a u.

 

. .2. r

‘ . 2m V...m,..._..§w.,.u~.&

 

 

 

 

LIBRARY
Michigan State
University

 

 

 

This is to certify that the

dissertation entitled

CHANGES IN THE LABOUR MARKET FOR PRIMARY
TEACHERS IN BRAZIL FOLLOWING THE FUNDEF REFORM

presented by

MARIBEL ALVES FIERRO SEVILLA

. has been accepted towards fulﬁllment
of the requirements for

Ph.D. degree in W Administration

090 1

Major professor

 

 

Date 10/14/2002

 

MS U is an Affirmative Action/Equal Opportunity Institution 0- 12771

 

PLACE IN RETURN BOX to remove this checkout from your record.
TO AVOID FINES return on or before date due.
MAY BE RECALLED with earlier due date if requested.

 

DATE DUE

DATE DUE

DATE DUE

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

6/01 c:/CIRCJDateDue.p65-p.15

 

CHANGES IN THE LABOUR MARKET FOR PRIMARY TEACHERS IN BRAZIL
FOLLOWING THE FUNDEF REFORM
By

Maribel Alves Fierro Sevilla

A DISSERTATION

Submitted to
Michigan State University
in partial fulﬁllment of the requirements
for the degree of

DOCTOR OF PHILOSOFY

Department of Educational Administration

2002

ABSTRACT

CHANGES IN THE LABOUR MARKET FOR PRIMARY TEACHERS IN BRAZIL
FOLLOWING THE FUNDEF REFORM
By

Maribel Alves Fierro Sevilla

During the 1990’s, under a context of economic adjustment, high inﬂation, and
high unemployment rates, educational policies in developing countries have focused on
the expansion of educational systems in order to provide education for all school-age
population In Brazil, the rapid expansion of primary schooling have also resulted in
increasing the number of teachers regardless of their quality, and consequently, there
have been a decline in the quality of the educational services provided and an increase in
learning opportunity inequalities. This paper focuses on the Brazilian policy to address
these problems: the reform of the ﬁnance of primary education through the institution of
the Fund for Maintenance and Development of Primary Education and Teacher
Enhancement (FUNDEF). The reform aims at reducing disparities in education ﬁnance
among school networks of primary education and guaranteeing minimum per pupil
revenues to support an adequate minimum average level of teacher earnings. This study
shows that the impact of F UNDEF on teacher earnings and teacher supply was
observable even only aﬁer two years of reform implementation. Analysis using the

PNAD/IBGE data suggests that, there were signiﬁcant increases in teacher wages of

municipal teachers due to the reform. The relative average wage rate of primary teachers
changed and there were heterogeneous reform effects depending on the teachers’ years of
education and school location. Surprisingly, there were also positive reform effects on the
supply of teachers, but these effects were very small. The impact of the reform on the
teacher supply occurred directly and through the wage differentials brought by the
reform. The fact that changing the distribution of resources and implementing a more
equitable school ﬁnance mechanism also supports making the teaching career more
attractive is one of the most interesting ﬁndings of this study. Increases on education
ﬁnance equity positively affects the functioning of education systems and its reserve of
workers, the teachers, which in turn might reflect in future improvements in the quality of

the teaching-leaming process in the Brazilian primary education.

Copynghtby
MARIBEL ALVES FIERRO SEVILLA

2002

ACKNOWLEDGEMENTS

I would like to acknowledge the ﬁnancial support provided by the Brazilian
agency for human resource deveIOpment, Coordenacao de Aperfeicoamento de Pessoal
de Ensino Superior (CAPES), during the four years of my studies that allowed me to
undertake the Doctorate Program at Michigan State University, and by the Graduate
School that ﬁnanced me during the research phase of my dissertation. I am very grateful
for both institutions.

I am also thankful to many people who have assisted me over the process of
acquiring a Ph.D. In special, I am thankful to the members of my advisory committee,
and especially to the chairman, Prof. David Nathan Plank, whose steady support and
conﬁdence guided me through the entire process, since the choice of my program, until
the conclusion of this dissertation and further career choices. Professors Gary Sykes and
David Arsen supported my endeavors through valuable advice and encouragement at
several stages of the project. Professor John Strauss has been extremely helpful in
guiding me through the methodological issues of my research.

Historically, I may also be thankful to Prof. Fernando Spagnolo, who brought me
to the fascinating area of Education and especially educational research. To my friends
from IPEA, J. Amara], A. C. Xavier and A. E. Marques, that Showed me the economic
emphases of educational planning. I would also thank to the support and comments of

Professor Céndido Gomes who helped me to keep update with the changes in the system

while I was at MSU. While in Brazil, I also profited greatly form discussions with Robin
Horn, Soﬁa Lercher, Robert Verhine, AntOnio Naspolini.

I would also like to express my thanks to GOmes Neto, AntOnio Carlos Moreno
and Paulino Motter, who helped me gather the data. I also thank my friends from the
Ministry of Treasury that explained the structure and functioning of the public accounts. I
liked to thanks the support of my colleague Ali M. Berker, who assisted me with the
organization and management of the statistical data, as well as gave me thoughtful
suggestions about the empirical speciﬁcations of my models. I am also grateful to the
writing group, Sharon Thomas, Andy McCullough, and Catherine Fleck, who supported
through the last three years the development of my writing Skills. Finally, I am grateful to

my family.

vi

TABLE OF CONTENTS

LIST OF TABLES .......................................................................................................... IX
LIST OF FIGURES ......................................................................................................... XI
CHAPTER 1
THE BRAZILIAN CASE .................................................................................................. 1
Introduction ..................................................................................................................... 1
Unequal Opportunities to primary education: The Brazilian Case .................................. 3
The Dimension and Structure of the Brazilian Education System ............................. 4
Causes of the unequal opportunity for education ..................................................... 13
Redistributive inefﬁciency in the system of educational ﬁnance ............................. 15
The solution proposed: The Fund For Maintenance and Development of Primary
Education and Teacher Enhancement — FUNDEF ....................................................... 18
Background ............................................................................................................... 1 8
The reform ................................................................................................................ 19
Goals and Objectives ............................................................................................... . 21
Target Population ...................................................................................................... 21
Operationalization ..................................................................................................... 22
Implementation ......................................................................................................... 23
Reform theory and expected reform effects .............................................................. 25
Immediate results .......................................................................................................... 28
CHAPTER 2
THE EFFECTS OF THE REFORM ON TEACHERS’ EARNINGS ............................. 33
Introduction ................................................................................................................... 33
Data and empirical speciﬁcation ................................................................................... 35
Data ........................................................................................................................... 35
Empirical speciﬁcation ............................................................................................. 39
Results ........................................................................................................................... 47
Determinants of earnings .......................................................................................... 47
The national average effects of the FUNDEF reform ............................................... 50
The effects of the FUNDEF reform and teachers’ years of education ..................... 54
The effects of the F UNDEF reform on newly hired teachers ................................... 56
Regional Differences ................................................................................................ 58
Conclusions ................................................................................................................... 65
CHAPTER 3
THE EFFECTS OF THE REFORM ON TEACHER SUPPLY ...................................... 68
Introduction ................................................................................................................... 68
The reform and teacher labor force participation ......................................................... 68
The model ..................................................................................................................... 71
Data and variables ......................................................................................................... 76

vii

Results ........................................................................................................................... 77

The determinants of primary teacher labor force participation. ............................... 77
The reform effects ..................................................................................................... 79
The structural model for labor force participation. ....................................................... 88
Wage equations ......................................................................................................... 88
Labor supply estimates using the structural model ................................................... 89
Conclusions ................................................................................................................... 90
CONCLUSION ................................................................................................................ 93
REFERENCES ................................................................................................................. 98
APPENDIX ...................................................................................... 104
APPENDIX A: DESCRIPTION OF TERMS ................................................................ 105
APPENDIX B: STRUCTURE OF THE BRAZILIAN SYSTEM ................................. 107

APPENDIX C: INTERPRETATION OF PARAMETERS UNDER THE
DIF F ERENCE-IN-DIFFERENCES (DD) AND DIFFERENCE-IN-DIFFERENCE-
IN-DIFFERENCES (DDD) F RAMEWORKS .............................................................. 108

APPENDIX D: TABLES ............................................................................................... l 12

viii

Table I:

Table 2:

Table 3:

Table 4:

Table 5:

Table 6:

Table IA:

Table 2A:

Table 3A:

Table 4A:

Table 5A:

Table 6A:

Table 7A:

LIST OF TABLES

Average Number of Completed Grades and Average Time Spent on

Primary Education, 1990-1995 ........................................................................... 9
Per Pupil Revenues in Primary Education by School Network, 1996-

1998 ................................................................................................................... 30
Estimated Reform Effects on Primary Teacher Wages using Per Pupil

Revenues as predictors ...................................................................................... 63
Estimates of the Reform Effects on Primary Teacher Labor Force

Participation ...................................................................................................... 81
Estimates of the Reform Effects on Primary Teacher Labor Force

Participation in Urban Areas ............................................................................. 84

Estimates of the Reform Effects on Primary Teacher Labor Force

Participation in Rural Areas .............................................................. . ................ 87
FUNDEF Balance ........................................................................................... 112
Total Sample Size and Sub-Samples with Restrictions .................................. 113
Description of Variables, Means and Standard Deviations for Primary

Teachers and Non-Primary Teachers, Brazil —- 1995 - 1999 ........................... 1 l4
Estimates of the Reform Effect on Primary Teachers’ Wages ....................... 116
Estimates of the Reform Effect on Primary Teachers' Wages ........................ 1 l9
Estimates of the Reform Effect on Primary Teachers' Wages ........................ 121

Estimates of the Reform Effect on Primary Teachers' Wages by Level

of Education .................................................................................................... 123

ix

Table 8A: Estimates of the Reform Effect on Primary Teachers' Wages by Tenure
Group .............................................................................................................. 125

Table 9A: Estimates of the Reform Effect on Primary Teachers' Wages by Age

Group .............................................................................................................. 128
Table 10A: The Reform Effects on Primary Teachers' Wages — Brazil .......................... 131
Table 11A: The Reform Effects on Primary Teachers' Wages — North Region .............. 132

Table 12A: The Reform Effects on Primary Teachers' Wages — Northeast Region ........ 133
Table 13A: The Reform Effects on Primary Teachers' Wages — Southeast Region ........ 134
Table 14A: The Reform Effects on Primary Teachers' Wages — South Region .............. 135
Table 15A: The Reform Effects on Primary Teachers' Wages — Center-West Region ...... 136

Table 16A: Descriptive Statistics for Labor Force Participation Estimation .................. 137

LIST OF FIGURES

Figure 1: Enrollment in Primary Education by Sector, 1995-2000. .................................... 6
Figure 2: Enrollment in Secondary Education by Sector, 1995-2000. ................................ 7
Figure 3: Enrollment in Higher Education by Sector, 1995-2000. ...................................... 8
Figure 4: Number of Primary Teachers by Qualiﬁcation Level, 1995-2000 ..................... 10
Figure 5: The Reform Model ............................................................................................. 27
Figure 6: Average Monthly Earnings by Occupation, 1995-1999 ..................................... 38
Figure 7: Structure of the Educational System ................................................................ 107
Figure 8: Teacher Wages Trend, Brazil 1995-1999 ......................................................... 108
Figure 9: Teacher Wages Gap Trend, Brazil 1995-1999 ................................................. 109
Figure 10: Difference-in-Difference Estimation Speciﬁcation ........................................ l 10
Figure 11: Differences-in-DifferenceS-in-Differences Speciﬁcation .............................. 1 11

xi

CHAPTER 1

THE BRAZILIAN CASE

Introduction

Both qualitative and quantitative research ﬁndings suggest that policies focused
on the quality of teachers may be related to the improvement of educational quality,
mainly measured by student achievement (Fuller and Clark, 1994). Within the context of
economic adjustment, high inﬂation, and high unemployment rates, educational policies
in Third World countries have focused on the expansion of the school system.
Investments in the educational sector have concentrated on infrastructure and input
strategies that promote the necessary rapid expansion of the systems by supplying schools
with more classrooms, more teachers, and more instructional materials (Fuller et al.,
1999). Constrained by scarce resources, the rapid expansion of primary schooling may
have also resulted in increasing the number of teachers regardless of their quality, and
consequently, there may have been a decline of the quality of the educational services
provided and increasing inequalities in learning opportunities.I

The Brazilian case exempliﬁes this trend. The net enrollment rate in primary
education increased from 67 percent in 1970 to 96 percent in 1999 (MEC, 1999a), even
though repetition rates are very high and completion rates low. At present, attention has
shifted from expanding the educational system to examining its quality, and, as a result,

teacher quality has emerged as a major issue. At the federal level, a series of programs has

 

I The importance of teachers in promoting opportunity to learn has been widely discussed. See Stevenson
and Stigler (1992), Anderson (1991), and Schmidt, et al. (1997).

been created to qualify current teachers,2 and a newly enacted law mandates that part of the
resources reserved for primary education be compulsorily allocated to supporting teacher
policies.

The new law created by the Brazilian federal government instituted the F undo de
Manutenciio e Desenvolvimento d0 Ensino Fundamental e de Valorizacc'z'o do Magistério
(Fund for Maintenance and Development of Primary Education and Teacher Enhancement)
(FUNDEF) in order to support this policy, guaranteeing minimum revenues per pupil in
primary education throughout the country, independent of the type of public school
Brazilians attend. F UNDEF was enacted by law in December 1996 and implemented
throughout the country in 1998. Sixty percent of this fund must be allocated toward paying
teachers, training them, and ensuring their qualiﬁcations. The recruitment and retention of
qualiﬁed teachers, through the development of policies to provide incentives and support to
the teaching profession and continuing education, are crucial for improving a country’s
quality of education (Mumane, 1991). Therefore, this study’s objectives are to determine
whether this new policy has produced any changes in the quality of the teacher labor force
(as deﬁned by level of education, years of experience, and average wage rates), to identify
where these changes may have had signiﬁcant effects, and to discuss its implications for
the quality of education in Brazil.

In order to achieve these objectives, this research addresses ﬁve main questions:
Does the reform affect teacher wages? Which teachers (municipal or state) have beneﬁted

most from the reform? Has the reform affected teacher supply? What types of teachers

 

2 Some of the programs are Pfograma TV Escola (1995), Pardmetros C urriculares Nacionais — PC NS
(1997), and PRO-FORMACAO (1998).

have been attracted to the education sector? Have the relative changes in teacher wages

inﬂuenced teacher participation decisions?

This chapter presents the Brazilian case. It discusses the characteristics of the
Brazilian educational system, the problem of primary teacher supply, and the speciﬁc
policy reform under consideration. Chapter 2 addresses the reforrn’s effects on teachers’
wages. The impact of the reform on teacher labor supply is discussed in Chapter 3. The
last section summarizes the main results and presents concluding remarks. This study is
important because it provides an empirical evaluation of the impact of a particular policy
reform on the teacher labor market. The ﬁndings of the study provide information that can
be used to determine whether the policy has resulted in improvements that might justify its

continuation when it is revised in 2003.

Unequal opportunities to primary education: The Brazilian Case

The quality of Brazilian primary education has shifted over time. While Brazilian
education was quite successful in the 19603, it experienced a period of stagnation in the
19803 and early 19905. By the year 2000, the Brazilian education sector had begun to
Show Signs of Si gniﬁcant recovery. In 1965, educational opportunities in Brazil were
superior to those of other developing countries. A large share of the school age
population was attending primary education and gender disparities in access to education
were minimal. Secondary education was not exceptional, but still, the, expected number
of years of schooling for a Brazilian student was above the international average (Birdsall
et al., 1996). By 1987, however, Brazil lagged behind with an educational system that

covered a much smaller share of its eligible population than it had in the past. Primary

education was barely above the international average and secondary education was far
below the international average (Birdsall et al., 1996).

Several authors attribute the hard times of the 19805 to the economic environment
and poor policy choices (Birdsall et al., 1996). The problems of low expenditures,
efﬁciency, efﬁcacy and equity in primary education, and the poor educational policies
directed to serve political purposes have been broadly discussed (Gomes, 1992, 1996,
1997; Plank, 1990, 1991, 1995, 1996a, 1996b). While there is a general consensus on the
structural problems the Brazilian educational system faces, the multiplicity of actors
involved in the ﬁnance, management, and delivery of the education system has made
addressing these problems complicated.

In 1988, the Federal Constitution Reform triggered a decade of signiﬁcant
expansion of primary education and improvement of the overall educational system. This
legislation created a framework for structural changes in the country and the effects of its
implementation began to become visible in the mid-19905. By 1994, the country was
experiencing a deep process of structural adjustment that launched a new period for the
education sector in 1995. Despite the overall scarcity of resources devoted to primary
education, Brazil began to enjoy the beneﬁts of economic stability, which made it
possible to promote substantial increases in the number of schools, students, and teachers

(MEC, 1997a).

The Dimension and Structure of the Brazilian Education System
In Brazil, parallel networks of schooling provide primary, secondary and tertiary

education. Some schools are managed and ﬁnanced by federal, state, or municipal

governments, and others, by the private sector. In primary and secondary education, in
1999, the federal school network is a residual of the centralized system and represents
only 0.1 and 1.6 percent of the primary and secondary schools in the country,
respectively (MEC, 2000a). Federal institutions are predominantly higher education
institutions (18.5% of higher education institutions) (MEC, 2000b). All these networks of
schooling follow the national education policy, which deﬁnes the structure, function, and
content of education.

In 1999, the public schools had, in all levels and modalities of basic education, a
total of 50 million students,3 with about 36 million in primary education (MEC, 2000a) (see
Figure 1). This represents a large expansion of the education coverage in compulsory
primary education. As Figure 1 indicates, most primary education students were enrolled
in schools sponsored by state and municipal govemments, as opposed to those ﬁJnded
privately. The net enrollment rate for the school-age population (7-14 years old) increased
from 67 percent in 1970 to about 93 percent in 1999 (MEC, 1998b, 1999a). Completion
rates at this level have also increased. With the increase in primary education enrollment
rates, the demand for teachers has also grown.

AS more students have been completing primary school, demand for secondary
education has increased signiﬁcantly. Between 1994 and 1999, enrollment at the
secondary level grew by 57.3 percent. This amounts to an additional three million students.
Between 1995 and 2000, secondary education enrollment grew from 5.3 million to 8.1

million (MEC, 1998b, 2001 a) (see Figure 2).

 

3 See Figure 7 — Structure of the system in appendix B.

Figure l: Enrollment in Primary Education by Sector, 1995-2000

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

40.0 '
+
35.0 , 4
30.0 .
E 25.0 ~
.9
E
.. 20.0
r: 4% .
o - ;
g \r
g 15.0 —
m
10.0 -
5.0 .
X— —)K + —)|(— 9K —)K
0.0 F a 1 a 1 i J
1995 1996 1997 . 1998 1999 2000
Year
[iTOEI +Federal —I_-—State +Municipa| +Privatej

 

Source: MEC, 1996, 1997a, 1998a, 1999b, and 2000a.

The private sector has decreased its participation in both primary and secondary
education in the last ﬁve years. Despite this trend, there has been an increase in overall
enrollment of 34.4 percent in secondary education in the last ﬁve years, all of which has

occurred in the public sector (see Figure 2).

Figure 2: Enrollment in Secondary Education by Sector, 1995-2000

 

 

7.0 -

6.0 a

5.0 i

4.0 -

Enrollment (Millions)

3.0 ~

2.0 -

 

 

10:-E

 

 

 

 

—)l6 )K—
r 1— T: i :r a
1995 1996 1997 1998 1999 2000
Year

 

 

 

i-O-—Total -;— Federal + State + Municipal + Private]

 

 

 

 

Source: MEC, 1996, 1997a, 1998a, 1999b, and 2000a.

In higher education, however, the private sector has markedly increased its
participation, even though the federal budget for this level of education has grown in the
last ﬁve years by 41.23 percent. This is signiﬁcantly more than its change in expenditures

on primary education, which increased by 32.25 percent during the same period.

Figure 3: Enrollment in Higher Education by Sector, 1995-2000

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

f __ .__ _ .__
3.0
73
c
.9
Li.
.- 1.5 -
c
o
E
'2'
C
I” 1.0 .
0.5 4 . 44
+ l + #1
I At .A_ ’1? ;
L ' '— M
0.0 r . i . . : a
1 995 1 996 1 997 1 998 1 999 2000
Year
1 l-O-Total -I—Federal +State —I-—Munrcrpal —)I(—Prrvata

 

Source: MEC, 1998c, 1999c, and 2000b.

Notwithstanding the quantitative expansion of enrollments, the quality of the
education provided has been the weakness of the Brazilian education system. The main
indication the system’s low level of quality is its high repetition rates. Table 1 below shows
the average number of completed grades for a cohort of students in primary education. On

average, a Student completed only 6.8 grades out of the 8 grades in 9.7 years.

Table 1: Average Number of Completed Grades and Average Time Spent on Primary

Education, 1990-1995

 

 

 

 

Year Average Number of Average Time in
Completed Grades Primary School

1990 . 6.16 9.32

1993 6.55 9.65

1995 6.77 9.69

 

Source: MEC, 1998d.

The expansion of the educational system accentuates the shortage of qualiﬁed
teachers to respond to the growing demand for education at all levels. In 1999, there were
1.5 million teachers in primary education,4 of which only 46.9 percent had any higher
education (MEC, 2000a). This corresponds to an average of 24.3 students per teacher in
primary education.

Teachers without higher education can be considered non-qualiﬁed teachers,
professores leigos, even though teachers with complete secondary education and teacher
training, curse normal, meet national requirements to teach pre-school and grades 1-4.5 In
secondary education, qualiﬁed teachers represent 88 percent of the total number of
teachers. However, many of these teachers with higher education do not have subject area
teacher training. There is a lack of qualiﬁed teachers in such subject areas as chemistry,

physics, and mathematics (MEC, 1999a).

 

4 Educational statistics count the number of teaching posts by educational system, referring to the number
of teachers in each system. However, a teacher may teach in more than one system, in varying types and
levels of education. Therefore, the total sum of teaching posts may upwardly bias the total number of
teachers. However, at present the number of teaching posts offers the best estimate available for the total
population of teachers in the country.

Before 1997, teachers with secondary education and teaching training were considered qualiﬁed to teach
in the ﬁrst six grades of primary education. Teachers with complete higher education with teaching training
(Licenciatura Plena) were qualiﬁed to teach the last two grades of primary education and secondary
education (LDB, 1997).

Figure 4: Number of Primary Teachers by Qualiﬁcation Level, 1995-2000

 

800.0

 

700.0 s '

 

 

600.0

 

500.0 -

400.0 —

300.0 .

Teachers (Thousands)

 

200.0 -

 

100.0 *

G 3—
m j)

v

 

 

 

 

0.0 . .w .
1995 1995 1997 1998 1999 2000

Year

 

 

 

i_-G-_ lncompletejrjrnaFy _-O——- Primary +ﬁ Seconds—EL—X—T::lig_her_ Education j

.__- w _-__ .__ _ J

 

 

 

 

 

 

 

 

Source: MEC, 1996, 1997, 1998a, .1999a, and 2000a.

The number of unqualiﬁed teachers has been dropping in recent years. In 1999,
only 2.1 percent of the total number of primary teachers in the country had less than eight
years of education (incomplete primary education) and 2.5 percent had only eight years of

education (primary education) (Figure 4). Their continued presence in some Brazilian

10

schools, however, may still threaten students’ opportunity to learn. In absolute numbers, in
1999 roughly 765,000 Children were taught by teachers with incomplete primary education
and 1,192,000 children were taught by teachers with only primary education.

The teacher supply problem is not entirely unique to Brazil; it is experienced in
both developed and underdeveloped countries. In Brazil, the problem has three main
aspects: (a) providing enough teachers to maintain the school systems’ expansion; (b)
replacing unqualiﬁed with qualiﬁed teachers, as well as expanding the teacher workforce
by attracting more qualiﬁed teachers; and (c) creating a system to maintain new and
higher levels of preparation of the teacher workforce, while providing in-service training
to prepare current teachers to reSpond to technological changes and higher standards.

In general, teaching remains an occupation with relatively low salaries and even
lower prestige (Sedlak & Scholssman, 1986) in almost every country. In the United
States, teachers’ wages have declined almost 15 percent relative to those of other college
graduates and nearly 20 percent relative to all other workers Since 1940 (Hanushek &
Rivkin, 1997). The quality of teachers has also declined (Manski, 1985; Weaver, 1983),
despite the increasing demand for education and the rise of educational attainment in
developed countries (Schultz, 1987). When the supply of teachers is limited, quality
might be sacriﬁced if salaries remain low relative to other professions (Baker & Smith,
1997)

In developing countries, relative teacher quality exhibits no clear patterns
(Schultz, 1987). Teacher earnings are even more difﬁcult to describe across countries in
Latin America (Liang, 1999). Studies that compare teacher salaries with those of other

professions in Latin America have reached mixed results. Depending on data, variable

11

deﬁnitions, and comparison groups, the results can be different even within the same
country (ILO, 1991; Liang, 1999; Psacharopoulos, 1996). In Argentina, Vegas (1999)
shows that in some metropolitan areas, teachers are paid less than their counterparts in
the labor force, but this is not true for all metropolitan areas in the country. In Bolivia,
Piras and Savedoff (1998) report that teachers receive higher hourly wages than their
counterparts in other occupations. Liang (1999) Shows that teachers are underpaid
relative to other workers in 12 Latin American countries if differences in hours of work
are not taken into consideration. In Costa Rica, Honduras, and Paraguay however, after
controlling for gender, schooling, experience, private sector, and location, differences in
wages disappear. In the same study, Liang (1999) also estimates wage equations
controlling for hours of work. In that case, teachers are underpaid relative to other
workers only in urban Ecuador and Brazil.

In Brazil, the trends are similar to those observed in developed countries.
Teachers’ wages have been low relative to those of other occupations; for example, in
1982, the average monthly earnings of a teacher was US$181.71 (Barreto, 1990); in
1995, it was US$161.00.6 Low teacher salaries have been signiﬁcantly related to the
country’s quality of education (Harbison, 1992), the teachers’ average education level,
and the average pupil-teacher ratio (26 students per teacher in 1999).

One of the barriers to improving teachers’ level of qualiﬁcation is their low
remuneration relative to that of other occupations in the country. Primary teachers cam
on average less than other workers outside the educational sector. Only for primary

teachers with completed higher education are teacher earnings superior to the average

 

6 The average monthly earnings in Brazil for 1995 were estimated using PNAD/IBGE data for 1995.

12

earnings of workers in other occupations. Teacher wage differences will be further

discussed in Chapter 2.

Causes of the unequal opportunity for education

As observed in the previous section, Brazil undertook an effort to expand its
access to education, but this represents only one pathway to assuring the right to quality
public education for all. In developing countries, because of the lack of resources,
educational policies focus on either expanding their educational systems or improving the
quality of basic education (Birdsall et al., 1996). This lack of resources then may
accentuate existing disparities in the system. The relative quality of education also can be
linked with the structure of the education systems. In Brazil, several factors may affect
the quality of education, such as the large number of independent schooling delivery
systems, the large variation in income per capita across and within regions (between rural
and urban areas), and an inefﬁcient redistributive system of educational ﬁnance (Plank,
1996)

As mentioned by Birdsall (1996), higher investments in education as a Share of
GDP would have been necessary if the quality of education was to be kept constant
during the expansion during the 19905. The rise in enrollment rates during the 19905
came at a signiﬁcant cost in terms of the quality of education provided and equity of the
system, as indicated by the great qualitative variation among municipalities within and
between states. For example, promotion and repetition rates from 1989 to 1997 show
signiﬁcant regional differences in quality. In 1989, the Northeast and North regions

presented promotion rates of 34 and 33% in the ﬁrst ﬁve grades of primary education,

13

while the Southeast achieved 70%. Repetition rates Show strong differences for the same
year. The Northeast and North presented 62 and 63% of repetition rates in the ﬁrst ﬁve
grades of primary education, while the Southeast presented 29%. In 1997, the gap was
still large between the regions, representing a difference of almost 50%. Promotion rates
for the Northeast and North were 45 and 43% while for the Southeast, 83%. Repetition
rates for the Northeast and North were 55 and 53% and for Southeast it was 38% (Klein,
1997).7 Slow economic growth and the continued increase of the school-age population
constrained the growth of spending on basic education per child. The expansion of the
system meant low expenditures per pupil at a time when more resources per child were
necessary to cover the costs of including the less advantaged students from lower socio-
economic backgrounds and distant, low population-density areas (Birdsall et al., 1996).

The wide variance in educational opportunities for Brazilians can be explained, at
least in part, by the fact that the government overlooked the unequal availability of
resources in the different country regions.8 Deep structural regional inequalities are
observed in almost every socioeconomic indicator, including demography, income levels
and distribution, schooling, and public service quality. Just as various states exist at

different levels of economic development, there is a large variance among municipalities.

 

7 For a description of the regional inequalities of the Brazilian educational system, see Castro (1999).

8 The decentralization trend has taken place Since the decline of the military government with a gradual
decentralization of revenues and public expenditures from the federal government to state and municipal
governments. In 1988, the Brazilian Federal Constitution established a structural reform in the ﬁscal
system. Essentially, this reform strengthened the political power and spending autonomy of state and
municipal levels of the education system. State and municipal governments can directly decide on changes
to their local revenues, creating or eliminating taxes (MEC, 1997b). Besides providing means, the
Constitution also deﬁnes the share of revenues that each level has to spend on education and the levels of
education for which each sphere of government would be mostly responsible. Once again, there was a
constitutional guarantee of revenues to the education sector. However, decentralized systems involve
substantial transition costs, including the administrative costs of the government structure itself, the
compliance costs of making decisions collectively through government, and the information problems
facing governments in discerning the public interest. The normative instrument basically reflects the
Brazilian administrative decentralization process and federal govemment’s strong distrust of state and local
political authorities' competence and motives to invest resources efﬁciently in the sector.

14

Changes in how the education system was ﬁnanced to support decentralization
provided incentives for many of these municipalities with low ﬁnancial capacity to create
and/or expand their municipal school networks. This is evidenced by the accelerated
increase in enrollment in primary and secondary municipal schools shown in Figures 1
and 2. As a result of economic inequalities among municipalities, different school
systems -- state, municipal and private -- with very different capacities, Share the
responsibility for providing education. Municipal schools, on average, provide
educational services as well as the private and state school networks, but many times

under worse economic conditions.

Redistributive inefﬁciency in the system of educational ﬁnance

According to the Federal Constitution of 1988, the federal government must apply
annually to the maintenance and development of education, at least 18 percent of the
resulting revenues of taxes, including the resources coming from inter and intra-
govemmental transfers. The state, federal district, and municipal governments must also
contribute at least 25 percent of the same revenues. Fifteen percent of the portion
allocated to education must be speciﬁcally spent on primary education (Brasil, 1998a). In
simpler terms, primary education maintained by municipal governments must invest 15
percent of total local tax revenues and state and federal transfers, while state governments
invest 15 percent of total state tax revenues and federal transfers. Primary education also
receives additional resources from social contributions as well as supplemental programs
such as the student lunch program and health programs targeted to students in primary

education (Brasil, 1988b).

15

In practice, compliance with these expenditure requirements is difﬁcult to assess.
The system of public accounting does not facilitate the identiﬁcation of irregularities in
expenditures (Gomes, 1992; Plank, 1996a). Nevertheless, analysis of public expenditures
in 1995 indicates that state and municipal governments exceeded their constitutional
obligation (MEC, 1997b). In fact, state and municipal regulations require that a larger
share of the tax revenues and intergovernmental transfers be assigned to education than
required by the federal government. In 1995, for example, the average state share of
effective expenditures in education was 31.4 percent of total state recurrent revenues
(MEC, 1997b).

The Ministry of Education (MEC) Shows that transfers play an important role in
drastically reducing the inequalities of tax revenues collected in the various states. In less
developed states, after transfers, the total tax revenues for primary education achieve
approximately twice the original share of tax revenues for primary education before
transfers (MEC, 1997b). The same report comments that this phenomenon is even more
accentuated in municipalities of the North and Northeast regions. For example, in four
states of these regions, the transfers are responsible for 90 percent of the recurrent
revenues.

However, even though inequalities have been reduced, equality is not achieved
because a large proportion of the taxes is returned to where they were collected. Richer
municipalities and states collected more and got more resources. Until 1997, the
distribution of federal resources devoted to equalization was based on negotiation with
states and municipalities. Gomes & Verhine (1996) point out that, contrary to their

objective, these transfers reinforced inequalities between the states since the distribution of

16

resources happened without technical criteria (Oliveira, 1992). For example, inequalities
within states were accentuated because rural municipalities with less access to the federal
administration of these grants also possessed less organization and technical capacity to
successﬁrlly apply for resources.

These characteristics indicate how the Brazilian education system was built upon
a decentralized structure and reinforced an implicit collaboration between systems that
provide basic education in Brazil. However, the implicit collaboration, under a context of
parallel and competitive networks of schooling (Gomes, 1992; Plank, 1996a), ended up
leaving decisions about who is responsible for providing primary education to the
discretion of state and municipal politics. Plank concludes that educational resources
were still contingent On factors such as regions, wealth, and political afﬁliation, where the
use of ﬁrnds for political interests undermined the objective of providing basic education
for all (Plank, 1996a). In this arena characterized by very unequal capacities, some states
and many municipalities received few or no extra resources at all.

Therefore, there was the need for educational reform to address these problems.
This reform, which took effect in 1996, changed the criteria for allocating
intergovernmental transfers to public school systems based on enrollment. This
mechanism changed how the educational system was ﬁnanced, leading to a more
equitable educational system in Brazil. In the long term, such a policy may promote

growth as well.

17

The solution proposed: The Fund For Maintenance and Development of Primary
Education and Teacher Enhancement - F UNDEF

Background

The current Brazilian system of ﬁscal federalism resulted from years of
institutional reforms that, for the most part, took place at the end of the authoritarian
regime (1964-1985) and during the transitional years to democracy. The decision-making
power of the federal government has become less concentrated, which resulted from a
gradual decentralization of revenues and public expenditures from the federal
government to the state and municipal governments (MEC, 1997). Even though federal
grants to state and local governments were common before 1995, the expenditures were
mainly dictated by federal regulations, leaving the autonomy of the other levels of
government arguable. In 1995, the return of the mandated compulsory allocation of a
share of the total tax revenues to the education sector with the Calmon Amendment
favored changes in how primary education would be ﬁnanced (Castro, 1998).

In the 19905, the collapse of central authority due to political crises, including the
apparent incapacity of the central government to guide educational policies and the
concerns about the efﬁciency of the existing allocation of resources, created increasing
demands for more decentralization. The Federal Constitution played an important role in
the country's effort to consolidate decentralization in the provision of education, despite
people’s mistrust in the capacity of state and local governments to invest efﬁciently in
education. The ﬁnance system lacked mechanisms to deal with equity disparities in the
structure of public ﬁnance, however. In the middle 19905, the federal government, under

the administration of Fernando Henrique Cardoso, assumed the role of compensating for

18

the high ﬁscal inequalities among municipalities and states in order to regain control over
the coordination of national policies. The result was a policy design that aims to
guarantee minimum school ﬁnancial equity in order to create conditions for
implementing such parallel reforms as national parameters for curriculum and
requirements for teacher’s qualiﬁcation levels.

The reform

As mentioned before, the Federal Constitution of 1988 initiated a period of reforms
in the Brazilian educational system. Another important reform was the revision of the Law
of Guidelines and Bases for National Education (LDB), approved in December 1996. This
reform, main policy guideline for Brazilian education, reinforces the responsibilities of
each educational system. It gives greater autonomy to Brazilian schools, including an
allowance for a more ﬂexible curriculum. In addition, it demands that teachers complete
higher levels of education. All of these changes were perceived as necessary in order to
improve Brazil’s quality of education.9

Because the implementation of these changes was still constrained by limited
capacity of the education ﬁnance system, a reform was necessary. This happened with the
Constitutional Amendment N. 14 approved in September of 1996. This amendment
created the Fund for Maintenance and Development of Primary Education and Teacher
Enhancement (FUNDEF), regulated by the Law 9, 424, December 1996. Legislative
process and politics were responsible for the late enactment of the LDB. In fact both
mandates were simultaneously under Congress’ consideration, so that both would act as

coherent policy instruments to support and reinforce each other.

 

9 For a summary of the Brazilian educational policies and reforms instituted in the Fernando Henrique
Cardoso administration see MEC (19991) Brazilian Education: Policies and Results.

19

The F UNDEF was supposed to be implemented throughout the country in January
1998, but states that wished to anticipate the process were able to start it in 1997 if the state
and federal district constitutions allowed it. By the end of 1996, the federal government
was providing ﬁnancial incentives to the states that would pioneer the implementation of
the F UNDEF.IO Only the state of Para anticipated the implementation of the FUNDEF in
1997. The federal governmentll proposed an educational ﬁnance reform that holds each
state and municipal government responsible for educating a number of students, which
previously determined the amount of funding for each state and municipal government.12
The FUNDEF strengthens the central role of the federal government. The standard of
quality is deﬁned centrally, mainly by a national minimum revenue per pupil.13 In order to
guarantee equal opportunities and the national minimum standard of quality education, all
resources are aggregated into a fund from which the federal government performs its
redistribution and/or complements the fund in order to guarantee a minimum revenue for
each pupil. Implementation is still the responsibility of the state and municipal
governments, which now have a great challenge to face Since providing education for all is

the means to maximizing their own educational revenues.

 

'0 The FUNDEF Operationalization guide mentions that additional resources would be provided by the
Ministry of Education mainly for improvement of school quality, including rebuilding, enlargement of
buildings, and equipment (MEC, 1997c).
” In Brazil, the federal government has the ﬁrst responsibility for Brazilian education policy, through the
actions of the Ministry of Education — MEC. The MEC is responsible for formulating and evaluating the
national policy, as well as for coordinating the actions of the different systems and levels of education.
'2 In 2000, 57 municipalities went to court to leave the FUNDEF, which distributes resources depending on
the number of students enrolled in the public systems in each municipality. On the contrary of Anapolis
(capital of the state of Goias), the 56 municipalities were in the process of losing resources with the
redistribution mechanisms of FUNDEF, as mentioned by the MEC. "The problem is not to win or lose
money, but to avoid that municipal revenues are managed by the Federal Government", commented the
rocurador-geral de Anépolis, Roldao Izael Cassimiro. Agéncia Estado, 05/01/2000, 15h 11min.
3 The minimum annual expenditure per pupil is deﬁned by law, based on the ratio between the estimated
total revenue from the FUNDEF for year i and the total enrollment of the previous year (H) in primary
education plus the estimated growth of the enrollment for the actual year 1'. See MEC (1997c).

20

Goals and Objectives

The changes to the ﬁnance system of primary education provide the mechanism by
which the federal government controls and perfonns its redistributive and complementary
role to assure equalization of educational opportunities for the entire school age
population and the illiterate, as well as minimum standards for quality of schooling.

These are the general goals of the reform. The objectives are to:

o Guarantee allocation of resources in primary education;

- Raise the bottom of the distribution of educational revenues to a national
minimum revenue per pupil;

0 Deﬁne the redistribution of responsibilities between state and municipal
school systems, based on their provision of schooling.‘4

The indirect objectives are to create a ﬁnance system capable of raising teacher
quality and guaranteeing minimum average teacher earnings.

Target Population

All states, municipalities and federal district are affected by the national policy.
Each state has its own school network (the state educational system) and practically each
municipality within the state has its own school network (the municipal educational
system) as well. Table 1 in the appendix lists the states and the number of municipalities
within each state, as well as each state’s total enrollment. In 1996, if the redistribution

mechanism had taken place, the number of school networks that would have received funds

 

H The Constitutional Amendment N. 14, passed in September 1996, changed the ﬁrst Paragrapht of Article
60 of the 1988 Federal Constitution and determined that the distribution of responsibilities and resources
between states and municipalities is assured by the institution of the FUNDEF. Before the amendment, the
public sphere had the responsibility for providing primary education together with the organized civil
sector through the use of at least 50% of the resources originated from taxes and governmental transfers.
The amendment specify that state, federal district and municipal governments have to invest not less than
60% of the revenues mentioned before (25% of taxes and governmental transfers) to guarantee universal
primary education and fair remuneration for teaching.

21

would have been 532, almost 10% of the total number of public school networks in the
country.” Once additional federal resources invested were minimal (roughly 4% of the
total), the increases in revenues for these 532 municipalities may have occurred by
transferring resources from other municipalities or state networks of schooling.

Operationalization

The fund. The instrument created by the ﬁnancial reform is, in fact, an accounting
procedure. It is a fund instituted in each state, municipality, and federal district, created as a
special account, whose resources can only be used for the maintenance and development of
primary education. The balance of this account can be invested, but all resources, including
those that originated from investments and the balances from previous years, must be spent
on primary education. The resources can also be used as counterparts in contracts to
ﬁnance projects and programs in primary education.

The resources. The ﬁrnd is composed of resources from the main state and
municipal taxes. From state taxes, it includes state sales taxes (ICMS and QP-IPI-EX),
income taxes (1P), and federal transfers to the state (FPE). From municipal taxes, the fund
is composed of municipal sales taxes (188), state transfers to local governments (QP-ICMS,
QP-IPI-EX), and federal transfers to local governments (FPM, ITR). The institution of the
fund does not free state and municipal governments from their obligation to spend 25
percent of the total revenues (tax revenues, inter- and intra-governmental transfers) in the
maintenance and development of education.

The redistribution mechanism. The redistribution of resources from the fund

between municipalities and state governments is performed in proportion to the number of

 

'5 Estimated by FACEM, Séo Paulo.

22

students enrolled in each school network.'6 When the value of the fund transferred per
pupil to the municipal or state government is below the annual minimum national value, the
federal government supplements these resources, so that within each state, municipal and
state govenunents will have the same minimum revenues per pupil. The objective is to
raise the low level of education expenditures to the minimum revenue per pupil necessary
to ﬁnance quality education. The transfers of resources are conducted monthly directly to
the municipal and state accounts linked to the fund account.

The value of national minimum revenues per pupil. The minimum national
revenues per pupil are deﬁned by presidential act and are based on the ratio between the
estimated total revenue from the FUNDEF for year i and the total enrollment of the
previous year (i-l) in primary education plus the estimated growth of the enrollment for the
actual year 1'. The total enrollment is based on the Educational Census, the information for
which is provided by the state, federal district, and municipal governments.‘7 For 1997, the
value estimated for the fund was $300 dollars per pupil (R$300,00 in Brazilian currency).'8

Implementation

The implementation of the policy begins with the deﬁnition of the criteria for
redistributing the resources, that is the value of national minimum revenues per pupil. State,
municipal, and federal district governments each create a speciﬁc account for the fund. The
resources are automatically transferred to this account, usually three times monthly. The
government budgetary procedures have to take fund into account, including and adapting

their Budget Plans. State, municipal and federal district governments also are required to

 

'6 The law considers only students in classroom.

'7 Rectiﬁcation of the number of enrollments cannot be done until after 30 days of the publication of the
Census results.

'8 In 1996, the exchange rate was US$1.00 to R5100.

23

institute boards that supervise and control the distribution, transfer, and use of resources.
Each government level must deﬁne career plans for the teaching profession that include
mechanisms by which the system 'will qualify its teachers and eliminate unqualiﬁed
teachers by 2003.

Within each state, state and municipal governments can reorganize their educational
systems by transferring students, schools, and human resources. The FUNDEF law also
establishes that education boards at the state, municipal, and school levels must be created
to exert social control over the use of these resources. Sixty percent of the monies from the
fund must be spent on teachers’ wages and training, and 40 percent on other school
expenditures. The share of 60 percent was deﬁned in accordance with previous a regulation
implemented in the country, the C amata Law, which stipulated that personnel expenditures
should not exceed this percentage. This regulation was inspired by studies that investigated
the percentage of public revenues spent on personnel expenditures for public services in
European countries (55 to 65 percent), as well as Japan, the United States, and Canada
(Dourado, 1999). The law was approved for all public services, despite the fact that in the
educational sector, most Brazilian municipalities were already spending more than 60
percent, sometimes 80 percent on personnel (Dourado, 1999). A study from UNESCO
(1991) indicated that, in developing countries, the average percentage of education
revenues Spent on teacher wages roughly varies from 80 to 90 percent. This suggests that

the 60 percent rule is not necessarily practicable in Brazil.

24

Reform theory and expected reform effects

The theory of reform describes the set of beliefs and underlying actions that
support what the policy states, what it expects to achieve, and also how it expects to
achieve it. The policy reform model is a construct of how the policy is supposed to work.

The F UNDEF is an accounting procedure that provides a mechanism for
accounting for the resources spent on primary education. Before the FUNDEF, it was very
difficult to verify how much was effectively spent on primary education versus the other
levels of education in the school systems. The F UNDEF increases resources to primary
education, since municipal and state systems, which spend less than the nationally deﬁned
minimum standard per pupil, receive additional resources from the federal government.
The increase of resources is justiﬁed on the grounds of the equalization principle and the
goal of ﬁnancing a quality education for all children.

The theory that links increasing resources and quality of education is represented in
Figure 5. Increasing resources for primary education through federal complementary funds
will raise the bottom of the distribution of educational revenues to national minimum
revenues per pupil. It is expected that the national minimum revenues per pupil will cover
the costs of quality schooling. Responsibility for primary education is decentralized to
state and municipal level, depending .on the number of students enrolled in each school
network. Each school network will have incentives to provide education for a larger
number of students, to maximize their revenues, and to attend to the demand for education.

Since 60 percent of F UNDEF resources must be spent on teachers, state and
municipal school systems will be able to deﬁne teacher policies, which enhance the quality

of education, and consequently will attract more students. These policies may include

using resources to qualify unskilled teachers and to guarantee minimum average salaries or
even increase salaries, which will help retain experienced teachers without the additional
costs of hiring new skilled teachers. The development of training and qualiﬁcation
programs is expected to reduce teacher tumover.

Increased salaries may induce teachers to give up their second jobs. Teachers who
used to moonlight in order to earn enough money may now put their full energies into
teaching. Greater energy and attention lead to more thoroughly prepared lessons, greater
variety in pedagogical strategies, more effective teaching, better student learning, and thus
better quality of schooling.

Another assumption of the reform is that the minimum per pupil revenues
established annually by the federal government will in fact be sufficient to ﬁnance
increases in teacher wages, and thus, the school networks will be able to attract more
qualiﬁed individuals into teaching. It may also be the case that school systems offering
higher salaries will lure good teachers away from other school systems. Competition
between two systems may lead to increased quality of schooling, better management of
education through cooperation, and more shared responsibilities among the systems. But
the very least, school systems offering higher salaries may induce current teachers to stay
on the job longer, thus preventing vacancies from opening up. This may also enhance the

quality of schooling.

26

Figure 5: Reform Theory

 

ﬂaw quality of education due to unequal and low per pupil revenuesl

New primary education ﬁnance implemented
FUNDEF

Within state redistribution of revenues implemented
and
complementary federal educational revenues transfered

[Minimum per pupil revenues guaranteed |

School quality standards are ﬁnanced I

i

Decentralization of education systems

 

 

 

 

 

 

 

 

based on the supply of education

State System Municipal System

 

 

 

 

l

 

 

 

 

 

 

 

 

 

 

 

I
jOther School Characteristich Teacher Policy J Teacher Policy l jOther School Characteristics
1 l
l W 1
Resources for teacher Resources for ﬁnancing Resources for ﬁnancing Resources for teacher
training and qualiﬁcation minimum average salary minimum average salary training and qualiﬁcation
are allocated are allocated are allocated are allocated

 

 

 

 

 

 

 

 

 

 

 

   
   

 

   
  

Average teacher wages Reserve of Skilled Workers Average teacher wages Unskilled teachers
are increased are increased are trained

Higher qualified workers _/_ :\ Higher qualified workers Teachers give up Trained teachers

4 second jobs stay in the system

attracted into teaching attracted into teaching
Municipal school quality
enhanced

Unskilled teachers
are trained

Trained teachers Teachers give up

stay in the system second jobs

State school quality
enhanced

      

 

 

 

 

        

   

 

 

 

Quality of Education

enhanced

 

 

 

 

Immediate results

In 1996, per pupil revenues ranged from 63 to 1,696 reals, across states and
municipal school networks. That is, a municipal school in the state of Roraima invested
27 times more per annum than a municipal school in the state of Maranhao. Even in the
single state of Roraima, a student enrolled in a state school received roughly 1/3 less in
terms of public revenues invested than a student attending classes in a municipal school.
In the North and Northeast regions, with the exception of the state of Roraima, state
schools receive higher per pupil annual revenues. In the more developed regions of the
Southeast, South, and Center-West, municipal schools spend Signiﬁcantly more than state
schooling systems.

Table 2 shows the differences between state and municipal per pupil revenues in
1996 (columns A and B) and the per pupil revenues in 1998 under the FUNDEF reform
(columns E and H). It also shows the predicted per pupil revenues in 1998 had the
reform not been implemented.'9 The comparison between the actual and hypothetical
amounts of how much would have been the per pupil revenues by school network within
the Brazilian states if F UNDEF had not been approved (columns D and G) help us to
observe the percentage change in per pupil revenues. School networks had to administer
(columns F and I) the re-distributive and compensatory allocation of resources by the
FUNDEF reform in 1998. This comparison shows impressive differences in revenues

between municipal and state networks of schools. Indeed, in 20 out of the 26 Brazilian

 

'9 Predicted revenues were calculated using data from STN/SIAF I with the same methodology of education
ﬁnance before the introduction of the F UNDEF reform. These figures include only the tax revenues that
were considered by the FUNDEF reform to allow for comparisons between before and after the reform.
Speciﬁc local tax revenues which may also be included in the per pupil revenues were not considered in
this estimate and represent less than 3% of the total amount, as estimated in MEC (1997b).

28

 

states, the average state school revenues per pupil would have been much higher than in
municipal schools if the reform had not taken place.

The marked differences between school networks suggest that, within the same
town, an individual with access to a public state school is likely to have a better quality
education than an individual enrolled in a municipal school in these 20 Brazilian states.

Because of the FUNDEF, now there are virtually no differences in revenues per
pupil within each state based on the main taxes allocated to primary education. The
FUNDEF thus corrected the within state disparities among public schools. There was an
average increase of 108.7 percent in per pupil revenues in the 20 states where municipal
schools would have had fewer resources. In the remaining six states, located mainly in
the most developed regions of Brazil, there was a contrasting decrease in municipal
school revenues.

In 1998, state school revenues per pupil in these six states would have been, on
average, 36.9 percent (SD=25.8) lower than those in the municipal school networks.
Differences among states and regions still exist in spite of the reform, but the disparities
in levels of educational revenues between school networks within states are being
considerably minimized20 (See Table 2, columns E and H).

The changes in per pupil revenues before and aﬁer the F UNDEF reform by school
networks within Brazilian states are composed of two main effects. The ﬁrst is the
redistribution effect within a state, where 15 percent of main state and municipal tax
revenues are aggregated and then returned to municipal and state educational

administrations in a manner proportional to each school network’s enrollment.

 

20 It is important to remember that there are still other taxes (e.g., local taxes) that are not included in the
FUNDEF; therefore, wealthier localities may still have higher per pupil revenues despite the reform.

ll
'11
11..

Table 2: Per Pupil Revenues in Primary Education by School Network, 1996-1998.

 

 

 

 

 

 

 

 

 

 

 

1996 1998
State Networks Municipal Networks
BT32“ and States State Municipal Difference Before After Change Before After Change
(%) FUNDEF FUNDEF (%) FUNDEF FUNDEF (%)
(A) (B) (C) (D) (E) (F) (G) (H) (I)

North 437 440 -0.8 409 369 -9.8 251 369 47.0
RondOnia 301 148 50.9 438 388 -11.4 289 388 34.3
Acre 639 241 62.3 754 607 -19.5 304 607 99.7
Amazonas 386 193 50.0 488 425 —12.9 319 425 33.2
Roraima 624 1696 —1 71.8 810 901 11.2 2,986 901 -69.8
Para 192 110 42.8 269 315 17.1 184 315 71.2
Amapa 584 448 23.4 767 690 -10.0 595 690 16.0
Tocantins 333 247 25.7 421 383 -9.1 309 383 23.9
Northeast 387 150 61.3 427 321 -24.8 170 321 88.8
Maranhzio 312 63 79.8 385 315 -18.2 101 315 211.9
Piaui 328 96 70.8 437 315 -27.9 159 315 98.1
Ceara 435 139 68.0 509 315 -38.1 152 315 107.2
Rio Grande do Norte 360 185 48.5 439 346 -21.2 245 346 41.2
Paraiba 376 209 44.4 485 325 -33.0 220 325 47.7
Pemambuco 350 165 52.7 422 315 -25.4 201 315 56.7
Alagoas 558 142 74.5 830 336 -59.5 151 336 122.5
Sergipe 453 187 58.8 529 395 -25.3 231 395 71.0
Bahia 313 162 48.3 350 315 -10.0 183 315 72.1
Southeast 561 667 -18.9 529 550 4.0 602 550 -8.6
Minas Gerais 271 461 -69.8 296 354 19.6 515 354 -3I.3
Espirito Santo 463 592 -27.9 448 463 3.3 496 463 -6. 7
Rio de Janeiro 1013 220 78.3 1,262 619 -51.0 270 619 129.3
850 Paulo 497 1397 -I8l.1 568 657 15.7 1039 657 -36.8
South 447 392 - 12.3 544 482 -I 1.4 407 482 18.4
Parana 428 269 37.0 499 418 -16.2 328 418 27.4
Santa Catarina 386 494 -28.2 620 561 -9.5 475 561 18.1
Rio Grande do Sul 526 411 21.8 486 477 -l.9 460 477 3.7
Center-West 366 331 9.6 373 371 -0.5 370 371 0.3
Mato Grosso do Sul 399 291 27.0 410 366 -l(). 7 306 366 19.6
Mato Grosso 405 367 9.2 445 421 -5.4 379 421 11.1
Goiés 296 335 -13.2 328 346 5.5 382 346 -9.4
Distrito Federal

 

 

 

Source: MEC (l997d). Per pupil revenues in 1996 estimated using ﬁnance data from the MF/STN and
enrollment data from MEC (l997c). (Nominal Per Pupil Revenues in Reals —Brazilian Currency).

30

This redistribution results in equal per pupil revenues within states. The second is
the complementary role of federal transfers to guarantee a national minimum standard for
expenditures per pupil. The federal government deﬁnes a national minimum per pupil
value depending on its capacity for investment in education. After redistribution, if the
total per pupil revenues within a state is below this national per pupil value, federal
transfers complement the resources to guarantee the national minimum. The national
minimum amount during the reform implementation year was R83 15,00 (real, Brazilian
currency). This was only worth about US$190.00 per pupil in 1998.

Verhine (1998) comments that the Brazilian government was unable to maintain
the originally deﬁned yearly value of US$300.00 per pupil due to tightly mandated
government spending restrictions from the International Monetary Fund (IMF). The lack
of mechanisms to protect the national minimum value from the instability of the economy
and the devaluation of currency may be one of the most ominous aspects of the new
ﬁnance design because it may continuously decrease the purchasing power of the public
educational management system. Monlevade and Ferreira (1997) point out that the
national minimum is very low and there is no guarantee that the amount will increase in
accordance with rising costs.

During 1998, eight states located in the poorest regions of Brazil, the North and
Northeast, were unable to achieve this revenue per pupil aﬁer the FUNDEF
redistribution. Consequently, they received federal complementary resources (MEC,
1999). A consolidated report from Secretaria do Tesouro Naeional (National Treasury
Secretary) shows that the total federal transfers were 424.4 million reals, representing

only 3.2 percent of the total revenues invested in primary education under the FUNDEF

31

reform. The same report indicates that six states would receive positive adjustments from
federal complementary funds in 1999 because their 1998 per pupil average was still
below the national minimum despite FUNDEF transfers (See Table 1A in Annex).

The effects of the F UNDEF have varied greatly by state. In the state of Roraima,
for example, the impact of FUNDEF is remarkable. It represents resource transfers from
municipal to state schools on the order of 70 percent. On the other hand, the highest loss in
revenues from state to municipal schools took place in the state of Alagoas. Municipal
schools in this state now have 123 percent more revenues than they would have had
without FUNDEF. Because of: a) the drastic changes in revenues, and b) the compulsory
allocation of part of these per pupil revenues to teacher training and wages some changes in

teacher supply and wages may be noticeable after only three years of the F UNDEF reform.

32

CHAPTER 2

THE EFFECTS OF THE REFORM ON TEACHERS’ EARNINGS

Introduction

The total ﬁnancial resources that a country chooses to invest in education is a
critical decision that raises many issues for debate, for example, whether and to what
extent education costs in a speciﬁc country guarantees quality education for all. Among
the costs of education, teachers’ salaries account for most of the public expenditure on
education (Psacharoupolous, 1987, 1996), which makes teacher earnings an important
policy question especially in countries under economic constraints, such as developing
countries. Teachers’ earnings are also important because the high share of education
expenditures spent on teachers may be related to education quality. Economic analysis
suggests that relative costs of several inputs should be proportional to the marginal
contribution of each separate input to the productivity. If schools are operating
efﬁciently and are paying teachers relative to their productivity in teaching, teachers’
earnings can also be used as a proxy for quality (Harbison & Hanushek, 1992). Research
has been investigating the relationship among low quality of education and low quality of
teachers, which in its turn has been related to the low pay of teachersz' Finally, salary
policies are commonly used by policymakers as a way of improving schools; by

increasing salaries education ofﬁcials and the public expect to attract and retain quality

 

2' Hanushek, Kain and Rivkin (1998) suggest that at least 7% of the variance in student achievement scores
may be explained by variation in teacher quality. Mumane, Willett and Levy (1995) have shown that
individual test scores is positively associated with teacher wages, even though studies of the relationship
between teacher wages and student outcomes has produced mixed results. In Brazil, however, Harbison and
Hanushek (1992) found a positive and signiﬁcant relationship between teacher salaries and student
achievement in schools in the rural Northeast.(p.109). They observed that teacher wages were very low in
rural areas, representing 60% of the minimum wage (p.108).

33

teachers. Especially in what concems beginning teachers, as the education sector
competes with other sectors for additional personnel to cope with growing enrollment and
an aging work force of experienced teachers, increased salaries potentially provide the
means of attracting and retaining the increased number of qualiﬁed young teachers who
will be needed in the years ahead.

As mentioned in the ﬁrst chapter of this dissertation, during the late 19905,
Brazilian education reform addressed these issues by deﬁning a new education ﬁnance
mechanism called the Fund of Maintenance and Development of Primary Education and
Teacher Enhancement (FUNDEF), enacted in 1996.22 The main objective of this
nationwide fund is to guarantee minimum revenues per pupil in primary education all
over Brazil such that, regardless of the type of public school attended, minimum
resources are guaranteed. There is a minimum standard of ﬁnancing aimed to provide
certain schooling opportunities for all Brazilians in primary education. The total annual
revenues committed by the federal government when establishing the national minimum
revenues per pupil23 deﬁnes the bottom of the quality distribution of primary education
schooling in the country. The policy instrument also addresses salary policies. In order
to attract and maintain teachers with higher educational levels, 60 percent of per pupil
revenues are compulsorily allocated to cover the training and wage expenditures of
teachers. The magnitude of the changes in per pupil revenues among school networks

within states is impressive in many states of the country, which may be reﬂected in

 

22 The Fund of Maintenance and Development of Primary Education and Teacher Enhancement (FUNDEF)
was formally created by the Constitutional Amendment N.14, on September 12'“, 1996. It was
subsequently regulated by the Federal Law N. 9424 (December 24th, 1996) and the Executive Decree N.
2264, in which the FUNDEF implementation phase is determined to begin in January 1“, 1998.

23 When the Fund was originally created in 1996, the national minimum revenues per pupil was deﬁned at
R$300.00, corresponding to US$300.00 per pupil year (Law N. 9,424, December 24, 1996).

34

 

changes in teachers’ earnings (See Chapter 1, Immediate Results in page 39). For 20 out
of 26 Brazilian states, there was an average increase of 108.7 percent in the per pupil
revenues in municipal schools. In the other six states, located in the most developed
regions, however, there was a 36.9 percent decrease in the per pupil revenues in
municipal schools.

This chapter analyzes the effects of the F UNDEF reform on the distribution of
primary teacher earnings in Brazil in state and municipal networks of public schooling.
This analysis is based on data from 1995 to 1999. It reveals important information about
teachers’ earnings and the differentiated reform effects by regions of the country. The
next section describes the data and empirical Speciﬁcation used in this analysis, followed

by a section with the results and concluding remarks.

Data and empirical speciﬁcation

Data

The Ministry of Education (MEC), the Brazilian Institute of Geography and
Statistics (IBGE), and the National Treasury Secretary (STN) responsible for public
accounts provided information for this analysis. 1 have constructed a data set by merging
individual records from all Brazilian households sampled in the National Research for
Sample of Households (PNAD-IBGE) from 1995 to 1999, with the accountability reports
from STN and The Educational Census data from MEC. By deﬁning an identiﬁcation
number for each data set consisting of year, state, and type of administration, it was

possible to merge all three ﬁles into the analysis data set.

35

The National Research for Sample of Households (PNAD-IBGE). This survey is
a random sample of Brazilian households with individual data on 1,659,403 people of
whom some are primary teachers in public or private schools during the years of 1995 to
1999 (See Table 2A, in the Appendix).24 For every individual in the sample, there is
demographic, economic, family and work information, among others. From this total
sample, I restrict the analysis to Brazilian workers who: 1) were employed in the year of
reference, 2) were at least 15 or less than 64 years old, 3) were in the non-agriculture
sector, and 4) worked during the day (SAM-10PM). A total of 302,172 observations
satisﬁed these criteria. They were utilized in the analysis of earnings.

The Educational Finance Data (5 TN). Data on educational revenues were
provided from the STN databank and yearly reports. They are basically yearly budgetary
balance reports of the federal, state, and municipal public accounts. Particularly, the data
on the FUNDEF were provided at the state level and by school administration. This data
was used to estimate the per pupil revenues used in the estimation.

The Enrollment Data (MEC). lnforrnation for this analysis was obtained from the
Ministry of Education. Speciﬁcally, the National Institute of Pedagogic Research (INEP)
is legally responsible for disseminating all educational data. It includes information in all
levels of education and from all school networks. Data on total primary school
enrollment from the Educational Census of 1995 to 1999 were used.

Variables used in the estimation. Table 3A in the Annex describes the variables
used in the estimation and reports means and standard deviations of the variables by
occupation over the ﬁve years studied. The proportion of primary teachers in the sample

varies from 0.045 in 1993 to 0.05 in 1998. The majority of primary teachers is female.

 

2" There was no PNAD research in 1994 (“Notas Metodologicas”, PNAD/IBGE, 1998).

36

Females represent around 91-93 percent of the primary teacher population across the
years, whereas female participation in non-teaching occupations in this sample
corresponds to 31-35 percent. The majority of the workforce—teachers and non-
teachers—is white. Minority groups (self-declared black, yellow and indigenous people)
represent 6 percent of non-teachers and 4 percent of teachers. In this sample, the
majority of teachers and non-teachers live in urban areas (around 91-93 percent for non-
teachers and 85-86 percent of teachers) where economic activities are concentrated.25
Across all years, around 71-83 percent of teachers are civil servants, whereas only 23-24
percent of the non-primary teachers work for the public administration. Among teachers
in the public sector, the sample varies from 41-47 percent of state primary teachers and
16-19 percent of municipal teachers. This is expected since the majority of state schools
is concentrated in urban areas.

Years of education and monthly earnings. Average levels of education for
teachers are higher than those of non-teachers and this disparity has grown over the years.
The average teacher, in 1999, had 13 years of education (complete secondary education),
whereas nOn-teachers on average only concluded 9 years of education (complete primary
education). In 1999, without controlling other factors, average monthly earnings for non-
teachers surpassed earnings of primary teachers, even though non-teachers had on
average ﬁve years less of education them their teacher-counterparts. In 1995, the average
gap in earnings between teachers and non-teachers was 28 percent, but in 1999, this

disparity was reduced to 15 percent. The exception is for teachers with higher education.

 

25 While the proportion of teachers living in urban areas is a good estimate of the entire population, the
proportion of non-teachers does not include individuals working in the agriculture sector. For this reason
this proportion may be upward biased.

37

This group of teachers receives monthly earnings superior to those of non-teachers (See
Figure 6).
Figure 6: Average Monthly Earnings by Occupation, 1995-1999

(Brazilian Currency)

 

 

 

 

 

 

 

 

 

Average Monthly Earnings

 

 

 

 

1 995 1 996 1 997 1 998 1999

 

i+All primary teaCher; I -________,.
i + Primary teachers with less than 8 years of completed education"
1+ Primary teachers with 8 years of completed education

;+ Primary teachers with 12 years of completed education

are Primary teachers with more than 15 years of completed education;
-0— Non-primary teachers

 

 

 

L.” ___- ,1”. __.1

Source: Estimates using PNAD/IBGE datasets from 1995 to 1999. Sample used in this analysis
(Nominal values in Brazilian Currency).

" Wages for primary teachers with less than 8 years of completed education do not include
individuals with less than one year of education.

 

 

Teacher monthly earnings have been increasing slightly over the years. Barreto
(1990), using data from 1982 to study primary teachers’ wages proﬁles, reported national
monthly average earnings for primary teacher of USS l 83. Average monthly earnings

ranged from US$35-8249, depending on the location of teachers’ residences—urban or

38

rural—and geographic region. In 1999, the PNAD data show that the average monthly
earnings for a teacher were US$254

In 1995, monthly earnings for primary teachers with 12 years of education were
R8250 (SD=166). For non-teachers with the same education level, monthly earnings
were R8517 (SD=539). By 1999, there had been a 39 percent increase for teachers.
Average predicted monthly earnings were R8347 (SD=218). For non-teachers with 12
years of education there was only an 8.7 percent increase in monthly earnings (average
predicted value is R8562 (SD=621). That is, a teacher with complete secondary
education still earns less than the average worker in other occupations, assuming the
same education level, but the earnings differential between primary teachers and all other

workers has diminished.

Empirical speciﬁcation

The methodology applied is a quasi-experiment. The F UNDEF reform, an
exogenous event, changes the per pupil revenues in public school networks in Brazil.
The objective of this analysis is to compare the mean wage difference between primary
teachers and individuals in other occupations before and after the reform. Therefore, the
treatment group includes primary teachers in Brazil, who ought to be affected by the
reform. The control group comprises Brazilians in non-teaching occupations. The
comparison group of workers in other occupations than teaching may not be a proper
control group because members in other professions may systematically differ from

teachers on human capital and other personal characteristics.26 However, despite this

 

26 Even controlling for these characteristics, as Psacharopoulos (1987) points out, the reference salary in the
control group profession may itself be subject to rents.

39

methodological issue, this approach has been commonly used in the teachers’ earnings
literature, because of the importance of considering changes in relative rather than
absolute teachers’ pay.27 The speciﬁcation applied in this analysis captures the impact of
the reform considering how teachers’ wages have fared in comparison to other
occupations. In order to control for systematic differences between control and treatment
groups overtime, l utilize data from ﬁve years: three years28 before reform
implementation and two years (1998 and 1999) after it. Additionally, I check for the
robustness of the results by comparing the treatment group of primary teachers with a
control group comprising teachers in pre-school and secondary education, because this
last group of teachers may be more homogeneous than a control group of individuals in
non-teaching occupations. For a graphical representation of the estimation methodology,
see Appendix C. Interpretation of parameters under the difference-in-difference
framework (DD) and difference-in-difference-in-difference framework (DDD).

The equation of interest is given as:

ln(wage),- = 50 + 50y95i + 51y97i + 52y98i + 53,1299, + Blprimtchi + 54y95 * primtchi

m l
+ 55y97 * primtchi + 56y98* primtchi + 57y99* primtchi + Z Xiij + ui ( )

3:2
where ln(wage) is the natural logarithm of individual wage rate. Parameter 00 measures
the average log wage of non-teachers in 1996. The model also includes year dummy
variables, y95, y97, y98, and y99, which equal one for their corresponding year and zero,
otherwise. There is no dummy variable for 1996, the year of the reform enactment, Since

it is the base year deﬁned for this analysis. These year dummy variables capture year

 

27 See Liang (1999), Piras and Savedoff(l998), Psacharopoulos (1996), Bee (1995), Komenan (1990), for
example.
2‘ The years were 1995, 1996, and 1997.

40

speciﬁc variation in wages. This takes into consideration the fact that the distribution of
such earnings may differ in different time periods. The parameters on these year
dummies, 60, 61, 52, 53 capture changes in each specific year’s earnings for all
workers compared to those of 1996, the base year. The dummy variable, primtch,
indicates ifthe individual has teaching as his/her main occupation, assuming value one if
the individual is a primary teacher and zero otherwise. The coefﬁcient on primtch, ,61 ,
measures the difference in earnings between primary teachers and non-teachers that is not
due to the implementation of the reform.

Interaction terms between year dummies and primtch are also included in the
model. Immediately in 1997, there was the opportunity for some school networks to
adapt themselves to manage the predicted gains and losses resulting from the 1996

reform before the 1998 implementation year. The model captures this possible effect

through the parameter, 55 , on the interaction term between y97 and primtch,

y97*primtch.29 The parameter 55 captures whether or not changes in primary teacher
earnings were already taking place in 1997.

The impact of the reform on teachers’ wages is estimated by the interaction terms,
y98 *primtch, —which equals one for all primary teachers in 1998 and zero, otherwise—,

and y99*primtch, —which equals one for all primary teachers in 1999 and zero,

otherwise. The main parameters of interest are 55 on y98*primtch, and 57 on

 

29 In fact, one state of Brazil, the state of Para located at the North Region, initiated the new ﬁnance system
in 1997, since the F UNDEF law granted the choice to any state to anticipate the reform implementation for
1997. The state of Para received from July to December 1997, federal complementary transfers of 3.8
million reals. (STN/SIAFI, 1998. Relatorio do FUNDEF). Federal Law N. 9424 (December 24, 1996). In a
separate model, I also included the dummy variable PA, indicating if individuals are from the state of Para,
and interactions to investigate if anticipating the reform implementation had promoted any interesting
effect on teacher earnings.

41

y99 *primtch. They capture the interaction effect ofbeing a primary teacher during the
ﬁrst and second year of the reform implementation, 1998 and 1999. They measure the
percent difference in monthly earnings between primary teachers and non-teachers due to
the reform in each respective year, provided that changes in earnings due to other factors

are constant for teachers and non-teachers.

In addition the model also incorporates a vector of control variables, X to take

1’ j -
into consideration human capital and job speciﬁc characteristics that are relevant to
workers’ productive attributes and earning differentials. The workers’ education
characteristics are incorporated into the model using year dummy variables for each year
of completed education (educI yr to educ] 5 yr) to capture non-linear effects of education
on earnings.30 The variables age, ageZ, age3 and age4 are also included to capture non-
linear productivity effects of skills, levels of ability, and other types of experience
acquired throughout the life cycle. Tenure and tenure2 are years of experience in the
current job and years of experience in the current job squared, respectively. The
quadratic form for tenure is also included to capture non-linear effects of tenure on
earnings. Even though tenure is a potentially endogenous variable —since it is likely to
be related to unobserved individual and job characteristics that affect the wage—, this
variable is included in the speciﬁcation because economic literature has shown that the

wage-tenure proﬁle is a key determinant of the extent to which individual earnings are

tied to speciﬁc jobs within different occupations} '

 

30 For detailed discussion on the wage-education distribution in Brazil, see Strauss, J. and Thomas, D.
(1996). The authors found out that the wage-education functions are not linear and not even high order
polynomials, and the convexity of the function may be reﬂection of a positive correlation between the
quantity and quality of schooling (p.148). For evidence on the impact of school quality see Behrman and
Birdsall (1983) and Berhman, Birdsall and Kaplan (1996).

3' For the wage-tenure discussion, see Hamermesh (1984), Antonji and Shakotko (1987).

42

There are still other factors that may affect earnings and by which teachers and
non-teachers may differ systematically over the years. These other factors are
incorporated into the model in the form of binary information: female, white, black,
yellow, indigenous, urban, municipal, and state. The dummy variable, female, equals one
if the individual is female and zero if male. The parameter on female represents the
percentage difference in earnings for females versus males. It determines if there are any

forms of gender discrimination by examining whether or not, on average, women earn
less (ﬁfemale < 0) or more (ﬂfemale > 0) than men, holding other factors constant.

In PNAD research, race and ethnicity characteristics are captured by the variable
color. The individual chooses one of the following identities on the survey questionnaire:
white, black, yellow, mixed, Indigenous, or ignored. Yellow refers to Asians or persons
with Asian descendants. Indigenous refers to native Brazilians or Indigenous groups.
White equals one if the individual declared being white and zero otherwise; black equals
one if the individual declared being black and zero, otherwise; yellow equals one if the
individual declared being yellow and zero, otherwise; indigen equals one if the individual
declared being Indigenous and zero, otherwise. The reference group chosen is mixed,
which represents roughly 36 to 38 percent of the total sample population over the years.
Urban describes an individual who lives in urban areas (urban=l); and zero, otherwise.

Furthermore, I discussed in the ﬁrst part of this paper how the reform
redistribution effect has differently affected state and municipal networks from different
states in the country. In order to control for these state differences, I included state

dummy variables.

43

In equation (1), the speciﬁcation assumes that the reform is the unique source of
variation on workers’ earnings during the year 1998 and 1999. However, it is likely that
the FUNDEF reform promoted changes in state and municipal educational policies which
may have also affected teachers’ earnings. In this case, to control for these possible
differentiated effects, I estimate the following equation using the difference-in-

difference-in-differences estimation (DDD):

ln(wage),~ = ,60 +60y95i +51y97i +52y98i +53y99i +Blprimtch +y1municipal +
+54y95*primtch +55y97’l‘primtch +56y98*primtch +57y99*primtch

+ y 2 y95* municipal + y3y9 7* municipal + y4y98* municipal + 75y99* municipal (2)
+ 76 * primtch‘" municipal + y7y95* primtclr’" municipal + 78 y9 7* prinztclf" municipal

m
+79y98*primtch* municipal +710.1'99*primtch*municipal + ZXiij +ui
1:2

The third-level interactions among the year dummies, the primary teacher dummy
variable and the administration dummy variables, particularly, state and municipal
dummy variables, allows for differences on the impact of the reform on teachers’ wages
depending on school network. The variable municipal equal to one indicates that a
worker serves under the municipal administration, otherwise it is zero. Teachers working
for municipal school networks are identiﬁed through this variable. If the worker is ruled
by state administration laws, the dummy, state, equals one. Otherwise, it is zero. This

variable identiﬁes if a teacher works for the state network of schools.

In this speciﬁcation”, the parameters of interest are y.) on

y98 *primtch *municipal, and 110 on y99 *primtch *municz‘pal. These parameters capture

the interaction effect of being a primary teacher working for a municipal network of

schools compared to primary teachers working for the state network of schools in the ﬁrst

 

32 The same specification was used replacing the municipal dummy variable by the state dummy variables.

44

and second year of the reform implementation. It measures the percent difference in
monthly wage rates between municipal and state primary teachers relative to the percent
difference in monthly earnings between municipal and state non-teachers in 1998,
compared to those teachers in 1996, due to the reform, provided that changes in earnings
due to other factors are constant for teachers and non-teachers. This parameter is

represented by the equation:

I l“(wagelteacher/ municipal" movagéteached state I98 "

[ ln(wagénon—teacher/ municipal— ]n(waganon—teacher/ state l98 ' (3)
79 =1 t
[ ln(wag€7teached municipal— ln(wag€)teacher/ state 196 '

 

 

KI 1n(WagC2non—teached municipal— ln(wag€lnon—teacheil state ]96 J

Some other variations of the speciﬁcations in equation 1 and equation 2 were also
deﬁned to investigate if there were differential reform effects by education group, tenure
year, and age group. First, as discussed in reform documents”, the reform Should beneﬁt
mainly qualiﬁed teachers identiﬁed by their level of education. In that case, wage
increases Should be observed for teachers with complete secondary education or more.
Teachers’ education level is acting as a proxy for quality. The relatively higher wages for
more qualiﬁed teachers should be proportional to their higher productivity in teaching,
even though institutional wage schedules are usually deﬁned considering also other

factors instead of actual teaching performance, such as experience, in-service teacher

training, and education level itself. Several studies have claimed that the low quality of

 

’3 Federal Law N. 9424 (December 24‘”, 1995).

45

teacher force also points to the relative low pay of teachers in Brazil, in other developing
countries, or in developed countries.34

Second, ifthe reform aims at attracting new and better people into the teaching
profession, and the means to achieve this is to promote more attractive wages for new
teachers, the reform effect should be more noticeable for newly hired teachers. The
equation 2 with the third interaction term between year dummies, primtch and tenure
years captures whether the refonn brought signiﬁcant differences in wages for newly
hired teachers than they would have received in the absence of the reform. The third
interaction term with age groups may explain if there are different effects depending on
the teacher age cohort.

I estimate equation (1), (2) and variations of these equations using pooled-
ordinary least squares and ﬁxed effects estimator. If the unobserved individual errors are

not correlated with all control variables (X i j ), OLS gives consistent estimates. Serial

correlation over time is expected, however. This problem would generate wrong standard
errors. To control for heteroskedasticity due to changes over time, I estimate the robust
standard errors. On the other hand if mean levels of unobserved earnings incentives vary

across states, and this variation is due to unobserved career incentives and policies which
are deﬁned at state level, then u,- and X1 j are correlated and OLS estimates are

inconsistent. I apply the state ﬁxed effects estimator to separate idiosyncratic unobserved
error u,- from the aggregate unobserved errors at state level when including state dummy

variables.

 

3" See Lockhead and Verspoor (1991), Harbison and Hanishek (1992) comments that salary policies are
often used as a way of improving schools; by increasing salaries policymakers expect to attract new and
better people into teaching (p. 108).

46

AS some may argue, it is possible that there are many unobservable characteristics
of non-teachers, including self-selection problems that would yield biased estimation.
Comparisons of primary teachers with a control group formed by teachers in other levels
of education are possible; however this comparison may still be problematic. The
problem of comparing primary teachers with pre-school and secondary teachers, who
should not be affected by the reform, arises from the fact that the F UNDEF reform is an
incentive to increase primary school enrollment which may reduce investments in other
levels of education. This may affect the demand for teachers in pre-school and secondary
education, consequently affecting teacher earnings in these levels of education. It is
interesting to observe whether or not there is a negative impact on teachers’ wages in
other levels of education.

The last model included the variable perpupil, indicating the value of per pupil
revenues in each school network per year and interactions. This is a direct measure of the
reform effect used here to assess how robust the results of the analysis are. In this case,
the sample includes only teachers from state and municipal schools which data on per
pupil revenues were available. The Federal District was also excluded because it only has
one network of schools (municipal network) so state and municipal comparisons
wouldn’t be possible. This robustness check is presented in the end of the chapter.
Results

Determinants of earnings

The estimation results in Table 4A (in appendix) are consistent with the economic
literature on determinants of earnings35 and these determinants are also signiﬁcantly

important to predict changes in relative teacher earnings. The estimated effects show the

 

35 Mincer (1985), Becker(l964).

47

direction expected. For example, the effect of schooling on earnings is positive and
signiﬁcant for all years of education dummies. Consistent with the results of Strauss &
Thomas (1996), when estimating log-wage functions by gender and region for Brazilian
workers using data from 1982, the conditional log-wage functions for all workers are not
linear. They capture important differences in returns across the education distribution.
Wage rates of workers with complete primary education (8 years of completed education)
are 48 percent superior to those of workers with less than one completed year of
education. If the worker has secondary education (1 1 years of completed education), the
wage differential between his/her wage and workers with no education is 85 percent. If
he/She has complete higher education or more (15 or more years of education), his/her
salary is 165 percent greater than that of workers with less than one year of completed
education.

As noted in previous research”, the effects of education on earnings increase
exponentially beyond primary school and these results have important implications for
income inequality in Brazil. In 1996, for example, there is an 80 percent difference
between the average wage rates for workers with higher education and those for workers
with complete secondary education, which is statistically signiﬁcant at the 1 percent
level. This outcome is relatively similar across the country, but the largest difference
found is in the Northeast. In that region, we observe wage differential of 88 percent for
workers with higher education. Earnings also increase with experience throughout the

country. This is indicated by the positive and signiﬁcant proxy variable for human

 

3" Strauss and Thomas (1996) pointed out that the returns of education have increased for recent cohorts.
(p.175), Lam (1999) analyses links between schooling inequalities and earnings inequalities in Brazil and
South Africa.

48

capital: age. As expected, years of work experience in the main job (tenure) also have a
positive and signiﬁcant effect throughout all regions of the country.

A number of demographic variables are signiﬁcant across regions, as well as
nationwide (See Table 4A). Female workers earn approximately 29 percent less on
average than males when all other factors are held constant. In the North region, this
differential is lower than in all other regions at 24 percent. Conversely, in the Northeast
region, one of the poorest regions of the country, the difference is 32 percent.

Nationwide and in all regions, these differences are statistically signiﬁcant at the 1
percent level. Although the North region also lags behind in terms of development, the
demand for labor is high because the relative demographic density is very low. This may
explain why gender-wage differentials are lower in this region. In the developed regions,
however, there are still signs of discrimination against women. The gender-wage
differential is around 29 percent and statistically signiﬁcant at the 1 percent level. There
are also signs of race discrimination. White individuals earn around 10 percent more
than members of other groups across the nation and its regions. Differences in wages by
sector also exist. Individuals working for the federal government receive on average 38
percent more than workers in the private sector. If an individual works for the state
government, on average, he/she receives 6 percent more than an individual in the private
Sector. In contrast, municipal employees receive 7 percent less than workers in the private
Sector. These wage-sector differentials also differ by region in the country. As expected,
i n the more developed Southeast, the wage-differential between the federal government
and private sector decreases to 3 percent. State and municipal employees, however,

I‘eceive 0.6 and 9.5 percent less, respectively, than those who work in the private sector.

49

This difference is signiﬁcant only for workers in federal and state administrations. The
North Region is the only region where municipal workers receive more than private
sector employees, by 3 percent, and this difference is statistically signiﬁcant at the 1
percent level.

The national average effects of the F UNDEF reform

The question of interest is whether the FUNDEF reform has a positive impact on
teacher earnings, after controlling for other observable factors such as years of education,
experience, region of the country, and location of workers’ residences. Table 4A reports
the estimated effects of FUNDEF on teacher wages for different speciﬁcations and
samples. Column I particularly shows the results from Simple difference estimation (SD)
for a sample of primary teachers (N=13,859). Before the reform enactment, in 1995,
primary teachers’ wages were 15.6 percent less than in the year 1996, and this difference
was statistically signiﬁcant at the 1 percent level. After the reform enactment, there was
an overall increase in primary teachers’ wages. For example, in 1997, teachers’ wages
were statistically superior to wages in year 1996 by 5.6 percent. This increase was
signiﬁcantly accentuated with the reform implementation in 1998 and 1999, representing
changes in wages of 18.9 and 24.2 percent, respectively, compared to year 1996.
These positive changes in primary teachers’ wages indicate that it is possible that, on
average, the F UNDEF reform has positively affected teacher earnings in primary
education all over the country.

However, SD (simple differences estimation) does not allow us to separate the
effects of the FUNDEF reform from aggregate time trends that may also affect teachers’

earnings. In order to separate the inﬂuence of the aggregate time trends from the reform

50

effects, I compared primary teachers with all other individuals (non-primary teachers),
whose wages may also be inﬂuenced by the country’s economic situation, but not by the
reform. In column 11, I estimate the difference-in-differences equation (DD), comparing
primary teachers with non-primary teachers in order to observe how the reform has
affected teachers in relation to all other workers in the country. As observed in the SD
estimation, the DD estimation also suggests that the introduction of the FUNDEF reform
increased primary teacher wages, but this increase started only after the reform
implementation in 1998 and 1999. Even for the state of Para, which initiated the reform
in 1997, teachers’ wages did not present any increase that was statistically signiﬁcant in
comparing Para with the rest of the country. However, if this state is compared to the
other states in the North Region anticipating the reform implementation favored teacher
wages in Para (See Table 5A, Columns I and II in appendix).

For 1998, holding other factors constant, the country observed a positive and
signiﬁcant reform effect of 9.2 percent in teachers’ earnings (Table 4A, Column 11). This
effect suggests that the mean wage difference between primary teachers and all other
workers in 1998 is 9.2 percent more than the mean wage difference in 1996. Such an
effect is economically, signiﬁcant considering that the inﬂation rate for this year was
about 2.5 percent (INPC/IBGE, 199.8). The FIPE report (1999) found an increase of 12.9
percent in 1998, but the study covered only teachers from all Brazilian state capitals and
200 other municipalities.

The estimates reported in my study shows that the reform effect was also positive
in 1999. There was an increase of 15.7 percent in teachers’ wages. As mentioned before,

if the reform had a positive impact, it may have happened mainly in the most distant and

51

small municipalities of the countryside, predominantly in the rural municipal networks of
schooling, where low revenues and low teacher earnings were predominant. In December
of 1996, the MEC preposal for the creation of the FUNDEF, using data from 1995,
predicted that with the institution of the F UNDEF , state networks of schooling would
lose resources to municipal networks in 20 out of 26 states of Brazil, due to the
redistribution mechanism of the law (MEC, 1997d). Table 2 shows this trend. In 1998
and 1999, the municipal networks of schooling (Column 111) in the poorest regions of the
country (See Table A1 1, Column IV and V) beneﬁted most from the reform.

In Table 4A, Column 111 shows the net effect of the F UNDEF reform across
municipal networks of schooling. Comparing municipal networks of schooling against all
other types of schools, teachers’ wages in municipal schools have increased since 1997,
surprisingly. In 1997 the increase was 6.4 percent and statistically signiﬁcant. In 1998
and 1999, there were also positive and statistically signiﬁcant increases in wages of 14.6
and 16 percent, respectively. The estimates for year 1997 may have resulted from
information about the reform disseminated by governmental reports in 1996 (MEC,
1997d). Positive estimates regarding the impact of the F UNDEF on schools’ network
budgetary constraints for 1997 may have also acted as incentives to review teacher wage
policies in municipal schools. After controlling for the other types of schooling (Column
VIII), the increases in municipal teacher wages were statistically Signiﬁcant, compared
only to primary teachers in state schools.

Column IV shows that wages of primary teachers in state schools decreased by

20.2 and 26.3 percent compared to primary teachers in other types of schools, and

52

relative to all other workers in 1998 and 1999 respectively, after the reform.37 There were
no signiﬁcant changes in wages for teachers in federal schools compared to all other
individuals (Column V), to municipal workers alone (Column VIII), and in relation
exclusively to the private sector (Column VI). These results should be interpreted
carefully since the number of cases in the sample for federal teachers is very small.

Wages for primary teachers in private schools also increased by 22.1 and 21.5
percent in the years 1998 and 1999 after the reform, compared with wages of all other
workers in the same period, relative to 1996. There have been annual increases in the
wages of primary teachers in the private sector during the period between 1995 and 1999
compared with all other teachers, but this difference has been decreasing over time since
1995. This information may suggest that due to decreasing demand for private education,
private schools have been forced to cut costs which may have negatively affected
teachers’ wages. The number of students in private schools has been decreasing Since
1995. In 1995, these students represented l 1.6 percent of total enrollment in primary
schools; the share has decreased to 9.1 percent in 2000.

If primary teacher wages in private schools were compared to primary teacher wages in
municipal schools (Column VIII), wages in private schools were Statistically signiﬁcantly
superior to wages in municipal schools before the FUNDEF reform. Aﬂer the reform, this
difference is small and non signiﬁcant.

The reform has increased the wages of municipal teachers, making changes on wages in

municipal schools similar as the ones in private schools. Controlling for all types of schools,

 

37 To be exact 20.2 is the percentage difference in monthly wage rates between state primary teachers and
non-state primary teachers relative to the percentage difference in monthly earnings between all other
workers at the state administratrion and all other workers in other types of administration in 1998 compared
to those in 1996, due to the reform.

53

average difference in primary teacher wages between 1996 and 1998, or 1996 and 1999 from state
schools are 25.7 and 29.7 percent less than the average difference in primary teacher wages in
municipal schools between 1996 and 1998, or 1996 and 1999, relative to the average wage
differences between all other workers in state and municipal administrations (Column VIII).

Table 6A presents the similar speciﬁcations as in Table 4A, but compares only primary
teachers with all other individuals who work for municipal and state administrations. Column I
shows the results of the DD estimation. They are consistent with the results in Table 4A.

The DD estimates indicate that primary teachers had increases of 8.8 and 14 percent in
1998 and 1999, respectively, and these results were statistically signiﬁcant at the 5 percent level. In
Column 11, which shows the difference-in—difference-in—diﬂ‘erences estimation results (DDD) for
the sample of municipal and state workers, the reform effects are also consistent with Table 4A.

The results indicate that the FUNDEF reform had strong and signiﬁcant effects on
municipal networks of schooling. For 1998 and 1999, teacher wages in municipal schools were
signiﬁcantly higher by 23 and 26 percent than wages in state schools in 1996, at the 1 percent level,
relative to non-teachers in state and municipal jobs. In Column III, I restrict my sample to compare
only primary teachers against pre-school and secondary teachers. The DD estimates Show positive
effects on primary teacher wages relative to preschool and secondary teachers, but these effects
were only statistically signiﬁcant in 1999. Municipal teachers had also an increase in wages relative
to preschool and secondary teachers of 2.3 percent, in both 1998 and 1999, but these differences
were not statistically signiﬁcant (Column IV).

The effects of the FUNDEF reform and teachers’ years of education

MEC data on primary teachers were indicating a progressive increase in levels of

qualiﬁcation of teachers (MEC/INEP, 1999). The PNAD data also conﬁrm this trend

54

(see Table 3A). It is possible that the refomr effects on wages observed in Table 4A have
accentuated this trend.

Table 7A reports the estimated effects of FUNDEF on teacher wages by level of
education. Column I shows the SD estimation for the sample of primary teachers with
years of education aggregated in age groups. As expected, the higher the level of
education, the higher the percentage of increase in teacher wages. These increments are
statistically signiﬁcant at the 1 percent level.

Column 11 reports the DD estimation for the sample of primary teachers versus
pre-school and secondary teachers The effects of the reform in this comparison are also
positive and signiﬁcant, consistent with the previous results reported in Table 4.

Column III shows that in 1998 there were no signiﬁcant reform effects on the
wages of primary teachers depending on years of schooling, compared with teachers with
less than eight years of schooling. There are no differences on wages between primary
teachers with 9 to 11 years of completed schooling and teachers working in other levels
of education with the same years of schooling. The same happens for teachers with 12 to
14 years of schooling and for teachers with 15 or more years of schooling. The effects of
reform on the average primary teacher wages may not be taking into consideration
teacher levels of qualiﬁcation.

In 1999, however, the reform effects on wages by teachers’ years of schooling
suggest that less qualiﬁed teachers received more than qualiﬁed teachers. Primary
teachers with 12 to 14 years of schooling compared to primary teachers with 0 to 8 years
of education received, in 1999, 35.9 percent less than primary teachers with 12 to 14

years of schooling compared to those teachers with 0 to 8 years of education in 1996,

55

relative to the wage difference of preschool and secondary teachers within the same
education groups between 1996 and 1999. For the group of primary teachers with more
than 15 years of completed schooling, the effect was -25.5 percent. Both effects were
statistically signiﬁcant at the l and 5 percent levels, respectively (Table 7A, Column 111).
These results suggest that the increases in wages due to the reform may have been
offset by increases in wages of pre-school and secondary teachers in the same educational
groups (Table 7A, Column 111) and by increases in wages of all other workers (Table 7A,
Column V). In other words, even though the reform may have increased average wages
for primary teachers, it doesn’t change the fact that primary teacher is still an occupation
with lower pay relative to other occupations. Therefore, individuals with higher levels of
education still may receive higher wages in other occupations, even aﬂer the reform.
Comparing with all other workers (column V), the results are consistent and the
dimension of the effect varies slightly indicating that even comparing with a less

homogeneous group, the results hold.

The effects of the FUNDEF reform on newly hired teachers

The question of interest here is whether there are differential reform effects
depending on how long teachers have been working in the current job. Whether newly
hired teachers are better off after 1998, may suggest that the reform has positively
affected the wages of these teachers. In Table 3A column 11, I compare primary teachers
with non-teachers. The effect of the reform on newly hired teachers is negative and
represents a decrease of 13 percent in teachers’ wages relative to non-teachers’ wages for

the year 1997. In 1998 and 1999, this tendency is accentuated. Teachers’ wages were 16

56

percent lower than in 1996 relative to non-teachers in both years 1998 and 1999, and
these results are signiﬁcant at 1 percent level. However, for teachers with one year of
experience relative to non-teachers, the reform effect is positive in 1997, 1998 and 1999,
representing an increase of 10 percent, 12 percent and 2 percent, respectively. These
effects are only signiﬁcant for the years 1997 and 1998.

In Table 3A, Columns III and IV, I run the same speciﬁcation for the sample of
primary, pre-school and secondary teachers. In this case, the reform effect is positive and
signiﬁcant only for teachers with three years of tenure. This may reﬂect changes in wage
schedules after the reform. For teachers who acquired two years of experience, the
observed increase of 17.8 percent on wages may serve as incentives for teachers to
continue in the school system”.

Column IV compares primary teachers with teachers in pre-school and secondary
education who work in the state or municipal administration. This comparison may be
more accurate to describe the reform effect. In this case, new hired state teachers received
23.4 percent less than their counterparts in municipal schools during the year 1999. This
difference is statistically signiﬁcant at 10 percent level. These results are consistent with
the results indicating positive effects of the reform on wages for municipal teachers.

I also run additional regressions include age group dummies to access if there are
changes in wages by cohort groups. The results show no effects. In fact, for all age

groups and years there are no signiﬁcant effects (Table A9, in appendix)

 

38 For teachers who are “funcionario publico estatutario”, after two years, they are granted job stability and
wages usually are upgraded.

57

Regional differences

In this section, we discuss the reform effect by regions of the country, since the
impact of the reform is likely to differ depending on the country’s region. Table A10
illustrates the net effect of the reform in Brazil and Table A1 1 to A15 reports the
estimates for the ﬁve main geographic regions of the country. The DD estimates show a
positive reform effect on primary teacher wages in almost all regions of the country,
except the Center-West. These effects were statistically signiﬁcant only in the Northeast
and Southeast Regions.

The Northeast region has one of the greatest educational deﬁcits. The percentage
of illiterates in 1995 was around 30 percent. The 1996 net enrollment rate was only 82.5
while the 1996 national average was 90 percent (MEC/INEP, 1999). Teacher level of
qualiﬁcation is also very low. Around 17.7 percent of the teachers have completed
primary education or less (MEC, 1998, p.92), equivalent to 8 years of education or less.

Not surprisingly, the effect of the reform is very high. It brought by an increase
of 18.3 and 28.6 percent in average earnings in this region (Table A12A, Column 11).
The DDD estimates also conﬁrm a positive effect for primary teachers in municipal
schools, whose wages increased on average 17.9 and 15.6 percent in 1998 and 1999,
respectively (Column IV). If compared to average teacher wages in private schools,
controlling for all other types of teachers, municipal teachers received an increase of 23.6
and 20.9 percent in the same years, which was statistically signiﬁcant at the 1 percent
level (Column V).

Even though teacher-eaming levels were extremely low in this region, such an

increase may imply fewer investments in teacher training where it is most needed. This

58

increase in earnings may serve to attract more qualiﬁed teachers, but it may also
constitute an undeserved premium to poorly qualiﬁed teachers in the system if changes
are not discriminatory.

The reform effect was negative for teachers in state schools compared to all other
types of schools, however (Column 111). There was an average decrease of 16.6 and 25.3
percent in the relative teacher wages in state schools, which was statistically signiﬁcant at
the 1 percent level. Despite this decrease in average wages, there were no statistically
signiﬁcant differences in wages between state and private schools for primary teachers if
we control for all types of schools (Column V). In this region, municipal primary teachers
were the greatest beneﬁciaries of the reform, conﬁrming that the redistribution reform
effect was able to minimize wage differences among the school networks.

In the Southeast Region, the DD estimates shows positive and statistically
signiﬁcant increases of 9 and 12 percent in primary teachers’ average wages in 1998 and
1999 (Table A13A, Column 11). The FUNDEF redistribution effect reduced municipal
school network revenues in favor of state networks”. The DDD (Table A13A, Column
111) shows that the impact of the reform is positive for teachers in state school networks
compared to all types of schools, representing increases of 10 percent and 7 percent in
1998 and 1999, respectively. These increases were only statistically signiﬁcant in 1998.
Furthermore, if comparing wages in state versus private schools, the increase in wages
was signiﬁcantly higher, on the order of 17 and 15 percent for average primary teacher
wages in state schools (Column V). The DDD estimates show negative effects, but not

signiﬁcant effects on wages for teachers in municipal schools after 1996 for this region

 

39 The exception is in the state of Rio de Janeiro, where municipal networks gain revenues with the
FUNDEF reform. (See Table 2).

59

(Column IV). Except for the state of Rio de Janeiro, the average municipal school
network lost resources in favor of state school networks. This may explain why there
were positive effects for teachers in state schools, instead of for those in municipal
schools. In this region, teachers have higher levels of education, so expenditure on
teacher training may not be as much in demand as in other regions. Consequently,
additional resources brought by the law after 1998 may have been predominantly directed
to increase teacher earnings in state schools. In fact, the regional average monthly
earnings for a primary teacher with 12 years of education was R$361.40 (SD=1.41),
whereas the monthly national average was R8308.05 (SD=1.48).

In the South Region, even though the DD estimates Shows positive reform effects,
the DDD estimates (Table A14A, Columns 11 and 111) show negative effects for state and
municipal schools in 1998 and 1999. Compared to private schools, the effects were
positive. No effects for this region were statistically signiﬁcant.

In the Center-West, Table 15A shows negative effects for primary teachers in
state and municipal schools, but these effects were not statistically signiﬁcant (Columns
11, III and IV). There were positive effects for primary teachers in federal schools
compared to private schools, but these estimates may be capturing policies for federal
employees, not F UNDEF reform effects (Column V).

For the North Region, there were positive reform effects of 3 percent and 14
percent in 1998 and 1999, respectively, using DD estimates (Table A1 1A, Column 11).
These effects were signiﬁcant, only for the year 1999. The DDD estimates (Table A10,
Columns III, IV and V) show non-signiﬁcant effects. As noted before, the state of Para,

which changed its ﬁnance system in 1997, presented signiﬁcant and positive reform

60

effects since 1997 comparing teachers in this state with the Region. Table A4 shows a
positive effect of 24 percent, in 1997; 10 percent in 1998, and 19 percent in 1999. These
effects were statistically signiﬁcant only in 1997 and 1999. The effects observed in 1997
are surprising because additional resources were transferred to Para only after July
1997.40 In 1998, however, the reform effect in this state was not signiﬁcantly different
than in the other states in the region. Para’s educational policy prioritized
decentralization of the ﬁrst four grades of primary education to municipal level in 1997
(Sevilla, 1997), which may have increased enrollment in municipal school networks, as
well as revenues, which were transferred according to a per pupil formula. However,
lack of adequate infrastructure was especially strong in this state. For instance, 20
percent of the municipal schools reported having no access to water supply and 80
percent reported having no access to electricity (MEC, 1999). This may have affected the
capacity of municipal networks to continue expanding school places in 1998 and
receiving additional students and monies in the ﬁrst year of the reform. Furthermore, it is
possible that additional resources may have been spent on teacher training and other
activities making'improvements in levels of remuneration trivial.

I also run a joint Si gniﬁcant test for changes on teacher earnings in the state of
Para in 1997, 1998 and 1999. In this case, there were statistically signiﬁcant positive
effects (p=0.091). The results are consistent with what educational reform theory
predicts. Reform effects for the North region, as a whole, may not be immediately
observed because the system’s structure—system management and administration, school
management, and teachers—demand time to adapt themselves to the top-down reforms

(Tyack and Cuban, 1995). The results for the state of Para may well be due to the small

 

4° STN/SIAFI, Monthly FUNDEF transfers to states.

61

sample size compared with the overall number of cases. The total number of primary
teachers in Para after the reform was 272.

As mentioned in the empirical speciﬁcation section, we added an additional
model for assessing the robustness of the results. Table 3, in the next page, shows the
results of this model. Column I presents a simple regression of lnwage on human capital
characteristics for a sample of primary teachers in state and municipal public
administrations. Column II presents the results for a sample of all workers in state and
municipal public administrations in Brazil. Consistent with the previous discussions, the
reform effect on primary teacher wages is positive and statistically Signiﬁcant in the order
of 8 and 12 percent in the years 1998 and 1999. Column 111 shows the results including
the variable per pupil revenues and the interactions with primtch and year dummies. The
effects of the reform due to the changes in per pupil revenues appears signiﬁcant only for
the year 1999, even though negative and very small (-0.0002). This may indicate that
changes in the wages of teachers happened due to new policies adopted with the
FUNDEF reform, not because of the amount of resources redistributed. The law
mandates that the average wage for qualiﬁed teachers were in 1998 roughly US$300,
which represents increases on average wage only for teachers in rural and remote areas
where wages were below this minimum. This statement may have inﬂuenced municipal
and state governments to review teacher career plans. However, a better explanation is
that the teaching profession attracted more qualiﬁed teachers and school administration

investments in teacher training have promoted increases on wages.

62

Table 3: Estimated Reform Effects on Primary Teacher Wages using Per Pupil Revenues

as predictors

 

 

 

Total
lwagel ( I ) (II) (III)

Coef. S.Err Coef. S.Err Coef. S.Err
educlyr 0.02852 0.2750 0.06617 0.0219 1”” 0.06566 0.0218 ***
educ2yr —0.21862 0.2864 0.10714 0.0176 *** 0.10446 0.0176 ***
educ3yr 0.18194 0.2298 0.17604 0.0153 1“” 0.17261 0.0153 ***
educ4yr -0.10157 0.2135 0.24536 0.0128 *** 0.23910 0.0128 1"”
educ5yr 0.01275 0.2179 0.31484 0.0150 *** 0.30960 0.0150 ***
educ6yr 0.10133 0.2239 0.37448 0.0183 **"‘ 0.36815 0.0184 **"‘
educ7yr 0.06529 0.2181 0.43894 0.0178 “‘1' 0.43300 0.0178 1“”
educ8yr 0.10136 0.2122 0.55408 0.0135 1”” 0.54632 0.0136 1””
educ9yr 0.21029 0.2132 0.65977 0.0186 *** 0.65196 0.0186 1“”
educlOyr 0.29687 0.2119 0.69804 0.0166 “‘1‘ 0.68887 0.0166 1“”
educl lyr 0.51992 0.2096 ** 0.95396 0.0123 *** 0.94415 0.0124 “W
educ12yr 0.63568 0.2102 *** 1.13414 0.0174 1”” 1.13037 0.0174 ***
educ13yr 0.71943 0.2108 *** 1.24366 0.0203 *** 1.23607 0.0203 *"
educl4yr 0.79008 0.2103 **"' 1.34756 0.0174 *"‘* 1.34437 0.0174 ***
educ15yr 0.89330 0.2100 *** 1.65417 0.0135 **"' 1.64833 0.0135 ***
age 0.16562 0.0646 ** 0.14860 0.0244 1“” 0.14864 0.0244 *"‘*
age2 -0.00571 0.0026 ** -0.00434 0.0010 1”” -0.00433 0.0010 ""'”"
age3 0.00009 0.0000 "' 0.00006 0.0000 1”” 0.00006 0.0000 1“”
age4 0.00000 0.0000 "' 0.00000 0.0000 “'1' 0.00000 0.0000 1””
tenure 0.01182 0.0023 “‘1‘ 0.01305 0.0011 *** 0.01299 0.0011 ***
tenure2 -0.00002 0.0001 0.00006 0.0000 0.00006 0.0000
female -0.06916 0.0159 1”” -0.29012 0.0053 *** -0.28866 0.0053 1””
white 0.03379 0.0109 **"‘ 0.08873 0.0057 1“” 0.08923 0.0057 ***
black 0.01731 0.0270 -0.03309 0.0107 1”” -0.03376 0.0107 *"”"
yellow -0.02611 0.0725 0.13584 0.0420 *** 0.13585 0.0418 "1'
indigenous 0.16359 0.1282 0.15176 0.0695 ** 0.15878 0.0697 1"
union 0.11853 0.0101 1'“ 0.15836 0.0056 1'" 0.15951 0.0056 1””
state 0.07368 0.0105 1‘" 0.14640 0.0052 1‘" 0.14738 0.0059 "‘**
urbanl 0.14842 0.0149 **"‘ 0.10221 0.0079 1'" 0.10485 0.0079 ***
primtch -0.11980 0.0137 1”” -0.03526 0.0229
perpupil 0.0001 1 0.0000 ** "‘
primrev -0.00026 0.0001 * * *
y95 -0. 16591 0.0151 *** -0.1 1941 0.0086 1“” -0.10694 0.0133 1‘"
y97 0.05732 0.0150 *** 0.06456 0.0089 *** 0.09338 0.0150 1“”
y98 0.21497 0.0144 "1‘ 0.12183 0.0084 "1‘ 0.16601 0.0213 1“”
y99 0.27538 0.0143 **"‘ 0.13928 0.0084 *** 0.21878 0.0231 1“”
y95prim -0.04320 0.0182 ** -0.09924 0.0304 ***
y97prim -0.01014 0.0184 -0.04014 0.0315
y98prim 0.08090 0.0178 ”1' 0.10477 0.0415 1““
y99prim 0.12071 0.0177 *"‘* 0.23993 0.0430 1“”

 

63

Table 3: (cont’d)

 

 

 

Total
lwagel (I) (II) (III)
Coef. S.Err Coef. S.Err Coef. S.Err
y95pupilrev -0.00003 0.0000
y97pupilrev -0.00008 0.0000 **
y98pupilrev -0.00012 0.0000 **
y99pupilrev -0.00020 0.0000 1"”
y95primrev 0.00017 0.0001 *"'
y97primrev 0.00010 0.0001
y98primrev -0.00002 0.0001
y99primrev -0.00020 0.0001 * "'
Rondonia 0.13401 0.0414 *** 0.28916 0.0261 *** 0.29053 0.0262 1””
Acre 0.01992 0.0522 0.19456 0.0306 “‘1‘ 0.19701 0.0310 1'”
Amazonas 0.34867 0.0343 1'" 0.19159 0.0206 “'1‘ 0.19258 0.0207 1“”
Roraima 0.46479 0.0851 *** 0.42857 0.0353 *** 0.44221 0.0364 ***
Para -0.06364 0.0310 ** 0.01410 0.0167 0.01707 0.0168
Amapa 0.51867 0.0563 *** 0.38647 0.0350 "1' 0.39821 0.0361 ***
Tocantins 0.02775 0.0330 0.05450 0.0187 ”1‘ 0.05676 0.0187 ***
Maranhao -0.11282 0.0375 *** -0.12335 0.0220 “'1‘ -0.12822 0.0220 "'l‘
Piaui -0.28257 0.0323 *"‘* -0.26505 0.0217 *** -0.26846 0.0217 ***
Ceara -0.17111 0.0295 1”” -0.1 1213 0.0148 1‘" -0.1 1498 0.0148 “*
RGNorte -0.40025 0.0433 1”” -0.28837 0.0221 *** -0.28703 0.0221 *"
Paraiba -0.24826 0.0374 1”” -0.20130 0.0194 "1' -0.20314 0.0194 1””
Pemambuco -0.08178 0.0271 *** -0.04549 0.0145 *** -0.04510 0.0145 "1'
Alagoas -0. 18775 0.0422 *** -0.15252 0.0228 *** -0.15906 0.0227 1'”
Sergipe 0.02313 0.0404 -0.01783 0.0197 -0.01572 0.0197
Bahia -0.07469 0.0230 *** -0.00317 0.0131 -0.00587 0.0131
Minas 0.30300 0.0225 *** 0.18210 0.0124 1”” 0.18238 0.0124 “‘1‘
Espirito 0.27434 0.0372 “* 0.26325 0.0190 *1" 0.26804 0.0193 "1'"
Rio 0.31068 0.0255 *" 0.15653 0.0140 1”” 0.15667 0.0153 "*
SaoPaulo 0.45227 0.0242 *** 0.36926 0.0126 "1' 0.36288 0.0155 ***
Parana 0.35703 0.0251 *** 0.22386 0.0138 **"‘ 0.22727 0.0140 ***
SantaCat 0.16841 0.0332 *** 0.27596 0.0179 “'1‘ 0.28016 0.0182 ***
RGSul 0.22384 0.0250 "1' 0.24448 0.0138 *** 0.25044 0.0146 ***
MGSul 0.08790 0.0348 ** -0.00648 0.0194 -0.00329 0.0195
MGrosso 0.20566 0.0319 1”” 0.19806 0.0180 *** 0.20040 0.0181 1"”
constant - 1 .90025 0.6004 *"“" -2.24629 0.2087 *** -2.28603 0.2087 *“
Observations 1 1380 65404 65404
F test 204.610 1352.590 1352.590
Prob > F 0.0000 0.0000 0.0000
R-squared 0.5418 0.5728 0.5735
Root MSE 0.4746 0.6037 0.6032

 

Note: Each speciﬁcation includes an intercept, the controls for human capital characteristics, years and
state effects. The marginal effects signiﬁcantly different from zero at 1 percent, 5 percent and 10 percent
level are indicated with 1‘", *“ and 1'. The table shows robust standard errors.

64

Conclusions

The educational ﬁnance reform has shown a signiﬁcant impact on teacher
earnings in Brazil during the ﬁrst year of its implementation and this effect has been
accentuated in 1999. The greatest beneﬁciaries of the FUNDEF reform were municipal
networks of schooling, which due to the redistribution mechanism within states,
increased their per pupil revenues. Wages for primary teachers in municipal schools
increased by 15 and 16 percent in 1998 and 1999, respectively, relative to wages of other
workers. Meanwhile, relative wages in state schools decreased. As a result primary
teacher wages in municipal schools are on average superior to wages for these teachers in
state schools, but Similar to wages in federal and private schools. After the FUNDEF
reform, wages that were increasing over time in the private sector, continued to increase
but at a decreasing rate. These results are robust when comparing teachers and non-
teachers who work for municipal and state administrations. If selecting a comparison
group of preschool and secondary teachers, the positive effects of the reform remain.

Primary teachers’ average years of education have been increasing over time. In
the ﬁrst year of the reform, there were no differential effects for teachers with different
levels of education. In 1999, however, the reform effect indicates that there was positive
effects on primary teachers wages, but the increases in wages for primary teachers with
complete secondary education were inferior to increases received by unqualiﬁed teachers
(teachers with primary education or less), relative to preschool and secondary teachers in

the same educational groups. The same effect was observed for teachers with higher

65

education. This suggests that the reform has most beneﬁted teacher with lower levels of
qualiﬁcation.

In each year after the reform, newly hired teachers have received 16 percent less
than their counterparts in other professions. Within the education sector, however, there
were no differences between wages of newly hired teachers in preschool, primary and
secondary education. Speciﬁcally comparing state and municipal teachers, newly hired
teachers in state schools received less than their counterparts in municipal schools,
relative to preschool and secondary teachers.

The reform effects, however, were not homogeneous around the country. It may
be the case that resources allocated to primary education are spent differently. Some
regions may spend more on teacher training than on increasing teacher earnings. The
municipal school networks in the Northeast region were the greatest beneﬁciaries of the
reform. Aﬁer the reform, municipal wages were signiﬁcantly higher than state wages for
primary teachers in this region. In the Southeast region, the changes in wages for
primary teachers in state school networks presented positive effects as compared to
wages for teachers in municipal schools. In the South and in the North region the reform
did not change relative primary teacher wages.

Individual state variation has not shown a larger greater impact on the reform
implementation Since OLS and ﬁxed effects estimators are almost the same. The reform
does not allow for variation across states, since it determines compulsory allocation of
resources for particular types of expenditures: 60 percent on teachers and 40 percent on

other recurrent and capital expenditures. Because the primary education coverage of the

66

country arrived in 1999 to 93 percent, additional capital expenditures may not be in as

high demand as in the last decades.

67

CHAPTER 3

THE EFFECTS OF THE REFORM ON TEACHER SUPPLY

Introduction

In 1996, the Brazilian government instituted an education ﬁnance reform, which
’ signiﬁcantly changed the average per pupil revenues in primary education in certain
school networks, and mandated a speciﬁc share of these funds for teacher training and
wages expenditures. As indicated in Chapter 2, during the ﬁrst two years of reform
implementation, there have been signiﬁcant increases in wage rates for primary teachers
in municipal schools compared to non-primary teacher workers and compared to primary
teachers in other school networks."l In this chapter, I examine to what extent these
reform effects are likely to affect the labor supply of primary teachers in Brazil. The
analysis is based on data from the PNAD/IBGE collected from 1995 to 1999. It reveals
important information about the determinants of teacher labor supply and the effects of
the FUNDEF reform on the participation of more highly educated teachers in the primary
education sector. In the subsequent section, I brieﬂy discuss the importance of the supply
of teachers, the model estimated and the data, following by the results, discussion

concluding remarks.

The reform and teacher labor force participation
Several factors have served to explain the importance of teacher labor supply. The

teacher is the primary input in the production of education, and an important item in the

 

4' See Sevilla, 2001 (Chapter 3).

68

expenditure categories in the public sector.42 Particularly in Brazil, after 1998, teacher

wages can represent as much as 60 percent of the public expenditure on education43

, and
a large Share of all public revenues invested in social areas. In addition, the quality of
educational systems in developed and developing countries is highly related to the
teacher labor supply. For example, the continuous growth of the population, and
consequently, the increasing number of people seeking admittance to schools often far
exceeds the number of places available, in developing countries. In order to attend to the
demand for education, the supply of teachers should follow the growth of the demand for
all levels of education.

However, increasing the number of teachers to address the growing demand for
education is not enough to guarantee the quality of education, if the quality of the
education. If the quality of the teaching work force must improve as well, school
personnel, teachers and administrators will also need to have higher levels of
qualiﬁcation to manage new tasks brought by administrative decentralization or school
autonomy movements, such as organizing the school and its relationship with the
communities it serves.

Brazilian legislators have adopted policies that aim to make teaching more
appealing, by providing the means for schools to review teacher career plans, provide
teacher training, and increase salaries. The FUNDEF reform assumes that changes in per
pupil revenues and the mandatory allocation of 60 percent of these revenues to teachers’

salaries will lead to increased wages which in turn will attract more skilled workers to the

education sector. The impact of the reform on teacher quality and wage rates depends on

 

’2 See Psacharoupoulos (1987), UNESCO (1991), Psacharoupolous (1996).
’3 Law 9,424, December, 1996. .

69

how the reform affects the labor supply and occupational choice decisions of individuals,
and which type of productivity endowments (such as experience, ability, education,
household characteristics --e.g. number of dependents, family income--, these individuals
bring to the education sector.

There are important features of the labor supply of teachers that must be known to
better understand the impact of the policy reform. First, what the determinants of labor
supply behavior and occupational choice are, and second, whether the reform affects
those decisions. Changes in primary teachers’ years of education were already being
observed prior to the reform in the public system (as Shown in Chapter 1, p.21). In 1991,
there were 1,295,965 primary teachers, of whom 72,285 (roughly 6 percent) had not
completed primary education and 67,087 (roughly 5 percent) had only completed primary
education. In 1999, there were 1,487,292 primary teachers and the percentage of teachers
with uncompleted primary education and completed primary education decreased roughly
to 2 and 4 percent respectively (MEC, 1996, 1999).

Furthermore, the reform increased the average per pupil revenues of municipal
schools in the country, because it introduced a redistribution and per pupil revenue
equalizing mechanism. All the main state and municipal taxes are now collected in a fund
within each state in Brazil, and then'the monies are returned to the educational systems in
proportion to the number of students enrolled. Consequently, revenue shifts from one
school network toward another occurred, and there were increases in average wages in
some schools. However, the reform seems to not affect the wages for primary teachers by
particular educational groups. There weren’t differential increases on wages based on the

education level of teachers in 1998. In fact, as Shown in Chapter 2, primary teachers with

70

12 to 14 years of education received less increases on wages than the group of primary
teachers with eight years of primary education or less in 1999, relative the same
comparison in 1996. This result suggests that, in the ﬁrst two years of implementation,
the reform has most beneﬁted teachers with lower levels of education in primary
education.

Since primary teacher wages for individuals with lower levels of education
increased, despite the policy intention to attract more highly educated teachers, this
chapter examines how the supply of teachers with different levels of education and,
particularly, qualiﬁed teachers (teachers with 12 or more completed years of education)

changed after the reform.

The model

In this analysis, I observe an exogenous event, the FUNDEF reform, which changes
the per pupil revenues in school networks in Brazil. This study estimates if the FUNDEF
policy reform has inﬂuenced workers and non-workers’ choice to participate in the teacher
labor market.

This policy change allows me to use the difference-in-differences framework to
observe if there are changes in the probability of being a primary teacher by educational
group in the years before and after the reform. I estimate the following equation by a probit

model:

P(primtch = l I X) = '30 +50y93i +51y95i +52y97i +§3y98i +64y99i

m ( 1 )

+55 yeardummies *edrrcationgoupsi + XXII/81' +u,~
Fl

71

where primtch is a binary variable taking the value of one, if the individual is currently
working as a teacher, and zero otherwise.

In order to control for systematic differences between control and treatment groups
over time, I include the dummy variables y95 and y97 for the years before reform
implementation (1995, 1996 and 1997), and y98 and y99 to capture the effect of the reform
after the policy implementation. These dummy variables assume a value of one for their
corresponding year and zero otherwise. There is no dummy variable for the year 1996,
because this is the base year deﬁned for this analysis. The reform was enacted in 1996,
even though implementation started in 1998. These dummy variables capture year speciﬁc
variation in the probability of being a primary teacher in each speciﬁc year. For example,
the parameter 63 measures the mean difference in the probability of being a teacher (who
is participating in the labor market) between 1996 and 1998.

In addition, the model includes a vector of control variables X 1" j to take into

consideration individual human capital characteristics such as education and age and also
household characteristics. The variable ed0to8 indicates if the individual has zero to eight
years of education; ed9to] 1 indicates if the individual has nine to eleven years of education
(the equivalent of having secondary education diploma); ed] 2toI 4 indicates if the
individual has twelve to fourteen years of education (corresponding to acquiring teacher
higher education diplomas or incomplete higher education), and ed15 indicates if the
individual has completed higher education or more (four-year-college or more). The
variable is truncated at 15 years of education indicating that the individual has 15 or more
completed years of education. The variables age, and its quadratic forms (agez, age3 and

age4) were included to capture non-linear effects of age on teacher participation decisions.

72

Other factors that affect labor supply are gender and race. I deﬁne the dummy
variable female, which equals to one if the individual is female and zero if male. This
variable indicates whether holding other factors constant there are gender differences in the
labor market for teachers. The reference group chosen for race/ethnicity is mixed, which
represents roughly 36 to 38 percent of the total sample population over the years. The
dummy variables applied was white equals one if the individual declared being white and
zero otherwise; black equals one if the individual declared being black and zero otherwise;
yellow equals one if the individual declared being Asian and zero otherwise; and
indigenous equals one if the individual declared being Indigenous and zero, otherwise.

Similarly, differences in family characteristics, e. g. marital status, number of
children and non-labor income, may affect the household budget constraint, which in turn
may inﬂuence the individual decision to participate in the primary teacher labor market. I
include the dummy variable married which indicates if the individual is manied. The
variable child114 indicates if the individual has children younger than 14 years old, and the
dummy variable childm14 indicates whether or not the individual has children who are 14
years old or more. Particularly for females, the labor market participation decision may also
depend on whether there are other sources of income in the family. I included the variable
nonlabor to control for differences in non-labor income including income from other
family members.

The main parameters ofinterest are, however, the interactions between the year
dummies and the education groups (yeardummies *education groups). These interactions
capture whether the reform had a differentiated impact on teacher participation behavior

depending on educational group (ed9tol 1, ed] 2toI 4, and ed15). For example, the

73

interaction between ed9t011 and the variable y98, aims to test whether the reform affected
the participation of teachers with completed secondary education compared to teachers
with eight or less years of education in the ﬁrst year of its implementation, relative to 1996.
I also estimated a structural model to predict teacher labor market participation to
analyze whether wage differences between primary teachers and non-primary teachers
due to the introduction of FUNDEF may affect the probability of becoming a teacher. In

this case, the equation of interest is given as:

P(primtch = l | X) = .30 +50y93, +5ly95, +52y97i +63y98, +54dijf
(2)

m
+ dsyeardummies * educationgroups + Ext/'81 + u,
i=1

where diff is the predicted difference between teachers and non-teachers wages for a
particular educational group, in a particular region and year estimated using equation 3,
below.

Since wages and labor supply are jointly determined, there is an identiﬁcation
problem. In order to address this problem I include the variable enroll (total enrollment in
primary education in the previous year) which may affect wages but not labor supply."4
The identiﬁcation strategy consists of estimating separately the wage for primary teachers
and non-primary teachers using equation ( 3 ) for each region of the country, using state-
ﬁxed effects. Speciﬁcally for the primary teacher wage equation, I include the variable
enroll, which indicates the total number of students enrolled in primary education in the
previous year. Enrollment in the previous year may affect educational planners’ decisions

regarding the demand for teachers for the next year, and consequently, inﬂuence school

 

’4 I also estimated the models including pupil/teacher ratio of the previous year and the results did not
change signiﬁcantly.

74

network capacity to pay teachers (teacher wages), but it should not inﬂuence individuals’

decisions to go into teaching (primary teacher participation decision).

ln(wage), = ’60 + Ooyeardummiesi +53educationgroups

(3)

m
+ 57yeardummies * educationgroups + Z XUBJ. + ui
i=2

In equation ( 3 ), ln(wage) is the natural logarithm of the individual wage rate.
Year dummies included are y95, y9 7, y98, y99. The education dummies included are
ed9t010, ed] 21014 and ed15. I also include interactions between year dummies and
education groups. In addition I included the human capital variables and indicators of
race as in chapter 2. The base year for this analysis is 1996.

Equation ( 3 ) is estimated separately for total sample of female teachers. There
was limited number of male teachers in the sample to run the same analysis for this group
and if the analysis included all sample, it would be neglecting the fact that there are
gender-speciﬁc differences in the effects of wage and non-labor income on labor market
participation decision and occupational choice.

Additionally, as observed in Chapter 2, the reform had positive and signiﬁcant
effects on wages for teachers in municipal schools, which are responsible for the majority
of enrollment in rural areas. Therefore, if the reform affected supply of teachers through
changes on wages, it is important to test whether primary teacher participation might
differ between rural and urban areas. In order to do this, separate sets of equations are

estimated for the sample of female teachers in urban and rural areas.

75

Data and variables

This empirical analysis employs data from the National Research for Sample of
Households (PNAD) for the years 1995 to 1999, conducted by the Brazilian Institute of
Geography and Statistics (IBGE). The survey is a random sample collected yearly, which
was designed to be nationally representative.45 It collects information on individual
demographics, labor force participation, household characteristics and income.

The sample consists of Brazilian households with individual data on 1,659,403
people, of which some are primary teachers in public or private schools (See Table A1, in
Appendix). From this total sample, I restricted the analysis to Brazilian workers who
were at least 20 and not more than 50 years old“. In addition, I dropped individuals that
have incomplete information on demographics characteristics. A total of 599,360
observations satisﬁed these criteria. They were utilized in the analysis of earnings and
labor force participation.

Table A16 shows means and standard deviations of the main variables used in this
analysis. The total number of primary teachers in the sample is 13,544, or approximately
2,708 each year, and the total number of non-primary teachers and non-workers is
585,816. Female teachers correspond to 92 percent of the teacher sample, whereas 57
percent of the non-primary teachers and non-workers are female. The average years of
education for teachers is 13, corresponding to teachers with completed secondary

education. In rural areas, the average number of years of education for a teacher is 11.

 

45 Except for the Region North, where the sample did not include individuals from rural areas.

’6 Age restrictions are different than the one used in previous chapter, even though the wage model
estimated is similar. For the labor supply estimations, I excluded individuals with less than 20 years old
because their labor supply behavior may be influenced by schooling decisions, likewise, I also excluded
individuals with more than 50 years old, since these individuals’ labor supply decisions may be inﬂuenced
by retirement decisions. Tables including the equations used to derive the predicted wage differences are
included in annex.

76

For non-primary teachers and non-workers, the average years of education is only 8
years, corresponding to completed primary education. in rural areas, the average number
of years of education is 7 years.47

Female primary teachers receive less than male primary teachers on average, both
in urban and rural areas. The average monthly earnings for a female primary teachers is
R8 394.20 (Brazilian Currency) whereas a male teacher receives on average R8456.59. In
rural areas, the wages are lower for both females and males, but female teachers still
receive less than male teachers (R8226.54 and R8248.73, respectively). For non-primary
teachers, females also receive lower monthly earnings on average than male teachers,
R8371.64 against R8575 .66. In rural areas males receive almost twice as much as
females.

The majority of the sample is composed of married individuals (77 percent). In
rural areas, 84 percent of the sample is married. The largest contrasts in family
characteristics are related to the number of children less than 14 years old. Fifty seven
percent of the female teachers have children less than 14 years old, as do the majority of
the non-primary teachers and non-workers (62 percent). Roughly 85 percent of the
teachers in this sample live in urban areas, whereas for non-primary teachers and non-

workers, 90 percent live in urban areas.

Results
The determinants of primary teacher labor force participation.
The estimated labor supply equations are presented in Tables 4, 5 and 6 for the

total sample of females, females in urban areas and rural areas, respectively. All the

 

47 Descriptive Statistics for the variables included in the estimation are described in Table l, in annex.

77

coefﬁcient estimates on the control variables have the expected signs, and most are
statistically Signiﬁcant at the 5 percent level.

Table 4 shows the results for the total number of females in the sample. Years of
education increases the probability of being a teacher, and these increases are statistically
signiﬁcant. Particularly for females, Column I Shows that more educated females are
more likely to be primary teachers. The average difference between female individuals
with 9 to 11 years of education and individuals with eight years of education or less in the
probability of being teacher is 9 percentage points. This difference increases to 29.1
percentage points and 25.3 percentage points for females with 12 tol4 years of education
and 15 or more years of education respectively. All these coefﬁcients are statistically
signiﬁcant.

Age also affects signiﬁcantly the probability of being a primary teacher in Brazil.
The estimated effect for females is 1.5 percentage points (Table 4, Column 1). For
example, the average female individual in the sample, whose age is roughly 30 years old,
is 15 percent points more likely to be a teacher compared to female individuals with 20
years old, holding other factors constant.

Type of family is also signiﬁcantly related to the probability of teacher wage
labor. Married females and females with children are more likely to be primary teachers.
For single mothers with children less than 14 years old, there is a 1.73 percentage points
decrease in the probability of participating in the labor market for females in rural areas,
but a 0.1 percentage points increase in the urban areas. Both of these effects are

statistically signiﬁcant. For individuals with children aged 14 years old or more there is a

78

signiﬁcant increase in the probability of being a teacher of 2 percentage points in urban
areas, but still there is a 0.6 percentage points decrease for females in rural areas.

The likelihood of being a primary teacher decreases for white, black, yellow and
indigenous compared to the mixed category. Analyzing the results for females in urban
and rural areas, the results are similar for all variables (See Table 5, Column I and Table
5, Column 11), except for females in rural areas, where the likelihood of being a white
teacher or indigenous teachers is higher than the mixed racial/ethnical group. Also, the
likelihood that a teacher has completed higher education or more is not statistically
signiﬁcant for a teacher in rural areas.

Lastly, the availability of non-labor income lowers the probability of being a
teacher for all individuals, and this is statistically signiﬁcantly different than zero. An
R8l,000 (Brazilian currency) increase in non-labor income is associated with a 0.004
decrease in the probability of being a female teacher, which is statistically signiﬁcant at
the 1 percent level for females.

The largesteffect of non-labor income takes place in rural areas for female
teachers. In rural areas, family size tends to be higher, so demands for in-home labor are
higher and the Opportunity cost of participating in labor market is higher too. An increase
in non-labor income for females in rural areas is associated to 0.008 percentage points
decrease in the probability of being a teacher, which is statistically Signiﬁcant at the 1
percent level.

The reform effects

Table 4, Columns 1, show the estimation results for total females in the sample.

The coefﬁcients of interest are the interaction terms between year and education group

79

dummy variables. As noted before, the reference year is 1996, and ed0to8 is the
reference education group.

Each interaction term measures the average difference in the likelihood of being a
primary teacher between teachers with eight years of education or less and the speciﬁc
group observed, before and after the reform. For example, the coefﬁcient ofy98ed9t011
captures the difference in the probability of being teacher with 9 to l 1 years of education
and the probability of being a teacher with less than eight years of education in the year
1998, relative to 1996. This coefﬁcient aims to answer whether the probability of being a
teacher changed in the ﬁrst year aﬁer the reform implementation, and the interest would
be to observe an increase in the group aged 9 to 11 years of education, i.e. more educated
group, aﬁer the reform.

In Table 4, Column 1, I present the results for females. For the years before the
FUNDEF reform was implemented, the year and educational group (ed12tol4)
interactions show increases on the teacher labor participation of 0.03 and 0.22 percentage
points, and these effects are not statistically Signiﬁcant. These results suggest that there
were no differences in the probability of becoming a primary teacher across educational
groups in the years before the reform implementation for all female individuals in the
country. However, after the reform implementation, in 1998, there were increases on the
probability of being a primary teacher for all education groups as compared to the
probability of being a female primary teacher in the same education groups in 1996,
relative to the probability of being a female primary teacher with eight or less years of

education. The estimated effects of the reform were increases of 0.37 to 0.65 percentage

80

Table 4: Estimates of the Reform Effects on Primary Teacher Labor Force Participation

 

 

 

 

 

P(primtch: 1) Total
( I ) ( III ) ( III )

dF/dx Std. Err dF/dx Std. Err dF/dx Std. Err
diff 0.0013 0.0017 -0.0012 0.0015
ed9toll 0.0901 0.0037 "”"* 0.0901 0.0037 ""‘ 0.0986 0.0014 ***
ed12tol4 0.2909 0.0150 *** 0.2932 0.0154 “‘1‘ 0.3110 0.0080 ***
ed15 0.2529 0.0098 *** 0.2599 0.0135 *** 0.2703 0.0104 ***
age 0.0150 0.0058 *"‘ 0.0150 0.0058 ** 0.0152 0.0059 **
age2 -0.0006 0.0003 ** -0.0006 0.0003 ** -0.0006 0.0003 **
age3 < 0.0000 0.0000 * < 0.0000< 0.0000 * < 0.0000 < 0.0000 *
age4 <-0.0000 0.0000 < -0.0000< 0.0000 <-0.0000 < 0.0000
white -0.0020 0.0003 1'" -0.0020 0.0003 "'** -0.0020 0.0003 "1"
black -0.0030 0.0006 *** 0.0030 0.0006 "1' -0.0030 0.0006 "‘"
yellow —0.0087 0.0007 *** -0.0087 0.0007 *"”" -0.0088 0.0007 ***
indigenous -0.0006 0.0039 -0.0006 0.0039 -0.0006 0.0039
Married 0.0055 0.0003 “‘1‘ 0.0055 0.0003 *"“" 0.0055 0.0003 1””
childll4 0.0009 0.0003 *** 0.0009 0.0003 *" 0.0010 0.0003 *"'*
childm14 0.0018 0.0003 *** 0.0018 0.0003 *** 0.0018 0.0003 ***
nonlabor(*1000) -0.0038 0.0002 1“” -0.0038 0.0002 1”” -0.0038 . 0.0002 1'"
y95 0.0009 0.0011 0.0010 0.0011 0.0018 0.0005 ***
y97 —0.0016 0.0010 -0.0017 0.0011 -0.0003 0.0004
y98 -0.0039 0.0010 1'” -0.0041 0.0010 1'" -0.0010 0.0005 **
y99 -0.0050 0.0010 1'" - -0.0054 0.0010 1”” -0.0015 0.0005 *"‘*
y95ed9toll 0.0013 0.0013 0.0012 0.0013
y95ed12tol4 0.0003 0.0019 0.0003 0.0019
y95ed15 0.0009 0.0015 0.0009 0.0015
y97ed9to1 1 0.0017 0.0014 0.0019 0.0015
y97ed12tol4 0.0022 0.0022 0.0024 0.0022
y97ed15 0.0017 0.0016 0.0018 0.0016
y98ed9toll 0.0037 0.0017 ** 0.0038 0.0017 **
y98ed12tol4 0.0039 0.0024 * 0.0041 0.0025 *
y98ed15 0.0051 0.0020 **"‘ 0.0053 0.0020 ***
y99ed9toll 0.0046 0.0018 1“” 0.0050 0.0019 *"‘*
y99ed12tol4 0.0057 0.0027 "”' 0.0064 0.0029 1””
y99ed15 0.0065 0.0022 *** 0.0070 0.0023 “'1'
Rondonia 0.0167 0.0038 *** 0.0167 0.0038 8‘” 0.0169 0.0038 ***
Acre 0.0224 0.0057 1”” 0.0225 0.0057 “‘1' 0.0226 0.0058 “'1'
Amazonas 0.0079 0.0022 “'1‘ 0.0079 0.0022 *** 0.0080 0.0022 ***

 

81

Table 4 (cont’d)

 

 

 

 

 

P(primtch=1) Total
( I ) ( III ) ( III )
dF/dx Std. Err dF/dx Std. Err dF/dx Std. Err

Roraima 0.0280 0.0076 *** 0.0280 0.0076 1"” 0.0284 0.0077 ""‘
Para 0.0067 0.0016 "1' 0.0067 0.0016 **"' 0.0068 0.0016 "*
Amapa 0.0244 0.0062 *** 0.0244 0.0062 “'1' 0.0247 0.0063 ***
Tocantins 0.0402 0.0051 *** 0.0401 0.0051 **"‘ 0.0406 0.0052 ***
Maranhao 0.0537 0.0056 *** 0.0541 0.0056 1”” 0.0538 0.0056 "”"*
Piaui 0.0550 0.0056 1'“ 0.0555 0.0056 1'” 0.0551 0.0056 ”1‘
Ceara 0.0152 0.0020 1”” 0.0154 0.0020 1“" 0.0152 0.0020 *"*
RGNorte 0.0165 0.0029 *** 0.0167 0.0030 “‘1‘ 0.0166 0.0030 1'"
Paraiba 0.0245 0.0033 *** 0.0248 0.0034 "”” 0.0245 0.0033 *"""
Pemam~o 0.0061 0.0014 *** 0.0062 0.0014 *"“" 0.0061 0.0014 *‘”
Alagoas 0.0344 0.0044 1”” 0.0348 0.0044 ”" 0.0343 0.0044 1””
Sergipe 0.0228 0.0034 1”” 0.0230 0.0035 *"'* 0.0228 0.0035 **"'
Bahia 0.0190 0.0020 1"” 0.0192 0.0021 *** 0.0190 0.0020 1'”
Minas 0.0103 0.0015 *** 0.0099 0.0015 ‘1'" 0.0107 0.0016 1‘"
Espirito 0.0108 0.0024 1“” 0.0104 0.0024 *** 0.0112 0.0025 1“”
Rio 0.0048 0.0012 *** 0.0045 0.0013 "* 0.0051 0.0013 ***
SaoPaulo -0.0007 0.0009 -0.0009 0.0009 -0.0005 0.0009
Parana 0.0033 0.0012 "* 0.0030 0.0013 “‘* 0.0035 0.0013 1"”
SantaCat 0.0085 0.0020 *** 0.0082 0.0020 ”* 0.0089 0.0021 ***
RGSul 0.0032 0.0012 *"'* 0.0030 0.0012 *"“" 0.0035 0.0012 ”*
MGSul 0.0050 0.0019 1‘" 0.0050 0.0019 *** 0.0052 0.0019 *"'*
MGrosso 0.0208 0.0030 *** 0.0208 0.0030 ”* 0.0210 0.0030 *"
Goias 0.0148 0.0021 *** 0.0148 0.0021 **"' 0.0149 0.0021 "*
obs. P 0.0361 0.0361 0.0361

pred. P 0.0112 0.0113 0.0114

N Of Observations 343,717 343,717 343,717

 

Note: Each Specification includes an intercept, the controls for human capital characteristics, years and
state effects. The marginal effects signiﬁcantly different from zero at 1 percent, 5 percent and 10 percent
level are indicated with 1“”, ** and ‘.

82

points, suggesting that a growing number of female individuals with more than 8 years of
education chose to be a primary teacher aﬁer the reform was introduced.

The average estimated effect for female individuals in 1999 indicates that the
effect of the reform remains positive and statistically signiﬁcant after two years for all
educational groups, as compared to the base group (ed8t010). There is an increase of
0.65 percentage points in the probability of being a teacher with completed higher
education or more compared to the likelihood of being a teacher with 15 or more years of
education in 1996 (Table 4, Column 1). These results suggest that the reform has attracted
more qualiﬁed female teachers to primary education.

One of the most important reform effects was the reduction in per pupil revenues
disparities within states, as discussed in Chapter 2. School networks in rural areas were
commonly poorer than in urban areas, but after the reform, there might be lower
differences among schools in rural and urban areas due to the guarantee of minimum per
pupil revenues. This suggests that the effect of the reform may differ in urban and rural
areas. Table 5 shows the estimated effects for urban areas, and Table 6, shows the
estimated effects for rural areas.

In Column 1, Table 5, the estimated reform effects for females in urban areas are
positive and not statistically signiﬁcant. I also run a joint signiﬁcant test for the year
dummy variables after the reform implementation and a test for the interaction between
year dummies and educational groups after the reform. Both were not statistically
signiﬁcant. These results suggest that even though there was redistribution of revenues
from urban to rural areas in the majority of the states, these revenue shifts did not

effected the probability of being a female teacher across education groups in urban areas.

83

Table 5: Estimates of the Reform Effects on Primary Teacher Labor Force Participation

in Urban Areas

 

 

 

 

P(primtch= 1) Urban
( I ) ( III ) ( III )

dF/dx Std. Err dF/dx Std. Err dF/dx Std. Err
diff 0.0048 0.0014 *** 0.0041 0.0013 1””
ed9toll 0.0948 0.0051 “‘1' 0.0991 0.0054 "”""' 0.1030 0.0022 *1"
edl2tol4 0.3309 0.0187 *** 0.3546 0.0205 *** 0.3689 0.0115 *"
ed15 0.2800 0.0132 *** 0.3263 0.0199 1”” 0.3377 0.0151 1””
age 0.0092 0.0049 1' 0.0092 0.0049 * 0.0092 0.0050 "
age2 ~0.0003 0.0002 -0.0003 0.0002 -0.0003 0.0002
age3 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000
age4 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000
white -0.0017 0.0003 1”” -0.0017 0.0003 *“‘ -0.0017 0.0003 *"""
black -0.0014 0.0006 ** -0.0014 0.0006 ** -0.0014 0.0006 "
yellow -0.0068 0.0005 *** -0.0068 0.0005 "1' -0.0068 0.0005 **"‘
indigenous 0.0006 0.0036 0.0005 0.0036 0.0005 0.0036
married 0.0039 0.0003 "* 0.0039 0.0003 *"‘"‘ 0.0039 0.0003 "*
childll4 0.0010 0.0003 1”” 0.0010 0.0003 1”” 0.0010 0.0003 **"'
childm14 0.0019 0.0003 *** 0.0019 0.0003 *"‘* 0.0019 0.0003 *"“"
nonlabor < 0.0000 0.0000 “* < 0.0000 0.0000 *** < 0.0000 0.0000 1“”
y95 0.0002 0.0013 0.0000 0.0013 0.0014 0.0004 ‘1'"
y97 -0.0008 0.0013 -0.0008 0.0013 -0.0002 0.0004
y98 -0.0020 0.0012 -0.0022 0.0012 * -0.0011 0.0004 1“”
y99 -0.0018 0.0012 -0.0030 0.0011 ** -0.0017 0.0004 *1"
y95ed9toll 0.0014 0.0015 0.0018 0.0016
y95ed12tol4 0.0006 0.0018 0.0013 0.0020
y95ed15 0.0013 0.0016 0.0015 0.0017
y97ed9toll 0.0006 0.0015 0.0007 0.0015
y97ed12tol4 0.0012 0.0020 0.0014 0.0020
y97ed15 0.0008 0.0016 0.0006 0.0016
y98ed9toll 0.0012 0.0016 0.0009 0.0016
y98edl2tol4 0.0013 0.0020 0.0013 0.0021
y98ed15 0.0022 0.0019 0.0021 0.0018
y99ed9toll 0.0002 0.0015 0.0010 0.0016
y99ed12tol4 0.0008 0.0019 0.0023 0.0022
y99ed15 0.0018 0.0018 0.0027 0.0019
Rondonia 0.0145 0.0034 "* 0.0148 0.0034 1”” 0.0147 0.0034 ”*
Acre 0.0190 0.0050 1‘” 0.0195 0.0051 1”” 0.0194 0.0051 "1‘
Amazonas 0.0069 0.0019 “'1' 0.0070 0.0019 *"'* 0.0070 0.0019 1””
Roraima 0.0237 0.0067 1“” 0.0241 0.0067 *** 0.0242 0.0067 1‘"
Para 0.0057 0.0014 "‘** 0.0058 0.0014 “'1' 0.0058 0.0014 *"
Amapa 0.0210 0.0055 "'" 0.0210 0.0055 *** 0.0210 0.0055 1'”
Tocant~s 0.0232 0.0040 **"‘ 0.0233 0.0040 *1” 0.0234 0.0040 1“”
Maranhao 0.0409 0.0056 "* 0.0420 0.0057 1”” 0.0418 0.0057 ***

84

Table 5 (Cont’d)

 

 

 

 

 

P(primtch: 1 ) Urban
( I ) ( III ) ( III )

dF/dx Std. Err dF/dx Std. Err dF/dx Std. Err
Piaui 0.0349 0.0045 *"‘* 0.0358 0.0046 1”” 0.0357 0.0046 1“”
Ceara 0.0072 0.0014 *** 0.0075 0.0014 “‘1‘ 0.0075 0.0014 1””
RGNorte 0.0100 0.0024 *** 0.0104 0.0024 *** 0.0103 0.0024 ***
Paraiba 0.0118 0.0023 *** 0.0124 0.0024 1“” 0.0123 0.0024 “‘1'
Pemambuco 0.0040 0.0011 *** 0.0043 0.0011 *** 0.0043 0.0011 **"'
Alagoas 0.0235 0.0038 *** 0.0243 0.0038 *"'* 0.0242 0.0038 1“”
Sergipe 0.0152 0.0028 *** 0.0158 0.0029 1”” 0.0158 0.0029 1””
Bahia 0.0114 0.0015 *** 0.0118 0.0016 *** 0.0117 0.0016 *"""
Minas 0.0072 0.0012 “‘1‘ 0.0059 0.0012 “'1' 0.0061 0.0012 *"
Espirito 0.0069 0.0019 1“” 0.0056 0.0018 *"‘* 0.0059 0.0018 *"
Rio 0.0035 0.0010 *** 0.0025 0.0010 **"' 0.0026 0.0010 *“
SaoPaulo -0.0005 0.0007 -0.0013 0.0007 * -0.0012 0.0007
Parana 0.0012 0.0009 0.0005 0.0009 0.0006 0.0009
SantaCat 0.0056 0.0016 1”” 0.0046 0.0015 *** 0.0048 0.0016 1””
RGSul 0.0018 0.0009 ** 0.0011 0.0009 0.0012 0.0009
MGSul 0.0039 0.0016 1”” 0.0040 0.0016 *"'* 0.0040 0.0016 ***
Mgrosso 0.0122 0.0023 1"” 0.0124 0.0023 **“‘ 0.0124 0.0023 1”"
Goias 0.0098 0.0017 *"‘* 0.0101 0.0017 *** 0.0101 0.0017 "1‘
obs. P 0.035 0.035 *** 0.035
pred. P 0.008 0.008 *** 0.008
Number of obs 305264 305264 *** 305264
(I chi2(57) 10645 10724 "* 10702
b > chi2 0.000 0.000 *** 0.000
Pseudo R2 0.233 0.233 *** 0.233

 

Note: Each speciﬁcation includes an intercept, the controls for human capital characteristics, years and
state effects. The marginal effects signiﬁcantly different from zero at 1 percent, 5 percent and 10 percent
level are indicated with ***, ** and *.

In Table 6, Column I Shows that the expected positive effects of the F UNDEF
reform in rural areas took place only in the second year of the reform implementation,
1999. More speciﬁcally, the estimated effect of the reform in the probability of being a
female teacher with 9 to 11 years of education compared to 0 to 8 years of education in
rural areas is 5.0 percentage points, and this effect is statistically signiﬁcant at the 1%

level. For female teachers with 12 to 14 years of education compared to teachers with 0

85

to 8 years of education, the effect is 0.61 percentage points, and this is not statistically
Signiﬁcant. The largest effect in rural areas occurred for teachers with 15 or more years of
education compared to female teachers with 0 to 8 years of education, and represents an
increase of 1.96 percentage points. This effect is statistically Signiﬁcant at 10% level.
Some fragility in the results for rural areas can be detected. First, the probability of being
a teacher for individuals with 9 to 1 1 years of education was positive before and after the
reform. The reform only accentuated the annual trend that has been manifested in
previous years during 1999. A second point to be careful about is that I did not observe
positive effects of the reform in the year 1998 for female individuals from the education
groups of 12 to 14 years of education and 15 or more years of education.

All results for urban areas are not statistically signiﬁcant.48 In rural areas,
however, the likelihood of being a primary teacher is positive and signiﬁcant for teachers
with 15 years of education or more in 1998 compared to teachers with 0 to 8 years of
education, relative to 1996. Because the participation decisions of males are different that
females, separate equations should be estimated for males. However, as mentioned
before, because of the small number of male teachers in the sample I prefer to not
consider this results reliable to discuss. I found no signiﬁcant reform effects, but these

results may not be accurate due to low number of cases”.

 

’8 The total number of male primary teachers for urban areas in this sample is 891 and 107 in rural areas.

’9 For total males, the reforrn effects through the wage differential is very small, negative and statistically
signiﬁcant. There are not statistically signiﬁcant effects through the year and education groups’
interactions. Similar results are found for male individuals in urban and rural areas. In these areas, the
estimated reform effect through wages is negligible, negative and not signiﬁcant. The estimation results do
not support the hypothesis that the change in the wage differential between teachers and non-teachers due
to the implementation of the reform may affect the participation of male primary teachers. These results are
not robust because of the limited number of cases per year in the sample.

86

Table 6: Estimates of the Reform Effects on Primary Teacher Labor Force Participation

in Rural Areas

 

 

 

 

P(primtch: 1) Rural
( I ) ( H ) ( 111 )

dF/dx Std. Err DF/dx Std. Err dF/dx Std. Err
diff -0.0067 0.0021 *** -0.0133 0.0011 “'1'
ed9toll 0.1004 0.0109 1"” 0.1486 0.0211 *"‘* 0.1960 0.0066 "*
ed12tol4 0.4248 0.0329 “‘1' 0.4993 0.0387 1'” 0.2863 0.0189 "”""‘
ed15 0.2510 0.0393 *** 0.2313 0.0395 *"‘"' 0.2013 0.0163 1‘"
age 0.0406 0.0227 "' 0.0411 0.0228 " 0.0984 0.0225 "””
age2 -0.0020 0.0011 "' -0.0020 0.0011 "' -0.0045 0.0011 1””
age3 < 0.0000 < 0.0000 ** < 0.0000 < 0.0000 '” 0.0001 < 0.0000 ”1'
age4 < -0.0000 < 0.0000 ** < -0.0000 < 0.0000 8* < -0.0000 < 0.0000 1””
white 0.0069 0.0013 1“” 0.0068 0.0013 ""' 0.0043 0.0013 "*
black -0.0083 0.0023 1“” -0.0083 0.0023 **"‘ -0.0090 0.0021 "1'
yellow -0.0080 0.0090 -0.0076 0.0094 -0.0046 0.0115
indigenous 0.0146 0.0366 0.0128 0.0346 0.0099 0.0328
married 0.0054 0.0014 1”” 0.0054 0.0014 ”1‘ 0.0031 0.0014
childll4 -0.0173 0.0017 1”” -0.0170 0.0017 **"' -0.0142 0.0017 ***
Childml4 -0.0057 0.0013 1'" -0.0056 0.0013 ”1' -0.0050 0.0013 1”"
nonlabor(*1000) -0.0083 0.0021 **"‘ -0.0080 0.0020 *** -0.0085 0.0021 1"”
y95 0.0036 0.0027 0.0087 0.0034 “'1' 0.0131 0.0022 ***
y97 -0.0038 0.0023 0.0003 0.0029 -0.0122 0.0013 "'"
y98 -0.0093 0.0021 *** -0.0034 0.0032 -0.0067 0.0016 1'”
y99 -0.0094 0.0020 ‘1‘" -0.0021 0.0039 0.0069 0.0025 ”1‘
y95ed9toll 0.0360 0.0092 1“” 0.0145 0.0079 **
y95ed12tol4 0.0132 0.0086 * 0.0009 0.0064
y95ed15 0.0147 0.0132 0.0173 0.0142
y97ed9toll 0.0307 0.0094 **"‘ 0.0018 0.0072
Y97ed121014 -0.0213 0.0011 "‘*"‘ -0.0214 0.0011 *"
Y97ed15 -0.0130 0.0022 1”" -0.0098 0.0040
Y98ed9toll 0.0387 0.0106 "* 0.0070 0.0087
Y98ed12tol4 -0.0029 0.0064 -0.0125 0.0031 **
Y98ed15 -0.0152 0.0020 1‘" -0.0133 0.0026 1'"
Y99ed9toll 0.0504 0.0134 1”” 0.0233 0.0111 1””
Y99ed12tol4 0.0061 0.0101 -0.0083 0.0053
Y99ed15 0.0196 0.0142 " 0.0083 0.0108
Maranhao 0.1102 0.0260 1'" 0.1164 0.0266 *"”" 0.1298 0.0277 1'"
Piaui 0.2587 0.0458 “‘1‘ 0.2684 0.0464 *"‘* 0.2812 0.0471 “'1‘
Ceara 0.1314 0.0288 "* 0.1364 0.0293 1‘" 0.1429 0.0302 **"‘
RGNorte 0.0739 0.0234 ”1‘ 0.0773 0.0240 *** 0.0818 0.0246 1””
Paraiba 0.2028 0.0398 **"' 0.2114 0.0405 1“” 0.2193 0.0411 ***

87

Table 6: (cont’d)

 

 

 

 

P(primtch= 1) Rural
( 1 ) ( 11 ) ( 111 )

dF/dx Std. Err dF/dx Std. Err dF/dx Std. Err
Pemambuco 0.0473 0.0164 *** 0.0523 0.0173 “'1' 0.0553 0.0175 1””
Alagoas 0.1430 0.0339 1”” 0.1507 0.0348 1”” 0.1606 0.0358 "'**
Sergipe 0.1470 0.0367 "”"* 0.1541 0.0376 1‘" 0.1612 0.0384 1'"
Bahia 0.1510 0.0281 *** 0.1589 0.0286 *** 0.1645 0.0288 "*
Minas 0.1470 0.0264 *" 0.1553 0.0269 1”" 0.1752 0.0287 “‘1‘
Espirito 0.0867 0.0293 *** 0.0962 0.0309 “'1' 0.1165 0.0338 ***
Rio 0.0277 0,0132 1“” 0.0292 0.0135 1”” 0.0302 0.0135 ***
SaoPaulo 0.0025 0.0068 0.0049 0.0073 0.0041 0.0070
Parana 0.0535 0.0185 1"” 0.0506 0.0178 *** 0.0408 0.0156 1””
SantaCatarina 0.0287 0.0150 *** 0.0245 0.0139 ** 0.0205 0.0122 **
RGSul 0.0271 0.0117 1”” 0.0264 0.0115 "* 0.0341 0.0129 **"‘
MGSul 0.0190 0.0169 0.0145 0.0161 0.0158 0.0165
MGrosso 0.2122 0.0427 1”” 0.2249 0.0443 1”” 0.2572 0.0468 ”"‘
Goias 0.1356 0.0304 *** 0.1408 0.0311 1”” 0.1198 0.0290 1””
obs. P 0.0518 0.0518 0.0518
pred. P 0.0168 0.0168 0.0168
N of Observations 42,095 42,085 42,085

 

Note: Each speciﬁcation includes an intercept, the controls for human capital characteristics, years and
state effects. The marginal effects signiﬁcantly different from zero at 1 percent, 5 percent and 10 percent
level are indicated with “'1', ** and *.

The structural model for labor force participation.

Wage equations

Already in 1998, changes in average wages were observed.50 The effect of the
reform on primary teacher wages controlling for other observable characteristics is
positive and statistically signiﬁcant. Since the wage difference between primary teachers
and all other workers is one of the main determinants of whether the individual chooses

primary teaching or not, the changes in the wage difference over time may inﬂuence the

 

5" See Sevilla, 2001. (Chapter 2).

88

probability of being a primary teacher. In order to test this hypothesis, I add the predicted
difference between primary teacher and non-primary teacher wages, to capture the effect
of the reform through wages.

Labor supply estimates using the structural model

The results for the structural model are presented in Columns 11 and III for
females (Tables 4, 5, and 6). In Columns 11, the speciﬁcation captures reform effects
through the predicted wage difference between primary teachers and all other workers
(variable diﬂ), and through the year dummy and education group variables interactions.
These interactions allow other reform effects which may affect the decision to go into
teaching that are not captured through the wage difference. In Columns III, I exclude the
interaction terms between year dummies and educational groups to observe if there are
changes in the effects captured speciﬁcally by the wage difference variable (diﬂ).

For female primary teachers, (Table 4, Column 11), when I use the total sample of
females, the predicted wage differential between primary teachers and all other workers
has a positive effect on the likelihood of being a primary teacher. The marginal change
for this variable is 0.0013, but it is not statistically signiﬁcant.

This positive effect becomes larger and statistically signiﬁcant for females in
urban areas (Table 5, Column 11). The estimated effect is 0.0048 and it is signiﬁcant at
the 1 percent level. For example, a 20% increase in the difference between primary and
non-primary teacher hourly natural logarithm of wage rates found would increase by
0.001 the labor force participation of female teachers, i.e. one additional teacher hired for
each 1000 newly hired teachers due to this 20% increase on the relative wage differential

after the reform. For females in rural areas, the effect of the predicted wage differential in

89

the probability of being a primary teacher is -0.0067 and this is statistically signiﬁcant.
These results suggest that while the effect of the reform captured by changes in the
predicted wage difference is positive in urban areas, it is negative for females in rural
areas. The estimated reform effect for rural areas is surprising, considering the fact that
the reform may increase wages for teachers and make this occupation more attractive.
One explanation would be that there might not be enough individuals that would be
qualiﬁed to become teachers. Since the reform effects are captured by the wage
difference variable (diﬂ), the year and education group interactions Show small effects.
Some of the variation explained by these interactions is already captured by the predicted

wage difference variable.51

Conclusions

In this chapter, I study the effects of the F UNDEF reform on the primary teacher
participation behavior. I examine to what extent the likelihood of being a primary
education teacher changed after the reform implementation in two ways: ﬁrst, I apply the
difference-in-differences framework to analyze whether there are differences in the
probability of being a primary teacher across education groups before and after the
reform, and second, I deﬁne the wage difference between primary teacher and all other
workers by regions and education group, and use this predicted wage difference to
measure the effect of the reform on the primary teacher participation.

On average, the difference-in-difference’s estimations suggest that the reform may

have been successful in attracting more qualiﬁed female teachers in the ﬁrst two years of

 

5' This is clear when I exclude the year and education group’s interactions, in Column 111. In that case, the
reform effect through wage remains signiﬁcant and stronger (-0.0133).

90

the reform implementation, but these effects vary by location. The reform effect is
stronger for females in rural areas compared to females in urban areas. However when I
attempt to capture the effect of reform by using predicted wage differences I found a
positive reform effect for females in urban areas and negative reform effect for females in
rural areas. One explanation for these contrasting effects can be that the pool of workers
that can be attracted into teaching is limited in rural areas compared to urban areas. If this
is the case, when the reform increased teacher wages, it was possible to observe
increasing number of individuals going into primary teaching in urban areas. This is
consistent with the estimation results found in this analysis. However, in rural areas the
increases on wages brought by the reform may not be larger enough to attract more
teachers. Qualiﬁed workers from other jobs or from the pool of unemployed in rural areas
might ﬁnd even better wages in other occupations. Nevertheless, since it is difﬁcult to
explain the results obtained for females in rural areas, this speciﬁcation would beneﬁt
from replicating the same empirical framework using different data set.

Even though I do not report the results for males, in urban areas, the estimation
results from both speciﬁcations suggest that the probability of being a male primary
teacher may not be altered by the FUNDEF reform. Again, this might be due to the small
number of cases observed that don’t allow me to drive reasonable conclusions for the
possible effects of the reform for this speciﬁc group.

For the speciﬁcation including the wage differential, the results are not
homogeneous for females from urban and rural areas. I found signiﬁcant effects of the
reform for females in urban areas. It may be the case that resources allocated to primary

education are differently spent, so increases in revenues may be invested in teacher

91

training rather than or in addition to increases on wages. In that case, the size of the wage
differential may be important to influence the participation decision of more qualiﬁed
individuals, particularly in rural areas. This may happen because of other job
opportunities that pay better in rural areas, or because of shortage of qualiﬁed teachers in
those areas.

The results of the labor supply analysis are important. The fact that changing the
distribution of resources and implementing a more equitable school ﬁnance mechanism
also supports making the teaching career more attractive is surprising. Even though the
impact on labor supply were very small, the fact that there was an impact at all, and this
happened after only two years of reform implementation, highlights already
improvements in the education policy that, in the future, might reﬂect on improvements

in the quality of the teaching-leaming process.

92

CONCLUSION

There was a large expansion of the Brazilian system in the last thirty years. This
expansion undermined the quality of the system, which declined over time. This decline
is associated with the expansion developed under a decentralized education structure,
which reinforced an implicit collaboration between systems that provide basic education.
However, the implicit collaboration, under a context of several parallel networks of
schooling offering the same service, ended up leaving decisions about who is responsible
for providing primary education to the discretion of state and municipal politics. The
education system expansion was also based on hiring a large number of low educated
teachers and without monitoring and accountability mechanisms to guarantee the use of
the constitutionally mandated tax revenues on primary education, for example, to train
the unqualiﬁed teachers.

The F UNDEF reform were able to address these problems by signiﬁcantly
changing the average per pupil revenues in primary education in certain school networks,
and by mandating that a speciﬁc share of these funds were invested on teacher training and
wages expenditures. This analysis demonstrated that the new policy produced changes in
the quality of the teacher labor force (as deﬁned by level of education, years of experience,
and average wage rates), even though the effects were not homogeneous in the country. In
fact, the positive impact on teacher wages and on the supply of teachers is still linked to

location and teacher’s level of education.

93

During the ﬁrst two years of reform implementation, there have been signiﬁcant
increases in wage rates for primary teachers in municipal schools compared to non-
primary teacher workers and compared to primary teachers in other school networks.
This may constitute the most important impact of the reform since wages in municipal
schools were below the Optimum, particularly in rural areas. Because of these low wages,
teachers attracted to the system in these areas were less qualiﬁed and used to moonlight,
absenteeism was high and quality of education was low.

The second remarkable ﬁnding is that for teachers with lower quality of
education, there were also increases on wages. In principle, this effect is what we would
like to avoid, but this might serve as incentives for less educated teachers to participate in
in-service trainings, improve their practice and remain in the system. In that case, despite
the low number of education years completed, these teachers may have concluded in-
service trainings and are now receiving higher wages.

Third, the effects were larger in the least developed regions in the country,
reﬂecting the large disparity in ﬁnancing predominant between North and South Regions
and between metropolitan urban areas and distant rural areas, now alleviated after the
reform. Because currently there are virtually no differences in revenues per pupil within
each state based on the main taxes allocated to primary education, the drastic disparities
among public schools revenues were overcame and low revenue municipalities were able
to increase expenditures on teacher wages. In addition, disparities among states were also
minimized since complementary role of federal transfers guarantee a national minimum

standard for expenditures per pupil.

94

Finally, and less expected were the reform effects on the labor supply of primary
teachers. Even though the effects were very small, it is surprising that we observe any
effect at all. It is unexpected because the gap between increases in revenues, policy
reform to improve wages, and dissemination of the results to the prospective teachers in
order to change labor participation decisions is usually large.

FUNDEF has proven a successful policy reform for primary education in Brazil.
However, there are still challenges this reform has to overcome in order to achieve a
positive impact on quality of learning. The ultimate goal of this policy is to reduce
inequalities and to enhance education quality through guaranteeing the means for
implementing minimum standards of quality education. The deﬁnition of the minimum
revenues per pupil based on the availability of resources may not address the needs.
Minimum revenues per pupil do not guarantee per se equal opportunity to learn. Basic
standards of quality schooling demand more than minimum revenues per pupil. If this
minimum is not able to cover all expenses for improving the physical environment of the
schools, enhancing the teaching-leaming and pedagogical conditions, and improving the
quality of the teaching force may not be guaranteed.52 Inequalities were minimized, but
there is the need to deﬁne operational minimum quality standards and assess how much
implementing these standards would cost. New policies should address the problem of
how to sustain and improve the education quality over time under scarcity of resources.

There is also the need to improve the system of educational statistics to guarantee
better data about students, schools, educational programs, and particularly teachers. Any

assessment of educational quality and the evaluation of the impact of policies and programs

 

52 For a conceptual framework of factors inﬂuencing school quality, see Carron, G. and Chau Ta Ngoc
(1996). The qualitLof primary schools in different develogment contexts. UNESCO/IIEP.

95

demand data at school or even classroom level. If a system want to invest in the quality of
its workforce, than a better information system on teacher training, supervision, and
incentives tailored to teacher outputs is necessary.

Another important aspect of the education policy reform process that cannot be
neglected is the relationship between interventions in primary education level and the other
education levels in the system. If guaranteeing resources for primary education limits or
reduces expenditures at other levels of education, this consequence may in the future affect
primary education too. There are several studies that discuss the importance of pre-primary
education for the development of children. Therefore, if this policy reduces the access of
children to pre-primary education, the policy may have negative impacts on the quality of
primary education, as well. Also, expansion of completion rates in primary education may
create larger demand for secondary education that will not be possible to meet, since lack

of resources may lead to less investment in the expansion of secondary schools.

Qualiﬁed teachers at other levels of education may decide to migrate to primary
education decreasing the quality of secondary education. Without good secondary
education, it is possible that low quality students tend to look for less demanding post-
secondary studies, for example, pedagogy and teacher preparation programs. Consequently,
the supply of teachers may be affected. Even though primary education may offer better
wages, neglecting secondary education reform may bring future shortage of qualiﬁed

workers even to attend the primary education.

The impact of the reform in terms of improving equality of revenues and wages of
teachers, and even a slight positive effect on supply of teachers was observed. There are

additional indirect gains with the reform. First, it facilitated the social control over the use

96

of resources exclusively for primary education; accountability is more efﬁcient. The
existence of a unique fund facilitates the identiﬁcation of the revenue sources, the amount
of monies and dates of entrance, the total expenditures, and the revenues of the ﬁnancial
applications. Second, not every state and municipal administrations possess, individually,
the ﬁnancial capacity for covering the full costs of quality education. The conjunction of
state and municipal administration efforts to cover the demand for education through the
FUNDEF ﬁrnd strengthened the collaboration and use of resources in education without
weakening school management autonomy. With resources linked to particular
responsibilities, like educating a speciﬁc number of students, responsibility between
municipal and state administrations within states is clear. Equal distribution of costs and
expenses are possible because the federal government is able to execute its
complementary role of providing minimum resources necessary in those places where
needs and poverty are greater.

The federal government assumes a strong regulatory role in the system. It deﬁnes
the minimum quality, i.e. minimum revenues per pupil spent annually, despite the fact
that the delivery systems is completely decentralized and have the responsibility to
implement the reform. It also intervenes in teacher policies and labor markets setting the
expected minimum wage for teachers in primary education. Even though the focus of the
analysis is on the ﬁnance of the system, the complex relations of all these features of the
education may be mentioned if we are to understand how money may affect the quality of
education. FUNDEF is a successful example of a ﬁrst step toward improving education

quality in the system.

97

REFERENCES

Arcia, G., Alvarez, C., & Scobia, T. (2001). Education Finance and Education Reform:
A Framework for Sustainability. Center for International Development. Research
Triangle Institute. North Carolina.

Baker, D. P. & Smith, T. (1997). Teacher Turnover and Teacher Quality: Refocusing the
Issue. Trend 2. Teachers College Record, 99(1): 29-35.

Barreto, A.M.R.F (1990). Professores do Primeiro Grau: Quem S50, Onde Estcio e
Quanto Ganham. [Primary Teachers: Who they are, Where they are, and How
much they earn]. Brasilia, DF: IPEA/CPS, October.

Becker, G. S. (1964). Human Capital. New York, Columbia University Press, (for
NBER).

Bee, M., & Dolton, P., (1995). The Remuneration of School Teachers: Time Series and
Cross-Section Evidence. The Manchester School, 63(1):l-22.

Behrman, J.R., & Birdsall, N (1983). The Quality of Schooling: Quantity Alone is
Misleading. American Economic Review, 73, pp. 928-946.

Behrman, J .R., Birdsall, N., Kaplan, R., (1996). The Quality of Schooling and Labor
Market Outcomes. In N. Birdsall & R. H. Sabot (Eds), Opportunity Forgone:
Education in Brazil (pp.245-266). Washington, DC: Inter-American
Development Bank.

Birdsall, N., Bruns, B. & Sabot, R. H (1996). Education in Brazil: Playing a Bad Hand
Badly. In N. Birdsall & R. H. Sabot (Eds.), Opportunity Forgone: Education in
Brazil (pp. 7-4 7). Washington, DC: Inter-American Development Bank.

Brasil (1988a) Federal Constitution. Senado Federal, 1988, Article 212.

Brasil (1988b) Federal Constitution. Senado Federal, 1988, Article 212. §5 (Constitutional
Amendment N. 1 4, September 12, 1996).

Brasil (1996a). Senado Federal. Constitutional Amendment N.14, on September 12m,
1996.

Brasil (1996b). Senado Federal. Federal Law N. 9424 (December 24““, 1996)

Brasil (1996c). Senado Federal. Executive Decree N. 2264.

98

Brasil (1997). Senado Federal. LDB: Lei de Diretrizes e Bases da Educaca'o Nacional.
Lei n. 9394, de 1996. [LDB: Law of National Guidelines and Basis of Education.
Law N.9,394, from 1996]. Brasilia, DF: Subsecretaria de EdicOes Te'cnicas.

Brasil (2000). IBGE. Sintese dos Indicatores Sociais — 2000. (www.ibge.gov.br).

Carron, G. and Chau Ta Ngoc (1996). The quality of primary schools in different
development contexts. UNESCO/HEP, Paris.

 

Dourado, L.F. et a1 (1999). Financiamento da Educaca'o Bdsica. Autores Associados,
Campinas, SP.

Fuller, B. & Clark, P. (1994). Raising School Effectiveness While Ignoring Culture?
Review of Educational Research, 64(1): 119-57.

Fuller, B. et al. (1999). How to Raise Children’s Early Literacy? The Inﬂuence of Family,
Teacher, and Classroom in Northeast Brazil. Comparative Educational Review,
43(1): 1-35.

Gomes, C.A. & Sobrinho, J .A (Ed.) (1992). Qualidade, Eﬁciéncia, e Equidade na
Educaca'o Basica, Brasilia, IPEA, pp.123-158.

Gomes, CA (1992) “F inanciamento e Equidade”, In: Gomes, C.A. & Sobrinho, J.A (Ed.)
Qualidade, Eﬁciéncia e Equidade na Educaca'o Basica (Brasilia, IPEA).

Gomes, CA (1997) “Panorama do Ensino Fundamental nas regiOes Norte, Nordeste e
Centro-Oeste”. Unpublished manuscript.

Gomes, CA. (1998) A nova LDB: uma lei de esperanca. Universa, Brasilia, DF.

Gomes, C.A. & Sobrinho, J.A (Ed.) (1992). Qualidade, Eﬁciéncia, e Equidade na
Educacao Basica, Brasilia, IPEA, pp.123-158.

Harbison, R. W., & Hanushek, E. A (1992). Educational Performance of T he Poor:
Lessons from Rural Northeast Brazil. Washington, DC: The World Bank.

Hanushek, E. A., & Rivkin, S. G (1997). Understanding the Twentieth Century Growth of
US. Education Spending. Journal of Human Resources, 32(1): 35-68.

Hanushek, E. A., Kain, J. F., & Rivkin, S. G (1998). Teachers, Schools, and Academic
Achievement (Working Paper No. 6691). Massachusetts, CA: National Bureau of
Economic Research.

International Labor Ofﬁce (ILO) (1991). Teachers in Developing Countries: A Survey of
Employment Conditions. Geneva: ILO.

Lakdwalla, D. ( 2001). The Declining Quality of Teachers. (Working Paper 8263).
Massachusetts, CA. National Bureau of Economic Research.

99

Lam, D (1999). Generating Extreme Inequality: Schooling, Earnings and
Intergenerational Transmission of Human Capital in South Africa and Brazil.
Ann Harbour, MI: Department of Economics and Population Studies Center,
University of Michigan.

Liang, X. (1999). Teacher Pay in 12 Latin American Countries: How does teacher pay
compare to other professions, what determines teacher pay, and who are the
teachers? (Human Development Department Paper Series No. 49). Washington,
DC. The World Bank, Latin America and the Caribbean Regional Ofﬁce

Manski, C. F. (1985) Academic Ability, Earnings, and the Decision to Become a
Teacher: Evidence from the National Longitudinal Study of High School Class of
1972. Working Paper 1539. Cambridge, MA. National Bureau of Economic
Research.

MEC (1993) Plano Nacional de Educacao (MECleEP, 1993).

MEC (1996a) Sinopse Estatistica da Educacao Basica. Censo Educacional I995.
Brasilia, DF: SEEC/MEC.

MEC (1996b) Relatério Alas de encaminhamento, criacdo e regulamentacdo do F undo
de Manutencao e Desenvolvimento do Ensino Fundamental e de Valorizacao do
Magistério. Brasilia, DF: Dezembro, 1996.

MEC (1996c). Perﬁl da Educacao Basica no Brasil. Brasilia, DF: INEP/MEC.

MEC (l997a) Relatério sobre a evoluca'o da Educacao Basica no Brasil: 1991-1997.
Unpublished manuscript.

MEC (l997b) Andlise preliminar acerca do padra'o regional deﬁnanciamento e
dispéndio com educaca'o dos governos subnacionais no Brasil. Unpublished
manuscript.

MEC (l997c) F undo de Manutencao e Desenvolvimento do Ensino Fundamental e de

Valorizacao do Magisterio: Guia para sua Operacionalizacao. Brasilia, DF :
FNDE/MEC.

MEC (l997d) Balanco do Primeiro Ano do F UNDEF . Brasilia, DF: FNDEMEC.
Unpublished manuscript.

MEC (1998a) Sinopse Estatistica da Educacao Basica. Censo Educacional 1997.
Brasilia, DF: INEP/MEC.

MEC (1998b) Informe Estatistico da Educacao Bdsica: Evolucdo Recente das
Estatisticas da Educacao Bdsica no Brasil. Brasilia, DF: INEP/MEC.

MEC (1998c) Sinopse Estatistica do Ensino Superior. Graduacao 1997. Brasilia, DF:
INEP/MEC.

100

MEC (l999a) “Educacao Brasileira: Sintese das Avaliacoes e dos Levantamentos
Estatisticos. Brasilia, DF: INEP/MEC.

MEC (1999b). Sinopse Estatistica da Educacao Basica. Censo Educacional 1998.
Brasilia, DF: lNEP/MEC.

MEC (l999c) Sinopse Estatistica do Ensino Superior. Graduacao 1998. Brasilia, DF:
INEP/MEC.

MEC (l999d) Levantamento sobre a condica'oﬁsica da escola. Brasilia, DF:
INEP/MEC.

MEC (19996). Avalz'acdo do Programa de Complementacdo de Recursos do F undo de
Manuntenca'o e Desenvolvimento do Ensino Fundamental e de Valorizacao do
Magistério (F UNDEF ) (Technical Report). Brasilia, DF: INEP/MEC.

MEC (1999f) Brazilian Education: Policies and Results (Technical Report). Brasilia, DF:
INEP/MEC.

MEC (2000a) Sinopse Estatistica do Ensino Superior. Graduacao 1999. Brasilia, DF:
INEP/MEC.

MEC (2000b). Sinopse Estatistica da Educacao Basica. Censo Educacional 1999.
Brasilia, DF: INEP/MEC.

MEC (2001). Sinopse Estatistica da Educacao Basica. Censo Educacional 2000. Brasilia,
DF: INEP/MEC.

Mincer, J. (1985). Intercountry Comparisons of Labor Force Trends and of Related
DeveIOpments: An Overview. Journal of Labor Economics, Part 2: Trends in
Women's Work, Education, and Family Building. 3(1), Sl-S32.

Monlevade, J. & F erreira, E (1997). O FUNDEF e seus pecados capitais. Ceilandia, DF:
Idéia, 1997.

Mumane, R. J. et. al. (1991). Who will teach? Policies that matter. Cambridge
Massachusetts: Harvard University Press.

Mumane, R. J., Willett, J. B. and Levy, F. (1995) The Growing Importance of Cognitive
Skills in Wage Determination. The Review of Economics and Statistics, 77(2), 251-
266.

Piras, C. and Savedoff, W.D. (1998). How much do teachers earn? (Working Paper Series
No. 375). Washington, DC. Inter-American Development Bank.

Plank, D.N (1990) Issues in Brazilian school ﬁnance. Journal of Education Finance, 15,
Spnng.

101

Plank, D.N., Xavier, A.C. da Ressurreicao & Sobrinho, J .A (1991) “Why Brazil lags
behind in educational development?”, paper presented to the conference on
Education, Growth, and Inequality in Brazil, Rio de Janeiro, March.

Plank, D.N. and Verhine, RE (1995) “Schooling for some: local ﬁnancial commitment to
Basic Education in Bahia, Brazil”. Journals Oxford.

Plank, D. N (1996a) The means of our salvation. Boulder, CO: Westview.

Plank, D. N., Sobrinho, J .A. & Xavier, A.C. da Ressurreicao (1996b) Why Brazil Lags
Behind in Educational Development. In N. Birdsall & R. H. Sabot (Eds),
Opportunity F orgone: Education in Brazil (pp.117-146). Washington, DC: Inter-
American Development Bank.

PNAD/IBGE (1998) Notas MetodolOgicas from the PNAD Data CD-ROM.

Psacharopolous, G (1987). Are teachers overpaid? Some evidence from Brazil. Teaching
& Teacher Education, 3(4), 315-318.

Psacharopolous, G., Valenzuela, J ., Arends, M (1996). Teacher Salaries in Latin America:
A Review. Economics of Education Review, 15(4), 401 -406.

Oliveira, J .B.A.(1992) “Uniéio e municipios e o sistema de ensino fundamental:
atribuicOes e perspectivas,” In: Gomes, C.A. & Sobrinho, J.A (Ed.) Qualidade,
Eﬁciéncia, e Equidade na Educacdo Basica, Brasilia, IPEA, pp.123-158.

Schultz, T. P. (1987). School Expenditures and Enrollments, 1960-1980: The Effects of
Income, Prices, and Population Growth. Population Growth and Economic
Development: Issues and Evidence. D.G. Johnson and R.D.Lee (Eds). Madison,
WI. University of Wisconsin Press. 413-476.

Sedlak, M. & Scholssman, S. (1986). Who will teach? Historical perspectives on the
changing appeal of teaching as a profession. Santa Monica, CA : Rand.

Sevilla, M.A.F (1997) Interview with the Chief of the Planning Unit of the Secretariat of
Education in the state of Para during FUNDESCOLA Planning Workshop. July,
1997.

STN (1998). Boletim — Ano 1998 Fundef, Outubro. Brasilia, DF: COFIN/DIREV/STN.
(hftp://www.tesouro.fazenda.gov.br)

Strauss, J. and Thomas, D (1996). Wages, Schooling and Background: Investments in
Men and Women in Urban Brazil. In N. Birdsall & R. H. Sabot (Eds),
Opportunity F orgone: Education in Brazil (pp. 147-191). Washington, DC: Inter-
American Development Bank.

Tyack, D.; Cuban, L (1995). T inkering Toward Utopia: A Century of Public School
Reform. Massachusetts, CA: Harvard University Press.

102

Vegas, E.L. et a1. (1999) Attracting and Retaining Qualified Teachers in Argentina:
Impact of the Level and Structure of Compensation. (Unpublished manuscript).

Verhine, RE (1998) Financing Public Primary Education in the 21" Century: An
Analysis of the Brazilian FUNDEF Experiment. Paper presented at the CIES,
Toronto — CA, 1998.

Weaver, W. T. (1983). America '5 Teacher Quality Problem: Alternatives for Reform.
New York: Praeger Publishers.

103

APPENDIX

104

APPENDIX A: Description ofTerms

Basic education — Organized predominantly in annual grades, but may occur in
cycles, semester periods, etc. Divided in to early childhood education (pre-school - for
children until three years of age, and kindergarten— for children until six years of age),
primary education (for children above seven years of age), and secondary education (for

children above ﬁfteen years of age).

Primary education - Composed of a minimum of eight years of compulsory

education. Free in all public schools for all Brazilian citizens, independent of age.

Public education — Education provided by public institutions. Public primary
education is provided by state and municipal school networks53 and ﬁnanced by state and

municipal taxes and inter and intra-govemmental transfers.

National educational system — The aggregate educational system. All schools,
institutions and governmental structures related to education. It includes the Ministry of
Education and state and municipal secretaries of education, public foundations, private

and public schools in the Brazilian territory.

School network — Schools under a speciﬁc educational system administration. For
example, state school network is all schools in the system under the ﬁnance and
management of the state government. Municipal school network is all schools in speciﬁc

municipality under the ﬁnance and management of the municipal government. Municipal

 

53 There are still some primary schools under the administration of federal government. They are residual
from the centralized system and correspond mainly to frontiers’ schools.

105

and state school networks follow the state educational legislation, and the national
educational legislation. Municipal school networks also follow speciﬁc municipal

educational legislation that cannot contradict the state and national legislation.

106

 

 

APPENDIX B: Structure of the Brazilian System

Figure 7: Structure of the Educational System

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Doctorate
C)
3
2'
a
(D Master
6th
5th
Higher Education 4th
(Duration from 3 to 6 years) 3rd
2nd
1 st
[ Selective Processes
18 4th year
17 Secondary 3rd year
16 Education 2nd year
15 lst year
14 8th grade
a: 13 7th grade
g. 12 6th grade
E 1(1) Primary Education 23:22:11:
2». 9 3rd grade
3 8 2nd grade
7 lst grade
4 - 6 Kindergarten
‘ Child Education
0 - 3 Pre-school

 

 

 

 

 

 

Source: Adapted from Gomes, 1998.

107

 

 

 

 

 

APPENDIX C: Interpretation of Parameters under the Difference-in-differences (DD)

and Difference-in-difference-in-differences (DDD) Frameworks

The example below is a representation of the methodology applied in the analysis
of independently pooled cross section data used in this study. For every year, we have a
random sample on hourly wages, education, experience, and so on, from the population
of workers and non-workers in Brazil. Since the sample is independently pooled cross
section, we allow the intercept to differ across years to capture differences in the
population distributions over time. The question of interest is whether the teacher wages
have increased with the FUNDEF implementation after controlling for human capital
individual characteristics and others. Figure 8 below shows a positive increase in average
teacher wages over time in Brazil.

Figure 8: Teacher Wages Trend, Brazil 1995-1999

w_ri__—__ . .__, .—____b.._w.__.._ ,,__.—___-

 

 

600

500 s

400 7‘ /—‘J/
300 /

200 .

Brazilian currency RS

100 -

 

1995 1996 1997 1998 1999

 

{*AII pn'mary teachers ]

 

 

 

 

108

However, it is not possible to afﬁrm that this increase on wages was due to the
reform since the growth seems to be a trend that might have happened with other workers
too. For that reason, we include the control group of other workers to observe if the
changes are particular to teacher wages, which ought to be affected by the reform.
Therefore, the question of interest become whether relative teacher wages have changed
before and after the reform implementation. The change on wages due to the reform is
represented in the shaded area below, corresponding to a change in the slope of the trend

line (dashed line).

Figure 9: Teacher Wage Gap Trend, Brazil 1995-I999

 

 

 

 

 

 

 

 

600 —}
500 ‘3'"
“ A 9
3 400 i A e
C I
e V
3 300 1
= l
2 1
'ﬁ 200 l
E .
m i
100 l
l
. l
j 0 __—_"___— I T ' I _% “T‘TW' _.__,_____1
l 1995 1996 1997 1998 1999
ill—ill primary teachers —l—Non-primary teachisj
F g _ .- _ _ -__- _._. __-____

 

 

 

 

A .____. __ _____ _ _.___‘_-____,_ik_+ﬂ___l

The estimation equation is be represented by the model:

In wage = .30 + ,Bly98 + ﬂzprimtch + aly98 * primtch + e (l)

109

The parameter of interest is al that captures (A) the difference between primary
teacher lnwage before and after the reform (ng-Pgb) relative to the same difference
between non-teacher lnwage before and after (093-096), or rewriting the equation, (B) the
relative difference between primary teacher and non-teachers lnwage between 1996 and

1998 (See Figure below).

Figure 10: Difference-in-differences Estimation Speciﬁcation

 

 

 

 

 

 

 

 

Before F UNDEF After FUNDEF A fter—Before
(1996) (1998)

Treatment Group
(Primary teacher) P96 P98 P98'P96
Control Group
(Other workers) 0% 098 098-096
(A) Difference-in-differences estimation (1. = (P98 - P96) — (093-096)
(B) Difference-in-differences estimation (1' = (P93 - 093) — (P96 -096)

 

 

In the equation (I), the speciﬁcation assumes the reform is the unique source of
variation on teacher wages. However, it is likely that municipal and state networks of
schooling responded differently to the gains or looses in revenues brought with
FUNDEF. In order to capture these differentiated reform effects, I estimated the
difference-in-difference-in-difference framework.

The estimation equation is be represented by the model:

In wage = ,60 + ﬂly98 + ,BZprimtch + B3 municipal + aly98 * primtch +
a2 y98 * municipal+ 7| y98 * primtch * municipal+ e

(2)

The parameter of interest is m that captures (A) the difference between primary
teacher lnwage in municipal schools before and aﬁer the reform (ng-Pgb) relative to the

same difference between non-teacher lnwage before and after (098-096) who work for

110

 

municipal administration, compared to the same difference for teachers and workers in
other administrations, or rewriting the equation, (B) the relative difference between
primary teacher and non-teachers lnwage between 1996 and 1998 in municipal
administrations compared to the same difference for workers (teachers and non-teachers)
in other administrations (Figure l I).

Figure 11: Difference-in-difference-in-differences Speciﬁcation

 

 

 

 

 

 

 

 

 

 

 

 

 

After
Bamrflggmm FUNDEF After-Before
(1998)
Treatment Group
(Primary teacher) PM96 PM98 (P984396) M
Municipal Control Group
(M) (Other workers) OM96 OM98 (098-096) A
Difference-in-difference [(ng- 093) - (P96-O%)]M
Treatment Group
(Primary teacher) PA96 PA98 (P 98‘P 96) M
admigitslifitions Control Group 0 O (O -O )
(A) (Other workers) A96 A98 98 96 A
Difference-in-difference [(ng- 093) - (P96-096)]A

 

A D'ff -' - 'ff -. -
( ) dilffeifgceeelgiﬂatiegnce m it = [(P98’ P96)M - (Pos- P96)A] - [(098'096) M ' (096'096)A]

 

 

(B) Difference-in-difference-in-
difference estimation

7: = [(P98' 098) - (P96'O96)]M - [(P98- 098) ' (P96'O96)]A

 

 

The equation (2) identiﬁes relative changes in wages caused by the reform
comparing the effect among the types of administration, municipal and other

administrations.

llI

 

APPENDIX D: Tables

Table 1A: FUNDEF Balance

 

Revenues Sources

 

 

FUNDEF Per Per

Total Pupil Pupil

States Enrollment Federal Averag (1n
FPE“ FPM* IPl-EXP* Transfers 87/96, ICMS“ Total 60” donars)

(2) real) (4)
AC 128650 56038 9706 7 0 707 11655 78112 607 304
AL 530316 68145 42175 355 1300 2822 63273 178070 336 168
AM 543515 45708 24615 1306 O 4194 155247 231071 425 213
AP 108368 55890 7125 110 O 2105 9513 74743 690 345
BA 2822720 153915 163974 13573 112330 14285 396113 854189 303 151
CE 1499510 120182 96025 2082 48202 6030 195529 468050 312 156
ES 533965 24571 32611 10945 0 11920 166924 246970 463 231
GO 1003255 46572 67952 1748 O 0 230900 347171 346 173
MA(2) 1338452 118238 75333 3466 130500 7040 52985 387561 290 145
MG 3468839 72967 240880 32616 0 50628 831643 1228734 354 177
MS 393004 21819 28154 964 0 10252 82564 143752 366 183
MT(2) 493594 37805 35140 2179 0 8912 123776 207811 421 211
PA(3) 1394025 100118 66408 12442 97502 27923 118541 432802 303 152
PB 691314 78445 59740 733 3164 0 80363 222444 322 161
PB 1511462 113029 93758 2456 10535 5318 249024 474119 314 157
P1 602203 70787 45822 434 21418 503 45099 184061 306 153
PR 1651437 47228 126877 . 23714 0 54313 438574 690707 418 209
RJ 1729001 25025 56319 12078 0 27841 948405 1069668 619 309
RN 544131 68436 45118 514 O 0 74297 188366 346 173
R0 275003 46121 15541 151 O 0 44906 106719 388 194
R 62375 40635 5038 43 0 150 10344 56210 901 451
RS 1578410 38573 125550 44212 0 62971 614519 885825 561 281
SC 888794 20964 71380 23908 0 16082 291264 423597 477 238
SE 371886 68066 25491 336 O 0 52833 146726 395 197
SP 5710410 16381 243953 47598 O 9 3446204 3754144 657 329
TO 323127 71092 28255 20 0 0 24369 123735 383 191
Total 30535072 1638053 1838315 237989 424949 314004 8758861 13222041 433 216

 

Source: SIAFI/Secretaria do T esouro Nacional
(2) Enrollment data deﬁned by Portaria n" 319, de 16.04.98.

(3) In the state of Para, the FUNDEF total includes the amount of RS 3.477.422,79 the federal

complementary transfers from July to December of 1997.
(4) Exchange rate US$1.00=R$0.50.
* F undo de Participac'ao do Estado (FPE) - Share of Income and Property tax from rural areas collected at
federal level; F undo de Participacao dos Municipios (PPM) - Share of Income and Property tax from rural
areas collected at federal level; Imposto sobre produtos para exportacao (IPI-EXP) - Export Tax; LC86/97 -
Complementary Sale T ax; Imposto sobre Circulacao de Mercadorias - ICMS - Sale tax collected at state

level.

112

Semﬁmnuzw ﬂue—33-5: 82:05 29:8 .82 .maozmﬁzmo 3&3 cops 05 com .

.38» .838 .momQQs/E ”ooczom

 

113

 

:5.— mwb omw.m $5 @222 Now; oSJN omega Nod—mm . whim? amt—N vvaam N86; 55
$0

36 wow 31...: mad 3.3 5.x code— 56 mda odo— 22 3.3 odo—
034 mm: ovmﬁ on: ”no.3 3w; ommﬁm ownﬂm omiwnv win—cm aﬂvm v2.20.— movdmc; :38.
mod 5% ame— mmw code ems code— wé mda 0.8. 21 mmda ode—
omm mum can com mm; 32.. $06 256 $5.2: mmwdo— mama v2.53” $3.9m moi
mm.m 2.6 8.2 mm.m 3.3 mm.» 8.2: 9v v.3 92: we; wwwa 92:
an N .3 mac mom coca oov avaq 3%.". mmadm E562 oooﬁ ovaﬁmm 3&6: moo—
N_ .v 3 .m 8.3 8% ”mac $8 8.2: .o.m 9.3 92: $.— vvda odo—
ﬂ: mmm mmo mmm 3mg 3m Ste 236 Snéw vmmdw omcé o3.$~ madam 3a—
:r.v $6 3.3 m»; 8.8 ova sodo— cé v.3 92: 22 5.3 odo—
Em Gm new vcm oom.~ mam mwmxa maﬁv owoéa Sada vmné vamdmm wm~._mm coo.
M26 2; 2x: 03. Snow mm.” 8.2: mé 0mm odo— Zl oﬁwa c.8—
nmm mm. own BE v36 ohm 3.ch mnmxa maﬁom SQ;— NEJ‘ cemdmm 3%.va 32
:ogozum m5 .mcozmoo> be :80 >38: 8 8-8 Mwnmmmh men use E283. 295m «.5 one 80:68... So

con—mi 50 banaooom n m .n— _ s a: L8... m z b .52 .88. a (r .52 _ ._r 53»

accuser—r .3 2953.35 . mcetgbmox £3» uEEmmbzm 295m .mEmtO

 

mcozoﬁmom 53> moEEmmbzm new ommm 038mm :38. “<N 03mg.

 

 

 

23.8 2.8 3.8 3.8 $8 $8 5.8 $8 428 38 e333 Becca: ca ammaaea
and «do «2 85 m3 m3 and m3 mg 3c 8:5
25.8 28.8 25.8 398 25.8 28.8 5.8 28.8 5.8 28.8 32.8858 88235
88 3o 88 88 88 8o 86 cod 2:. 88 25592:
:98 25.8 25.8 28.8 898 258 $98 $28 23.8 Go.8 32.032. 88888
86 3o 23 85 cod 3o 88 86 88 85 32.“;
28.8 $8 2:8 2&8 2:.8 2&8 8.8 2&8 22.8 $.18 8.33 .6 8825:
22 3o 88 86 8.9 88 23 8.3 88 3o 8%
8.8 23.8 23.8. 23.8 23.8 22.8 23.8 23.8 28.8 23.8 855% .5885
and ”no one :5 22¢ 28 2.0 83 and So Baa
84.8 28.8 84.8 28.8 32.8 28.8 28.8 28.8 28.8 28.8 £23 .8 5885
$8 and one one one ad and 2.3 and $8 223
28.8 22.8 28.8 23.8 G~8 23.8 28.8 23.8 22.8 23.8 case 80 .5885
:3 and Nod 28 3o 83 ad :3 mad and 22.8
208 5.8 . 28.8 28.8 and 3.8 28.8 58 2a: 2 n8 8:838 52: s 8:253... VEa, co 38>
8; .3 88 Ea a; 8:. m; S.“ 8 .22 Ba 222
22.8 2“: 8 22.8 2 2 8 5.8 2.3 8 , 2228 2a.... : 23.8 2.: : awe.
2.3” 5% 3.: and 3% 2.: on: ”2m 32 2.2 uma
28.8 2:8 28.8 a _ .8 5.8 288 $38 E8 and 2:8 8:828? as;
8.2 a; . 3.2 3.x 8.2 86 8.8 m; 5.2 mm.” 8:828
:3: 22:8 and: 28.88 5.88 2:38 22.88 33: 23.88 688 x83 .8 stoic 2.5:
8.3 2.? 8.8 3.9. 3% 2.8. 8.8 8.9 8 .3 Rd. 32:2
28.8 25.8 28.8 2G8 5.8 28.8 5.8 28.8 28.8 28.8 ee 833% 22553 3532
NS 2.2 :._ m3 2: 8.0 8.9 83 Fe 36 Ewe:
30:23,... mMWMFNMM—H ﬂosomoh. mMWMMMW“. whonumoh ”WWW—0M. 95:0qu mmuhmwm. whozuaoP ammuawﬂom.

metm .52 .0955 .8? basin -=oZ baa—rm .202 2:25.:— -=oZ «33:3

88 :8 88 38 88
82

nm0©~ .. Zthm .muOr—UNDF bmemumrﬁoz ﬂaw muoﬁﬁomor—t Namath how mcomamm>oa ﬁhmﬁgwm ﬁg mime: .mﬂﬁmmnmxw MO COUQtomoQ .<m BAH—GP

ll4

 

 

.88 888 ER 8 .8 88 .8 . m $8 $83 .88 88“ 880 8 8882
.88 . .88 28.8 .88 .88 .88 .88 .88 .58 .88 888 28.... 2.. c. 89.83... 8.88:.
.2... E... m . ... R... 8... R... . .2... 8... 8.... 2... . 28.8
.88 28.8 .88 .88 2...... .88 .88 .88 .88 .58 8.... .88....8 a. 8.8... 8 8.88:.
2.... 8... .8... 8... 8m... 8... mm... 8... 8.... 8... .888...
.88 28.8 .88 28.8 .88 .88 .88 28.8 .88 28.8 8.... 888 a. 88.83.... 8.88:.
.8... 8... .8... 8... 8... 8... 8... 8... 5.... 8 ... 88...
.88 .88 .88 28.8 .88 .88 .....8 .88 .88 28.8 8.... 88.8. a. 22.83 8 8.88:.
S... 8... S... 8... 8... 8... 5... 8... 8... 8... .828.
.28 28.8 .38 28.8 .28 28.8 .38 .88 .28 .38 .88 :88 a. 8.3. 83.8.: 8 8.88:.
8... 8... 8... a... 2.... 8... 8.... 8... 2.... 8... =8...
muonomoh. mmuhhwnwr genome? mMWMMHWML whosomuh mMWMMMWM. 30:23? mMWMMNWM. whosomoh. mmmpmmwm.
ESE...— -:o../. .0555 .8? 8.3.55 .5? 58.5 .5? .0855 A82 o_nm_.m>
88 .8. .8. 88. m8.

 

€8.88 <m £an

115

 

Table 4A: Estimates of the reform effect on primary teachers' wages (Dependent variable: Natural Logarithm of Wage rates)

 

 

lwagel

 

 

educlyr
educ2yr
educ3yr
educ4yr
educSyr
educ6yr
educ7yr
educSyr
educ9yr
educlOyr
educllyr
educl2yr
educ13yr
educl4yr
educ15yr
age

age2
age3
age4
tenure
tenure2
knuﬂe
whhe
black
yeHow
indigenous
union
urbanl
municipal
state
federal
private
pruntch
y95

y97

y98

y99
y95prun
y97prnn
y98prun
y99pnn1
y95mun
y97rnun
y98mun
y99rnun

( I ) ( ll ) ( 111 ) ( 11/ ) ( \/ ) ( \ll ) ( \lll ) ( \lIlI )

Coeﬁ SKlErL Coeﬁ SKLErr Coeﬁ SKlErL Coeﬁ SulErn Coeﬁ SKlEn: Coeﬁ SKLEn: Coeﬁ SKLErn Coeﬁ SKLEU.

-0.081 (0.237) 0.082 (0.010)*** 0.082 (0.010)*** 0.082 (0.010)*** 0.084 (0.010)*** 0.084 (0.010)*** 0.082 (0.010)*** (1084 (0.010)***
—0.233 (0.259) 0.104 (0.007)*** 0.102 (0.007)*** 0.103 (0.007)*** 0.108 (0.007)*** 0.106 (0.007)*** 0 106 (0.007)*** 0.106 (0.007)***
0166 (0202) 0157 (0007)*** 0155 (0007)*** 0156 (0007)*** 0163 (0007)*** 0159 (0006)*** 0159 (0007)*** 0159 (0006)***
-0.147 (0 187) 0.226 (0.006)*** 0.222 (0.006)*** 0.224 (0.006)*** 0.230 (0.006)*** 0.225 (0.005)*** 0.227 (0.006)*** 0.225 (0.005)***
-0.034 (0.191) 0.281 (0.006)*** 0.277 (0.006)*** 0.280 (0.006)*** 0.285 (0.006)*** 0.279 (0.006)*** 0.284 (0.006)*** 0 279 (0.006)***
0.035 (0.196) 0.347 (0.007)*** 0.342 (0.007)*** 0.345 (0.007)*** 0.349 (0.007)*** 0.342 (0.006)*** 0.348 (0.007)*** 0 342 (0.006)***
0.058 (0.190) 0.392 (0.006)*** 0.387 (0.006)*** 0.390 (0.006)*** 0.392 (0.006)*** 0.385 (0.006)*** 0.393 (0.006)*** 0.385 (0.006)***
0.040 (0.185) 0.488 (0.006)*** 0.483 (0.006)*** 0.485 (0.006)*** 0.485 (0.006)*** 0.477 (0.006)*** 0.487 (0.006)*** 0.477 (0.006)***
0.131 (0.186) 0.563 (0.007)*** 0.559 (0.007)*** 0.560 (0.007)*** 0.554 (0.007)*** 0.544 (0.007)*** 0.559 (0.007)*** 0.544 (0.007)***

0.270 (0.185)

0.459 (0.182)**
0.567 (0.183)***
0.683 (0.183)***
0.752 (0.183)***
0.869 (0.183)***
0.221 (0.057)***
-0.008 (0.002)***
0.000 (0.000)***
0.000 (0.000)***
0.016 (0.002)***
0.000 (0.000)

-0.113 (0.016)***
0.060 (0.010)***
-0.008 (0.024)

-0.051 (0.081)

0.178 (0.120)

0.150 (0.010)***
0.130 (0.014)***
—0.172 (0.014)***
-0.117 (0.014)***
0.356 (0.074)***

-0.156 (0.014)***
0.056 (0.014)***
0.189 (0.014)***
0.242 (0.013)***

0.642 (0.007)***
0.879 (0.006)***
1.128 (0.010)***
1.286 (0.011)***
1.377 (0.010)***
1.717 (0.007)***
0.208 (0.009)***
-0.006 (0.000)***
0.000 (0.000)***
0.000 (0.000)***
0.032 (0.001)***
0.000 (0.000)***
-0299 (0.002)***
0.105 (0.003)***
0044 (0.005)***
0.173 (0.022)***
0.054 (0.029)*

0.189 (0.003)***
0.089 (0.004)***

—0.134 (0.012)***
-0.120 (0.004)***
0.037 (0.004)***
0.067 (0.004)***
0.053 (0.004)***
—0.037 (0.016)**

‘ 0.005 (0.016)

0.092 (0.016)***
0.157 (0.016)***

116

0.638 (0.007)***
0.873 (0.006)***
1.124 (0.010)***
1.280 (0.011)***
1.371 (0.010)***
1.712 (0.007)***
0.209 (0.009)***
-0.006 (0.000)***
0.000 (0.000)***
0.000 (0.000)***
0.033 (0.001)***
-0.001 (0.000)***
—0.290 (0.002)***
0.104 (0.003)***
-0.044 (0.005)***
0.170 (0.022)***
0.052 (0.029)*

0.187 (0.003)***
0.081 (0.004)***
-0.145 (0.009)***

-0102 (0.012)***
—0.116 (0.004)***
0.036 (0.004)***
0.063 (0.004)***
0.046 (0.004)***
-0.063 (0.018)***
-0.022 (0.018)

0.023 (0.019)

0.069 (0.018)***
0052 (0.013)***
0.018 (0.013)

0.054 (0.012)***
0.089 (0.012)***

0.638 (0.007)***
0.874 (0.006)***
1.123 (0.010)***
1.280 (0.011)***
1.371 (0.010)***
1.710 (0.007)***
0.209 (0.009)***
-0.006 (0.000)***
0.000 (0.000)***
0.000 (0.000)***
0.032 (0.001)***
0.000 (0.000)***
-0.301 (0.002)***
0.105 (0.003)***
—0.044 (0.005)***
0.174 (0.022)***
0.056 (0.029)*

0.190 (0.003)***
0.091 (0.004)***

0.006 (0.009)

—0.131 (0.012)***
-0.124 (0.004)***
0.034 (0.004)***
0.061 (0.004)***
0.047 (0.004)***
0.022 (0.020)

0.063 (0.019)***
0.156 (0.018)***
0.239 (0.018)***

0.628 (0.007)***
0.862 (0.006)***
1.098 (0.010)***
1.246 (0.011)***
1.339 (0.010)***
1.669 (0.007)***
0.219 (0.009)***
0007 (0.000)***
0.000 (0.000)***
0.000 (0.000)***
0.030 (0.001)***
0.000 (0.000)***
—0.291 (0.002)***
0.104 (0.003)***
-0.046 (0.005)***
0.175 (0.021)***
0.053 (0.029)*

0.181 (0.003)***
0.097 (0.004)***

0.342 (0.011)***

-0.095 (0.012)***
-0.125 (0.004)***
0.038 (0.004)***
0.066 (0.004)***
0.054 (0.004)***
0.033 (0.016)**
0.006 (0.016)

0.096 (0.016)***
0.157 (0.015)***

0.619 (0.007)***
0.849 (0.006)***
1.085 (0.010)***
1.228 (0.011)***
1.320 (0.010)***
1.648 (0.007)***
0.224 (0.009)***
-0.007 (0.000)***
0.000 (0.000)***
0.000 (0.000)***
0.030 (0.001)***
0.000 (0.000)***
-0.289 (0.002)***
0.104 (0.003)***
~0.047 (0.005)***
0.175 (0.021)***
0.052 (0.029)*

0.179 (0.003)***
0.091 (0.004)***
-0.115 (0.009)***
0.038 (0.009)***
0.347 (0.011)***

-0.071 (0.013)***
—0.127 (0.004)***
0.032 (0.004)***
0.053 (0.004)***
0.037 (0.004)***
0.207 (0.030)***
0.208 (0.028)***
0.175 (0.030)***
0.223 (0.026)***
—0.041 (0.013)***
0.024 (0.013)*

0.067 (0.013)***
0.103 (0.012)***

0.637 (0.007)***
0.871 (0.006)***
1.113 (0.010)***
1.268 (0.011)***
1.358 (0.010)***
1.694 (0.007)***
0.214 (0.009)***
-0.006 (0.000)***
0.000 (0.000)***
0.000 (0.000)***
0.030 (0.001)***
0.000 (0.000)***
—0.307 (0.002)***
0.106 (0.003)***
-0.045 (0.005)***
0.179 (0.022)***
0.057 (0.029)**
0.189 (0.003)***
0.095 (0.004)***

-0.045 (0.006)***
—0.140 (0.013)***
-0.102 (0.008)***
0.058 (0.008)***
0.121 (0.008)***
0.120 (0.008)***
-O.116 (0.018)***
-0.076 (0.018)***
0.013 (0.017)

0.068 (0.017)***

0.619 (0.007)***
0.849 (0.006)***
1.085 (0.010)***
1.228 (0.011)***
1.320 (0.010)***
1.648 (0.007)***
0.224 (0.009)***
-0.007 (0.000)***
0.000 (0.000)***
0.000 (0.000)***
0.030 (0.001)***
0.000 (0.000)***
-O.289 (0.002)***
0.104 (0.003)***
—0.047 (0.005)***
0.175 (0.021)***
0.052 (0.029)*

0.179 (0.003)***
0.091 (0.004)***

0.462 (o.014)***
0.153 (0.012)***
0.115 (0.009)***
-0.071 (0.013)***
-0.168 (0.013)***
0.057 (0.013)***
0.120 (0.012)***
0.140 (0.012)***
0.042 (0.024)*

0.032 (0.023)

0.160 (0.021)***
0.221 (0.021)***

 

 

Table 4A (cont‘d)

 

 

 

lwagel (1) (11) (111) (IV) (V) (W) (W) (v111)

“ Coef. Std.Err. Coef. Std.Err. Coef. Std.Err. Coef. Std.Err. Coef. Std.Err. Coef. Std.Err. Coef. Std.Err. Coef. Std.Err.
y955m 0.037 (0.012)*** 0.048 (00mm 0.089 (0.017)***
y97sra 0.032 (0.013)** 0.044 (0.013)*** 0.020 (0.017)
y98sta 0.063 (0.012)*** 0.081 (0.013)*** 0.014 (0.017)
y99sta 0.071 (0.013)*** 0.094 (0.013)*** 0.009 (0.016)
y95fed 0.059 (0.016)*** 0.062 (0.016)*** 0.103 (0.020)***
WM 0001 (0.016) 0.005 (0.016) 0.019 (0.020)
y98fed 0.025 (0.015) 0.035 (0.015)** 0.032 (0.019)*
y99fed 0.033 (0.016)” 0.048 (0.016)*** 0.055 (0.019)***
395an 0.025 (0.009)*** 0.041 (0.013)***
y97priv 0.027 (0.009)*** 0.024 (0.013)*
y98priv -0.069 (0.009)*** 0.067 (0.013)***
y99priv 0.084 (0.009)*** 0.103 (0.012)***
y95primun 0.114 (0.025)*** -0.166 (0.034)***
y97primun 0.064 (0.024)*** 0.176 (0.032)***
y98primun 0.146 (0.023)*** 0.015 (0.032
y99primun 0.161 (0.022)*** 0.002 (0.028)
y959rista 0.153 (0.024)*** 0.394 (0.032)*** 0.228 (0.027)***
y97prim 0.158 (0.024)*** 0.357 (0.031)*** -O.181 (0.027)***
y98prista 0.202 (0.023)*** 0.273 (0.032)*** 0.257 (0.025)***
y99prim 0.263 (0.022)*** 0.300 (0.029)*** 0.297 (0.024)***
y95prifed 0.089 (0237) 0.175 (0.238) 0.010 (0.238)
y97prifed 0.005 (0181) 0.220 (0.182) 0.044 (0.181)
y981>rifed 0.036 (0149) 0.137 (0.149) 0.122 (0.148)
y99prifed 0.264 (0175) 0.173 (0.176) 0.175 (0.175)
y95pripriv 0.383 (0.030)*** 0.166 (0.034)***
y97pripriv 0.345 (0.028)*** 0.176 (0.032)***
y98pripriv 0.221 (0.030)*** 0.015 (0.032)
y99pripriv 0.215 (0.026)*** 0.002 (0.028)
Rondonia 0.533 (0.046)*** 0.280 (0.014)*** 0.271 (0.014)*** -0.281 (0.014)*** 0.247 (0.014)*** 0.242 (0.014)*** 0.280 (0.014)*** 0.242 (0.014)***
Acre —0.635 (0.058)*** 0.350 (0.023)*** 0.340 (0.023)*** 0.357 (0.023)*** 0.304 (0.023)*** 0.307 (0.023)*** 0.358 (0.023)*** 0.307 (0.023)***
Amazonas 0.348 (0.040)*** 0.345 (0.012)*** 0.333 (0.012)*** 0.345 (0.012)*** 0.304 (0.011)*** 0.293 (0.011)*** 0.343 (0.012)*** 0.293 (0.011)***
Roraima 0.250 (0.079)*** 0.120 (0.023)*** 0.108 (0.023)*** 0.126 (0.023)*** 0.121 (0.024)*** 0.123 (0.023)*** 0.137 (0.023)*** 0.123 (0.023)***
Para 0.757 (0.036)*** 0.521 (0.009)*** 0.510 (0.009)*** 0.521 (0.009)*** 0.482 (0.009)*** 0.471 (0.009)*** 0.518 (0.009)*** 0.471 (0.009)***
Amapa 0.181 (0.056)*** 0.187 (0.023)*** 0.173 (0.023)*** 0.189 (0.023)*** -O.176 (0.023)*** 0.172 (0.022)*** 0.200 (0.023)*** 0.172 (0.022)***
Tocantins 0.654 (0.039)*** 0.490 (0.013)*** 0.474 (0.013)*** 0.491 (0.013)*** 0.441 (0.013)*** 0.431 (0.013)*** 0.494 (0.013)*** 0.431 (0.013)***
Maranhao 0.809 (0.042)*** 0.687 (0.015)*** —0.669 (0.015)*** —0.689 (0.015)*** 0.640 (0.015)*** 0.625 (0.014)*** -0.687 (0.015)*** 0.625 (0.014)***
Piaui 0.970 (0.038)*** 0.763 (0.014)*** 0.747 (0.014)*** 0.764 (0.014)*** 0.717 (0.013)*** 0.706 (0.013)*** 0.765 (0.014)*** -0.706 (0.013)***
Ceara 0.891 (0.034)*** 0.622 (0.008)*** 0.612 (0.008)*** 0.622 (0.008)*** 0.574 (0.008)*** 0.560 (0.008)*** 0.614 (0.008)*** 0.560 (0.008)***
RGNone -1.086 (0.046)*** 0.721 (0.012)*** 0.707 (0.012)*** 0.721 (0.012)*** 0.683 (0.012)*** -O.671 (0.012)*** 0.721 (0.012)*** -0.671 (0.012)***
Paraiba 0.955 (0.042)*** 0.666 (0.012)*** 0.646 (0.012)*** -0.667 (0.012)*** 0.633 (0.012)*** 0.616 (0.012)*** -0.671 (0.012)*** —0.616 (0.012)***
Pemambuco 0.830 (0.032)*** 0.612 (0.008)*** 0.601 (0.008)*** 0.611 (0.008)*** 0.566 (0.008)*** 0.553 (0.008)*** 0.605 (0.008)*** 0.553 (0.008)***
Alagoas 0.886 (0.044)*** -0.656 (0.013)*** 0.638 (0.013)*** 0.656 (0.013)*** 0.602 (0.013)*** 0.587 (0.013)*** 0.654 (0.013)*** 0.587 (0.013)***
Sergipe 0.682 (0.045)*** 0.589 (0.012)*** 0.572 (0.012)*** 0.589 (0.012)*** 0.541 (0.012)*** 0.527 (0.012)*** 0.589 (0.012)*** 0.527 (0.012)***
Bahia 0.754 (0.031)*** 0.543 (0.007)*** 0.532 (0.007)*** 0.543 (0.007)*** 0.493 (0.007)*** 0.481 (0.007)*** 0.536 (0.007)*** 0.481 (0.007)***
Minas 0.381 (0.030)*** 0.405 (0.007)*** 0.396 (0.007)*** 0.402 (0.007)*** 0.349 (0.007)*** 0.335 (0.007)*** 0.393 (0.007)*** 0.335 (0.007)***

117

 

 

Table 4A (cont‘d)

 

 

 

lwavel (1) (11) (111) (IV) (V) (VI) (VII) (VIII)

5 Coef. Std.Err. Coef. Std.Err. Coef. Std.Err. Coef. Std.Err. Coef. Std.Err. Coef. Std.Err. Coef. Std.Err. Coef. Std.Err.
Espirito -0.376 (0.041)*** -0.389 (0.011)*** -0.375 (0.011)*** -0.386 (0.011)*** -0.332 (0010)*** -0.315 (0.010)*** -0.380 (0.010)*** -0.315 (0.010)***
Rio 0436 (0.031)*** —0.347 (0.007)*** -0.339 (0.007)*** -0.344 (0.007)*** -0.302 (0.007)*** -0.287 (0.007)*** -0.334 (0.007)*** -0.287 (0.007)***
SaoPaulo -0.226 (0.030)*** -0.095 (0.007)*** -0.088 (0.007)*** -0.092 (0.007)*** -0.030 (0 007)*** -0.016 (0.007)** —0.079 (0.007)*** -0.016 (0.007)**
Parana -0.343 (0.032)*** -0.267 (0.007)*** —0.256 (0.007)*** —0.265 (0.007)*** —0.209 (0007)*** -0.193 (0.007)*** -0.254 (0.007)*** -0.193 (0.007)***
SantaCat -0.516 (0.037)*** -0.236 (0.009)*** -0.227 (0.009)*** -0.233 (0.009)*** -0.174 (0 008)*** -0.159 (0.008)*** -0.222 (0.008)*** -0.159 (0.008)***
RGSul —0.458 (0.031)*** -0.302 (0.007)*** -0.294 (0.007)*** -0.299 (0.007)*** -0.249 (0 007)*** -0.234 (0.007)*** —0.287 (0.007)*** -0.234 (0.007)***
MGSul «0.587 (0.039)*** -0.463 (0.011)*** -0.449 (0.011)*** -0.462 (0.011)*** -0.420 (0.010)*** -0.406 (0.010)*** -0.458 (0.010)*** -0.406 (0.010)***
MGrosso -0.478 (0.037)*** -0.336 (0.010)*** -0.323 (0.010)*** -0.334 (0.010)*** -0.283 (0.010)*** -0.269 (0.010)*** -0.329 (0.010)*** -0.269 (0.010)***
Goias -0.652 (0.033)*** —0.470 (0.008)*** —0.459 (0.008)*** -0.469 (0.008)*** -0.418 (0 008)*** -0.406 (0.008)*** -0.463 (0.008)*** -0.406 (0.008)***
constant -1.474 (0.521)*** -2.351 (0.076)*** -2.363 (0.076)*** -2.357 (0.076)*** -2.521 (0.076)*** -2.556 (0.075)*** -2.363 (0.076)*** -2.671 (0.076)***
N Obs 13,859 302.172 302,172 302,172 302,172 302,172 302,172 302,172
R-squared 0.541 0.576 0.578 0.577 0.584 0.586 0.578 0.586
Root MSE 0.493 0.597 0.596 0.596 0.591 0.590 0.595 0.590

 

118

 

 

Table 5A: Estimates of the Reform Effect on Primary Teachers' Wages

(Dependent Variable: Natural Logarithm of Wage Rates)

 

 

A11 Sample North Region

lwagel ( 1 ) ( II )

Coef. Std. Err. Coef. Std.Err.
educlyr 0.083 0.010‘" 0.137 0037*”
edchyr 0.106 0007*” 0.110 0.028 ***
educ3yr 0.160 0006*" 0.138 0.025 "*
educ4yr 0.227 0005*" 0.177 0.021 ***
educSyr 0.281 0006*" 0.241 0021’"
educ6yr 0.344 0006*" 0.289 0.025 ***
educ7yr 0.387 0006"" 0.331 0.023 ***
educByr 0.479 0006*" 0.422 0.021 **"‘
educ9yr 0.547 0007*" 0.488 0026*"
educlOyr 0.622 0007*“ 0.578 . 0.025 ***
educl lyr 0.853 0006*” 0.823 0020*”
educ12yr 1.087 0.010‘” 1.105 0.038 *"
educl3yr 1.233 0011*” 1.262 0.040"“""I
educl4yr 1.321 0010*" 1.321 0046*"
educ15yr 1.652 0007*" 1.677 0.025""""I
age 0.224 0009*" 0.192 0038*"
age2 —0.007 0000*" -0.005 0.002 ""'
age3 0.000 0000*“ 0.000 0.000
age4 0.000 0000*“ 0.000 0.000
tenure 0.030 0001*" 0.023 0.002 ”*
tenure2 0.000 0000*" 0.000 0.000 *"”"
female -0.288 0.002 ”* -0.242 0010‘"
white 0.104 0.003 "”” 0.085 0010""
black -0.047 0005"" -0.048 0.025 "
yellow 0.174 0021*" 0.118 0.095
indigenous 0.051 0.029 "‘ 0.072 0.090
union 0.179 0003*" 0.181 0012“"
urbanl 0.090 0004"" -0.042 0024"
federal 0.376 0005*" 0.490 0.019 "“""'
state 0.065 0004*“ 0.152 0014*"
municipal -0.072 0004*" 0.029 0.014”
primtch -0.090 0012*” -0.020 0.048
Para -0.471 0009"" -0.276 0.022 “W
primtch*Para -0.077 0.064 -0.233 0079*"
y95 -0. 121 0004*" -0.104 0014*"
y97 0.039 0004*“ 0.024 0.015 *
y98 0.068 0.003 *** 0.033 0.014 ”
y99 0.057 0.003 *" 0.045 0014*"
y95*primtch -0.042 0.016" -0.111 0.063 "'
y97*primtch 0.006 0.016 -0.138 0.064"
y98*primtch 0.101 0016*“ -0.001 0.063
y99*primtch 0.162 0016*" 0.077 0.061

 

119

Table 5A (cont’d)

 

 

A11 Sample North Region

lwagel ( I ) ( II )

Coef. Std. Err. Coef. Std.Err.
y95*primtch*Para 0.167 0086* 0.219 0.105”
y97*primtch*Para 0.080 0.083 0.241 0.104“
y98*primtch*Para -0.027 0.089 0.102 0.108
y99*primtch*Para 0.086 0.084 0.192 0.102 *
Rondonia -0.242 0014*" -0.053 0.025"
Acre -0.307 0.023 *** -0. 141 0.031 ***
Amazonas -0.295 0.011 **"‘ -0.105 0.023 ***
Roraima -0.125 0.023 *** 0.041 0.031
Amapa -0.172 0.022 *** ***
Tocantins -0.433 0.013 **"' -0.277 0.025 “‘*
Maranhao -0.624 0014*” ***
Piaui -0.706 0.013 *** **"'
Ceara -0.560 0.008 *** *“'
RGNorte -0.672 0.012 *** **"'
Paraiba -0.616 0012*“ ***
Pemambuco -0.553 0.008 *** *"‘*
Alagoas -0.587 0.013 *” ***
Sergipe -0.528 0012*” *"”"
Bahia -0.480 0007*” "*
Minas -0.338 0007*" **"‘
Espirito -0.317 0010*" ***
Rio -0.288 0007*" ***
SaoPaulo -0.018 0007*” **"‘
Parana -0.195 0007*“ ***
SantaCat -0.161 0008*“ ”*
RGSul -0.237 0007*" *"
MGSul -0.407 0010*" H"
MGrosso -0.272 0010"" **"‘
Goias -0.408 0.008 “‘"' *”
constant -2.563 0075*" -2.491 0308*"
Prob 302,172 20,691
R-squared 0.585 0.561
Root MSE 0.591 0.616

 

120

 

 

Table 6A: Estimates ol‘ the ret‘omi effect on primary teachers' wages (Dependent variable: Natural Logarithm of Wage rates)

 

 

 

[WM] (1) (11) (111) (1V)
5 Coef. Std.Err. Coel‘. Std.Err. Coef. Std.Err. Coef. Std.Err.

educlyr 0.064 0.023 *** 0.066 0.023 *** 0.121 0.249 0.125 0.243
educ2yr 0.087 0.018*** 0.085 0.018*** 0.280 0.232 0.286 0.231
educ3yr 0.158 0.016*** 0.155 0.016*** 0.085 0.208 0.083 0.206
educ4yr 0,233 0.013W 0.226 0.013 *** 0.162 0.192 0.154 0.190
educ5yr 0.297 0.015*** 0.289 0.015*** 0.153 0.195 0.154 0.194
educ6yr 0.361 0.019*** 0.352 0.019*** 0.009 0.198 0.008 0,197
educ7yr 0.413 0.018 *** 0.403 0.018*** 0.022 0.195 0.025 0.194
educ8yr 0.539 0.014*** 0,525 0.014*** 0.061 0.190 0.059 0.189
educ9yr 0.646 0.019*** 0.633 0.019*** 0.153 0.192 0.149 0.191
educlOyr 0.681 0.017*** 0.666 0.017*** 0.236 0.190 0.228 0.188
601161 lyr 0.949 0.013*** 0.933 0.013 *** 0.470 0.188** 0.459 0.186**
educ12yr 1.122 0.018*** 1.110 0.018*** 0.585 0.188*** 0.574 0.187***
educl3yr 1.228 0.021*** 1.214 0.021*** 0.683 0.18 0.671 0.188***
educl4yr 1.332 0.018*** 1.322 0.018*** 0.760 0.189*** 0.750 0.187***
educ15yr 1.637 0.014*** 1.624 0.014*** 0.867 0188*” 0.857 0.187***
age 0.149 0.025*** 0.143 0.025*** 0.155 0.059*** 0.162 0.059***
age2 0.004 0.001*** 0.004 0.001*** 0.005 0.002** -0.006 0.002**
age3 0.000 0.000*** 0.000 0.000*** 0.000 0.000** 0.000 0.000**
age4 0.000 0.000*** 0.000 0.000** 0.000 0000* 0.000 0000*
tenure 0.014 0.001*** 0.013 0.001*** 0.012 0.002*** 0.013 0.002***
tenure2 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000
female 0.305 0.006*** 0.304 0.006*** 0.081 0.014*** -0.081 0.014***
white 0.089 0.006*** 0.090 0.006*** 0.042 0.010*** 0.041 0.010***
black 0.031 0.011*** 0.029 0.011*** 0.002 0.024 0.002 0.024
yellow 0.131 0.043 *** 0.127 0.043 *** 0.004 0.062 0.001 0.063
indigenous 0.116 0068* 0.121 0068* 0.163 0.109 0.159 0.108
union 0156 0,006*** 0.156 0.006*** 0.117 0.009*** 0.118 0.009***
urbanl 0.103 0.008*** 0.113 0.008*** 0.149 0.014*** 0.150 0.014***
municipal 0.149 0.005*** 0.162 0.012*** 0.061 0.010*** 0.109 0.021***
primtch 0.129 0.014*** 0.125 0.014*** -0.080 0.025 *** 0.077 0.024***
y95 0.128 0.009*** 0.091 0.012*** 0.155 0.031*** 0.143 0.033***
y97 0.057 0.009*** 0.066 0.012*** 0.045 0.030 0.035 0.034
498 0.113 0.009*** 0.118 0.012*** 0.160 0.029*** 0.134 0.032***
499 0.125 0.009*** 0.119 0.012*** 0.176 0.028*** 0.133 0.032***
y95*primtch 0.032 0018* 0.120 0.021*** 0.006 0.035 0.013 0.036
y97*primtch 0.008 0.018 0.081 0.021*** 0.010 0.034 0.015 0.037
y98*primtch 0.087 0.018*** 0.025 0.021 0.051 0.032 0.035 0.035
499*primtch 0.131 0.018*** 0.002 0.021 0.092 0.032*** 0.065 0.035 *
y95*municipa1 0.084 0.017*** -0.038 0.055
y97*municipa1 0.018 0.017 0.025 0.049
y98*municipa1 0.009 0.016 0.069 0.046

W 0.014 0.016 0.124 0.043 ***

 

 

 

 

 

Table 6A (cont‘d)

 

 

 

lwagel (I) (111) (IV)

‘ Coef. Std.Err. Coef. Std.Err. Coef. Std.Err. Coef. Std.Err.
y95"primtch*mun1cipal 0.203 0026*” 0.025 0.056
y97*primtch*municipal 0.157 0036*“: -0.013 0.049
y98"primtch*municipal 0.227 0.02-1*“ 0.023 0.045
y99*primteh*munieipal 0.261 0.023 *** 0.023 0.042
Rondonia -0.397 0028*“ -0.393 0028*” -0.571 0.040“M -0.57l 0.0-10*“
Acre -0.459 0033*” -0.458 0.033“M —0.708 0.052“M -0.710 0.052“M
Amazonas «0.488 0.023 *** -0.483 0.023 *** -0.399 0040*M -0.401 0040*“
Roraima -0.241 0036*” -0.233 0.036*** 0.317 0087*M -0.310 0.088"M
Para -0.651 0.019*** -0.648 0019*M ~0.784 0034*M -0.786 0034*”
Amapa -0.279 0039*“ -0.270 0.039"M —0. I96 0.056“M -0. 196 0056*“
Tocantins -0.609 0.021 *** —0.604 0021*“ -0.716 0035*“ —0.721 0.035***
Maranhao -0.794 0.024“M -0.795 0.024*** -0.900 0039*“ —0.903 0039*M
Piaui -0.923 0.023“M -0.918 0.023"M -l.074 0.035“M -l.075 0.035***
C3913 -0.794 0.017*** -0.794 0017*“ —0.963 0032*“ -0.965 0.031 ***
RGNorte ~0.970 0.024*** -0.964 0024*“ -l .170 0042*” -l .174 0.042***
Paraiba —0.867 0.021*** -0.863 0.021*** — | .045 0037*” - l .046 0.037***
Pemambuco -0.722 0016*“ —0.720 0.016*** -0.851 0029*“ -0.853 0.029***
Alagoas -0.824 0.024*** —0.822 0.024*** -0.998 0.043 *** —l .001 0.042“M
Sergipe —0.694 0.022*** -0.689 0022*M -0.770 0038*” —0.774 0.038“M
Bahia -0.682 0.015"M —0.683 0.015*** -0.840 0028*M —0.844 0.027***
Minas -0.504 0.014*** -0.499 0.014*** -0.479 0.027*** -0.481 0.026***
ESpirito -0.418 0.021*** -0.411 0.021*** -O.479 0036*“ -0.480 0.036***
Rio -0.525 0.016*** —0.524 0.016*** -0.499 0029*” -0.500 0.029***
SaoPaulo -0.314 0.014*** -0.306 0.014*** -0.323 0.027*** —0.326 0.026***
Parana -0.457 0.016*** -0.455 0.016*** —0.429 0.028“M -0.426 0.028***
SantaCat -0.406 0.019*** -0.399 0019*“ -0.606 0034*” —0.606 0.033 ***
RGSUI -0.430 0.016*** -0.426 0.015*** -0.583 0028*” -0.583 0.027***
MGSul 0.672 0.021*** -0.668 0.021*** 0.704 0.035*** 0.702 0.035 ***
MGFOSSO -0.473 0020*M —O.467 0.019*** -0.566 0034*” -0.567 0.034***
GOiaS -0.664 0.017*** —0.659 0.017*** -0.793 0029*“ -0.793 0.029***
CODStant -1.413 0.213*** -1.358 0.213*** -0.818 0.546 -O.843 0.545
N of Observations 61895 61895 14576 14576
R‘Squared 0.587 0.589 0.563 0.565
Root MSE 0.602 0.600 0.484 0.483

 

122

f‘""‘\

 

 

 

 

 

Table 7A: Estimates of the reform effect on primary teachers' wages by level of education (Dependent variable: Natural Logarithm of Wage rates)
Primary teachers All teachers All teachers All workers All workers
lwagel (I) (11) (111) (IV) (V)

Coef. Std.Err. Coef. Std.Err. Coef. Std.Err. Coef. Std.Err. Coef. Std.Err.
ed9toll 0.434 0022*” 0.430 0018*” 0.441 0065*“ 0.485 0.003*** 0.472 0.006***
ed12tol4 0.597 0.026*** 0.640 0022*M 0.653 0.081*** 0.838 0.006*** 0.831 0.016***
ed15 0.826 0.024*** 0.884 0.020*** 0.935 0062*M 1.287 0.005*** 1.302 0.010***
age 0.288 0.057*** 0.302 0.049*** 0.279 0.049*** 0.230 0.010*** 0.228 0.010***
age2 -0.010 0002*“ -0.01 l 0002*“ -0.010 0.002*** -0.007 0.000*** -0.007 0.000***
age3 0.000 0.000*** 0.000 0.000*** 0.000 0.000*** 0.000 0.000*** 0 000 0.000***
414.94 0.000 0.000*** 0.000 0.000*** 0.000 0.000*** 0.000 0.000*** 0.000 0.000***
tenure 0.016 0002*“ 0.018 0.002*** 0.018 0.002*** 0.031 0.001*** 0.031 0.001***
tenure2 0.000 0000* 0.000 0.000*** 0.000 0.000*** 0.000 0.000*** 0.000 0.000***
female -0.117 0.016*** -0.l73 0.013*** —0.168 0.013*** -0.269 0.002*** -0.271 0.002***
white 0.063 0.010*** 0.082 0.009*** 0.080 0.009*** 0.126 0.003*** 0.125 0.003***
black -0.009 0.024 -0.040 0022* -0.038 0022* -0.055 0.005 *** -0.054 0.005***
Yellow -0.049 0.083 0.071 0.064 0.062 0.063 0.206 0.022*** 0.193 0.022***
indigenous 0.152 0.114 0.118 0.093 0.106 0.094 0.046 0.030 0.046 0.030
L1111011 0.154 0.010*** 0.160 0.009*** 0.158 0.009*** 0.198 0.003*** 0.200 0.003***
urbanl 0.146 0.014*** 0.142 0.013*** 0.153 0.013*** 0.130 0.004*** 0.139 0004*“
federal 0.355 0.075*** 0.213 0.042*** 0.182 0.042*** 0.397 0.006*** 0.385 0.006***
state -0.117 0.014*** -0.164 0.011*** -0.176 0.012*** 0.087 0.004*** 0.089 0.004***
municipal -0.176 0.014*** —0.201 0.012*** -0.211 0.012*** -0.077 0.004*** -0.088 0.004***
primtch -0.110 0.021*** -0.079 0.070 -0.077 0.012*** 0.157 0.044***
1’95 -0.157 0.014*** -0.191 0.026*** -0.234 0.092** —0.127 0.004*** -0.140 0.004***
y97 0.058 0.014*** 0.029 0.026 0.028 0.079 0.042 0.004*** 0.043 0.005***
y98 0.193 0.014*** 0.094 0.025*** 0.138 0.086 0.071 0.004*** 0.065 0.004***
y99 0.247 0.013 *** 0.100 0.025*** 0.094 0.078 0.064 0.004*** 0.066 0.004***
y95prim 0.035 0.030 0.000 0.109 -0.033 0.016** -0.090 0.062
y97prim 0.026 0.029 0.043 0.099 0.010 0.016 0.021 0.063
y98prim 0.093 0.028*** 0.157 0.106 0.103 0.016*** 0.207 0.062***
y99prim 0.141 0.028*** 0.340 0.096*** 0.167 0.016*** 0.353 0.060***
pri9t011 0.003 0.079 -0.048 0.046
pril2t014 0.026 0.096 -0.178 0.057***
pri15 —0.079 0.076 -0.564 0.049***
y95ed9t011 -0.020 0.104 0.027 0.008***
y95ed12tol4 0.139 0.128 0.039 0023*
y956d15 0.081 0.099 0.042 0.014***
y97ed9t011 -0.090 0.091 -0 003 0.009
y97edl2t014 0.068 0.116 -0.005 0.022
y97ed15 0.045 0.086 -0.004 0.014
y98ed9toll —0.192 0.098 * -0.001 0.008
y986d12tol4 0.035 0.117 0.000 0.022
y98ed15 0.023 0.093 0.064 0.014***
y99ed9toll -0.042 0.090 -0.021 0.008***
y99ed12t014 0.023 0.109 0.018 0.021
y99ed15 0.033 0.085 0.046 0.014***

123

 

 

 

 

 

Table 7A (eont‘d)
Primary teachers All teachers All teachers All workers All workers
[“3861 (I) (11) (111) (IV) (V)

Coef. Std.Err. Coef. Std.Err. Coef. Std.Err. Coef. Std.Err. Coef. Std.Err.
y95pri91011 0.114 0.121 0.057 0.066
)‘95pril2tol4 -0.100 0.147 -0.043 0.079
y95pr115 0.008 0.117 0.043 0.069
y97pri9toll 0.051 0.111 —0.030 0.067
y97pri|2tol4 -0. 126 0.137 -0.065 0.080
y97pr115 —0.015 0.108 0.046 0.070
y98pri9toll 0.114 0.117 -0.064 0.066
y98pril2tol4 -0.l79 0.138 -0.123 0.079
y98pri15 0.173 0.113 0.173 0.068**
y99pri9toll —0.108 0.108 -0.116 0063*
1091311131014 -0.359 0129*” -0.344 0.075***
y99pr115 -0.255 0.104M -0.233 0.066***
ROUdOHia —0.541 0.046*** -0.449 0.039*** -0.448 0.039*** -0.257 0.014*** -0.258 0.014***
Acre -0.662 0.059“M -0.584 0.052*** -0.582 0.053*** -0.334 0.023*** -0.334 0.023***
A”132011215 -0.364 0.040*** —0.303 0.038*** -0.310 0038*“ -0.298 0.012*** -0.300 0.012***
Roraima -0.270 0.080*** -0.235 0.072*** -0.227 0.072*** -0.141 0024*M -0. 144 0024*”
Para -0.772 0.037*** -0.674 0.033 *** —0.683 0.033*** —0.482 0.009*** —0.486 0.009***
Amapa -0.189 0.056*** ~0. 124 0.052** -0.128 0.052** —0. 191 0023’"M —0.l94 0.022***
Tocantins —0.669 0.040*** -0.614 0.037*** -0.619 0.036*** -0.450 0.013 *** -0.459 0.013 ***
Maranhao -0.836 0.042“M -0.816 0.039*** -0.817 0.038*** -0.629 0.015*** -0.640 0.015***
Piaui 0.993 0.038*** 0.963 0.035*** 0.969 0.035*** 0.741 0.014*** 0.751 0.014***
C3313 -0.910 0.034*** —0.869 0.030*** -0.865 0030*” -0.588 0.008*** -0.591 0.008***
RGNOITB -l .093 0.046*** -1.026 0.041*** -1.028 0.040*** -0.704 0.012*** -0.704 0.012***
Paraiba -O.970 0.042*** -0.930 0.037*** -0.930 0.037*** -0.656 0.012*** -0.660 0.012***
Pemambuco -0.835 0.033*** -0.785 0.029*** -0.790 0.028*** -0.571 0.008*** -0.571 0.008***
Alagoas -0.900 0.045*** 0.887 0.041*** -0.894 0.040*** -0.623 0.013*** -0.626 0.013***
Sergipe -0.693 0.045*** -0.709 0.037*** -0.699 0.037*** -0.558 0.012*** -0.561 0.012***
Bahia -0.759 0.031*** -0.711 0.028*** -0.722 0.028*** -0.494 0.007*** -0.500 0.007***
M11135 -0.388 0.030*** -0.380 0.027*** -0.375 0.027*** -0.348 0.007*** -0.347 0.007***
ESpirito -0.382 0.042*** -0.343 0.037*** -0.337 0.037*** -0.323 0.011*** -0.323 0.01 l ***
Rio —0.436 0.031*** -0.437 0.028*** —0.441 0.028*** -0.281 0.007*** -0.280 0.007***
SaoPaulo 0.244 0.030*** 0.243 0.026*** 0.241 0.026*** 0.024 0.007*** 0.022 0.007***
parana 0.355 0.032*** 0.360 0.028*** 0.358 0.028*** 0.210 0.007*** 0.208 0.007***
SantaCat 0.529 0.037*** 0.502 0.032*** 0.502 0.032*** 0.175 0.009*** 0.172 0.009***
RGSul 0.460 0.031*** 0.454 0.028*** 0.452 0.028 *** 0.239 0.007*** 0.235 0.007***
M63111 -0.610 0.040*** —0.581 0.036*** -0.573 0.036*** —0.433 0.011*** -0.429 0.011 ***
MGTOSSO —0.493 0.037*** -0.462 0.033*** -0.460 0.033*** -0.286 0.010*** -0.285 0.010***
601143 0656 0.033 *** -0.628 0.030*** -0.632 0.029*** —0.424 0.008*** —0.426 0.008***
_cons —2.1 19 0.490*** —2.206 0.420*** -2.017 0.420*** -2.413 0.077*** —2.399 0.077***
N Obs 13.859 20,030 20,030 302,172 302,172
R-Squared 0.534 0.525 0.530 0.565 0.568
Root MSE 0.497 0.537 0.535 0.605 0.602

124

 

 

 

Table 8A: Estimates of the reform effect on primary teachers' wages by tenure group (Dependent variable: Natural Logarithm of Wage rates)

 

 

 

(1) (11) (111) (IV)
lwagel Coef. Std.Err. Coef. Std.Err. Coef. Std.Err. Coef. Std.Err.
educlyr 0.088 0.010*** 0.088 0.010*** -0.167 0179 0.108 0.248
educlyr 0.110 0007*** 0.110 0.007*** 0.280 0169* 0.277 0.232
educ3yr 0.165 0007*** 0.165 0007*** 0.063 0.154 0.095 0.208
educ4yr 0232 0.006*** 0231 0.006*** 0.131 0.137 0.151 0.192
educiw 0.289 0.006*** 0.289 0.006*** 0.166 0.140 0.135 0.196
educéyr 0354 0.007*** 0355 0007*** 0.030 0.144 0.029 0.198
educ7yr 0.397 0.006*** 0.397 0.006*** 0.019 0139 0.020 0.195
educ8yr 0.489 0.006*** 0.489 0.006*** 0.059 0.134 0.077 0.190
educ9yr 0.558 0.007*** 0.558 0.007*** 0.112 0.136 0.167 0.192
educlOyr 0.632 0.007*** 0.630 0007*** 0.229 0134* 0.250 0.190
educllyr 0.865 0.006*** 0.864 0.006*** 0.456 0132*** 0.494 0188***
educ12yr 1.099 0.010*** 1.097 0.010*** 0.593 0.132*** 0.602 0189***
educl3yr 1.244 0011*** 1.240 0.011*** 0.728 0.133*** 0.699 0189***
educl4yr 1.335 0.010*** 1.336 0.010*** 0.796 0132*** 0.782 0.189***
educlﬁyr 1.668 0.007*** 1.668 0.007*** 0.917 0132*** 0.906 0.188***
889 0.196 0009*** 0.191 0009*** 0.163 0.050*** 0.126 0.060**
age2 0.006 0.000*** 0.006 0.000*** -0.006 0.002*** 0.004 0002*
age3 0.000 0.000*** 0.000 0.000*** 0.000 0.000*** 0.000 0000*
age4 0.000 0.000*** 0.000 0.000*** 0.000 0.000*** 0.000 0000*
female 0.293 0.002*** 0.293 0.002*** 0.169 0013*** 0.101 0.014***
white 0.104 0.003*** 0.104 0003*** 0.077 0009*** 0.042 0.010***
black 0.047 0.005*** 0.047 0.005*** 0.040 0022* 0.000 0.025
yellow 0.178 0.022*** 0.177 0.022*** 0.068 0.064 0.015 0.064
indigenous 0.059 0.029** 0.060 0.029** 0.132 0.099 0.154 0.110
union 0.185 0.003*** 0.186 0003*** 0.160 0009*** 0.126 0009***
urbanl 0.090 0004*** 0.090 0.004*** 0.122 0013*** 0.141 0.014***
federal 0.413 0.005*** 0.410 0.005*** 0.220 0.042***
state 0.094 0004*** 0.096 0004*** 0.150 0011*** 0.078 0.010***
municipal 0.069 0004*** 0.073 0.004*** 0.191 0.012***
tenureO 0.267 0003*** 0.258 0007*** 0.041 0.056 0.145 0074*
tenurel 0.206 0.004*** 0.196 0.008*** 0.235 0.068*** 0.186 0104*
termreZa 0.156 0004*** 0.148 0.009*** 0.002 0.069 0.055 0.102
tenure3 0.127 0004*** 0.107 0.010*** 0.004 0.070 0.033 0.087
tenure4 0.100 0.005*** 0.085 0.012*** 0.135 0078* 0.157 0.102
primtch 0.085 0012*** 0.177 0.015*** 0.096 0.027*** 0.111 0.029***
495 0.122 0.004*** 0.136 0.016*** 0.185 0105* -0.064 0.092
y97 0.039 0004*** 0.036 0.016** -0.084 0.088 0.208 0.138
498 0.068 0.003*** 0.059 0.008*** 0.216 0096** 0.284 0120**
W9 0.056 0003*** 0.043 0.015*** 0.189 0097* 0.325 0133*4
y951>rim 0.044 0.016*** 0.041 0021* 0.015 0.039 0.024 0.044
y97prim 0.008 0.016 0.010 0.021 0.014 0.038 0.006 0.041
y989rim 0.102 0.016*** 0.086 0020*** 0.024 0.036 0.031 0.039
m 0.167 0.015*** 0.156 0020*** 0.103 0.036*** 0.093 0039**

 

125

 

Table 8A (cont’d)

 

 

 

 

 

 

 

 

(I) (ll) ([11) (IV)

lwagel Coef. Std.Err. Coef. Std.Err. Coef. Std.Err. Coef. Std.Err.
y95tenure0 0.014 0.017 -0. 150 0.125
y97tenure0 -0.004 0.018 -0.065 0.112 -0.243 0.167
y98tenure0 -0.002 0.011 -0.312 0.118*** -0.048 0.150
y99tenure0 -0.010 0.016 -0.219 0120* -0.020 0.169
y95tenure1 0.007 0.018 0.149 0.135 0.134 0.168
y97tenure1 0.010 0.018 0.193 0.123 -0.239 0.200
y98tenure1 -0.068 0.127 -0.258 0.179
y99tenure1 -0.011 0.017 -0.052 0.128 -0.090 0.182
y95tenure2 0.010 0.019 0.067 0.139 -0.074 0.158
y97tenure2 -0.003 0.019 0.161 0.126
y98tenure2 0,001 0.013 -0.283 0129’” 0.196 0.183
y99tenure2 0.014 0.018 -0.185 0.126 —0. 194 0.178
y95tenure3 -0.005 0.020 -0.071 0.140 -0.212 0.144
y97tenure3 -0.027 0.021 -0.339 0180*
y98tenure3 -0.008 0.015 -0.099 0.138 -0.313 0169*
y99tenure3 -0.002 0.019 —0.158 0.132 -0.204 0.172
y95tenure4 -0.061 0.164
y97tenure4 0.138 0.129 —0.217 0.178
y98tenure4 -0.006 0.018
y99tenure4
y95tenure5 0.024 0.017 0.019 0.110 -0.070 0.100
y97tenure5 0.013 0.017 0.157 0094* -0.155 0.143
y98tenure5 0.029 0.010*** —0.065 0.100 -0.123 0.125
y99tenure5 0.042 0.016*** -0.048 0.102 -0.161 0.137
primtenureO 0.481 0037*“ 0.042 0.066 0.147 0.084"
primtenure] 0.198 0.045*** 0.049 0.079 —0.008 0.116
primtenure2 0.133 0.042*** -0.162 0.078** -0.074 0.109
primtenure3 0.129 0.043 *** -0.086 0.080 -0.069 0.097
primtenure4 0.090 0.046 ** 0.030 0.087 0.048 0.111
y95prim*tenure0 -0116 0.053” 0.053 0.091 -0.193 0115*
y95prim*tenure1 0.042 0.062 -0.073 0.107 -0.151 0.160
y95prim*tenure2 0.069 0.058 0.018 0.111 0.044 0.146
y95prim*tenure3 -0.048 0.058 0.020 0.112 0.100 0.131
y95prim*tenure4 0.048 0.063 0.049 0.124 0.035 0.155
y97prim*tenure0 -0.127 0.049“ 0.110 0.089 -0.024 0.113
y97prim*tenure1 0.104 0062* 0.071 0.108 0.241 0.165
y97prim*tenure2 0.076 0.059 0.046 0.110 -0. 129 0.154
y97prim*tenure3 0.000 0.062 0.129 0.109 0.137 0.136
y97prim*tenure4 -0.056 0.064 -0.044 0.115 0.034 0.134
Y98prim*tenure0 -0.157 0.051*** 0.125 0.090 —0.l33 0.110
y98prim*tenure1 0.121 0.060** 0.106 0.104 0.269 0152*
Y98prim*tenure2 0.119 0.056** 0.297 0.104*** 0.168 0.152
y98prim*tenure3 0.092 0.062 0.074 0.118 0.213 0.138
X989rim*tenure4 0.124 0067* 0.052 0.117 0.010 0.140

 

 

 

 

 

Table 8A (cont‘d)

 

(1) (11) (111) (IV)

 

 

lwagel Coef. Std.Err. Coef. Std.Err. Coef. Std.Err. Coef. Std.Err.
y99prim*tenure0 -0.158 0.051*** -0.004 0.091 ~0.234 0.123 *
y99prim*tenurel 0.024 0.060 0.046 0.104 0.066 0.143
y99prim*tenure2 0.063 0.055 0.178 0099* 0.078 0.134
y99prim*tenurc3 0.109 0057* 0.136 0.109 0.100 0.127
y99prim*tenure4 0.088 0.062 -0.009 0.116 -0.105 0.150
Rondonia -0.247 0.014*** -0.246 0.014*** -0.438 0.038’“M -0.566 0040*”
Acre -0.303 0.023*** —0.302 0.023*** —0.542 0.051*** -0.687 0052*”
Amazonas —0.299 0.011*** -0.297 0.011*** —0.288 0038*** -0.385 0040*”
Roraima —0.143 0.024*** -0.l42 0.02-1*M -0.232 0.074*** -0.332 0.090’“M
Para -0.469 0.009*** -0.467 0.009*** -0.647 0033*M -0.749 0035*M
Amapa -0.179 0.022*** -0.l79 0.022*** —0.109 0.052** -0.l73 0.057***
Tocantins -0.434 0.013*** -0.434 0.013*** —0.597 0.036*** -0.697 0.036***
Maranhao -0.620 0.015*** -0.621 0.015*** -0.787 0.038“M -0.870 0040*M
P151111 -0.696 0.014*** -0.695 0.013*** -0.916 0.035“M -1.030 0.036***
Ceara -0.552 0.008*** -0.551 0.008*** -0.832 0.029*** -0.905 0.032***
RGNorte -0.664 0.012*** -0.662 0.012*** -1.003 0.041*** -l.l27 0043*M
Paraiba -0.609 0.012*** -0.605 0.012*** -0.906 0037*M -l.010 0038*”
Pemambuco -0.548 0.008*** -0.547 0.008*** -0.767 0.028*** -0.820 0.029***
Alagoas -0.580 0.013*** -0.577 0.013*** -0.853 0.040*** -0.955 0.043***
Sergipe -0.521 0.012*** -0.518 0.012*** —0.685 0.037*** -0.745 0039*“
Bahia -0.477 0.007*** -0.476 0.007*** -0.700 0.027*** —0.800 0.028***
Minas -0.335 0.007*** -0.333 0.007*** -0.360 0.027*** —0.449 0.027***
ESpirito -0.315 0.011*** -0.313 0.010*** -0.331 0.036*** -0.449 0.037*"*
Rio —0.288 0.007*** -0.286 0.007*** -0.426 0.028*** -0.460 0.029***
SaoPaulo -0.021 0.007*** -0.019 0.007*** -0.220 0.026*** -0.287 0.027*"*
Parana —0.l96 0.007*** -0.l93 0.007*** —0.339 0.028*** -0.385 0.029***
SantaCat -0.160 0.008*** -0.159 0.008*** -0.488 0.032*** -0.573 0034*”
RGSUl -0.234 0.007*** -0.232 0.007*** -0.437 0.027*** —0.559 0.028***
MGSUl -0.406 0.010*** -0.405 0.010*** -0.553 0.036*** -0.681 0.036***
MGrosso -0.272 0.010*** -0.272 0.010*** -0.445 0.033*** -0.546 0.034*"*
Goias -0.404 0.008*** —0.403 0.008*** -0.615 0.029*** -0.763 0030*”
900318.111 2.059 0.076*** -2.019 0.076*** -0.735 0.447 -0.503 0.557

***

N of Observations 302,172 302,172 20,030 *** 14,790
R-Squared 0.582 0.583 0.532 *** 0.562
Root MSE 0.592 0.592 0.534 *** 0.493

 

127

 

 

Table 9A: Estimates of the Reform Effect on Primary Teachers' Wages by Age Group

 

 

 

( I ) ( II ) ( 111 )
lwagel

Coef. Std.Err. Coef. Std.Err. Coef. Std.Err.
educlyr 0.077 0.010*** -0.162 0.181 -0.097 0.252
educ2yr 0.101 0007*” -0.296 0.177 * -0.296 0.241
educ3yr 0.153 0007*" 0.090 0.155 0.107 0.209
educ4yr 0.221 0.006*** -0.129 0.137 -0.145 0.193
educ5yr 0.267 0.006*** -0.l76 0.141 -0.143 0.196
educ6yr 0.320 0007*" 0.011 0.145 0.011 0.199
educ7yr 0.361 0.006*** -0.030 0.140 0.027 0.196
educ8yr 0.470 0.006*** 0.047 0.134 0.076 0.191
educ9yr 0.514 0007*” 0.071 0.137 0.146 0.193
educlOyr 0.608 0007*" 0.209 0.135 0.232 0.191
educl lyr 0.862 0.006*** 0.459 0.132 *** 0.485 0.189"
educ12yr 1.099 0.010*** 0.599 0.133 *** 0.600 0189‘"
educl3yr 1.246 0011*" 0.738 0.133 *** 0.699 0190""
educl4yr 1.340 0.010*** 0.803 0.133 *** 0.779 0190*”
educ15yr 1.660 0.007‘" 0.923 0132*” 0.884 0189*“
tenure 0.034 0.001‘" 0.020 0002*" 0.015 0.002 ***
tenure2 -0.001 0.000*** 0.000 0.000*** 0.000 0.000
female -0.285 0002*" -0.172 0.013 *** —0.081 0014""
white 0.103 0.003 *** 0.079 0009*" 0.043 0.010***
black -0.044 0.005*** -0.034 0.022 0.006 0.024
yellow 0.169 0021*” 0.061 0.063 0.008 0.064
indigenous 0.061 0.029** 0.148 0.100 0.157 0.111
union 0.187 0003*" 0.156 0009*" 0.120 0009*"
urbanl 0.091 0004*" 0.129 0.013 *“ 0.153 0014"”
federal 0.366 0.006*** 0.200 0.042 ***
state 0.066 0004*" -0.166 0011*" 0.062 0.010***
municipal -0.074 0004*” -0.198 0.012 ***
primtch -0.209 0072*" -0.284 0147* -0.177 0.160
agecatl -0.394 0.016‘" -0.511 0140*” -0.430 0161*"
agecat2 -0.098 0016*" -0.298 0.136 " -0.297 0.150”
agecat3 0.027 0.016" -0. 199 0.138 -0.l48 0.152
agecat4 0.056 0017‘" -0.143 0.141 -0.037 0.156
y95 -0.127 0020*" -0.162 0.193 0.003 0.242
y97 0.072 0022*" -0.181 0.197 -0.156 0.184
y98 0.073 0020*" -0.013 0.167 0.058 0.194
y99 0.116 0020"" 0.040 0.161 0.090 0.179
y95prim -0.087 0.116 -0.052 0.219 -0.246 0.264
y97prim 0.097 0.098 0.302 0.215 0.198 0.204
y98prim 0.079 0.102 0.161 0.192 0.053 0.216
y99prim 0.147 0.096 0.229 0.181 0.174 0.198
primagecatl 0.443 0076*" 0.286 0.155 "‘ 0.183 0.176
primagecatZ 0.140 0.074 "' 0.180 0.151 0.159 0.164
primagecat3 -0.054 0.075 0.110 0.153 0.042 0.167
primagecat4 -0.046 0.078 0.150 0.157 0.028 0.171

 

128

 

Table 9A (cont’d)

 

 

 

( 1 ) ( 1H )
lwagel

Coef. Std.Err. Coef. Std.Err. Coef. Std.Err.
y95agecat1 -0.001 0.021 0.023 0.202 -0.094 0.258
y95agecat2 0.001 0.021 -0.021 0.198 -0.116 0.248
y95agecat3 0.01 1 0.022 -0.066 0.200 0.219 0.249
y95agecat4 0.014 0.023 -0.017 0.204 -0.210 0.252
y97agecat1 -0.038 0.023 * 0.205 0.205 0.272 0.204
y97agecat2 -0.041 0.023 * 0.255 0.202 0.279 0.191
y97agecat3 -0.036 0.023 0.164 0.203 0.155 0.192
y97agecat4 -0.024 0.025 0.219 0.207 0.112 0.197
y98agecat1 —0.019 0.021 0.062 0.177 0.182 0.214
y98agecat2 -0.003 0.021 0.130 0.172 0.139 0.199
y98agecat3 0.003 0.021 0.101 0.174 0.049 0.201
y98agecat4 0.022 0.023 0.136 0.178 0.078 0.206
y99agecat1 -0.079 0021*" 0.063 0.171 0.249 0.203
y99agecat2 -0.063 0021*” 0.093 0.166 0.170 0.184
y99agecat3 -0.049 0.021" 0.032 0.168 0.052 0.186
y99agecat4 -0.026 0.023 0.026 0.172 -0.089 0.190
y95primage~1 0.049 0.121 0.021 0.229 0.142 0.281
y95primage~2 0.016 0.119 0.049 0.224 0.171 0.270
y95primage~3 0.080 0.120 0.174 0.226 0.350 0.272
y95primage~4 0.071 0.124 0.088 0.232 0.305 0.276
y97primage~1 -0.105 0.104 -0.306 0.225 -0.287 0.226
y97primage~2 -0.062 0.101 -0.304 0.220 -0.248 0.211
y97primage~3 -0.073 0.102 -0.223 0.222 -0.133 0.213
y97primage~4 -0.097 0.107 -0.311 0.227 -0.128 0.219
y98primage~1 0.093 0.108 0.023 0.203 -0.007 0.237
y98primage~2 0.026 0.105 -0.093 0.197 -0.035 0.222
y98primage~3 0.051 0.106 -0.045 0.199 0.073 0.224
y98primage~4 -0.053 0.109 -0.181 0.204 -0.080 0.230
y99primage~l 0.041 0.102 -0.115 0.193 -0.179 0.223
y99primage~2 0.019 0.099 -0.128 0.187 -0.172 0.204
y99primage~3 0.055 0.100 -0.040 0.189 -0.038 0.207
y99primage~4 -0.030 0.104 -0.075 0.195 0.040 0.212
Rondonia -0.249 0.014 *** -0.439 0.038 *** -0.568 0040*"
Acre -0.315 0.023 *** -0.547 0051*" -0.705 0051*"
Amazonas -0.291 0011*" -0.279 0.038 *** -0.392 0040*"
Roraima -0.125 0.024 *** -0.211 0.072 *** -0.316 0.088 ***
Para -0.471 0.009 *** -0.652 0.033 *** -0.776 0034*"
Amapa -0.179 0.022 *** -0.110 0.052 ** -0.185 0056*"
Tocantins -0.441 0.013 *** -0.590 0.036*** -0.708 0.036 ***
Maranhao -0.631 0.015 *** -0.784 0039*" -0.894 0040*"
Piaui -0.711 0014*“ -0.931 0.035 *** -1.067 0.036 ***
Ceara -0.561 0.008*** -0.843 0030*“ -0.960 0.032***
RGNorte ~0.672 0.012 *** -1.014 0040*“ -l.165 0042*"
Paraiba -0.617 0.012 *** -0.913 0037*" -1.037 0.038 ***
Pemambuco -0.555 0.008*** -0.777 0.028*** -0.849 0.029***

 

129

Table 9A (cont’d)

 

 

 

 

( I ) ( 11 ) ( 111 )
lwagel
Coef. Std.Err. Coef. Std.Err. Coef. Std.Err.

Alagoas -0.591 0013*” -0.871 0040*” -0.995 0042*”
Sergipe -0.529 0012*“ -0.688 0037*" -0.767 0038“”
Bahia -0.485 0007*" —0.705 0028*" -0.836 0.028***
Minas -0.341 0007*" -0.367 0.027“‘Ml -0.475 0.027***
Espirito -0.323 0011*" -0.342 0.036*** -0.475 0.036***
Rio -0.287 0007*” -0.436 0.028*** -0.496 0.029***
SaoPaulo -0.025 0007*“ -0.221 0026*" -0.318 0.027***
Parana -0.201 0007*" -0.343 0.028 *** -0.421 0.029***
SantaCat -0.171 0.008*** -0.494 0.032 *** -0.607 0034*"
RGSul -0.238 0007*” -0.450 0.028 *** -0.582 0.028 ***
MGSul -0.412 0.010*** -0.553 0.036*** -0.695 0035*"
MGrosso -0.282 0.010*** -0.445 0.033 *** -0.560 0034*”
Goias -0.414 0.008*** -0.619 0.029 *** -0.786 0.030’"
constant 0.271 0017*" 1.269 0187*” 1.019 0.238 ***
N 302,172 20,030 14,576

R-squared 0.579 0.532 0.563

Root 0.595 0.534 0.484

 

130

 

 

 

 

 

 

 

 

 

Table 10A: The reform effect on primary teachers’ wages - Brazil (Dependent variable: Natural Logarithm of Wage rates)
lwage Brazil
7 Coef. Std. Err. Coef. Std. Err. Coef. Std. Err. Coef. Std. Err. Coef. Std. Err.
constant -2.560 (0.075)*** -2.563 (0.075)*** -2.357 (0.076)*** -2.362 (0.076)*** -2.556 (0.075)***
federal 0.376 (0.005)*** 0.376 (0.005)*** 0.351 (0.012)***
state 0.064 (0.004)*** 0.065 (0.004)*** 0.022 (0.009)** 0.064 (0.009)***
municipal -0.072 (0.004)*** —0.072 (0.004)*** —0. 155 (0.010)*** -0. 120 (0.010)***
primtch -0.044 (0.006)*** -0.093 (0.012)*** -0.074 (0.016)*** -0.128 (0.015)*** 0.141 (0.026)***
primtchfed -0.3 l 7 (0.130)”
primtchsta -0. 135 (0.025)*** -O.344 (0.032)***
primtchmun 0.082 (0.026)*** -0.164 (0.034)***
y95 -0.l23 (0.003)*** —0.121 (0.004)*** -0123 (0.004)*** -0.117 (0.004)*** -0.l24 (0.004)***
3'97 0.039 (0.004)*** 0.039 (0004)*** 0.036 (0.004)*** 0.035 (0004)*** 0.035 (0.004)***
y98 0.073 (0.003)*** 0.068 (0.003)*** 0.063 (0.004)*** 0.062 (0.004)*** 0.056 (0.004)***
1’99 0.065 (0.003)*** 0.057 (0.003)*** 0.049 (0.004)*** 0.045 (0.004)*** 0.040 (0.004)***
y95*primtch -0.037 (0.016)** —0.035 (0.023) -0.037 (0.020)* -0.004 (0.038)
y97*primtch 0.008 (0.016) 0.007 (0.022 0.004 (0.020) -0.003 (0.036)
y98*primtch 0.100 (0.016)*** 0.099 (0.021)*** 0.049 (0.020)** -0.036 (0.038)
y99*primtch 0.165 (0.015)*** 0.183 (0.020)*** 0.095 (0.020)*** 0.012 (0.035)
y95*federal 0.059 (0.016)***
y97*federa1 0.002 (0.016)
y98*federal 0.032 (0.015)**
y99*federal 0.045 (0.016)***
y95*state 0.022 (0.013)* 0.022 (0.013)*
y97*state 0.016 (0.013) 0.018 (0.013)
y98*state 0.047 (0.013)*** 0.055 (0.013)***
y99*state 0.056 (0.013)*** 0.069 (0.013)***
y95*municipal -0.042 (0.0 141-*** —0.036 (0.014)***
y97*municipal 0.028 (0.014)“ 0.030 (0.014)**
y98*municipa1 0.064 (0.013)*** 0.072 (0.013)***
y99*municipa1 0.099 (0.013)*** 0.109 (0.013)***
y95*primtch*federa1 0.141 (0.271)
y97*primtch*federal 0.097 (0.223)
y98*primtch*federa1 0.180 (0.198)
y99*primtch*federal 0.489 (0.219)* *
y95*prithh*state 0.018 (0.034) 0.051 (0.045)
y97*prithh*state 0.023 (0.034) 0.014 (0.044)
y98*primtch*state -0.067 (0.034)” 0.070 (0.046)
y99*primtch*state -0.128 (0.033)*** 0.044 (0.043)
y95*primtch*munic1pa1 0.033 (0036) 0.002 (0.048)
y96*primtch*municipal —0.0 1 8 (0.036) -0.012 (0.046)
y97*primtch*municipa1 0.065 (0.034)* 0.149 (0.046)***
y98*primtch*municipa1 0.080 (0.034)” 0.162 (0.044)***
N ofObservations 302,172 302,172 302,172 302,172 302,172
Note: All equations include year dummy variables (educlyr, edchyr, educ3yr, educ4yr, educSyr, educ6yr, educ7yr, educ8yr, educ9yr, educlOyr, educl lyr,
educ12yr, educl3yr, educl4yr, educ15yr), age variables (age, age2, age3, age4), tenure and tenure squared, female, race variables (white, black. yellow,
indigenous), union, urbanl, state dummy variables (Rondénia, Acre, Amazonas, Roraima, Para, Amapa, Tocantins, Maranhao, Piaui, Ceara, Rio Grande do
Norte, Paraiba, Pemambuco, Alagoas, Sergipe, Bahia, Minas Gerais, Espirito, Rio de Janeiro, Sﬁo Paulo, Parana, Santa Catarina, Rio Grande do Sul, Mato
Grosso do Sul, Mato Grosso, Goias) and a constant. (R-squares vary from 58.4 to 58.9 percent).
131

 

Table 1 1A: The reform effect on primary teachers' wages ! North Region (Dependent variable: Natural Logarithm of Wage rates)

 

 

 

North
1W'4L’e Coef. Std. Err. Coef. Std. Err. Coef. Std. Err. Coef. Std. Err. Coef. Std. Err.
constant -2.485 (0.307)*** -2.491 (0.307)*** -2.385 (0.312)*** -2.404 (0.312)*** -2.463 (0.307)***
federal 0.490 (0.019)*** 0.491 (0.019)*** 0.423 (0.038)***
state 0152 (0.014)*** 0.153 (0.014)*** 0.032 (0.032) 0.132 (0.033)***
municipal 0.029 (0.014)“ 0.028 (0.014)* -0.107 (0.034)*** -0.045 (0.034)
primtch —0.078 (0.018)*** -0.097 (0.040)” —0.046 (0.063) -0.100 (0.048)M 0.106 (0.134)
primtchfed -0.133 (0.190)
primtchsta -0.105 (0.086) -0.252 (0.145)*
primtchmun 0.056 (0.094) -0.104 (0.156)
y95 -0.107 (0.014)*** -0.104 (0.014)*** -0.100 (0.016)*** 0.092 (0.015)*” -0.120 (0.017)***
y97 0.020 (0.014) 0.024 (0.015)* 0.023 (0.016) 0.024 (0.016) 0.020 (0.018)
y98 0.035 (0.014)” 0.033 (0.014)” 0.017 (0.015) 0.019 (0.015) 0.004 (0.017)
y99 0.054 (0.013)*** 0.045 (0.014)*** 0.025 (0.015)* 0.016 (0.015) 0.003 (0.017)
y95*primtch -0.040 (0.052) 0.026 (0.086) -0.076 (0.063) 0.114 (0.170)
y97*primtch -0.058 (0.053) -0.055 (0.082) -0.097 (0.064) -0.071 (0.166)
y98*primtch 0.031 (0.053) -0.028 (0.082) 0.035 (0.066) 0.055 (0.163)
y99*primtch 0.141 (0.051)*** 0.137 (0.078)* 0.089 (0.063) 0.024 (0.157)
y95*federal 0.122 (0.050)**
y97*federal 0.016 (0.052)
y98*federal 0.095 (0.050)*
y99*federal 0.126 (0.050)**
y95*state 0.022 (0.042) 0.031 (0.043)
y97*state 0.006 (0.043) 0.01 1 (0.044)
y98*state 0.063 (0.043) 0.074 (0.044)*
y99*state 0.046 (0.041) 0.081 (0.042) *
y95*municipal -0.038 (0.050) -0.010 (0.050)
y97*municipa1 0.014 (0.048) 0.022 (0.048)
y98*municipa1 0.082 (0.045)* 0.099 (0.045)”
y99*municipal 0.154 (0.044)*“ 0.177 (0.045)***
y95*primtch*federal -0.095 (0.277)
y97*primtch*federal -0.225 (0.351)
y98*primtch*federal —0.308 (0.287)
y99*primtch*federal -0.018 (0.249)
y95*primtch*state -0.138 (0.115) -0.213 (0.186)
y97*primtch*state -0.038 (0.113) -0.013 (0.182)
y98*primtch*state 0.025 (0.116) -0.037 (0.181)
y99*primtch*state -0.077 (0.1 10) 0.048 (0.174)
y95*primtch*municipa1 0.114 (0.129) -0.097 (0.203)
y96*primtch*municipa1 0.071 (0.124) 0.034 (0.195)
y97*primtch*municipal -0.130 (0.125) -0.156 (0.193)
y98*primtch*municipal —0.004 (0.120) 0.052 (0.185)
N ofObservations 20,691 20,691 20,691 20,691 20,691

 

Note: All equations include year dummy variables (educlyr, educ2yr, educ3yr, educ4yr, educ5yr, educ6yr, educ7yr, educ8yr, educ9yr, educlOyr, educl lyr,
educ12yr, educl3yr, educl4yr, educ15yr), age variables (age, age2, age3, age4), tenure and tenure squared, female, race variables (white, black, yellow,
indigenous), union, urbanl, state dummy variables (Rondénia, Acre, Amazonas, Roraima, Para, Amapa, Tocantins, Maranhao, Piaui, Ceara, Rio Grande do
Norte, Paraﬂoa, Pemambuco, Alagoas, Sergipe, Bahia, Minas Gerais, Espirito, Rio de Janeiro, sao Paulo, Parana, Santa Catarina, Rio Grande do Sul, Mato
Grosso do Sul, Mato Grosso, Goias) and a constant. (R-squares vary from 54.4 to 56.2 percent).

132

 

 

Table 12A: The reform effect on primary teachers' wages - Northeast Region (Dependent variable: Natural Logarithm of Wage rates)

 

 

 

Northeast
lwage Coef. Std. Err. Coef. Std. Err. Coef. Std. Err. Coef. Std. Err. Coef. Std. Err.
constant -2.608 (0.162)*** -2.603 (0.162)*** -2.364 (0.165)*** -2.350 (0.165)*** -2.591 (0.162)***
federal 0.467 (0.011)*** 0.468 (0.011)*** 0.465 (0.024)***
state 0103 (0.008)*** 0.104 (0.008)*** 0.043 (0.018)” 0.097 (0.018)***
municipal -0.057 (0.008)*** -0.058 (0.008)*** -0.190 (0.019)*** -0.l38 (0.019)***
primtch -0.051 (0.010)*** -0.152 (0.020)*** -0.l97 (0.026)*“ -0.146 (0.026)*“ 0.033 (0.040)
primtchfed 0.384 (0.291)
primtchsta -0.026 (0.044) -0.238 (0.053)***
primtchmun -0.009 (0.044) -0.l67 (0.053)***
y95 -0.129 (0.007)*** -0.127 (0.007)*** -0.128 (0.008)*** -0.121 (0.008)*** -0.133 (0.009)***
y97 0.048 (0.007)*** 0.046 (0.008)*** 0.036 (0.008)*** 0.034 (0.008)*** 0.034 (0.009)***
y98 0.100 (0.007)*** 0.089 (0.007)*** 0.090 (0.008)*** 0.081 (0.008)*** 0.071 (0.008)***
y99 0.091 (0.007)*** 0.072 (0.007)*** 0.066 (0.008)*** 0.051 (0.008)*** 0.047 (0.008)***
y95*primtch -0.023 (0.028) -0.026 (0.037) -0.001 (0.037) 0.035 (0.059)
y97*primtch 0.035 (0.028) 0.067 (0.034)* 0.026 (0.037) 0.050 (0.055)
y98*primtch 0.183 (0.027)*** 0.217 (0.033)*** 0.049 (0.037) -0.008 (0.062)
y99*primtch 0.286 (0.026)*“ 0.334 (0.032)*** 0.137 (0.037)*** 0.082 (0.055)
y95*federa1 0.162 (0.032)***
y97*federal 0.029 (0.034)
y98*fedcra1 -0.040 (0.030)
y99*federai -0.070 (0.032)**
y95*state -0.003 (0.024) 0.003 (0.024)
y97*state 0.033 (0.026) 0.038 (0.026)
y98*state 0.038 (0.025) 0.059 (0.025)**
y99*state 0.055 (0.026)** 0.081 (0.026)*“
y95*municipal -0.049 (0.026)* 0.039 (0.026)
y97*municipal 0.061 (0.026)** 0.065 (0.026)**
y98*municipa1 0.120 (0.024)*** 0.133 (0.024)***
y99*municipa1 0.185 (0.024)*** 0.195 (0.024)***
y95*primtch*federal -0.892 (0.957)
y97*primtch*federa1 -0.405 (0.301)
y98*primtch*federal -0.405 (0.358)
y99*primtch*federa1 -0.138 (0.361)
y95*primtch*state 0.023 (0.062) —0.043 (0.077)
y97*primtch*state —0.106 (0.063)* -0.095 (0.075)
y98*primtch*state 0.166 (0.060) *** 0.057 (0.079)
y99*primtch*state -0.253 (0.062)*** -0.005 (0.075)
y95*primtch*municipa1 0.003 (0.062) -0.035 (0.077)
y96*primtch*municipa1 -0.01 1 (0.060) —0.039 (0.072)
y97*primtch*municipal 0.179 (0.058)*** 0.236 (0.076)***
y98*primtch*municipal 0.156 (0.057)*** 0.209 (0.069)***
N ofObservation 75,501 75,501 75,501 75,501 75,501

 

Note: All equations include year dummy variables (educlyr, educ2yr, educ3yr, educ4yr, educ5yr, educ6yr, educ7yr, educ8yr, educ9yr, educlOyr, educl lyr,
educ12yr, educl3yr, educl4yr, educ15yr), age variables (age, age2, age3, age4), tenure and tenure squared, female, race variables (white, black, yellow,
indigenous), union, urbanl, state dummy variables (Rondénia, Acre, Amazonas, Roraima, Para, Amapa, Tocantins, Maranhao, Piaui, Ceara, Rio Grande do
Norte, Paraiba, Pemambuco, Alagoas, Sergipe, Bahia, Minas Gerais, Espirito, Rio de Janeiro, Sﬁo Paulo, Parana, Santa Catarina, Rio Grande do Sul, Mato

Grosso do Sul, Mato Grosso, Goias) and a constant. (R-squares vary from 55.1 to 56.5 percent).

 

 

Table 13A: The reform effect on primary teachers' wages — Southeast Region (Dependent variable

: Natural Logarithm of Wage rates)

 

 

 

Southeast
[wage Coef. Std. Err. Coef. Std. Err. Coef. Std. Err. Coef. Std. Err. Coef. Std. Err.
constant -3.l69 (0.119)*** -3.l72 (0.119)*** -3.110 (0.119)*** ~3.095 (0.119)*** -3.l70 (0.119)***
federal 0.266 (0.010)*** 0.266 (0.010)*** 0.254 (0.021)***
state -0.006 (0.007) -0.005 (0.007) -0.005 (0.016) 0.007 (0.017)
municipal -0.095 (0.007)*** -0.096 (0.007)*** —0.142 (0.016)*** -0.126 (0.016)***
primtch 0.019 (0.010)* -0.024 (0.021) 0.064 (0.030)** -0.102 (0.025)*** 0.200 (0.048)***
primtchfed —0.462 (0.351)
primtchsta -0.227 (0.044)*” -0.362 (0.058)***
primtchmun 0.217 (0.047)*** -0.069 (0.062)
y95 -0. 122 (0.005)*** -0.121 (0.006)*** -0.122 (0.006)*** -0.118 (0.006)*** —0.l2l (0.006)***
y97 0.038 (0.006)*** 0.037 (0.006)*** 0.037 (0.006)*** 0.035 (0.006)*** 0.036 (0.006)***
y98 0.068 (0.005)*** 0.064 (0.006)*** 0.063 (0.006)*** 0.061 (0.006)*** 0.059 (0.006)***
y99 0.064 (0.005)*** 0.060 (0.006)*** 0.055 (0.006)*** 0.056 (0.006)*** 0.052 (0.006)***
y95*primtch -0.031 (0.028) -0.035 (0.041) -0.037 (0.033) -0.056 (0.065)
y97*primtch 0.036 (0.029) -0.024 (0.042) 0.051 (0.035) -0.041 (0.066)
y98*primtch 0.092 (0.029)*** 0.002 (0.040) 0.093 (0.035)*** -0.068 (0.068)
y99*primtch 0.118 (0.027)*** 0.037 (0.039) 0.115 (0.033)*** -0.045 (0.063)
y95*federal 0.015 (0.028)
y97*federa1 -0.009 (0.030)
y98*federal 0.01 1 (0.029)
y99*federa1 0.071 (0.029)**
y95*state 0.022 (0.022) 0.021 (0.022)
y97*state -0.003 (0.023) 0.002 (0.023)
y98*state 0.024 (0.023) 0.026 (0.023)
y99*state 0.049 (0.023)M 0.052 (0.023)“
y95*municipal -0.032 (0.022) -0.029 (0.022)
y97*municipal 0.022 (0.023) 0.022 (0.023)
y98*municipal 0.045 (0.022)** 0.048 (0.022)**
y99*municipa1 0.027 (0.021) 0.033 (0.021)
y95*primtch*federal 0.145 (0.372)
y97*primtch*federal (dropped) ***
y98*primtch*federal -0.453 (0.688)
y99*primtch*federal 0.204 (0.354)
y95*primtch*state —0.01 1 (0.059) 0.009 (0.078)
y97*primtch*state 0.093 (0.061) . 0.110 (0.080)
y98*primtch*state 0.099 (0.060)* 0.171 (0.081)”
y99*primtch*state 0.068 (0.058) 0.152 (0.076) **
y95*primtch*municipal 0.041 (0.065) 0.060 (0.086)
y96*primtch*municipal -0.088 (0.066) 0.003 (0.087)
y97*primtch*municipa1 -0.085 (0.063) 0.074 (0.085)
y98*primtch*municipal '0-053 (0-061) 0404 (0-081)
N ofObservation 116,020 116,020 116,020 116,020 116,020

 

 

Note: All equations include year dummy variables (educlyr, educ2yr, educ3yr, educ4yr, educ5yr, educ6yr, educ7yr, educ8yr, educ9yr, educlOyr, educl lyr,
educ12yr, educl3yr, educl4yr, educ15yr), age variables (age, age2, age3, age4), tenure and tenure squared, female, race variables (white, black, yellow,
indigenous), union, urbanl, state dummy variables (Rondénia, Acre, Amazonas, Roraima, Para, Amapa, Tocantins, Maranhao, Piaui, Ceara, Rio Grande do
Norte, Paraiba, Pemambuco, Alagoas, Sergipe, Bahia, Minas Gerais, Espirito, Rio de Janeiro, Sao Paulo, Parana, Santa Catarina, Rio Grande do Sul, Mato
Grosso do Sul, Mato Grosso, Goias) and a constant. (R-squares vary from 56.3 to 56.8 percent).

134

 

 

 

Table 14A: The reform effect on primary teachers' wages ~ South Region (Dependent variable: Natural Logarithm of Wage rates)

 

 

 

South
lwage Coef. Std. Err. Coef. Std. Err. Coef. Std. Coef. Std. Coef. Std.Err.
constant -2.879 (0.164) *** -2.880 (0.164) *** -2.825 (0.165) *** «2.808 (0.165) *** -2.884 (0.163) ***
federal 0.285 (0.013) *** 0.285 (0.013) *** 0.256 (0.028) ***
state 0.015 (0.011) 0.015 (0.011) 0.003 (0.024) 0.027 (0.024)
municipal -0.058 (0.009) *** -0.058 (0.009) *** -0.104 (0.021) *** -0.086 (0.021) ***
primtch —0.128 (0.014) *** -0.130 (0.029) *** -0.009 (0.034) -0.254 (0.041) *** 0.310 (0.060) ***
primtchfed 0.083 (0.105)
primtchsta -0.384 (0.058) *** -0.696 (0.076) ***
primtchmun 0.247 (0.058) *** -0.299 (0.073) ***
y95 -0.117 (0.008) *** -0.114 (0.008) *** -0.116 (0.008) *** -0.110 (0.008) *** -0.111 (0.008) ***
y97 0.044 (0.008) *** 0.043 (0.008) *** 0.043 (0.008) *** 0.043 (0.008) *** 0.037 (0.009) ***
y98 0.063 (0.007) *** 0.063 (0.008) *** 0.056 (0.008) *** 0.062 (0.008) *** 0.055 (0.008) ***
y99 0.043 (0.007) *** 0.041 (0.008) *** 0.037 (0.008) *** 0.035 (0.008) *** 0.030 (0.008) ***
y95*primtch -0.079 (0.039) ** -0.076 (0.048) —0.103 (0.056) * -0.091 (0.106)
y97*primtch 0.008 (0.040) —0.029 (0.050) 0.021 (0.057,) -0.061 (0.101)
y98*primtch 0.021 (0.039) -0.003 (0.047) 0.031 (0.056) -0.102 (0.094)
y99*primtch 0.059 (0.040) 0.059 (0.049) 0.058 (0.054) -0.060 (0.086)
y95*federal -0.024 (0.039)
y97*federal 0.053 (0.039)
y98*federa1 0.064 (0.040)
y99*federa1 0.076 (0.039) *
y95*state 0.032 (0.032) 0.025 (0.032)
y97*state 0.032 (0.034) 0.039 (0.034)
y98*state 0.086 (0.033) ** 0.086 (0.033) **
y99*state 0.056 (0.034) * 0.062 (0.034) *
y95*municipa1 -0.060 (0.029) ** -0.059 (0.029) **
y97*municipal 0.018 (0.031) 0.027 (0.031)
y98*municipa1 -0.01 1 (0.029) -0.002 (0.029)
y99*municipal 0.057 (0.029) * 0.065 (0.029) **
y95*primtch*federal ***
y97*primtch*federal ***
y98*primtch*federal ***
y99*primtch*federal -0.385 (0.374)
y95*primtch*state -0.019 (0.079) —0.001 (0.123)
y97*primtch*state 0.018 (0.082) 0.050 (0.119)
y98*primtch*state -0.014 (0.082) 0.086 (0.115)
y99*primtch*state —0.069 (0.081) 0.052 (0.107)
y95*primtch*municipal 0.100 (0.080) 0.088 (0.120)
y96*primtch*municipal -0.059 (0.083) 0.022 (0.117)
y97*primtch*municipal -0.006 (0.081) 0.125 (0.110)
y98*primtch*municipal -0.037 (0.083) 0.082 (0.106)
N Observation 56,380 56.380 56,380 56,380 56,380

 

Note: All equations include year dummy variables (educlyr, educ2yr, educ3yr, educ4yr, educ5yr, educ6yr, educ7yr, educSyr, educ9yr, educlOyr, educl lyr,
educ12yr, educl3yr, educl4yr, educ 1 5yr), age variables (age, age2, age3, age4), tenure and tenure squared, female, race variables (white, black, yellow,
indigenous), union, urbanl, state dummy variables (Rondénia, Acre, Amazonas, Roraima, Para, Amapa, TocantinS, Maranhao, Piaui, Ceara, Rio Grande do
Norte, Paraﬂaa, Pemambuco, Alagoas, Sergipe, Bahia, Minas Gerais, Espirito, Rio de Janeiro, Sao Paulo, Parana, Santa Catarina, Rio Grande do Sul, Mato
Grosso do Sul, Mato Grosso, Goias) and a constant. (R-squares vary from 55.1 to 54.8 percent).

U1

 

 

 

Table 15A: The reform effect on primary teachers' wages — Center-West Region (Dependent variable: Natural Logarithm of Wage rates)

 

Center-West

 

 

 

lwage Coef. Std. Err. Coef. Std. Err. Coef. Std. Err. Coef. Std. Err. Coef. Std. Err.
constant —2.554 (0.221)*** —2.552 (0.221)*** -2.276 (0.225)*** -2.271 (0.224)*** —2.543 (0.220)
federal 0.420 (0.014)*** 0.420 (0.014)*** 0.379 (0.026)
state 0143 (0.011)*** 0.143 (0.011)*** 0.024 (0.023) 0.122 (0.024)
municipal -0.071 (0.013)*** —0.071 (0.013)*** -0.150 (0.034)*** -0.102 (0.034)
primtch —0.043 (0.016)*** -0.009 (0.036) 0.037 (0.051) -0.054 (0.043) 0.252 (0.077)
primtchfed -0.881 (0.190)
primtchsta -0.176 (0.073)** -0.374 (0.092)
primtchmun 0.124 (0.085) -0.161 (0.104)
y95 -0.131 (0.011)*** -0.131 (0.011)*** -0.140 (0.012)*** -0.130 (0.011)*** -0.148 (0.013)
y97 0.029 (0.011)*** 0.034 (0.011)*** 0.022 (0.012)* 0.029 (0.012)” 0.035 (0.013)
y98 0.057 (0.010)*** 0.059 (0.010)*** 0.040 (0.011)*** 0.047 (0.011)*** 0.040 (0.012)
y99 0.043 (0.010)*** 0.044 (0.010)*** 0.022 (0.011)* 0.028 (0.011)“ 0.016 (0.012)
y95*primtch -0.012 (0.049) 0.026 (0.073) -0.002 (0.057) 0.082 (0.107)
y97*primtch -0.109 (0.047)“ -0.089 (0.070) -0.098 (0.056)* 0.016 (0.113)
y98*primtch -0.039 (0.046) -0.035 (0.068) —0.050 (0.056) -0.104 (0.103)
y99*primtch -0.006 (0.047) 0.020 (0.074) 0.001 (0.057) -0.047 (0.112)
y95*federa1 0.047 (0.035)
y97*federal -0.010 (0.037)
y98*federal 0.080 (0.035)
y99*federal 0.114 (0.035)
y95*state 0.073 (0.033)M 0.080 (0.033
y97*state 0.030 (0.034) 0.024 (0.035)
y98*state 0.059 (0.032)* 0.065 (0.032)
y99*state 0.079 (0.032)“ 0.092 (0.032)
y95*municipal 0.003 (0.045) 0.013 (0.045)
y97*municipa1 -0.033 (0.043) -0.042 (0.043)
y98*municipal 0.030 (0.044) 0.036 (0.044)
y99*municipa1 0.052 (0.043) 0.073 (0.043)
y95*primtch*federa1 0.484 (0.206)
y97*primtch*federal 0.41 1 (0.210) ‘
y98*primtch*federal 0.953 (0.313)
y99*primtch*federal 1.810 (0.887)
y95*primtch*state -0.091 (0.101) -0.160 (0.127)
y97*primtch*state -0.047 (0.098) -0.166 (0.131)
y98*primtch*state -0.056 (0.096) 0.012 (0.122)
y99*primtch*state -0.059 (0.100) -0.003 (0.129)
y95*primtch*municipal -0.032 (0.118) -0.105 (0.146)
y96*primtch*municipal -0.023 (0.109) -0.1 18 (0.144)
y97*primtch*municipal -0.002 (0.111) 0.071 (0.138)
y98*primtch*municipal -0.039 (0.110) 0.041 (0.144)
N Observation 33,580 33,580 33,580 33,580 33,580

 

Note: All equations include year dummy variables (educlyr, educ2yr, educ3yr, educ4yr, educ5yr, educ6yr, educ7yr, educ8yr, educ9yr, educlOyr, educl lyr,
educ12yr, educl3yr, educl4yr, educ15yr), age variables (age, age2, age3, age4), tenure and tenure squared, female, race variables (white, black, yellow,
indigenous), union, urbanl, state dummy variables (Rondénia, Acre, Amazonas, Roraima, Para, Amapa, Tocantins, Maranhao, Piaui, Ceara, Rio Grande do
Norte, Paraiba, Pemambuco, Alagoas, Sergipe, Bahia, Minas Gerais, Espirito, Rio de Janeiro, S50 Paulo, Parana, Santa Catarina, Rio Grande do Sul, Mato
Grosso do Sul, Mato Grosso, Goias) and a constant. (R-squares vary from 62.9 to 63.1 percent).

136

 

 

Table 16A

: Descriptive Statistics for Labor Force Participation Estimation

 

 

 

 

Females
Total Urban Rural

. Non- . Non- . Non-
anary Primary Primary Primary Primary Primary
Teachers T h Teachers T h Teachers Teach rs

eac ers eac ers e
eamingsl 394.20 371.64 419.16 387.02 226.54 225.55
(313.14) (538.25) (318.44) (551.50) (208.52) (307.64)
lnwagel 1.01 0.46 1.09 0.50 0.44 0.12
(0.73) (0.94) (0.69) (0.94) (0.76) (0.97)
hourwwl 29.05 20.33 29.36 21.09 24.03 14.77
(10.51) (22.13) (10.65) (22.22) (12.05) (20.33)
education 13.07 7.91 13.40 8.23 11.00 6.71
(2.40) (4.20) (2.12) (4.18) (2.65) (4.54)
ed0to8 0.06 0.64 0.03 0.62 0.19 0.73
(0.24) (0.48) (0.17) (0.49) (0.39) (0.44)
ed9toll 0.49 0.25 0.48 0.26 0.58 0.13
(0.50) (0.43) (0.50) (0.44) (0.49) (0.34)
ed12tol4 0.12 0.03 0.12 0.03 0.14 0.06
(0.32) (0.17) (0.33) (0.17) (0.35) (0.25)
ed15 0.34 0.08 0.37 0.09 0.09 0.07
(0.47) (0.27) (0.48) (0.28) (0.28) (0.26)
age 33.65 32.78 33.90 32.88 29.48 32.03
(7.89) (8.40) (7.88) (8.40) (8.09) (8.42)
white 0.59 0.54 0.60 0.55 0.56 0.50
(0.49) (0.50) (0.49) (0.50) (0.50) (0.50)
black 0.03 0.06 0.04 0.06 0.02 0.05
(0.18) (0.23) (0.19) (0.23) (0.12) (0.21)
yellow 0.002 0.004 0.002 0.004 0.001 0.001
(0.05) (0.06) (0.05) (0.06) (0.03) (0.04)
mixed 0.38 0.40 0.36 0.39 0.42 0.45
(0.48) (0.49) (0.48) (0.49) (0.49) (0.50)
indigenous 0.001 0.001 0.001 0.001 0.000 0.001
(0.03) (0.04) (0.03) (0.04) (0.02) (0.04)
married 0.77 0.76 0.76 0.74 0.84 0.85
(0.42) (0.43) (0.43) (0.44) (0.37) (0.35)
single 0.23 0.24 0.24 0.26 0.16 0.15
(0.42) (0.43) (0.43) (0.44) (0.37) (0.35)
child114 0.57 0.62 0.55 0.61 0.48 0.71
(0.50) (0.48) (0.50) (0.49) (0.50) (0.45)
childm14 0.45 0.44 0.46 0.45 0.53 0.44
(0.50) (0.50) (0.50) (0.50) (0.50) (0.50)
urbanl 0.86 0.89 1.00 1.00 0.000 0.000
(0.34) (0.31) (0.00) (0.00) (0.00) (0.00)
N of Observations 12409 331308 10712 294552 2181 39914

 

137

Table 16A (cont’d)

 

 

 

Males
Total Urban Rural
. Non- . Non- . Non-
anary Prima Primary Prima Primary Prim
Teachers ry Teachers ry Teachers ary
Teachers Teachers Teachers

 

eamingsl
lnwagel
hourwwl
educauon
ed0108
ed9tol 1
ed 1 2to 14
ed15

age

white
black
yellow
mixed
indigenous
married
single
child114
childm14

urbanl

N of Observations

456.59 575.66 479.08 580.30 248.73 418.32

(394.84) (762.21) (400.59) (765.67) (205.17) (452.49)
1.11 0.75 1.17 0.76 0.53 0.51
(0.73) (0.91) (0.72) (0.91) (0.69) (0.84)
30.70 38.51 30.98 38.60 29.30 38.82
(12.40) (19.98) (12.58) (19.97) (11.55) (21.05)
13.41 7.97 13.73 8.10 11.02 7.09
(2.39) (4.16) (2.06) (4.13) (3.26) (4.84)
0.06 0.65 0.03 0.64 0.28 0.71
(0.23) (0.48) (0.16) (0.48) (0.45) (0.46)
0.40 0.24 0.38 0.25 0.47 0.10
(0.49) (0.42) (0.48) (0.43) (0.50) . (0.30)
0.15 0.03 0.16 0.03 0.10 0.07
(0.35) (0.17) (0.37) (0.17) (0.31) (0.25)
0.40 0.08 0.44 0.08 0.15 0.12
(0.49) (0.28) (0.50) (0.28) (0.36) (0.33)
31.54 32.50 31.69 32.48 28.82 30.93
(8.18) (8.42) (8.12) (8.41) (7.50) (8.04)
0.53 0.52 0.53 0.51 0.47 0.49
(0.50) (0.50) (0.50) (0.50) (0.50) (0.50)
0.04 0.07 0.04 0.06 0.03 0.05
(0.20) (0.25) (0.20) (0.25) (0.17) (0.22)
0.004 0.003 0.003 0.003 0.000 0.005
(0.07) (0.06) (0.06) (0.06) (0.00) (0.07)
0.42 0.41 0.42 0.42 0.50 0.45
(0.49) (0.49) (0.49) (0.49) (0.50) (0.50)
0.004 0.001 0.003 0.001 0.000 0.002
(0.07) (0.04) (0.06) (0.04) (0.00) (0.04)
0.74 0.81 0.72 0.81 0.82 0.85
(0.44) (0.39) (0.45) (0.40) (0.38) (0.35)
0.26 0.19 0.28 0.19 0.18 0.15
(0.44) (0.39) (0.45) (0.40) (0.38) (0.35)
0.45 0.54 0.42 0.54 0.59 0.60
(0.50) (0.50) (0.49) (0.50) (0.49) (0.49)
0.49 0.45 0.50 0.45 0.44 0.46
(0.50) (0.50) (0.50) (0.50) (0.50) (0.50)
0.84 0.91 1.00 1.00 0.000 0.000
(0.37) (0.28) (0.00) (0.00) (0.00) (0.00)
1135 254508_ 891 211355 107 17384

 

138

IIIIIIIIIIIIIIIIIIIIIIIIIIIIIIII

111111(1)1111(11111111111111