Application for Education

 

After reading this handout, please answer the following questions.

1. What are the economic arguments for and against state involvement in financing and producing education?
2. What social and economic characteristics of a state might influence the choice of how to produce education?

CHAPTER

HEADLINES

The last chapter does not involve specific services or expenditures, but

rather focuses on the overall economic and fiscal effects of individual state

and local government fiscal behavior. Although economic conditions in a

jurisdiction are different and separate from fiscal conditions of the govern-

ment for that jurisdiction, it is important to consider the relationship

between economic and fiscal conditions. That states compete for economic

activity is obvious; whether that competition is effective in increasing
welfare is not.

EDUCATION

An Act
To close the achievement gap with accountability, flexibility, and choice, so that
no child is left behind. . . . Each State plan shall demonstrate that the State has

adopted challenging academic content standards and challenging student acade-
mic achievement standards . . . Each State plan shall demonstrate that the State
educational agency, in consultation with local educational agencies, has imple-

mented a set of high-quality, yearly student academic assessments . . 1

—NO CHILD LEFT BEHIND ACT OF 2001

494

“A STATE JUDGE RULED LAST NIGHT THAT AN ADDITIONAL $5.6 BILLION MUST BE SPENT ON

[NEW YORK CITY’S] PUBLIC SCHOOLCHILDREN EVERY YEAR TO ENSURE THEM THE OPPORTUNITY

FOR A SOUND BASIC EDUCATION THAT THEY ARE GUARANTEED UNDER THE STATE CONSTITUTION,

BEYOND THAT, ANOTHER $9.2 BILLION MUST BE SPENT OVER THE NEXT FIVE YEARS TO

SHRINK CLASS SIZES, RELIEVE. OVERCROWDING, AND PROVIDE THE CITY’S 1.1 MILLION STUDENTS

WITH ENOUGH LABORATORIES, LIBRARIES, AND OTHER PLACES IN WHICH TO LEARN,

TILE DECISION IS A LANDMARK IN ONE OF THE NATIONS BIGGEST SCHOOL-FINANCE

CASES. . . THOUGH VIRTUALLY EVERY STATE IN THE NATION HAS BEEN EMBROILED IN LAWSUITS

OVER SCHOOL SPENDING, THE NEW YORK SUIT HAS BEEN MORE CLOSELY WATCHED, IN PART

BECAUSE OF THE NUMBER OF STUDENTS AND THE DOLLAR FIGURES AT STAKE. 2 ”

‘U.S. Department of ‘Education. Public Law print of I’L 197-110, the No Child left hint Ad Of 2007,
http: // www.eclgov /policy/elsealeg/esea02/inciex.html.

`Winter, Greg, “Judge Orders Billions ut Aid to City Schools.” The New York Tin http://nytimes.com ,

Febrvaty 15, 2005.

495

PART V ■
APPLICATIONS AND POLICY ANALYSIS

ER NINETEEN ■ EDUCATION

816 2,162 19:1
375

3,849 2,055 22:1 2,275
‘Total enrollment in fall of that school year.

‘Current expenditures per pupil in average daily attendance.

SOURCE: U.S. Department of Education.
Digest of Education Statistics,

2003, 2004.

Education is, by almost any measure, thepr
rn primary se rvice p by state-local

governments in the United States. You have already learned rovided

that expenditures on
elementary and secondary education represent the single largest category of
state-local government spending, equal to nearly a quarter of aggregate subna-
tional government direct general expenditure in 2003. Elementary

and

secondary

education is an even larger fraction of local government spending, nearly 41 per-
cent in 2003. This is five times as great as local spending for police and fire protec-
tion and about nine times as great as local spending on roads. Public elementary
and secondary education teachers represent about 15 percent of total state-local
employees and about 25 percent of local government employees.

In the 2001-2002 academic year, expenditures for public

elementary and

secondary schools were nearly $431 billion, equal to about 4.3 percent of GNP and
$8,203 per student in average daily attendance at those schools, as shown in
Table 19.1. Public-school expenditures have increased substantially over this
period–by more than 100 percent just since 1990-but generally remained
between 3.5 and 4.0 percent of GDP during this time. Expenditures per pupil also
have increased substantially over the past 40 years, even in real terms (after
adjustment for inflation). In fact, real expenditures per pupil by

public elementary

in 1960 (see Figure 19.1). and secondary schools were about three and one-half times as great in 2002 than

In Fall of 2001, 47.7 million students were enrolled in these public schools.
Public-school enrollment generally increased in the 1950s and

1960s-peaking in elementary schools in the late 1960s and in secondary schools in the mid-1970s, as
shown in Figure 19.2. After that time, public- (and private-) school enrollment
decreased, largely because of demographic factors, until the mid-1980s. Elemen-
tary school enrollment began to increase again in 1985 and secondary school
enrollment increased starting in 1991. Even more importantly, the fraction of the

2002 $430.6
2000 381.8
1995 279.

0

1990 212.8
1985 137.0

Spending as a Percentage per Pupilb Pupil- Year (billion $) (thousands) of GDP Teachers per Pupil
Teacher

Pupils
Spending Spending Spending Puil’

Overview Ot Public Elementary and Secondary Education, Various years

47,688 4.3%
46,857 4.1
44,111 4.0
40,543 3.9
39,208 3.5
41,651 3.7
45,5

50

35,182

4.1
3.1

(current $) (2001–2002$)b (thousands)
Ratio

$8,203 $8,203

7,394 2,998 16:1

5,989
7,782 2,911 17:1

4,980
7,095 2,552 18:1

3,470
6,988 2,357 17:1

2,272 5,214
5,847 2,168 18:1

7,000

8,000
Constant 2001-02 dollars

……….._____…-4-” 5,000 -• – • • •- •- -• • •- –

4,000

1,408 26:1
3,000

2,000

1,000

Current dollars

1970-71 1975-76 1980-81
0 i . 1 t

1985-86 1990-91 1995-96 2001-02

School year

souRcE: U.S. Department of Education. Digest of Education Statistics, 2004.

u.

10

E

20

c

40

8 30 –

50

………

Elementary ——
——-

Total

Secondary

0
1960 1965 1970 1975 1980 1985 1990 1995 2002

School year beginning

T
ea

ch
er

s,
in

m
ill

io
ns

2-

0 • I.

3-

1960

……. __

1965 1970 1975 1980 1985

School year beginning

Number of teachers

————– 7 ————- Pupil/teacher ratio —–

– 35

• 30 o

– 25 2

20

-15 w

– 10 7f5.

-5

0
1990 1995 2002

SOURCE: U.S. Department of Education. Digest of Education Statistics, 2004.

497

496

9,000

8,000

1970 40.7
I 1980 96.0

1960 15.6

Figure 19. 1

Current expendi-

ture per pupil in

average daily

attendance in

public elementary
and secondary

schools:

1970-1971 to

2001-2002

Figure 19.2

Enrollment,

number of

teachers, and

pupil/teacher ratio

in public schools:

1960-1961 to

2002-2003

498

able • .2

Governmental Organization of Public Schoolai

Number of

Percent of Public

Independent Elementary

Districts’ Schools

Number of
Public

Secondary
Schools

ER NINETEEN ■ EDUCATION

90.0% 65,228 22,1

80

90.4 64,601 21,994
91.1 60,808 20,904
90.8 59,015 21,135
90.6 58,827 23,916
91.7 59,326 22,619
91.5 65,800 25,352
93.7 91,853 25,784

1992, 1997, 2002.

strictly comparable to earlier years; different survey coverage.

SOURCES: U.S. Department of Education, Digest of Education Statistics, various years; U.S. Bureau of the Census, Governmental Organization, various years.

population aged five and under began to rise in 1988, suggesting that the number
of students enrolled in elementary schools would continue to rise. Importantly,
total expenditures by these schools, per-pupil expenditures, and even real per-
pupil expenditures continued to increase in the 1970 to 1985 period when school
enrollment was declining.

Salaries for teachers and other workers (administrators, librarians, counselors,
maintenance persons, bus drivers) comprise the bulk of the expenditures by pub-
lic schools. Recall from Chapter 7 that employee compensation represented about
63 percent of the noncapital direct expenditure of school districts in 2002. The
number of public elementary and secondary school teachers also has increased
over the past 40 years, including the period between 1970 and 1985 when the num-
ber of students was decreasing. As a consequence, the pupil-teacher ratio also
decreased over the past 40 years, from nearly 26 students per teacher in 1960 to
about 16 in 2002, a decrease of about 40 percent (see Figure 19.2).

Public-school services in the United States are provided both by independent
school districts and by dependent school systems that are part of general-
purpose local governments such as cities, townships, or counties. The number of
school districts decreased substantially over the past 40 years and particularly
between 1960 and 1970, as shown in Table 19.2. 3 Also, substantial decreases
occurred in the number of public elementary schools until 1980. At the same
time, the number of public secondary schools decreased slightly. Since 1995, the

3
The decreases in the 1960s were a continuation of the trend operating at least since 1930. See US. Department of Education (May 1987).

499

ber of secondary schools has been increasing again, reflecting growth in the

ber of students. Thus, the picture that emerges of the provision of public

c
ation since 1960 is one of increasing spending per pupil, largely because of
ases in class sizes and consolidation of both school districts and elementary

Dols within districts.
udgetary data about education spending really do not capture the impor-
ce placed on public education and state-local government educational insti-

. 0xis . Education has been identified as an important means of altering the

co
me distribution, generating social mobility, improving economic growth,
easing the “international competitiveness” of firms in the United States, and

yert improvirig the operation of the political public-choice system in a democ-
itic society. A substantial economic literature shows that the perception of local

schools is an important factor influencing locational choices of both individuals
and firms, and through that, perception also influences property values in

“specific jurisdictions. And perhaps no local government fiscal or political issue
generates as much or as intense public interest and comment as consideration

o
f closing or consolidating local public schools or the results of educational

assessment tests.
State and local governments continue to debate and experiment with three

broad and important public policy issues regarding education: (1) How should
public education be financed, including the relative role for state as opposed to
local governments, the appropriate structure for state aid to local schools, and the
relative roles for various taxes and charges; (2) How can education be produced
most effectively and efficiently, including questions about school and class size,
teacher compensation and training, and the role of technology; and (3) How to
measure and assess educational results and enforce accountability, including
issues about the appropriate structure and uses for student testing, the use of
incentives or penalties for school systems, and the role of the federal as opposed to

state-local government.

FINANCING EDUCATION

Current Practice
Nearly half of the revenue for financing public elementary and secondary schools
in 2002 was provided by state governments, on average; local governments—the
school districts—generated a slightly smaller share from their own sources at
about 43 percent of public-school spending. The federal government has a rela-
tively minor role in financing elementary and secondary education, providing only
8.4 percent of public school spending in 2002, and even private sources of spend-
ing (for private schools) represented only about 8 percent of total school spending
in that year. This division of revenue sources for 2002-2003 is shown in Fig-
ure 19.3a. Importantly, local taxes (mostly property taxes) constitute the bulk of
local revenue for schools and amount to 37 percent of total revenue. Also, fees and
charges represent a very small fraction of revenue for public education. As shown

Year

Number of
School

District?
2001 15,014
2000 15,178
1995 15,834
1990b 15,367
1985 15,747
1980 15,912
1970 17,995
1960 40,

520

‘For the Census of Governments years: 1962, 1972, 1977, 1982, 1987,

‘Data since 1990 are not

■ APPLICATIONS AND POLICY ANALYSIS

50 1

500

00 1.01 NINETEEN ■ EDUCATION

PART V ■ APPLICATIONS AND POLICY ANALYSIS

Local

40.1
40.1
42.8
40.7
40.7
40.3
47.5
52.8

Private

7.0
7.2
7.7

10.1
8.6
7.3

1

0.5

12.3

7.3
6.8
6.1
6.6
9.8
7.2
3.7

43.2
46.4
46.6
44.4
43.4
51.8
56.8

495
46.8
47.3
48.9
46.8
40.9
39.5

………….

Local governments

……………………
_„••- …..

.State governments

1995-96 2001-02
0

1970-71 1975-76 1980-81 1985-86 1990-91

School year

Federal State

6.8 46.2

6.8 45.9
6.2 43.2
5.6 43.5
6.1 44.7
9.1 433
7.4 34.6
3.9 31.1

Table 1J.3

Public Schools
Percent Financed By

Federal State Local
73 49.7 43.1

Figure t9.4

Sources of

revenue for public

elementary and

secondary schools:

1970-1971 to
2001-2002

All Schools
Percent Financed By

Capital outlay 11.1%

(b) Expenditure

SOURCE: U.S. Census Bureau.

by the data in Table 19.3, the federal government’s role has always been relatively
small, increasing a bit from 1960 to 1980 but relatively constant

since.

The relative roles of state compared to local governments in financing educa-
tion changed dramatically in the 1970s, with the two levels of government effec-
tively switching positions, as shown in Figure 19.4. Prior to the 1970s, state
governments provided about 40 percent of school revenue, on average, and local

SOURCE: U.S. Department of Education. Digest of Education Statistics, 2004.

governments provided more than 50 percent. Responding to a number of forces,
state governments attempted to equalize educational opportunity across districts
In the 1970s, which resulted in increased state financial commitments and corre-
sponding decreases in local financial responsibility. The increased state share was
accomplished both by changing the magnitude and type of state grants to school
districts, discussed in detail later. Because the primary local revenue source for
schools (and only source in many states) is the property tax, the increased state

Figure 19.3

Distribution of

public elementary-

secondary

education revenue

and expenditure,

2002-2003

Total: $440.3 billion

State sources
49.0%

Taxes and appropriations
37.2%

Current charges 2.6%

Other local sources 3.0%

Federal sources
8.4%

(a) Revenue

Total: $453.6 billion

Other expenditure 2.9%

Instruction
52.0%

Support services
29.3%

Other 4.6%

Year

2001

2000

1995
1990

198 5
1980

1970
1960

S o
urces of Elementary and Secondary School Funding

SOURCE: U.S. Department of Education, Digest of Education Statistics, various years.

Federal government
…………….. ———————————

100

80

; 60

0

C

40

a.

20

U.S. average = 49.0 %

Median states are New Hampshire (49.0) and Georgia (48.5)

SOURCE: U.S. Census Bureau. 2003 Public Elementary-Secondary Education Finance Data.

role in financing education reduced the demand for property tax increases in
these years and, in some cases, resulted in property tax reductions. For a number
of years, the state government share of school finance was a bit larger than the
local share. After being about equal in the first half of the 1990s, the state share has
again exceeded the local share since.

Great diversity exists among states in the relative role of the state government in
financing education. In fact, the varied roles that state governments play in educa-
tion are even greater than for most other services. The distribution of states by
the state government’s share of public-school expenditures in 2002 is shown in
Table 19.4. State government provides more than 60 percent of revenue in ten states
and less than 40 percent in another nine states. At the opposite extreme, the public
schools are substantially financed by local governments in South Dakota (the state
share is 34.1 percent of revenue), Nebraska (34.5 percent), Illinois (35.6 percent),
Connecticut (36.3 percent), North Dakota (36.5 percent), and Pennsylvania
(36.7 percent). In contrast, elementary and secondary education is a state govern-
ment function in Hawaii where local school districts do not exist, and the state
generates 90.1 percent of revenue for school expenditures (the federal share is rel-
atively large in Hawaii because of the substantial U.S. military presence in the
state). Other states with a substantial state share include Arkansas (74.2 percent),
Minnesota (73.7 percent), and New Mexico (72.6 percent). The U.S. average is
49.0 percent, and the median states are Georgia (48.5 percent) and New Hampshire
(49.0 percent).

State Government Percentage of Public Elementary-Secondary

South Dakota
Nebraska

34.1
34.5

Massachusetts
Rhode Island

41.4
4

1.5

Wyoming
Oregon

50.9

513

North Carolina
West Virginia

60.3
60.9

New Mexico
Minnesota

72.6
73.7

Hawaii 90.1

Illinois 35.6 Maine 42.1 Oklahoma. 51.4 Washington 62.4 Arkansas 74.2
Connecticut 363 New Jersey 42.5 Mississippi 53.9 Michigan 63.2
North Dakota 36.5 Colorado 43.4 Wisconsin 54.8 Delaware 65.8
Pennsylvania 36.7 Ohio 44.1 Utah 55.9 Vermont 69.3
Maryland 38.2 Tennessee 44.4 Alaska 57.0
Texas 39.1 Florida 44.5 Alabama 57.1
Virginia 39.6 Arizona 44.9 Indiana 57.1

Missouri 45.4 California 58.0
Montana 46.2 Idaho 59.0
New York 46.2 Kansas 59.0
Iowa 46.8 Kentucky 59.6
Louisiana 48.2 Nevada 59.9
South Carolina 4

8.4

Georgia 48.5
New Hampshire 49.0

l
School Revenue, 2002-03

PART V ■ APPLICATIONS AND POLICY ANALYSIS pTER NINETEEN ■ EDUCATION

503

States that are considered similar in many other ways have a number of
vesting differences in the state government role in financing

education.

michigan, the state share of school revenue is 63.2 percent, but only 35.6 percent

neighboring Illinois. Texas has a low state share (at 39.1 percent), and Louisiana
d Oklahoma are about at the national average (48.2 and 51.4), however,
kansas has the second highest state share (74.2). Vermont is another state with a

e state share (69.3), whereas Connecticut has one of the lowest state shares

.3). Major changes have occurred over time in the role of state governments in
ninny of these states as new financing systems were put in place. For instance, in

1992, Michigan had among the lowest state shares (at 26.6 percent), but had the
seventh highest in 2002 (at 63.2 percent). Similarly, New Hampshire had the low-

est state share at only 8.5 percent in 1992; by 2002, New Hampshire was at the
national average. The obvious conclusion is that there is no one or even typical
way that states finance elementary and secondary education. As you will discover
in this chapter, the economic, political, and social factors that underlie these finan-

cial differences extend as well to the states’ role in regulating education.
Just as differences exist among states in how elementary and secondary educa-

tion is financed, substantial differences also exist among the states in the level of
educational spending, as demonstrated in Figure 19.5. Per-pupil spending on cur-
rent services by all public schools in aggregate was $8,019 in the 2002-2003 school
year, but per-pupil spending averaged less than $6,500 in ten states and more than
$9,500 in seven states and the District of Columbia. At the extremes, per-pupil
spending was $4,840 in Utah but $12,202 in New Jersey. The coefficient of varia-
tion, a comparative measure of variation in distributions equal to the standard
deviation divided by the mean, was .22 for 2002-2003, meaning that among the
states, there was an average of about 22 percent variation in per-pupil spending
around the mean. There has been little long run change in the degree of difference
among states in the level of education spending over the past 40 years, as the inter-
state coefficient of variation for per-pupil spending was .25 in 1992 and 1980, .21 in
1970, and .22 in 1960.

The differences in per-pupil spending among different school districts within
states appear to be about as large as the differences among states. Wayne
Riddle and Liane White (1994, p. 358-62) report, for example, that the ratio of
per-pupil expenditures for districts at the 95th percentile to those at the 5th
percentile had a median value of about 1.5 in 1990 for those states with local
school districts and varied from 3.1 to 1.3. (The ratio is 1 in Hawaii, which has
a state school system.) Similarly, Linda Hertent et al. (1994) report the coeffi-
cient of variation for per-pupil revenues among districts within states varied
from .07 (West Virginia) to .35 (Montana) in 1990, with a median of about .175.
Seventeen states had coefficients exceeding .20. Recall from Chapter 7 that dif-
ferences in expenditures can result from differences in input prices and envi-
ronmental conditions as well as from differences in demand, so that these
differences in per-pupil spending may not correspond to equivalent differences
in educational results.

Eighty-six percent of spending by public elementary and secondary schools in
2002-2003 was for current services to students, as shown earlier in Figure 19.3b.

Figure 19.5

Current spending

per pupil in public

elementary and

secondary schools,

2002-2003

0 0

0
0

0
0

0
0
0

0
0 0

0

0 0 0
0 0 0
‘1,

0 0
0

01
0
0 0

0

0 0
os

00
o a in
LA . se rsi rsi

0 0
0 0

0
0
0

N.

0

us

0
0
In
ct■

PART V ■ APPLICATIONS AND POLICY ANALYSIS copTE• NINETEEN ■ EDUCATION

505

NJ _
NY
CT

MA –
AK
DE
PA _11
RI

MD
ME –
MI

OH
IL

NH

HI
MN

United States
IN

NE
GA —
CA
IA

OR

CO
K5

MO –
ND
WA
TX
SC

NM
LA
KY
NC
SD
FL

– AR
AL
TN
OK
NV
ID

MS
AZ

I
12,202

12,140

.

•

13,328

Types oil State Aid
Unless state governments want to operate the public-school system directly (as in
Hawaii), states have to rely on intergovernmental grants to assist local govern-
ments in financing public education, and those grants must be one of the two gen-
eral forms-lump-sum or matching-as described in Chapter 9.

foundation Aid

Lump-sum school grants are usually referred to as foundation aid because the per-
pupil grant represents a minimum expenditure level; the state aid is thought of as pro-
viding a basic foundation on top of which local revenue supplements may be
added. Prior to the 1970s and then starting again in the 1990s, states generally used

lump-sum per-pupil grants to support local education. Those grants were some-
times equal per-pupil amounts provided to all school districts, but more commonly
the amount of the per-pupil grant for each district was directly related to educa-
tional costs in the district or inversely related to some measure of district wealth.
still, the grant is lump-sum because the size of the grant (per pupil) is independent
of the district’s choice about the level of spending (and thus taxes).

In general, a foundation aid program requires a basic grant per pupil and per-
haps a way of reducing the grant for richer districts. A generic formula for a foun-
dation aid grant is

G i = F[1 + Ci] – [RI[V M ]

where

Gi = per-pupil grant to district i

F = basic per-pupil grant or foundation level

C, = cost index for district i

R’ = basic property tax rate set in the formula

Vi = per-pupil property tax base in district i

Suppose, for instance, that a state establishes such a program with F = $5,000
and R’ = $10 per $1,000 of taxable property value (assuming all C, = 1 for a
moment). The largest (per-pupil) grant any district could receive is $5,000, but only
if V, is zero. Compare two school districts, one with per-pupil property value of
$50,000 and the other $100,000. The first would receive a per-pupil grant of $4,500
[$5,000 – ($10 X 50)] and the second $4,000 [$5,000 – ($10 X 100)]. Because the
only district-specific factor in the formula is the property tax base per pupil, which
is outside the direct control of the district, these are lump-sum grants. If both dis-
tricts had identical property tax rates equal to the basic rate in the formula ($10),
both would end up with $5,000 per student to spend. The first would collect $500
in property taxes per pupil and receive $4,500 in grant funds; the second would
generate $1,000 from property taxes and $4,000 from the grant program. Thus, all
districts are guaranteed $5,000 per pupil, which is the foundation amount. If dis-
tricts want to spend more than the guaranteed $5,000 per pupil, they must collect
local taxes to finance all the additional

spending.

4,860

7,047
6,8701
6,8681
1 6,647 =

‘ 6,635
6,

532

6,450
6,4081
6,395;

6,201
■ 6,127

6,084
16,034

5,816
5,672

.•

10,372
10,322

10,223
9,919

9,669
9,36
9,315

1 9,202
8,99

18,921
8,847

8,588
8,555
8,

8,285
8,218

8,100
8,073

18,019
7,948

17,832
7,743
7,724
7,691

7,534
7,460

‘ 7,449
7,316 I

‘ 7,292
7,262
7,153

‘ 7 101
7,076

SOURCE: U.S. Census Bureau.

The bulk of current spending was for direct instructional expenses, which repre-
sented 52 percent of total spending, while support services accounted for another
30 percent of total spending. In contrast, capital expenditure on such things
as buildings, technology, and transportation equipment represented only about
11 percent of total spending by public elementary and secondary schools.

PART V ■ APPLICATIONS AND POLICY ANALYSIS

In states where costs among districts are substantially different, the nominal
foundation level must be greater in districts with relatively higher costs to ensure
equal real foundation spending. For instance, if costs are 10 percent greater than
average in one district (so C, = .10), and that district has a per-pupil property value
of $100,000, the district’s per-pupil grant would be $4,500 [(1.1 x $5,000) – ($10 x
100)]. This grant, combined with $1,000 of local property tax, would provide $5,500
per student to spend. A similar-wealth district with average costs would receive
only $4,000 in grants. When combined with $1,000 of local property tax, this dis-
trict has $5,000 per student to spend. Per-student spending is 10 percent greater
in the first case because costs are 10 percent greater. In implementing such a for-
mula, one might adjust for two types of cost differences-differences in input
prices (especially for labor) and differences in environment (such as the nature of
the students who are to be educated), as discussed in Chapter 7. 4

Under what conditions would a district’s grant be $0? A district gets no grant if
its per-pupil property tax base is equal to or greater than (F /R .)/ (1 + C1 ). If a dis-
trict’s per-pupil property value is $500,000 and C, = 1 for the example, then the per-
pupil grant is $0 ($5,000 – ($10 x 500)]. The reason is simple: With a per-pupil tax
base of at least $500,000 and standard costs, the basic tax rate of $10 would gener-
ate the full foundation amount in taxes; no grant is required to bring such a district
up to the foundation level. 5

Under foundation aid programs, however, districts often may choose tax rates
greater (but often not less) than the basic rate in the formula. Again, compare two
districts with $50,000 and $100,000 property tax bases per pupil. If they both select
property tax rates of $40 per $1,000 of taxable value, the first collects $2,000 of
property taxes per pupil and receives a grant of $4,500, allowing spending equal to
$6,500 per pupil. The second collects $4,000 per pupil in property taxes and
receives a grant of $4,000, allowing spending of $8,000 per pupil. The difference in
grant amounts does not fully offset the difference in property taxes. Equal property
tax rates do not generate equal amounts of per-pupil spending if those tax rates are
greater than the basic rate in the aid formula. These two districts are not required
or expected to select equal tax rates. In fact, the wealthier district might even select
a higher tax rate.

Guaranteed Tax Ba6e Aid

An entirely different type of state aid to education, called the Guaranteed Tax Base
(GTB) or District Power Equalizing plan, is intended to provide an equal, basic
per-pupil property tax base to each district, rather than basic per-pupil minimum
expenditure level of the foundation program. Per-pupil spending may still differ
among school districts if they choose different property tax rates, but the aid pro-
gram effectively provides the same basic tax base to which the selected rate is

4 But the “costs” must not be determined solely by the recipient government. With respect to labor, for instance, the
index might be based upon average wages for all jobs in the region of the school district.

Note that districts with per-pupil tax bases greater than (F /R1/ (1 + CO will generate more than $5,000 per-pupil in
real revenue and will be able to spend more than the foundation.

AFTER NINETEEN ■ EDUCATION

applied. A GTB plan involves matching grants that reduce the price of education
to the school districts, which is the important economic difference from foundation

grants.
A GTB grant formula requires, at least, that the GTB and the allowed tax rate be

specified. The general formula for grants of this type is

G, = B + (V` – V,)R,

where

B = basic or, loundation grant

guaranteed per-pupil tax base

Vi = per-pupil tax base in district i

R i =- property tax rate in district i or maximum rate allowed for the guarantee

In a pure GTB program, B = 0, and R, is the local tax rate without any maximum.
in that case, districts receive positive grants if their per-pupil tax base (V,) is less

than the GTB (V), with the grants being positively related to the tax rate selected
by the district. Although theoretically these grants could be negative, requiring
that districts with V, > transfer funds to the state for redistribution, only in a
few cases has such recapture of funds been tried. In one variation on this program,
some states mix the foundation and GTB styles by providing a basic per-pupil
grant in addition to the guaranteed base, that is, they set B > 0. These are some-
times called two – tier programs. In that case, a district receives a per-pupil grant
exactly equal to the foundation amount if Vi = V, with that grant being reduced if

V,> V. until G is zero (negative grants again are not used). In one other variation,
the guaranteed base V applies only to some maximum, state-specified tax rate;
districts may set a higher rate, but it generates only more local tax revenue and not
additional grant funds.

To illustrate the operation of the basic GTB formula, suppose that a state pro-
gram guarantees a tax base of $200,000 per pupil (V = $200,000) and sets no max-
imum on the tax rate that is eligible for that guarantee. Districts with a per-pupil
property tax base of $200,000 or more would receive no education grants from the
state government. For districts with V, < $200,000, the grant is inversely related to per-pupil wealth. For instance, a district with a per-pupil property tax base of $50,000 and a tax rate of $40 per $1,000 of taxable value would collect $2,000 per pupil [$50,000 x ($40/$1,000)] from property taxes and receive $6,000 per pupil [$150,000 x ($40 /$1000)] from the state grant program. A district with a per-pupil tax base of $80,000 and the same tax rate would collect $3,200 per pupil [$80,000 x ($40/$1,000)] from property taxes and $4,800 per pupil [$120,000 X ($40/$1,000)] in state aid. Both receive $8,000 per pupil in total, which is the revenue generated from a base of $200,000 and a tax rate of $40. In essence, all districts are guaran- teed $200 per pupil for each $1 of property tax rate selected. Any portion of that amount that is not provided by the local property tax base is made up by a state grant.

It follows from this discussion that an increase in a district’s tax rate also leads
to a larger grant per pupil for districts with V, < V. Continuing the numerical

507

PART V ■ APPLICATIONS AND POLICY ANALYSIS
CHAPTER NINETEEN ■ EDUCATION

509

illustration, suppose that the district with a per-pupil tax base of $50,000 increases
its property tax rate to $41 per $1,000 of taxable value. That additional $1 in the tax
rate generates an additional $50 per pupil from local property taxes and $150 per
pupil from state aid; again, the net effect is an increase of $200 per pupil for each
$1 of tax rate, the guarantee amount. The local district’s share of the additional per-
pupil revenue is generally, and 0.25 in this specific example. The district
with a per-pupil value of $50,000 pays only 25 percent of the cost of increased
school expenditures per pupil; the remainder is financed by the aid program. In
contrast, the district with per-pupil value of $80,000 would pay 40 percent of the
cost of increasing per-pupil spending ($80,000/$200,000). As previously men-
tioned, one effect of a GTB aid program is to reduce the local price of providing
education. The marginal cost or price to the local district of increasing per-pupil
spending by $1 is V,/ V” if V, < and $1 otherwise.

Economic Ettecta oh Equalizing State Aid

An important economic and policy issue about different state aid programs is
their expected influence on recipient school districts to alter educational expen-
ditures. Do state education grants induce school districts to spend more on edu-
cation, and if so, by how much? This was the rationale for why many states
adopted guaranteed tax-base systems in the 1970s. Perhaps the best way to
understand the potential economic effects of different grant types is to actually
work through the responses of specific districts given some assumptions about
economic and fiscal conditions. The following educational grant simulation does
just that. Information about the demand for educational service and the initial
expenditure choices of several representative schools is first presented, and
then a new proposed state education grant program is described. The effect of
that grant program on each school district’s behavior is then analyzed, given
the demand restrictions. The simulation will be most useful if you attempt to
analyze the expected outcomes before reading the analysis in the text. Some
suggestions about how you might proceed to do that are offered after the simu-
lation is set up.

Education Grant Simulation

Suppose a state consists of four school districts, denoted A through D, each financ-
ing education solely with local property taxes. The initial fiscal situation in each of
those districts is shown in the following table, with V equaling the per-pupil tax-
able property value in each district, R equaling the property tax rate in each district
specified in dollars of tax per $1,000 of taxable value, and E equaling the per-pupil
school expenditure in each district:

A B C D

V = $100,000 V = $130,000 V = $200,000 V = $225,000
R = $55 R = $53.85 R = $45 R = $60
E = $5,500 E = $7,000 E = $9,000 E = $13,500

Thus, district A is the low-wealth, low-spending district, while D is the opposite-
high-wealth, high-spending district. Note that the product of the per-pupil value

and tax rate equals the per-pupil expenditure in each district, which is required if
local property taxes fully finance the schools.

suppose the (absolute value of the) price elasticity of demand for educational
Spending is known to be the same in each district and equal to .5, so that demand for
education is price inelastic. This value is consistent with the evidence reported in
chapter 4; if anything, it may be relatively high. Similarly, suppose that the income

elasticity of demand for education in each district is 1.0, and that the average family
income in each district is half as large as the per-pupil property value (such would
be the case if all the property is residential and consumers buy houses valued at
twice their income, so a consumer with a $50,000 income has a $100,000 house).

The state government is considering introducing a program of state education
grants to these school districts, to be determined by

Grant per Pupil = $500 + ($200,000 – V)R

where V and R correspond to the per-pupil value and tax rate in each district, and

the per-pupil grant may not be smaller than zero (no recapture). The policy ques-
tion is to analyze what the expected effect of such a grant program would be on
educational spending and property taxes in each district, and given that, what the
potential advantages might be from the state’s point of view.

At this point you should stop reading and think about how you would
do such an analysis if you were assigned this task as an economic or policy
analyst for the state. Consider the following suggestions:

i. Determine whether the grant for each separate district is matching or
lump sum. Lump-sum grants are fixed amounts that do not change
in response to a recipient government’s fiscal reactions, whereas
matching grants explicitly depend on the fiscal decisions of those
governments.

2. If the grant is lump sum, use the income elasticity to determine the
effect on per-pupil spending and the required local property tax rate.

3. If the grant is matching, determine the marginal cost or “price” to the
locality of increasing education spending and note how the grant has
changed that “price.” Use the price elasticity to compute the expected
effect on per-pupil spending and the tax rate in the district.

4. If you follow steps 1-3, you will estimate new levels of spending and

taxes in each district. Now evaluate those changes. Has education
spending increased on average? Has spending become more equal?
To what degree? Have local taxes decreased on average? What has
happened to the distribution of tax rates? Has the state received a
good return on the use of its funds? Would you recommend the
adoption of this grant program?

PART V • APPLICATIONS AND POLICY ANALYSIS TER NINETEEN ■ EDUCATION

Now let’s see how you did. Consider the districts in order of ease of the analysis.
District D receives no grant because its per-pupil value is greater than the $200,000
base guaranteed in the grant formula (D’s grant from the formula is negative, but
the smallest a grant can be is $0). Therefore, the grant program is expected to have
no effect on education spending or property taxes in district D. 6

District C receives a lump-sum grant of $500 per pupil because its per-pupil
value exactly equals the guarantee amount [G = $500 + (0)R]. Thus, district C
receives the foundation amount but no matching aid from the GTB component of
the formula. The lump-sum aid means that this district now has $500 more per
pupil in income, which can be spent to buy more education services or other
things. The per-pupil income in district C is $100,000, so the $500 grant represents
an income increase of 0.5 percent [($500/$100,000) X 100%]. With an income elas-
ticity of demand for education equal to 1, an increase in income of .5 percent caus-
es an increase in educational spending of .5 percent. Thus, per-pupil spending is
expected to increase by $45, from $9,000 to $9,045. Although the district receives a
grant of $500 per pupil, only $45 of that amount gets spent on more educational
spending. What happens to the rest of the grant? It goes for lower local property
taxes and thus more private spending by taxpayers. The new level of spending is
financed both by property taxes and the grant, so that

E’ = $500 + (V)R’

$9,045 = $500 + $200,000 X R’

R’ = $42.725 per $1,000 of taxable property value

The grant allows district C to lower its property tax rate to $42.725 from $45.00. The
district collects $8,

545

per pupil in property taxes and receives $500 per pupil in
state aid for per-pupil education expenditures of $9,045. Spending rises slightly,
but local property taxes decline by a greater amount.

District A receives both the full foundation amount of $500 per pupil and
matching aid from the GTB part of the formula because its per-pupil value is less
than the guaranteed amount. The grant to district A given the initial conditions
would be $6,000 [$500 + ($100,000)($55/$1000)J, but that grant amount will
change as district A changes its property tax rate in response to the grant itself.
First, district A receives the $500 of foundation aid, which it would continue to
receive even if its property tax rate (and spending) was $0. That $500 grant repre-
sents a 1 percent increase in per-pupil income [($500/$50,000) X 100%], which is
expected to increase per-pupil spending by 1 percent or $55 because the income
elasticity of demand for education spending is assumed to be 1. 7

6D would get a positive grant if it lowered its tax rate to less than $20 per $1,000 of value, but education spending per
pupil would fall drastically.

‘Douglas Wills has pointed out to me that there is some ambiguity about this example because the analysis of the

lump-sum component of the grant assumes a tax price of one, even though the matching component of the GTB

grant reduces the tax price. The analysis presented here is equivalent to assuming that the lump-sum grant

occurs first. This seems appropriate because if a district selects a tax rate (R) equal to zero, it still receives the
lump-sum grant.

In addition, the matching grant from the GTB formula reduces the “price” of edu-
onal spending to the residents of district A. Following the discussion of the pre-
us section, the new price is VA / $200,000, or 0.50. To increase per-pupil spending
$1, district A must collect an additional $.50 in local property taxes per pupil and

cold receive an additional $.50 per pupil in state aid. Without the grant program,
e local price was $1, so that the effect of the grant is to lower the education price in
by 50 percent. If the price elasticity of demand for education spending is .5, then

4jerpupil spending is expected to increase by 25 percent as a result of the matching
ant. Through this effect, per-pupil spending in district A would increase by

$1 ,388.75. Thus, the new level of per-pupil education spending in district A is expect-
ed to be about $6,944, an increase of about $1,444 due to the grant. Again, district A
finances that expenditure with property taxes and the grant, so that

$6,944 = ($100,000)R’ + $500 + ($100,000)R’

R’ = $32.22 per $1,000 of taxable value

District A lowers its property tax rate to $32.22 from $55.00 as a result of the grant.
The district collects $3,222 per pupil in property taxes and receives $3,722 per pupil
in state aid, allowing spending of $6,944 per pupil. Of the total education grant of
about $3,722, only about $1,444 goes for higher-education spending and the rest
into lower taxes. The grant causes a larger expenditure increase in district A than
district C because district A receives a matching grant in addition to the founda-
tion amount.

District B also receives both the full foundation amount of $500 per pupil and
matching aid from the GTB part of the formula. First, district B receives $500 of
lump-sum foundation aid, which it would continue to receive even if its property
tax rate (and spending) was zero. This component of the grant increases income
by .77 percent [($500/$65,000) X 100%] and desired spending by an additional
.77 percent, or $53.90.

In addition, district B faces a price effect from the matching component. In this
case, the price effect is smaller because the district’s per-pupil tax base is larger.
Because district B has a per-pupil tax base of $130,000, its price for additional
school spending is $0.65 [$130,000/$200,000]; district B can increase spending by
$1 by collecting an additional $.65 in property taxes and receiving, as a result, an
additional $.35 in state aid per pupil. The grant has lowered the tax price by 35 per-
cent (from $1 to $.65), which is expected to increase desired spending by 17.5 per-
cent given the price elasticity. This represents an increase of $1,234 [.25 x $7,053.90].
Thus, the new level of per-pupil education spending in district B is expected to be
about $8,288, an increase of about $1,288 due to the grant. The new property tax
rate is determined by

$8,288 = ($130,000)R’ + $500 + ($70,000)R’

R’ = $38.94 per $1,000 of taxable value

District B collects $5,062 per pupil in property taxes and receives $3,226 per pupil
in state aid to fund spending of $8,288.

AFTER NINETEEN ■ EDUCATION PART V ■ APPLICATIONS AND POLICY ANALYSIS

The expected effects of the grant program on these school districts are summa-
rized in the following table:

A B C D Average
Initial Spending $5,500 $7,000 $9,000 $13,500 $8,750
New Spending 6,944 8,288 9,045 13,500 9,444.25
Per-Pupil Grant 3,722 3,226 500 0 1,862
Initial Tax 5,500 7,000 9,000 13,500 8,750
New Tax 3,222 5,062 8,545 13,500 7,583.63
Initial Tax Rate 55.00 53.85 45.00 60.00 53.46
New Tax Rate 32.22 38.94 42.725 60.00 46.66

On the basis of this analysis, the proposed education grant program is expected
to have the following effects in the state:

■ Per-pupil education spending increases slightly by about 7.9 percent, on
average, although spending rises in only three of the districts. A little less
than 40 percent of the state grant funds go for higher spending on
education.

■ The variance in per-pupil spending among the districts in the state is
reduced only slightly. The ratio of the highest- to lowest-spending level is
reduced to 1.94 from 2.45, about a 21-percent change; however,the dollar
difference between those districts is still nearly $6,556.

■ Property taxes are reduced in all districts that receive state grants, resulting in
about a 13 percent decrease in property tax rates, on average. About 60 per-
cent of the state education grant funds go to reduce local property taxes.

■ Property tax rates are reduced more in districts with lower per-pupil
property values, so that effective tax rates now increase with property
value. The ratio of tax rate to per-pupil expenditure—which represents the
tax rate required to provide per-pupil spending of $1—is made much more
equal across the districts. Without the grants, those ratios were .01 for
district A, .0077 for district B, .005 for district C, and .0044 for district D.
Thus, a tax rate of $.01 per $1,000 of taxable value was required in order to
spend $1 per pupil in district A, but a rate of only about $.005 was required
in district C. With the grants, the required rates are $.0046 in district A, and
$.0047 in districts B and C.

It is also interesting to note what the effect would have been on district D if
recapture—that is, negative grants—were allowed. In that case, the price to local
residents per dollar of per-pupil spending would have been $1.125 ($225,000/
$200,000). Residents of district D would have had to increase local property taxes
by $1.125 per pupil to increase spending by $1 per pupil because the district would
also have to pay additional funds to the state. Thus, the price of education to resi-
dents of district D rises by 12.5 percent, which is expected to cause a 6.25 percent
decrease in per-pupil spending if the price elasticity is .5. Thus, per-pupil spending
in district D would have fallen to $12,656. Although that would have generated

more spending equality than without recapture, the interdistrict differences would
oil be large, and the increased equality would be achieved by worsening educa-
tional opportunity in one district.

po licy Implications
one of the traditional criticisms of foundation grants is that such aid programs do

not equalize resources across districts and thus are not expected to equalize spend-
ing unless the basic tax rate in the foundation aid formula is set high relative to the
actual rates employed by school districts (which requires that the foundation level
of spending also be set high) or unless district choice of R is limited. Beginning in
the late 1960s and early 1970s, this effect of the traditional foundation aid programs

used by states led to a series of court challenges in various states to the educational
systems in place. In these cases, plaintiffs argued that per-pupil spending on
local education was dependent on and generally varied by the per-pupil taxable
wealth of the school district and not exclusively on the wealth or income of the
family. Because state aid programs did not offset this dependence, those bringing
the cases argued that students were being denied equal protection under the law.
These cases were successful in a number of states, the courts finding that the state
aid systems violated the equal protection clause of the Fourteenth Amendment to
the U.S. Constitution or similar equal protection provisions in state constitutions.

The Serrano decision in California in 1971 was the most influential and often
cited. The courts ordered the states to devise state aid programs that would elimi-
nate (or at least reduce) the relationship between property wealth and per-pupil
spending in school districts. 8

As a result of these decisions and other forces encouraging states to equalize
educational opportunities, some states increased the basic grant amount in their
foundation programs, while many others switched to forms of guaranteed tax-
base aid programs because those programs provided the wealth neutrality that the
courts had demanded. That is, with GTB aid, each district is guaranteed a mini-
mum tax base, usually property value per student, so that spending need not
depend on district property wealth.

However, the results of the grant simulation that you completed previously rep-
resent accurately the actual results obtained in many states that adopted grant pro-
grams of this type. Often substantial equalization of per-pupil spending among
school districts did not occur. The economic reasons for this are clear. Because the
demand for education spending is price inelastic, the price reductions that are
caused by the matching grants do not influence consumption very much. Similarly,
given the magnitude of income effects, lump-sum grants also do not influence
education-spending levels substantially. As a result, most of these state education
grant funds went to reduce local property taxes rather than to increase educa-
tion spending. Therefore, the wide differences in educational spending between

512

‘For more detail about these court challenges and decisions, see Lukemeyer (2004) and Huang, Lukemeyer, and

Yinger (2004).

513

PART V ■ APPLICATIONS AND POLICY ANALYSIS

515
copTER NINETEEN ■ EDUCATION

districts within states were not reduced substantially. As Richard Murnane (1985,
p. 133) has noted:

. . . It seems clear that the main lesson from the first ten years of school finance
(reform] is that GTB finance plans which lower the price of education to property-
poor communities, but leave the communities free to choose between more spending
on education or lower tax rates, will not produce an equalization of per-pupil spend-
ing levels across school districts and will not result in districts spending enough to
provide their students with a strong basic academic program.

This difficulty cannot be changed by increasing the size of state aid programs if
the structure of those programs remains the same. If demand is price inelastic, a
substantial portion of the grants will go to reduce taxes regardless of how much
the price of education spending is reduced. The simulation understates the mag-
nitude of the problem in several ways. If incomes are increasing over time, then the
demand for education spending is also rising in many and perhaps all districts.
Many states provide education grants for purposes other than equalization—such
as grants for special education or transportation—and these often increase spend-
ing in all districts. As you learned in Chapter 14, state property tax credits can
induce tax and spending increases in high-tax jurisdictions. Those economic forces
may serve to widen the spending disparities, so that the modest equalizing force
from state aid may serve only to preserve the existing distribution and prevent the
increased variance that would otherwise occur.

Therefore, even though many states adopted GTB aid programs at one point,
which are theoretically wealth neutral, court challenges to state education finance
systems have continued. Huang, Lukemeyer, and Ymger (2004) report that as of
2003, only 5 states (Delaware, Hawaii, Mississippi, Nevada, and Utah) had not
faced any litigation regarding education finance. State education finance systems
have been rejected or overturned by the courts in 18 states, and systems have been
upheld in 16 states, although litigation continues at this writing in a number of
cases. Thus, Huang, Lukemeyer, and Ymger (2004, p. 329) report “Of these thirty-
four states, at least eleven have ongoing litigation in which plaintiffs are seeking
further reform or . . . presenting new evidence or legal theories. In an addi-
tional four states, state supreme courts have issued interim decisions favorable to
plaintiffs and litigation continues. Finally, suits are pending in three states. . . .”

Anna Lukemeyer (2004) reports that the basis for continuing court challenges to
state education finance systems has changed over time. Many challenges still focus
on the wide differences in educational spending per student that exist within
states, although often no longer using an “equal protection” argument, but rather
based on clauses in state constitutions about the required state role in providing or
guaranteeing adequate or appropriate education for all students. Cases taking this
approach are essentially arguing on equity grounds that education differences
should be reduced. The most recent set of court challenges focus on a different
issue, however: whether or not the education system in a state serves to provide
an “adequate” or “efficient” education to students in aggregate. In such cases,
plaintiffs are less concerned about spending differences, per se, and more about the
level of education spending, services, and outcomes in the state overall. For

instance, Lukemeyer (2004) notes the 1989 Kentucky Supreme Court decision in
ouch the court found that the constitution’s requirement of “art efficient system

of common schools throughout the state” was not being met and ordered changes

to provide each student “an equal opportunity to have an adequate education”

and then defined an “adequate education” at a quite high leve1. 9

sununarizing the actual types of aid programs used by the states is difficult
because each state typically has a number of different components, and the struc-

ture of the aid programs often includes fiscal features specific to each state. For a

conference held at Syracuse University in 2002, Yao Huang (2004) summarized

state aid systems for education for all states in 2001. Focusing only on general aid

for schools (ignoring specific categories such as transportation or capital invest-
ment), Huang reports that 30 states were using foundation aid programs, all but

two of which included some type of adjustment for cost differences. Three states
used a GTB aid system exclusively and another 11 states used a GTB system in
addition to a foundation level, to create an incentive for equalization. The remain-

ing states used other systems, including flat per-student grants in North Carolina,
Pennsylvania, and Rhode Island, and full state funding in Hawaii. Whatever sys-

tem is used, limits on spending or revenue are common- Three states effectively
permit no local supplementation of revenue beyond that in the state formula, with
another 25 states limiting local supplementation by setting maximum tax rates, by
setting limits on the per-student amount of supplemental funds that can be col-
lected, by limiting the growth of revenue or spending, or by requiring that part of
any local supplemental revenue be recaptured by reducing state aid.

Thus, most states that switched from GTB aid programs back to foundation aid
coupled the foundation level with caps on spending or revenue. The simplest
foundation plan, of course, is one that sets equal spending in all districts. Such a
plan could set targeted per-pupil spending in each district at F and pay state grants

to each district equal to the difference between F and the local property tax col-
lected at some mandated level. Other options include foundation amounts (F) that
vary with district costs, again with some maximum allowed expenditure or limit
on local supplements to the foundation. These spending caps essentially are nec-
essary to prevent growing spending differences among districts (or to bring about
additional equalization) if the foundation is to be below the highest district spend-

ing levels.
What are the options for state policymakers who want to equalize education

opportunities or spending among school systems in their state or who want to
increase the level of spending throughout the state? In general, there are three
approaches. First, a state government can assume the responsibility for directly
Providing elementary and secondary education, effectively having a single state
school district, as in Hawaii. This would certainly involve the most dramatic and
traumatic change to the fiscal system among the alternatives. There are at least
two economic reasons why this alternative may not be desirable. If cost differ-
ences exist among different school districts, then equal per-pupil expenditures

9See Lukemeyer (2004) and Flanagan and Murray (2004).

PART V ■ APPLICATIONS AND POLICY ANALYSIS APTER NINETEEN ■ EDUCATION

5 1 7

may not generate equal educational service. And politically, it would likely be
very difficult not to have equal per-pupil spending in all areas with a state
system. The advantage of local districts is that such cost differences as well as
differences in individual desires about emphasis in education can be recognized
and acted on.

The second option is for states to mandate a minimum amount of per-pupil
spending through their aid programs and to set that minimum relatively high
compared to actual spending levels in that state. The second prescription is crucial
because unless the minimum applies to a number of school districts, little equal-
ization will occur. States can do this using either a foundation or GTB program.
With foundation aid, the state can require that districts at least levy the specified
tax rate in the formula, with both that rate and the foundation amount set rela-
tively high. For instance, if the foundation amount is set at $5,000 per pupil and the
required tax rate is $50 per $1,000 of taxable value, districts with per-pupil values
less than $100,000 per pupil ($5,000 / ($50/$1,000)) would receive foundation
grants. But the minimum any district could spend is $5,000 per pupil. With GTB
aid, this result can similarly be accomplished by setting a relatively high minimum
required tax rate. Returning to the simulation, if the minimum were set equal to the
average rate of about $47 that prevailed after the grants were received, districts A,
B, and C would have to increase their tax rates and per-pupil spending. By requir-
ing a number of local districts to increase spending up to the minimum amount,
the state government is restricting local choice but to a lesser extent than results
from direct state provision of education.

This second option, to narrow school spending differences by raising the mini-
mum allowed spending or tax rate, often is accompanied by limits on maximum
allowed spending (or maximum allowed growth of spending) for high-spending
districts. Such spending limits are intended to prevent or reduce spending in-
creases that would occur in these districts (due to income growth or other factors)
to assist in narrowing the differences. Such spending limits have at least three dif-
ficulties. By preventing some districts from raising local taxes to support additional
desired education service, states may reduce support for the education finance
system overall. In addition, such spending caps may reduce the overall level of
spending on education. And finally, these limits might induce residents of the
limited districts simply to purchase more education service in a different way—
from the private market or through school-parent associations or foundations, for
instance. Of course, this last difficulty is the ultimate reason why it is impossible to
cap spending by higher-income families; the state may limit school spending, but
not spending on education.

Evidence reported by William Evans, Sheila Murray, and Robert Schwab (1999)
and by Caroline Hoxby (2001) suggests that states have in fact pursued this second
option, so that court-ordered changes in state systems to finance education did
lead to equalization of education resources among districts. Evans, Murray,
and Schwab examined education provision in 46 states during the period from
1972 to 1992; 11 of those states experienced court-ordered school finance reform in
those years. They report that those education finance reforms, all of which
involved increases in the state government role in financing schools, reduced

spending differences between districts in those states substantially—on average
about 20 to 30 percent. Evans and colleagues also report that for these cases, the

uced differences between districts occurred as a result of the lowest-spending dis-

tricts increasing spending substantially—what is called leveling up. Hoxby modeled
the state education finance systems in every state in 1990 and then related the char-
acteristics of each state’s financing system to actual education spending in the state.
She reports that spending is increased by high foundation levels and by GTB pro-
grams that reduce tax prices substantially for low-wealth, low-spending districts.

But Hoxby also finds that a substantial amount of the equalization of spending

among districts arises by limiting or restricting spending by the highest-spending
districts—what is called leveling down.

A final alternative is for states to mandate minimum educational conditions but
not minimum spending levels in local school systems. For instance, a state might
se minimum standards all teachers must satisfy, or a state might establish mini-
mum course requirements that students must satisfy to graduate. If those mini-

mum standards are set relatively high compared to the actual performance of
many districts in the state, then those local districts will be required to adjust the
educational service provided, which might require increased per-pupil expendi-
tures in some districts. The difficulty with this alternative, as we will examine next,
is discovering just what conditions matter for educational results and thus how to
set the minimum standards.

STATE ATTEMPTS TO REFORM EDUCATION FINANCE:

THE CASES OF CALIFORNIA AND MICHIGAN

Many states have continued to wrestle with changes occur almost continuously. The

the fundamental policy problem of providing experiences in California and Michigan are

for an equitable and efficient level of educa- particularly illuminating in showing both

tion to all children in the state, while recog- the forces that have operated over the past

nizing the role for local school districts and 30 years and those that are likely to continue

differences between districts in educational into the future.

costs and demands. In most states, this has

California was among the first states in

been a continual process involving interac- recent years to have the courts find that the

tion between state government, the courts, state system of financing and providing edu-

and the local districts. Occasionally, states cation was unconstitutional and order spe-

make radical or dramatic changes in the edu- cific changes in that system. 1° In a series of

cational system, but smaller marginal legal decisions between 1969 and 1976

“‘This section draws heavily from Thomas A. Downes, “Evaluating the Impact of School Finance Reform on the
Provision of Public Education: The California Case,” National Tax Journal, 45, (December 1992): 405-19.

Application 19

51 8

PART V ■ APPLICATIONS AND POLICY ANALYSIS ER NINETEEN ■ EDUCATION

Application 19.1 — State Attempts to Reform Education Finance

Application 19.1 — State Attempts to Reform Education Finance

incentives of the GTB plan (demand was very

price inelastic) by residents of low-spending

districts. In addition, state equalizing aid did

not increase sufficiently to fund local desired

spending on education, so local property

taxes provided an increasing share of local

school revenue. In 1978, local property taxes

provided about half of school revenue; by

1994, this share had increased to about

66 percent. Property tax burden in Michigan

relative to income was seventh highest in the

nation.

As a consequence of high property taxes

and growing spending disparities among dis-

tricts, Michigan changed its system entirely

again in 1994. The new system is based on a

foundation guarantee for each district, which

is the allowed per-student spending, deter-

mined by spending in 1993-1994, plus

allowed annual increases. Districts above the

state’s basic foundation (initially $5,000 per

student in 1994-1995) receive annual lump-

sum per-student increases equal to the per-

centage growth of state school aid revenue

multiplied by the basic foundation. Districts

spending less than the basic foundation

receive up to double those annual per-

student amounts. To finance the districts

foundation guarantee, each district receives a

lump-sum per-student grant from the state

equal to the difference between that district’s

guarantee and an 18 mill local tax on non-

homestead taxable property. Districts spend-

ing more than $6,500 per student in 1994-

1995 (the highest 6 or 7 percent) also levied

an additional local property tax on home-

steads only to fund the differences between

$6,500 and the district’s guarantee. Over time,

the foundation amount grows, so that the

minimum level of per-student spending also

(the Serrano v. Priest cases), state courts essen-

tially found that “any education financing

scheme that allowed for a positive correlation

between a district’s taxable wealth [property

tax base] and per-pupil expenditures would

be unconstitutional” (Downes, 1992, p. 406).

In response to these decisions, the state

adopted a financing system providing a foun-

dation level of spending to districts and limits

on revenue per student excluding categorical

aid and local property taxes. These changes

alone were not sufficient to bring the state’s

school finance system into compliance with

the court mandate because local districts still

had the option to collect local taxes to exceed

the spending limits.

But in 1976, California voters adopted a

major tax limitation proposal (Proposition 13)

that, among many things, imposed tight lim-

its on both the level and growth of local prop-

erty taxes. Compliance with the new limits

required large reductions in property taxes

and effectively prevented schools (and other

localities) from replacing those property tax

revenues in the future. As a result, the local

government share of school revenue fell in

1978-1979 (the first year Proposition 13 was in

effect) to about 30 percent from about 54 per-

cent in prior years. Because the state govern-

ment replaced much of the lost property tax

revenue for schools (from state surpluses ini-

tially and general state taxes later), the state’s

share of education revenue rose from less

than 40 percent to about 60 percent in 1978-

1979, where it has essentially remained

since.

These changes had a number of important

effects in California. First, the property tax

limit meant that the state and revenue limit

for school districts, exclusive of categorical

aid, was tight and became a force for equaliz-

ing spending differences between districts.

Downes (1992) reports substantial reductions

in the differences between districts in both

the revenue limit per student and total

expenditures per student. Second, the state

government became the dominant level

for financing education, especially for any

growth of spending. As a result, the relative

level of school spending in California

declined. Fabio Silva and Jon Sonstelie (1993)

report per-pupil spending in California went

from 13 percent above the U.S. average in

1970 to about 10 percent below average in

1990. Per-student spending remained 5 per-

cent below the national average in 2002-

2003. Third, because districts have been pre-

vented from using local taxes to increase

school spending to desired levels, many dis-

tricts have instead turned to increased fees,

parent contributions to schools or fundrais-

ing by school associations or foundations,

and generation of school revenue in other

ways, such as renting out school facilities.

A disappointing aspect of the changes in

California is that even though per-student

spending differences between districts have

been reduced substantially, there has not

been a corresponding equalization of stu-

dent performance, at least as measured by

test scores. This failure to affect performance

as much as spending seems to result from

three factors: (1) wealthier districts used non-

tax sources to maintain spending, (2) low-

wealth districts used the increased resources

to lower dropout rates (which may actually

cause test scores to fall, if the retained stu-

dents are worse-than-average academically,

(3) costs of educating students increased rel-

atively in the low-wealth districts, partly due

o
demographic changes in the student

po pulation.
Michigan also made fundamental changes

to its school finance system in the early

19705, but because the subsequent develop-

ments in Michigan were different than in

California, a completely new and radically dif-

ferent financing system was put in place in

1 994. Beginning with 1974, Michigan had

changed its state aid program for schools from

a foundation program to a power-equalizing/

GTB plan, which then was continued with

only minor modification until 1993. Under the

state’s GTB aid plan, the aid formula parame-

ters were altered each year so that between

50 and 65 percent of the local school districts

received aid and had a marginal reduction in

tax prices. Districts generated local revenue

from property taxes, which were limited only

slightly, and there were no limits on school

spending.

The results of the new (1974) financing sys-

tem in Michigan were disappointing on at

least two fronts. Differences in spending

among districts were not reduced (although

differences in local taxes per pupil were

reduced); in fact, spending differences

increased over time. Prior to 1974, the coeffi-

cient of variation for operating expenditures

per pupil among Michigan districts was about

0.16; by 1980, it was about 0.17 and by 1994,

it had increased to 0.23. Spending differences

increased rather than decreased due to con-

tinued use of state categorical aid (which was

not equalizing), state property tax credits that

applied to wealthy as well as poor localities,

uaeunse

local tax increases adopted by voters in dis-

tricts who wanted to increase spending

(because

and lack of

or other personal

influences), response to the price

PART V ■ APPLICATIONS AND POLICY ANALYSIS
520
Application 19.1 — State Attempts to Reform Education Finance

rises. For 2003, the bask foundation was

$6,700. The maximum foundation grant was

$8,000, and only 45 school districts levied

local taxes to spend more than that amount.

Michigan’s new financing system has had

four primary long-run effects, as noted by

Julie Cullen and Susanna Loeb (2004). First,

the state government generates more than

75 percent of funding for schools, more than

double its share before the change. Second,

as the state funding comes mostly from state

sales taxes, a state property tax, and the state

income tax, the importance of property taxes

(and especially local property taxes) was sub-

stantially reduced. Third, the level of educa-

tional spending has increased substantially,

by more than 9 percent in real terms between

1990 and 1998. Average real revenue per stu-

dent grew from about $5,700 in 1991 to more

than $7,200 in 2000. Finally, relative spending

differences between districts have been

reduced as low-spending districts were raised

to the basic foundation level, which is indexed

annually, and as the growth of spending in

high-spending districts was limited. The coef-

ficient of variation for per-student revenue

fell from 0.22 in 1991 to 0.13 in 2000.

Even with these changes, a number of

concerns and controversies remain. Although

spending differences have narrowed (as in

California), it is not clear that corresponding

educational outcomes have equalized as

well (California did not see an equalization,

for instance). Because the state sets allowed

spending in each district, some high-

spending districts have not been allowed to

raise local taxes to spend as much as is

demanded. These high-spending districts

seem unwilling to accept the restriction on

new education spending and have worked

to eliminate the restrictions or seek new

nontax methods to fund additional services

(as also happened in California). Some of the

previously low-spending districts that

received substantial amounts of new funds

are essentially forced to spend more than

what the voters selected. Some of these dis-

tricts may now use the general spending

funds for capital projects rather than collect-

ing separate revenue for that purpose.

Because the grants are in per-student terms,

decreases in enrollment cause proportionate

decreases in state funding, but costs may not

fall proportionately. Finally, with the state

government providing nearly 80 percent of

revenue for schools, K-12 education must

compete with other state services for the

available state revenues.

PRODUCING EDUCATION

The Paradox of Declining Performance

We have learned that per-pupil spending in real terms by public schools
increased nearly continuously over the past 40 years, in part because average class
sizes declined. The paradox, however, is that student performance, at least as
measured by a variety of average test scores, has not increased proportionately.

AFT ER NINETEEN ■ EDUCATION

521

anges in the scores on the Scholastic Aptitude Test (SAT)—a test purporting
measure preparation for college given to high school seniors and with

Which many of the readers of this book are intimately familiar—have been given
riment attention. The now well-known story is that those average scores, for

Pheth
eg

verbal and mathematics skills, declined from 1963 t 1980. Over that period,

the average SAT verbal score declined by more than 11 percent from 478 to

424, and the average math score declined by more than 7 percent from 502 to 466
(possible SAT scores range from 200 to 800 on each component of the test). Since
too, average SAT verbal scores have remained about the same, whereas average
math scores have increased modestly. Similarly, American College Testing Pro-
gram (ACT) average scores also declined from 1966 to 1975 and have remained
about constant since 1995.

Although not as widely reported, the SAT score changes also were being reflected

b y changes in scores of other standardized tests given to students at various
grade levels over that period. For instance, Hanushek (1986) noted that scores on
the Iowa Tests (standardized tests used in many states and given to students in

grades 5, 8, and 12) also declined beginning in the mid-1960s through the 1970s.
Interestingly, Hanushek also noted that the timing of improvements in those test
scores and others he discusses are consistent: Fifth-grade scores started to rise in
1975, eighth-grade scores in 1977, and twelfth-grade scores in 1980. These average
scores do mask some differences by subject matter. Murnane (1985) discussed a set
of tests sponsored by the national government called the National Assessment of
Educational Progress (NAEP) given to students aged 9, 13, and 17 in various years.
Those results show that reading skills have remained essentially constant since
1971 for 13- and 17- year olds and improved for 9-year-olds. Over the same period,
students’ mathematics skills remained stable for 17-year-olds but have improved
substantially for 9- and 15-year-olds. More recently, attention has focused on a set
of student assessments used internationally. By those measures, achievement by
students in the United States remains below that of students in many other nations
that have lower educational spending (see the “International Comparison” section
later in this chapter).

What are the possible explanations for why student-achievement test scores did
not keep pace with increases in public school spending relative to both enrollment
and inflation? Part of the explanation for the change in college-entrance test scores
lies in changes in the number and mix of students who were taking the test and
going on to college, which was important in the 1960s but not the 1970s. Some of
the explanations offered for the broader trend include shortages of qualified teach-
ers, especially in mathematics and science, the nature of teacher-training programs
emphasizing education over academic classes, social factors that altered interest or
participation in education, and changes in the characteristics of schools and public-
school programs themselves, such as the introduction of broader, less academic
curricula or new teaching methods. But the evidence is inconclusive or even neg-
ative on some of these factors.

The real task in resolving the paradox is discovering in a general sense just
what inputs into the education process affect educational outcomes and by what

ER NINETEEN ■ EDUCATION

inlum-profit restriction, but even in those cases, the objective is clear and easily

arttifiable.
With respect to government services and education particularly, the govern-

ment’s objective is not so easily defined. Even if an objective can be agreed on,

the measures of output and thus success in meeting the objective are imprecise.

The output or result of education is usually measured in one of four ways: by
scores on standardized tests, by numbers of students achieving a particular

level of education (number graduating from high school and number entering
college, for example), by economic achievements such as rate of employment or

level
of income, or by subjective measures (often through surveys) of individ-

ual satisfaction. Among the numerous studies attempting to relate education

inputs and methods to educational results, test scores are easily the most com-
monly used measure of performance or output, partly because they are readily

available for many students and because they make comparisons over time
relatively easy.

Analyses relating economic achievements to education level certainly suggest,
at least on the surface, that more education leads to economic gains. For instan

ce,

the basic data shown in Figure 19.6 indicate that unemployment rates are lower
and incomes higher among those who have completed more years of school. These
correlations have two qualifications, however. First, some have argued that rather
than producing education, the primary effect of the school system is to serve as a
screening device, identifying more able individuals by the fact that they are allowed
to pursue more education. By this viewpoint, the role of schools is to select the
more able and provide that information to the market. If that is the case, those with
more education do better economically because they are more able, not because
additional years of school made them more skilled.

Second, these correlations do not distinguish the quantity of education from

the quality of result. Measures of numbers of students graduating on time, the
percentage entering college, the number of school years completed, or the
number who are employed x years after graduating are predominately quantity
measures, which do not distinguish very well the quality of education. After
all, there are different types of colleges, and the fact that someone is emplo-
yed does not indicate the type of job or level of satisfaction. This is, of course,
another reason for the attractiveness of test scores that can be interpreted as
reflecting an entire range of outcomes. Whether test scores do, in fact, reflect
educational quality is controversial and problematic. The evidence shows, for
instance, that test scores do not necessarily correlate with later economic success
by students.

Even if a measure (or several measures) of educational output from this list can
be agreed on, however, it is not clear what the objective of the school system is or
should be. This difficulty arises because there is typically a wide range of students
in any school system, so that one might be interested in the distribution of results
among those students as well as the average result. This point has been empha-
sized by Byron Brown and Daniel Saks (1975) who suggest that schools might be
interested in both the mean and variance of test scores, for instance. Suppose that

PART V ■ APPLICATIONS AND POLICY ANALYSIS

magnitude. With that information, it may be possible both to understand what
happened in the 1960s and early 1970s and to improve the provision of education
in all types of schools.

A Production Function Approach to Education

A production function characterizes the relationship between inputs and the range
of possible outputs that can be produced with each input combination (as dis-
cussed in Chapter 7). If that technology of producing “education” can be identified
and quantified—that is, if the effect of different educational inputs on educational
results can be determined—then a mechanism is revealed to evaluate how dif-
ferent schools go about educating and why educational results differ for different
students or at different times. The concept of education production analysis by
economists, then, is to statistically relate education outputs to education inputs.
Mathematically,

Q = 12, 13, .

where

Q = the educational outcome

I = educational inputs

Although it seems natural to economists to examine the production of education
in the same way that one might study the production of automobiles, computers,
or agricultural products, this approach when applied to education remains some-
what controversial.

Measuring Outcomes

The necessary first step in analyzing and evaluating production decisions is iden-
tifying both the objective of the organization and some way of measuring output. As
discussed in Chapter 7, neither of these decisions is straightforward in the case of
many services provided by governments, including, and perhaps especially for,
education. Moreover, the appropriate way to measure output depends on the gov-
ernment’s objective in providing the service. For instance, a discovery that schools
do not do a good job of improving students’ scores on standardized tests may not
be surprising or very useful if, in fact, schools do not care about test scores and
thus do not try to improve them.

In doing production analysis for private firms, particularly those in manufac-
turing, these decisions seem clearer. Economists typically assume that the objective
of the firms is to produce the amount of product that generates the highest possi-
ble profit. Output can either be measured by the number of physical units pro-
duced or by the dollar volume of sales. If profit rises, then the firm is moving in
the direction of achieving its goal. Production changes that increase profits are
deemed desirable. Economists also sometimes consider objectives other than max-
imizing profit, such as increasing market share or maximizing sales subject to a

522

10
8.4

5 2 4

Figure 19. 6

The relationship

between educa-

tion level attained

and economic

status, 2001-2002

P
er

ce
n
t

u
n
e
m

p
lo

y
e

d

0

All education levels 4.6

Some high
school,

no diploma

High school
completed

■ Men
Women

Master’s
degree

Bachelor’s
degree

Associate
degree

PART V ■ APPLICATIONS AND POLICY ANALYSIS

Less than high High school
school

completed,
completion

no college

Highest level of education

(a) Unemployment rates of persons 25-years-old and over by highest degree attained.

570,000

60,000

50,000

40,000

0

30,000

20,000

10,000

0
Highest level of education

(b) Median annual earnings of workers 25-years-old and over by years of schooling completed

and sex.

A
pTER NINETEEN ■ EDUCATION

525

Sample Alternative Teat Score Distributions

Student Case A Case B A – B Percent Change

1 700 600 -100 -14.3%

2 650 570 -80 -12.3

3 600 550 -50 -8.3
550 520 -30 –

5.5

5 500 490 -10 –

2.0

6 450 450 0 0.0

7 400 410 +10 +

2.5

8 350 380 +30 +8.6

9 300 350 +50 +16.7

10 250 320 +70 +28.0

Average 475 464 – 11 (Loss) -2.3

Standard Deviation 143.6 92.1 -51.5 (Gain) -35.9

Std. DevJAve. 30.2 19.8 -10.4 -34.4

the two alternative sets of test scores shown in Table 19.5 are both possible out-
comes that arise from different allocations of the teacher’s time and other
resources, for a school or class. The average test score (or equivalently, the sum of
scores) is maximized in case A by applying more of the educational resources to
the better students. Although the resulting average score is high, the variation
among the students is also very large; the coefficient of variation is 30.2, meaning
an average of 30.2-percent variation in scores around the mean score. Case B rep-
resents the results of an alternative application of the same educational resources,
perhaps applying those resources more evenly among the students. The result is a
2-percent lower average score but much less variation among the students (about
20 percent around the mean). In essence, what has happened is that the top scores
have fallen by more than the bottom scores have risen, but the percentage gains by
the students at the bottom of the distribution outweigh the percentage decreases
by those at the top.

Which distribution is better? Which do you prefer? There may be no clear
answer. One often hears about equal opportunity in education or society, and an
explicit economic objective of government is to alter the distribution of income or
resources in society. If that is the case, then individuals and government may be
willing to accept lower average test scores or educational outcomes in exchange for
a more even distribution of those outcomes. This issue implies one of the difficul-
ties in evaluating teachers or schools. If teachers are evaluated or paid, or are if
districts are rewarded with state aid, based on the average score of their students
On some standardized test, then there is an incentive to maximize those average
sc ores by allocating teaching time or resources to those students whose test scores
improve the most; however, the resulting distribution of student performance may
not be that which is most desired.

Some college,
no degree

Associate
degree

Bachelor’s or
higher degree

PART V ■ APPLICATIONS AND POLICY ANALYSIS

C HAPTER NINETEEN ■ EDUCATION

527 5z6

Measuring Inputs

The second requirement for analyzing educational production is to identify and
measure the inputs into the production process, which are those factors that are
expected to influence educational results. In general, three types of inputs are iden-
tified: those provided by the schools, those provided by society (broadly defined),
and those provided by the student. The following equation reflects these inputs:

Q = q(School Inputs, Social Inputs, Student Inputs)

Examples oil each type ob input are listed here:

School Inputs Social Inputs

Student Inputs

Teachers

Family Experiences

Innate Ability
Books

Cultural Factors

Effort
Computers

Nonschool Learning
Classroom Hours

Books at Home
Curricula
Other Students

At least three important issues must be resolved before this general model can be
applied. First, one factor that differentiates the production of education from the pro-
duction of many other commodities is that the inputs are expected to have a cumula-
tive effect. The educational achievement of a student at a particular grade or age is
expected to depend on all the previous education inputs applied to that person, not
just on the most recent or those from a particular grade. In other words, for a statisti-
cal analysis based on test scores, one should not relate the score at a particular grade
to the inputs provided by that year’s class, but rather to all past education received
by that student. This is another difficulty in using test scores or achievement results
to evaluate teachers or school systems because a student’s achievement at one time
may depend on the work of past teachers or other schools. This is another reason why
focusing on the change in achievement in a particular period may be more useful.

Second, the school inputs either can be measured by the actual numbers of
inputs used (number of teachers per student, number or percentage of teachers
with a Master’s degree, number or percentage of teachers with more than five
years’ experience, number of school days or hours per year, types of subjects
taught) or by the amount of money spent by the school on those inputs (instruc-
tional expenditures per student). However, it may be that additional spending will
improve educational outcomes only if those resources are applied in particular
ways. Finally, it must be decided whether the unit of analysis is to be the class-
room, thus focusing on specific teachers, or the school or school system.

Evidence on Educational Production: What Matters?
Hanushek (1986) has identified about 150 different studies—prepared over the past
20 years using the basic approach outlined previously—of the factors influenc-
ing educational production. Although these studies use different data sources and

different theoretical and statistical models, some relationships among inputs and
results have been noted consistently, although other hypotheses about relation-
ships have consistently not been supported by the research. Accordingly, a con-
sensus has developed about what factors appear to be important in improving
educational results.

First is a surprising result about some factors that apparently have not been
associated with improved educational outcomes. As stated by Hanushek (1986,
p.1162), “There appears to be no strong or systematic relationship between school
expenditures and student performance.” As stated previously, the instructional
expenditures of schools are largely composed of the costs of teachers. So higher
per-pupil expenditures most likely arise from smaller class sizes, paying all teach-
ers higher salaries, or hiring teachers with more education (which requires higher
salaries). The absence of a relationship between per-pupil expenditures and stu-
dent performance is also found when expenditures are decomposed into these
characteristics. So there also appears to be no strong or systematic relationship
between student performance and smaller class sizes, teachers with more graduate
education, or higher teacher salaries generally.

That per-pupil expenditures per se do not appear to matter for student perfor-
mance is certainly surprising, at least to economists, because it implies that addi-
tional inputs do not lead to additional output. It is important to note, however,
that although the result suggests that increased per-pupil expenditures have not
led to improved performance, increased spending still might lead to improved
performance if those additional resources were spent differently, that is, on dif-
ferent inputs that do affect performance. For instance, smaller classes might
improve performance if the time in those classes was used differently than it is in
larger ones, whereas the finding that graduate education of teachers does not
improve performance may say more about the current nature of graduate educa-
tion than it does about the value of more training generally. Therefore, what these
studies suggest about how to improve educational performance is particularly
important.

Second, the “skill” of the teacher is one factor that apparently is related to stu-
dent performance. As Murnane has noted (quoted in Brown and Saks 1981,
p. 222), “Virtually every study of school effectiveness finds that some attributes of
teachers are significantly related to student achievement. . . . In particular, the
intellectual skills of a teacher as measured by a verbal ability test or the quality of
college the teacher attended tend to be significant.” A similar theme is cited by
Hanushek (1986, p. 1164) who writes that “The closest thing to a consistent find-
ing among the studies is that ‘smarter’ teachers, ones who perform well on verbal
ability tests, do better in the classroom. . . .” The practical difficulty with this find-
ing is that it may not always be easy to identify ahead of time “more skilled” or
“smarter” people and then to induce more of those people into teaching. In fact,
it may be that there are several ways for individuals to be successful teachers, so
identifying a single characteristic as indicative of whether someone will be a
“good” teacher is not feasible.

The third general conclusion of these studies is that the school curriculum can
be related to student performance, at least on standardized tests. As noted by

PART V ■ APPLICATIONS AND POLICY ANALYSIS pTER NINETEEN ■ EDUCATION

529

52 8

c lot

Mumane (1985, p. 120), “The best documented schooling change contributing to
the [SAT] score decline is a reduction in the number of academic courses students
take…. Subsequent research supports the link between the number of academic
courses students take and their scores on standardized tests.” By “academic
courses,” this finding refers to the basics—reading and writing, mathematics,
science, social studies—as opposed to vocational and other courses students can
select (the arts, sports, and so on). This finding should not be surprising because
these academic skills are primarily tested by standardized tests. Nonetheless, it is
comforting that the statistical studies come to such a common sense conclusion: If
one wants students to read and write well and do mathematics, then those are the
courses students must take and the skills they must practice in school.

Policy Implications
In large measure, these results have spurred many of the actual and proposed
changes in state education policies in recent years. Most of these changes and pro-
posals focus on teachers and courses. Regarding teachers, the policy issues concern
how teachers are trained, certified and evaluated, and paid. A number of colleges
and universities have now agreed that students working to become teachers will
take fewer education classes and more classes in the specific disciplines they plan
to teach. Thus, for example, someone who plans to be a high school math teacher
might major in mathematics in college and take some specialized education classes
in addition (rather than majoring in education and taking a few math classes). All
states have some procedure to certify teachers as eligible to teach in that state.
A number of states have acted to toughen certification requirements by raising the
basic education requirement, creating certification exams, and/or using a proba-
tion period coupled with an on-the-job evaluation. In 1991, 40 states required
teachers to pass specific tests for initial certification, all but 3 of which took effect
since 1980.

Regarding teacher pay, the two common proposals are for higher teacher
salaries generally and for the adoption of a merit-pay system for salary increases,
with those increases depending on some measure of a teacher’s “success.” The first
is intended to attract more skilled people into teaching, whereas the second is
intended both as an incentive for teachers to be more successful and as a reward
for teachers who are. The average annual salary of public elementary and sec-
ondary school teachers was $45,822 in 2002-2003 (U.S. Department of Education,
2003). 11 Although the average nominal salary of teachers has increased essentially
continually since 1960, real average salaries have risen and fallen over this period.
For instance, real salaries in 1985 ($40,636 in 2002-2003 dollars) were less than in
1970 ($41,587). Of course, the average real salary of all workers declined some in
the 1970s, although Hanushek (1986) presents evidence that suggests that the real
salaries of teachers declined slightly more than those of all workers in those years.

“In contrast, the median annual earnings in 2001 of workers 25-years-old and over with at least a Bachelor’s degree
was about $55,930 for men and $40,995 for women.

since 1985, real average salaries increased in the period 1985 to 1991, fell from 1991
to 1997, and have increased again after 1997. Still, real average teacher salaries
have increased by only about 13 percent in the 20 years since 1985, less than 1 per-
cen t per year.

There seem to be at least three important economic issues about these proposals

to alter teacher pay. First, increased salaries may not be successful in attracting
more skilled people into teaching soon if no mechanism is in place to create job

vacancies for these individuals, and if teacher certification requirements prevent

s
ome people from moving into teaching without additional specialized training.

second, increases in teacher pay generally may not succeed in attracting more of
the scarcest teachers, those in mathematics and science. The opportunity costs for
people trained in those disciplines may require paying different salaries to teach-
ers of different subjects, even if they have the same education and experience.
Third, although merit pay is likely to induce teachers to spend more time generat-
ing the results on which the merit evaluation is based, that will improve education
only to the extent that the performance test is valuable or appropriate. If the merit
pay is based on the average performance of students, then teachers have an incen-
tive to maximize test scores and may be less concerned with the distribution of
those scores, as previously discussed.

SCHOOL SIZE AND PERFORMANCE12

Although the number of school districts in approximately 4,000 schools comprising

the United States has decreased substan- 18 percent of all secondary schools covers

tially as a result of consolidations, the struc- five or six years of grades, essentially com-

ture of school districts has not changed sub- bining the junior and senior high school

stantially. The Census Bureau reported that grades. The U.S. Department of Education
in 2002, 77 percent of independent school further reports that of all the regular junior-
districts and 71 percent of dependent (city and senior high schools (18,456 in number),

or county) districts provided both ele- about 8 percent had enrollment of less than

mentary and secondary grades. The great 100 students, 27 percent enrollment of less

bulk of high schools—nearly 14,000 repre- than 300 students, and 50 percent enroll-

senting about 75 percent of all secondary ment of less than 600.

schools—cover three or four years involving

A substantial number of economies of

grades 10 to 12 or 9 to 12. Another set of scale or size studies in the U.S. context have

“This section draws from Ronald C. Fisher, “Organization of Educational Production: Schools, School Districts, and
Consolidation.” Working paper presented at the annual conference of the International Institute of Public

Finance, Milan, 2004.

Application 19.2

C HAPTER NINETEEN ■ EDUCATION

531

Application 19.2 — School Size and Performance

the result in a variety of statistical studies

that an optimal size high school is between

600 and 1,000 students seems reasonable

and understandable.

Evidence also shows that the cost struc-

ture for providing primary education is differ-

ent from that for providing secondary educa-

tion, as the optimal size elementary school

(300 to 500) is about half that of the optimal

size high school (600 to 1,000). Unless a highly

skewed age-distribution of children exists

in a school district, there is a standard rela-

tionship between district and school size at

each level. For districts that provide both ele-

mentary and secondary education, elemen-

tary students represent about 54 percent of

the total (7 out of 13 grades) and high school

students about 31 percent (4 out of 13).

Districts with 2,000 to 3,500 students are

expected to have approximately 1,100 to

1,900 elementary students and 600 to 1,000

high school students, permitting an “opti-

mal” size high school and multiple “optimal”

size elementary schools. After districts

exceed 4,000 students or so, single high

schools may become too large, or it may be

efficient to operate multiple high schools.

Districts with fewer than 2,000 students, on

the other hand, may be too small to operate

an efficient size high school.

The situation in Michigan illustrates the

difficulties posed by district and school

organization. Of the approximately 550 inde-

pendent public school districts in Michigan,

95 percent provide both elementary and

secondary education. The enrollment in

these districts varies dramatically. The result-

ing size distribution for the approximately

460 public high schools that cover grades 9

through 12 is shown in Table 19.6. About

40 percent of Michigan’s traditional high

schools are smaller than the optimal size

range for high schools (less than 600 stu-

dents) and about 40 percent are larger than

the optimal size range (more than 900 to

1,000 students). Similar circumstances exist

in other states, as well.

This suggests a whole range of school

organization options—including separate

primary and secondary districts, small pri-

mary districts contracting with larger inte-

grated districts for high school service, high

schools that are jointly operated by separate

K-8 districts, and a possible increased state

role for secondary as opposed to primary

education.

Michigan Public High Schools, Grades 9-12 .

Rank/Percentile

Minimum
20th Percentile
40th Percentile
Median
60th Percentile
80th Percentile
Maximum

461 total schools with mean enrollment of 897

Enrollment

127
406
624
769
951

1,416
2,596

PART V ■ APPLICATIONS AND POLICY ANALYSIS

530

Application 19.2 — School Size and Performance

been conducted to examine both size effects

for local school districts as well as individual

schools. That research was initially reviewed

and summarized by Fox (1981) and most

recently by Andrews, Duncombe, and Yinger

(2002). In some cases, these studies examine

the effects of size (enrollment) on costs, hold-

ing output constant, whereas other studies

examine the effect of size on output (student

performance), holding cost constant.

In their review of production function

studies, Andrews etal. (2002, p. 258) conclude

that”… decreasing returns to size may begin

to emerge for high schools above 1,000 stu-

dents and elementary schools above 600 stu-

dents.” Lee and Smith (1997) find that high

schools of between 600 and 900 students

maximize student performance, while Eberts

et al. (1984) find that elementary schools of

between 300 and 500 students seem optimal.

Taking all the studies into account, Andrews

et al. (2002, p. 246) conclude that there likely

are “… potentially sizeable cost savings up

to district enrollment levels between 2,000

and 4,000 students” and that” … moderately

sized elementary schools (300-500 students)

and high schools (600-900 students) may

optimally balance economies of size with

negative effects of large schools.”

Monk and Haller (1993) carefully examine

the effect of high school size on the variety of

classes in the high school curriculum, disag-

gregating effects both by academic discipline

and by the target audience (advanced vs.

remedial). They report that “… there are

stronger positive relationships between

school size and course offerings in foreign

languages and the performing and visual arts

than in mathematics and social studies” and

practically no relationship in English and

science. Specifically, their results show

that “… the largest schools offer more than

fifteen additional foreign language and more

than sixteen additional … arts courses than

do the smallest schools.” Focusing on the frac-

tion of classes in each disciplinary area that

are “specialized: that is targeted either to

advanced or remedial students, Monk and

Haller find that “the percentage share of the

courses earmarked for either remedial or

advanced students increases with school

size.” A particularly striking pattern emerges

for mathematics, which is the area with the

least degree of class specialization in the

smaller schools but the area of greatest

specialization in the largest schools.’ 3

A substantial number of studies have

emerged recently in the education litera-

ture suggesting that large high schools can

have substantial negative effects on student

performance (and/or cost), especially for

disadvantaged students. The issues usually

include concern about an environment that

effectively discourages student and staff

motivation and effort in large schools, the

potential for less parental involvement, the

potential for higher labor costs, and possibly

greater opportunity costs for students due to

greater transportation distances. Many of

these studies seem to focus on large high

schools with enrollments of approximately

1,500 students or more. From this perspective,

13lnterestingly, Monk and Haller find that the increased specialization in mathematics classes in larger high schools
arise primarily because of the offering of more remedial classes. Thus, the students needing the most help may
be most disadvantaged in mathematics by small schools.

PART V ■ APPLICATIONS AND POLICY ANALYSIS C HAPTER NINETEEN ■ EDUCATION

533

Assessment and Accountability

You have learned previously in this chapter about the changes many states have
made in their financing systems in an attempt to improve education or to meet
court-ordered expectations. The other major development in education policy in
the past decade has been an increased focus on evaluating students, schools, and
educational results. State governments initiated the emphasis on accountability
during the 1990s, which in many ways was the natural result of those legal deci-
sions that forced states to take more fiscal responsibility for distributing educa-
tional resources and for ensuring adequate educational production. The state
emphasis on accountability also arose from the now well-documented long-run
trend of increasing real per-student spending by public schools and decreasing
average class sizes, which is coupled with student performance that—as mea-
sured by a wide variety of tests comparing students in the United States as well as
comparing U.S. students internationally—either declined or did not improve
nearly as fast as spending grew. This fact has induced states to want to improve the
results of public education systems and to find ways to ensure that the increasing
state spending is being used in the most effective manner.

With the approval of the No Child Left Behind Act in 2002, advocated by the
Bush administration, the federal government became an additional force encour-
aging educational assessment and accountability. The act requires states to
adopt education-assessment systems and accountability measures that impose
consequences—such as denying federal education grants—on school systems that
show poor assessment results.

“No Child Lett Behind”

Even before the No Child Left Behind Act was passed and signed in 2002, many
states had adopted and implemented new educational standards and assessment
mechanisms. Ladd (2001, p. 385) reported that “Forty-five states now [in 2001]
have report cards on schools, and 27 of them rate schools or identify low perform-
ing schools.” Hanushek and Raymond (2001, p. 369) noted that “The basic skele-
ton of accountability systems involves goals, standards for performance, measure-
ment, and consequences….” They further reported that although few states had
set clear goals for their accountability systems, essentially all states had established
standards for performance and tested students in some form. In addition, most of
the states also evaluated and reported on the performance of schools, but perhaps
only about half of the states had explicit consequences for poor performance by
either students or schools. The No Child Left Behind Act put the force of the fed-
eral government behind this trend and made inescapable the increased focus by
states on educational performance.

The No Child Left Behind Act established a series of assessment provisions and
procedures that states are to pursue, as follows:

■ Each state identifies an assessment mechanism (usually some test or set of
tests) and sets specific academic achievement levels (called “proficiency
levels”) for reading and mathematics based on the state’s assessment method.

■ Each state sets performance goals, measured as the percentage of students,
by grade level and for both reading and mathematics, who meet the
proficiency level on the state assessment mechanism. The performance
goals are to be increased each year so that all students are at the proficient
level by the end of the 2013-2014 school year.

■ The results of the state assessment (test scores or the percentage of students
achieving the proficient level) are to be reported for all schools and for
specific subgroups of students in each school, including low-income
students, racial or ethnic minorities, students with disabilities, students
with limited English language capability, and others.

■ By the 2005-2006 school year, states are to have a system in place to test
every student in grades 3 through 8 annually and students in grades 10
through 12 once in both reading and mathematics. Testing in science must
be added by the following year.

■ Schools and school districts will be evaluated each year as to whether they
are making “adequate yearly progress” toward the ultimate goal of full
proficiency for all students by 2014. Adequate yearly progress is defined as
all student groups (including each subgroup) in that school or district
meeting the proficiency level percentage goal for that year. If the percentage
of students in any group failing to achieve a proficiency level score on the
test exceeds that year’s goal, then the entire school and district are deemed
to be not making adequate yearly progress.

■ At least 95 percent of the students in a school and in each subgroup must
take the assessment test for that school to be deemed as making adequate
yearly progress.

■ Schools that fail to make adequate yearly progress face a number of
special requirements and procedures, including such things as developing
and implementing a school improvement plan, allocating additional
federal funds to teacher improvement, providing the opportunity for
students to transfer to other schools, provide special tutoring assistance
to selected students, and ultimately facing the possibility of school staff
replacement, consolidation with other schools, state takeover, or private
firm management.

The NCLB act has a number of other provisions and expectations concerning
testing students with disabilities, testing students with limited proficiency in
the English language, measuring and certifying teacher quality, and applying the
schedule and set of required remedies and actions to schools that fail to
make adequate yearly progress, called “schools in need of improvement.”
The carrot or stick, depending on one’s point of view, that enforces NCLB is the
$33 billion of federal government support for public primary and secondary
education in 2002. Although only about 8 percent of total revenue for public
1(-12 education, this is still an amount of funds that provides the federal govern-

ent substantial clout.

PART V ■ APPLICATIONS AND POLICY ANALYSIS HAPTER NINETEEN ■ EDUCATION

NCLB is controversial, to say the least. States have expressed concern about a
variety of issues, most notably the costs associated with implementing and operat-
ing such an extensive assessment system, the seeming rigidity of requiring annual
improvements for all groups of students in all tested areas, the complicated proce-
dures for dealing with students with special circumstances, the costs associated
with educational improvements needed to make adequate yearly progress, th e

notion of 100 percent proficiency as the ultimate goal (almost assuring that every
school system will fail), and what some see as an unwarranted and heavy-handed
intrusion by the federal government into matters that traditionally have been the
responsibility and purview of state and local governments. All these issues cannot
be resolved here, and likely will not in practice for several years at least, but the
economic and fiscal analysis presented in this chapter can be used to clarify these
issues.

Aaseasing and Accountability Issues

Four issues about educational accountability seem to be the most difficult and con-
tentious: (1) who should be evaluated—students, teachers, schools, or school dis-
tricts; (2) how should the evaluation be structured—that is, what are the relative
advantages of various evaluation methods; (3) what level of government should be
primarily responsible for setting standards, conducting the evaluation, and paying
the costs—states or the federal government; and (4) what remedies or conse-
quences should apply to schools or students who fail to meet assessment stan-
dards. Each is considered next.

Unit ob Evaluation

The production function approach to education presented earlier in this chapter
can be used to illustrate the difficulties of using districts, schools, teachers, or
students as the unit of evaluation. Districts and teachers clearly seem the worst
options. Education is produced in schools, not districts, and focusing on district
average results would allow poorly performing schools to be hidden by other
schools that perform well. A given teacher in a single grade is but one input
among many that affect a student’s learning, and thus it would seem nearly
impossible to attribute a student’s score on an assessment test to one single
teacher.

Between students and schools, Ladd (2001, p. 398) argues that “schools are the
most logical starting point for a top-down accountability system. . ..” First,
schools are the production unit that controls educational resources and can act to
reallocate those resources in an attempt to improve educational performance. Sec-
ond, what school to attend is the educational unit that families most directly select
and that families can change if performance is unacceptable. Third, poorly per-
forming schools cannot escape notice and attention if schools are the focus of
assessment. Finally, as Ladd (p. 389) puts it, “school-based incentive rewards pro-
vide an incentive for all school personnel to work cooperatively toward a well-
specified goal.”

The concerns about focusing on schools also are easy to note. Student learning
and knowledge, either that which students actually possess or that which may be
represented or measured by test scores, is a cumulative result both of learning
done in all schools attended and of learning done outside of school through per-
onal or family experiences and activities. Scores on tests administered in the mid-
dle of the 8th grade, for example, can hardly be attributed only to the 15 months
that the student attended that particular middle school. Certainly the elementary
schools that the students attended, which might be in a different district, different
state, or even , a different country, is expected to have had an impact. In addition,
one cannot minimize the importance of the learning that students achieve
through interactions with their fellow students as well the learning that arises
front private activity—parental reading or teaching, books in the home, family
travel or other experiences, athletics, music, and so on. Finally, not all students are
in a personal or community environment that places equal value on education or
provides equal motivation. Surely, the score a student achieves on a test in the 3rd
grade, 8th grade, or 11th grade reflects the combined effects of all of these influ-
ences. Why, then, should those scores be attributed solely to the contribution of
the last school?

In the end, one of the strongest arguments for assessing schools may be the dif-
ficulties with assessing and penalizing only students. Although students certainly
influence their own educations by their behavior, they have little direct control or
influence over the allocation of educational resources; students do not control cur-
ricula, hire teachers, maintain facilities, and so on. Some of the potential problems
of focusing assessment on schools may be mitigated by appropriately structuring
the assessment instrument and information, as discussed next.

Method6 of Evaluation

A host of controversial and well-known issues exist about student testing. How-
ever, assuming that one believes in the value of a particular test, results may be

to shed light on the school contribution. One option is to compare schools
only with the same populations—schools in rural, low-income communities to
those in similar communities; predominately minority, central city schools to the

e; and so on. A second option is to correct, using statistical methods, for dif-
t student populations. Usually this involves comparing the results in a par-
r school or district to the results that might be expected or predicted for the

ulation of students in that district. A third possibility is to look only at the
vement of students in a particular district. If a district’s average score is at

50th percentile of the statewide average for students in 3rd grade, say, but
the 75th percentile for that same group of students in the 11th grade, then the

t would seem to have made relative improvement in the students’ perfor-
e. Which is having a greater impact—such a district or one that has students
perform at the 85th percentile in both grades?

issue, then, is whether to focus on the level of outcome or result by a student
ool or on the change in that level by a student or school over some time. The
ction is important because factors specific to a student (innate ability, effort,

CHAPTER NINETEEN ■ EDUCATION PART V ■ APPLICATIONS AND POLICY ANALYSIS

536 537

family circumstances, social environment) are expected to influence the student’s
level of achievement, and those factors may be difficult to measure and thus con-
trol for in studies of educational outcomes. By focusing on the change in achieve-
ment for a given student or set of students over time, those other student-specific
factors are held constant, so that the change in achievement may reflect the value
added by the educational system.

Many times, the same standardized test is given to students at different times—
for instance, in grades 5, 8, and 12—and the scores at each grade level compared to
some average or norm for that level. The student’s score relative to the norm at one
grade level (90 percent of the norm in grade 5) compared to the same student’s
score in a later grade (110 percent of the norm in grade 12) may reflect the improve-
ment caused by the school system. The fact that the average 12th-grade score for
two schools is both 110 percent of the norm may not mean that both schools are
doing an equally good educational job if the students in one school started at a
lower level. The change in scores for the same students may be the preferable mea-
sure. In fact, it may be that a school with a lower average 12th-grade score has a
greater value added than some other school with a higher average score, but one
whose students started at a higher level.

A preference for measuring educational performance by value added led the
state of Utah to adopt legislation in 2005 giving Utah’s state education evaluation
system priority over the procedures specified in NCLB, even at the risk of losing
$76 million of federal education aid to the state. Utah measures student achieve-
ment by value added, comparing achievement for the same set of students as they
progress from grade to grade. NCLB, on the other hand, requires states to compare
test scores for one set of students in a given grade to scores for a different group of
students in that grade the following year. By one report, as many as half the states
have been considering legislation urging changes in NCLB or permitting state offi-
cials not to follow its requirements (and risk losing federal aid). Vermont has
adopted a law that gives local school districts that option.”

Local, State, or National Responsibility

States remain concerned about the federal government imposing a single evaluation
mechanism on all schools regardless of differences in circumstances or expectations
of state residents. Many state officials have argued that state assessment systems are
better able to accommodate differences among states and differences among districts
within states. Alternatively, state officials have asked the federal government to
build in more flexibility to the NCLB system. You have seen evidence earlier in this
chapter about the substantial variation in educational levels and emphasis between
states and even between districts within states. A report of a task force on the NCLB
act composed of state officials and organized through the National Conference of
State Legislatures (NCSL) even questions the constitutionality of NCLB because the
conditions that the federal government imposes through NCLB on receipt of federal

“See Alan Greenblatt. “The Left Behind Syndrome.” Governing, September 2004.

education grants may violate the 10th Amendment, which reserves certain powers
to the states, because it uses coercion to attain state participation. 15

States also are extremely concerned about the costs of administering the assess-
ments required by NCLB as well as the costs that schools will face to improve
achievement toward the 100-percent proficiency goal. The NCSL Task Force notes
that the federal government’s share of K-12 education funding has risen by about
2 percentage points (to 8 percent of total education revenue) since adoption of
NCLB. Estimates of the administrative costs to schools of implementing the assess-
ment procedures of NCLB seems to be have about the same magnitude. So, one can
argue that the federal government has provided roughly sufficient additional
funds to cover the administrative costs of NCLB. Additional state and local gov-
ernment costs to bring all students to a proficiency level are likely to be of much
greater magnitude, however, which the federal government seems unwilling to
provide. The NCSL Task Force therefore questions whether NCLB should be eval-
uated as an unfounded mandate imposed by the federal government on states,
which would require additional federal action by federal law.

Consequences and Remedies

Assessment consequences can include releasing information about the perfor-
mance of schools, providing rewards (perhaps in the form of additional resources)
to schools that meet specific performance objectives, or imposing penalties on
schools that do not meet assessment standards. Some states have incorporated
reward/penalty provisions in their education grant systems, essentially threaten-
ing to reduce resources to poorly performing schools. This might be seen as coun-
terproductive by some, although the concept is that resource reductions would
induce additional students to leave poorly performing schools and to move to
other schools with better performance ratings, which would then receive addi-
tional resources.

Perhaps the most important long-run issue is how students, teachers, schools,
school districts, states, and the federal government will respond to cases where
schools are found to be performing poorly, whether measured by adequate yearly
progress as mandated by NCLB or by some other standard. The options through
NCLB are relatively limited. Students can switch schools, low-achieving students
may receive additional tutoring, school staff can be changed, the school day can
be lengthened, or curricula can be revised. Unfortunately, the economic research
about education production discussed earlier in this chapter does not provide
much guidance. That research, as well as the body of research by educational spe-
vralists, has identified only a limited number of curriculum, teacher, and technology

ptions that clearly seem to improve educational results.
The fact that research about education production has not identified many spe-
ic factors that can be used to boost teacher productivity and educational results

caused a number of states to experiment with entirely new approaches. Some

Report of the Task Force on No Child Left Behind.” National Conference of State Legislatures, February 2005.

ER NINETEEN ■ EDUCATION

ecOadary Education Comparieona,
Selected Nation&

Percentage of
15 Year-
Olds in

Secondary
School
1999 2001 2000

92.0% 10.8 $6,894

91.9 9.9 8,578

93.4 172 5,947

92.8 13.9 7,726

95.3 11.2 7,636
96.7 13.7 6,826

91.5 – 4,638

78.7 10.4 7,218

95.1 14.0 6,266

94.2 92 8,476
85.4 – 5,185
96.8 16.6 6,339

90.5 9,780

84.5 12.3 5,991

88.5 14.8 8,855

Pupils Per Secondary Average Average Average

Teacher in Education Literacy Score* Literacy Score Literacy Score

Upper- Expenditure for 15-Year- for 15-Year- for 15-Year-

Secondary Per Olds in Olds in Olds in

Education Student Reading Mathematics Science

Australia
Austria
Canada
Denmark
France
Germany
Ireland
Italy
Japan
Norway
Spain
Sweden
Switzerland
United Kingdom
United States

2000

2000

2000

528

533

528

507

515

519

534

533

529

497

514

481

505

517

500

484

490

487

527

503

513

487

457

478

522

557

550

505

499

500

493

476

491

516

510

512

494

529

496

523

529

532

504

493

499

‘Scales were designed to have an average score of 500 points and standard deviation of 100.

SOURCE: US. Department of Education.
Digest of Education Statistics. 2003.

The information reported in Table 19.7 and Figures 19.7 and 19.8 (and other
background data from the same sources) suggest the following comparative obser-
vations about primary and secondary education in the United States:

■ The United States spends about 4 percent of GDP on

primary and

secondary education, which is about average among the developed nations
represented by the Organization for Economic Cooperation and
Development (OECD).

■ Expenditure per student is relatively high in the United States (second only
to Luxembourg for primary education and fourth behind Luxembourg,
Switzerland, and Norway for secondary education).

■ The United States is about average in terms of the percentage of students
attending or graduating from secondary schools, with about 89 percent
of 16-year-olds in school (but compared to 97 percent in Germany and
Sweden, 95 percent in France and Japan, and 94 percent in Canada) and
about 75 percent of students graduating from secondary school at the
typical age (compared to an OECD average of 78 percent).

■ Typical class sizes in the United States also are about average, substantially
smaller than in such nations as Canada and Japan, but larger than in such
nations as Denmark and Switzerland.

PART V ■ APPLICATIONS AND POLICY ANALYSIS

states, such as Kentucky, have experimented with grouping students differently
(eliminating traditional elementary grades) and less-structured classroom activity.
Other states are experimenting with changes to the length of both the school
day and school year; perhaps it would be preferable if students attended school for
fewer hours (providing time for more personal study and work) but more days
(eliminating long breaks away from school). The results of these and other experi-
ments are likely to be important in improving education in the future.

States also have acted to change school curricula and the types of courses stu-
dents take, largely by altering graduation requirements imposed by state govern-
ments. According to the U.S. Department of Education (2005), all but six states
(Colorado, Iowa, Massachusetts, Michigan, Nebraska, and Pennsylvania) have
substantial state-set academic requirements for high school graduation. Many of
those state standards have been established or strengthened since 1985. The other
six states essentially leave those standards as an option for local districts. Among
those states with course requirements imposed by the state government, common
requirements are three to four units (years) of English and two to three units of
mathematics, science, and social studies each. One of the most dramatic changes
occurred in Florida, which now has among the most stringent requirements. Local
school districts previously determined requirements, but now all high school grad-
uates are required to have four units of English; have three units each of social
studies, mathematics, and science; and pass a minimum competency test. Indeed,
20 states now require students to pass a competency test to graduate.

Therefore, schools already have incentives to use any new methods or technolo-
gies known to improve educational performance, although they may not always
have sufficient resources to apply all of them effectively. In addition, in the cases of
some students, it just may not be possible through the application of additional
school resources to offset a variety of environmental factors-the social inputs and
peer effects in the educational production function-that work against those stu-
dents’ educational achievements. Therefore, simply identifying that a school (and
its students) is performing poorly and not meeting assessment standards is not
particularly helpful unless some clear mechanisms are known that will improve
education results for that particular group of students, and unless adequate
resources are available to implement those mechanisms.

INTERNATIONAL COMPARISON

Not surprisingly, the structure and financing of primary and secondary education
varies greatly among industrialized nations. Comparisons are difficult because of
differences in the structure of government generally, because of problems in con-
verting financial amounts to comparable units, and because of obvious cultural
differences. Still, comparing primary and secondary education in the United States
to other nations both illustrates many of the issues discussed in this chapter and
suggests options that the United States might consider for altering its educational
system.

Figure 19.7

Expenditure on

primary and
secondary

education as a

percentage of

GDP, 2002

Figure 19.8

Public expenditure on educational institutions (2002)

0 Private expenditure on educational institutions (2002)

♦ Public and private expenditure on educational institutions (1995)

a- –
•

Country mean
-•'” • •

,
ct`

‘ 4,1■• ,,,ZP ce „.0 44- e oke 4,5 6 Ib 6″, <3, •c 6 ts c, 0e• c, ,be \40 0 ‹,‘.

0:5 .4. ,tve ,45, .0`0, Asa *15 4, 6? ,473′ zo 6••
N.- ,0″,0 ?se.> 4.56,0e. ei

0+ in 4;
(..”

1. Public subsidies included in private expenditure.

SOURCE: OECD, Education at a Glance, 2005.

ES Primary education

Lower secondary education

•

0
I

o e 6 .0 (A 9 49 _.scs dib “AV’ ,(•

6 0 ‘6 e
40.0,4,e IN- e q, r, .4z e ecv ,..z9 441″ 0Iz’

ce,

1. Public institutions only.

SOURCE: OECD, Education at a Glance, 2005.
5.5

4.5 –

4.0 – –

0 3.5
a,
fir 3.0

2.5
2.0
1.5

1.0

0.5

0.0 .

40

“3 30

C
• 20

0

• 10
Z

PART V ■ APPLICATIONS AND POLICY ANALYSIS APTER NINETEEN ■ EDUCATION

Average class

size in educational

institutions

by level of

education, 2003

■ Students in the United States attend school for more hours per year (about
1,150) than in any other nation.

■ Teacher salaries in the United States relative to per-capita GDP are lower
than average at about 130 percent of per-capita GDP, compared to 180
percent in Germany, 160 percent in Japan, and 140 percent in Australia, for
instance.

■ Scores of U.S. students on standardized tests administered specifically for
international comparisons are about average, but substantially lower than
for students from Australia, Canada, Japan, and the United Kingdom, for
instance.

The picture that emerges for the United States is a nation that spends an average
fraction of its income on education, but because income (GDP) is very high, spend-
ing per student is also very high. Data reported by the OECD show clearly that
education spending per student is positively related to national income, as mea-
sured by GDP per capita. That relatively high spending in the United States funds
a larger than average amount of time in school for students rather than substan-
tially smaller than average class sizes or higher than average teacher salaries
(which could be alternative uses of the funds). The longer time in school per year
by U.S. students arises not because U.S. students go to school more days per year,
generally, but rather because of more hours per day. The comparison with Japan is
particularly dramatic. Japan spends roughly 25 percent less on education than
does the United States (3 percent of GDP in Japan compared to 4 percent in the U.S)
and has roughly 25 percent larger classes (at least in primary and lower secondary
schools). Students in Japan attend school fewer hours per year than in the United
States, but teachers are paid more relative to national GDP.

Two other important structural differences in the educational system in the
United States become clear when compared to most other nations not illustrated
by the data. First, the United States uses one of the most decentralized systems to
provide education of all nations, even compared to the other nations with federal
systems of government (having federal, state, and local governments, such as in
Australia, Canada, and Germany). Local school districts in the United States gen-
erate about 43 percent of revenue for primary and secondary education and are
responsible for spending essentially 100 percent. In comparison, reliance on local
finance is substantially lower in Germany, and state governments in Australia gov-
ern, finance, and operate the schools (as is done only in Hawaii in the United
States). Second, U.S. primary and secondary education is a uniform or nonstrati-
fied system, with students at any given age and location all participating in the
single-school system and taking a similar curriculum. A number of other nations,
notably most European nations, operate stratified systems, with students sorted
into various educational tracks at relatively young ages (as young as 10 years of
age in Austria and Germany).

Interestingly, there seems to be no clear overall relationship between these char-
acteristics of educational production systems and educational results, at least as
measured by standardized tests. As noted previously, U.S. students perform about

Primary, secondary, and post-secondary non-tertiary education

6.0 –

PART V ■ APPLICATIONS AND POLICY ANALYSIS AFTER NINETEEN ■ EDUCATION

54 2 543

average on these tests, although there is some variation by subject area; U.S. stu-
dents score relatively better in reading than mathematics. The OECD reports that
although students in stratified educational systems tend to perform relatively less
well, this tendency is small and not statistically significant. However, the OECD
also concludes “in countries that separate students at an early age into schools of
different types, students’ social background tends to be relatively strongly related
to their performance. Disadvantaged students are more likely to be placed in low-
status schools with less demanding curricula.. . and then to end up with relatively
poor performance. Socially advantaged students are more likely to be placed
in high-status schools with demanding curricula and then to end up with rela-
tively high quality performance…. In countries that keep students together in
comprehensive schools, the relationship between social background and educational
performance is weaken …” (OECD, Education at a Glance, 2005 Edition, p. 399).

In 2002, public elementary and secondary schools served about 48 million stu-
dents. Public school spending amounted to about $8,200 per student, 4.3 percent of
GDP and 40 percent of local government spending. Expenditures per pupil, even
after adjustment for inflation, increased substantially in the past 40 years.

State governments generated about half of the revenue for financing public ele-
mentary and secondary schools in 2001-2002, and local governments generated
about 43 percent. The federal government provided the remaining 7 percent. Since
1995, the relative role of state governments has increased, and the relative role of
local governments has decreased.

Lump-sum per-pupil grants to support local education are referred to as foun-
dation aid because the per-pupil grant represents a minimum expenditure level;
the state aid is intended to provide a basic foundation on top of which local rev-
enue supplements may be added.

Guaranteed Tax Base (GTB) or District Power Equalizing aid plans are intended
to provide an equal, basic per-pupil property tax base to each district, rather than
basic per-pupil minimum expenditure level. A GTB plan involves matching grants
that reduce the price of education to the school districts. Because the demand for
education spending is price inelastic, the price reductions that are caused by the
matching grants generally did not influence education spending very much.

Little relationship appears to exist between rising school expenditures and
improved student performance, given how those funds have been used, including
spending for smaller classes or higher teacher salaries. The intellectual skills of a
teacher as measured by a verbal ability test or the quality of college the teacher
attended tend to have a significant effect on student performance. A third general
conclusion is that the school curriculum can matter because of the link between the
number of academic courses students take and their scores on standardized tests.

In the 1970s, the primary educational policy issue concerned the differences in
per-pupil spending among districts. States altered their educational grant
programs and spent more money on education, but spending differences among

districts were not reduced and educational performance generally did not
improve. Recently, the primary issue moved from focusing on educational spend-

ing to educational results. Expenditures are a very imperfect measure of the out-

put of government in providing services, and consistent with that observation,
increasing expenditures may be necessary, but certainly are not sufficient, for
improving service results.

DISCUSSION QUESTIONS

1. Per-pupil spending often varies among school districts in a given state.
Suppose that one district spends $5,000 per pupil for instruction (excluding
transportation, lunches, administration, and so on), while another district
of about the same size spends $8,000 per pupil. What could account for this
difference? Consider factors in the categories of the quantity of inputs, the
type of inputs, the prices of inputs, and the type of output.

2. The role of state governments in providing public primary and secondary
education varies greatly. In one case, the state government operates the
school system; in a number of others, the state government provides a
substantial amount of the revenue for local schools (half or more) and sets
minimum graduation or teacher requirements; and in other cases, the state
provides either a relatively small amount of revenue or sets few standards
or both. What are the economic arguments for and against state
involvement in financing and producing education? What social and
economic characteristics of a state might influence the choice of how to
produce education? Do these help explain the cases of Hawaii and New
Hampshire or Washington compared to Oregon?

3. Refer to the “Education Grant Simulation” section earlier in this chapter.
In that illustration, a program of matching grants was not effective in
equalizing per-pupil spending because demand was relatively inelastic.
What other means might be used to narrow these spending differences?
Outline the specifics of a state program that you believe would be
successful in setting a minimum per-pupil spending level of $9,444, the
average level in the illustration. Explain the effect of that program on each
district and discuss whether you would support such a change in your
state.

4. Suppose your college or university decides to evaluate its undergraduate
program to determine how successful it is at educating students. How
should the output of a university be measured? In terms of education only,
what characteristics do you think show how good of a job a college does?
How should the teaching output or quality of individual professors be
measured? Does your university attempt to measure education output or
teaching success? Does your university have a merit pay system for faculty,
and if so, what role does education output or teaching quality play?

SUMMARY

TRANSPORTATION

CHAPTER 20

. In no other major area are pricing practices so
irrational, so out of date, and so conducive to waste as

in urban transportation.1
—WILLIAM S. VICKREY

HEADLINES

“CONGESTION CONTINUES TO GROW IN AMERICA’S URBAN AREAS. DESPITE A SLOW GROWTH IN

JOBS AND TRAVEL. IN 2003, CONGESTION CAUSED 3.7 BILLION HOURS OF TRAVEL DELAY AND

2.3 BILLION GALLONS OF WASTED FUEL . . . TO A TOTAL COST OF MORE THAN $63 BILLION.

IN GENERAL, TRAFFIC CONGESTION IS WORSE IN THE LARGER URBAN AREAS THAN IN THE

SMALLER ONES. TRAFFIC CONGESTION LEVELS HAVE INCREASED IN EVERY AREA SINCE 1982.

CONGESTION EXTENDS TO MORE TIME OF THE DAY, MORE ROADS, AFFECTS MORE OF THE TRAVEL

AND CREATES MORE EXTRA TRAVEL TIME THAN IN THE PAST. AND CONGESTION LEVELS HAVE.

RISEN IN ALL SIZE CATEGORIES, INDICATING THAT EVEN THE SMALLER AREAS ARE NOT ABLE TO

KEEP PACE WITH RISING DEMAND.

THE AVERAGE COST PER TRAVELER IN THE 85 URBAN AREAS WAS $794 IN 2003. . .

2.3 BILLION GALLONS OF FUEL WERE WASTED IN THE 85 URBAN AREAS. THE AVERAGE DELAY PER

PEAK TRAVELER IN THE 85 URBAN AREAS IS 47 HOURS.
2′

1-Pricing in Urban and Suburban Transport.” AnteriCatt ECOIWYPIW Review (May 1963); p. 452.

25chrank, David and Tim Loinax. The Urban Mobility Report. Texas Transportation Institute, The Texas A&M

University System, http://mobility.tainmedu , May, 2005.

PART V ■ APPLICATIONS AND POLICY ANALYSIS

544

SELECTED READINGS

Fisher, Ronald C. and Leslie Papke. “Local Government Responses to Education Grants.”
National Tax Journal, 53 (March 2000): 153-168.

“Forum on School Accountability Programs.” National Tax Journal 54, (June 2001). This
Forum includes three papers, by Eric Hanushek and Margaret Raymond; by Helen

Ladd; and by Richard Murnane and Frank Levy.

Hanushek, Eric A. “The Economics of Schooling.” Journal of Economic Literature 24
(September 1986): 1141-77.

Hanushek, Eric A. Making Schools Work. Washington, D.C.: Brookings Institution, 1994.

Ladd, Helen F. and Janet S. Hansen, eds. Making Money Matter: Financing America’s Schools.
Committee on Education Finance, Commission on Behavioral and Social Sciences and

Education, National Research Council. Washington, D.C.: National Academy Press,
1999.

Murnane, Richard J. “An Economist’s Look at Federal and State Education Policies.” In

J. Quigley and D. Rubinfeld, eds., American Domestic Priorities: An Economic Appraisal.
Berkeley: University of California Press, 1985, 118-47.

Yinger, John, ed. Helping Children Left Behind: State Aid and the Pursuit of Educational Equity.
Cambridge MA: The MIT Press, 2004.

545

• Page 1
• Page 2
• Page 3
• Page 4
• Page 5
• Page 6
• Page 7
• Page 8
• Page 9
• Page 10
• Page 11
• Page 12
• Page 13
• Page 14
• Page 15
• Page 16
• Page 17
• Page 18
• Page 19
• Page 20
• Page 21
• Page 22
• Page 23
• Page 24
• Page 25
• Page 26