Various suggested reforms for
elementary and secondary schooling in the United States are driven by increased
parental choice. These reforms include intra-district and inter-district
choice, vouchers for private and charter schools. Supporters of such reforms
expect that competition would improve schools while those against it worry that
the students’ method of choosing among schools may harm the educational views
of some students. The reforms broaden the traditional method of school choice
in the United States, which is a process regarded to be a main example of the
Tiebout process, by which residential choices identify the spending on, and
quality of local public goods.
The aims of this paper are to
explain the existing system and to show the general attributes of school choice
that contributes in coming up with reforms.
There are likely trade-offs given
any forms of school choice. On one hand, it may offer students the chance to
self-distribute in such a way that it may assist in learning or it may hinder
the education of at least some peers. On the other, choice may strengthen the
competitive mechanism that reinforces schools having high pupil achievement per
spending or high productivity.
In addition, since there is a need
to know whether the Tiebout choice impact relies on its financial implications
or just on parents being more able to match their children with schools, this
paper deals with some questions:
1. Is
there less possibility that parents enroll their children in private schools
provided they have more choice among public schools?
2. Do
states' school finance programs that reduce the Tiebout process’ financial
significance decreases the choice impact?
I. The Importance
of Analyzing Tiebout Choice
The analysis of Tiebout choice
allows understanding of the general-equilibrium impact of choice among schools.
It would take years before any reform could have the permeating impact, as any
short-term effect of reforms are misleading since they rely unduly on the pupil
actively making a choice in the years right after a reform. Till numerous
children have an augmented degree of choice, it is improbable for reforms to
have an impact on public schools much, either through competitive pressure or
through distribution.
Furthermore, understanding the
Tiebout process is crucial since reforms are layered on top of the existing
system. The majority of reforms would extend, not initiate, choice. Ignoring
Tiebout choice is like disregarding the fact that some of the expected effects
of reforms are have already been attained by Tiebout choice.
Finally, study of Tiebout choice
is also for practical reasons. There is difficulty in determining the choice
impact, as opposed to the causes and correlates of choice programs, except for
the case of Tiebout choice. Recognizing the problem is more manageable since
the Tiebout choice in an area mainly relies on historical conditions that are
arbitrary with regards to modern schooling.
II. What Theory
Predicts About Tiebout Choice?
Tiebout choice can have an effect
on private allotment efficiency, social allotment efficiency, and schools'
productivity. It is observed that Tiebout choice tends to increase private
allotment efficiency, even when combined with political mechanism, such as
voting on local property tax rates. With more school districts, it is easier
for households to distribute themselves into groups that are relatively uniform
with respect to their preferences on schooling and property. Hence, equilibrium
is wherein households are likely to have access to schools near to what they
privately prefer. The impact of free-rider problems, which distorts the Tiebout
equilibrium, is lessened given that there are more school districts.
One of the implications of the
private allotment efficiency result is that to a point, greater sorting
decreases the magnitude to which households pay for school programs they do not
value, Tiebout choice increases the value of school quality that households
want to buy. In addition, given greater private allotment efficiency in public
schools, household will have lesser tendencies to enroll their children to
private schools. Also, better Tiebout choice means more sorting in terms of
taste for or ability to benefit from education. For instance, a school district
may come up with a group of households having the same desire for progressive
curricula, hence to a certain degree, greater sorting enhances match quality
between pupils' needs and schools' offerings, and thus Tiebout choice will
increase the mean pupil performance.
Meanwhile the increase in private
allotment efficiency brought about by Tiebout choice does not automatically
lead to a rise in social allotment efficiency given there are human capital
spillovers among students or neighbors. If thrusting more knowledgeable
students into an encounter with less knowledgeable students would raise social
welfare, an equilibrium in which more educated pupils are self-sorted may be
socially ineffective. Unfortunately, the true impact of the contact between
over-achieving and under-achieving pupils is unknown. Thus, it is uncertain
whether social allotment efficiency always, sometimes or never increases as
private allotment efficiency augments.
In general, the theory expects
that Tiebout choice enhances school productivity. Provided that households have
difficulty identifying producers' effort, validating the quality of schooling
inputs and verifying schooling outputs, then schooling producers can earn rent.
Such agent-principal concern is relieved by Tiebout choice since school budgets
grounded on property taxes form a method that naturally integrates various
households' private observations of schooling output and schooling inputs.
These observations regarding schools establish property prices in all school
districts, and schooling producers charging excessive rent are reprimanded
through cutbacks in their school funds
Lastly, there is also uncertainty
as to whether Tiebout choice increases or decreases total expenses on schools.
Although it is expected for households to purchase greater achievement, it
however cannot be predicted as to whether they will pay more in total, provided
the fall in price.
III. An Empirical
Version of the Theoretical Predictions
Theory implies that Tiebout choice
directly influences the segregating and the incentives for productivity of
schools and indirectly impacts productivity, achievement, school expenditure,
and private schooling.
As of now it is assumed that
choice is measured accurately and that all variation in choice is exogenous.
The first prediction has the ratio of the reward provided by the market to
administrators who develop productivity to how much Tiebout choice is present
in the particular educational market to be greater than zero. Such prediction
can be only indirectly examined through looking at the connection between
choice and productivity and attempting to remove the impact of sorting.
The second prediction is that
choice stimulates self-segregating, thus each school district has households
that are more uniform in terms of their preferences for a schooling type and
school expenses. Theory deals with the
joint impacts of household features (such as income) on education preferences
that become more homogeneous. Given more choices in an educational market, the
homogeneity of household attributes in its districts will vary and some
properties are likely to become more homogeneous in every district. This
prediction implies that Tiebout choice has an effect with an unknown direction
to the heterogeneity of a particular attribute in each school district, in
relation to the heterogeneity that would exist if an educational market is only
one district.
IV. The
Identification Problem
The two identification problems that
can potentially affect the analysis of Tiebout choice are:
1. Threat of omitted variables bias.
The magnitude of choice in an
educational market is due to several factors that affect the supply of school
districts and the demand of the population for school districts. In assessing
the choice impact, there is a need to depend only on the disparity in choice
that comes from the supply side. Note that the level of the bias declines as
other factors are included, but it is impossible to determine when all bias had
been removed, also the sign of such omitted variables bias cannot be predicted.
2. Threat that the observed choice to be endogenous.
The degree of
choice noticed in a market can also be, partially, a reaction to schools'
observed productivity. This strict endogeneity can negatively bias estimates of
the effect of choice on productivity.
The best solution to the
identification problem is a set of valid tools that is, a set of variables that
impact the supply of jurisdictions but are uncorrelated with factors that
impact the demand for jurisdictions.
V. Measuring the
Degree of Tiebout-Style Choice
The key to assessing the degree of
Tiebout choice is to consider how households make residential decisions. First,
account for the boundaries of the educational market over which households
exercise choice. Second, consider the expenses related with implementing the
choice. For instance if households'
endowments are regarded to be a given, then each household experiences two main
limitations on its residential choice: income and job location. Hence the
educational market on which it practices Tiebout choice includes all school
districts in a viable commuting distance from/to its job(s). Such markets are
inclined to correspond to Census-characterized metropolitan areas of the United
States. The empirical work is limited to the analysis of metropolitan schools
and pupils.
Third, take to account the
implementation cost of Tiebout choice in a metropolitan area. The crucial
expenses include the costs of selecting a residence for its associated schools
rather than for its other attributes. The household can pick various degree of
school spending only by selecting among different school districts. In general
cost is naturally a function of the number, extent, geographic setting, and
housing stock of school districts within the metropolitan area. If households
have the ability to select among different school districts offering equivalent
residences, where comparability is based on all these attributes, then the cost
of exercising choice is low, otherwise the cost is high.
Meanwhile, as theory implies that
incentives for productivity rely on the financial outcomes related with
productivity gains and losses, such repercussions are experienced at the
district level since they toil through property prices, which have an effect on
districts' tax revenue and districts' budgets. Thus, it is expected that
productivity be influenced more by the degree of choice among districts than by
the degree of choice among schools. On the contrary, sorting is expected to be
impacted at least as much, if not more, by the degree of choice among schools
as by the degree of choice among districts.
VI. Instruments for
Measures of Tiebout Choice
The study proposes variables,
particularly, streams which reflect the number of natural school district
boundaries found in a metropolitan area, since the eighteenth and nineteenth
centuries, when school district boundaries were originally established in
America, a crucial consideration was students' commute time to school. However,
the rudimentary significance of natural barriers is maintained since they
defined initial school district boundaries, which are the main supply side
attribute that determines present boundaries. As a result, the number of school
districts in a given area and at a given time of settlement was an increasing
function of the number of natural barriers.
VII. School-District
Characteristics and Aggregation Issues
A. Aggregation
When All the Covariates Are Exogenous
First think of the situation that
would be present if measures of choice and all other covariates were exogenous.
Through equation manipulation it was observed that the gain in efficiency, in
relation to an aggregate regression, is because of the fact that an individual
level regression creates more accurate estimates of the covariances between
variables that differ at the individual or district level. Naturally, the
individual level regression takes up more information for instance the fact
that black children are more likely to be underprivileged.
B. Should
One Worry That District-Level Means Are Endogenous to Choice?
Tiebout choice in an educational
market influences the average individual qualities in each school district,
where in the district mean variables are endogenous to the degree of choice in
the educational market. However, the endogeneity of district mean variables to
choice is irrelevant to the estimated coefficient on the magnitude of choice.
By intuition, when greater choice augments the average of some attribute in one
district, there is a precisely offsetting decline in the average of that
attribute in other districts. Thus, although mean district qualities are
endogenous to choice, the impact of choice through its effect on average
attributes is automatically equal to zero.
C. Aggregation
When Choice Needs to be Instrumented
Naturally, the effects of
aggregation given that instruments are needed since measures of choice are
likely to been endogenous. Although the insight is that the impact of choice through its effect on
district average attributes is automatically equal to zero, however it would be
inaccurate, to view the coefficients on district mean characteristics as if
they were exogenous. Hence, district average attributes are included merely to
develop the fit of the equation. Nonetheless there is no interpretation
included in the result.
D. Measures
of District Heterogeneity
The average level of district
heterogeneity in an educational market is likely correlated with choice. Bear
in mind that choice does not necessarily increase the within-district
homogeneity of all attributes. Hence, the mean level of district heterogeneity
in an educational market relies on the interaction between choice and the
heterogeneity of the market's population. If an educational market is initially
homogeneous, then a rise in choice cannot increase the homogeneity of its mean
district. On the other hand, if an educational market is initially
heterogeneous, then a rise in choice potentially changes the homogeneity of its
mean district.
VIII. Data
The paper made use of several
data, all coordinated geographically at the school district or metropolitan
area level. The two sources of data on schools and school districts are:
1. Census
of Governments (COG) (U.S. Department of Commerce, 1984), which consist of
administrative data on the expenses, enrollment, and instructional employee of
each district in the United States;
2. National
Center of Education Statistics Common Core of Data (CCD) (U.S. Department of
Education, 1995), which has administrative data on enrollment, instructional
employee, and pupil demographics for each school in the United States. The only
demographic information accessible at the school level is gender, race, and
free lunch eligibility which is a common proxy for poverty.
The special school district
tabulation of the Census of Population and Housing (SDDB) (U.S. Department of
Commerce, 1983a; U.S. Department of Education, 1994b) are also utilized for
data on the percentage of private school students, average demographic
attributes of each district, and measures of the demographic heterogeneity of
each school district. While the demographic measures at the metropolitan area
level are obtained from the City and County Data Book (CCDB) (U.S. Department
of Commerce, 1983b).
Lastly, to guarantee that the
measure of choice is not just seeing larger metropolitan areas, measures of
metropolitan-area size based on population as well as land area are included,
and to make sure that the measure of choice is not just seeing regional
effects, indicator variables for the nine Census regions are incorporated.
Meanwhile, the streams variables
are from the U.S. Geological Survey's (USGS) 1/24,000 quadrangle maps.
The holdup in this study (as well
as with other studies) is obtaining data on achievement. In order to have
student data that is matched to individual school districts, the
restricted-access version of the National Education Longitudinal Survey (NELS)
(U.S. Department of Education, 1994a) for 8th, 10th, and
12th grade test scores was used. In addition, student achievement
measures were taken from the restricted-access version of the National
Longitudinal Survey of Youth (NLSY) (U.S. Department of Labor, 1998), where in
the samples are older.
The school and district data have
nearly no missing observations as all 6,523 regularly functioning metropolitan
districts are incorporated in the regressions. Discrepancy in the accessibility
of the attainment measures describes the variation in the number of
observations among regressions, and most of the disparity in availability is
not because of the missing observations but of the survey framework or the
nature of the achievement measure.
IX. Results
The output shows that the level of
choice differs broadly across metropolitan areas in the United States, not only
for the reason that some metropolitan areas are bigger than others. Also there
is not much variation among metropolitan areas in the degree of choice among
schools. The measures of choice across districts are highly correlated. Having
the significant correlation across the district-based measures, the vital
disparities among them are the substantive ones, for instance, the greater
information and greater likelihood for endogeneity in the enrollment-based
measure. However, a metropolitan area's degree of choice among districts is not
greatly correlated with its degree of choice among schools. Metropolitan areas
having greater choice across districts do not essentially provide more choice
among schools. The lack of correlation is crucial for two reasons, (1) peers
whom a student actually has contact with relies on the school he attends, and
(2) the lack of such is the key to interpreting some of the results.
A. First-Stage
Results
The district-based choice index is
statistically significantly related to the streams variables. Meanwhile, the
streams variables have a weak statistical connection with the school-based
choice index. Results indicate that if streams impact schools, it is because
they affected district boundaries, not because they are otherwise important as
of present.
B. The
Effect of Tiebout Choice on Student Achievement
Since the interest is on the
schools' productivity, it makes sense to consider first the two productivity
components: student performance and school spending. Pupils belonging to households
with income that is 10 percent higher have test scores that are about 0.15
standardized points greater. Females’ reading scores are 2 standardized points
greater than those of males. Reading scores of Black and Hispanic pupils are
5.5 and 2.9 standardized points lower than those of white non- Hispanic pupils,
respectively. Pupil whose parents have some college education but no
baccalaureate degree, and pupils having parents with at least one baccalaureate
degree have reading scores that are 2.3 and 5.5 standardized points greater
than student whose parent has no college education, respectively. Most of the
attributes of the metropolitan-area do not have statistically significant
effects on achievement.
Result implies when there exists
more choice among districts then the student achievement is greater. In
addition, such rise in the degree of choice creates educational achievement
that is 1.4 grades higher and income that is about 15 percent higher at age 32.
These results are all statistically significantly different from zero.
Meanwhile, the OLS results expose
the sign of the bias due to omitted variables and endogeneity. The successful
districts do draw in households with students, as well as attract other
districts into integration. Additionally, unobserved factors, such as
dissension, that increases the demand for districts in a metropolitan area may
have a negative impact on achievement. However if measures of a district
heterogeneity are omitted from the equation, the estimated coefficient on choice
does not vary by a statistically significant amount. Thus, the main choice
impact on pupil attainment does not seem to be working through the impact of
choice on the heterogeneity of districts.
The results offer little evidence
of heterogeneous effects. The estimates are faintly lesser for low income
families but not statistically significantly different from the estimates of
the high-income families.
Another observation is that
Tiebout choice does not have a statistically significant impact on the 8th-grade
reading scores or the 10th-grade math scores of Black and Hispanic pupils,
while the effect on non-minority pupils are statistically significant and
positive. The minority-nonminority difference in the effect on 8th-grade
reading scores is not statistically significant, while it is statistically
significant with regards to the effect on 10th-grade math scores. However, such
weak evidence of heterogeneous effects is not supported by the other measures
of achievement.
Lastly, various tests failed to reject
the null hypothesis that streams impact student attainment only through their
effect on choice. Meanwhile the test of the exogeneity of the larger streams
variable is based on the assumption that one has more a priori confidence in
the exogeneity of the smaller streams variable since they are too little to
have an impact on modern life. The test fails to reject the null hypothesis
that the bigger streams variable is a valid instrument.
In general, an increase in Tiebout
choice has a statistically significant, positive effect on measures of
attainment that range from test scores to wages. Raw OLS estimates of the
choice impact on pupil achievement are potentially downward biased and the
stream variables seem to be valid instruments. It is likely that the effect on
minority and low-income pupils is smaller than the effect on other pupils, but
the evidence for such a conclusion is only suggestive. Furthermore, sorting
caused by Tiebout choice may have negative effects on underprivileged students
that offset some of the gains they experience from competition. On the other
hand, it could also be that the choice measure is particularly erroneous for
such students.
C. The
Effects of Tiebout Choice on Per-Pupil Spending
Metropolitan areas with higher
household incomes have greater per pupil spending, as do metropolitan areas
that are more Hispanic and more racially homogeneous. Also, metropolitan areas
with bigger populations have a greater private schools enrollment, as do
metropolitan areas whose adults have more heterogeneous educational
achievement.
Meanwhile the impact of choice on
attainment seems to be significantly influenced by endogeneity and omitted
variables bias, and seems there is only a slight affect on the effects of
choice on per pupil expenditure.
However, an increase in choice among districts
decreases per pupil spending with no decline in student attainment – this has
great implications for productivity.
Interestingly, the estimated
reduction in per-pupil spending is linked to a decrease in the pupil-teacher
ratio or to an increase in teaching resources per student. Finally, Tiebout
choice pushes districts to allot money away from other inputs and towards the
reduction of pupil-ratio.
D. The
Effect of Tiebout Choice on Private-School Enrollment
The estimates suggest that a rise
in the choice index leads to a fall of private school enrollment, in other
words, Tiebout choice can have a significant impact on the percentage of
private schools students. However, raw OLS estimation indicates that Tiebout
choice has no impact on private schooling. An ineffective public school
district seems to push its pupils into private schools. Since such phenomenon
augments the concentration of public school pupils in some districts, it
endogenously reduces the choice index based on district enrollment. The key
sources of endogeneity are possibly district consolidation and the residential
decisions of households with school-aged children.
E. The
Effect of Tiebout Choice on Productivity
The results imply that a raise in
the index of Tiebout choice among districts increases productivity. However
there was no observed statistical evidence that productivity effects are
different for pupils from privileged and under privileged families. Although
the estimates suggests that productivity effects may be faintly higher for
students from privileged families, there is still weak evidence to support that
minority students encounter inferior productivity boost than other students.
Ultimately, the estimated impact
of productivity for mostly state-controlled districts are constantly lower than
those for mostly locally controlled districts, but none of the disparities is
statistically significant. Nevertheless, the result hints that Tiebout choice
has greater impact in states where districts have more financial independence.
F. Further
Evidence on Tiebout Choice and Student Sorting
So far there has been little
evidence that Tiebout choice has an effect on productivity via sorting. This
may suggest that peer effects are small or that there are offsetting benefits
and losses when pupils face heterogeneous peers.
Another observation shows that
pupils are just as segregated in schools in metropolitan areas that have a
small number of districts as they are in metropolitan areas that have more than
a few districts. Households distribute themselves into school attendance areas
within districts so that district boundaries have little impact on the racial
heterogeneity faced by students.
Overall, selection among districts
may only have little impact on the peers a pupil actually encounters since
households distribute themselves into school-attendance areas whether or not
they have much choice among districts. Thus, the impact of choice on
productivity is more likely to be due to competitive pressure among districts
than by sorting of students.
X. Conclusion
The first conclusion is perceived
to be practical. Estimates that do not consider the school districts
endogeneity are biased towards observing no effects. This is potentially
because of the tendency of successful school districts to draw in households
with school-aged children, thus increasing their market share and reducing the
observed level of choice.
The primary finding of this study
is that Tiebout choice among public school districts increases school
productivity. The most striking observation is the conflicting sign of the
attainment and spending results: Tiebout choice boosts productivity by
simultaneously increasing achievement and decreasing spending. Also, if the
potential of Tiebout choice as a policy is considered, the impacts on
productivity, student performance, and per-pupil spending are significant in
size.
There is also indicative evidence
that Tiebout choice must have financial effects if it aims generate the
productivity results described. Tiebout choice seems to have greater
productivity impacts in states where school districts are more financially
independent.
Meanwhile, households having
Tiebout-style choice are less inclined to choose private schools and are more
likely to stay in the public-school system. Although Tiebout choice among
districts allows more sorting of households by district, such rise in district
homogeneity have little net impact on attainment, per student spending, or
productivity.
The estimated impact of Tiebout
choice is not significantly affected by controlling for district-level measures
of demographic heterogeneity. This is because sorting of pupils by racial and
income groups is at the school level, not the district. Metropolitan areas with
little Tiebout choice among districts have roughly the same level of sorting
among schools as metropolitan areas with plenty of Tiebout choice among
districts. Such areas having little Tiebout choice among districts could face
the productivity benefits of choice with little variation in the nature of
student sorting among their schools.
The results of Tiebout choice are
insignificantly different for lower-income and higher-income families and for
minority and nonminority families. In addition, although the sorting brought
about by Tiebout choice has little general impact; it has negative results on
underprivileged pupils that counterbalance some of the gains of experience from
competition.
Source:
Caroline M. Hoxby, “Does
Competition among Public Schools Benefit Students and Taxpayers?” The American Economic Review, Vol. 90,
No. 5, (December 2000), pp. 1209-1238.
0 comments:
Post a Comment