School Choice and School Competition: Evidence from the United States

The motivation of assessing the first generation school choice programs in the United States from 1988 to 1994 is the upcoming activity of the school choice legislation due to the June 2002 decision of the United States Supreme Court. The Supreme Court endorsed the constitutionality of a voucher program in Cleveland, Ohio that also consisted of private schools having religious affiliation. Also, the Court listed the stipulation in which it would be able to find future programs to be constitutional – the school choice program must not be drafted in such a way that a pupil who opted to pick a secular school was provoked to go to a religious school. Since school choice plans under consideration must satisfy this condition, there was a pile up of several years of legislation.

Given that the second-generation school choice programs are in an active planning mode, it is logical to determine what have been learned from the first generation, and explain the attributes needed for the second-generation program for the learning to develop further.

The three questions the study attempts to answers are:

1. Does students’ performance increase when they go to voucher or charter schools?

2. Do public schools react constructively to competition stimulated by school choice, through increasing their own efficiency? 

3. Do voucher and charter schools wind up with variety of better pupils?

After dealing with these queries through relying on credible empirical methods, five simple and easily implemented aspects the second-generation programs must have are: full financing, transferability of compensatory funding, integration with school finance, universality, and integration with accountability systems. Although having these features doesn’t make the program optimal, it somewhat eliminates the most disabling problems linked with first-generation designs. The frequent poor planning of first-generation programs is due to distressed compromise between opponents and proponents of school choice. Although the opponents of school choice, who attempt to load programs with crippling features, often succeed, there are still proponents who are successful.

I. Does students’ achievement rise when they attend voucher or charter schools?

This question is essentially obstinate for two reasons. First, the concern should be with regards to productivity of and not achievement at schools regardless of the available resources. If school choice is to be public policy, then the concern would be whether pupil’s achievement will increase when they go to voucher or charter schools than to regular public schools, with all else constant (resources, students’ skills, motivation, and other attributes). Unfortunately, since no private schools taking part of voucher program and few charter schools have equivalent resources to that of public schools, the study is forced to center on the achievement concern without holding the resources equal.

Second, the economic theory expects that school choice will increase the productivity of the public schools competing with voucher or charter schools. In the face of competition, the best response function of public school is to move toward better productivity and less rents, such as salaries. The main motivation for school choice proponents is the positive impact on public school achievement provided with the resources at hand. Hence it is unclear as to what dealing with the question meant to achieve. 

On the other end of the spectrum, it is believed that parents constantly choosing a voucher or charter school is a robust indication of their beliefs about the school’s productivity that as long as there is considerable demand for such school it should not be easily concluded that a choice school’s success is substandard.

Lastly, the primary challenge for the study is the fact that pupils who apply to voucher or charter programs may vary greatly from individuals who do not apply, for different observable as well as unobservable factors.

A. Evidence from the best available empirical method: randomized control group of students

For this challenge, an appropriate method would be the comparison of choice program pupils to pupils who applied to the similar choice program but were not arbitrarily assigned to a voucher or charter school. The random assignment or lotteries is due to the oversubscriptions in choice programs, and randomization is often required in dealing with excess demand.

The study demonstrates an upper and lower bound on the gains from voucher or charter schools. Upper bound, if arbitrarily accepted pupils who in the end do not enroll in choice school are the least motivated among their peers, while it is lower bound if they are those whose attributes are considered the best.

Furthermore, a rational conclusion of the study would be that although, private (voucher) schools yield rather equivalent achievement in underprivileged pupils everywhere, the quality of high poverty public school differs across districts. For instance, vouchers seem to generate better performance gains in worst high poverty schools.

Notable results appear to be limited to black and Hispanic pupils, while for white and Asian pupils, the estimated impacts are constantly statistically insignificant. Such result might have been affected by the fact that the student bodies in the location of the voucher programs that have been evaluated are mostly black. The white and Asian student body in public schools with high poverty levels is usually smaller and less homogenous in its school experience than the black and even Hispanic student body. Therefore, given the smaller sample size and greater noise, it would be inadequate to consider vouchers not to have any impact on white and Asian pupils.

B. Achievement gains may be concentrated among students particularly likely to be exploited now

A speculation can be made that due to bias in the housing market, underprivileged black and Hispanics are less able to apply Tiebout choice compared to underprivileged whites and Asians. Another assumption is that due to unfair treatment of teachers and administrators, employees seem to be more at ease with under-serving pupils when the schools they work in services black and Hispanic pupils, than white and Asian pupils. With this in mind, vouchers would potentially reflect a greater positive effect to blacks’ and Hispanics’ choice sets than to whites’ and Asians’ choice sets.

C. Evidence based on similar “control” schools

The problem with this type of evidence is the positive selection bias, since pupils attending choice school will be more motivated. Hence, instead of presenting such evidence, the study will just illustrate an example of Edison Schools.

What make Edison interesting are the following reasons: (1) it is an unusual for-profit school management firm, where majority of choice school operators are non-profit organizations comprised of parents, community leaders, and former public school teachers. (2) It is convenient for evaluating, since although its schools are broadly distributed geographically and face different local choice conditions, it provides constant data. Even though Edison functions in several varieties of charter schools and of schools under intra-district choice, in all cases, it is compensated a fixed price per pupil. The schools are managed with a consistent model of school management, and the pupils are from low to very-low earning backgrounds. Such traits are helpful for evaluating choice school impact using a selection of areas. (3) Most importantly, Edison’s schools seem to face negative pool of pupils, in terms of both their observable and unobservable attributes. This is rather expected, since Edison is often employed to aid failing schools or handle charter schools in troubled neighborhoods. Comparison between Edison’s schools and closely matched comparison schools are more likely to experience negative rather than positive selection bias.

In comparison of Edison schools to others, Edison students seem to grade higher and are more prone to be deemed proficient on their state exam’s than other pupils.

D. Final thoughts on achievement comparisons between choice schools and regular public schools

Studies evaluating the disparities between choice schools and regular public schools frequently show parents’ self-reported satisfaction with their schools. However it has been observed that when children (especially those who shifted to choice schools in grade 7 or latter) evince dissatisfaction, the parents sometimes convey satisfaction with the new school. The grievances of children are due to the higher achievement standards and more restraining code of behavior. During the shift, while parents remain pleased, it is expected to detect an initial decrease in a pupil’s performance and satisfaction, which eventually increases. This implies that parents anchor their choices on their long-term expectations for their child, which may be influenced by changes in their child’s attitudes that are not immediately reflected in test scores. It would be beneficial to compare the achievement of choice and non-choice pupils on long-term outputs such as high school graduation rates, college-going rates, college graduation rates, and salaries. 

II. Does competition force public schools to be more productive?

School’s productivity is set as achievement per dollar spent, provided that the incoming achievement differences of its pupils are constant. Empirically, it is important to determine whether school choice cause regular public schools to be more productive. This is because, in the short term, public schools’ demonstrating a positive productivity would definitely extend the school choice benefits beyond the pupils who are first to assume the chance to go to voucher or charter schools. Another reason is that supporters of school choice depend on the notion that school productivity would rise enough so to flood any negative distribution impacts that some pupils might suffer.

If a school could improve a pupil’s achievement given the same amount as of that student’s current school, then the student is expected to be drawn away from the current school. Such would reduce the less productive and inflate the more productive schools until the less productive schools are replaced by the more productive or until the less productive would increase its productivity thus maintaining its student population. 

Evidently, there are pseudo-choice plans that would not let anything similar to a competitive process to happen, such are:

a.  plans wherein little or no money follows pupils that a school admitting an extra pupil cannot deal with its marginal costs;

b. plans wherein schools are not able to enter, exist, expand and contract;

c. plans wherein schools must get approval or financial aid from other schools, which were supposed to be its competitors; and so forth.

Some empirical study on these pseudo-choice plans, observes that the plans somewhat demonstrate conducts that are expected from a competitive market place.

A. Is it reasonable to think that public schools could be more productive?

Direct productivity estimates, attained by dividing the National Assessment of Educational Progress scores by per student cost and adjusted by the CPI, indicate that the efficiency in 1970 – 1971 is 65 percent greater compared to 1998 – 1999. If the study adjusts for family background, increased cost for employing individuals, teacher salaries, and so on, separately, the productivity gains ranges between 50 and 69. This implies that regular public schools are capable of considerable productivity boosts.

B. Empirical approaches to estimating the effect of school choice and competition on productivity

Due to the greatly local nature of their school districts and finance, some American metropolitan areas have usually benefited from Tiebout choice to a certain degree.

In the short term, attempt to increase the school’s productivity in response to competition has only limited options: persuade the employees to work harder, remove inefficient staff and programs, and distribute resources away from non-achievement designed activities. For the medium term, the option could be renegotiating the teacher’s contract to make the school more productive. Pursuing these options may be able to raise the school’s productivity significantly.

However, choice can have an effect on productivity through various long-term, general equilibrium mechanisms that are not easily available to administrators. For instance, the financial stress of choice may increase the salaries of teachers who improve productivity and draw in parents. Higher wages could keep or attract people into teaching, who would have otherwise chosen other careers. The need to pull in parents potentially pressures schools to supply more information regarding their achievement and thus eventually turns parents into better “consumers”.  Meanwhile, the need to generate competitive outputs may compel schools to determine and get rid of pedagogical methods and curricula that are philosophically appealing but are actually ineffective.  Lastly, for the long term, choice can influence the school’s size and existence, while in short term, the main observation would be on how the present supply of schools alters its actions.

It is preferred to parse the productivity impacts of school choice into short- and long-run, general equilibrium effects, however it is yet impossible. Even though traditional forms of school choice can provide information on long-run, general equilibrium impacts, it yields weaker incentives than other reforms like vouchers or charter schools. Thus, the study concentrates on empirical evidence from current school choice reforms.

C. The endogenous availability of choice options

The main challenge is that choice options are not arbitrary, rather it is a reaction to school behavior; for example, when individuals are disappointed with a certain school’s behavior, they attempt to build another school for themselves. 

Such occurrence is apparent to the construction of private, charter and vouchers schools. In a district where the public school is disappointing, parents are willing to allot more effort or money to attain alternative schooling.

Endogenous school choice in areas having displeasing public schools yields bias if the impact of choice on productivity is naively estimated – since schools with bad productivity force the creation of choice, it can seem as if choice induces low productivity, instead of the other way around. Bias can be best avoided by comparing the same school before and after a choice reform, while having a control group of schools that did not undergo any reforms but faced the same events of the period.

D. Identifying reforms that produce competition among schools

Since the hypothesis that competition among schools increases productivity cannot be tested by looking at choice reforms that do not establish the structure that could produce competitive incentives, the focus should be on reforms where:

1. At least a considerable fraction of pupil’s financing follows him from his regular public school to his choice school,

2. Choice schools can grow and regular public school can  contract,

3.  Choice schools do not rely on the regular public schools which they consider competitors,

4.  Must have been implemented for a long period,

5. Regular public  schools could have likely lose several percentage of its pupils, and

6. Ex ante data are available.

The study presents three American choice reforms that satisfied the criteria: vouchers in Milwaukee, charter schools in Michigan, and charter school in Arizona.

E. The effect of vouchers on productivity in the Milwaukee Public Schools

In summary, the improvements in the Milwaukee public schools after the 1998 voucher reforms are remarkable and have been maintained.  This suggests that public schools can have a strong, positive productivity reaction to competition from vouchers.

F. The Effect of Charter Schools on Productivity in the Michigan and Arizona Public Schools

In 1994, both Michigan and Arizona implemented charter school laws. In both states, charter schools have a reasonable degree of autonomy and can obtain their charters from state-wide institutions, thus allowing them to compete with the local public schools. Furthermore, for both states, the competition is greater in elementary since charter fees sufficiently cover more the elementary costs and it is easier to reach the efficient scale. Also, elementary school pupil is more likely to attend charter school than a secondary school pupil.

In general, the results on the case of Michigan and Arizona suggest that public schools facing charter competition improved their productivity as well as achievement, to an extent not only beyond their own prior performance but also in relation to other schools not under the pressures of charter competition. The developments in productivity and achievement take place once charter school reaches a vital level that coincides with the enrollment at which charter school’s taking pupils would be easily noticeable and potentially begin creating consequences for staff.

III. Do voucher and charter schools “Cream-Skim”?

Economic models expect that school choice programs impact that distribution of pupils among school in various ways contingent on the assumptions regarding:

1. The amount of money that follows the student,

2. The relation between a student’s voucher or charter school fee with the student’s attributes,

3. Whether the attributes of the student’s neighborhood, previous school and current school have an impact on voucher or charter school fee,

4. Whether choice school can apply selection,

5. The political process that identifies how the reduction of pupils to choice school influences the government aid for regular public schools,

6. The relationship between the local housing market and the support for regular public schools, and

7. The significance and functional form of peer impacts.

A. Approaches to dealing with the multiplicity of allocation outcomes

There are three useful approaches:

a. First is to admit that theory oftentimes provides a series of predicted outcomes and the task is to empirically explain the allocation outputs and attempt to recognize any existing patterns.

b. Second is to devise choice programs so to meet certain assumption and yield a desirable array of allocation outputs.

c. Last is to obtain outputs for a set of assumptions that are realistic as possible for parameters and associations that have an available evidence, and to test various assumptions for parameters and associations that have little or no evidence available.

The study will take the purely empirical approach. Cream-skimming is not a general expectation of choice model and potentially occurs in models having quite certain assumptions:

a. Assume that there is not much sorting in public schools;

b. Assume for voucher size and allocation that eligibility is wide and vouchers are rather uniform; and

c. Assume peer effects with respect to ability.

B. Evidence about cream-skimming from patterns of race, ethnicity, and poverty

The findings illustrate the odds ratio that a pupil is black, Hispanic or underprivileged, relative to the area in which his charter school is located or relative to the public school that is logistically nearest to the charter school. The odds ratio that is greater than one means that the charter disproportionately attracts students. The odds ratio of white and Asian pupils is smaller than one, a little higher than one for Hispanic and considerably greater than one for blacks and underprivileged pupils.

This strongly indicates that charter schools are not cream-skimming in any usual racial, ethnic, or economic way; instead they are disproportionately attracting pupils who have likely experienced discrimination and bias in the public schools.

C. Evidence about cream-skimming from longitudinal data on achievement

Looking for evidence of cream-skimming in terms of ability, there is a need for longitudinal achievement data for both choice schools and regular public schools since students’ performance can constantly change. Provided that a few states keep such information, the study relies on Michigan as well as on evidence from Texas’ charter schools.

Using the conduct observed on the Michigan Educational Assessment Program tests, it appears that future charter school pupils answer lesser reading and match questions correctly.

In summary, it seems that choice schools are not cream-skimming in the United States; rather they are disproportionately pulling in pupils who are perceived to be underachieving.

IV. Five features that school choice programs need to include, if we are to learn more

A. Full funding of vouchers or charter schools

The little knowledge available regarding full funding is from a few states (such as Michigan) that offer fairly commensurate funding to their charter schools that are independent of their local areas.

Majority of the first-generation choice plan provide lesser than a full financing. Voucher programs tend to give merely a portion of local per pupil spending. Even enrollment program not needing randomization simply offers way below full financing that they end up having low pupil participation rates. This is due to the “receiving” districts aversion in admitting pupils with no sufficient funding.

B. Students’ taking all compensatory funds with them when they attend choice schools

If there is concern on impacts of allocation of choice, then it is reasonable to let pupils’ take their compensatory funding when they attend choice schools. This funding comes from the state and federal government and a pupil produces his compensatory funding in the regular public school by being underprivileged, in need of remedial aid, disabled, disadvantaged, and so forth.

Opponents of choice do not agree with pupils’ bringing with them their compensatory funding when moving to choice schools. However, they also argue that able pupils should not be allowed to exit regular public schools because compensatory funding is insufficient; the school needs its privileged pupils in order to do cross-subsidizations – implicitly getting funds from able pupils and using to pupils who are costly to teach. Hence, if such funding is insufficient, then a regular public school should be satisfied with pupils leaving with their compensatory funds, since the school would gain financially.

Nonetheless, many prohibit pupils from taking their compensatory funding. Despite this loss, underprivileged pupils are likely to go to choice schools. Moreover, parents are more willing to give up the special services in the regular public school; so as to have their child enrolled in a school that they believe provides greater quality education.

C. School choice integrated with school finance

Nearly all first generation school choice programs are scantily integrated with their state’s system of school finance. Partially due to the fact that choice school are an after effect and are often not included in basic formulas. Also, opponents of school choice have often placed choice school on an odd standing so that public schools are held harmless. The lack of integration does not only provide protection for regular public schools from possible loss of revenues, but it also offers the states discouragement to develop their choice programs.

Obviously, in order to pursue school choice as a serious policy, the programs should not be at odds with the states’ systems of school finance.

On the other side, it is actually easier to do state school finance for choice schools than for regular public schools. This is owing to the fact that the fundamental objective of state school is to provide compensation for schools for disproportionate circumstances outside their control.

D. Universal choice

Many odd concerns come about when choice programs are limited to a small fraction of possible qualified pupils. Limiting choice, may lead to peculiar outcomes, such as parents migrating in order to work around the program’s limits.

It would be helpful to observe a program of universal choice, wherein all parents and if possible all regular public school must participate. A universal choice program should also permit endogenous expansion and contraction of schools, if it is to give any valuable information.

The little knowledge regarding universality in choice program is due to what is learned from the within-district choice programs which are often universal. However such programs commonly weight parents’ neighborhoods greatly, and do not allow expansion and contraction of schools thus pressuring parents to migrate and to strategically list their preference (listing a school to be their first choice, when it is not).

E. School choice integrated with accountability

School choice programs could potentially work better if parents have the information regarding schools that can be produced by accountability. Parents can also assess schools on holistic basis, and on just one grade that may have been poorly adjusted for school’s circumstances and statistical sources of bias.

Parents must be able to learn the basic information on choice schools without being skilled data analysts or internet users. To use a choice program, parents merely need to evaluate local schools.

Knowledge regarding choice and accountability comes greatly from charter schools, which bear changes in demand that are rather responsive to their published achievement.

V. Final thoughts on school choice and competition in the United States

Analysis of first-generation school choice programs explains as to why on average pupils’ performance improves when they attend choice, why public school can be productive in the face of competition, and why current choice schools is not cream-skimming.

Finally, it is significant that as more is learned regarding school choice, program structures must be clearly defined, in order to be able to understand the features separately rather than be swamped with contradictory evidence.

Source:

Caroline M. Hoxby, “School Choice and School Competition: Evidence from the United States”, Swedish Economic Policy Review,  Vol. 10, No. 2 (2004), pp. 11-67.