.2
Descriptive Statistics
The population in Washington, DC peaked in the 1950s at about 802,000, and declined steadily
until 2000, when population was approximately equal to 572,000. The last ten years have
witnessed some population gains, as the estimated 2009 population is approximately equal to
600,000. In particular, the population is estimated to have grown from 577,000 in 2003 to
588,000 in 2007.10 The racial breakdown of the city has changed as well over the last two
decades, going from 32, 66 and 5 percent White, Black and Hispanic in 1990 to 40, 55 and 8
percent respectively in 2007.11 Despite these changes, the city remains highly segregated by race
and income. In 2006, median household income was $92,000 for whites, $34,500 for blacks (21st
Century School Fund et al, 2008).
2.2.1. Basic trends
In 2007, 57 percent of students attended public schools, 22 percent charter schools, and 22
percent private schools.12 Though the private school enrollment share is substantial, in this draft
we focus on public and charter schools exclusively. We will include private schools in future
drafts. Hence, in what follows “total enrollment” refers to the aggregate between public and
charter schools (i.e., total enrollment in the public system), and shares are calculated relative to
this total.
10
Source: Population Division, U.S. Census Bureau. Estimates released on March 2009.
These percents do not add to 100 because “Hispanic” is an origin and not a race. Hence, a “Hispanic” person may
be counted as black or white from the point of view of ethnicity, and as having a Hispanic origin from the point of
view of origin.
11
12
Own calculations, based on data from OSSE and the Private School Survey. Our analysis of the Private School
Survey data between 03 and 07 indicates that the number of children enrolled in private schools remained roughly
constant during this period.
10
In national assessments, DC public schools have ranked consistently at the bottom of the
nation in recent years. For instance, in 2009 the percent of 4th or 8th graders who are at or above
the proficiency level in reading according to the National Assessment for Educational Progress
was higher in all 50 states than in DC (similarly for math). Perhaps this explains why charter
schools have grown rapidly in DC since their inception in 1996. During our sample period alone,
the number of charter school campuses almost doubled, going from 31 to 59 (see the top line in
Figure 1), whereas the number of public school campuses declined slightly as a result of a few
closings and mergers (see Table 2). Even though total enrollment in the public system has
declined over time,13 enrollment in charter schools has risen (see Figure 2), and the enrollment
share of charters has risen from 16 to 28 percent (see Figure 3).
As Table 1 shows, Blacks and Hispanics account for almost all the enrollment in the
public system, although charters tend to enroll proportionately more Blacks and fewer Hispanics
than public schools. They also tend to enroll a higher fraction of low-income students. Moreover,
they tend to locate in neighborhoods (Census tracts) with lower average household income (see,
also, Figure 4). On average, achievement is very similar in public and charter schools. Note that
public schools are highly heterogeneous, as are charters.
2.2.2. Variation by grade level and focus
As column (1) Table 3 shows, about two thirds of public schools are elementary. In contrast, 43
percent of charter campuses are elementary. Furthermore, while only 4 percent of public schools
mix levels, about a third of charter campuses do. In general, elementary schools are smaller in
terms of enrollment than middle schools, which are in turn smaller than high schools.
Nonetheless, at every level charters are smaller than public schools.
During the sample period, 55 percent of charter enrollment corresponds to preschool
through 6th grade, 19 percent to grades 7 and 8, and 26 percent to grades 9 through 12. Figure 3
breaks the charter market share down by these three categories. Importantly, the charter share
13
The population of children aged 5-17 declined by about 5,000 between 2003 and 2007 according to estimates
from the U.S. Census Bureau (estimates released on March 2009). Since the number of children enrolled in private
schools has remained roughly constant over this period (own calculations based on the Private School Survey), the
enrollment decline in the public system is most likely due to the decline in school-aged population.
11
differs greatly among individual grades (see Figure 5), a feature of our data which we intend to
exploit in the estimation. It peaks for 7th grade, followed by 6th and 8th grade. Among the
remaining grades, there is a peak for kindergarten. During our sample period, market share has
grown the most for early childhood and middle school grades.
It is quite common for charter schools to add grades over time until completing the grade
coverage stated in the charter. For instance, a school’s charter may state that in a few years the
school will offer kindergarten through 5th grade, yet the school may open with kindergarten and
first grade, add second grade the following year, third grade the year after, and so on. In fact, of
the 30 campuses that entered between 2003 and 2006, 26 added grades during the sample period.
While most additions were at the top (in the example above, adding second grade the second
year of operations), some of them were at the bottom (as it would be the case if in the example
above, the school added prekindergarten).
The variation in charter share across grades raises the questions of whether the student
body varies across grade levels. The answer to this question is yes, as Table 4 shows. In general,
middle and high schools have higher concentrations of black and Hispanic students than
elementary schools yet lower fractions of low-income students. Reading and Math proficiency
decline with the school level. Middle and high schools, in turn, tend to be located in Census
tracts with higher average income than elementary or elementary/middle schools. While charters
replicate some of these broad patterns, they also differ in that a full 72 percent of high-school
students are low-income (as opposed to only 58 percent among public school students).
Interestingly, whereas public schools surpass charters in reading and math achievement at the
elementary level, the reverse happens at the other levels. The gap between public and charter
proficiency is greatest at the middle school level – while approximately 40 percent of middle
school students are proficient in public schools, more than 50 percent are in charter schools.
Moreover, charters achieve their highest proficiency with middle- and elementary/middle-school
students.
As column (1) of Table 5 shows, over the sample period approximately 60 percent of the
campuses offered a core curriculum. Among the remaining campuses, the most popular foci are
language, arts and vocational. However, this pattern of program specialization differs by level.
While 56 and 74 percent of elementary and middle schools offer a core curriculum respectively,
12
only 21 percent of high schools do. Among the mixed-level charters, elementary/middle and
elementary/middle/high charters are almost exclusively devoted to a core curriculum, but only 29
percent of middle/high schools offer one. Moreover, language and arts are the more popular foci
after core among elementary schools, yet vocational and civics play this role among high
schools.
As expected, different foci attract different students (see Table 6). For instance, language
charters (usually focused on Spanish) tend to attract more Hispanic students. At the elementary
and middle school level, math and science charters attract lower fractions of low-income students
than the other foci, though the reverse happens at the high school level. At the high school level,
vocational charters have the highest fraction of low-income students. At the middle and high
school level, students in math and science charters outperform students in other charters both in
math and reading. If one excludes math and science, the highest average achievement
corresponds to core schools at the elementary and middle-school level, and both to core and
civics charters at the high school level.
While average achievement is very similar for public and charter schools, the topperforming public and charter schools are quite different (see Table 9). The top-performing
charter schools are located in much poorer Census tracts, enroll poorer students and are chosen
by much larger fractions of African-American students than the top-performing public schools.
Interestingly, they offer a core curriculum.
2.2.3. Relocations, closings and multiple-campus charters
A few campuses have relocated during our sample period (see Table 7). According to the SPR’s,
most of these campuses started out in a temporary location while they were renovating or
preparing their permanent facilities, and then moved there. The median distance move is 1.91
miles. Nonetheless, charters that plan to relocate usually inform parents of their plans ahead of
time, so that the families can plan accordingly.
Furthermore, six campuses (and four schools) closed during our sample period, and six
additional campuses have closed since 2008. Of the closings that took place during our sample
period, four were due to mismanagement and correspond to schools authorized by the Board of
13
Education. The two other closings were due to academic reasons and correspond to schools
authorized by PCSB.
As Table 8 shows, our sample includes 45 schools and 63 campuses. In general, multicampus charters run multiple campuses for the sake of serving different grade levels. For
instance, Friendship has two elementary school campuses (Southeast Academy and
Chamberlain), one elementary/middle school campus (Woodridge), one middle school campus
(Blow-Pierce), and one high school campus (College). The 10 multi-campus schools in our
sample accounted for by 26 campuses and 46 percent of the charter enrollment during the sample
period. Multi-campus schools tend to focus more on a core curriculum (70 percent of them do, v.
54 percent among single-campus schools) and to enroll slightly more African-Americans (89
percent of their students are African-American, relative to 82 percent among single-campus
schools). A slightly higher fraction of their students is proficient in Reading or Math (39 and 37
percent in single-campus charters v. 40 and 40 percent respectively in multi-campus charters).
2.2.4. Early v. recent entrants
Given our focus on charter entry, an important question is whether the 27 campuses that entered
before our sample period (“early entrants”) are systematically different from the 37 campuses
that entered during our sample period (“recent entrants”). As Table 10 shows, these two groups
are indeed different along a number of dimensions. Recent entrants tend to be smaller and more
concentrated at the elementary or middle school level. They are also more likely to belong to a
multi-campus charter organization. They enroll lower fractions of blacks and low-income
students, and their achievement is lower on average. They are also less likely to offer a core
curriculum.
To summarize, our data set is unique and draws from a variety of sources. It has not been
compiled or used by any other researcher before. Moreover, it features a rich variation over time
and across schools that will help us identify the parameters of our model. In the next section we
formulate the model.
14
3.
Model
In this section we develop our model of charter school entry and equilibrium. In this model, the
economy is Washington, D.C. We divide the economy into neighborhoods. Each neighborhood
has either a charter or a prospective charter school entrant. An existing charter school makes a
decision of whether to continue operation or exit. A prospective charter school entrant chooses
which grades to serve and the best thematic focus for the location in case of entering, and then
decides whether to enter. There is a discrete set of potential grade levels (e.g., elementary,
middle, etc.) and thematic foci (e.g., math and science, arts, etc.) from which the entrant can
choose. The prospective entrant forecasts demand and revenue under each configuration and
enters if the expected net revenue is positive.14 Public schools in the model are passive;15 the
prospective entrant takes the locations, grades served, and thematic foci of public schools as
given.
Households must choose among the public schools available given its residential
location, and the charter schools (all of which are available, since charter schools have no
residence requirements).16 For a given location, the prospective entrant forecasts its revenue
based on a model of household choice that predicts the clientele that would be attracted with
each potential thematic focus in competition with the incumbent public schools in that location.
The prospective entrant confronts uncertainty about its own quality in the form of an unknown
shock that will affect demand if it should enter.
The model thus has two stages, an entry stage and a household choice stage. Since the
latter is used in the former, we begin by presenting the model of household choice of school. Our
school choice framework follows closely the model of Bayer and Timmins (2007).
14
A potential extension is to have a charter school authorizer that imposes a minimum threshold for net revenue to
internalize the externalities that a failing school imposes on its students.
15
Between 1992 and 2007, DCPS had seven superintendents. This high turnover, coupled with financial difficulties
and instability, suggest that DPCS did not react strategically to charters during our sample period.
16
At the moment we are not modeling out-of-boundary enrollment in public schools, though we will do so in future
versions. During our sample period, approximately 18 percent of public school students attend and out-of-boundary
school.
15
3.1
School Choice and Market Equilibrium
A student i is described by the tuple ( g , d , ) , where

g 1, , G is the grade of the student, such that students in grade g only choose from
schools offering the corresponding grade.

d d1 , , d K  is a collection of vectors to describe a set of K discrete demographics,
taking values of 0 everywhere except a 1 in the place corresponding to a particular
demographic. Our data contain ethnicity and poverty status.

1, , L is the location of the household in one of the L possible areas of the school
district. In our empirical application, each location corresponds to a census block group.
Let j and t subscripts denote respectively a school and year. A student’s school choice depends
on several variables:

X j denotes school characteristics that result from a school’s decisions, such as the type
of thematic focus used in the curriculum. Henceforth, for presentational clarify, we will
refer to X j as focus.

Yij reflects characteristics of the match between the household and a school, such as the
distance from the household to the school, and whether the household is “in-boundary” or
“out-of-boundary” for a particular school.

 jt represents endogenous student peer characteristics of the student body at the school
that are a result of aggregate household choices, such as the demographic composition of
a school. These peer characteristics naturally change over time as students’ school
choices change.

 ijgt an idiosyncratic utility shock.
While both X j and  jt represent school-specific characteristics, an important distinction is that
X j is chosen by the school, whereas  jt results from household choices. The household utility
function is then:
16
1. U ijgt   jgt  ijgt   ijgt
where  jgt is the mean utility over households with children eligible for that school and grade
and μijgt is a household-specific deviation from the mean school-grade utility. In what follows we
discuss each of these components.
The mean utility component of a school and grade is common to all students, and depends
on school and peer characteristics as follows:
2.
 jgt  Q jt1  X j 1'   jt1'   'jgt
Here, 1 is a scalar parameter and 1 and 1 are vectors of parameters, and  jgt is a demand
shock for school j grade g at date t. Variable Q jt is school quality, determined as follows: 
3. Qjt  X j 1"   jt1"  X j jt    "j   "jt
In this equation, school quality is a function of school characteristics (focus X j and peer
composition  jt ), their interaction, and school productivity shocks  j and  jt (note that  j is a
time-invariant school shock, and  jt is a productivity shock for school j at date t).
The
interaction of X and  captures the fact that a particular thematic focus may be more suitable to
some students than others.17 We can think of school quality as the expected achievement for
children in the school. In our empirical application we take the average of math and reading
proficiency rates to proxy for a school’s quality.
Substituting (3) into (2), we obtain:
4.
 jgt  X j 1   jt1  X j jt    jgt
where the coefficients can be expressed as:
1  1'  1"
1  1'  1"
'
 jgt   jgt
 1  "jt   "j 
Hence  jgt impounds both a demand and a productivity shock. In a particular period  jgt is the
sum of a time-invariant mean school shock (e.g., the mean quality of the teachers across all
17
It is possible to derive equation (3) by starting from an individual-level model of achievement (where achievement
is a function of individual characteristics, school focus and demographics, school productivity shock, the interaction
between individual characteristics and school focus and demographics, and an individual idiosyncratic shock), then
aggregating it up to the school level and taking expectations of the school-level aggregate. Detailed notes are
available from the authors.
17
grades), a school-time specific deviation from this mean (e.g., differences over time in the
quality of the facilities), and a school-time-grade deviation (e.g., differences from year-to-year in
the quality of teachers across grades). Note that the peer characteristics enter utility at the school
level as opposed to the grade level. We do not expect significant variation to exist within a
school across grades over basic demographic variables, and our data lack information on such
characteristics by grade level.
Parents observe  jgt at the time they make their school choice decisions. However,
potential entrants do not observe it at the time of their entry decision. Thus, a prospective entrant
commits to the process of opening a school in the face of substantial uncertainty and well before
any parents make school choice decisions. However, by the time the entrant has been authorized
to open and begins to recruit students, much uncertainty has been resolved. For instance, new
charters conduct aggressive neighborhood searches, community meetings with parents, open
houses, etc., in order to advertise their services and recruit students. Hence, when parents choose
a school for their children they have much better information on the school’s chances for success
than the prospective entrant did at the entry stage. This asymmetry results from the timing of
decisions, as parents make their school choice after the entrant decides whether to enter. We
capture the asymmetry by assuming that the entrant does not observe  jgt at the entry stage, yet
households and the entrant observe it at the households’ school choice stage.
Given suitable instruments, the parameters of (3) can be estimated by a school fixedeffects regression of proficiency rates on school demographics, focus and interactions between
demographics and focus. This regression estimates the productivity effects of school
characteristics (including the shocks). In addition, as we explain in Section 4, by relying on
market shares for each school, grade and time we can estimate the parameters 1 and 1 in (4).
These parameters impound both taste and productivity effects of peer and school characteristics
respectively. Parameter 1, which measures the weight on school quality relative to preferences
for school characteristics, is not identified. We note that the identification of peer effects in (4)
rests heavily on the assumption that there are no unmeasured peer characteristics.
Next we discuss the household-specific component of utility:
5. ijt  Yij  0  di X j  0  di jt 0
18
Note that the interaction of X j and d i captures the variation in attractiveness of a school’s focus
across students of different demographic groups. The interaction of d i and  jt captures the
potential variation in preferences for school peer characteristics across different demographic
groups.
Note that the setup above captures an interesting potential conflict between school
characteristics that enhance productivity and school characteristics that attract students. For
example, a long school day may enhance achievement, but parents and students may not like the
longer day. The former would be captured in (3). The latter would be captured by 1 in (2) and
possibly 0 in (5).
With a slight abuse of notation, let X , Yi ,  t and  t denote concatenation of the
corresponding school-specific vectors X j , Yij , jt, and jgt, and let D be the collection of student
demographic variables. Let J t denote the set of schools available to a given household in period
t and let J tg denote the subset of schools offering grade g. A household chooses from among the
set of schools J tg that are relevant depending on the age of the student to maximize its utility:
arg max U ijgt
jJ tg
Assuming the idiosyncratic error terms in equation (1) are i.i.d. type I extreme value, we can
integrate over these shocks to form the probability that household i chooses school j in grade g at
date t:
Pijgt  X , Yi , di ,  t , t ;  
exp  jgt  ijgt 
J tg
1   exp  kgt  ikgt 
k 1
Let h( d , ) be the joint distribution of students over demographics and locations, and h(d , | g )
be the joint distribution conditional on a particular grade. Aggregating over all households yields
equilibrium enrollment shares, sjgt, and equilibrium values of each of the peer characteristic, kjt,
19
in each school j for each date t:
s jgt   Pijgt ( X , Yi , di ,  t , t ,  it )dh( d , | g )
 
k
jt
d
ik
Pijgt ( X , Yi , di ,  t , t ,  it )dh(dik , | g )
g,

Pijgt ( X , Yi , di ,  t , t ,  it )dh(dik , | g )
, k .
g , dik ,
where dik is an indicator that a household belongs to a particular demographic. We will discuss
how we match the empirical counterparts of the above quantities to those predicted by the model
in Section 4.
3.2
Charter Schools
We distinguish between potential entrant and incumbent charter schools. Apart from the actual
decision to enter, the main differentiating factor is that entrants are uncertain about their mean
demand shocks.
3.2.1 Entrants
The prospective entrant j at location
uses the preceding model to predict its enrollment share
for each potential school configuration: school focus Xj. and grades served G j . Recall there is a
one-to-one correspondence between an entrant and a location since each unoccupied location
holds one prospective entrant. For the purpose of estimation, we classified each school as serving
a particular grade level (e.g., elementary, middle) according to the set of grades the school served
at the end of our sample period. This is because when requesting authorization, prospective
entrants must specify the full set of grades they intend to serve.
A prospective entrant is uncertain about the precise value of its school-specific demand
realization  j (the assumption that it does not observe jgt means that it does not observe any
component of this shock, which is why he does not observe  j ). Although the prospective
20
entrant does not observe its shock, he knows the distribution of shocks in the population of
potential entrants. Further, we assume the distribution of shocks is common knowledge to the set
of potential entrants.
Although the entrant does not know its demand shock, it can form a belief about it based
on elements such as neighborhood searches for students, conversations with parents, etc. The
entrant’s belief about its demand shock (and hence about its potential success) is denoted by  jgt .
In each period a potential entrant observes a signal of their demand shock of the following form:
 jgt   jgt   jt ,
 jt
N (0, 2 )
where  jt is a noisy shock to the mean quality signal and 2 represents an entrants degree of
uncertainty over their beliefs. Entrants must integrate over this uncertainty when making their
entry, thematic focus, and grade level decisions. For simplicity, we assume the noisy shocks to
charters’ beliefs are distributed normally. We denote the distribution of the perceived demand
shocks,  jgt , for entrants to be w( ; ) . The normality assumption on the distribution of  jt
implies that  jgt
w( ; )  N ( jgt , 2 ) , such that the distribution of a charter’s beliefs about its
demand shock also follows a normal distribution centered at their unbiased belief  jgt . In our
estimation, we set  jgt   jgt , such that a school’s belief is equal to the realized value of its shock.
Potential entrants know the quality of incumbent schools, but are also uncertain about the
school-specific demand shocks of other potential entrants in different locations---since charter
schools can accept students from anywhere in the district, the demand shocks of entrants in other
locations can affect the realized enrollment for the current potential entrant. The entrant has
observed past realizations of shocks from this distribution, and so assumes potential entrants’
shocks will be equal to the mean of this distribution,  . We assume potential entrants do not act
strategically with respect to other entrants’ decisions.18
18
There is little evidence, anecdotal or otherwise, that prospective charter schools are strategic in their entry
decisions with respect to other entrants’ decisions in the same period. This assumption allows us to avoid the
complications associated with estimating a game of strategic interaction between the entering charter schools.
21
Note that the subscripts imply that this distribution varies over grades and can change over
time. Hence, the entrant forms the following expectation of its enrollment share for each
possible choice of X j :
E  s jgt | X j    s jgt ( X j , X  j , Y , D,  ,  j ,   j ) w( j ; )d  j
7.
j
E  kjt | X j     kjt ( X j , X  j , Y , D,  kjt ,  j ,   j ) w( j ; ) d j

j
Let Rg denote the reimbursement per student in grade g that a charter school obtains, and let
V ( X j , G j ) denote variable costs and F ( X j , G j ) fixed costs associated with focus Xj and choice
of which grades to serve G j .
Charter schools face different decisions depending on their status. A prospective entrant in a
given location, choosing its thematic focus X j and set of grades to offer G j , will enter if its
maximum attainable expected revenue net of fixed and variable costs is non-negative:
 G j

8. E    max  N gt E  s jgt | X j   Rg  V ( X j , G j )   F ( X j , G j )   0
X j ,G j
 g 1

e
jt
where N gt is the number of students eligible for grade g. The school district adjusts the base
reimbursement per student for inflation each year but otherwise reimbursements have largely
stayed constant over time. Thus a key reason a school would enter in one period versus another is
that expects to receive high enough demand shocks in one year to yield a sufficient supply of
students.
This concludes the description of prospective charter entrants, and so we next discuss
how incumbent charter schools decide whether to stay in their present location, move to a new
location, or exit the market entirely.
3.2.2 Incumbents
Having previously entered, an incumbent school no longer faces uncertainty over the value of its
demand shocks. In addition, incumbent schools do not change their thematic focus or their grade
levels. An incumbent charter school decides whether it will continue operation with its current
22
focus or not, and if so, whether to stay in the current location or move. An incumbent charter will
remain in operation if its expected revenue net of variable costs is non-negative.
Gtj
E  ijt | X j    N gt s jgt  Rg  V ( X j , G j )   0
g 1
An incumbent school may also change location. For simplicity, we assume an incumbent charter
school has a single alternative location that emerges exogenously. The charter school may
relocate from j to ℓ if alternative ℓ has net revenue gain sufficient to offset relocation cost mjℓ:
9. E  it   E  ijt 1   m j
4.
Estimation
This section describes our strategy to estimate the household choice and school choice sides of
the model. For now, we estimate the household choice model first, taking their preference
parameter estimates given when estimating school’s parameters.
4.1
Demand Estimation
We formulate household’s decisions as a discrete choice problem and estimate the model’s
parameters using an approach based on Berry, Levinsohn, and Pakes (1995). An important point
of departure is our inclusion of the endogenous peer characteristics of each school. BLP allows
for endogeneity in prices but assumes all other observed product characteristics are exogenous.
In our setting, the peer characteristics are the result of aggregate student choices, such as the
fraction of a school’s student body that is low income or a particular ethnicity. The peer
characteristics are an equilibrium outcome yet simultaneously determine aggregate choices and
represent an additional source of endogeneity. These endogenous school characteristics are
similar to the local spillovers in the Bayer and Timmins (2007) model of sorting.
Our data are a panel of markets over time because we consider each grade as a separate
“market.” This generates variation in the choice set across households within a time period. To
form the necessary moment conditions, we must first calculate a school’s predicted enrollment
23
share. Conceptually, we can think of each location
being inhabited by a representative
household from each grade g with demographic profile d i . Given that we possess an estimate of
the distribution of households over locations, grades, and demographics, the model’s predicted
enrollment share is:
sˆ jgt   Pijgt  X , Yij , di ,  t , t ;   h(dik , | g )
L
K
1 k 1
The equilibrium enrollment shares for the students simultaneously determine the levels of the
student peer characteristics:
G jt
10. ˆ 
k
jt
L
 d
P
ik ijgt
g 1 1
G jt L K
 X , Y , d ,  ,  ;    h( d
ij
i
t
t
ik
 P  X , Y , d ,  ,  ;   h(d
g 1 1 k 1
ijgt
ij
i
t
t
ik
, | g)
, | g)
Although we are able to impute local demographics according to grade level, we only observe
student peer characteristics at the school level. ˆ kjt is the model’s predicted value of the peer
characteristic using the observed value  kjt of the peer characteristic. We appeal to rational
expectations on the part of households to use the observed value on the right-hand side of
equation (6). In general we do not expect the model to perfectly predict the peer characteristic,
implying a measurement error exists that relates our predicted and observed peer characteristics.
Taking this error into account and substituting observed enrollment shares into the denominator
of (6), the observed peer characteristic is
 kjt 
1
s jt
 P  X ,Y , d ,
L
1
ijgt
ij
i
k
jt

, t ;  h(dik , | g )  u kjt
 ˆ kjt  u kjt
Our basis of estimation, as in BLP, is the conditional moment condition
E  jgt | Z , X   0,
where Z is a set of instruments that are assumed conditionally independent of the demand
shocks. The instruments in our application are critical because schools and households likely
24
observe numerous factors that influence household school choice but are unobserved by the
econometrician.
To this moment we add a condition relating the observed and predicted peer
characteristics. Under the assumption that conditional on observables the error term u kjt is mean
zero, the peer characteristic moment condition is:
E u jt | Z , X   0, k .
For estimation, we stack the moments above into a single vector. The respective sample moment
conditions average, as appropriate, over time periods, grades, and schools:
m1 ( , Z ) 
1 T G 1

TG t 1 g 1 J tg
m2 (u, Z ) 
1 T 1

T t 1 J t
J tg

j 1
Jt
u
j 1
jt
jgt
 h( Ztj )
 h( Z tj )
To estimate the model, we reformulate the problem as a mathematical program with
equilibrium constraints (MPEC). Dubé, Fox, and Su (2010), building on Su and Judd (2010),
show how to estimate the aggregate demand model in BLP as a constrained optimization
problem that simultaneously calculates the mean utilities and estimates the utility preference
parameters. Thus, a key benefit in following the MPEC approach is that we avoid numerically
inverting the market shares to obtain the mean utilities within a nested optimization, which
significantly improves the speed and robustness of estimation. We briefly describe the method
below and direct the reader to Dubé, Fox, and Su (2010) for details.
Let S and  be the observed market shares and observed peer characteristics stacked in
vector form across years t, schools j, and grades g. The MPEC objective function minimizes an
analog of a GMM objective subject to a set of constraints:
min
 , ,
s.t.
 'W 
S  s  , ; 
    ; 
  m  , u ,  
In contrast to the nested fixed point optimization in BLP, the MPEC algorithm simultaneously
searches over values for the demand shocks  and preference parameters  . Given values for
25
these, we can calculate the predicted market share. The first constraint ensures the observed
enrollment shares S match the enrollment shares predicted by the model given values for the
preference parameters, demand shocks, and peer characteristics.
The second constraint enforces the necessary equilibrium condition that observed peer
characteristics match expected peer characteristics. We plug-in the observed values of the peer
characteristics into consumer’s utility functions, so we are searching for parameter values that
make the predicted peer characteristics consistent with their observed counterparts. The third
constraint is used to simplify computation by converting the objective function into a quadratic
form in λ.
4.2
Supply Estimation
A charter school entrant chooses among a discrete set of alternatives consisting of whether to
enter or not, and conditional upon entry, a thematic focus. Entrant charter schools are uncertain
over their mean demand shock, which we express as TBD
A charter school imperfectly measures its profits when making decisions, introducing an
error term:
E  ejt   E  ejt   jt
The probability that a charter school enters and chooses a thematic focus x jt is:


exp E  ejt | x jt 
P( X jt  x jt ) 
1   exp E  ejt | x j 't 


x j 't
Need to build in the integration with the beliefs parameter, and indicate decisions in some way.
Recall our assumption that a potential entrants exists in each location  L . Then we form a loglikelihood function jointly over all time periods and locations such that
T
L
LTL ( )   Pr( X )
t 1 l 1
4.3
Identification and Instruments
26
In most empirical IO settings, price is the sole endogenous product characteristic. In contrast, we
must address the endogeneity of school entry and (potentially) multiple peer characteristics. A
school’s decisions to enter a specific location and choose a thematic focus are likely influenced
by the local factors observed by the school and households but unobserved by the researcher.
We consider several instruments to resolve the endogeneity of grade choice, thematic
focus, and demographic composition. First, we use data on businesses and employment from the
County Business Patterns. This database provides zip-code level data annually on employment
by industry and firm size. We specifically use data on local employment in industries where we
would expect schools to prefer to locate away from, such as drinking establishments (e.g., bars
and nightclubs) and heavy industries. Second, demographic information from the 1990 Census
pre-date the charter school movement in Washington DC and are clearly correlated with the
demographics in our data. Third, similar to demographics, characteristics of the housing stock
(age, condition, density, multi-family) should also be relevant for schools’ demographic
compositions today.
[To be completed]
4.4
Discussion and Shortcomings
In translating our theoretical model to one that we can estimate empirically, we are forced to
make a variety of assumptions and simplifications. Some of these assumptions have already been
stated explicitly in the text so far, although others are implicitly buried in our model’s
formulation. In this subsection we highlight what we believe to be the key assumptions and
discuss their implications.
One important shortcoming is the role of charter authorizers. First, we do not explicitly
model the charter authorization process. We lack adequate data to be able to accurately recover
the rules authorizers use to evaluate charters. A second issue is that, although the PSCB acted as
the main charter authorizer for most schools in the district, the BOE was still responsible for
authorizing some schools in a number of years. Our concern is that the observed set of charter
schools might be subject to some unobserved selection criterion due to the authorizers. If each
27
authorizing agency used the same evaluation rules and charter schools randomly applied to one
of the authorizers, than the presence of two authorizers would not pose a problem. However, it is
unclear whether the PSCB and BOE used the same standards to evaluate potential charters, and if
prospective entrants new about some of these differences, then an entrant could strategically
apply to one versus the other. This unaccounted for selection effect could lead our supply-side
estimates to be biased in some way.
A second shortcoming is that many charter campuses are administered by the same charter
entity, i.e., some charters operate multiple campuses much like a retail chain while others are
independent, single-campus entities. A few of these chains (e.g., Knowledge Is Power Program,
KIPP) are highly successful and likely benefit from operating multiple locations, both in terms of
reducing startup costs and improving education quality. Our model does not account for multicampus charters, and thus ignores the potential synergies that some schools enjoy as part of such
a chain.
[To be completed]
5.
Results
Counterfactuals
6.

Changes in funding

Facilities

No special theme allowed

Changes in transportation funding

Student targeting through charters

Changes in accountability requirements
Conclusion
First theoretical and structural treatment of charters
28
•
•
•
Equilibrium
Endogenous product quality
Post-entry outcome
Significant policy interest in charter schools
29
References
Bayer, P. and C. Timmins (2007), “Estimating Equilibrium Models of Sorting Across Locations,”
Economic Journal, 117(March), 353-374.
Berry, S., J. Levinsohn, and A. Pakes (1995), “Automobile Prices in Market Equilibrium,”
Econometrica, 63, 841-890.
Bettinger, Eric P. (2005), “The Effect of Charter Schools on Charter Students and Public Schools,"
Economics of Education Review, 24, 113-147.
Bifulco, Robert and Helen F. Ladd (2006), “The Impacts of Charter Schools on Student
Achievement: Evidence from North Carolina," Education Finance and Policy, 1 (1), 123-138.
Booker, K., S. M. Gilpatric, T. Gronberg, and D. Jansen (2007), “The Impact of Charter School
Attendance on Student Performance," Journal of Public Economics, 91 (5/6), 849-876.
_____ (2008), “The Effect of Charter Schools on Traditional Public School Students in Texas: Are
Children Who Stay Behind Left Behind?," Journal of Urban Economics, 64 (1), 123-145.
Buddin, Richard and Ron Zimmer (2005a), “Is Charter School Competition in California Improving
the Performance of Traditional Public Schools?," RAND Working Paper.
______, (2005b), “Student Achievement in Charter Schools: A Complex Picture," Journal of Policy
Analysis and Management, 24 (2), 351-371.
Cardon, James H. (2003) “Strategic Quality Choice and Charter Schools," Journal of Public
Economics, 87 (3), 729-737.
Clark, Damon (2009), “The Performance and Competitive Effects of School Autonomy," Journal of
Political Economy, 117 (4), 745-783.
Dubé, J. P., J. Fox, and C.-L. Su (2010), “Improving the Numerical Performance of BLP Static and
Dynamic Discrete Choice Random Coefficients Demand Estimation,” Working paper No. 09-07,
University of Chicago Booth School of Business.
Downes, Thomas A. and Shane M. Greenstein (2006), “Understanding the Supply Decisions of
Nonprofits: Modelling the Location of Private Schools,” RAND Journal of Economics, Vol. 27(2),
Summer, pp. 365-390.
Draganska, Michaela, Mike Mazzeo, and Katja Seim (2009), “Beyond Plain Vanilla: Modeling Joint
Product Assortment and Pricing Decisions,” Quantitative Marketing and Economics, 7(2), 105-146.
Draganska, Michaela Sanjog Misra, Victor Aguirregabiria, Pat Bajari, Liran Einav, Paul Ellickson,
Dan Horsky, Sridhar Narayanan, Yesim Orhun, Peter Reiss, Katja Seim, Vishal Singh, Raphael
Thomadsen, and Ting Zhu (2008), “Discrete choice models of firms’ strategic decisions,” Marketing
Letters, 19, 399–416.
30
Ferreyra, Maria M. (2009). “An empirical Framework for Large-Scale Policy Evaluation, with an
Application to School Finance Reform in Michigan”. American Economic Journal: Economic Policy,
vol. 1 no. 1: 147–80.
Glomm, Gerhard, Douglas Harris, and Te-Fen Lo (2005), “Charter school location,” Economics of
Education Review, 24, 451–457.
Hanushek, Eric A., John F. Kain, Steven G. Rivkin, and Gregory F. Branch (2007), “Charter School
Quality and Parental Decision Making With School Choice," Journal of Public Economics, 91 (5/6),
823-848.
Holmes, George M., Jeff DeSimone, and Nicholas G. Rupp (2003), “Does School Choice Increase
School Quality?," NBER Working Paper No. 9683.
Hoxby, Caroline M. (2004a), “Achievement in Charter Schools and Regular Public Schools in the
United States: Understanding the Differences," Dept. of Economics, Harvard University, mimeo.
Hoxby, Caroline M. and Jonah E. Rockoff (2004b), “The Impact of Charter Schools on Student
Achievement," Dept. of Economics, Harvard University, mimeo.
Hoxby, Caroline M. and Sonali Murarka (2009) “Charter Schools in New York City: Who Enrolls
and How They Affect Their Students' Achievement," NBER Working Paper No.14852.
Imberman, Scott A. (forthcoming), “Achievement and Behavior of Charter Students: Drawing a
More Complete Picture," Review of Economics and Statistics.
________ (2009), “The Effect of Charter Schools on Achievement and Behavior of Public School
Students”, Working paper, University of Houston.
National Alliance for Public Charter Schools (2010), “How State Charter Laws Rank Against the
New Model Public Charter School Law,” January, downloaded from www.publiccharters.org.
Rincke, Johannes (2007), “Policy diffusion in space and time: The case of charter schools in
California school districts,” Regional Science and Urban Economics, 37, 526–541.
Sass, Tim R. (2006), “Charter Schools and Student Achievement in Florida," Education Finance and
Policy, 1 (1), 123-138.
Su, C.-L. and K. L. Judd (2010), “Constrained Optimization Approaches to Estimation of Structural
Models,” Working paper, University of Chicago Booth School of Business.
Weiher, Gregory R. and Kent L. Tedin (2002), “Does Choice Lead to Racially Distinctive Schools?
Charter Schools and Household Preferences," Journal of Policy Analysis and Management, 21 (1),
79.
Zimmer, Ron and Richard Buddin (2003), “Academic Outcomes," in Charter School Operations and
Performance, RAND, pp. 37-62.
31
TABLE 1
Demographics and Achievement at Public and Charter Schools
Avg.
Percent White
Percent Black
Percent Hispanic
Pct. Low Income
Reading Prof.
Math Prof.
Tract Income
4.86
83.37
10.04
66.22
41.61
41.67
$54,000
All Schools
10th
pctile.
0
41.61
0
30.07
15.84
13.51
$28,200
90th
pctile.
21.36
100
31.64
89.32
72.97
73.98
$98,400
Public Schools
10th
90th
pctile.
pctile.
5.60
0
22.50
81.82
36.79
100
10.61
0
35.09
65.09
27.44
88.70
41.55
14.70
77.52
41.54
12.99
76.10
$56,300 $29,300 $136,600
Avg.
Charter Schools
Avg.
10th
90th
pctile.
pctile.
2.19
0
4.00
89.03
58.00
100
7.98
0
26.00
70.34
50.00
95.00
41.83
25.32
63.39
42.18
21.05
66.78
$46,000 $25,600 $65,600
Notes: The unit of observation is a campus-year. “Reading Prof.” is the percent of students who are proficient in
Reading. “Tract income” is the average household income in the Census tract where the school is located.
TABLE 2
School Openings and Closings
Public School Campuses
Year Open Closed Merged
2003
0
0
0
2004
1 (1)
1 (1)
0
2005
0
5 (5)
0
2006
0
0
1 (1)
2007
0
0
1 (1)
Charter School Campuses
Open
Closed
Merged
4 (4)
0
0
9 (8)
1 (1)
0
9 (9)
1 (1)
0
8 (7)
1 (1)
0
6 (6)
3(1)
0
Notes: each cell indicates number of campuses; number of schools is in parentheses. For instance, in 2004 9 charter
campuses and 8 new charter schools opened. The three campuses of Washington Academy were taken over by
Howard Road in Spring 2008. In this table we count them as three campuses (and one school) that close in 2007.
32
TABLE 3
Grade Levels at Public and Charter Schools
Percent
Elementary
Elementary/Middle
Middle
Middle/High
High
Elem./Middle/High
(1)
68.62
4.28
14.41
n/a
12.70
n/a
Public Schools
Pct. of
Avg.
Students Enrollment
(2)
(3)
58.03
315
4.96
432
15.23
394
n/a
n/a
21.78
639
n/a
n/a
Percent
(4)
42.61
20.87
11.74
6.09
14.78
3.91
Charter Schools
Pct. of
Avg.
Students Enrollment
(5)
(6)
30.79
226
23.17
347
12.56
334
6.85
352
19.55
413
7.08
566
Notes: The unit of observation is a campus-year. For instance, on average during the sample period 68.62 percent of
public schools are elementary, 4.28 are elementary/middle, 14.41 are middle, and 12.70 are high schools. Among
public school students, on average 58.03 percent attend elementary schools, 4.96 attend elementary/middle schools,
etc.
33
TABLE 4
Demographics and Achievement at Public and Charter Schools, by Level
Avg. Pct. White
Avg. Pct. Black
Avg. Pct. Hispanic
Avg. Pct. Low Income
Avg. Pct. Proficient Reading
Avg. Pct. Proficient Math
Avg. Tract Hh. Income
Elem.
6.38
80.40
11.30
68.96
46.18
45.29
$54,000
All Schools
E/M
Middle
2.57
4.00
90.20
85.25
6.56
9.08
71.14
65.71
42.44
39.95
43.69
38.83
$40,700 $61,000
High
3.30
85.11
9.73
57.61
31.89
35.25
$56,500
Elem.
6.85
79.70
11.41
68.32
46.86
46.08
$54,600
Public Schools
E/M
Middle
1.43
4.38
91.41
85.19
6.74
8.58
71.31
66.66
36.92
37.21
39.51
35.60
$39,500 $62,300
High
4.07
82.92
10.77
53.94
31.51
34.06
$61,000
Elem.
3.15
85.20
10.55
73.33
41.49
39.84
$50,000
Charter Schools
E/M
Middle
3.46
2.32
89.25
85.52
6.42
11.29
71.01
61.53
46.75
52.03
46.94
53.07
$41,600 $55,513
High
.16
93.97
5.54
72.47
33.44
40.06
$39,000
Notes: The unit of observation is a campus-year. “Elem.” and “E/M” denote elementary and elementary/middle school, respectively. Weighted averages; weight
is school-level enrollment.
34
TABLE 5
Program Foci at Charter Schools
Focus
Core
Language
Arts
Vocational
Civics
Math and Science
All Levels
(1)
61.30
11.30
9.57
7.83
6.52
3.48
Elem.
(2)
56.12
22.45
20.41
0
0
1.02
E/M
(3)
97.92
2.08
0
0
0
0
Middle
(4)
74.07
11.11
0
3.70
3.70
7.41
M/H
(5)
21.43
0
14.29
35.71
28.57
0
High
(6)
20.59
0
0
35.29
29.41
14.71
E/M/H
(7)
100
0
0
0
0
0
Notes: the unit of observation is a campus-year. “Elem.”, “E/M”, “M/H” and “E/M/H” denote elementary,
elementary/middle, middle/high and elementary/middle/high school respectively. For instance, among elementary
charter campuses, on average 56.12 percent focus on a core curriculum, 22.45 percent on language, etc.
35
TABLE 6
Student Demographics and Achievement by Program Focus and Level
a. Elementary Schools
Arts
Core
(1)
Avg. Percent White
Avg. Percent Black
Avg. Percent Hispanic
Avg. Percent Low Income
Avg. Pct. Proficient in Reading
Avg. Pct. Proficient in Math
Avg. Tract Hh. Income
Language
(2)
3.98
90.13
4.74
69.60
43.73
43.08
$52,800
.65
94.57
4.28
84.40
36.98
31.50
$38,700
(3)
3.54
44.06
50.65
73.80
37.49
38.46
$56,800
Math and
Science
(4)
0
96.00
2.00
46.00
58.41
33.63
$30,800
b. Middle Schools
Civics
Core
(1)
Avg. Percent White
Avg. Percent Black
Percent Hispanic
Avg. Percent Low Income
Avg. Pct. Proficient Reading
Avg. Pct. Proficient Math
Avg. Tract Hh. Income
0
68.00
32.00
90.00
30.43
28.26
$63,800
Language
(2)
2.59
85.83
10.67
60.49
52.88
54.47
$57,200
(3)
0
55.02
44.79
92.22
25.56
16.06
$59,600
Math and Vocational
Science
(4)
(5)
0
0
96.44
100
2.56
0
54.34
92.00
59.29
21.43
58.07
20.00
$30,800
$30,500
c. High Schools
Avg. Percent White
Avg. Percent Black
Avg. Percent Hispanic
Avg. Percent Low Income
Pct. Proficient Reading
Avg. Pct. Proficient Math
Avg. Tract Hh. Income
Civics
Core
(1)
(2)
.50
76.37
22.31
69.92
38.07
43.39
$35,300
0
99.24
.51
65.78
33.82
44.11
$34,900
Math and Vocational
Science
(3)
(4)
.30
.06
99.00
98.94
.52
.94
100
73.14
46.73
19.24
55.26
18.02
$31,900
$56,600
Note: Unit of observation is a campus-year.
36
TABLE 7
Number of Charter Campuses that Move
Year
Number of Campuses
2004
2005
2006
2007
39
47
54
59
Number of Campuses that
Move
3
5
7
4
Note: unit of observation is a campus. For instance, 3 of the 39 campuses in 2004 have a different location than in
2003.
TABLE 8
Number of Campuses per School
Number of Campuses per
School
1
2
3
5
Total
Number of Schools
35
4
5
1
45
Note: unit of observation is a charter school organization. Since the number of campuses changes over time for some
schools, the table depicts information pertaining to the highest number of campuses that each school has during the
sample period. For instance, 35 charters have one campus, 4 have 2, etc.
TABLE 9
High-Performing Schools
Public Schools
Avg. Tract Household Income
$103,700
Avg. Percent White Students
35.56
Avg. Percent Black Students
47.68
Avg. Percent Hispanic Students
10.44
Avg. Percent Low Income Students
25.02
Charters
$19,600
.36
98.48
.57
74.21
Note: unit of observation is a campus-year.
37
TABLE 10
Early versus Recent Charter Entrants
Number of campuses
Avg. Enrollment
Pct. Focused on Core
Pct. Elementary
Pct. Elementary/Middle
Pct. Elementary/Middle/High
Pct. High
Pct. Middle
Pct. Middle/High
Avg. Tract Hh. Income
Pct. belonging to multiplecampus charters
Pct. White Students
Pct. Black Students
Pct. Hispanic Students
Pct. Low Income Students
Pct. Proficient Reading
Pct. Proficient Math
Early Entrants
27
450
66.67
14.81
33.33
11.11
22.22
11.11
7.41
$48,800
25.93
Recent Entrants
37
192
56.76
59.46
13.51
0
2.70
18.92
5.41
$50,500
56.76
1.07
92.54
5.94
73.78
42.49
40.49
5.24
85.05
8.55
64.95
40.90
35.55
Note: unit of observation is a campus. Information tabulated here corresponds to the last year that the campus is
observed in the data. Weighted averages; weight is enrollment.
38
0
20
40
60
80
100 120 140
FIGURE 1
Number of Public and Charter School Campuses
2003
2004
2005
Y EAR
C harter
2006
2007
Public
39
60,000
40,000
20,000
Enrollment
80,000
FIGURE 2
Enrollment in Public and Charter School Campuses
2003
2004
2005
Y EAR
Public Enrollment
Total Enrollm ent
2006
2007
C harter Enrollm ent
40
0
10
20
30
FIGURE 3
Charter School Enrollment Shares
2003
2004
2005
Y EAR
PS-6
9-12
2006
2007
7-8
Total
Notes: percentages calculated relative to total enrollment, where total is the aggregate over public and charter
schools for all grade levels. The figure depicts shares for preschool through 6th grade, grades 7 and 8, and grades 912.
41
FIGURE 4
Geographic Location of Public and Charter Schools
Schools in
Washington, DC 2007
%
%
%
%
%
%
%
^
_
%%
^
_
%
%
%
^% %
_
%%
%
%
^% _
_
^
%
%
^
%_
%
^
_
%
%
%
%
%
%
%
%
%
%
^^
_
^ _
_
^^
_
% _
%
^
_
^
_
^% _
_
^_
_
^^ % %
%
%% _
^%
%
^_
^ _
_
^ % %
%
^
_
%
^%
^_
_
%
% %
%
^%
_
^
_
%
^
_
^%
_
% %
%
%
%
^%% %
_
%
^
_
%
%
%
%
%
^
_
^
_
% %
^
_
%
%
%
%
% %
%_
^
%
^ _
_
%
^%
%
%
%
^
_
^% %
_
%
% %%
%
% ^
^% _
_
%
%
%
%%
%%
^
_
%
^% % _
_
^
^%
_
%_
^% % %
^
%_
%%
%
^%
_
%% %
%
%
%
^
_
^
%_
%
%
^
_
%
%
^
_
%
%
% %
%
%
Average Household Income - 2000
$0 - $19,999
^
_
Charter Schools
$20,000 - $34,999
%
Public Schools
$35,000 - $54,999
$55,000 - $89,999
0
2.5
5 Miles
$90,000 - $161,000
42
FIGURE 5
Charter Enrollment Share by Grade
50.00%
45.00%
40.00%
Percentage
35.00%
30.00%
2003
2004
25.00%
2005
20.00%
2006
15.00%
2007
10.00%
5.00%
0.00%
PS PK
K
1
2
3
4
5
6
7
8
9
10 11 12
Grade
43
Appendix I: Data
Public Schools
The starting point for this dataset was audited enrollments from the District of Columbia Office
of State Superintendent of Education (OSSE), available at
http://osse.dc.gov/seo/cwp/view,a,1222,q,552345,seoNav,|31195|.asp. From here we obtained the
list of public schools along with their enrollment for preschool, pre-kindergarten, kindergarten,
each grade between 1 and 12, and enrollment of ungraded and adult students.
Our list only includes regular schools. This means that relative to the original OSSE list, we have
excluded the following:
1. alternative schools
2. special education schools
3. early childhood centers
We excluded schools listed as alternative by the OSSE, schools that report their program as
alternative in the Common Core of Data (CCD), and schools listed as alternative during the
corresponding years by DCPS. We also excluded schools that offer residential programs, such as
Ballou Stay and Spingarn Stay. In total, we eliminated 11 alternative schools from the original
OSSE lists.
We excluded schools listed as special ed by OSSE, and schools that report their own type as
special ed in the CCD. Some schools report their type as special ed only during one or two years
during our sample period, in which case we called the individual schools to gauge their type and
removed them from the data if they were special ed. In total, we eliminated 22 special ed schools
from the original OSSE lists.
Early childhood centers include one or several of the following: pre-school, pre-kindergarten,
and kindergarten, and do not include regular grades. We excluded early childhood centers as
long as they never had enrollment in regular grades during the sample period. For instance, if an
early childhood center added first grade in 07, then the school was included for all 5 years.
Below is the list of information and sources for each school, along with the construction of the
corresponding variables:
Address: geographic address of the school. The main source for this variable was the
CCD. When the address was not consistent over time, we used Google Earth and Google
Maps to see whether the different addresses corresponded to the same geographic
location or not. If not, we consulted the school’s web site, or called the school to track
down the history of its location. In the absence of other information, we used the most
recently reported address. We geocoded all addresses.
School enrollment: total school enrollment, excluding ungraded and adult students.
Source: own calculations based on OSSE.
44
Grade-level enrollment (grades preschool through 12). Source: OSSE.
Percent of white students: percent of White students in the school. Source: own
calculations based on CCD. For the few cases in which the CCD data was not available,
we used the demographics reported to fulfill NCLB requirements (found in
http://www.nclb.osse.dc.gov/). Notice, however, that the NCLB requirements pertain to
the students enrolled in the grades tested for the sake of NCLB accountability, not to the
entire student body. When using the NCLB web site, we computed this percent as the
ratio between the number of white students in the grades tested and the total number of
students in those grades.
Percent of black students, percent of Hispanic students, percent of students of other
ethnicities: constructed similarly to percent of white students.
Percent of Low Income students: calculated as the percent of students who receive
either free or reduced lunch. Source: own calculations based on CCD. For the few cases
in which the CCD data was not available, we used the demographics reported in
fulfillment of NCLB requirements.
Reading proficiency: percent of students who are proficient in Reading. Source:
http://www.nclb.osse.dc.gov/. In 03 and 04, proficiency levels were determined
according to the Stanford-9 assessment. To be considered proficient, a student was
supposed to score at the national 40th percentile or higher. Since 05, proficiency has been
determined according to DC CAS (Comprehensive Assessment System). See
http://www.nclb.osse.dc.gov/faq.asp for more information.
Prior to 05, the grades tested were 3, 5, 8 and 10 (according to the School Performance
Reports for PCBS-authorized charter schools, and according to our own calculations
comparing grade-level enrollment with number of students tested). Since 05, the grades
tested have been 3, 4, 5, 6, 7, 8 and 10 (according to
http://www.nclb.osse.dc.gov/aboutayp.asp, School Performance Reports and our own
calculations comparing grade-level enrollment with number of students tested).
The achievement website, http://www.nclb.osse.dc.gov/faq.asp , reports proficiency for
elementary and secondary schools. According to http://www.nclb.osse.dc.gov/faq.asp,
schools are classified as elementary or secondary as follows. Since 05, elementary
schools are schools with a 3rd and/or 5th grade. If the school goes above the 6th grade it
must also have a 3rd grade. Secondary schools are schools with no 3rd and a grade above
6th. Prior to 2006, schools with a wide grade range were required to meet both the
elementary and secondary targets. The general rule for most schools was that they were
classified as elementary if they had no grade above 6 and as secondary if they had no
grade below 5. Schools that did not meet either of these criteria had to meet the targets at
both levels.
45
For some schools and years, proficiency data are not available for one of the following
three reasons: 1) the school only includes early childhood enrollment; 2) the school only
includes grades that are not tested (for instance, the school includes kindergarten and first
grade only); 3) the school includes grades that are tested, but enrollment in those grades
is below the minimum threshold for reporting requirements. The last reason is the most
prevalent cause of missing proficiency. See below for the imputations made in those
cases.
Math proficiency: percent of students who are proficient in Math. Constructed similarly
as reading proficiency.
Percent of out-of-boundary enrollment: percent of students enrolled through the outof-boundary process for years 04, 05, 06 and 07 (we do not have information for 03).
Source: own calculations based on DCPS data.
Year of Opening: year the school opened, if it was not open in 03. Using the CCD
“status” variable and web searches we verified the school’s initial year. This is the first
year for which we have records.
Year of Closing: year the school closed, if it was not open in 07. The variable stores the
first year that the school was no longer open. We verified the content of this variable
using the CCD “status” variable and web searches.
Year of Merge: year the school merged with another school. The variable stores the first
year that the school no longer operates separately, which is the first year for which we
have joint records.
Ethnic composition and low-income status of the student body were missing for 2 and 4 (out of
701) observations respectively; these are schools with very low enrollment. Since ethnic
composition varies very little for any given school over time, to the cases of missing ethnic
composition we imputed the school’s average ethnic composition over the years for which we do
have data. Whenever possible, we imputed the predicted value coming from a school-specific
linear trend.
Achievement was missing in 16 out of 701 observations. To these observations, we imputed the
predicted achievement coming from the regression of school-level proficiency rates on year
dummies, ethnic composition variables, percent of low income students, enrollment, and school
fixed effects. In cases in which we had no proficiency data for the school at all, we ran a similar
regression excluding school fixed effects and including dummies for school level, and used the
resulting predicted values for our imputations.
Charter Schools
As with public schools, the starting point for this dataset was audited enrollments from the
District of Columbia Office of State Superintendent of Education (OSSE), available at
http://osse.dc.gov/seo/cwp/view,a,1222,q,552345,seoNav,|31195|.asp. From here we obtained the
46
list of public schools along with their enrollment for preschool, pre-kindergarten, kindergarten,
each grade between 1 and 12, and enrollment of ungraded and adult students.
For consistency with public schools, we excluded alternative schools. We identified alternative
schools based on OSSE’s classification, the schools’ mission statements, and the self-reported
program type in the CCD. We excluded schools in which ungraded or adult students constitute
the majority of the student body, and schools with residential programs.
By law, charter schools cannot serve special education students exclusively. However, they can
offer services targeted to specific populations even though their enrollment must be open to all
students. Only one charter school in D.C., St. Coletta, serves special ed students exclusively, and
they made this agreement with the Board of Education (BOE) when they were chartered. We
excluded schools whose services target mostly special ed students. We identified those schools
based on OSSE’s classification, the schools’ mission statements, the self-reported program type
in the CCD, and phone conversations with staff from the Public Charter School Board (PCSB).
We excluded early childhood schools only if they never added regular grades. We also excluded
online campuses.
Some non-early childhood charters opened an early childhood campus during the sample period.
We only included these campuses if at some point they added regular grades. Community
Academy-Amos II is an early childhood campus that did not include regular grades (at least not
during our sample period) and is hence not included in our data. Similarly, Roots Academy
opened an early childhood campus in 07, which we did not include in the data. KIPP-LEAP is an
early childhood campus that opened in 07 and is not included either. In contrast, E.L. Haynes
opened a campus in 07 for early childhood and first grade; this campus is included in our data.
Below is the list of information and sources for each campus, along with the construction of the
corresponding variables:
Parent organization: if the school belongs to a multi-campus charter, this is the name of
the charter organization.
Authorizer: entity which authorized the charter. Before 07, there were two authorizers –
the Public School Charter Board (PCSB) and the Board of Education (BOE). After 07,
there was only one authorizer (PCSB).
Address: geographic location of the campus. For PCSB schools, the main source was the
SPRs. For BOE schools, the main source was the CCD. We supplemented these sources
with web and Internet archive searches, and phone calls to the individual schools. We
geocoded all addresses.
Several schools moved in the middle of the school year, temporarily relocated, or closed.
We consulted the SPRs and various web sites to handle these cases. If the school moved
in the middle of a school year, the address variable contains the new address. Some
47
schools relocated some students for a few months during renovations. Since this was a
temporary arrangement and parents knew it, we did not consider this a change in address.
School enrollment: total school enrollment, excluding ungraded and adult students.
Source: own calculations based on OSSE.
Grade-level enrollment (grades preschool through 12). Source: OSSE.
Percent of white students: percent of White students in the school. For PCSB charters,
the source is the School Performance Reports (SPRs) when available. Only for one
campus and year (Tri Community in 07), does the SPR not include demographic
information, in which case we use the NCLB data from http://www.nclb.osse.dc.gov/
otherwise.
For BOE charters in 2007, the source is the SPRs (in 2007, the PCSB began including the
BOE-authorized charters in its reports). For BOE charters before 2007, the source is the
NCLB web site. Note that the NCLB information pertains only to students in grades
tested, not to the whole student body of a school. These sources were supplemented by
the CCD when needed.
When the school has multiple campuses but we only have one set of ethnic composition
data, we impute it to all campuses.
Percent of black students, percent of Hispanic students, percent of students of other
ethnicities: constructed similarly to percent of white students.
Percent of Low Income students: percent of low-income students, equal to the fraction
of students on free or reduced lunch. For PCSB charters, the source is the SPRs. For BOE
charters, the source is NCLB information in http://www.nclb.osse.dc.gov/. These sources
were supplemented by the CCD when needed and possible, in which case case pctLowInc
is the fraction of students with either free or reducedlunch.
When the school has multiple campuses but we only have one set of low-income
variables, we impute it to all campuses.
Reading proficiency: percent of students who are proficient in Reading. See public
schools for sources and construction.
Some charters span elementary and secondary grades, and this affected their reporting.
For instance, Capital City had to meet elementary and secondary targets before 05, and
for those years it reported two sets of scores. Since 05, Capital City has had to meet only
one target, has been considered an elementary school for the sake of reporting, and has
had to report only one set of scores. In general, when the school had only one campus but
separate proficiencies for elementary and secondary students, we combined them into a
single proficiency indicator for comparability with other years in which we had a single
proficiency indicator.
48
For multi-campus charters we usually had achievement data for each campus. When we
did not, we imputed the available data to all the campuses.
As with public schools, we did not have proficiency data for some campus and years for
the reasons enumerated above. In the case of many PCSB schools for which the NCLB
web site did not report test scores due to low enrollment, we obtained proficiency rates
from the SPRs. This was not possible for BOE schools with low enrollment since the
SPRs only cover PCSB schools before 07. In cases in which we could not find
proficiency data, we made imputations (see below).
Math proficiency: percent of students who are proficient in Math. Constructed similarly
as reading proficiency.
Year of Opening: year the campus opened. Source: SPRs, FOCUS, web searches. The
variable stores the first Fall that the school is open.
Year of Closing: year the school (as opposed to the campus) closed, even if it closed
after 07. The variable stores the first year that the school is no longer open. The sources
are Center for Education Reform
(http://www.edreform.com/accountability/charters/CER_ClosedCharterSchools2009.pdf),
current SPRs and PCSB listings of charter schools, current NCLB reports, and web
searches.
Note that Washington Academy was taken over by Howard Road in Spring 08 – i.e., in
the middle of school year 07. Hence, the year of closing is 2007 for Washington
Academy because it the first year for which Washington Academy no longer exists. In
fact, for the school year 07/08, data were reported by Howard Road, and the SPRs report
data for Howard Road campuses noting that they were former campuses of Washington
Academy. We chose not to add campuses to Howard Road for 07 because the campuses
did not physically move – they simply changed ownership.
Reason for Closing: reason for closing (academic / financial / mismanagement). Source:
Center for Education Reform, SPRs, web searches.
Statement: the school’s mission statement. Source: schools’ web sites, FOCUS, SPRs.
The percent of low-income students was missing for 9 out of 230 observations respectively. As
in the case of public schools, these schools had low enrollments. In most cases we imputed the
school’s average percent of low-income students, the average calculated over the years for which
the school does have data.
In the case of missing proficiency rates (36 out of 230 observations), we made imputations
similar to those described for public schools, the only difference being that we used school- (as
opposed to campus-) fixed effects in the predicting regressions.
49
Appendix II: Charter Statements and Foci
School Name
ACADEMIA
BILINGUE DE LA
COMUNIDAD PCS
ACADEMY FOR
LEARNING
THROUGH THE
ARTS PCS
ARTS AND
TECHNOLOGY
ACADEMY
BARBARA
JORDAN PCS
CAPITAL CITY
Statement
ABC will graduate adept learners, effective
communicators, and community leaders
who are culturally aware and prepared to
use their academic skills and bilingual
proficiency to succeed in rigorous high
schools, post-secondary education, and
society.
The Academy for Learning through the Arts
Public Charter School provides students in
grades Pre-K through 6th with an
outstanding academic and social education
through a rigorous arts-integrated program.
The Arts & Technology Academy Public
Charter School (ATA) provides an
academically challenging, technologically
rich, student-centered learning environment,
where each student develops a strong
intellectual, moral, environmentally
conscious, and artistic foundation. ATA
strives to meet student academic needs
through an exciting instructional program
and varied extra-curricular activities. The
school’s accountability process is
characterized by a combination of parental
involvement and performance measurement.
To provide a rigorous, academic program
that promotes high levels of student
achievement and fosters high self-esteem,
inspiration, and motivation for lifelong
learning. BJPCS is proud to offer an array
of services and programs for bilingual
instruction to small class sizes in a secure
school environment.
The Capital City Upper School offers a
unique opportunity C Public Charter School
Board parents and education professional as
a school dedicated to best practices in
education reform. The School implements a
comprehensive model known as
Expeditionary Learning
(www.elschools.org) , which emphasizes
project-based instruction to help students
meet rigorous academic and character
standards. Capital City also uses the
Responsive Classroom social curriculum,
which provides a consistent school-wide
approach to classroom management and
Focus
Language
Arts
Arts
Core
Core
50
CESAR CHAVEZ
PCS FOR PUBLIC
POLICY
CHILDRENS
STUDIO SCHOOL
CITY
COLLEGIATE PCS
focuses on respectful social interactions. All
students receive instruction in music,
drama, visual art and fitness.
Drawing on the vast policy resources in the
nation’s capital, Chávez schools challenge
students with a rigorous college preparatory
curriculum that fosters citizenship and
prepares them to excel in college and in life.
Chávez schools use public policy themes to
guide instruction and provide students with
direct experience working in the public
interest. Chávez schools provide nurturing,
safe learning environments where every
student is expected to meet rigorous
academic and personal behavior standards.
It is challenging, but Chávez is a public
school of choice and among both the
students and faculty there is a sense of pride
and excitement in being part of a network of
schools that is beating the odds and
achieving success.
Children's Studio School is a 30-plus yearold community-based arts organization,
studio complex, and a full-day school of the
arts and architecture for young children.
Engaging highly accomplished artists of
diverse disciplines and cultures (who are
also scientists, authors, mathematicians,
anthropologists, etc.) as complete educators,
the school is internationally recognized and
has received numerous awards for its
unique pedagogy of transcultural Arts As
Education©. The School was awarded its
charter by the DC Board of Education in
1996 and reopened its full-day school in
1997. It is one of the first accredited schools
for young children in DC. Studio School's
Urban Arts Complex also provides evening
and week-end Artists and Communities
programs for families, teachers, artists and
community; City As Studio is open to all
young searchers in the summer.
City Collegiate Public Charter School, a
District of Columbia public charter school
serving grades 7 and 8, will value and
support the unique needs and capabilities of
adolescents as they transition to adulthood
and create a community of lifelong learners
through a rigorous standards-based
curriculum explored in a collaborative
Civics
Arts
Core
51
COMMUNITY
ACADEMY
DC BILINGUAL
DC
PREPARATORY
ACADEMY
E.L. HAYNES
environment with the support of students'
families.
The mission of Community Academy
Public Charter School (CAPCS) is to create
a family-centered, community-focused
learning environment that offers a worldclass preschool through secondary
education. CAPCS carries out its mission
through a rigorous, standards-based
academic program delivered by caring,
supportive and highly competent educators
and administrators. Careful attention is
given to assessment and the use of
technology in helping students to master
skills, acquire knowledge, and develop a
love of learning. CAPCS expects
excellence from its students, staff and
families.
DC Bilingual Public Charter School opened
in September 2004 with 120 students from
pre-school through kindergarten. The
school will add an additional grade each
year, through 5th grade in 2009, eventually
serving over 350 students. DC Bilingual’s
mission is to expand educational
opportunities and choices for children and
their families who seek an academically
stimulating environment that fosters
bilingualism, illiteracy, critical thinking,
problem-solving and a joy of learning. The
school’s curriculum and program design
encourage students to be self-motivated,
life-long learners, who will express their
creativity, develop strong leadership skills,
and value the diverse cultures of their city
and world. DC Bilingual’s philosophy and
approach flow out of the two decades of
work in bilingual early childhood education
of its founding organization, CentroNía.
DC Prep provides a developmentally
appropriate, academically challenging
education for preschool and elementary
children. The key ingredients of highquality early childhood programs have been
incorporated into DC Prep's proven model,
creating a coherent and exciting learning
environment that lays the foundation for
college preparation and success from the
earliest years.
E.L. Haynes Public Charter School was
Core
Language
Core
Core
52
EARLY
CHILDHOOD
ACADEMY
ELSIE WHITLOW
STOKES
COMMUNITY
FREEDOM
founded in 2004 and has a program based
on nationally-recognized best practices for
advancing student achievement. At the core
of the school’s program is the unwavering
belief that every student, regardless of race,
socioeconomic status, or home language, is
capable of reaching high levels of academic
performance. In the 2008-2009 school year,
E.L. Haynes will serve 370 students in
grades Pre-K through 6 and will grow,
adding a grade each year, until it serves 650
students in grades Pre-K through 12. The
school is the first year-round public school
in Washington, DC and is named for Dr.
Euphemia Lofton Haynes, the first AfricanAmerican woman to receive a doctorate in
mathematics, a teacher in the District of
Columbia's school system for 47 years, and
the first woman to serve as President of the
District of Columbia Board of Education.
E.L. Haynes was recently chosen from
DC’s 56 charter schools as the winner of
Fight for Children’s first-ever Quality
Schools Initiative Award, and was also
ranked 6th among a consortium of 99
charter schools nationwide in student
achievement gains, as recognized in New
Leaders for New Schools’ Effective
Practice Incentive Community. The school
completed construction of a new 46,000
square foot facility in May 2008.
A literacy-focsed curriculum using the
Creative Curriculum and the Core
Knowledge sequence. The school culture
values respect, compassion, curiousity, and
first-hand experience
The Elsie Whitlow Stokes Community
Freedom Public Charter School, prepares
culturally diverse pre-kindergarten and
elementary school students in the District of
Columbia to be leaders, scholars and
responsible citizens who are committed to
social justice. We teach children to think,
speak, read, write and learn in two
languages: English and French or English
or Spanish. With a dual focus on academic
excellence and community service, the
Stokes School accomplishes its mission by
creating an environment of achievement,
respect and non-violence.
Core
Language
53
FRIENDSHIP PCS
- BLOW-PIERCE
CAMPUS
FRIENDSHIP PCS
- CHAMBERLAIN
CAMPUS
FRIENDSHIP PCS
- COLLEGIATE
ACADEMY
FRIENDSHIP PCS
- SOUTHEAST
ACADEMY
Blow Pierce is committed to developing
high achievers for a successful transition
from middle to high school. Through a
rigorous curriculum including pre AP
programs, Blow Pierce builds on students’
individual strengths and provides the
academic and developmental programs to
ensure each child’s success. Like all
Friendship Public Charter School campuses,
the school provides both academic and
wrap-around services to its students and
provides a technologically rich environment
to support student achievement.
Chamberlain Elementary provides a rich
learning environment purposely crafted to
develop early academic success. With more
than 700 students, Chamberlain provides
academic instruction, extended learning and
extra curricular activities to students in
grades PK to 7. Like all Friendship Public
Charter School campuses, the school
provides both academic and wrap-around
services to its students and provides a
technologically rich environment to support
student achievement.
Collegiate Academy is a college preparatory
high school serving students in grades nine
through 12. Challenging and relevant,
Collegiate’s rigorous core curriculum
prepares students for college and the world
of work in a global economy. Students
benefit from a broad offering of Advanced
Placement and Honors courses as well as
Career Academy courses that provide
pathways in technology, science,
engineering, law and communications. The
Collegiate Early College program allows
highly motivated students starting in the
ninth grade to take college courses and earn
a maximum of two years college credit.
Southeast Elementary provides a rich
learning environment purposely crafted to
develop early academic success. With more
than 500 students, Southeast provides
academic instruction, extended learning and
extra curricular activities to students in
grades PK to 6. Like all Friendship Public
Charter School campuses, the school
provides both academic and wrap-around
services to its students and provides a
Core
Core
Core
Core
54
FRIENDSHIP PCS
- WOODRIDGE
CAMPUS
HOPE
COMMUNITY
PCS
HOSPITALITY
PCS ROOSEVELT
HIGH
HOWARD ROAD
ACADEMY PCS
HOWARD
UNIVERSITY MIDDLE SCHOOL
OF MATH
technologically rich environment to support
student achievement.
Woodridge Elementary and Middle
provides a rich learning environment
purposely crafted to develop early academic
success and to prepare middle school
students for a high school and college. With
more than 600 students, Woodridge
provides academic instruction, extended
learning and extra curricular activities to
students in grades PK to 8. Like all
Friendship Public Charter School campuses,
the school provides both academic and
wrap-around services to its students and
provides a technologically rich environment
to support student achievement.
The mission of Hope Community PCS is to
positively shape the hearts and minds of our
students by providing them with an
academically rigorous, content rich
curriculum, an environment in which
character is modeled and promoted, and a
community in which to build trusting
relationships with others.
Hospitality High School is the nations’ first
and only public charter school designed to
introduce students to careers in hospitality
management and culinary arts. The school
provides a rigorous academic program to
promote success in college and the
hospitality industry. The goal is to empower
students with academic, critical thinking,
technological and interpersonal skills
necessary to successfully compete in a
diverse global economy.
Howard Road was conceived by a group of
professionals committed to improving the
quality of education available to residents in
Ward 8. Managed by Mosaica Education,
Inc., the educational program features a
researched-based, direct- instruction
curriculum. Chartered for 1,270 students in
grades kindergarten through twelve, the
school began operations by serving 500
kindergartens through sixth grade students.
The school added seventh grade in 2004.
The Howard University Public Charter
Middle School of Mathematics and Science
(MS)2 is the first charter middle school in
Washington, DC to be established by a
Core
Core
Vocational
Core
Math and Science
55
HYDE
LEADERSHIP PCS
IDEAL
ACADEMY
university. In an enriched educational
environment, it will provide a sound
foundation in all academic subjects, with a
concentration in mathematics and science
that will prepare students to succeed in high
school and beyond. (MS) 2 highlights
include: an academic model designed to
prepare middle school students for college
and careers in math, science, and
engineering; a highly qualified instructional
staff with ongoing professional
development; an extended school day and
school year; small class sizes with a low
student to teacher ratio; advanced
technology designed specifically for middle
school students, and active parent and
community involvement in support of high
student achievement. (MS)2 will be located
on Howard University’s elite college
campus and will be open to sixth grade
students this fall.
Hyde Leadership Public Charter School is
Pre K-12 College preparatory school
committed to developing socially
responsible leaders who are equipped with
the knowledge, skills, and personal integrity
necessary for life and leadership in the 21st
century. For this reason we have worked
hard to develop a high quality,
comprehensive, school program rooted in
the three pillars of character education,
family renewal, and academic rigor. At the
core of Hyde’s work with our students and
families is the belief that each of us is gifted
with a unique potential that defines a
destiny. It is around this core belief that we
have shaped our program goals and
objectives. Our focus on character
development and family education is a
critical element in our success. We have
created a school community that seeks
challenge and opportunity, where our
families feel like partners in the pursuit of
their sons’ and daughters’ academic success
and character development. Our parents
understand that they are the primary
teachers and that their home is the primary
classroom in their children’s lives.
Mission of the Ideal Academy Public
Charter School is to create and implement
institutions and programs, which help
Core
Core
56
INTEGRATED
DESIGN
ELECTRONIC
ACADEMY
KAMIT
INSTITUTE
students to develop fully in areas related to
their mind, body, spirit and emotions.
Students who receive “Ideal” education are
empowered to excel in all academic and
personal growth activities; to positively and
productively handle challenges and stresses;
to appreciate the fine arts; and to express
their own creativity, intuition and critical
thinking skills in order to solve problems
and positively contribute to the quality of
life. Ideal students are able to understand,
accept and respect themselves and others; to
appreciate and live in accordance with the
very finest socially acceptable and life
supportive standards, values and moral; and
to work together with others regardless of
sex, race, color or creed in order to make
positive contributions to their family,
community, the city in which they live, and
ultimately, the world.
The IDEA Public Charter School is a career
focused, academic, secondary school which
prepares students (grades 7-12) for college
or for careers in electronics, drafting,
computer networking, or computer repair.
The computer networking and repair classes
can lead students towards national
certifications. The electronics class
provides students with training to become
electrician apprentices. The AutoCAD class
prepares students for architecture,
engineering, and drafting. IDEA's JROTC
program includes participation in JROTC
summer camp. Self-discipline, leadership
training, and citizenship skills are addressed
through the JROTC program. In addition to
the career focused courses of study, the
school relies on partnerships with
employers in the industries. These business
partners provide summer/part-time jobs and
internships for students as well as providing
mentorship, tutorial, instructional, and
curriculum supports for teachers and
students.
Kamit Institute for Magnificent Achievers
PCS provides a challenging and stimulating
academic course of study for college and
career-bound adolescents, supplemented by
hands-on engagement with their global
environment through cultural exploration,
social research and outdoor education.
Vocational
Core
57
KIPP DC
LATIN
AMERICAN
MONTESSORI
BILINGUAL
MARY MCLEOD
BETHUNE
Smaller class sizes and Individual Learning
Plans (ILP) ensure that each student
receives personalized cultivation. Kamit
Institute’s accountability plan encompasses
articulated goals for an ongoing process of
looping information about performance
back to students, teachers and parents.
KIPP DC’s mission is to empower our
students with the academic and character
skills necessary to succeed in high school,
college, and in life. In order for students to
succeed academically, KIPP DC schools
provide a safe and structured environment
with innovative and engaging teachers. By
using mandatory summer school, Saturday
school, and extended hours during the
week, the academic program offers students
an intensive foundation in the core
academic subjects. Teachers are available
via cell phone to parents in the evening. In
addition, parents receive weekly reports
updating them on their child's progress.
KIPP DC has a fair and consistent discipline
plan that holds students accountable for
their actions.
The Latin American Montessori Bilingual
Public Charter School's (LAMB) goal is biliteracy in English and Spanish. LAMB
currently has pre-school through second
grade. In school year 2007 – 2008, LAMB
will have three classrooms of 3-, 4- and 5year-olds, and two elementary classrooms
of 1st, 2nd and 3rd graders. LAMB utilizes
the Montessori curriculum - core academic
and non-academic subject matter. Each
class has a Spanish dominant Montessori
teacher and an English dominant
Montessori teacher. Children are immersed
in Spanish for 80% of the time during the
primary years of Montessori; and 60%
Spanish and 40% English in elementary
(1st, 2nd, and 3rd grade). Each year LAMB
will add a grade level until 6th grade.
Mary McLeod Bethune PCS features a partday Spanish language immersion program
for students in preschool – grade 2 and a
challenging academic program for all
students in a supportive, diverse learning
environment that is parentally involved, to
enable each student to achieve academic
Core
Language
Language
58
MAYA ANGELOU
MERIDIAN PCS
NEW SCHOOL
FOR ENTERPRISE
&
DEVELOPMENT
NIA
COMMUNITY
PCS
PAUL PCS
success, foreign language acquisition,
talent, and positive social development.
Mary McLeod Bethune Day Academy PCS
celebrates diversity and ensures
opportunities for students to experience a
global education. The school has a staff
representing more than 14 nations of the
world.
A year-round program serving at-risk youth
by integrating academics and work
experience.
The primary goal of Meridian Public
Charter School is to ensure that each student
achieves at the highest possible level.
Reading and Mathematics serve as the
academic core of the school. The
curriculum is further enhanced and
supported through Science, Social Studies,
the arts, Technology, Foreign Languages
and Character Development programs. Each
classroom has both a lead teacher and a
teaching assistant who work with students
simultaneously. The entire Meridian
teaching team meets periodically to
examine its teaching strategies in light of
student performance on standardized tests.
The love of learning, the foundation of
knowledge, and skills that our students
acquire will serve them throughout their
lifetime!
Train low income students to become
entrepeneurs.
Nia Community Public Charter School is
committed to providing comprehensive
educational services that are studentcentered and focused on the individual
educational needs of each child. Teachers,
staff and parents will work together to meet
the needs of each student. NCPCS also
commits to building community
partnerships with organizations, business
and government agencies within the
metropolitan area.
The mission of Paul Public Charter School
is to educate our students and to develop in
them the capacity to be responsible citizens,
independent thinkers, and leaders. Paul
offers a rigorous technologically rich liberal
Vocational
Core
Vocational
Core
Core
59
POTOMAC
LIGHTHOUSE
ROOTS
SEPTIMA CLARK
arts curriculum that is to designed to meet
the needs of every student. It focuses on
educating the whole student through its
Triple “A” Program: Academics, Arts,
and Athletics. The Cecile R. Middleton
Ninth Grade Academy integrates
technology and global awareness in its core
curriculum and emphasizes service learning.
The Potomac Lighthouse Public Charter
School provides children with a rigorous,
arts-infused educational program that
prepares them for college. The school
model is based on the five (5) ideas that
together form the Lighthouse Cycle of
Success:The arts engage
students.Engagement leads to higher levels
of achievement.Achievement increases
expectations; higher expectations, in turn,
further engage everyone.Engaged students
benefit from extra time in school.Visionary
leadership makes these pieces fit together.
The Roots Charter School (Roots) provides
an intellectually challenging, Africancentered learning environment. Roots PCS
provides a program for students to develop
exemplary character, academic excellence,
and strong social and moral responsibility.
Roots strives to 00meet student academic
needs through an exciting heritage,
culturally- rich instructional program and
varied extra-curricular activities. The
school’s accountability process is
characterized by a combination of academic
success, performance measurement and
parental involvement.
The Septima Clark Public Charter School is
the first public charter school for boys in
Washington, DC. Our philosophy is simple:
Education is freedom. Every student
deserves an exceptional education that
prepares him for academic mastery and
achievement. An unrelenting focus on
excellence--personal and intellectual-ensures our students will have the crucial
tools to build the future of their dreams. Our
college preparatory mission ensures that
your son is prepared to compete, achieve,
and contribute in the world. MISSION: At
the Septima Clark Public Charter School,
preschool to 8th grade boys master
Arts
Core
Core
60
advanced academic skills and knowledge to
compete, achieve, and contribute in
academic institutions of distinction.
SOUTHEAST
ACADEMY OF
SCHOLASTIC
EXCELLENCE
THE VILLAGE
LEARNING
CENTER
THE
WASHINGTON
LATIN PCS
THURGOOD
MARSHALL
ACADEMY
TREE OF LIFE
COMMUNITY
PCS
Rigorous age- and grade-appropriate
academic program.
Village Learning Center started in 1998 as
an expanded and secularized version of
what had been a muslim private school.
Washington Latin Public Charter School
provides a challenging, classical education
that is accessible to students throughout the
District of Columbia. Challenging, classical,
and accessible are key words in the mission
of our school. Our talented and caring
faculty and staff challenge students with
high academic and personal expectations.
Thurgood Marshall was proposed and
developed by a group of Georgetown
University law students and faculty
members. An outgrowth of a popular, extracurricular program, “Street Law Clinic,” the
Academy is founded on the belief that all
children must be afforded equal educational
opportunity in order for our democracy to
live up to its promise. To this end, the
school will prepare 400 high school
students, grades eight through twelve, for
civic participation through the America’s
Choice standards-based curriculum
embedded with lessons about law,
democracy, and human rights. Instruction
will model the democratic ideals of fairness,
participation, and respect for diverse
perspectives.
The Tree of Life Community Public Charter
School signed its charter on April 6, 2000
and opened in September 2000 as a District
of Columbia Charter School. The Tree of
Life Community Public Charter School
provides holistic academic and
nonacademic programs and services for
children and families of the District. The
school offers small classes of 18-20
students for grade levels pre-kindergarten
through eight. A content-rich, childfocused, culture-rich and relationshipfocused curriculum model is utilized to
achieve higher academic performance and
Core
Core
Core
Civics
Core
61
TRI-COMMUNITY
PCS
TWO RIVERS
develop productive behaviors and skills for
one's contribution to the world community.
The integration of academic content (and
individualized instruction), particularly
reading and literacy, within the context of
stronger family/community services is
emphasized. It is our vision to prepare our
children to become exemplary citizens of
our community and world through:
academic and experiential exploration;
technological proficiency; exposure to
foreign languages; building positive
community and world relationships;
community and world service; and national
and world travel. A content-rich, childfocused, culture-rich and relationshipfocused curriculum model is utilized to
achieve higher academic performance and
develop productive behaviors and skills for
each child's contribution to the world
community. We emphasize the integration
of academic content, particularly reading
and literacy, within the context of stronger
family/community services. The school is
free and open to the DC public.
A school founded with community
involvement, the school focuses on literacy
development and small
class size, using the America’s Choice©
school design.
Two Rivers will be a vibrant educational
environment where students and staff
become a community of learners on a
journey of discovery. Founded by an
energetic and committed group of D.C.
parents, Two Rivers intends to use
Expeditionary Learning Outward Bound,
and educational model that emphasizes
interactive, hands-on, project-based
learning. The school will focus on the
whole child, recognizing the importance of
character education and the socialemotional needs of children while helping
them achieve academic excellence.
Through a diverse student body and a
demonstrated commitment to servicelearning, Two Rivers’ students will be
empowered to make positive contributions
to their community and develop a strong
sense of self-awareness.
Core
Core
62
WASHINGTON
ACADEMY PCS
WASHINGTON
MATHEMATICS,
SCIENCE,
TECHNOLOGY
WILLIAM E.
DOAR PCS
YOUNG
AMERICA
WORKS
An educational environment that enhances
and supports the intellectual and social
growth of children through the development
of innovative curriculum concepts, creative
instructional techniques, collaborative
educational partnerships, and the
incorporation of state of the art technology
into all phases of the curriculum
The mission of WMST is to offer a
rigorous, standards-based, college
preparatory education which will prepare
young people in the District with an interest
in math, science and technology for success
in work and life. WMST is an open
enrollment, liberal arts public charter high
school specializing in the preparation of
students for higher education and,
subsequently, careers in the fields of math,
science and technology. The summer
program runs from the first week of July
through the second week of August. Our
computer labs are open before school, on
the lunch hour, and after school. Many
school activities are offered, and we have a
renowned AFROTC program.
The William E. Doar, Jr. Public Charter
School for the Performing Arts (WEDJ
PCS) is designed to provide a collegepreparatory academic and artistic learning
environment that challenges students to
reach their maximum intellectual, artistic,
social, and emotional development as
rapidly as their talents permit. We firmly
believe that an ideal education addresses the
fact that students have multiple learning
faculties, instructs students in a number of
disciplines in the Renaissance tradition and
challenges students to engage and relate to
the world around them beyond an insular
definition of community. Combining a
multi-disciplinary arts program with the
highest educational standards, WEDJ PCS
is committed to graduating well-rounded,
liberally educated, responsible young men
and women.
The Young America Works Public Charter
School’s (YAW) mission is to instill and
cultivate a passion for lifelong learning with
each student through a rigorous academic
program combined with invaluable career
Core
Math and Science
Arts
Vocational
63
training and direction that inspires
academic, professional and personal
excellence. Our year-round program not
only prepares students for college, it
prepares them for adult life. Strong
relationships with the business community
offer students opportunities to explore and
excel in the world while developing a strong
work code of ethics. Upon graduation, each
student will have the education,
occupational skills and self-confidence to
identify and accomplish his or her personal
goals.
64