Do Private Schools Provide Competition for Public Schools?

I. Introduction

It has been discussed that competition among schools for students enhances school quality. In this paper the quality of public school is measured by ultimate educational attainment, wages, and high school graduation rates. The proof on competition’s effect on the quality relies on exogenous variation in the costs of private school (serving as an alternative to public school).

Two main ways public schools are affected by raising the private school competitiveness:

1. Public schools are obliged to boost school quality.

2. Increased distribution of students among schools occurs (even though parents see private school as a more competitive substitute for public school, it doesn’t imply that private schools satisfy economic concepts of perfect competition). The distribution depends on factors such as student’s ability/personality, and /or parent’s preference regarding school setting and curriculum.

Simple comparison between public school student’s outcomes in areas with and without considerable private school enrollment actually blurs the effect of increased private school competitiveness with the rise in demand for private schools in areas where there’s poor quality in public school.

Instrumental variables approach is used in obtaining unbiased estimates of the effect of private school competition.[1] Religious composition is considered to be a good exogenous measure of possible competition for public schools because it is:

1. Correlated with costs of private schools.

2. Uncorrelated with other sources of demand of private schools, such as poor public school quality.

II. A Simple Model for Public Schools, Private Schools and Parents

The paper deals with the relationship among public schools, private schools and parents, which determines the sorting of students amongst schools and the quality of each school.

Two important extensions can be derived from using the Tiebout-type model of local public goods provision.

Assume each town to be school district having its own fixed housing stock and a local education market which is considered by the household to be places of residence given its job condition. The local education market has different attributes, urban or rural in nature, town boundaries, income and educational spread, racial and religious composition. Households’ children have known ability and personality. The utility of a household is a function of the education produced in its children, its general interest in education and particularly in private or religious education, its house quality, job location and school location relative to the home and other services supplied by the residential city.

Children’s abilities and personalities, household income, and job location serve as restraints in maximizing the household utility when they allocate themselves among school districts – under the condition there is absence of private schools. Given fixed town boundaries, the level of public school provisioning per town ought to be equilibrated through local property tax and local public schooling value capitalization into house prices. However by doing so would not generate an efficient equilibrium in which the public schooling is offered at a minimum mean price and the household’s demand for it is precisely satisfied. Inefficient equilibrium to a certain extent will lead the demand for private schooling.

Private schooling occurs when it can provide for the schooling demands of an adequate number of households who are willing cover the costs on top of other taxes needed to pay for public schools. Though private schools may have the same production function as of public schools, it may also be different from each other, for example, due to greater degree of selecting and disciplining pupils.

The first extension of the model is the theory on the extent of variation of private school competition among areas. Schooling is most frequently publicly offered for two reasons: (1) fixed costs are a large portion of total costs and (2) people hope to distribute the cost of educating their children over their lifetimes. Public school funded via property tax is one way of managing such conditions. Private schools will be more competitive alternatives since they need to find ways to reduce fixed costs and distribute the expenses over people’s lifetime.

The second extension of the model is the theory on private school competition’s influence on public school conduct. There are two possible effects on public school administrators and teachers:

1. Given difficulty in evaluating school personnel productivity, principal (residents) – agent (school employee) problem tends to exist. However, the more private schools are a competitive substitute to public school, the more information on agents’ efficiency will be contained in the private schools’ enrollment share. Greater competitiveness of private schools provides more information on school productivity thus allowing residents’ response to potentially impose better productivity.

2. Greater competitiveness may also furnish higher financial incentives for school personnel. Also, the number of public school pupils will also differ more with residents’ satisfaction.

Evidently, private school competitiveness makes the total school budget rely positively on public school productivity; however for per student budget, the case is quite ambiguous as it may perversely increase when the approval with public schools falls.

If total school and per-pupil budgets would depend positively (negatively) on public school productivity or if school staffs prioritize total school budgets (per pupil budgets) rather than per pupil, greater private school competitiveness will enhance (weaken) the financial incentives experienced by schools (respectively).

III. Implications of the Model

If public schools do improve productivity due to private school competition it would by raising quality and not by reducing taxes. The reduction of taxes would only lead to higher disposable income which can be spent on private school tuition.

Furthermore, the model implies that greater private school competitiveness will raise the degree of pupil sorting among schools. Greater number of substitutes can only have a non-negative impact on the level to which parents’ inclinations and pupils’ abilities determine the choice of school.

It is important to mention that the model do not suggest anything with regards to the type of and the significance of sorting. 

The model also suggests that given all else constant, lower public school quality will lead more parents to demand for private schooling.

Also the homogeneity of the population will impact the extent to which public schooling provision is effective as well as affect the private school competitiveness.  Hence, there is a need to consider the probability of confusing the impacts of population homogeneity on public schools with the effects of private school competition.

IV. Empirical Strategy

First step is to determine the sign and magnitude of the private school competitiveness impact on public schools; the next step would be to distinguish between the mechanisms that can produce the observed sign and size.

The representations of the model indicate that private school enrollment share is likely endogenous to public school quality.

V. How Private School Costs Depend on Religious Composition

The empirical strategy of the paper relies on the fact that the greater a denomination’s population density in an area; its schools are better able to contend with public schools. This is because a substantial portion of a denominational school’s revenues normally come from church offerings which are from all denominational members not considering whether their children attend the schools. Such system (which has an adequate aspect wherein members greatly valuing the public goods will donate more) provides for cost distribution over parents’ lifetimes. Another reason is that denominational schools lessen fixed costs by sharing infrastructures, equipment, and staff with churches and synagogues. Third reason would be the parents’ preference wherein they would favor schools affiliated with their own religious congregation. Lastly, an area heavily populated by a denomination’s member can support schools at a shorter distance from each other so that parents are more able to find a private school substitute located close by.

The four points can be examined with particular reference to the Catholic Church, which is considerably the largest single religious congregation in the U.S. The three types of Catholic schools are parochial run by parish with 96% being elementary, diocesan which is under the control of a bishop with 70% being elementary, and private owned by an order such as Jesuits, with 85% being either secondary or a combination of elementary and secondary.

The study indicates that households living in an area densely populated by Catholics are more likely to live in a parish that is geographically small, supports a school, holds a great number of Catholics and has more people in religious orders eager to offer their services to Catholic schools. Moreover, parents experience lower tuition, more accessible school places, a shorter distance to the nearest Catholic school, lesser school transportation troubles, and an area wherein other children also attend Catholic school. However, between areas having comparable Catholic population densities, there are still disparities in whether Catholic schools actually exist.

In addition, there is evidence that a considerable share of parents keenly compare public and private schools, and they do not send their children to private denominational schools merely for religious motives, majority of them are likely to cite academic standards and courses as the main reason for not enrolling their children to public schools. However for some religious parents, particularly conservative Christian parents, they deliberately forgo academic quality for religious/moral values.

Another finding shows that various denominations have opportunities to build critical masses which allow furnishing denominational schooling simple and economical. However, only few denominations exhibit interest in offering schooling.

Generally districts having denser Catholic population shares that are more urban have bigger shares of households headed by females, and more are Hispanic. However, counties with greater Lutheran population shares are less urban and more homogenous with regards to race, income and family composition.

Results indicate that the higher denomination’s population shares in a district, the greater the chance that the religious congregation’s schooling will be offered.

Catholic variables are expected to offer the best instrument since the essential connections between population and school costs and accessibility are observable. Catholic variables are also expected to have the least correlation with unobservable variables that may be linked to school quality.

VI. Data

The data required is on individuals’ schooling outputs, their backgrounds, and (attended) school area’s attributes. The paper uses five datasets which are merged geographically and focused on year 1980.  Data on individuals are derived from the National Longitudinal survey of Youth (NLSY) which is a panel of 12,686 men and women aged 14 – 22, surveyed yearly since 1979. The backgrounds obtained from NLSY are the respondent’s race, gender, number of siblings, parent’s educational attainment, household’s denomination and frequency of religious attendance. It is significant to control for the impact of the religion in which the participant was raised and there is an importance in sorting out the impact of living in a greatly Catholic area from the impact of living in a Catholic household.

The second dataset is the Survey of Church and Church Membership in the United States, 1980 which is 231,708 Judeo-Christians parishioners in the U.S. The data provides each denomination’s devotees and churches/synagogues for each of the 3,101 U.S. districts.

The third set is the 1980 National Center for Education Statistics (NCES) Private Schools in America survey of 20,050 private schools in the U.S. It includes denominational or other association, enrollment, grade levels taught, tuition, subsidies and geographic location.

The fourth dataset is from the 1982 Census of Governments which contains 16,270 public school districts in the U.S., consisting of 85,000 schools.

Lastly, the data was derived from the 1980 census and 1983 City and County Data Book which provide all the district attributes there than religious composition.

About 2,097 observations are lost due to missing geographic information, respondents didn’t attend public school, neither parent’s highest level completed was reported and lacking information on district attributes.

 

VII. Results

The observation suggests that Catholic schools face higher subsidies in areas more heavily populated by Catholics. Moreover a Catholic school’s tuition seems to rely greatly on the extent of subsidy it experiences – tuition decreases as subsidies increase.

Meanwhile, Catholic secondary school enrollment as a portion of its district is decreasing in the mean tuition. Contributed teaching serves as a good instrument for tuition since it lets a school provides lower tuition while giving the same quantity of schooling and since the share of teachers who supply their services is determined rather randomly by the control of the school. The enrollment share is rising in the Catholic population share and the Catholic Church density.

Although similar results can be said for the elementary school enrollment, it seems as if private school competition at the secondary level has greater impacts on public school pupils’ outputs than at an elementary level. As have been repeatedly mentioned, Catholic area supports schools that draw in more enrollment, and this appeal is partially brought about by lower tuition and other factors.

Main results of the study indicate that Catholic school enrollment shows both increased demand for private schooling given that public school quality is low and positive impacts of private school competition on how public schools function. In addition, a change in the Catholic population of a district, such as an increase in the share of enrollment in Catholic schools yields an extra percentage point increase in the mean years of education for public school pupils.

With regards to frequency of religious attendance per households, a percentage point increase of it in the share of the population is linked to a below 1 year rise in public school pupils’ educational attainment.

Finally, Catholic secondary enrollment share increases teacher salaries, decreases per-resident spending, and no considerable impact on per-student spending.

VIII. What Explains These Results?

Holding the teacher wages and per-student spending are likely to raise the positive impact of the Catholic school enrollment share on student outputs. Exclusive of the indirect impacts on wage and spending, a percentage point increase in the Catholic school enrollment share produces an additional percentage point years of educational attainment, and some percent higher salaries. Although the impact of teacher wage and per-student spending on pupil outputs are too poorly estimated that it does not merit discussion, it can be assumed that salary increases alone can explain the improvements in schooling outputs.

In terms of sorting, given that the private school competition has a positive impact on public school, consider the “sorting story” wherein it assumes that increasing private school competition would raise its enrollment of student who would otherwise perform below average in the public schools. Hence, any factors that can reduce the cost of providing private schooling enhance the pupil population remaining in the public schools. The finding from testing this hypothesis shows that sorting or selection is actually insignificant, and to accept such assumption would be to establish acceptance that private school are extremely efficient. Further observation however, shows that the sorting story implies that Catholic schooling, offering considerably lower teacher salaries and using substantially less per student, admits students who would perform below standard or be disorderly in the public schools, and generate much better graduates than that of public schools. In other words, if student sorting between public and private school is prohibited to have an effect on the estimates, the private school enrollment share has a similar effect on pupil outputs.

Lastly, increase private school competitiveness may stimulate sorting that yields greater or smaller variation among pupils’ outputs; however the paper found no evidence for this probably due to the fact that the sample size is not sufficiently large.

IX. Conclusion

The findings support the conclusion that increasing the private school competitiveness to public schools has advantageous impact on public schooling outcomes and on how public school operates.

(Note however that these results are only suggestive of the impacts of policies that raise the public schools’ abilities to compete with one another.)

Source:

Caroline M. Hoxby, “Do Private Schools Provide Competition for Public Schools?”, NBER Working Paper 4978, (December 1994).

---

[1] Admission into private schools is likely to be endogenous to public school quality, and this endogeneity will result to downward-biased estimates of the competitive effect of higher private school enrollment. (Carolyn Hoxby)