RESPONSABILIDAD Y LOCAL

CONTROL: RESPONSE TO

INCENTIVES WITH AND WITHOUT

AUTHORITY OVER RESOURCE

GENERATION AND ALLOCATION

yo

D
oh
w
norte
oh
a
d
mi
d

F
r
oh
metro
h

t
t

pag

:
/
/

d
i
r
mi
C
t
.

metro

i
t
.

/

/

F

mi
d
tu
mi
d
pag
a
r
t
i
C
mi
–
pag
d

yo

F
/

/

/

/

/

2
1
1
0
1
6
8
9
0
0
2
mi
d
pag
2
0
0
7
2
1
1
0
pag
d

F

.

.

.

.

.

F

b
y
gramo
tu
mi
s
t

t

oh
norte
0
8
S
mi
pag
mi
metro
b
mi
r
2
0
2
3

Susana Loeb

(Autor correspondiente)

Universidad Stanford

Escuela de Educación

CERAS Building, 5th Floor

stanford, California 94305-3084

sloeb@stanford.edu

Katharine Strunk

Universidad Stanford

Escuela de Educación

CERAS Building, 5th Floor

stanford, California 94305-3084

kos@stanford.edu

Abstracto
This article examines the interaction between school
accountability and local control over revenue raising
and resource allocation. En particular, it asks whether
accountability policies are more or less effective at im-
proving student outcomes in states with stronger local
control. Local control is operationalized with multiple
measures, including the percentage of education fund-
ing from categorical grants, state supreme court rulings
overturning school finance systems, local entities’ tax-
ing authority, and principals’ self-assessments of their
ability to control school resources. Using the National
Center for Education Statistics’ Schools and Staffing
Survey and National Assessment of Educational
Progress, the article finds that accountability policies
are more effective when there is greater local control.

10

C(cid:1) 2007 American Education Finance Association

Susanna Loeb and Katharine Strunk

INTRODUCCIÓN

1.
Accountability and high-stakes testing have been at the forefront of education
policy discussions for the past two decades. These reforms tie student, escuela,
and/or district performance on assessments (usually standardized tests) to var-
ious consequences, including refusal to promote “underperforming” students
to the next grade, the withdrawal of funding from poorly performing schools
and districts, o, conversely, rewarding schools or districts for meeting high
performance standards. At the most extreme, districts and schools are subject
to closure or reconstitution. The theory behind accountability posits that these
consequences (and rewards) will motivate school and district actors to enact
the necessary changes to enhance student performance.

As increasing numbers of states focused on high-stakes school account-
ability policies throughout the 1990s, researchers began studying the effects
of accountability programs (Dee 2002; Figlio 2002; Figlio and Getzler 2002;
Jacob 2002; Amrein and Berliner 2003; Carnoy and Loeb 2003). With the
passage of the No Child Left Behind Act (NCLB) en 2001, the federal govern-
ment began requiring all states to use assessment and accountability systems,
placing accountability policies at center stage in national education policy de-
bates. This federal action has spurred a proliferation of research on school
assessment systems, in conjunction with discussions of the goals, equity, y
consequences of such programs (Hanushek and Raymond 2002; Hanushek
and Raymond 2004).

Although the theory behind accountability policies is fairly simple, estos
programs are based on a number of complex assumptions about the ability of
district and school personnel to enact reforms. In order to improve student
actuación, districts and schools must have sufficient resources as well as
the flexibility to control these resources. A key assumption behind account-
ability plans, entonces, is that local actors have sufficient amounts of control over
their own resources and actions to improve schools. Sin embargo, a number of
factors constrain educational actors, causing wide variations in the amount of
local autonomy enjoyed by schools and districts. Constraints include, but are
not limited to, the amount of control districts have over the allocation of their
resources, the degree to which local districts can raise revenues, and princi-
pals’ ability to control their school resources through, Por ejemplo, spending
allocation and teacher hiring decisions.

Although education policy analysts and researchers have explored how
accountability policies affect students, profesores, and administrators, también
as the creation and implementation of such policies, there is only a small
body of literature on the effects of local control on student outcomes, y
no articles that we know of focus on the interaction between accountability
policies and measures of local autonomy. This study focuses particularly on

yo

D
oh
w
norte
oh
a
d
mi
d

F
r
oh
metro
h

t
t

pag

:
/
/

d
i
r
mi
C
t
.

metro

i
t
.

F

/

/

mi
d
tu
mi
d
pag
a
r
t
i
C
mi
–
pag
d

yo

F
/

/

/

/

/

2
1
1
0
1
6
8
9
0
0
2
mi
d
pag
2
0
0
7
2
1
1
0
pag
d

.

.

F

.

.

.

F

b
y
gramo
tu
mi
s
t

t

oh
norte
0
8
S
mi
pag
mi
metro
b
mi
r
2
0
2
3

11

ACCOUNTABILITY AND LOCAL CONTROL

the interactions between accountability and local control. We measure local
control, or the lack thereof, by the percentage of state aid that comes from
categorical grants, local citizen governance over revenue raising, whether or not
a state has seen its school finance system overturned by a state supreme court
ruling, and principals’ autonomy over school policies. We first ask whether
stronger accountability reforms were more likely to be implemented in states
that had stronger or weaker local control, and whether accountability increased
or reduced this control. The article then explores the effects of accountability
and control on student outcomes and asks whether accountability systems are
more or less effective in states with greater local control.

2. BACKGROUND
This section provides a brief review of the relevant literature on how account-
ability systems, local governance, and principals’ autonomy affect student
resultados.

Accountability Policies

Accountability policies are premised on the theory that an important factor
contributing to low student performance is a lack of effort on the part of
teachers and administrators toward reaching specific state-set goals. This lack
of effort may be caused by a misalignment between the goals of teachers
and administrators and those of the community or state. To combat this
misalignment, accountability reforms are structured to guide teachers’ and
administrators’ goals toward alignment with those of the state. They reward
or sanction schools for their students’ performance on standardized tests,
which are designed to measure how successfully students are reaching state
objetivos. These consequences, both positive and negative, are intended to provide
incentives to teachers and administrators to improve student performance,
thereby achieving the set goals.

Whether high-stakes accountability policies work in the manner in which
they were intended is not clear. For accountability policies to succeed, el
underlying misalignment must significantly and negatively affect student out-
comes, and the required tests must both effectively measure state goals and be
implemented in such a way as to minimize gaming of the system by teachers
and administrators.

Some researchers have found that accountability policies may not be the
hoped-for solution. Their studies show that accountability programs are asso-
ciated with decreases in student outcomes. Sin embargo, many of these studies,
such as Dee 2002, examine an earlier set of reforms, including those utilizing
Course Graduation Requirements and Minimum Competency Tests. Porque

12

EDUCATION FINANCE AND POLICY

yo

D
oh
w
norte
oh
a
d
mi
d

F
r
oh
metro
h

t
t

pag

:
/
/

d
i
r
mi
C
t
.

metro

i
t
.

F

/

/

mi
d
tu
mi
d
pag
a
r
t
i
C
mi
–
pag
d

yo

F
/

/

/

/

/

2
1
1
0
1
6
8
9
0
0
2
mi
d
pag
2
0
0
7
2
1
1
0
pag
d

.

F

.

.

.

.

F

b
y
gramo
tu
mi
s
t

t

oh
norte
0
8
S
mi
pag
mi
metro
b
mi
r
2
0
2
3

Susanna Loeb and Katharine Strunk

these reforms are so different from the more recent accountability programs,
their results may not be applicable to the recent policies. Amrein and Berliner
(2003) study more recent reforms and conclude that these reforms also are as-
sociated with negative consequences for students. Sin embargo, their study is not
designed to estimate the causal effects of accountability reforms because they
look at only those states with strong accountability policies and thus have no
comparison group. Other research on recent accountability reforms has found
positive effects (Carnoy and Loeb 2003; Hanushek and Raymond 2004). Usando
the National Assessment of Educational Progress (NAEP) mathematics tests,
these studies find that states with strong accountability systems saw greater
increases in fourth- and eighth-grade student performance during the 1990s.
Although the Carnoy and Loeb (2003) and Hanushek and Raymond (2004)
studies find positive effects of accountability strength on student performance,
there is still reason for concern. Researchers are finding evidence that teachers
and administrators are “gaming” the system, so increased test scores may not
accurately reflect progress toward state goals but rather manipulations by local
actores. Por ejemplo, there appear to be increased classification of students into
special education programs, causing more students to be exempt from state
exámenes (Figlio and Getzler 2002; Hanushek and Raymond 2004). Carnoy and
Loeb (2003) and Hanushek and Raymond (2004) make accommodations for
this possibility in their work. Jacob and Levitt (2003) find that some Chicago
public school teachers cheated in order to boost their students’ test scores. fi-
finalmente, some research points to increased student dropout rates, though Carnoy
and Loeb (2003) do not find strong evidence of this behavior. Students may
drop out either because the school encourages them to do so in order to in-
crease overall test scores or because students themselves give up, thinking
they cannot meet the requirements for a high school degree.

In addition to concerns about the unintended consequences of account-
ability reforms, there is evidence that accountability is more effective in some
contexts than in others. While research shows that many strong-accountability
states saw test score gains, it also finds that others did not (Carnoy and Loeb
2003). One factor that may affect the implementation of accountability poli-
cies is the amount of control the local community, administradores escolares, y
teachers have over the generation and allocation of resources.1

1.

Local control is not the only factor that may affect the implementation of accountability policies. Para
ejemplo, it is likely difficult to enact school reforms without sufficient revenues. Existing research
has examined the effect of revenues on student outcomes, y, as a body of literature, has found
inconclusive results (Coleman 1966; Hanushek 1986; Card and Krueger 1992; Hanushek 1997;
Loeb and Page 2000).

yo

D
oh
w
norte
oh
a
d
mi
d

F
r
oh
metro
h

t
t

pag

:
/
/

d
i
r
mi
C
t
.

metro

i
t
.

/

/

F

mi
d
tu
mi
d
pag
a
r
t
i
C
mi
–
pag
d

yo

F
/

/

/

/

/

2
1
1
0
1
6
8
9
0
0
2
mi
d
pag
2
0
0
7
2
1
1
0
pag
d

.

.

F

.

.

.

F

b
y
gramo
tu
mi
s
t

t

oh
norte
0
8
S
mi
pag
mi
metro
b
mi
r
2
0
2
3

13

ACCOUNTABILITY AND LOCAL CONTROL

Governance of Revenue Raising and Centralization of School Finance Systems

Districts with the ability to raise funds and utilize them as they see fit may be
more able to tailor their educational systems to the specific needs of the stu-
dents in their communities and to use resources to meet state accountability
standards. Without the flexibility to raise and use funds to fit their needs, local
actors may be forced to allocate resources inefficiently, leading to decreased
student performance. Figlio (1997) finds that schools in states facing property
tax limits on revenue raising have higher student-teacher ratios and offer sig-
nificantly lower starting salaries to entering teachers. What is more, estudiantes
in states subject to limitations perform consistently worse on reading, ciencia,
matemáticas, and social studies exams than do students in states without revenue-
raising limits. Figlio and Reuben (2001) find additional evidence that local tax
limits reduce the average quality of education majors and of new public school
profesores, which is a possible cause of reduced student performance.

While tax limitations are one form of constraining local revenue-raising
activities, the centralization of school financing may also limit local actors’ abil-
ities to raise and spend funds in community-specific ways. The recent trend in
education finance that has shifted revenue-raising authority away from locali-
ties to state governments has decreased the flexibility of local administrators.
Research on the effects of school finance reform on student outcomes sug-
gests that student outcomes have not been improved by the centralization of
revenue-raising authority. Case studies of Kansas, Michigan, California, y
Texas all point to the same conclusion: there has been little to no reduction
in performance disparities after the enactment of centralizing reforms that re-
stricted local control over revenue raising (Downes 1992; Sonstelie, brunner,
and Ardon 2000; Duncombe and Johnston 2002; Imazeki and Reschovsky
2004; Cullen and Loeb 2004).

This centralization of revenue-raising authority is often due to court rulings
in which opponents of local school financing have successfully challenged
state school finance systems, arguing that such locally funded systems are
unconstitutional (Murray, evans, and Schwab 1998). Before 1993, fourteen
states had court rulings that overturned locally funded school finance systems
in favor of more centralized funding. Murray, evans, and Schwab (1998) usar
the variation in the timing of state supreme court decisions to investigate
the effect of such reforms on the distribution of education resources. Ellos
find that these state supreme court rulings reduced within-state inequality in
spending by 19 a 34 por ciento.

The trend toward centralization has led to an increased state share of total
education revenue. As state governments provide an increasing share of the
fondos, many states are placing more restrictions on the money sent to dis-
tricts. While unrestricted block grants make up the majority of state-provided

14

EDUCATION FINANCE AND POLICY

yo

D
oh
w
norte
oh
a
d
mi
d

F
r
oh
metro
h

t
t

pag

:
/
/

d
i
r
mi
C
t
.

metro

i
t
.

F

/

/

mi
d
tu
mi
d
pag
a
r
t
i
C
mi
–
pag
d

yo

F
/

/

/

/

/

2
1
1
0
1
6
8
9
0
0
2
mi
d
pag
2
0
0
7
2
1
1
0
pag
d

.

.

.

F

.

.

F

b
y
gramo
tu
mi
s
t

t

oh
norte
0
8
S
mi
pag
mi
metro
b
mi
r
2
0
2
3

Susanna Loeb and Katharine Strunk

funds, a substantial portion comes through categorical grants, restricting lo-
cal discretion over resource allocation (Sonstelie, brunner, and Ardon 2000).
When states give education dollars in the form of categorical aid, local actors
are constrained by the specific requirements attached to the monies. The per-
centage of education funding that comes in the form of categorical funding,
entonces, may serve as another measure of local control in schools and districts.

Tax limitations, court rulings overturning locally funded schools, central-
ized revenue raising, and state-imposed restrictions on resource allocation
all serve to limit local actors’ flexibility to raise and spend revenues. Yet dis-
tricts and schools with flexibility in the raising and spending of revenues may
be more responsive to the community-specific needs raised by accountability
políticas. Tal como, district personnel may be constrained in their abilities to
successfully enact accountability-inspired reforms by their (lack of) ability to
raise and spend revenues for their schools. Highly centralized state revenue-
raising systems that remove the direct connection between local property taxes
and local school revenues, and place restrictive tax limitations and heavy con-
straints on the manner in which districts and schools can spend revenues may
not only fail to enhance student outcomes on their own but may also impede
the ability of district personnel to raise sufficient funds or interest to enact
necessary accountability-inspired reforms.2

To assess whether accountability systems are more effective in states with
more local control, we use a measure of whether or not citizens or local boards
have the authority to raise tax revenues for schools. We also consider which
states have potentially restrictive tax limits and court rulings that overturn
the localized financing of schools. Finalmente, we explore whether accountability
systems are more effective when state aid is less restricted as measured by a
lower percentage of categorical aid.

Principal Influence over School Policies

Once districts have (or do not have) resources, it may matter who has control
over how to spend it, as well as over other important school-level decisions. El
theory behind accountability policies assumes that school-level administrators
will have the means to respond to incentives in order to increase student per-
rendimiento. Sin embargo, the ability of school-level actors to successfully implement
accountability policies may be hindered or helped by their degree of autonomy
over important school-level decisions. If principals do not have the flexibility
to enact reforms, accountability policies may be less effective.

2.

Some research finds that spending equalization leads to improved test scores for certain student
grupos. Por ejemplo, Card and Payne (2002) find spending equalization leads to a reduction in the
test score gaps between students of different family backgrounds.

yo

D
oh
w
norte
oh
a
d
mi
d

F
r
oh
metro
h

t
t

pag

:
/
/

d
i
r
mi
C
t
.

metro

i
t
.

F

/

/

mi
d
tu
mi
d
pag
a
r
t
i
C
mi
–
pag
d

yo

F
/

/

/

/

/

2
1
1
0
1
6
8
9
0
0
2
mi
d
pag
2
0
0
7
2
1
1
0
pag
d

.

.

F

.

.

.

F

b
y
gramo
tu
mi
s
t

t

oh
norte
0
8
S
mi
pag
mi
metro
b
mi
r
2
0
2
3

15

ACCOUNTABILITY AND LOCAL CONTROL

There is little literature regarding the extent of principal control or its as-
sociation with state-level variables. Hannaway (1993) espouses a theoretical
relationship between principal and teacher influence over school policies and
revenues, parent and union influence over school policies, the proportion of
minorities in a school, and the school’s urbanicity. She posits that “political
pressure” variables may moderate the abilities of principals and teachers to
influence school outcomes. Hannaway examines how the level of perceived au-
thority of school-level decision making is related to these “pressure” variables
and concludes that principals in schools that are more dependent on state and
federal aid, located in an urban center, and faced with stronger unions are
less likely to have authority over school-level decision making. The findings
point to the possibility of important interactions between local influence over
school-level decisions and accountability policies. The work also highlights
the need for more research on the interactions between principal and teacher
autonomy and political context variables.

Summary

There is a good deal known and being discovered about the direct effects of
accountability and local discretion over revenue raising on student and school
resultados. Less is understood about the effects of principal autonomy and
local discretion over resource allocation more broadly, and almost no work
exists on the interaction effects of all three variables together. In order to un-
ravel the questions of whether, cómo, and why accountability policies enhance
student outcomes, this article asks how state accountability, governance of rev-
enue raising, and the degree of local control over budget and hiring decisions
separately and together influence student outcomes.

3. ANALYTIC METHODS
This study has two parts. The first explores the relationship between account-
ability and measures of local control, asking whether stronger accountability
reforms were more likely to be implemented in states that had stronger or
weaker local control and whether accountability increased or decreased this
control. The second assesses the effect of accountability and control on stu-
dent outcomes and asks whether accountability is more or less effective in
states with greater or lesser local control. We define greater local control as
discretion over spending (fewer funds in categorical aid), decentralization of
school finance systems (local ability to vote on education finance, the absence
of tax limits for education, and court rulings overturning local finance struc-
turas), and greater principal autonomy over school-level decisions, specifically
regarding hiring and spending.

16

EDUCATION FINANCE AND POLICY

yo

D
oh
w
norte
oh
a
d
mi
d

F
r
oh
metro
h

t
t

pag

:
/
/

d
i
r
mi
C
t
.

metro

i
t
.

/

/

F

mi
d
tu
mi
d
pag
a
r
t
i
C
mi
–
pag
d

yo

F
/

/

/

/

/

2
1
1
0
1
6
8
9
0
0
2
mi
d
pag
2
0
0
7
2
1
1
0
pag
d

.

.

F

.

.

.

F

b
y
gramo
tu
mi
s
t

t

oh
norte
0
8
S
mi
pag
mi
metro
b
mi
r
2
0
2
3

Susanna Loeb and Katharine Strunk

Although these measures gauge aspects of local control, there are other,
especially nonfinancial, facets of local control that are not captured by these
measures. Decentralized states and states without tax limits may have more
control over revenue raising, yet it is possible that districts in such states
still do not have control over other important decisions such as curriculum.
States may be making the curricular decisions even when the local districts are
collecting the revenues. We focus here on the financial aspects of local control,
leaving other aspects of control for later research.

The Relationship between Accountability and Control

This simple analysis first tests whether states that implemented stronger ac-
countability systems had more or less local control. We then ask whether
accountability changed local control, using a regression framework in which
local control in 1999 is modeled as a function of local control in the early
1990s and accountability strength. We do not control for the strength of
accountability policies in the early 1990s because in many states no account-
ability policies either existed or were carefully documented. These analyses
are most effective for our measures of principal control because voting rights,
tax/revenue-raising limits, and the status of court decisions did not change
substantially over this time period. Además, we only have data on cate-
gorical grants for one year and are therefore unable to assess whether or not
accountability policies affected the percentage of these revenues.

Effects on Student Outcomes

In the second part of the study we estimate student achievement gains at
the state level as a function of accountability, control, and other measures
found to affect student outcomes and accountability (p.ej., percentage of black
and Hispanic students and population size). The estimates are based on the
following equation:

(cid:1)Yt−(t−1) = γ0 + γ1Ys(t−1) + (cid:1)Xs(t−(t−1))γ2

+γ3As + Lsγ4 + (As × Ls)γ5 + εst

where Yst is an aggregate student outcome measure in state s at time t; Xst
represents a vector of time-varying measures found to be significant predictors
of student achievement and to covary with accountability strength, como
student demographics and dollars of per pupil revenues;3 As is a measure of

3. We also ran models that included controls for the proportion of schools’ funds coming from the
state rather than local sources in 1995 and the percentage of the state that was in poverty in 1993.
Neither of these variables was significant in predicting either accountability strength or outcomes.

yo

D
oh
w
norte
oh
a
d
mi
d

F
r
oh
metro
h

t
t

pag

:
/
/

d
i
r
mi
C
t
.

metro

i
t
.

F

/

/

mi
d
tu
mi
d
pag
a
r
t
i
C
mi
–
pag
d

yo

F
/

/

/

/

/

2
1
1
0
1
6
8
9
0
0
2
mi
d
pag
2
0
0
7
2
1
1
0
pag
d

.

.

.

.

.

F

F

b
y
gramo
tu
mi
s
t

t

oh
norte
0
8
S
mi
pag
mi
metro
b
mi
r
2
0
2
3

17

ACCOUNTABILITY AND LOCAL CONTROL

accountability strength; and Ls is a vector of local control measures. εst is the
residual error. Because of the small state-level sample with which we are able
to work, we lose power by including all of the local control variables at once.
En cambio, we run separate regression analyses for the effect of accountability
and different local measures on student outcomes, such that Ls signifies just
one measure of local control in most specifications.

We run the analyses separately for white and black students and for fourth-
and eighth-grade math performance. We also “stack” our data, expanding our
sample from 51 states to 204 observaciones, with four observations for each
estado: one each for black fourth graders, black eighth graders, white fourth
calificadores, and white eighth graders. This gives our regression models added
fuerza. Sin embargo, because “stacking” the data generates four observations for
each state, we cluster our standard errors at the state level to avoid possible
misestimation of our standard errors.

As is the case with any state-level analysis of policy effects, this study is
subject to a variety of challenges. Small sample size at the state level limits
the number of controls we can include. Our regressions include only those
additional variables found to affect the coefficients on the variables of interest
in exploratory analyses and in Carnoy and Loeb’s earlier study. Además,
accountability policies vary substantially from state to state, and the account-
ability strength index we use is an imperfect measure of these differences. Nuestro
measures of local control are also subject to criticism, as they are clearly mea-
sured with error. Sin embargo, this study is intended to provide a starting point
for assessing a potentially important policy question. Further work is needed
to improve these measures and to find alternative methods for assessing the
interactions between accountability policies and local control, perhaps making
use of variation across districts and schools.4

4. DATA
We combine eight sets of data for the analyses: the National Assessment of Edu-
cation Progress (NAEP); Carnoy and Loeb’s accountability index; the Common
Core of Data; data from the National Center for Education Statistics’ (NCES)
Public School Finance Programs of the United States and Canada: 1998–1999;
local taxing authority data from Randall Reback; David Figlio’s 1997 catego-
rization of states that have imposed “tax revolt–era limits on school districts”;
Murray, evans, and Schwab’s (1998) classification of states’ supreme court

Due to the limited power of our sample, we left both control variables out of our final analyses.
Analyses including these measures are available upon request from the authors.

4. We attempt to examine within-state variation through the use of the National Longitudinal School-
Level State Assessment Database (www.schooldata.org). Sin embargo, outcome data going back to the
1993 school year are only available for three states, and the data for those states are very limited.

18

EDUCATION FINANCE AND POLICY

yo

D
oh
w
norte
oh
a
d
mi
d

F
r
oh
metro
h

t
t

pag

:
/
/

d
i
r
mi
C
t
.

metro

i
t
.

/

F

/

mi
d
tu
mi
d
pag
a
r
t
i
C
mi
–
pag
d

yo

F
/

/

/

/

/

2
1
1
0
1
6
8
9
0
0
2
mi
d
pag
2
0
0
7
2
1
1
0
pag
d

.

F

.

.

.

.

F

b
y
gramo
tu
mi
s
t

t

oh
norte
0
8
S
mi
pag
mi
metro
b
mi
r
2
0
2
3

Susanna Loeb and Katharine Strunk

rulings on the constitutionality of school finance systems; and the Schools and
Staffing Surveys (SASS) from 1993–94 and 1999–2000.

Outcomes

For student data, we use the posted NAEP mathematics test scores by state.
These data are collected approximately every four years in mathematics and
reading in the fourth, eighth, and twelfth grades. Students are sampled with
a multistage stratification design on a representative sample of student pop-
ulations of interest.5 NAEPdata have only been available by state since 1990,
and states were not required to participate in the testing until the passage
of the No Child Left Behind Act. Because the NAEP math test was given
in both 1996 y 2000, it provides a measure of the effects of state ac-
countability systems, many of which were enacted in the mid-1990s. Nosotros
use the percentage of students that receive a “basic” or “proficient” score on
the NAEP mathematics test. NAEP also reports the percentage of students
reaching an advanced level, but these numbers are too small to be used ef-
fectively as separate outcomes for the analyses. The National Center for Ed-
ucation Statistics (NCES) provides the NAEP scores divided by ethnicity and
carrera.

In addition to using the percentage achieving basic and the percentage
achieving proficient, we use an average scale score adjusted for exclusion
tarifas. McLaughlin (2001) has estimated an imputed set of fourth- and eighth-
grade math NAEP scale scores for 1996 y 2000 by state, assuming that all
excluded students had taken the test without accommodation. His imputations
are made on the basis of information provided on whether or not each student
in the sample was included or excluded from the tests. McLaughlin uses data
taken from student and teacher survey responses, including questions about
why the students are excluded from the tests. We use his imputed math scores
to reestimate the regression equations. These three outcome measures can
be roughly interpreted as an examination of students at the low (porcentaje
passing at the basic level), significar (adjusted scale scores), and high points of the
student performance distribution curve.

Mesa 1 shows that between 28 y 43 states have math test results in each
NAEP year. The scores are slightly higher in 2000 than in 1996, and the scores
for white students are higher than those for black students. The average gain
de 1996 a 2000 in the percentage of students achieving at least the basic
level was 5.5 por ciento. En 2000, a state average of 76.7 percent of white and
73.7 percent of black fourth-grade students and 74.4 percent of white and 33.1

5.

See http://nces.ed.gov/nationsreportcard/pubs/guide (accessed July 2005).

yo

D
oh
w
norte
oh
a
d
mi
d

F
r
oh
metro
h

t
t

pag

:
/
/

d
i
r
mi
C
t
.

metro

i
t
.

/

F

/

mi
d
tu
mi
d
pag
a
r
t
i
C
mi
–
pag
d

yo

F
/

/

/

/

/

2
1
1
0
1
6
8
9
0
0
2
mi
d
pag
2
0
0
7
2
1
1
0
pag
d

.

.

.

F

.

.

F

b
y
gramo
tu
mi
s
t

t

oh
norte
0
8
S
mi
pag
mi
metro
b
mi
r
2
0
2
3

19

ACCOUNTABILITY AND LOCAL CONTROL

Mesa 1. Descriptions of Outcome Variables

Variables (Dakota del Sur)

Obs Min Max Mean (Dakota del Sur) Obs Min Max Mean

Fourth Grade

Eighth Grade

33
33
37
37

33
33

40
30
39
28
40
30
39
28

33
23
33
23

261.3 285.7 275.8
227.0 275.0 243.6
263.0 290.0 278.4
227.0 273.8 247.6

−3.5
−9.7

11.1
23.0

2.2 (3.2)
3.8 (6.6)

55
15
59
17
12
1
14
1

−1
−8
−2
−3

81
43
85
49
37
11
43
12

14
18
13
5

70.5
27.3
74.4 (6.9)
33.1
26.7
4.1
29.7
5.5

4.5 (3.7)
5.1 (6.4)
3.7 (3.0)
1.3 (2.2)

NAEP McLaughlin-Adjusted

Level Scores
1996 Whites (6.9)
1996 Blacks (11.2)
2000 Whites (7.0)
2000 Blacks (11.0)

Change in McLaughlin-Adjusted

Scores 1996–2000
Whites
Blacks

NAEP Math Levels1

43
43
40
41

218.5 238.4 227.1 (5.2)
185.3 215.6 199.5 (6.8)
220.5 239.5 229.9 (5.2)
191.1 229.2 204.2 (7.8)

35
36

−3.6
−3.8

10.1
14.6

3.4 (3.2)
4.6 (5.1)

43
1996 % Basic—Whites (7.2)
1996 % Basic—Blacks (7.1)
35
2000 % Basic—Whites
40
2000 % Basic—Blacks (8.7)
31
1996 % Proficient—Whites (6.8) 43
1996 % Proficient—Blacks (2.3) 35
2000 % Proficient—Whites (7.2) 40
2000 % Proficient—Blacks (2.5) 31

Change in NAEP Levels 1996–2000

% Basic—Whites
% Basic—Blacks
% Proficient—Whites
% Proficient—Blacks

35
27
35
27

62
18
65
21
13
1
16
2

−2
−9
−3
−5

86
46
89
61
38
10
41
11

13
20
12
7

71.8 (6.5)
30.3 (6.5)
76.7 (6.2)
37.7 (8.9)
23.8 (5.9)
3.8 (1.8)
28.0 (6.5)
5.3 (2.6)

5.3 (3.9)
7.4 (6.8)
4.5 (3.6)
1.7 (2.5)

Stacked Fourth and Eighth Grades

NAEP McLaughlin-Adjusted

Level Scores
1996
2000

Change in McLaughlin-Adjusted

Scores 1996–2000

NAEP Math Levels
1996 % Basic
2000 % Basic
1996 % Proficient
2000 % Proficient

Change in NAEP Levels 1996–2000

% Basic
% Proficient

152 185.3 285.7 233.5 (28.5)
155 191.1 290.0 238.9 (28.2)

137 −9.7

23.0

3.5 (4.8)

148
138
148
138

15
17
1
1

118 −9
118 −5

86
89
38
43

20
13

52.6 (22.1)
58.4 (21.3)
15.6 (11.7)
18.8 (12.8)

5.5 (5.2)
3.0 (3.2)

1. NAEP test score is taken from http://nces.ed.gov/nationsreportcard/states.

percent of black eighth-grade students reached this basic level. De término medio,
solo 28.0 percent of white and 5.3 percent of black fourth-grade students and
29.7 percent of white and 5.5 percent of black eighth-grade students achieved
the proficient level in 2000.

Accountability

Our measure of accountability strength comes from a database developed by
Carnoy and Loeb (2003) using 1999–2000 information from the Consortium

20

EDUCATION FINANCE AND POLICY

yo

D
oh
w
norte
oh
a
d
mi
d

F
r
oh
metro
h

t
t

pag

:
/
/

d
i
r
mi
C
t
.

metro

i
t
.

/

/

F

mi
d
tu
mi
d
pag
a
r
t
i
C
mi
–
pag
d

yo

F
/

/

/

/

/

2
1
1
0
1
6
8
9
0
0
2
mi
d
pag
2
0
0
7
2
1
1
0
pag
d

.

.

.

F

.

.

F

b
y
gramo
tu
mi
s
t

t

oh
norte
0
8
S
mi
pag
mi
metro
b
mi
r
2
0
2
3

Susanna Loeb and Katharine Strunk

for Policy Research in Education (CPRE).6 The scale of accountability ranges
de 0 a 5, based on year 2000 accountability conditions, with states such
as Iowa and Nebraska, that did not have any state-level accountability require-
ments for schools or districts, coded zero, and states with the “maximum” state
level demands on schools and that require a high school competency exam
for graduation, such as Texas, North Carolina, New Jersey, and Florida coded
five.

The zero-to-five scale captures the degree of state external pressure on
schools to improve student achievement according to state-defined perfor-
mance criteria. States receiving a zero did not test students statewide or did
not set statewide standards for schools or districts. States that required state
testing in the elementary and middle grades and the reporting of test results to
the state, but no school (or district) sanctions or rewards (weak external pres-
sure or none at all), score a one. States that tested at the elementary and middle
school levels and had moderate accountability sanctions/rewards or, alterna-
activamente, a high school exit exam (that sanctions students but pressures schools to
improve student performance) earn a two. States that tested at the lower and
middle grades, have moderate accountability repercussions for schools and
districts, and require an exit exam in high school receive a three. Those that
tested and placed strong pressure on schools or districts to improve student
logro (threat of reconstitution, principal transfer, loss of students) pero
did not require a high school exit test score a four. States receiving a five tested
students in primary and middle grades, strongly sanctioned and rewarded
schools or districts based on improvement in student test scores, and required
a high school minimum competency exit test for graduation. Mesa 2 muestra
that the states’ average index in 2000 era 2.12 on a scale of zero to five.

Local Control

We measure local control by (1) the percentage of total education revenue tied
to categorical aid, (2) tax limitations, (3) whether or not a state’s supreme court
has overturned the constitutionality of the state’s school finance system, (4)
whether or not local voting for school revenues is allowed, y (5) principals’
reports of control over school-level decisions.7

6. See http://www.cpre.org/Publications/Publications Accountability.htm (accessed July 2005).
7. Realizing that teachers’ unions may also play an important role in the amount of local control
school or district personnel have over decision making and the effect of accountability policies on
student outcomes, we also run analyses looking at the extent of unionization in the state. Usamos
several measures of unionization: SASS’s designation of districts as having collective bargaining
agreements, meet-and-confer agreements, or neither; whether or not a state allows collective bar-
gaining; and whether or not a state is a right-to-work state. Although we find some evidence that
unions are weaker in states with stronger accountability policies, our analyses show inconclusive
and insignificant effects of the degree of unionization and its interaction with accountability policy

yo

D
oh
w
norte
oh
a
d
mi
d

F
r
oh
metro
h

t
t

pag

:
/
/

d
i
r
mi
C
t
.

metro

i
t
.

/

/

F

mi
d
tu
mi
d
pag
a
r
t
i
C
mi
–
pag
d

yo

F
/

/

/

/

/

2
1
1
0
1
6
8
9
0
0
2
mi
d
pag
2
0
0
7
2
1
1
0
pag
d

.

.

.

.

.

F

F

b
y
gramo
tu
mi
s
t

t

oh
norte
0
8
S
mi
pag
mi
metro
b
mi
r
2
0
2
3

21

ACCOUNTABILITY AND LOCAL CONTROL

Mesa 2. Descriptions of Explanatory Variables (norte = 50)

1993

1999

Variables

Accountability1

State Education Finance

Per pupil revenue (1000s)2
Expenditures per pupil (1000s)
Instructional only per pupil (1000s)

Measures of Teacher Quality3

’93 district base salary (1000s)
’93 ln base salary

Centralization of School Finance

Just citizen vote
Just board vote
Both vote
Neither vote
States with tax limits
States with court decisions
Percent of funds—Categorical
% of ed. financing from state4

Principal Influence Measures5

Principal

Average
Hiring
Spending

Principal by district (avg.)
Principal by state (avg.)
District by state

Average
Hiring
Spending

Significar

—

6.36
5.67
2.52

24.47
10.10

—
—
—
—

—

.43
.27

Dakota del Sur

—

1.31
1.29
0.56

3.26
0.13

—
—
—
—
—
—
—

0.41

0.18

4.25
4.46
3.99
1.49
2.20

1.68
2.15
1.57

0.17
0.32
0.31
0.19
0.37

0.24
0.35
0.30

Significar

2.12

8.40
7.33
3.18

28.77
10.26

.25
.20
.41
.12
.43

—

.12
0.53

4.32
4.65
4.24
1.43
1.94

1.49
2.01
2.15

Demographic Variables

Percent of students in poverty6
Percent black and Hispanic
1995 state population (100,000s)

0.45
0.23
52.44

0.12
0.18
57.59

0.36
0.26

—

Dakota del Sur

1.44

1.78
1.57
0.68

3.44
0.12

—
—
—
—
—
—

.06
0.16

0.10
0.27
0.26
0.15
0.26

0.21
0.31
0.35

0.11
0.18
—

1. http://www.cpre.org/Publications/Publications Accountability.htm. The Accountabil-
ity index ranges from 0 a 5, con 5 implying the strongest level of accountability.
2. Per pupil revenue, expenditures, and instructional expenditure data are taken from
the Common Core of Data at http://nces.ed.gov. The “before” data are from 1996, y
the “after” data are from 2001, chosen to best match the NAEP outcomes data.
3. District pay scale. All salary numbers are in unadjusted nominal dollars.
4. The percentage of total spending on education finance funded by the state govern-
ment can be found at http://nces.ed.gov. It is important to note that the “1993” data
are actually from 1963 and the “1999” data are from 1995. De nuevo, these are the closest
year matches to SASS feasible with these data.
5. These variables are taken from the SASS data set at http://nces.ed.gov/SASS.
6. This variable is from the SASS data set at http://nces.ed.gov/SASS.

We obtain the percentage of education revenues tied to categorical aid
from the NCES Public School Finance Programs of the United States and Canada:
1998–1999 (NCES 2001). There is a brief chapter on each state’s school finance

strength on student outcomes. This is possibly the result of our limited data sample and weak
unionization indicators and is a topic for further research.

22

EDUCATION FINANCE AND POLICY

yo

D
oh
w
norte
oh
a
d
mi
d

F
r
oh
metro
h

t
t

pag

:
/
/

d
i
r
mi
C
t
.

metro

i
t
.

/

F

/

mi
d
tu
mi
d
pag
a
r
t
i
C
mi
–
pag
d

yo

F
/

/

/

/

/

2
1
1
0
1
6
8
9
0
0
2
mi
d
pag
2
0
0
7
2
1
1
0
pag
d

F

.

.

.

.

.

F

b
y
gramo
tu
mi
s
t

t

oh
norte
0
8
S
mi
pag
mi
metro
b
mi
r
2
0
2
3

Susanna Loeb and Katharine Strunk

programa, from which it is possible to generate an estimate of the percentage
of categorical aid. We define categorical aid as the specific categorical aid
programs in each state plus funds that are earmarked for technology, aid to
private and alternative schools, and accountability and standards programs.
It is important to note that we can calculate this measure as a percentage
of total reported state and local revenues. We cannot account for private do-
nations from parents and/or foundations. The mean percentage of categor-
ical aid is 11.94%, with a wide range of categorical funding, de 0.44%
(Oregón) a 31.61% (Washington, corriente continua). Mesa 2 presents descriptive statistics
of our explanatory variables, and table 3 gives the local control variables by
estado.

We measure the degree of centralized governance over education revenue
raising using data from a survey gathered by Randall Reback at Columbia
Universidad. From this survey we are able to ascertain whether or not citizens
directly vote on local taxes to support regular instruction in public schools,
and whether or not elected boards such as school boards or town meeting
representatives vote on local taxes to support regular instruction. It is difficult
to determine which voting status—whether or not citizens alone, citizens and
boards together, or just boards have the right to vote on local taxes to support
instruction—indicates the lowest level of centralization. Sin embargo, we believe
that it is accurate to assume that states with neither citizen nor board votes
have the highest level of centralization. As table 2 indicates, 12 percent of states
allow neither citizen nor board voting for local revenue raising for schools (No
vote), mientras 41 percent allow both types of votes (both vote). A quarter of states
just allow citizens to vote (just citizen vote), and the remaining 20 por ciento
allow just board voting (just board vote).

We also use a measure of whether or not a state has imposed a potentially
restrictive tax limitation, taken from Figlio 1997. The states that he considers to
have imposed “tax revolt–era limits on school districts” are outlined in table 3.
Teóricamente, it is important to be able to ascertain the effect of varying levels of
district discretion over taxation on student outcomes. The state-by-state indices
from which the dummy variables are created do so. Practically, while it would
be preferable to be able to utilize a more in-depth measure, none exists at this
tiempo. Mesa 2 shows that in 1993, 43 percent of states were subject to restrictive
tax limits.

Además, we use a measure, taken from Murray, evans, and Schwab
1998, of whether or not a state’s supreme court had overturned the state’s
school finance system. Opponents of locally financed school systems had suc-
cessfully challenged the state school finance system by 1993 in California,
Connecticut, Kansas, Kentucky, Massachusetts, Montana, New Hampshire,
New Jersey, Tennessee, Texas, Washington, West Virginia, Wisconsin, y

yo

D
oh
w
norte
oh
a
d
mi
d

F
r
oh
metro
h

t
t

pag

:
/
/

d
i
r
mi
C
t
.

metro

i
t
.

/

/

F

mi
d
tu
mi
d
pag
a
r
t
i
C
mi
–
pag
d

yo

F
/

/

/

/

/

2
1
1
0
1
6
8
9
0
0
2
mi
d
pag
2
0
0
7
2
1
1
0
pag
d

.

.

F

.

.

.

F

b
y
gramo
tu
mi
s
t

t

oh
norte
0
8
S
mi
pag
mi
metro
b
mi
r
2
0
2
3

23

ACCOUNTABILITY AND LOCAL CONTROL

Mesa 3. Local Control Variables by State

Percent

Percent
Categorical Total Rev Voting
Fondos

Successful Average
Principal

Tax Court

Average
Principal

Estado

from State Provisions Limits Challenge Control—’93 Control—’99

0.06
Alabama
0.06
Alaska
0.04
Arizona
0.01
Arkansas
0.16
California
0.04
Colorado
0.09
Connecticut
0.19
Delaware
0.32
D.C.
0.11
Florida
0.16
Georgia
0.15
Hawaii
0.07
Idaho
0.10
Illinois
0.10
Indiana
0.22
Iowa
0.11
Kansas
0.17
Kentucky
0.11
Luisiana
0.10
Maine
0.10
Maryland
Massachusetts 0.14
0.12
Michigan
0.15
Minnesota
0.15
Mississippi
0.14
Misuri
0.05
Montana
0.12
Nebraska
0.16
Nevada
0.02
New Hamp.
0.13
New Jersey
0.25
New Mexico
0.20
Nueva York
0.13
norte. Carolina
North Dakota 0.06
0.12
Ohio
0.19
Oklahoma
0.00
Oregón
0.11
Pensilvania
0.06
Rhode Island
0.11
S. Carolina
South Dakota 0.06
0.03
Tennessee
0.16
Texas
0.20
Utah
0.21
Vermont
0.13
Virginia
0.16
Washington
West Virginia 0.05
0.06
Wisconsin
0.12
Wyoming

Significar

0.12

0.71
0.71
0.49
0.67
0.58
0.45
0.38
0.68
—
0.53
0.55
0.97
0.66
0.36
0.54
0.49
0.60
0.69
0.60
0.48
0.39
0.31
0.71
0.54
0.66
0.40
0.55
0.33
0.66
0.07
0.38
0.84
0.42
0.69
0.45
0.41
0.65
0.50
0.41
0.42
0.50
0.29
0.52
0.46
0.58
0.26
0.43
0.73
0.69
0.46
0.51

0.53

Ambos
Board
Citizens
Ambos
Neither
Citizens
Ambos
Citizens
Neither
Board
Board
—
Citizens
Ambos
Citizens
Ambos
Ambos
Board
Ambos
Ambos
Neither
Ambos
Citizens
Citizens
Ambos
Ambos
Citizens
Board
Neither
Citizens
Ambos
Neither
Ambos
Board
Ambos
Ambos
Citizens
Citizens
Board
Board
Ambos
Ambos
Ambos
Board
Ambos
Citizens
Neither
Citizens
Citizens
Ambos
Ambos

0
0
1
1
1
1
0
0
—
0
0
0
1
1
1
0
1
0
1
0
0
1
1
1
1
1
0
0
0
0
1
1
0
0
1
1
0
1
0
0
0
0
0
0
1
0
0
1
0
0
0

0
0
0
0
1
0
1
0
0
0
0
0
0
0
0
0
1
1
0
0
0
1
0
0
0
0
1
0
0
1
1
0
0
0
0
0
0
0
0
0
0
0
1
1
0
0
0
1
1
1
1

0.43 0.27

4.08
4.26
4.29
4.14
4.37
4.43
4.26
4.23
3.93
4.29
4.21
4.58
4.34
4.32
4.24
4.38
4.30
4.24
4.13
4.45
3.93
4.37
4.19
4.27
4.26
4.29
4.33
4.34
4.36
4.56
4.30
4.28
4.11
4.20
4.20
4.08
4.13
4.40
4.10
3.95
4.40
4.32
3.94
4.27
4.42
4.43
4.24
4.33
3.65
4.29
4.46

4.25

4.33
4.18
4.40
4.20
4.26
4.37
4.34
4.26
4.47
4.37
4.37
4.51
4.45
4.31
4.33
4.29
4.32
4.42
4.40
4.36
4.31
4.28
4.32
4.26
4.37
4.19
4.35
4.33
4.42
4.33
4.39
4.39
4.35
4.36
4.32
4.22
4.22
4.29
4.21
4.14
4.35
4.29
4.32
4.49
4.41
4.27
4.29
4.41
3.96
4.28
4.22

4.32

Wyoming. Murray, evans, and Schwab (1998) provide a thorough explanation
of these cases and their effects on the distribution of educational resources.

Our principal control variables come from the 1993–94 and the 1999–
2000 NCES Schools and Staffing Survey (SASS). SASS is the largest cross-
sectional sample survey of public, public charter, privado, and Native American

24

EDUCATION FINANCE AND POLICY

yo

D
oh
w
norte
oh
a
d
mi
d

F
r
oh
metro
h

t
t

pag

:
/
/

d
i
r
mi
C
t
.

metro

i
t
.

/

F

/

mi
d
tu
mi
d
pag
a
r
t
i
C
mi
–
pag
d

yo

F
/

/

/

/

/

2
1
1
0
1
6
8
9
0
0
2
mi
d
pag
2
0
0
7
2
1
1
0
pag
d

.

.

.

.

F

.

F

b
y
gramo
tu
mi
s
t

t

oh
norte
0
8
S
mi
pag
mi
metro
b
mi
r
2
0
2
3

Susanna Loeb and Katharine Strunk

elementary and secondary schools in the United States.8 We average princi-
pals’ perceptions of their own control over school management issues such as
curriculum development, spending, teacher hiring, standards, teacher evalua-
ción, professional development, and student discipline policies. In this analysis
we use both principals’ perceptions of their own control and their perceptions
of their control relative to their assessment of the influence of district and state
Departments of Education. These variables are ranked by principals on a one-
to-five scale, with one being the perception of the least amount of influence.
When a relative measure is used, uno (1) signifies that the principal believes
that he or she has the same level of influence as the state or district, and less
(más) than one (1) signifies that the principal believes that he or she has less
(más) influence than the state or district.

Principals believe they have substantial control over most areas of school-
En g. En 1993, principals rank themselves as having an average control level of
4.25 on a scale from one to five across measures (ver tabla 2). This is more
than double the control they attribute to the state and one and a half times the
control they attribute to the district. The average measure of principal control
and principal control relative to the district stayed approximately the same over
tiempo, while principals’ assessment of their authority relative to the state fell on
average from 2.2 times as much control in 1993 a 1.94 times in 1999. It is not
surprising that in a time period marked by the introduction of accountability
policies that principals would feel that they lose authority relative to states.
Mesa 2 indicates that principals feel that districts’ influence relative to states
dropped as well, decreasing from 1.68 en 1993 a 1.49 en 1999. Mesa 3 presents
average principal authority by state for both 1993 y 1999.

While all our measures of local control capture some aspect of the under-
lying flexibility of schools and school districts, they are by no means the same
measure. There is very little overlap between states with high levels of principal
control and either low percentages of categorical grants or citizen control over
voting. Por ejemplo, the correlation between percentage of categorical grants
and principal control is 0.16 y 0.12 for control over spending and hiring,
respectivamente. Half of the states with no voting provision for revenue raising
have low levels of control over allocation as measured by a high percentage of
categorical grants, but the others do not. To ensure that we are not picking
up effects due to high concentrations of poverty within a state, we also check
correlations and interactions between the percentage of that state population
in poverty in 1993 and our local control variables. We find no evidence of

8. We use only the public and public charter school surveys found at Schools and Staffing Survey Web
site, via the National Center for Education Statistics Web site, at http://nces.ed.gov/surveys/SASS
(accessed July 2005).

yo

D
oh
w
norte
oh
a
d
mi
d

F
r
oh
metro
h

t
t

pag

:
/
/

d
i
r
mi
C
t
.

metro

i
t
.

F

/

/

mi
d
tu
mi
d
pag
a
r
t
i
C
mi
–
pag
d

yo

F
/

/

/

/

/

2
1
1
0
1
6
8
9
0
0
2
mi
d
pag
2
0
0
7
2
1
1
0
pag
d

.

.

.

.

.

F

F

b
y
gramo
tu
mi
s
t

t

oh
norte
0
8
S
mi
pag
mi
metro
b
mi
r
2
0
2
3

25

ACCOUNTABILITY AND LOCAL CONTROL

Mesa 4. 1993 Local Control Measures by Accountability Strength

Bajo
Accountability

Medium
Accountability

High
Accountability

Mean per pupil revenues

Mean per pupil expenditures

Mean per pupil instructional expenses

% of states with no vote

% of funding from categorical aid

Estado % of state and local funding

% of states with tax limits

% of states with court decisions

Mean principal control

Mean principal by state control

Mean principal by district control

Mean district by state control

$5,703 $5,133

$2,302 3.85% 11.05% 52.55% 0.36 0.27 4.30 2.03 1.45 1.55 $5,473

$4,880 $2,197

7.14%

10.23%

50.12%

0.64

0.21

4.23

1.86

1.40

1.44

$5,877 $5,259

$2,360

30.00%

14.69%

57.00%

0.30

0.40

2.03

1.81

1.41

1.38

significant correlation between poverty and any of the other variables. In ad-
condición, including poverty in our regression analyses has little effect on our
resultados.

All of the models in this study control for the percentage of eighth-grade
students in the state that are black or Hispanic in the 1995–96 school year,
the yearly growth in the proportion of black or Hispanic students, the state
population in 1996 (in hundred thousands), yearly population growth, y
per pupil revenues in 1993. These measures were collected and utilized by
Carnoy and Loeb (2003). We control for the proportion and growth of minority
students in order to account for any possible omitted variables due to the effects
of demographic background characteristics on student outcomes. We control
for population size because it is a strong predictor of accountability strength.
We include a control for per pupil revenue levels as a measure of states’ capacity
niveles.

5. RESULTADOS
Initial Local Control Characteristics of States That Adopted Stronger

Accountability Policies

This section addresses whether the level of local control in 1993 (before the
implementation of most accountability policies) was systematically higher or
lower in states that implemented stronger accountability policies. Para esto
análisis, we divide states into three groups: el 26 states with an accountability
rating of less than two (bajo); el 14 with an accountability rating of two to four
(medio); y el 10 with an accountability rating of four or higher (alto).
Mesa 4 shows that there is little initial difference in revenues or expenditures

26

EDUCATION FINANCE AND POLICY

yo

D
oh
w
norte
oh
a
d
mi
d

F
r
oh
metro
h

t
t

pag

:
/
/

d
i
r
mi
C
t
.

metro

i
t
.

/

/

F

mi
d
tu
mi
d
pag
a
r
t
i
C
mi
–
pag
d

yo

F
/

/

/

/

/

2
1
1
0
1
6
8
9
0
0
2
mi
d
pag
2
0
0
7
2
1
1
0
pag
d

.

.

.

.

F

.

F

b
y
gramo
tu
mi
s
t

t

oh
norte
0
8
S
mi
pag
mi
metro
b
mi
r
2
0
2
3

Susanna Loeb and Katharine Strunk

per pupil across the three accountability groups. A pesar de 1993 per pupil
revenue, expenditures per pupil, and instructional expenditures all appear to
be slightly greater in high accountability states than in low or medium states,
simple F statistics show that these differences are not statistically significant
at conventional levels.

There are significant differences across accountability groups in some
measures of local control, though this relationship does not hold for other
measures. Por ejemplo, 30 percent of high-accountability states had no local
control over revenue raising, compared to only 4 percent of low-accountability
y 7 percent of medium-accountability states. Since there are a manageable
number of states to look at individually, we can assess the consistency of
this finding that states with lower local revenue-raising control, on average,
implemented stronger accountability. No es sorprendente, the finding does not
hold for all states. Mesa 3 shows that, En particular, New Jersey had provisions
for both board and citizen voting, which places it in the relatively high local
control group, but implemented strong accountability.9 Florida, Texas, y
North Carolina (other states that rank as fives on the accountability strength
escala) had just board voting provisions, indicating that they had relatively low
levels of local control over school financing.10

States that implemented strong accountability policies also had a higher
proportion of their resources tied up in categorical grants. Approximately 15
percent of overall education funding was in the form of categorical aid for
high-accountability states, whereas the states with the weakest accountability
policies had 11 percent of overall education funding in the form of categorical
aid, and medium-accountability states had 10 percent categorical aid. Estos
differences are not statistically significant across the three accountability clas-
sifications, but the medium-accountability and high-accountability states are
statistically different in the percentage of funding they receive from categorical
subsidios.

We do not find significant differences in the prevalence of supreme
court rulings overturning local school finance systems between accountability
grupos, though the trend is consistent with greater accountability in states
with less local control. Twenty-seven percent of low-accountability states had

9. New York also appears to be an outlier having both strong local control and strong accountability;
sin embargo, this accountability-strength rating is based on New York’s 2001 standards and is one
of the two state rankings that were revised upon further examination by Margaret Goertz of the
University of Pennsylvania.Goertz’s revised rankings do not significantly change the overall results
but do show that New York fits into the overall pattern, leaving New Jersey as the sole outlier.
10. Tax limits are another measure of district revenue-raising authority, though we find no relationship
between tax limits and accountability. Districts in 36 percent of low-accountability states were
subject to tax limitations in 1997. This compares with 64 percent for medium-accountability states
y 30 for high-accountability states. These differences are not significantly different than zero.

yo

D
oh
w
norte
oh
a
d
mi
d

F
r
oh
metro
h

t
t

pag

:
/
/

d
i
r
mi
C
t
.

metro

i
t
.

/

/

F

mi
d
tu
mi
d
pag
a
r
t
i
C
mi
–
pag
d

yo

F
/

/

/

/

/

2
1
1
0
1
6
8
9
0
0
2
mi
d
pag
2
0
0
7
2
1
1
0
pag
d

.

F

.

.

.

.

F

b
y
gramo
tu
mi
s
t

t

oh
norte
0
8
S
mi
pag
mi
metro
b
mi
r
2
0
2
3

27

ACCOUNTABILITY AND LOCAL CONTROL

such rulings, compared to 21 percent of medium-accountability states and
40 percent of high-accountability states. Mesa 4 also shows similar differ-
ences in 1993 principal-assessed control over school operations by account-
ability level. The only statistically significant difference across accountability
groups is that principals in states that implemented stronger accountabil-
ity perceived themselves as having somewhat less average influence relative
to their district, but the trend is evident across all measures of principal
control.

States that implemented stronger accountability appear to have had some-
what less local control over revenue raising and resource allocation. Sin embargo,
perhaps due to small sample sizes and imperfect measurement, not all of the
differences are statistically significant. When we conduct a similar analysis
but control for other state characteristics including the percentage of black
and Hispanic students (levels and growth), población (levels and growth),
and per pupil revenues, only average principal control is significantly related
to accountability strength.11

The Relationship between Accountability Policies and Changes in Local Control

This section evaluates whether accountability reforms were associated with
changes in local control from 1993 a 1999. Desafortunadamente, we can only
do this for local control as measured by principal assessments because for
most other measures there was very little change in local control over this
time period. While weak differences in principal control by accountabil-
ity strength were evident in the early 1990s, accountability reforms appear
to have changed principal autonomy. Mesa 5 shows systematic changes
in control by accountability strength. Strikingly, although not unexpect-
edly, states gained more control over most aspects of schooling in states
with stronger accountability policies, according to the principals surveyed.
Less predictably, sin embargo, principals in higher-accountability states expressed
greater gains in perceived control, especially in the areas of professional
desarrollo, spending, and teacher hiring. Mesa 5 shows that principals
in stronger accountability states gained significantly more perceived control
over spending and hiring than did principals in lower-accountability states,
and they gained more influence over hiring relative to the state, aunque
they lose on curriculum, evaluation of teachers, and average authority rel-
ative to the state. Districts in stronger accountability states lost perceived
authority over all categories (besides evaluation of teachers) relative to the
estado.

11. Results are available from the authors upon request.

28

EDUCATION FINANCE AND POLICY

yo

D
oh
w
norte
oh
a
d
mi
d

F
r
oh
metro
h

t
t

pag

:
/
/

d
i
r
mi
C
t
.

metro

i
t
.

/

/

F

mi
d
tu
mi
d
pag
a
r
t
i
C
mi
–
pag
d

yo

F
/

/

/

/

/

2
1
1
0
1
6
8
9
0
0
2
mi
d
pag
2
0
0
7
2
1
1
0
pag
d

.

.

.

F

.

.

F

b
y
gramo
tu
mi
s
t

t

oh
norte
0
8
S
mi
pag
mi
metro
b
mi
r
2
0
2
3

Susanna Loeb and Katharine Strunk

Mesa 5. Linear Regression of Principal Control Variables on Accountability Index and Control Variables’
93 Value (norte = 50)

1999

Average Curriculum Discipline Prof. desarrollador. Spending Evaluations Hiring

1999 Principal Influence

Accountability

1993 Principal
influence

0.02∗∗

0.37∗∗

0.02

0.32∗∗

−0.00

0.21

0.02∗

0.23∗∗

0.07∗∗

0.52∗∗

0.01

0.35∗∗

0.03∗∗

0.80∗∗

Constant

2.70∗∗

2.73∗∗

3.47∗∗

3.20∗∗

2.02∗∗

3.03∗∗

1.01∗∗

1999 Principal Influence Relative to State

Accountability −0.03∗∗ −0.05∗∗

−0,03

−0,03

−0,02

−0.06∗

1993 Principal
by state

0.56∗∗

0.28∗∗

0.57∗∗

0.36∗∗

0.63∗∗

0.71∗∗

0.05∗∗

0.70∗∗

Constant

0.76∗∗

0.90∗∗

0.82∗∗

1.15∗∗

0.98∗∗

0.62∗∗

0.89∗∗

1999 Principal Influence Relative to District

Accountability

0.01

0.01

1993 Principal
by district

0.53∗∗

0.81∗∗

0.02

0.53∗∗

−0.00

0.18

0.03

0.75∗∗

0.00

0.56∗∗

0.02

0.64∗∗

Constant

0.61∗∗

0.22

0.66∗∗

0.98∗∗

0.44∗

0.70∗∗

0.54∗∗

1999 District Influence Relative to State

Accountability −0.03∗∗ −0.04∗∗

−0.05∗∗

−0.03∗

−0,03

1993 District
by state

0.71∗∗

0.41∗∗

0.64∗∗

0.50∗∗

0.82∗∗

0.05∗∗

0.74∗∗

−0.06∗∗

0.62∗∗

Constant

0.38∗∗

0.71∗∗

0.52∗∗

0.79∗∗

0.40∗∗

0.46∗∗

0.83∗∗

∗ p < .10, ∗∗ p < .05. Effects on Student Outcomes The analyses of the relationship between student outcomes and the interac- tions of accountability strength and local control are based on five sets of state-level data: joint data consisting of black and white fourth and eighth graders; separate black and white fourth-grade data; and separate black and white eighth-grade data. We cluster the standard errors at the state level when using the “stacked” data set. Table 6 gives the average effects of accountability and local control on student outcome measures without including the interaction of these factors. The table reports the results for the change in the McLaughlin-adjusted scale scores, controlling for the 1996 score, the percentage of black and Hispanic students in the 1995–96 school year, the average yearly growth in the percent- age of black and Hispanic students, the 1996 state population, and the average yearly population growth.12 Most striking in this table is the finding that the 12. Similar results for the change in the percentage of students passing at and the change in the percentage of students passing at the proficient the basic level level as a l D o w n o a d e d f r o m h t t p : / / d i r e c t . m i t . f / / e d u e d p a r t i c e - p d l f / / / / / 2 1 1 0 1 6 8 9 0 0 2 e d p 2 0 0 7 2 1 1 0 p d f . . . . . f b y g u e s t t o n 0 8 S e p e m b e r 2 0 2 3 29 ACCOUNTABILITY AND LOCAL CONTROL Table 6. Analysis of 1996–2000 Change in Adjusted NAEP Math Scores Stacked Data (n = 129) White 4th Grade (n = 33) Black 4th Grade (n = 34) White 8th Grade (n = 31) Black 8th Grade (n = 31) 1.84∗∗ (0.67) –0.16 (0.65) –0.08 (0.99) 1.32∗∗ (0.57) 2.97∗∗ (0.99) –0.05 (0.06) –0.12 (0.15) 0.15 (0.17) 0.00 (0.12) –0.23 (0.14) Accountability 1996 NAEP math score % of 8th grade –5.28 (5.96) 8.00 (7.53) –5.63 (10.02) 4.28 (5.60) black/Hispanic ’96 –27.27∗∗ (10.34) Population in 1995 0.01 (0.01) 0.01 (0.02) 0.03 (0.02) –0.01 (0.01) –0.00 (0.02) (100,000s) 1996 per pupil 0.36 (0.40) 0.51 (0.81) –0.81 (0.91) 0.82 (0.70) 1.47 (1.01) revenue (1,000s) Yearly growth –45.69 (30.95) –11.27 (33.48) –55.29 (47.51) –15.68 (34.09) –89.93 (61.84) % black/Hispanic Yearly population growth 129.90 (84.61) 5.14 (96.94) 203.85 (133.63) 53.63 (110.23) 301.16 (187.77) No vote Court legislation % of funding— categorical Average principal control—’93 –0.55 (1.74) –0.18 (2.18) 0.12 (3.09) –2.59 (1.94) 1.20 (3.22) –1.59 (1.13) 0.02 (1.79) 3.29 (2.24) –0.33 (1.31) –2.06 (2.27) –16.13 (11.31) –15.93 (13.23) –22.58 (18.38) –27.30∗∗ (13.20) –3.86 (22.85) 0.80 (3.18) 3.50 (4.90) –4.49 (7.13) 0.16 (4.38) 6.39 (7.71) White 8th grader (W8) 2.49 (3.14) — Accountability × W8 –0.81 (0.78) — Black 4th grader (B4) 1.98 (3.93) — Accountability × B4 –1.26 (0.78) — White 4th grader (W4) 2.24 (2.55) — Accountability × W4 –1.19∗ (0.71) — — — — — — — — — — — — — — — — — — — Constant –1.19 (6.85) 13.76 (37.50) 0.79 (38.09) –4.56 (29.49) 21.91 (35.23) Adjusted R2 .22 –.16 –.05 .15 0.23 ∗ p < .10, ∗∗ p < .05. Standard errors in stacked model are clustered at the state level. positive effect of accountability holds up at significant levels using all three outcome variables in both eighth-grade specifications and the stacked data. This suggests that accountability policy strength is an important predictor of student performance at all points of the distribution curve, even when control- ling for the measures of local control, and especially so for students at the basic (lower) and mean levels. The results also show that accountability strength has function of accountability and the local control variables are available from the authors upon request. 30 EDUCATION FINANCE AND POLICY l D o w n o a d e d f r o m h t t p : / / d i r e c t . m i t . f / / e d u e d p a r t i c e - p d l f / / / / / 2 1 1 0 1 6 8 9 0 0 2 e d p 2 0 0 7 2 1 1 0 p d . . . . f . f b y g u e s t t o n 0 8 S e p e m b e r 2 0 2 3 Susanna Loeb and Katharine Strunk a larger positive effect on black than on white eighth graders on average and at the basic level. We do not find consistent evidence of direct effects of the local control measures on student outcomes, though the analysis is not designed to find these, since all of the local control measures are included simultaneously. The next series of analyses address the interaction of accountability and local control. Note that the models include all measures in table 6 except for the local control measures; for simplicity, though, only the local control and the accountability interaction coefficients are provided. Table 7 shows that lack of local control over revenue, as measured by increases in the percentage of total education revenue from categorical grants, negatively interacts with accountability strength to affect student performance. This finding is signifi- cant in our clustered stacked regressions as well as in our separate analysis of white fourth graders and is consistently negative in the rest of the analyses. In addition, while the linear effect of accountability was evident for eighth graders but not fourth graders, this negative interaction is just as strong for the fourth grade once these interactions are included. This suggests that al- though students perform better in states with higher accountability strength on average, this positive effect is less strong when districts have less discretion over resource allocation. The stacked estimations predict that a point increase in accountability will increase the percentage of students achieving at the basic level by 4.4 points on average in states with no categorical aid. In states with 10 percent of revenues through categorical aid, the effect of accountability drops to approximately 3.0 points. In states with 20 percent categorical revenues, the effect of a one-unit increase in accountability drops to only 1.5 points. Examin- ing the nonstacked predictions, we see that a point increase in accountability will increase the percentage of white fourth graders achieving at the basic level by 2.6 points in states with no categorical funding, whereas in states with the mean proportion of their overall funding in the form of categorical aid (12 percent), the effect of accountability drops to only one-third of a point. Table 7 also looks at the effects of an interaction between voting rights and accountability strength. The importance of local control is evident here as well. Voting rights are measured by whether states have no local voting provisions (the weakest level of local control). The estimates show that accountability is less effective in states that do not allow local voting. This effect is particularly strong for white eighth-grade students, but the negative interaction between accountability and local control is evident in the point estimates for all groups. For example, in a state that allows local voting for revenue raising, a level increase in accountability strength is associated with a 2.8 percent increase in the likelihood that a student will pass the NAEP math exam at the basic level, and with a 1.0 percent increase in the likelihood that a student will pass the NAEP math exam at the proficient level. However, when voting is l D o w n o a d e d f r o m h t t p : / / d i r e c t . m i t . / f / e d u e d p a r t i c e - p d l f / / / / / 2 1 1 0 1 6 8 9 0 0 2 e d p 2 0 0 7 2 1 1 0 p d . . f . . . f b y g u e s t t o n 0 8 S e p e m b e r 2 0 2 3 31 ACCOUNTABILITY AND LOCAL CONTROL ∗ ∗ 1 6 . 2 ) 7 8 . 0 ( 6 4 . 6 ) 4 7 . 0 1 ( 1 6 . 1 – ) 5 0 . 3 ( 4 2 . ∗ ∗ 2 0 . 4 ) 9 7 . 0 ( ∗ ∗ 1 9 . 2 2 ) 1 7 . 9 ( ∗ ∗ 5 4 . 6 – ) 2 9 . 2 ( ∗ ∗ 2 2 . 1 ) 0 5 . 0 ( ∗ 5 2 . 0 1 ) 2 0 . 6 ( ∗ ∗ 2 1 . 4 – ) 1 7 . 1 ( 5 2 0 . 9 2 0 – . ∗ ∗ 0 7 1 . y t i l i b a t n u o c c A ∗ ∗ 7 6 4 . ∗ ∗ 7 5 2 . s e r o c S l e a c S d e t s u d A n j i l h g u a L c M ) 9 8 0 . ( ) 7 5 0 . ( ) 8 6 0 . ( ) 2 7 1 . ( ) 0 0 1 . ( 2 8 . 5 9 2 4 . 5 9 5 . e t o v o N 1 6 . 8 4 9 6 4 – . ) 4 3 8 . ( ) 9 4 5 . ( ) 4 2 4 . ( ) 9 4 5 3 . ( ) 3 8 0 2 . ( 8 0 2 – . 5 6 1 – . ∗ 8 1 2 – . ∗ y t i l i b a t n u o c c A . 5 3 8 1 – . 5 5 0 1 – ) 3 4 2 . ( ) 0 6 1 . ( ) 0 3 1 . ( e t o v o N ) 2 7 2 1 . ( ) 1 2 7 . ( 4 2 . 1 ) 2 6 . 1 ( 9 4 . 5 – ) 9 8 . 6 2 ( 7 8 . 8 – ) 1 1 . 1 1 ( 0 3 . 1 ) 0 6 . 0 ( 1 4 . 4 1 ) 1 5 . 7 1 ( ) 0 1 . 7 ( ∗ 5 9 . 2 1 – ∗ ∗ 2 4 . 3 ) 1 8 . 0 ( 9 8 . 3 1 ) 6 2 . 6 1 ( ∗ ∗ 6 4 . 3 1 – ) 6 2 . 5 ( l a c i r o g e t a c — g n d n u f i f o % l a c i r o g e t a c % × y t i l i b a t n u o c c A y t i l i b a t n u o c c A ∗ ∗ 9 5 . 1 ) 4 5 . 0 ( ∗ 7 2 . 2 1 ) 6 4 . 6 ( ∗ ∗ 5 2 . 4 – ) 3 8 . 1 ( 9 6 0 . 3 3 0 . ∗ ∗ 9 7 2 . y t i l i b a t n u o c c A ∗ ∗ 2 5 4 . ∗ 8 3 2 . l e v e L c i s a B e h t t a g n i s s a P t n e c r e P ) 1 3 1 . ( ) 2 6 0 . ( ) 6 5 0 . ( ) 4 3 2 . ( ) 9 1 1 . ( 1 0 9 . 4 2 . 0 1 ∗ ∗ 9 3 3 1 . e t o v o N 5 4 . 0 4 4 2 1 – . ) 4 1 3 1 . ( ) 3 1 6 . ( ) 4 7 4 . ( ) 0 6 5 5 . ( ) 2 3 4 2 . ( 0 8 . 3 – ∗ 4 5 3 – . ∗ ∗ 7 5 4 – . ∗ y t i l i b a t n u o c c A 0 1 6 – . 7 7 6 – . ) 6 1 4 . ( ) 7 7 1 . ( ) 2 3 1 . ( e t o v o N ) 6 1 8 1 . ( ) 5 4 8 . ( 1 9 . 3 ) 3 8 . 2 ( 8 4 . 2 3 ) 3 1 . 8 4 ( ) 9 4 . 0 2 ( 1 9 . 8 2 – ∗ ∗ 0 6 . 2 ) 3 2 . 1 ( 4 1 . 4 3 ) 0 7 . 0 2 ( ∗ ∗ 0 8 . 8 1 – ) 0 3 . 8 ( ∗ ∗ 3 4 . 4 ) 7 1 . 1 ( 2 6 . 4 2 ) 4 0 . 7 1 ( ∗ ∗ 0 6 . 4 1 – ) 7 2 . 7 ( l a c i r o g e t a c — g n d n u f i f o % l a c i r o g e t a c % × y t i l i b a t n u o c c A y t i l i b a t n u o c c A 3 2 . 2 1 . – 9 0 . – 0 2 . 2 R 0 3 0 . 3 2 . 4 0 . – 5 0 . 4 2 . 2 R 4 5 . 0 3 . 5 0 . – 3 1 . 5 3 . 2 R d e t s u d A j 8 3 0 . 9 1 . 2 0 . 6 1 . 9 2 . 0 2 R d e t s u d A j l D o w n o a d e d f r o m h t t p : / / d i r e c t . m i t . f / / e d u e d p a r t i c e - p d l f / / / / / 2 1 1 0 1 6 8 9 0 0 2 e d p 2 0 0 7 2 1 1 0 p d f . . . . . f b y g u e s t t o n 0 8 S e p e m b e r 2 0 2 3 8 B 8 W 4 B 4 W d e k c a t S 8 B 8 W 4 B 4 W d e k c a t S g n i t o V n e z i t i C l a c o L d n a l a c i r o g e t a C t n e c r e P d n a y t i l i b a t n u o c c A n o s e r u s a e M h t a M P E A N n i e g n a h C 0 0 0 2 – 6 9 9 1 f o i s s y l a n A . l 7 e b a T 32 EDUCATION FINANCE AND POLICY Susanna Loeb and Katharine Strunk ∗ ∗ 7 3 . 1 ) 2 2 . 0 ( ∗ ∗ 0 4 . 1 ) 8 4 . 0 ( 4 3 . 4 ) 4 6 . 2 ( ) 0 8 . 0 ( 8 1 . 1 − 8 1 . 7 ) 6 8 . 5 ( ∗ 6 8 . 2 − ) 7 6 . 1 ( 2 3 . 0 . 4 1 0 − ∗ ∗ 7 9 0 . y t i l i b a t n u o c c A 7 7 0 . ∗ ∗ 1 7 2 . l e v e L t n e i c fi o r P e h t t a g n i s s a P t n e c r e P ) 3 4 0 . ( ) 1 6 0 . ( ) 5 2 0 . ( ) 9 4 0 . ( ) 3 9 0 . ( 0 9 . 0 − 1 7 4 . 4 8 3 . e t o v o N . 8 3 0 − 1 6 3 . ) 4 8 4 . ( ) 9 4 . 1 ( . 1 3 0 − ) 5 7 1 . ( ) 7 9 0 . ( e t o v o N ) 5 9 3 . ( ) 3 7 6 . ( ) 3 0 6 . ( ) 1 8 2 . ( ) 2 1 1 1 . ( ) 3 5 9 1 . ( . 3 9 1 − ∗ 4 6 1 − . ∗ y t i l i b a t n u o c c A 3 1 4 . . 0 4 0 1 − 5 5 . 1 ) 4 0 . 1 ( 4 9 . 3 2 ) 1 3 . 7 1 ( ) 5 5 . 7 ( 6 7 . 9 − 0 7 . 1 ) 7 1 . 1 ( 9 0 . 3 3 ) 4 7 . 9 1 ( ) 7 8 . 7 ( ∗ 8 9 . 4 1 − ∗ ∗ 9 1 . 2 ) 7 6 . 0 ( ∗ ∗ 4 6 . 9 1 ) 4 5 . 9 ( y t i l i b a t n u o c c A l a c i r o g e t a c — g n d n u f i f o % ) 2 6 . 4 ( ∗ ∗ 5 5 . 0 1 − l a c i r o g e t a c % × y t i l i b a t n u o c c A 8 B 8 W 4 B 4 W d e k c a t S 8 B 8 W 4 B 4 W d e k c a t S d e u n i t n o C . l 7 e b a T e t a t s e h t n i s r e d a r g h t h g e i e h t f o t n e c r e p , n o i t a u p o p l e t a t s 6 9 9 1 i , g n s s a p e g a t n e c r e p r o e r o c s t s e t 6 9 9 1 e h t r o f l o r t n o c l s e d o M . 5 0 . < p ∗ ∗ , 0 1 . < p ∗ 4 7 . 8 1 . 6 0 − . 4 0 . 5 3 . 2 R d e t s u d A j 4 7 0 . 4 2 . 2 0 . − 1 1 . 6 3 . 0 2 R d e t s u d A j . h t w o r g n o i t a u p o p l y l r a e y d n a 6 9 9 1 d n a , n o i t a u p o p l e t a t s 6 9 9 1 , s t n e d u t s i i c n a p s H / k c a b l f o n o i t r o p o r p n i h t w o r g y l r a e y , 6 9 9 1 n i c i n a p s H i r o k c a b l e r a o h w . s e t a m i t s e r o r r e d r a d n a t s r o f l e v e l e t a t s e h t t a g n i r e t s u c l e s u d n a s n o i t c a r e t n i d n a i s e m m u d e d a r g / e c a r e d u l c n i l s e d o m d e k c a t S l D o w n o a d e d f r o m h t t p : / / d i r e c t . m i t . f / / e d u e d p a r t i c e - p d l f / / / / / 2 1 1 0 1 6 8 9 0 0 2 e d p 2 0 0 7 2 1 1 0 p d . . . . . f f b y g u e s t t o n 0 8 S e p e m b e r 2 0 2 3 33 ACCOUNTABILITY AND LOCAL CONTROL Table 8. Analysis of 1996–2000 Change in Math Scores on Accountability and Court Legislation Stacked Data White 4th Grade Black 4th Grade White 8th Grade Black 8th Grade McLaughlin Adjusted Scale Scores Accountability Court legislation Accountability × court legislation R2 Accountability Court legislation Accountability × court legislation R2 Accountability Court legislation Accountability × court legislation R2 2.03∗∗ (0.62) 2.29 (2.22) −1.67∗∗ (0.76) .22 3.03∗∗ (0.63) 2.03 (2.47) −1.42 (1.09) .28 1.15∗∗ (0.27) 1.93 (1.30) −0.95∗ (0.57) .33 0.03 (−0.11) 3.86 (2.91) −1.79 (1.05) −.02 0.83 (0.92) 3.03 (4.03) −2.45 (1.56) .00 Percent Passing at the Basic Level 0.50 (0.72) 2.56 (3.28) −1.39 (1.27) .02 1.29 (1.25) 5.28 (7.13) −4.21 (2.61) .14 Percent Passing at the Proficient Level 0.12 (0.67) 2.50 (3.03) −1.41 (1.18) .03 0.58 (0.48) 2.23 (2.88) −1.20 (1.06) −.04 1.38∗∗ (0.61) 3.88 (2.75) −1.79 (1.07) −.00 1.65∗∗ (0.65) 2.51 (2.88) −1.21 (1.12) −.13 1.58∗∗ (0.55) 3.36 (2.47) −1.47 (0.96) .08 2.84∗∗ (0.93) −0.66 (4.17) −0.84 (1.61) 0.28 3.98∗∗ (1.05) 0.06 (6.19) −0.37 (2.18) 0.35 1.19∗ (0.25) −0.99 (1.47) 0.49 (0.51) 0.71 ∗ p < .10, ∗∗ p < .05. Models control for the 1996 test score or percentage passing, 1996 state population, percentage of the eighth graders in the state who are black or Hispanic in 1996, yearly growth in proportion of black/Hispanic students, 1996 state population, and 1996 and yearly population growth. Stacked models include race/grade dummies and interactions and use clustering at the state level for standard error estimates. not authorized, accountability has a negative effect. A one-point increase in accountability strength results in a 1.8 percentage point decrease in the number of students who pass the NAEP exam at the basic level and in a 0.7 percentage point decrease in the percentage achieving proficiency. These findings are consistent with our hypothesis that local control over resources is an important factor in allowing schools and districts to respond to the incentives created by accountability reforms.13 Table 8 looks at the effects of accountability and whether or not states’ supreme courts have overturned school finance systems that heavily rely on 13. We ran similar analyses for tax limitation but did not find any effect of the interaction of account- ability and tax limits on student outcomes. The results are available upon request. 34 EDUCATION FINANCE AND POLICY l D o w n o a d e d f r o m h t t p : / / d i r e c t . m i t . / / f e d u e d p a r t i c e - p d l f / / / / / 2 1 1 0 1 6 8 9 0 0 2 e d p 2 0 0 7 2 1 1 0 p d . . . . f . f b y g u e s t t o n 0 8 S e p e m b e r 2 0 2 3 Susanna Loeb and Katharine Strunk local financing. We see a consistently negative interaction between accountabil- ity strength and states’ court legislation status, although this is only significant in our stacked regressions in the “adjusted level” and “percentage passing at the proficient level” analyses. (This effect is almost significant at the 10 percent level for both white scale score specifications.) This tells us that in a state with the mean level of accountability strength (accountability equals 2.12), students, on average, are likely to score 4.3 points higher on the NAEP math exam. How- ever, in states where courts have overturned local school financing, students are only likely to perform 3.05 points higher on the NAEP math exam—1.25 points lower than students would in states without such court legislation. This indicates that accountability policies may be less effective as school funding be- comes more centralized, and perhaps more so for average- and high-achieving students than for low-performing students. Table 9 describes the effects of accountability and perceived principal con- trol measures. The results continue to show more positive effects of account- ability in states with greater local control. We see a consistently positive effect of the interaction between principal control and accountability strength. Specif- ically, stronger accountability reforms are associated with higher proportions of students scoring at or above the basic and proficient levels in states in which principals have greater control over spending. This finding is consistent with the theory that principals need control over resource allocation in order to effectively respond to accountability. In summary, these analyses provide evidence that the estimates of ac- countability are robust to the inclusion of local control measures. Most of our measures of local control do not show a statistically significant direct effect on student outcomes, although this could easily be due to a combination of a lack of power, colinearity, and measurement error. There is, however, sub- stantial evidence that accountability policies are more effective when there is more local control. This comes from the negative interactions of accountability with increases in the percentage of education funding from categorical grants, state supreme court legislation, and no voting provisions, as well as from the positive interaction of accountability and principal influence measures.14 14. One concern with the analysis is that because the sample size is so small, a single state might be driving the results. In order to check for this, we reran the analyses eliminating a different state each time. In general, the results were not affected by this. The elimination of individual states had no substantive effect on our estimate of the interaction between accountability and either percentage categorical or principal control. The elimination of Virginia from the estimate of the interaction between accountability and no local voting authority causes the point estimate to lose significance at all conventional levels, though it is still negative. We also use the dfbeta command in Stata to determine whether our results are sensitive to a single or a small number of observations. We find no evidence that this is the case. l D o w n o a d e d f r o m h t t p : / / d i r e c t . m i t . f / / e d u e d p a r t i c e - p d l f / / / / / 2 1 1 0 1 6 8 9 0 0 2 e d p 2 0 0 7 2 1 1 0 p d . . . . f . f b y g u e s t t o n 0 8 S e p e m b e r 2 0 2 3 35 ACCOUNTABILITY AND LOCAL CONTROL 4 0 . 1 ) 1 2 . 6 ( 5 3 . 2 ) 4 0 . 4 ( 8 0 . 0 ) 0 5 . 1 ( 8 5 . 9 − ) 0 1 . 2 1 ( ) 7 8 . 7 ( 8 7 . 6 − 0 6 . 2 ) 1 9 . 2 ( 1 7 . 2 0 . 9 8 . 1 ) 2 9 . 2 ( 4 3 . 1 ) 6 8 . 1 ( ) 7 6 . 0 ( 2 1 . 0 − ) 0 1 . 6 ( 9 6 . 4 − ) 3 9 . 3 ( 2 0 . 3 − 7 3 . 1 ) 0 4 . 1 ( 2 0 . 8 − ) 0 5 . 1 1 ( ) 5 2 . 7 ( 4 4 . 1 − 6 0 . 2 ) 6 7 . 2 ( 5 0 . − ) 4 9 . 4 ( 8 1 . 5 − ) 5 1 . 3 ( 7 0 . 0 − 6 3 . 1 ) 3 1 . 1 ( 7 0 . 0 2 − . 1 2 9 − y t i l i b a t n u o c c A 6 4 . 2 1 − . 8 5 9 − ) 1 8 3 1 . ( ) 9 6 6 . ( . 2 5 0 1 − . 2 5 5 − ) 0 7 8 . ( ) 0 3 4 . ( l a p i c n i r p e g a r e v A 3 9 ’ l o r t n o c ) 2 9 7 2 . ( ) 7 7 3 1 . ( . 7 0 4 − . 4 2 7 − ) 5 6 6 2 . ( ) 2 0 9 . ( 7 7 4 . 6 4 2 . × y t i l i b a t n u o c c A 5 0 4 . 3 6 2 . ) 1 3 3 . ( ) 2 6 1 . ( l o r t n o c . n i r p . g v a ) 3 6 6 . ( ) 1 3 3 . ( 6 0 . 3 3 . 2 R d e t s u d A j 8 3 0 . 1 1 . . 2 5 1 1 − . 5 3 4 − y t i l i b a t n u o c c A . 5 9 2 − . 3 6 2 − ) 2 9 6 . ( ) 6 7 2 . ( . 6 4 5 − . 2 1 2 − ) 4 3 4 . ( ) 0 8 1 . ( 1 6 2 . ∗ 4 2 1 . ) 8 5 1 . ( ) 4 6 0 . ( - l e r i h − o r t n o c . n i r p × y t i l i b a t n u o c c A l o r t n o c l a p i c n i r P 3 9 ’ e r i h — ) 9 1 2 1 . ( ) 8 9 6 . ( . 9 4 1 − . 9 2 2 − ) 0 3 0 1 . ( ) 5 5 4 . ( 5 6 1 . 2 9 0 . ) 6 7 2 . ( ) 0 6 1 . ( ) 1 0 . 9 2 ( 0 3 . 0 2 − 0 0 . 6 − ) 4 2 . 8 1 ( 2 1 . 5 ) 9 9 . 6 ( 7 0 . − ) 1 0 . 4 1 ( 8 6 . 7 1 − ) 5 0 . 9 ( 4 9 . 6 − 2 3 . 4 ) 2 2 . 3 ( 1 7 . 7 − ) 7 4 . 5 1 ( ) 0 7 . 9 ( 5 0 . 4 − 9 8 . 1 ) 1 7 . 3 ( 2 0 . − ) 7 8 . 7 ( 5 3 . 5 − ) 0 9 . 4 ( 2 0 . 3 − 6 2 . 1 ) 0 8 . 1 ( 0 7 . 3 0 . 5 1 . 8 0 . 3 3 . 2 R d e t s u d A j 8 3 0 . 0 1 . 2 0 . 1 0 . − 6 0 . 2 ) 1 0 . 4 ( 2 5 . 0 ) 8 1 . 3 ( ) 8 0 . 6 ( ∗ 5 1 . 2 1 − ) 9 9 . 3 ( 6 2 . 6 − ) 4 0 . 1 ( 0 2 . 0 − ∗ ∗ 9 4 . 3 ) 8 5 . 1 ( 8 6 . 0 2 . ) 3 0 . 6 ( 7 5 . 7 − ) 4 2 . 4 ( 3 3 . 5 − 4 0 . 2 ) 6 5 . 1 ( 3 0 . − ) 8 6 7 . ( ) 8 1 4 . ( ∗ ∗ 8 2 . 0 1 − ∗ 7 0 6 − . ) 6 0 5 . ( ) 8 8 2 . ( 3 9 ’ d n e p s — l o r t n o c l a p i c n i r P ) 1 1 9 1 . ( ) 9 0 7 . ( . 7 9 2 2 − ∗ 9 9 7 − . ) 1 7 9 1 . ( ) 7 6 4 . ( ∗ ∗ 4 6 . 7 1 − ∗ 9 9 8 − . y t i l i b a t n u o c c A . 7 7 3 1 − . ∗ 6 4 2 1 − ∗ ∗ 6 5 4 . 1 6 2 . × y t i l i b a t n u o c c A 5 4 4 . ∗ 0 6 3 . ) 1 0 2 . ( ) 9 0 1 . ( d n e p s — o r t n o c l . n i r p ) 1 9 4 . ( ) 4 8 1 . ( 6 1 . 7 3 . 2 R d e t s u d A j 2 4 0 . 2 2 . ) 6 4 . 6 1 ( 8 9 . 0 1 − 5 0 . 9 − ) 2 3 . 1 1 ( 8 9 . 2 ) 0 3 . 4 ( 8 0 . − ) 5 0 . 9 ( 2 3 . 0 1 − ) 2 0 . 6 ( 1 0 . 6 − 4 7 . 2 ) 6 3 . 2 ( 2 0 . ) 4 5 . 1 1 ( 9 8 . 1 1 − ) 2 6 . 3 ( 2 8 . 7 − 3 5 . 3 ) 6 8 . 2 ( 7 2 . 0 ) 7 1 . 4 ( 3 1 . 6 − ) 1 4 . 3 ( 0 2 . 4 − ∗ ∗ 6 0 . 2 ) 2 0 . 1 ( 8 2 . 0 ∗ 3 2 . 9 − ) 1 5 . 5 ( ∗ 8 4 . 8 − ) 8 6 . 4 ( 5 1 . 3 ) 1 0 . 2 ( 8 1 . 0 l o r t n o c . n i r p . g v a × y t i l i b a t n u o c c A 3 9 ’ l o r t n o c y t i l i b a t n u o c c A 2 R d e t s u d A j l i a p c n i r p e g a r e v A y t i l i b a t n u o c c A l o r t n o c l i a p c n i r P 3 9 ’ e r i h — - l e r i h − o r t n o c . n i r p × y t i l i b a t n u o c c A d n e p s — o r t n o c l . n i r p × y t i l i b a t n u o c c A 3 9 ’ d n e p s — l o r t n o c l i a p c n i r P y t i l i b a t n u o c c A 2 R d e t s u d A j 2 R d e t s u d A j ∗ ∗ , 0 1 . < p ∗ r o k c a b l e r a o h w e t a t s e h t n i s r e d a r g h t h g e i e h t f o e g a t n e c r e p , n o i t a u p o p l e t a t s 6 9 9 1 i , g n s s a p e g a t n e c r e p r o e r o c s t s e t 6 9 9 1 e h t r o f l o r t n o c l s e d o M . 5 0 . < p l D o w n o a d e d f r o m h t t p : / / d i r e c t . m i t . / / f e d u e d p a r t i c e - p d l f / / / / / 2 1 1 0 1 6 8 9 0 0 2 e d p 2 0 0 7 2 1 1 0 p d . f . . . . f b y g u e s t t o n 0 8 S e p e m b e r 2 0 2 3 e d a r g / e c a r e d u l c n i l s e d o m d e k c a t S l . h t w o r g n o i t a u p o p y l r a e y d n a 6 9 9 1 d n a , n o i t a u p o p e t a t s 6 9 9 1 , s t n e d u t s l i l i c n a p s H / k c a b f o n o i t r o p o r p n i h t w o r g y l r a e y , 6 9 9 1 n i i c n a p s H i . l e v e l e t a t s e h t t a d e r e t s u l c e r a s r o r r e d r a d n a t s d n a s n o i t c a r e t n i d n a i s e m m u d 8 B 8 W 4 B 4 W d e k c a t S 8 B 8 W 4 B 4 W d e k c a t S l e v e L t n e i c fi o r P e h t t a g n i s s a P t n e c r e P l e v e L c i s a B e h t t a g n i s s a P t n e c r e P y t i r o h t u A l a p i c n i r P f o s e r u s a e M d n a y t i l i b a t n u o c c A n o l s e v e L t n e i c fi o r P d n a c i s a B e h t t a i g n s s a P e g a t n e c r e P n i e g n a h C 0 0 0 2 – 6 9 9 1 f o i s s y l a n A . l 9 e b a T 36 EDUCATION FINANCE AND POLICY Susanna Loeb and Katharine Strunk 6. CONCLUSIONS AND POLICY IMPLICATIONS Our intent with this study is to begin assessment of the relationship between accountability and local control. We first find evidence that stronger account- ability policies were implemented in states with weaker local control, as mea- sured by local provisions for revenue raising and allocation. This finding seems intuitive but is relevant to the policy discussion surrounding accountability programs. If the ability for districts and schools to respond to accountability provisions depends on the amount of flexibility and control they have over their school- and district-based activities, then the states that are least able to respond to accountability programs are exactly those in which accountability policies are being instituted most strongly. Next we look at changes in local control associated with the implemen- tation of accountability reforms. We find little difference in principal control over school operations in the states that implemented different strength ac- countability policies. However, the implementation of stronger accountability corresponds to increases in principals’ perceived control over school opera- tions, through drops in principals’ assessment of their power and their dis- trict’s power relative to the state. This points to the possibility that principals do not feel disenfranchised by the implementation of strong accountability policies, as one might have guessed, but instead seemingly feel more in control with these policies. The main analyses in this article address the differential effect of account- ability reforms on student math achievement in states that had greater and lesser local control. The results are directly relevant to current policy discus- sions on accountability. First, we provide supporting evidence to Carnoy and Loeb’s (2003) findings that accountability policies led to increases in student performance on NAEP math exams. Accountability policies may be tools for educational policy makers to use in the quest to improve student outcomes. However, our own and other research suggests that accountability policies do not exist in a vacuum: many factors constrain the implementation and effec- tiveness of accountability programs. Our analyses, while imperfect, point to the importance of local control over revenue raising and allocation in the success- ful implementation of accountability programs. We find that accountability policies were substantially more effective in states with stronger local control. It appears that without some local control, even well-thought-out accountabil- ity policies will be less effective, and sometimes ineffective and harmful. This problem is further augmented by the recent trend away from local and toward state control over education funding. As education finance continues to be cen- tralized at the state level, citizens and districts lose control over revenue raising and allocation, potentially impeding the positive effects of state-implemented accountability policies. l D o w n o a d e d f r o m h t t p : / / d i r e c t . m i t . / f / e d u e d p a r t i c e - p d l f / / / / / 2 1 1 0 1 6 8 9 0 0 2 e d p 2 0 0 7 2 1 1 0 p d . . . f . . f b y g u e s t t o n 0 8 S e p e m b e r 2 0 2 3 37 ACCOUNTABILITY AND LOCAL CONTROL We appreciate the helpful guidance from David Figlio and Jim Guthrie, as well as two anonymous reviewers. This work was partially funded by the American Educa- tion Research Association Dissertation Grant. Any remaining errors are the authors’ own. REFERENCES Amrein, Audrey L., and David C. Berliner. 2003. The effects of high-stakes testing on student motivation and learning. Educational Leadership 60:32–38. Card, David, and Alan B. Krueger. 1992. Does school quality matter? Returns to edu- cation and the characteristics of public schools in the United States. Journal of Political Economy 100:1–40. Card, David, and A. Abigail Payne. 2002. School finance reform, the distribution of school spending, and the distribution of SAT scores. Journal of Public Economics 83:49– 82. Carnoy, Martin, and Susanna Loeb. 2003. Does external accountability affect student outcomes? A cross-state analysis. Education Evaluation and Policy Analysis 24 (4): 305–31. Coleman, James S. 1966. Equality of educational opportunity. Washington, DC: U.S. Government Printing Office. Cullen, Julie B., and Susanna Loeb. 2004. School finance reform in Michigan: Evalu- ating proposal A. In Helping children left behind: State aid and the pursuit of educational equity, edited by John Yinger, pp. 215–50. Cambridge, MA: MIT Press. Dee, Thomas S. 2002. Standards and student outcomes: Lessons from the “first wave” of education reform. Cambridge, MA: Kennedy School of Government. Downes, Thomas A. 1992. Evaluating the impact of school finance reform on the provision of public education: The California case. National Tax Journal 45 (4): 405–19. Duncombe, William, and Jocelyn M. Johnston. 2002. Is something better than nothing? An assessment of school finance reform in Kansas.Working paper, Syracuse University. Figlio, David N. 1997. Did the “tax revolt” reduce school performance? Journal of Public Economics (65):245–69. Figlio, David N. 2002. Funding and accountability: Some conceptual and technical issues in state aid reform. Paper presented at the Conference on State Aid to Education, Syracuse University, April. Figlio, David N., and Lawrence S. Getzler. 2002. Accountability, ability and disability: Gaming the system. NBER Working Paper No. 9307. Figlio, David N., and Kim S. Reuben. 2001. Tax limits and the qualifications of new teachers. Journal of Public Economics 80: 49–71. Hannaway, Jane. 1993. Political pressure and decentralization in institutional organi- zations: The case of school districts. Sociology of Education 66 (3): 147–63. Hanushek, Eric A. 1986. The economics of schooling: Production and efficiency in public schools. Journal of Economic Literature 24 (3): 1141–77. 38 EDUCATION FINANCE AND POLICY l D o w n o a d e d f r o m h t t p : / / d i r e c t . m i t . / f / e d u e d p a r t i c e - p d l f / / / / / 2 1 1 0 1 6 8 9 0 0 2 e d p 2 0 0 7 2 1 1 0 p d . . . f . . f b y g u e s t t o n 0 8 S e p e m b e r 2 0 2 3 Susanna Loeb and Katharine Strunk Hanushek, Eric A. 1997. Assessing the effects of school resources on student perfor- mance: An update. Educational Evaluation and Policy Analysis 19 (2): 141–64. Hanushek, Eric A., and Macke E. Raymond. 2002. Lessons and limits of state account- ability systems. Paper presented at “Taking Account of Accountability: Assessing Policy and Politics” Conference, Harvard University, June. Hanushek, Eric A., and Macke E. Raymond. 2004. Does school accountability lead to improved student performance? NBER Working Paper No. 10591. Imazeki, Jennifer, and Andrew A. Reschovsky. 2004. School finance reform in Texas: A never ending story? In Helping children left behind: State aid and the pursuit of educational equity, edited by John Yinger, pp. 251–82. Cambridge, MA: MIT Press. Jacob, Brian A. 2002. Test-based accountability and student achievement gains: Theory and evidence. Working paper, Harvard University. Jacob, Brian A., and Steven D. Levitt. 2003. Rotten apples: An investigation of the prevalence and predictors of teacher cheating. Quarterly Journal of Economics 118 (3): 843–77. Loeb, Susanna, and Marianne E. Page. 2000. Examining the link between teacher wages and student outcome: The importance of alternative labor market opportunities and non-pecuniary variation. Review of Economics and Statistics 82 (3): 393–408. McLaughlin, Don. 2001. Exclusions and accommodations affect state NAEP gain statis- tics: Mathematics, 1996–2000. American Institutes for Research (mimeo). Murray, Sheila E., William M. Evans, and Robert M. Schwab. 1998. Education finance reform and the distribution of resources. American Economic Review 88:789–812. National Center nance programs of the United States and Canada: http://nces.ed.gov/edfin/state finance/StateFinancing.asp. Accessed 11 July 2006. school fi- 1998–99. Available at for Education Statistics (NCES). 2001. Public Sonstelie, Jon, Eric Brunner, and Kenneth Ardon. 2000. For better or for worse? School finance reform in California. San Francisco: Public Policy Institute of California. l D o w n o a d e d f r o m h t t p : / / d i r e c t . m i t . / f / e d u e d p a r t i c e - p d l f / / / / / 2 1 1 0 1 6 8 9 0 0 2 e d p 2 0 0 7 2 1 1 0 p d . . . . . f f b y g u e s t t o n 0 8 S e p e m b e r 2 0 2 3 39
Descargar PDF