What Makes KIPP Work? A Study of Student Characteristics, Attrition, and School Finance
Gary Miron, Jessica L. Urschel, and Nicholas Saxton College of Education and Human Development Western Michigan University
Paper Jointly Released by the National Center for the Study of Privatization in Education, Teachers College, Columbia University and the Study Group on Educational Management Organizations at Western Michigan University www.wmich.edu/leadership/emo (contact: [email protected]
Executive Summary To date, most research on KIPP has focused on outcomes (i.e., student achievement). This study does not question the body of evidence on student achievement gains made in KIPP schools. Instead of looking at outcomes, this study examines two critical inputs: students and funding. Understanding more about these inputs allows for a better understanding of how KIPP works and whether the model can or should be replicated. This study’s research questions examine KIPP’s student characteristics and compare them with those of the local school districts. The study, which is national in scope, also compares student attrition at KIPP schools with local school districts, and finds that high levels of attrition are pervasive across the KIPP network—a finding congruent with findings presented in earlier research. A second topic addressed in this study is an analysis of KIPP revenues and patterns of expenditures. Using the most recent federal dataset on school finance (2007-08), we compared KIPP schools’ revenues and expenditures relative to local districts and to national means for charter schools and traditional public schools. Beyond the federal dataset, we systematically reviewed IRS Form 990 tax filings from KIPP schools so that we could calculate the amount of private revenues KIPP has received, most of which comes from philanthropic groups. Key Findings Student Characteristics: A Cross-Sectional Look at KIPP Schools Relative to Local Districts •
• • •
During 2008-09, KIPP enrolled a significantly higher proportion of African American students (55%) than did the respective local school districts (32%). However, KIPP schools served a substantially lower proportion of Hispanic students (39%) compared with local districts (50%). KIPP also enrolled substantially fewer white students (2%) compared with local districts (11%). KIPP schools enrolled a higher percentage of students eligible for free and reduced-price lunch (77%) than did the local school districts (71%). KIPP schools enrolled a lower percentage of students with disabilities (5.9%) than did their local school districts (12.1%). KIPP enrolled a lower percentage of students classified as English Language Learners (11.5%) than did their local school districts (19.2%).
Distribution of Students Across Grades in 2008 • •
Although a few of the KIPP schools now serve students at the primary and upper secondary levels, the overwhelming majority of KIPP’s enrollment is still at the middle school level. KIPP schools have experienced a sharp increase in enrollment and in the number of schools, with enrollment tripling between 2005-06 and 2008-09.
Student Attrition • • •
KIPP schools have substantially higher levels of attrition than do their local school districts. Our analysis revealed that on, average, approximately 15% of the students disappear from the KIPP grade cohorts each year. Between grades 6 and 8, the size of the KIPP grade cohorts drop by 30%. The actual attrition rate is likely to be higher since some of the KIPP schools do fill in some of the vacated places after grade 6. When these figures are further broken out by race and gender, we can see that a full 40% of the African American male students leave KIPP schools between grades 6 and 8. Overall a higher proportion of African American students than other ethnic groups leave the KIPP schools, and girls are much more likely remain in the KIPP schools across all ethnic groups. Attrition rates for students qualifying for students eligible for free and reduced-price lunch are approximately equal for KIPP schools and their host districts.
Analysis of KIPP Revenues and Expenditures Revenues Using the federal dataset on school finance (2007-08), we were able to obtain detailed revenue from 25 KIPP schools and their local districts. • During the 2007-08 school year, KIPP received more per pupil in combined revenue ($12,731 per student) than any other comparison group: the national average for all schools ($11,937), the national charter average ($9,579), or the average for KIPP schools’ local school districts ($11,960). • KIPP received more in per-pupil revenue from federal sources ($1,779) than did any other comparison group: the national average ($922), the national charter district average ($949), or KIPP schools’ host districts ($1,332). • None of the 12 KIPP districts reported any private revenues in the NCES finance survey; however, a separate analysis of these districts’ 990 tax forms for 2007-08 revealed large sums of private contributions. Per-pupil contributions for the 11 KIPP districts that we could include in this analysis equaled an average of $5,760, much more than the $1,000 to $1,500 additional per-pupil revenue KIPP estimates is necessary for their program. Two KIPP districts or groups received more than $10,000 per pupil in private revenues. • Combining public and private sources of revenue, KIPP received, on average, $18,491 per pupil in 2007-08. This is $6,500 more per pupil than what the local school districts received in revenues. Expenditures • •
As a whole, KIPP districts spend more per pupil in total current expenditures ($10,558) than do other charter school districts ($8,492), slightly more than their host districts ($10,101) and more than the national average for all schools ($10,121). KIPP spends more on instruction ($5,662) than the average for charter schools ($4,617) but less than the national average ($6,196) or KIPP host districts ($5,972).
• • • •
KIPP’s per-pupil spending on student support services ($460) is comparable to that of charter schools nationally ($464), but much less than the national average ($1,003) and even less than KIPP’s host districts ($1,179). KIPP’s per-pupil spending on administration ($972) is more than the national average ($746) or KIPP host districts ($687), but lower than the average for charter schools ($1,336). KIPP spends more on operations per pupil in dollars and as a percentage of total current expenditures than any other comparison group. KIPP’s additional spending in this area is focused in transportation, food services, and other support services. When spending on salaries is examined on a district-by-district basis, 11 of 12 KIPP districts spend less per pupil on salaries. The same pattern emerges when examining employee benefits. Eleven of the 12 KIPP districts spend less on employee benefits than do their host districts. KIPP also spends less per pupil on special education teachers’ salaries than does any other comparison group. The finding likely reflects the fact that KIPP enrolls fewer students with disabilities, particularly students with moderate or severe disabilities. As noted above, KIPP receives an estimated $6,500 more per pupil in revenues from public or privates sources of revenues. Our evidence on expenditures, show that KIPP reports spending $457 more per pupil than local school districts. From publicly available sources of information, however, we cannot determine whether or how KIPP spends its private sources of revenues.
How and Why KIPP is Successful at Improving Student Performance •
Selective entry of students. The findings in our report show that students with disabilities and students classified as English language learners are greatly underrepresented. The relative absence of students with disabilities and English language learners results in more homogenous classrooms. Secondly, in both traditional public schools and KIPP schools, the additional costs for these students—especially students with moderate or severe disabilities—is typically not fully funded, and therefore some of the costs for regular education is devoted to students requiring additional remediation. Because traditional public schools have a higher proportion of students with disabilities, and a higher concentration of students with severe and moderate disabilities, the burden of having to subsidize their education falls more heavily on them. High rate of student attrition with nonreplacement. The departure of low-performing students helps KIPP improve its aggregate results. Unlike local school districts, KIPP is not replacing the students who are leaving. When a student returns to a traditional public school after the autumn head count, KIPP retains most or all of the money (the amount depends on the particular state) allocated for educating that student during that school year. Traditional public schools do not typically benefit in the same way when they experience attrition, since vacancies are typically filled by other mobile students, even in mid-year. The discussion of findings at the end of this paper describe how “peer effects” play to KIPPs advantage, especially given its practice of filling few of the large number of vacancies from students who leave. High levels of funding that KIPP schools receive from both public and private sources. The additional resources KIPP receives are further compounded by the cost advantages it enjoys based on the students it serves compared with traditional public schools. Such advantages may be offset in part by the additional resources KIPP requires for its program’s longer
school day and longer school year. KIPP estimates that the additional costs for its expanded hours of instruction amount to between $1,000 and $1500. KIPP’s practices that result in selective entry and exit result in homogeneous groups of students that mutually benefit from peers who are engaged, have supportive families, and are willing and able to work hard in school.
Policy Implications If KIPP wishes to maintain its status as an exemplar of private management of public schools, rather than a new effort to privatize public schools, it will need to convince policymakers and the public that it intends to recruit and serve a wider range of students and that it will be able to do so with sustainable levels of funding comparable to what other traditional public schools receive. Before KIPP can be considered a model to be widely replicated, it has to be committed to serving all the students it admits and to serving a portion of the students who are mobile, including those who require a place in the middle of the school year, after the cut-off time for public funding to follow the student. Furthermore, to be considered a viable contributor to a system of public schools, KIPP also needs to recruit and serve a reasonable share of students who are more costly to educate, especially students with disabilities and students who are not native English speakers. The limited range of students that KIPP serves, its inability to serve all students who enter, and its dependence on local traditional public schools to receive and serve the droves of students who leave, all speak loudly to the limitations of this model. Furthermore the funding KIPP receives from public and private sources—more than $6,500 more per pupil in addition to what local school districts receive—is not likely to be sustainable in the longer run. KIPP’s only effort to take over a traditional public school—with a representative range of students and with the responsibility to serve all students who came and went during and between school years—ended in failure after only two years. This short-lived experiment with Cole Middle School in Denver speaks loudly about the viability of the KIPP model for public schools. Even though the KIPP model may not be replicable on a larger scale, its example does serve as a lever for change. The existence of KIPP schools has pushed the conversation about the value and importance of more instructional time for low-income students. Similarly, KIPP’s practice of recruiting and preparing administrators who can lead urban schools is another aspect of KIPP that is changing thinking about our public schools in a positive way. While much has been known about the outcomes of the students who persist in KIPP schools, this study provides a better understanding of two critical inputs: students and funding. The results of this study help answer questions related to how KIPP functions and how it succeeds in raising the achievement levels of students who persist in its schools.
Introduction In recent years, the Knowledge is Power Program (KIPP) has become one of the most widely praised models for school reform. Strong support for KIPP has come from both the Bush and Obama administrations, and hardly a day passes that KIPP’s success is not touted in the national media. The news and information about KIPP has largely focused on student achievement and the fact that KIPP targets high-poverty communities. Unfortunately, beyond this news, much is still not known about the program. This study seeks to fill in more of the missing information about KIPP. It addresses three very broad questions: How does KIPP work? What are the reasons for its success? Does KIPP provide a model that can be adapted more widely? To date, most research on KIPP has focused on outcomes (i.e., student achievement). This study does not question the body of evidence on student achievement gains made in KIPP schools. In fact, it is our view that KIPP’s claims that it improves test results of its students faster than traditional public schools are supported by rigorous and well-documented studies. Instead of looking at outcomes, this study will examine two critical inputs: students and funding. Understanding more about these inputs will allow for a better understanding of the how KIPP works and whether the model can or should be replicated.
Brief Description of KIPP The Knowledge is Power Program (KIPP) began as a fifth-grade public school program in Houston, Texas, in 1994. It was started by two Teach for America alumni, Mike Feinberg and Dave Levin. This Houston program became KIPP Academy Middle School. The following year, Levin opened a second school, KIPP Academy, in New York City. Both of the original KIPP schools are still in operation. In 2000, the KIPP Foundation was established in partnership with Gap Inc.’s founders, Doris and Donald Fisher, to expand the programs operating in New York City and Houston. The KIPP Foundation has grown considerably in the last decade. The foundation does not manage the network of schools, but rather facilitates sharing among schools and regions and works to fulfill its mission “to recruit, train and support educators to open KIPP schools across the United States.1” To this end, the KIPP Foundation runs the KIPP School Leadership Program, made of five specialized leadership programs. Two of these programs are designed to train leaders who can take responsibility for expanding the network of KIPP schools. According to the foundation’s website, over 100 leaders have been trained to open KIPP schools through these fellowships. The other three leadership programs are designed to promote leadership among current KIPP employees. The KIPP Foundation also organizes and hosts teacher retreats, compiles national outcome data from the KIPP regional organizations and independent schools, and provides support to the rapidly expanding network, including “legal support, real estate, technology, finance, corporate governance, operations, communications, marketing, and development” (Mathematica, 2010, pp. 1-2).2 At the beginning of the 2010-2011 school year, there were 99 KIPP schools in operation in 20 states and the District of Columbia. In its early years, KIPP focused exclusively on middle school education, but it has since expanded to elementary and high school education. However, the majority of KIPP schools are still serving middle school grades. According to its website,
KIPP schools are often placed in “high need” urban and rural neighborhoods. These are charter schools and are tuition-free, open-enrollment public schools, meaning there is no application or selection process. If more students are interested than there are spaces available, students are chosen through a lottery. KIPP schools have an extended school day and school year. This model touts the “No Excuses Method” (Thernstrom & Thernstrom, 2003)3. Typically the school day lasts from 7:30 a.m. until 5:00 p.m. weekdays and includes mandatory Saturday school every other week. According to Angrist, Dynarski, Kane, Pathak, and Walters (2010),4 each KIPP school sets its own curriculum, but many KIPP schools share important features. One such feature is that KIPP students earn points for good work and receive paychecks (in the form of scrip that only KIPP can redeem) based on those points. Paychecks can be spent on field trips and other rewards. Another feature is that both parents or guardians and students are asked to sign the “Commitment to Excellence.” For students, the commitment promises that students will arrive at school on time, dress according to the dress code, take responsibility for their actions, and conform to other such behaviors. The commitment requires parents to check homework every night and to make sure the child is on time and attends Saturday school and summer school, among other commitments. One can learn more about KIPP from its website: www.kipp.org.
Review of Relevant Literature Student achievement. Most of the scholarly literature focusing on KIPP has concentrated on achievement outcomes.These achievement studies have generally found positive results in both math and reading (Angrist et al, 2010; Doran & Drury, 2002; Mac Iver & Farley-Ripple, 2007; Mathematica, 2010; SRI International, 2008). 5 For example, the initial findings of a multiyear, rigorous evaluation of KIPP middle schools by Mathematica reported that 15 of 22 KIPP middle schools studied experienced significantly positive gains in math during the first year; only one had a significantly negative impact. In math, 8 of 22 schools experienced significantly positive gains in reading, and only two schools experienced significantly negative impacts in reading. These results were even better during the second year of study. As the most comprehensive, the most rigorous, and the most recent of studies that have examined student achievement in KIPP schools, this study deserves considerable attention. Even though the study is funded by KIPP, the technical report presents sufficient details that suggest that the study was carefully conducted and fairly reported. Since this Mathematica evaluation of KIPP is an ongoing study, we include recommendations for the evaluation team to consider in the concluding section of this report. Doran and Drury (2002)6 assessed achievement gains in three KIPP schools. Focusing on fifth-grade performance in the school’s start-up year, the authors found that these students experienced greater gains than the year before entering KIPP and also experienced greater gains than their host districts. These sorts of results are typical of achievement studies focusing on KIPP schools. Selection bias. The “No Excuses Method,” with its longer school days and year, along with strict dress and behavior codes, has led some researchers to question whether selection bias helps explain KIPP’s favorable achievement data. The suggestion is that the characteristics of the children who apply to KIPP schools or their families may be more responsible for academic outcomes than the school the child attends. Carnoy, Jacobsen, Mishel, and Rothstein (2005)7
used New York State assessment scores from 2002 to compare those who entered fifth grade in KIPP-Bronx Academy with their peers in a two-mile radius. The authors found that those entering the KIPP school had higher passing rates on the fourth-grade reading test than those who did not apply to KIPP. The authors also conducted interviews with teachers in feeder schools in New York City, Washington, DC, and Houston. These teachers stated that they encouraged more able and motivated students and those with supportive families to apply to KIPP schools. Similarly, Doran and Drury (2002) found that although fifth-grade students made greater gains in their first year at KIPP than they did the year before entering the school, the students entering KIPP had already been making above-average gains on the SAT 9 before enrolling. These results led the authors to conclude that there had been selection bias favoring at least one of the three KIPP schools they studied. A 2008 study by SRI International8 examined the issue of selection bias in five KIPP schools in the San Francisco Bay area. All five schools had higher rates of minority students, and four had higher rates of low-income students, than did their host districts. However, the KIPP schools also had lower rates of English language learners and special education students. Four of the five schools also had significantly higher rates of female students than males. Mac Iver and Farley-Ripple (2007)9 evaluated KIPP Ujima Village in Baltimore, MD using scores on the Maryland School Assessment, controlling for student characteristics. The authors found that KIPP students were similar to students from feeder schools in terms of race, gender, low-income status, and attendance, but that KIPP students were less likely to have special education status. Henig (2008)10 weighed the evidence available on selection bias at the time and concluded there was little evidence that KIPP exhibited systematic bias in terms of race, class, or past performance. He concluded, however, that there was evidence to suggest females are more likely to enroll and that there may be still additional measures on which KIPP students differ from those in traditional public schools. The initial findings of the Mathematica evaluation of KIPP schools released in 2010 assessed student characteristics in 22 of the 35 KIPP middle schools open in 2005. The authors found these schools have significantly higher concentrations of racial minorities than the surrounding districts. Although there was variation among the schools, KIPP schools were more likely to enroll low-performing students, and KIPP students were more likely to score below the host district’s average prior to enrollment in a KIPP school. Attrition of students. Related to selection bias is the issue of student attrition. Others11 have questioned the magnitude of KIPP’s achievement effects in light of its retention rates. Henig (2009) discussed the lack of evidence regarding attrition in KIPP schools, and explained why such information was important: “If those who leave KIPP schools are disproportionately those who are struggling academically, or whose families lack the supportive attributes that often predict success, then their absence could account for higher test scores or group gains over time” (p. 6). An internal memo written by the KIPP Foundation in 2008 examined attrition in 45 KIPP schools in operation in both 2006-07 and 2007-08. The percentage of students leaving these schools during those years ranged from 4% to 36%, with an average of 16%. Established regions (regions with more than 3 schools, an executive director or superintendent responsible for the oversight of all regional schools, a shared governing board, and shared back office support) exhibited lower attrition, an average of 10% during this period. Independent, single-site schools had an average attrition rate of 19%.
Mathematica (2010)12 found that cumulative rates of attrition varied widely among KIPP schools (10% to 76%), but that overall, their attrition rates were similar to those of traditional public schools. Approximately one-third of the KIPP schools studied had significantly higher rates, and approximately one-third had significantly lower rates. However, the Mathematica evaluation reported that lower-performing KIPP students were more likely to leave the school than higher-performing students. This pattern may inflate achievement outcomes over time, especially if these lower-performing students are not replaced.
Methods and Data Sources The methods used for the study are spelled out in this section. When appropriate and necessary, further details about the methods are included in the sections containing actual findings. Data and information required for this study all came from publicly available sources that are clearly documented. We used the national Common Core of Data (CCD) as the primary source.13 The most recent year for which we could obtain CCD data was 2008-09. Enrollment, race/ethnicity, and free and reduced-priced lunch (FRL) status were gathered from school-level datasets. FRL is the most commonly used proxy for family poverty levels and is the metric used for that purpose in this study. While the KIPP annual report for 2008 indicated that 64 KIPP schools or campuses were operating in 2008-09, we were only able to obtain data from 60 KIPP schools (94%) in the federal dataset. District-level datasets were used to obtain special education (defined as those having Individual Education Plans- IEPs) and English Language Learner (ELL) information. Because IEP and ELL data was only available from the district level dataset, we could only include KIPP schools that were designated as their own district (or LEA). In a few states, charter schools do not have LEA or district status and therefore do not report IEL and ELL data that is specific to the charter schools, but instead have their data joined together with other public schools in local district. In total we were able to gather IEP and ELL data that covered 37 of the KIPP schools/campuses in 2007-08. This corresponds to58% of the KIPP schools operating at the time. Each KIPP school was compared with its local traditional public school district. To identify the host or local traditional public school district for each KIPP school, we first identified the closest traditional public school using greatschools.org. The closest school’s name was put into the Common Core of Data to identify the district in which it was located. A database of all the host districts was then compiled, with the same information as was collected on the KIPP schools to allow for comparisons. Grade-level cohorts were created for KIPP schools and local school districts by linking grade-level groups as they progressed over successive years and grades. To build these cohorts we gathered corresponding CCD data from 2005-06, 2006-07, 2007-08 and 2008-09. KIPPs cohorts were designed in three- and four-year segments (grades 6-8 or grades 5-8); for any KIPP school to be included in the cohort, it was required to have students enrolled at each of the specified grades in the cohort. For this reason, KIPP schools that did not have a full middle school implemented were excluded from the sample. Districts corresponding to the appropriate KIPP schools were also sorted into matching cohorts. In reporting and comparing KIPP schools with local school districts, we calculated weighted averages, and we also tallied that number of match-ups that either favored KIPP or the local district. When we calculated our estimated attrition or drop off in enrollments, we report
the first year of the grade cohort as 100% and then report the relative size of enrollments in the subsequent years as a proportion of the first year enrollment. For example, if a KIPP school had 100 students in the grade 6 in 2006-07, and it enrolled 75 students in grade 7 in 2007-08, the estimated proportion of students that remain is 75%. The same calculations were made for each local school district so that we could examine the estimated attrition or drop off in enrollment with each consecutive year of a grade cohort. In addition to our analysis using Common Core of Data, we also examined data specific to the District of Columbia. The demographic data on KIPP DC schools’ and the local school district (i.e, District of Columbia Public Schools) was obtained from the following website: http://www.nclb.osse.dc.gov/. Study on KIPP revenues and expenditures. Financial data come from the NCES Common Core of Data School District Finance Survey (F-33), School Year 2007–08 (Fiscal Year 2008), the most recent year for which national school finance data are available. Spending by category is reported both as a percentage of Total Current Expenditures (TCE) and as a per-pupil amount. Twenty-five KIPP schools are included in the financial analysis. These schools include those that could be matched using the NCES Common Core of Data School District Finance Survey (F-33), and report financial data unique from their host districts. Twelve KIPP regions, groups, or individual schools that had LEA/district status had financial data reported in the federal data set. These 12 KIPP units with district status comprised 25 of the individual KIPP schools or campuses. This represents 42% of the 59 schools KIPP reported in operation during the 2007-08 school year. When results have been aggregated across districts, weighted averages based on enrollment have been used. The influence of a district’s financial results on the aggregate results is proportional to its enrollment. The same process that was used to identify host or local traditional public school districts for student characteristics was used to determine host districts for financial data. In the case of KIPP Inc. Charter in Houston, four different host districts were identified based on the geographical locations of the KIPP schools within the district. All other KIPP districts included in the financial analysis have one host district. In the NCES F-33 survey, none of the 12 KIPP districts included in this analysis reported revenues from private sources, which from various media sources we knew was not the case. In an effort to determine the amount of private contributions to KIPP schools, we performed Internet searches for reports or news coverage of donations to KIPP schools. This search led to school websites, press releases, and news articles; however, this approach did not lead to the identification of relatively small donations, and for the majority of KIPP schools included, no web presence regarding donations could be found. Because this approach was not comprehensive, we abandoned the Internet search and turned to a more systematic approach to identify private revenues. On guidestar.org, a website that collects information and tax forms on nonprofit entities, we searched for each KIPP organizational unit (region, district, or individual school) included in the analysis. Then we downloaded each district’s IRS Form 990, an annual reporting that tax-exempt organizations must file, and recorded the amount shown as “public support.” The figures reported do not include funds that were provided directly to the KIPP Foundation.
Limitations Before describing limitations of the study, it is important that we comment on obstacles we faced in analyzing and reporting findings that relate to names and definitions. First of all, depending on the data source, the names used for a KIPP region, school or campus names can vary. This made our task of merging and creating our datasets more complicated. When we refer to the demographic data we try to uniformly report to the KIPP organizational units as schools. When we present findings on finance, we used by KIPP schools and KIPP districts interchangeable. There are a number of general limitations that need to be recognized. Completeness of data. For each grade level cohort we created, we sought to include as many of the KIPP schools as possible. If a KIPP school was missing students in any of the grade levels, it was removed from the cohort. This was an important consideration since KIPP’s approach is to grow schools from below: in other words, starting a new school with one or two grades and then adding a grade each year as student progress until the desired grade configuration is complete. In any case, the number of schools included in each cohort varies, and it was never possible to include 100% of the KIPP schools in any of the grade level cohorts that we created. While data on student ethnic background and free and reduced-price lunch status were rather complete, the special education data were not. This was particularly problematic in states where charter schools are not considered Local Education Authorities (LEAs) or districts and thus did not have the legal responsibility to provide special education services. While many indicators in the Common Core of Data are reported at the building level, finance data are reported at only the district level. This has implications for this study, since in many states charters are not organized into their own districts. Instead, they have autonomy but remain legally part of a public school district for reporting purposes. NCES statistical reports on finance categorize districts in three ways: (1) districts including only individual charter schools or groups of charter schools, (2) districts with both charters and traditional public schools, or (3) districts with no charter schools at all. This categorization represents a critical obstacle to accurate comparisons of financial data, since there is no way to disaggregate data for districts containing both charters and traditional public schools. Therefore, data from such mixed districts are excluded from our analyses. Selection of comparison groups. For this study, we use each KIPP school’s local school district as a comparison group. In most instances, the local district provides a fair and sensible comparison group. We recognize, nonetheless, that large differences can exist within districts, especially large urban districts. In some cases, the district profile may not be similar to the immediate community from which a school draws its students. Capturing and measuring attrition. Because we are not working with student-level data, we cannot determine with certainty the specific attrition rates. We do estimate attrition based on the drop off in the number of students in the grade cohorts. This assumes that in the KIPP schools places are not filled after students leave. This also assumes that students do not repeat grades. Given that some KIPP schools are filling vacated places, our estimates of attrition at KIPP schools are clearly underestimates. It could be that the rate of attrition is higher, but replacement compensates for some of those leaving KIPP or district schools. There is also the general assumption that students are not held back at any point. If that assumption is incorrect, it is possible that a student may fall out of the cohort yet not leave the KIPP school; the same could
be true for a student in the local public district school system for that matter. Children could plausibly skip a grade as well, thus falling out of the cohort. Evolving and changing group of schools. Although we examine data that included 94% of the KIPP schools in operation in 2008-09 in terms of student characteristics and 42% of the schools in operation in 2007-08 in terms of the finance data, it is important to note that the network of KIPP schools is changing and growing rapidly.
Student Characteristics Cross-Sectional Look at KIPP Schools Relative to Local Districts KIPP reports that more than 95% of its students are minorities, and more than 80% qualify for free or reduced-price lunch. The KIPP 2008-2009 Annual Report Card reports that 60% of students are African American, 35% Latino, 3% Caucasian, and 2% Asian.14 Figure 1 represents the ethnic compositions of 59 KIPP schools and their local public school districts. As one can see in Figure 1, both KIPP schools and their host districts are predominantly comprised of African American and Hispanic/Latino students. The main difference between the aggregate data for the local districts and the aggregate data for KIPP is that KIPP has a higher proportion of African American students than the district (55% compared with 32%). While KIPP enrollments are 39% Hispanic, the host district enrollments are 50% Hispanic. Also, while only 2% of KIPP’s students are white, 11% of the host districts’ enrollment is white.
60% 50% 40%
Ethnicity of Students at KIPP Schools and Host Districts, 2008-09 55.4% 50.0% KIPP 38.9% 31.6%
Figure 1. Ethnic Background of KIPP Schools Compared with Local School Districts, 2008-09.
While the results in Figures 1 and 2 summarize weighted averages across all KIPP schools and all host districts, we also tallied the direct comparisons of between the schools in each KIPP region or independent school relative to the local host district (see Appendix B). In 30 of 39 comparisons between the schools in a given community compared to the local district we found that had a higher proportion of African American students. In a quarter of the comparisons, KIPP had a higher proportion of Hispanic students. As illustrated in Figure 2, the proportion of students qualifying for free or reduced-price lunch is slightly higher—by about 5.6 percentage points—at the KIPP schools than at their host districts. A side-by-side comparison of each KIPP district and its host district can be found in Appendix B. In 20 out of 36 comparisons between KIPP regions or schools and their respective local districts, KIPP had a higher proportion of students qualified for free-lunch than in their comparison local KIPP schools. Meanwhile in 27 out of 36 of the KIPP regions/schools, a higher proportion of students qualified for reduced-price lunches than in their respective comparison districts.
Students with Free and Reduced Lunch at KIPP Schools and Host Districts, 2008-09 80% 70%
KIPP Host Districts
60% 50% 40% 30%
0% Free Lunch
Figure 2. Proportion of Students Qualifying for Free and Reduced-Priced Lunch at KIPP Schools and Local Districts, 2008-09 As illustrated in Figure 3, KIPP schools enroll a smaller population of students with special needs and students classified as English Language Learners (ELL) than do their host districts (11.5% compared to 19.2% in the host districts). The percentage of students with an ELL classification is also substantially less across the board (5.9% compared with 12.1%), with the exception of the KIPP Charter school district and KIPP Aspire Academy. Since the ELL and special education data are reported at a district level, we were only able to gather this data for 20 KIPP districts and their host districts. Data for each of these KIPP regions or schools and their host districts can be found in Appendix B. Here we can see that in only 2 instances did KIPP schools have a higher proportion of ELL students than the local district. Similarly, there were also two instances where a KIPP region or school reported a higher proportion of students with IEPs relative to the local school district.
English Language Learners and Special Education, 2008-09 25% 19.2%
KIPP Host Districts 12.1%
10% 5.9% 5% 0% ELL
Students with Disabilities (IEP)
Figure 3. Percent of Students Classified as English Language Learner Status and Percent of Students with Disabilities in KIPP and Local School Districts
Distribution of Students Across Grades in 2008 The enrollment distribution of students by grade in all of the KIPP schools is depicted in four separate cross-sectional snapshots (see Figure 4). This illustrates the pattern of growth of KIPP, starting initially with a focus on middle schools, and then in the 2006-07 school year, beginning to expand to serve students in a small number of elementary and upper secondary schools. The sharp drop in enrollment by grades is a result of attrition and also due to the fact that a portion of the KIPP schools have not grown to include all grades. In our latter grade cohort analysis, we will only consider schools that have students enrolled at all grade levels considered in the cohort. Figure 4 also demonstrates that the KIPP schools are experiencing a growth spurt in number and size. Both the number of schools and the total number of students enrolled appears to be on the rise throughout the four-year span; enrollments tripled over that time, and this growth pattern has continued since. At the same time, however, individual cohorts of students have decreased; this is particularly noticeable in the sixth through eighth grade cohort from 2006-07 to 2008-09. So, from this one can see that although enrollment is on the rise, sizes of individual grade levels is decreasing in incremental yearly progressions. Figure 4 also makes more apparent the fact that KIPP schools are predominantly middle schools. They have a tendency to cater more to fifth through eighth graders rather than prekindergarten through fourth grade or ninth through 12th grade.
Total Enrollment: KIPP Schools 2006‐07
Total Enrollment: KIPP Schools 2005‐06 4,000
10 11 12
Total Enrollment: KIPP Schools 2007‐08 4,000
10 11 12
Total Enrollment: KIPP Schools 2008‐09
10 11 12
10 11 12
Figure 4. Cross-Sectional Snapshots of the Total Enrollment in KIPP Schools Across Four Years
Attrition from Grade to Grade The 2009 KIPP Annual Report Card reports that 88% of KIPP students returned to KIPP schools or finished a KIPP program in the 2008-09 school year. Because the more common natural breaking point between elementary and middle school is between fifth and the sixth grade, we designed a middle school cohort for the sixth through the eighth grades, with the eighth grade being measured from the most recent year of data provided by the Common Core of Data. The cohort contains 30 individual KIPP schools, nationwide, from 2007 to 2009 (see Figure 5). Included in this cohort is the progression of students by year, starting from the sixth grade and ending in the eighth grade. Enrollment was tracked by year as a percentage of the number of students originally enrolled. In other words, the number of students enrolled in the sixth grade represents 100% of the enrolled population. Assuming no students are replaced and none are held back, the data for each subsequent year reflects the percentage of students remaining in the cohort with respect to the number of students that originally started.
The comparison data were obtained from the Common Core of Data. The closest public school (distance-wise) was found using Greatschools.org; each school’s host district was found by using Greatschools.org as well and was also cross-checked with the Common Core of Data. Enrollment by grade was found for the district within the Common Core of Data, and data for each district was broken out in the same manner as for the KIPP schools to allow proper comparisons. Additional comparisons were made with 30 public school districts (one local public school district per KIPP school). The attrition rates for the local school districts, as with KIPP schools, were calculated in terms of the “original” enrollment, again starting with sixth grade, and employing the same assumptions about replacement and retention in grade from year to year. To generalize about the trends in detected attrition rates of KIPP schools and of their local public school districts, the total number of students enrolled by grade, per year, was summed across the entirety of the cohort for the included KIPP schools, and similarly for the local public districts. As is readily seen in Figure 5, the rate at which KIPP students appear to return from year to year seems dramatically less than that of their local public district counterparts. It would appear that approximately 15% of the originally enrolled students disappear from the KIPP cohort every year, while the local districts experience a drop of 3% per year. The same is apparent in the other two cohorts included in this study (see Appendix A), with the exception of Cohort B, as this cohort is inclusive of the fifth grade. One may observe an increase in enrollment between the fifth and sixth grade. Perhaps this is justified due to the fact that the sixth grade appears to be a more natural starting place for middle school.
Percent Change in Enrollment from Grade 6 to Grade 8 in KIPP Schools and Host Districts, 2006-07 to 2008-09 100%
60% 50% 40% 30% 20% 10% 0% KIPP
Figure 5. Estimated Attrition of Students Across the Middle School Cohorts for KIPP Schools and Local School Districts
Not all KIPP schools exhibited decreasing cohort sizes over time. In fact, there was considerable variation across the KIPP regions and schools. The data in Appendix B makes it clear that the practice of backfilling places that are vacated is a decision made at the local level. In a number of instances, 5 cases the KIPP region or school actually had more students enrolled in grade 7 than in grade 6. Furthermore, 2 of the schools showed an increase in the number of students between grades 7 and 8, which is also contrary to the overall declining enrollment trend. A more in-depth look at what appears to be happening demographically within the KIPP schools and the local districts is highlighted in Figure 6. We found that 34% of African American students at KIPP leave the cohort between the sixth and eighth grades, while 21% of Hispanic students and 26% of Caucasian students leave. By contrast, the local, public district schools experience an 11% drop-off in African American students, a 5% drop-off in Hispanic/Latino students, and a 6% drop-off in White students. As can be seen, KIPP experiences a greater proportional loss of students than its counterpart public districts, particularly in terms of African American students. Similar characteristics are found in the other two cohorts included in Appendix A, with regards to enrollment patterns from the sixth to eighth grade. Percent Change in Enrollment from Grade 6 to Grade 8 in KIPP Schools and Host Districts, 2006-07 to 2008-09 100%
100% 96% 95%
94% 94% 89%
79% 74% 66%
70% 60% 50% 40% 30% 20% 10% 0% KIPP
Figure 6. Patterns of Attrition by Ethnicity for KIPP Schools and Local School Districts An even more detailed view that breaks out the results by ethnicity and gender is included in Figure 7. It would appear that approximately 40% of African American males enrolled in the KIPP schools vanished from the original cohort enrolled in the sixth grade. For other groups, the attrition rates are as follows: 28% of African American females, 30% of Hispanic males, 12% of Hispanic females, 25% of White males, and 26% of white females. The other cohorts, included in Appendix A reflect this trend between the sixth and eighth grade.
However, Cohort C demonstrates a substantially greater drop-off in White males although this subgroup only comprised 26 boys in grade 6 in 2006-07.
100% 90% 80%
2006-07 (Grade 6) 2007-08 (Grade 7) 2008-09 (Grade 8)
70% 60% 60% 50% 40% 30% 20% 10% 0% Black Male
Figure 7. Patterns of Attrition by Gender and Ethnicity for KIPP Schools (Cohort A Results) By contrast, the local district schools lose 11% of their African American males, 8% of their African American females, 5% of their Hispanic males, 4% of their Hispanic females, 6% of their White males, and 2% of their White females between the sixth and eighth grades. In short, the local public districts are far more successful at retaining their enrolled students within the cohort. An attempt was made to track students who qualify for either free or reduced-price lunch over time. Unfortunately, the federal datasets did not break out this data by grade: instead it was only reported for the whole school. Therefore, we could not follow specific grade cohorts over time. We did, however, examine four separate cross-sectional snapshots to see if there were changes over time in the overall proportion of students served by KIPP who qualified for free or reduced-price lunch. Our review showed very minor fluctuations from year to year that did not merit reporting. Similar to the data on free and reduced-price lunch, there were not grade-specific data on English language learners (ELL) or special education. Instead we examined overall trends for all KIPP schools over a four-year period. Figure 8 illustrates the enrollment trend for students classified as ELL between 2005-06 and 2008-09. From this, one can observe that each type of school followed the same basic trend. However, the greatest difference is apparent in the percentages; KIPP schools appear to have fewer students, as a percentage, with an ELL classification than the local district schools.
Percent of Students Classified as ELL
5.6% 5% 0% 2005-06
Figure 8. Trends in the Proportion of ELL Students Served by KIPP and Local Host Districts between 2005-06 and 2008-09 (Note: This includes all KIPP schools and not only schools grouped in Cohort A) Figure 9 illustrates the enrollment patterns for students engaged in an IEP program in both KIPP and local district schools. From this one can observe two things: first, the percentage of total students who have an IEP classification in KIPP appears to be smaller across the board. Second, in the local district schools, the percentage of students who meet the IEP classification appears fairly steady from year-to-year (about 11%), while in KIPP the year-to-year trend in IEP enrollment appears to be slightly downward.
Percent of Students with Disabilities KIPP Host Districts
4% 2% 0% 2005-06
Figure 9. Trends in the Proportion of Special Education Students Served by KIPP and Local Host Districts between 2005-06 and 2008-09 (Note: This includes all KIPP schools and not only schools grouped in Cohort A)
A Closer Look at KIPP DC In addition to looking at the national Common Core of Data, we constructed similar cohorts and analyses with data from the District of Columbia. This data was newer by one year and also allowed us to look at the numbers of students classified as ELL and the number of students that qualify for special education services by grade. This analysis of KIPP DC data covered three separate KIPP schools: DC PCS Aim Academy, DC Key Academy, and DC Will Academy. Similar cohorts and analysis were also completed for the traditional public schools in the District of Columbia. The estimated attrition over the middle school grades was slightly lower than what we found in the national analysis. There was only one student in the KIPP DC schools classified as ELL, and over time this student either left the school or was reclassified as not ELL. The number of students with a special education classification was small but remained fairly steady though the duration of the cohort; by the eighth grade, there was only one fewer student than the schools started with in the sixth grade. When we examined the numbers of males and females in the KIPP DC cohort, we found that the number of male students dropped by 20%, while the number of female students dropped by just 5.8%. The findings related to student characteristics and our estimates of student attrition are discussed in the final section of this report. Before we get to the discussion and conclusion, however, we will review our findings from the analysis of data related to revenues and expenditures.
Analysis of KIPP Revenues and Expenditures In this section, we provide a comprehensive and detailed review of financial data for KIPP schools, including revenues, total current expenditures, and private revenues. While our analysis largely is based on the required reporting on the public resources received and spent on behalf of the schools, we do also attempt to capture private revenues received by KIPP districts/networks as well as those received by independent KIPP schools. The KIPP Foundation is a 501(c)3 nonprofit organization, as are the 20 regional organizations and 12 independent schools that comprised the KIPP network in 2007-08, the most recent year for which we could obtain finance data from the federal sources. The level of per-pupil funding that KIPP receives through federal, state, and local sources varies considerably across the country because of different funding formulas in different states. Our financial analysis focuses on examining and comparing the amounts and sources of revenues and expenditures for KIPP schools, other charter schools, and traditional public schools. To accomplish this we report data for four comparison groups: (1) all public schools in the country, (2) charter schools, (3) KIPP schools, and (4) KIPP host districts. As described in the methods section, financial data is only reported by school districts or local education authorities (LEAs). Since charter schools in some states do not have LEA status, and because charter school data in some states is combined together with local district finance data, we were only able to obtain data from the federal district finance dataset for 25 KIPP schools, which comprise a total of 12 KIPP regions, groups, or individual schools that have LEA or district status. It should be noted that several factors make the examination of traditional public school and charter school finance difficult. These include:
Funding formulas for both public and charter schools tend to be complex and vary considerably from state to state. Many types and sources of revenues are not easily captured, are not reported by schools and state agencies, or both. For example, schools’ general operating funds may be supplemented by allocations for capital investments, or for such supplemental services as transportation, vocational programs, or school health programs. Moreover, many charter schools secure large sums of private revenues, often kept outside the purview of analysts.15
Cost Advantages and Disadvantages All of the KIPP schools included in this analysis are charter schools. Charter proponents and opponents have aired a wide variety of arguments about whether or not the “marketplace” in which charter and traditional schools compete is a level playing field, based on assumed financial advantages or disadvantages that either type of school experiences. These contentions are worth bearing in mind when considering the comparisons in this study for two reasons. First, some of the following points may help explain or justify different findings for different types of schools. Second, and perhaps more importantly, the findings may help either support or refute the validity of common claims about charter finance and reform. Claimed Cost Advantages That Charters Offer: • • •
Charter schools’ increased autonomy—particularly in employment of teachers—permits the flexibility needed to be more responsive and cost-efficient. Charter school teachers typically receive lower salaries and fewer benefits than traditional public school teachers, saving money. Charter schools can limit enrollments to ensure an efficient match with existing facilities and instructors. For example, a charter school with four teachers can choose to admit only 100 students to ensure that each class will have 25 students. A public school with four teachers might end up with 80 students or 115 students. Adjustments can be made and more staff hired, but the teacher-student ratio in traditional public schools is not always predictable and not necessarily the most cost-efficient. Charter schools are community-based, better able to solicit in-kind contributions from families, community partners, businesses, and private organizations. Charter schools can apply for additional federal funding for start-up and implementation, and also for the dissemination of their ideas. While the possibility of additional federal monies does not make charter schools less costly to operate, it provides an opportunity not afforded traditional public schools.
While these advantages apply to KIPP, it is also true that some features of the KIPP model require additional resources: for example, the longer school day and longer school year will increase costs across all categories of spending. KIPP estimates that it costs somewhere between $1,000 and $1,500 per year per pupil to provide the extras that KIPP schools provide— including extra long school days, Saturday school, summer school, and field trips.16
Claimed Cost Disadvantages That Charters Face: • •
Most charter schools are start-up schools requiring substantial initial funding, particularly for facilities, and federal start-up grants are often insufficient to cover all such costs. Charter schools tend to be small and lack economies of scale that local school districts have. For example, when charter schools must have specialized staff, such as a certified administrator or a school nurse, the resulting cost is distributed over a smaller number of students. If a charter school is required to provide and fund transportation, it might not be able to achieve the same efficiency as district schools with more geographically concentrated students. In some states, funds that charters receive from local districts are based on spending levels in the previous year, and in some cases are not increased with inflation, even if the local district’s funds are. Furthermore, in a few states, the charter’s local funds are calculated based on what the local district spends, not what it collects. Thus, if a local district does not spend all of its funds in a given year, it pays correspondingly less to charter schools. In most states charter schools are disadvantaged when it comes to publicly funded capital resources. For example, charter schools lack access to low-interest bonds in some states to finance facilities or are not permitted to use state money to maintain their buildings, as public schools can, but instead must use operating or grant money.
Challenges to Common Claims about Lower Cost As if the oppositional claims above weren’t enough to muddy the waters, many counterarguments about charters’ presumed efficiencies demonstrate the difficulty of getting a clear picture of charter school financing. Challenges like the following illustrate the wide variety of factors that affect charter school finance: •
Lower teacher salaries are often the result not of greater efficiency but of lesser quality. While some schools may enjoy a loyal and talented staff who stay when the school simply does not have money for better salaries, it is fair to say that lower salaries often result from a lower level of qualifications—especially in years of experience—of teachers recruited by or seeking employment in charter schools. Thus, the cost advantage of lower salaries may be offset by a loss in valuable expertise, and as such they may be seen as a disadvantage rather than an advantage. Lower costs may stem from lesser services. Unlike public schools, charter schools are not obligated to provide such additional services as adult education or vocational education. Lower costs may come from greater student selectivity. With some exceptions, charter schools generally serve students who are less costly to educate than students in traditional public schools. Enrollments in charters schools are more concentrated at the elementary level, where per-pupil costs are lowest.17 Charter schools also have considerably fewer students classified as English Language Learners, fewer English-speaking students with special education needs, or both. Those students with disabilities who are enrolled in charter schools tend to have mild and less-costly-to-remediate disabilities.18 While traditional public schools do receive special education funds from state and federal sources, those seldom cover
all the costs incurred; districts thus must cover additional special education costs as part of their current operating expenses. The examples presented above underscore how complex and even confusing charter school finance can be. Nevertheless, we trust that the detailed findings presented here will add clarity to KIPP’s financial patterns as compared with their host districts, all charter schools in the country, and all public schools in the country (traditional public schools and charter schools).
Revenues This section presents findings from a comparison of (1) revenues relative to the number of students enrolled, and (2) revenue sources. It is important to reiterate that comparing only charter school revenues with those of traditional public schools can produce a misleading picture, as discussed above. Traditional public school revenues include funds for mandatory programs, like adult education, not required of charter schools; traditional public schools receive and spend substantially more on special education and student support services; some traditional public school revenues include money earmarked for transportation of district students to charter schools, private schools, or both;19 and charter schools are less likely than traditional public schools to report private revenue sources. We explore this latter point later in this section. Another caution worth reiterating is that states vary dramatically in the amount, sources, and pattern of revenues for charter schools and traditional public schools alike. Each state has a unique funding formula. Some states fund schools largely based on local taxes, while others have shifted most funding to state tax sources. States that rely on local taxes to fund schools typically have a formula that directs supplemental state revenues to districts with higher levels of poverty and a weaker tax base. In these instances, the funding mechanism requires districts to share a specific portion of local tax revenues with charter schools. State funding formulas also can vary in the degree to which they fund differentials in teacher salaries, including increments for such characteristics as advanced degrees or years of teaching experience. Further, they vary in financial support for educating students with special needs. While the accounting formulas of some states do allocate for such services,20 other states’ formulas are crude and create inherent incentives not to enroll students with special needs. The biggest difference among states relates to costs for facilities. Many states allocate separate funding for facilities or capital improvements. Charter schools have access to federal Public Charter School Program funds for start-up during the initial years of operation, although these funds are insufficient to purchase or build a new facility. While some states are generous in financing charter school facilities, others offer little or no such financing. We explore the issue of facilities in comparing expenditures later in this report. Combined revenues. All public schools, including charter schools, receive revenue from four major sources: federal, state, and local governmental sources and private sources. In many states, schools are supposed to report private revenues as a component of local revenues. We present data on public sources of revenue before moving on to private revenue. Overall, KIPP schools receive more in governmental sources of revenue than do any of the other comparison groups. On average, KIPP schools received $12,731 per pupil from federal, state, and local sources in 2007-08, while the national average for all public schools was $11,937. The average for KIPP’s host districts was $11,960 and the average for charter schools
nationally was significantly lower, $9,579. When revenues are examined district by district, differences between KIPP districts and their host districts vary. Seven of the 12 KIPP districts or groups of schools have higher combined revenues from public sources than their host districts: all five of the KIPP schools located in Texas, KIPP Endeavor in Missouri, and KIPP Charlotte in North Carolina. KIPP schools that receive less revenue than their host districts include two Team and Freedom Academies in New Jersey, KIPP DC, KIPP New Orleans, and KIPP Gaston College Preparatory in North Carolina. KIPP New Orleans and KIPP Freedom Academy in New Jersey stand out as the two KIPP schools/groups with the greatest discrepancy between their own level of funding and that of their host district. In both cases, the KIPP schools receive less than half of the perpupil revenues reported by the host district. There was no clear pattern as to whether KIPP groups/districts comprised of multiple schools fared better than single KIPP schools. Federal, State, and Local Revenue. Figure 10 shows per-pupil revenues by source for the four comparison groups. KIPP schools receive nearly twice as much in federal dollars ($1,779) than the national average ($922) or charter schools ($949). Because KIPP schools are often situated in high-need areas, the traditional public school districts in which KIPP schools operate may be the best comparison group. KIPP schools also receive more federal dollars per pupil than do their host districts ($1,332). On average, KIPP schools receive slightly less revenue per pupil from state sources ($5,581) than the national average ($5,666) and less than charter schools nationally ($6,230), but more than KIPP’s host districts ($5,164). In terms of local sources of funding, KIPP reports approximately the same per-pupil amount ($5,372) as the national average ($5,349) and their host districts ($5,464). However, KIPP schools receive more than twice as much from local sources as charter schools nationally ($2,400). In Appendix C, the revenue for each KIPP district or group of schools is compared with its respective host district. $15,000
Per Pupil Revenue by Source, 2007-2008
Per Pupil Revenues
$12,000 $9,000 $6,000 $3,000 $0
Federal State Local
USA: All Public School Districts (N=15,367)
Charter Schools (N=1,736)
KIPP Schools (N=12)
KIPP Host Districts (N=13)
$922 $5,666 $5,349
$949 $6,230 $2,400
$1,779 $5,581 $5,372
$1,332 $5,164 $5,464
Figure 10. Per Pupil Revenue Broken Out by Source for KIPP Schools and Comparisons Groups (2007-08)
Private sources of revenues. In the School District Finance Survey dataset, private revenues are considered a form of local revenues. Although some states break out revenue sources in four categories (federal, state, local, and private), states generally group private revenues with local revenues. By design, charter schools have been considered more able to attract or obtain funding from private sources because of their community roots, entrepreneurial spirit, and flexibility to create new partnerships. While several studies indicate that charter schools receive a large amount of funding from private sources,21 others find only a few charter schools receiving substantial private revenue.22 Such disparities are linked to the socio-economic status of the populations that various charters serve as well as to differences in the social capital of various charters’ founders and leaders. In a 1998 review of charter school research,23 Wells et al. noted that schools located in predominantly middle- and upper-middle-class communities tended to have easier access to financial and in-kind resources. In 2007, Miron et al.24 similarly found that charter schools serving minority and low-income families had less social capital and were less able to attract private revenues than schools serving middle-class populations. However, because of KIPP’s reported successes with achievement outcomes in such populations, KIPP schools may be exceptions to the finding. Because charter schools were designed to be entrepreneurial and because they were expected to use their autonomy and flexibility to build partnerships with diverse groups, it has been assumed that charter schools would seek private funding to bolster revenues. Dickerson, Mason, and Martucci (2000)25 identified diverse benefits for charter schools that attract private revenues. Huerta and d'Entremont (2010)26 note that charter schools can partner with foundations and businesses to build civic capacity with local community organizations and with an educational management organization to develop institutional legitimacy. Ascher et al. (2003)27 have identified risks for charters that rely on private partners. For example, funders may create budget problems by delaying or canceling anticipated revenues, or they may threaten school autonomy by attaching conditions that strain time and other school resources. Charter schools are not, however, particularly forthcoming about private revenues. Earlier state evaluations that the lead author has headed in Delaware, Connecticut, Illinois, Michigan, Ohio, and Pennsylvania28 found that fewer than half of charter schools report private revenues. In fact, it has become an increasingly common practice for charter schools to establish independent nonprofits to collect and spend private contributions on the school’s behalf. Although these funds mostly offset facility or other costs, they are not publicly reported. It can be argued that while charters may be accountable to their private contributors, as public charter schools they are also expected to be accountable to public authorities for use of these private monies—but in practice are not. This study has identified some very limited information on private contributions, listed as a sub-category of local revenues in the federal dataset. Those data indicate that the national average for per-pupil private revenue is $17, while the average for charter schools is $50 and the average for KIPP host districts is $5 per pupil. However, none of the KIPP schools examined reported any private revenues in this dataset. The KIPP Foundation’s Donations and Support department was contacted both by phone and by email and requests were made to obtain this data; KIPP did not respond to these requests, however. In order to estimate the amount of revenue KIPP schools received from individuals and philanthropic groups during 2007-08, the IRS Form 990 reports filed for each school, KIPP
regional organization, or both were downloaded from guidestar.org. The Form 990 line for “direct public support,” defined as “contributions, gifts, grants, and bequests received directly from the public; includes amounts received from individuals, trusts, corporations, estates, foundations, public charities, or raised by an outside professional fundraiser,”29 was used to calculate approximately how much KIPP schools were receiving per pupil from private sources of revenue. Financial documents for one of the KIPP districts, KIPP Southeast Houston, could not be located on guidestar.org. To err on the side of caution, we included KIPP Southeast Houston’s two schools in operation in 2007-08 under KIPP Inc. Charter, which also operates in Houston. Because of this merge, only 11 KIPP districts are included in this analysis. As illustrated in Table 1, the amount of private revenue per pupil varies widely; though 9 of 11 districts are receiving more in private dollars than the $1,000 to $1,500 KIPP estimates is required to support its model. In fact, Team Academy in New Jersey and KIPP Truth in Texas received more than $10,000 per pupil during the 2007-08 school year, even though the federal dataset reported no private revenue for these schools. In total, the private revenues for these KIPP districts amount to more than $37 million. On average, $5,760 per pupil in private revenues were received in 2007-08 by these KIPP districts. While the results in Table 1 only track the KIPP schools that had public revenues reported in the federal district finance dataset, Appendix D has a complete list of the private revenues for all KIPP districts/groups and independent schools. Here one can see that total private revenues received by KIPP schools in 2008 was $51.7 million. According to the IRS 990 Forms, these KIPP districts and schools received a total of $156,554,696 in both public and private revenues in 2008. One third, or $51.7 million came from private sources, and $104.7 million came from government sources (i.e., public revenue from local, state, or federal sources).30 Table 1. Private Revenues Received by KIPP and Reported in 990 Forms, 2007-08 Total Private 2007-08 Per Pupil KIPP Districts Revenue Enrollment Private Revenue KIPP DC PCS (District of Columbia) $3,541,058 834 $4,246 KIPP New Orleans (New Orleans, LA) $1,582,192 712 $2,222 KIPP Endeavor (Kansas City, MO) $477,681 75 $6,369 Team Academy (Newark, NJ) $8,148,163 621 $13,121 Freedom Academy (Camden, NJ) $0 257 $0 Gaston College Prep (Gaston, NC) $293,397 593 $495 KIPP Charlotte (Charlotte, NC) $285,993 96 $2,979 KIPP Inc. Charter & SE Houston (Houston, TX) $17,271,178 2,358 $7,325 KIPP Aspire (San Antonio, TX) $964,549 353 $2,732 KIPP Truth (Dallas, TX) $2,239,092 197 $11,366 KIPP Austin (Austin, TX) $2,412,688 365 $6,610 Average per Total $37,215,991 6,461 pupil =$5,760 It should be noted that monies donated directly to the KIPP Foundation are not included in this analysis. The KIPP Foundation is responsible for recruiting and training KIPP
administrators. During the 2007-08 school year, which corresponds with our analysis, the KIPP Foundation collected an additional $15,320,750 in private revenue, according to its IRS Form 990. Since 2007-08, philanthropic and government grants to the KIPP Foundation have grown in size and even include a $50 million grant from the US. Department of Education in 2010. The KIPP Foundation, based in San Francisco, California, also contributes to some of the administrative costs in the schools. It does this by paying salaries of intern administrators and by paying for pre-service and in-service training of administrators.31 On its website, the KIPP Foundation reports that it spent the equivalent of $1,175 per pupil. Most of this was spend on leadership development, but KIPP’s annual report also indicates that the KIPP Foundation devoted resources for such things as grants to schools, school support services, and school start-up support.32
Expenditures In line with common practice among researchers who compare financial data across districts and states, this study also examines spending across diverse categories as a proportion of total current expenditures (TCE). TCE excludes capital outlay, which can increase and decrease dramatically from year to year. It also typically limits data to expenditures on elementary and secondary education, excluding such services as adult education and community services that are often neither required of nor offered by charter schools. As a whole, KIPP schools spend more per pupil in total current expenditures ($10,558) than do other charter schools ($8,492), and slightly more than their host districts ($10,101) and the national average ($10,121). When we look at total current expenditures per pupil for each KIPP district, 8 of the 12 districts actually reported lower total current expenditures than their host districts. Four districts or groups of KIPP schools (i.e., KIPP Charlotte, KIPP Truth, KIPP Austin Public Schools Inc., and the largest KIPP district, KIPP Inc. Charter) reported higher total current expenditures than their host districts. As with revenues, there seems to be no pattern as to whether KIPP groups/districts spend more or less compared with single KIPP schools. Our findings related to revenues indicated that KIPP schools were receiving more in public and private sources of revenues than local districts. In total, our estimate is that KIPP schools receive about $6,500 more in revenues per pupil. Our analysis of expenditures shows that KIPP is reporting spending only about $450 more per pupil than local districts. This raises the question about what is happening to the rest of the revenues. Our guess is that the KIPP schools are not reporting how the private revenues are being spent, just as they did not report receiving private revenues within the regular public school finance accountability system. The NCES Common Core of Data School District Finance Survey contains 68 indicators related to expenditures. We have grouped these indicators into four categories: (1) instruction and instruction-related activities, (2) student support services, (3) administration, and (4) operations. Differences among the comparison groups in terms of spending on these four major categories are illustrated in Table 2 and Figure 11 below. In Appendix C, we compare each KIPP district’s spending on these four categories with KIPP’s host districts’ spending. NCES’ School District Finance Survey defines instruction expenditure as …payments from all funds for salaries, employee benefits, supplies, materials, and contractual services for elementary/secondary instruction; excludes capital outlay, debt service, and interfund transfers for elementary/secondary instruction.
Instruction covers regular, special, and vocational programs offered in both the regular school year and summer school; excludes instructional support activities as well as adult education and community services (p. B-5). KIPP spends more on instruction ($5,662 per pupil) than other charter schools ($4,617) but less than the national average ($6,196) or KIPP host districts ($5,972). When per-pupil amounts are converted into percentage of total current expenditures, KIPP schools spend less on instruction than any comparison group. When we look at spending on instruction at the district level, 9 of the 12 KIPP districts spend less per pupil on instruction than their host districts. The remaining three districts that reported higher spending on instruction than their host districts, KIPP Inc. Charter, KIPP Aspire, and KIPP Austin Public Schools, are all located in Texas. Table 2. Break Out of Expenditures Across Four Broad Categories of Spending
USA: All Public School Districts (N=15,367) Charter Schools (N=1,736) KIPP Districts (N=12) KIPP Host Districts (N=13)
Instruction & Instructionrelated
Student support services
Instruction & Instructionrelated
Student support services
Spending on student support services is comprised of pupil support services and instructional staff support services. NCES’ School District Finance Survey defines pupil support services as, “expenditure for attendance record keeping, social work, student accounting, counseling, student appraisal, record maintenance, and placement services. This category also includes medical, dental, nursing, psychological, and speech services” (p. B-10). KIPP’s perpupil spending on student support services ($460) is comparable to that of charter schools nationally ($464), but much less than the national average ($1,003), and KIPP’s host districts ($1,179). This may reflect that charter schools and KIPP serve different types of students than do traditional public schools. When we look at spending on student support services district by district, KIPP DC and KIPP Team Academy in New Jersey reported no spending in student support services, which decreased the aggregate spending reported. Even so, nine of the 10 remaining KIPP schools spent less than their host districts on student support services. KIPP Inc. Charter of Houston reported spending more per pupil on student support services than one of its host districts, Alief ISD, but less than its other two host districts, North Forest ISD and Houston ISD.
Total Current Expenditures Per Pupil
Total Current Expenditures Per Pupil Across Four Main Categories of Spending 2007-2008
$12,000 $9,000 $6,000 $3,000 $0
Operations Administration Student support services Instruction and instruction-related
USA: All Public School Districts (N=15,367)
Charter Schools (N=1,736)
KIPP Schools (N=12)
KIPP Host Districts (N=13)
$2,176 $746 $1,003
$2,074 $1,336 $464
$3,464 $972 $460
$2,264 $687 $1,179
Figure 11. Expenditures Broken Out Across Four Main Categories of Spending. 2007-08 The administration category is made up of school administration costs, defined by NCES as “expenditure for the office of the principal services” (p. B-10), and general administration, defined as “expenditure for board of education and executive administration (office of the superintendent) services” (p. E-8). KIPP spending on administration ($972) is more than the national average ($746) or KIPP host districts ($687), but is significantly lower than the charter school average ($1,336). While KIPP schools generally lack the economies of scale of a fullsized district, the majority of KIPP schools are located in districts with multiple schools. Therefore, administration costs can be spread over more students than a typical independent charter school, but can be spread over a smaller number of students than in a typical traditional public school district. When examined at a district level, once again KIPP DC reported no spending in this category. Of the remaining 11 districts that reported data, eight reported spending more dollars per pupil on administration than their host districts. Three districts, KIPP New Orleans, KIPP Gaston in North Carolina, and KIPP Southeast Houston, reported spending less per pupil on administrative costs. The final category of total current expenditures is operations. Operations includes spending for the operation of buildings, the care and upkeep of grounds and equipment, vehicle operation, student transportation, food services, maintenance, security, and enterprise operations (activities financed at least in part by user charges, for example). Unlike administration, KIPP spends more on operations per pupil in dollars and as a percentage of total current expenditures than any other comparison group. Eight of 12 KIPP districts included in this analysis spent more on operations than their host districts.
The operations category is further broken down into categories of (1) operation and maintenance of plant, (2) student transportation, (3) other support services, (4) food services, and (5) enterprise operations. From this level, we can see that KIPP’s additional spending is focused in transportation, food services, and other support services. “Other support services” is defined by NCES as “expenditure for business support, central support, and other support services.” Business support services include payments for fiscal services (budgeting, receiving and disbursing funds, payroll, internal auditing, and accounting), purchasing, warehousing, supply distribution, printing, publishing, and duplicating services. Central support services include planning, research, development, and evaluation services. They also include information services, staff services (recruitment, staff accounting, non-instructional in-service training, staff health services), and data processing services” (p. B-9). Given KIPP’s longer school day and school year, it is not surprising the KIPP schools would need to spend more on transportation and food services than traditional public schools. Greater spending in other support services may indicate fees for business support from KIPP network and/or it may be due to larger investments in research and evaluation of its schools relevant to the comparison groups. Spending on salaries. Table 3 describes the patterns of spending on salaries in dollars per pupil. KIPP schools spends less on salaries for instruction, $3,671 compared with the national average, $4,140, or their host districts, $4,299. However, KIPP schools spend significantly more per pupil on instruction salaries than the average of all charter schools ($2,429). As a percentage of the total current expenditures, KIPP schools spend more (34.8%) than do other charter schools (28.6%) but substantially less than the national average (40.9%) or KIPP’s host districts (42.6%). When spending on salaries is examined district by district, 11 of 12 KIPP districts spend less per pupil on salaries. KIPP DC is the only KIPP district that reported spending more per pupil on instruction salaries than its host district. Not surprisingly, the same pattern emerges when examining employee benefits. Eleven of the 12 KIPP districts spend less on employee benefits than do their host districts, and KIPP DC is the only exception. Table 3. Per-Pupil Spending on Salaries Across Key Categories of Staff
USA: All Public Schools (N=15,367) Charter Schools (N=1,736) KIPP Districts (N=12) KIPP Host Districts (N=13)
Regular Education Teachers
Salaries Specia Education Teachers
Support ServicesSchool Admin.
Benefits All BenefitsEmployees Instruction Staff
Also a notable finding, KIPP spends substantially less per pupil on special education teachers’ salaries than do any other comparison group, including the national average for charter schools. Each KIPP district’s salaries is compared with its host district(s) in the report’s appendices. In the appendices, we see that none of the districts located in Texas (host district or
KIPP district) report any spending on special education salaries. In addition to the Texas districts, KIPP DC, KIPP Endeavor (MO), KIPP Team (NJ), KIPP Freedom (NJ), and KIPP Charlotte (NC) reported no special education salary data. It is unknown whether these districts actually spent no money on special education salaries or whether they reported this in another category. Because of the missing data, one should be careful in drawing conclusions on spending differences on special education teacher salaries. Our finding that KIPP spends substantially less on teacher salaries relative to the host districts is surprising given that KIPP has a longer school day, plus Saturday school twice a month, and an overall longer school year. Nevertheless, the KIPP work force is generally younger and has fewer years of experience; therefore, the average KIPP teacher is lower on the pay scale than the average teacher in the surrounding public school district.33
Conclusion and Discussion How and Why KIPP is Successful at Improving Student Performance The findings in this study provide new insights into how KIPP schools are able to raise student results on standardized tests. One factor is selective entry of students; another is a high rate of student attrition. A third factor is the high levels of funding that KIPP schools receive from both public and private sources. Selective Entry Although KIPP enrolls large numbers of minority students and students from low-income families, the findings in our report show that students with disabilities and students classified as English language learners are greatly underrepresented. An earlier study by Carnoy et al. (2006)34 showed that substantial self-selection of families results in differences between KIPP and local school districts in terms of student and family characteristics. There is no evidence that KIPP purposely selects which students to enroll, however, decisions about location, and marketing of a rigorous college prep model promotes self-selection by families. Selective entry aids the performance of KIPP schools in two ways. First the relative absence of students with disabilities and English language learners provides more homogenous classrooms of learners. Secondly, in both traditional public schools and KIPP schools, the additional costs for these students—especially students with moderate or severe disabilities—is typically not fully funded, and therefore some of the costs for regular education is devoted to students requiring additional remediation. Because traditional public schools have a higher proportion of students with disabilities, and a higher concentration of students with severe and moderate disabilities, the burden of having to subsidize their education falls more heavily on them. This results in a cost advantage for KIPP schools. High Rate of Attrition with Non-Replacement KIPP Schools also benefit from its well-documented practice of not replacing the large numbers of students who leave its schools during and between school years. This practice benefits KIPP in a number of ways.
Selective attrition of low-performing students. The departure of low-performing students helps KIPP improve its aggregate results. Considerable research has shown that highly mobile students are much more likely to be lower-performing students. Specific to KIPP, this was verified by the Mathematica (2010) 35 evaluation, which noted that KIPP had attrition rates similar to those of the local districts and that the performance of students leaving was lower than average. What the Mathematica report overlooked, however, was that while both KIPP and the local district schools were losing lower-performing students due to attrition, the resulting vacancies were filled in the traditional public schools with low-performing students leaving KIPP or other schools; at the same time KIPP was not replacing its departing students. Retention of funds for students that leave after autumn head count. Students who leave after autumn head count (typically 4 weeks into the school year), have an impact on performance of the school as well as its economic bottom line. In most states, when students leave charter schools after the autumn head count, the funding allocated to the charter school remains for the school year, even though these students return to the district, another charter, or a private school, none of which will receive funding for the students for that particular school year. In other words, KIPP retains all or most of the money allocated for educating that student, even while the student has returned to traditional public schools or another school to receive instruction. Traditional public schools do not typically benefit from this practice, since places vacated are then filled by other mobile students, even in the middle of the school year. KIPP’s rather common practice of not filling empty places left vacant during or between academic years can positively and negatively affect school funding. Although KIPP can gain resources for students who leave after the autumn head count, its practice of not replacing students the following year may result in class sizes that are not optimally cost-efficient. Growing from below. For KIPP, not replacing students who leave also helps to maintain the school culture being established by the school. Filling empty places in the upper grades means new students would be entering with new values, expectations, etc. This is a common burden for classroom teachers in traditional public schools that KIPP teachers do not face after the initial entry year. Since admitting students mid-year would be disruptive to learning and the norms being set in the school. Of course this is a great disadvantage to the surrounding traditional public schools, since they are losing and receiving students throughout the year. Peer effect. There is a broad body of research that has confirmed the importance of peer effects on student performance.36 Essentially, a student enrolled with a group of engaged and high performing students is more likely to also perform well. In their international study of public and private schools, Zimmer and Toma (2000)37 found that peer effects were a significant determinant of educational achievement. They also found that the effects of peers appear to be greater for low-ability students than for high-ability students. Hanushek, Kain, Markman, and Rivkin (2003)38 found that peers had a positive effect on achievement growth: students throughout the school test score distribution appeared to benefit from higher-achieving schoolmates. KIPP’s practices that result in selective entry and exit result in homogeneous groups of students that mutually benefit from peers that are engaged, have supportive families, and are willing and able to work hard in school.
Higher Levels of Funding KIPP receives more in public revenues and it receives substantially more in private revenues. This allows for a more lengthy program with longer school days, Saturday instruction twice a month, and a longer school year. The fact that KIPP receives so much money per pupil ($6,500 per pupil more than local districts and some of the KIPP schools receive more than $10,000 per pupil in additional private revenues) is rather surprising. This amount does not include funding received or spent by the KIPP Foundation on behalf of its schools. While it is clear that the KIPP Foundation spends its resources on developing new leaders, marketing, and expanding the network, it is not clear that the local regions and schools are expected to devote a portion or their resources for such activities. The federal dataset that covers how the public or government revenues are spent by public schools, apparently does not include details on how KIPP is spending its privately raised revenues. This is a topic that needs to be further explored. The extra public and private revenues that KIPP receives relative to KIPP’s local school districts suggests an important factor that can explain the higher performance level of students enrolled in KIPP. Interesting as this finding appears, our analysis of expenditures indicated that KIPP is still spending less per pupil on instruction than local school districts, even while it receives more in both public and private revenues. In addition to the advantage of receiving more in revenues, KIPP also has additional cost advantages relative to the traditional public schools based on the students it serves: • KIPP has considerably fewer students with disabilities and students classified as English language learners. • Except for a few of its schools, it does not provide high school grades, where costs are higher per pupil. • KIPP does not have technical or vocational education programs, which are costly to provide. On the other side of the cost advantage/disadvantage issue, KIPP does require more resources for its program, which includes a longer school day and longer school year. Although KIPP does not spend more than the local districts on instruction, it does spend more per pupil on transportation and food service relative to the local districts and other charter schools. These costs represent its effort to facilitate involvement from lower-income families that are more dependent on transportation as well as on the existence of food services so that the school can participate in the federally funded subsidized lunch program. A number of for-profit EMOs spend little to nothing on transportation and food services, making their schools less attractive to lower income families.39
Waiting Lists at KIPP As the KIPP website states and as the media regularly reports, there is a high demand for KIPP schools, evidenced by long waiting lists. The co-founder of KIPP, Mike Feinberg stated in December 2010 that KIPP schools in Houston area had a waiting list of 3,400.40 He also indicated that this was “a heatbreaking number.” In October 2010, a news story indicated that the KIPP schools in Austin, Texas had a waiting list of 300 students.41
If the demand is so high, why do few KIPP schools choose to fill the empty places created by students who flee during and between school years? Traditional public schools are expected and obliged to fill places when students turn up at their doors. If the same is not required of KIPP and if KIPP is making it a practice to only serve students who have the ability and supportive families to handle its rigorous program, then perhaps it should not be receiving more in public dollars per pupil than traditional public schools. Perhaps KIPP’s reported demand for places is overstated, since by not filling vacancies, it forces more students to remain on the waiting list when places are available. Still, KIPP’s approach to growing its schooling from below (i.e., by only admitting students in the lower grades) is a sensible business approach used by many of the for-profit education management organizations (Miron & Nelson, 2002).
Questions for the Mathematica Evaluation KIPP has commissioned Mathematica Policy Research to conduct a multiyear evaluation of KIPP schools. To facilitate this process, KIPP has provided extensive student level data to the evaluators, and this data has also been provided by several of the local host districts. The initial report prepared by Mathematica involved a comprehensive technical report that carefully detailed the methods. While the methods and data sources were suitable for the evaluation questions being addressed, it is clear that other important questions are not being asked— questions that would require student-level data sets such as Mathematica now possesses. Below we list and briefly discuss some additional questions that we would hope that KIPP and Mathematica will consider in subsequent years of this ongoing evaluation. Student attrition and replacement • How many students leave during the academic year, and what portion of these depart after the autumn headcount? • Does KIPP admit any students after the autumn head count? If so, what are KIPP’s practices concerning how it fills places? • How many of the students leaving KIPP are replaced between school years and why does this practice vary so much by region and school? • What are the characteristics of the small number of students that KIPP invites in after the initial entry grade? Grade retention. While KIPP’s position to retain students who are not making sufficient progress can be seen as positive policy that applies “tough love,” it is likely that many of these students leave KIPP for another school or else drop out of schooling altogether. Because of the importance of understanding whether KIPP’s grade retention practices are benefiting the students or inducing them to leave, we suggest the following questions be addressed: • What are the characteristics of students who are retained or not promoted to the next grade? • What happens to the students who are retained? What proportion of students retained accept this decision and repeat a grade and what proportion leave the school?
By asking Mathematica to address these questions and to openly share the results, KIPP would be providing valuable information to policymakers and others that are interested in KIPP as a model for public education.
Policy Implications KIPP has demonstrable success in raising the performance of the students who remain in its schools. Based on the evidence presented in this report, however, it is important that policymakers and school boards take a closer look at the KIPP model before becoming advocates for it, before allocating further financial resources, and before signing new contracts to expand the number of KIPP schools. This study provides a better understanding of two critical inputs, students and funding, which help answer questions related to how KIPP functions and how it succeeds in raising the achievement levels of students who persist in its schools. While attention has been focused on the preparation and role of administrators in the KIPP model, we have just begun to explore the role and importance of teachers in it.42 We still need to know more about KIPP’s efforts to recruit and retain teachers as well as instructional practices in its schools. Before KIPP can be considered a model to be widely replicated, it has to be committed to serving all the students it admits and to serving a portion of the students who are mobile, including those who require a place in the middle of the school year, after the cut-off time for public funding to follow the student. Furthermore, to be considered a viable contributor to a system of public schools, KIPP also needs to recruit and serve a reasonable share of students who are more costly to educate, especially students with disabilities and students who are not native English speakers. The limited range of students that KIPP serves, its inability to serve all students who enter, and its dependence on local traditional public schools to receive and serve the droves of students who leave, all speak loudly to the limitations of this model. Furthermore the funding KIPP receives from public and private sources—more than $6,500 more per pupil in addition to what local school districts receive—is not likely to be sustainable in the longer run. Although many assume that KIPP wishes or intends to expand and become a fixture in the public school system, this is not necessarily the case. Even with the extensive private resources its garners, KIPP is insistent that it will not take over existing schools that are struggling. Instead, it intends to remain focused on new start-ups. KIPP’s only effort to take over a traditional public school—with a representative range of students and with the responsibility to serve all students that come and go during and between school years—ended in failure after only 2 years. This short-lived experiment with Cole Middle School in Denver43 speaks loudly about the viability of the KIPP model for public schools. Given KIPP’s reputation, many in Detroit were hoping that KIPP would bid to take over some of the 41 schools that are likely to be turned over to private management companies. KIPP rejected this opportunity and stated that it wishes to focus on new start ups.44 New start up charter schools, of course benefit from selective entry and selective exit (i.e., charter schools are not required to fill places of students who flee during and between school years). Even though the KIPP model may not be replicable on a larger scale, its example does serve as a lever for change. The existence of KIPP schools has pushed the conversation about the value and importance of more instructional time for low-income students. Similarly KIPP’s
practice of recruiting and preparing administrators who can lead urban schools is another aspect of KIPP that is changing thinking about our public schools in a positive way. KIPP states that it is aware of pressing issues such as high levels of student attrition and hopes to address these. If KIPP wishes to maintain its status as an exemplar of private management of public schools, rather than a new effort to privatize public schools, it will need to convince policymakers and the public that it intends to recruit and serve a wider range of students and that it will be able to so with sustainable levels of funding comparable to what other traditional public schools receive.
Appendices Appendix A. Detailed Results from Grade Cohort B and C Analyses
Appendix B. Student Enrollment Data for KIPP Schools, or Groups of Schools Sorted by State
Appendix C. Analysis of Revenues and Expenditures Broken Out for Each KIPP School or Regional Group of Schools
Appendix D. KIPP Revenues from Private Sources Broken Out for Each KIPP School or Regional Group of Schools
KIPP Foundation. (2010, September 9). KIPP secures $10 million in matching funds for federal grant. Retrieved February 28, 2010, from http://www.kipp.org/news/kipp-secures-10-million-in-matching-funds-forfederal-grant 2 Mathematica Policy Research. (2010, June). Student characteristics and achievement in 22 KIPP middle schools. Princeton: Author. 3 Thernstrom, A., & Thernstrom, S. (2003). No excuses: Closing the racial gap in learning. New York: Simon & Schuster. 4 Angrist, J. D., Synarski, S. M., Kane, T. J., Pathak, P. A., & Walters, C. R. (2010). Inputs and impacts in charter schools: KIPP Lynn. American Economic Review: Papers & Proceedings 100, 1-5. 5 Angrist, J. D., Synarski, S. M., Kane, T. J., Pathak, P. A., & Walters, C. R. (2010). Inputs and impacts in charter schools: KIPP Lynn. American Economic Review: Papers & Proceedings 100, 1-5. Doran, H.C., & Drury, D.W. (2002). Evaluating Success: KIPP Educational Program Evaluation. Alexandria, VA: New American Schools. Mac Iver, M.A., & Farley-Ripple, E. (2007). The Baltimore KIPP Ujima Village Academy, 2002-2006: A Longitudinal Analysis of Student Outcomes: Johns Hopkins University Center for Social Organization of Schools, prepared for The Abell Foundation. Mathematica Policy Research. (2010, June). Student characteristics and achievement in 22 KIPP middle schools. Princeton: Author. SRI International. (2008). San Francisco bay area KIPP schools: A study of early implementation and achievement. Final report. Menlo Park, CA: Author. 6 Doran, H.C., & Drury, D.W. (2002). Evaluating Success: KIPP Educational Program Evaluation. Alexandria, VA: New American Schools. 7 Carnoy, M., Jacobsen, R., Mishel, L., & Rothstein, R. (2005). The charter school dust-up: Examining the evidence on enrollment and achievement. New York: Teachers College Press. 8 SRI International. (2008). San Francisco bay area KIPP schools: A study of early implementation and achievement. Final report. Menlo Park, CA: Author. 9 Mac Iver, M. A., & Farley-Ripple, E. (2007, June). The Baltimore KIPP Ujima Village Academy, 2002-2006: A longitudinal analysis of student outcomes. Baltimore: The Abell Foundation. 10 Henig, J. R. (2008). What do we know about the outcomes of KIPP schools? (EPIC/EPRU Policy Brief).Retrieved March 3, 2009, from http://www.greatlakescenter.org/docs/Policy_Briefs/Henig_Kipp.pdf 11 National Education Policy Center (2010). New KIPP study underestimates attrition effects: Review of Mathematica evaluation of KIPP Schools. Retrieved February 5, 2011, from http://nepc.colorado.edu/newsletter/2010/06/new-kipp-study-underestimates-attrition-effects-0 Kalenberg, R. (2011). “Myths and realities about KIPP.” Answer Sheet Blog Retrieved January 10, 2011 http://voices.washingtonpost.com/answer-sheet/charter-schools/myths-and-realities-about-kipp.html 12 Mathematica Policy Research. (2010, June). Student characteristics and achievement in 22 KIPP middle schools. Princeton: Author. 13 For more information about the Common Core of Data see http://nces.ed.gov/ccd/ 14 Kipp Foundation. (2010). Report Card, 2009. Retrieved February 27, 2011 at http://kipp.org/00/docs/KIPP_ReportCard_2009/KIPP_ReportCard2009.zip 15
This is possible since private funds are not incorporated in state purchasing and accounting systems. Furthermore, private funds are often held and spent on behalf of the charter school by a trust or foundation set up to serve the school or to secure a facility. 16 Mathews, J. (2008, April 20). New report from KIPP charters. The Washington Post. Retrieved March 7, 2011, from http://www.kippla.org/news/1386/233/no 17 Miron & Nelson (2002) estimated that high schools had per pupil costs that were $750, on average, higher than elementary schools. This is due to demands on the high schools to provide vocational lines as well as the fact that the teacher-student ration at high schools is less cost efficient, in part, due to the need for more single
subject certified teachers. High schools are also more likely to offer sports and other extra curricular programs. See Miron, G. & Nelson, C. (2002). What’s public about charter schools? Lessons learned about choice and accountability. Thousand Oaks, CA: Corwin Press, Inc. 18 Miron, G., Urschel, J. L., Mathis, W, J., & Tornquist, E. (2010). Schools without Diversity: Education Management Organizations, Charter Schools and the Demographic Stratification of the American School System. Boulder and Tempe: Education and the Public Interest Center & Education Policy Research Unit. Retrieved February 15, 2010, from http://epicpolicy.org/publication/schools-without-diversity Howe, K. R., & Welner, K. G. (2002). School choice and the pressure to perform: Déjà vu for children with disabilities? Remedial and Special Education , 23(4). 212-222. Fiore, T. A., Harwell, L. M., Blackorby, J., & Finnigan, K.S. (2000). Charter schools and students with disabilities: A national study (Final Report). U.S. Department of Education. Miron, G. & Nelson, C. (2002). What’s public about charter schools? Lessons learned about choice and accountability. Thousand Oaks, CA: Corwin Press, Inc. 19 For example, in a handful of states, such as Connecticut and Illinois, a large portion of the costs for special education services provided by the charter schools is actually paid by local districts. 20 For example, in Ohio, the funding formula provides supplemental funding for children with disabilities. There are actually 7 variable levels of supplemental funding based on the estimated costs for students based on the nature of their disability and the commonly recognized mean costs for providing education for students in this category. 21 Mayor’s Office, City of Indianapolis. (2007). Accountability report on mayor-sponsored charter schools. Indianapolis: Mayor's Office. Nelson, F.H., Rubenstein, M., Mahoney, L. (2004). Financing autonomy: Charter school finance case studies. Washington, DC: American Federation of Teachers. 22 Akey, T., Plucker, J.A., Hansen, J.A., Michael, R., Branon, S., Fagen, R., & Zhou, G. (2008). Study of the effectiveness and efficiency of charter schools in Indiana. Bloomington, IN: Center for Evaluation & Educational Policy. 23 Wells, A.S., Artiles, L., Carnochan, S., Cooper, C.W., Grutzik, C. Holme, J., Lopez, A., Scott, J., Slayton, J., & Vasudeva, A. (1998). Beyond the rhetoric of charter school reform: A study of ten California school districts. Los Angeles: UCLA. 24 Miron, G., Cullen, A., Applegate, B., & Farrell, P., (2007). Evaluation of the Delaware charter school reform. Kalamazoo: The Evaluation Center, Western Michigan University. 25 Dickerson, J.F., Mason, J., & Martucci, F.J. (2000). The promise and risk of charter school project financing. New York: Fitch IBCA 26 Huerta, L. & d'Entremont, C. (2010). Charter school funding in New York: Seeking Institutional Legitimacy in a Marketplace of Resources. In Lubienski, C. & Weitzel, P. (eds.) The charter school experiment: Expectations, evidence, and implications. Cambridge, MA: Harvard University Press. 27 Ascher, C., Echazarreta, J., Jacobowitz, R., McBride, Y., & Troy, T., (2003). Governance and administrative infrastructure in New York City charter schools. New York: Institute for Education and Social Policy, Steinhardt School of Education, New York University. 28 Miron, G. & Nelson, C. (2002). What’s public about charter schools? Lessons learned about choice and accountability. Thousand Oaks, CA: Corwin Press. Miron, G., Cullen, A., Applegate, B., & Farrell, P. (2007). Evaluation of the Delaware charter school reform: Final Report. Dover, DE: Delaware State Board of Education. Miron, G. (2005). Evaluating the performance of charter schools in Connecticut: A report commissioned by ConnCAN. New Haven, CT: Connecticut Coalition for Achievement Now. Sullens, C. & Miron, G. (2005). Challenges of starting and operating charter schools: A multicase study. Cleveland, OH: The Cleveland Foundation. Miron, G. & Horn, J. (2002). Evaluation of Connecticut charter schools and charter school initiative: Final report. Hartford, CT:, Connecticut State Department of Education. Miron, G., Nelson, C., & Risely, J. (2002). Strengthening Pennsylvania’s charter school reform: Findings from the statewide evaluation and discussion of relevant policy issues. Harrisburg, PA: Pennsylvania Department of Education.
Nelson, C. & Miron, G. (2002). The evaluation of Illinois charter school reform: Final report. Springfield, IL: Illinois State Board of Education. Horn, J. & Miron, G. (2000). An evaluation of the Michigan charter school initiative: Performance, accountability, and impact. Lansing, MI: Michigan Department of Education. These and other related reports can be downloaded at http://homepages.wmich.edu/~miron/publics.html. 29 Guidestar. (2011). Glossary. Retrieved March 8, 2011, from http://www2.guidestar.org/rxg/help/glossary.aspx 30 We also examined the IRS 990 forms for a sample of Foundations or Nonprofit groups whose purpose is to support one of the KIPP host districts. The per pupil revenues for these nonprofit organizations who were serving the host district typically totaled between $1.50 and $35 per pupil. 31 Phone call with Maggie Mangen from KIPP Foundation. March 23, 2011. 32 Kipp Foundation. (2010). Report Card, 2009. Retrieved February 27, 2011 at http://kipp.org/00/docs/KIPP_ReportCard_2009/KIPP_ReportCard2009.zip 33 Wilder and Jacobsen (2010) interviewed KIPP teachers and found them to be you and devoted but nearly all of them reporting that they were likely to leave KIPP within a year or two. Wilder, T., & Jacobsen, R. (2010). The short supply of saints: Limits on replication of models that ‘‘Beat the Odds’’ Reading & writing quarterly, 26:237–263. 34 Carnoy, M., Jacobsen, R., Mishel, L., & Rothstein, R. (2006). Worth the price? Weighing the evidence on charter school achievement. Education Finance and Policy, 1, 151-161. 35 Mathematica Policy Research. (2010, June). Student characteristics and achievement in 22 KIPP middle schools. Princeton: Author. 36 Burke, M. A., & Sass, T. R. (2008, June). Classroom peer effects and student achievement. Working paper, National Center of Longitudinal Data in Education Research. Sacerdote, B. (2001). Peer effects with random assignment: Results for Dartmouth roommates. The Quarterly Journal of Economics, 116(2), 681-704. 37 Zimmer, R. W., & Toma, E. F. (2000). Peer effects in private and public schools across countries. Journal of Policy Analysis and Management, 19(1). 38 Hanushek, E. A., Kain, J. F., Markman, J. M., & Rivkin, S. G. (2003). Does peer ability affect student achievement? Journal of Applied Econometrics, 18(5), 527–544. 39 Miron, G. & Nelson, C. (2002). What’s public about charter schools? Lessons learned about choice and accountability. Thousand Oaks, CA: Corwin Press, Inc. 40 “This year, after we filled the seats in KIPP Houston schools through an open lottery, we still had 3,400 students on the wait list. That's a heartbreaking number.” Feinberg, Mike. (2010, December 4). Charter schools can't do it alone. Houston Chronicle. Retrieved from http://www.kipp.org/news/houston-chronicle-houston-tx-charter-schools-can-t-do-it-alone 41
Zaragoza, S. (2010, October 22). Austin charter school to expand as success brings increased demand. Austin Business Journal. Retrieved from http://www.bizjournals.com/austin/print-edition/2010/10/22/charterschool-to-expand-as-success.html. 42 Wilder, T., & Jacobsen, R. (2010). The short supply of saints: Limits on replication of models that ‘‘Beat the Odds’’ Reading & writing quarterly, 26:237–263. 43 Smarick, A. (2010). The turnaround fallacy. Education next. 10 (1). Retrieved January 25, 2011 http://educationnext.org/the-turnaround-fallacy. Colorado Charter Schools (2011). What's Wrong with the Turnaround Model. Blog dated January 1, 2011. Retrieved January 25, 2011. http://coloradocharters.blogspot.com/2011_01_01_archive.html. Wilder, T., & Jacobsen, T. (2010). The short supply of saints: Limits on replication of models that ‘‘Beat the Odds’’ Reading & writing quarterly, 26:237–263. 44 Zehr, M A. (2011). New Urban Playbook: Hand Over Schools to Charter Operators. Education Week, Online edition (March 21, 2011).