 

a}
. .15.
3
"(.muwﬂ
.~ 11—h?”-

‘x P.
: i:
(all.

i
l’. ‘. 1.-
»xflina’. .
$3....

$4. . n

it 5:221:
.1 .1 .
3:2 is

.4. :13.

o A

Iln~nlibav¢ x
r.

no: rain ..

 

xh3-u..1
S .

y :50. :0! ‘
I I .J

, 2. 3&8 «if... 3» . .ﬁ
be»; 21.19.47 :

5?. .1 3‘ \ .

4.. a .3] A

'm

1004
téjans?

 

UBRARY
Michigan State
University

 

 

 

This is to certify that the
dissertation entitled

ESTIMATING RETURNS TO SCHOOLING IN INDONESIA:
EVIDENCE FROM THE INDONESIA FAMILY LIFE SURVEY
1993-2000

presented by
GREGORIUS DAAN VINCENT PATTINASARANY

has been accepted towards fulfillment
of the requirements for the

Ph.D. degree in Economics

/ [ﬁg-J
/ ajor P essor 3 Signature

3%,}ny 524 02003

Date

 

 

 

MSU is an Affirmative Action/Equal Opportunity Institution

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:/CIRC/DateDue.p65-p.15

ESTIMATING RETURNS TO SCHOOLING IN INDONESIA:
EVIDENCE FROM THE INDONESIA FAMILY LIFE SURVEY 1993-2000

By

Gregorius Daan Vincent Pattinasarany

A DISSERTATION

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

DOCTOR OF PHILOSOPHY
Department of Economics

2003

ABSTRACT

ESTIMATING RETURNS TO SCHOOLING IN INDONESIA:
EVIDENCE FROM THE INDONESIA FAMILY LIFE SURVEY 1993-2000

By

Gregorius Daan Vincent Pattinasarany

The rate of return to schooling investment in Indonesia is estimated using the
Indonesia Family Life Survey (IFLS) of 1993, 1997 and 2000. This study explicitly takes
into account measures to overcome omitted variable and sample selection biases.
Parental schooling information is used to correct for omitted variable bias, while sample
selection bias is corrected by modeling employment outcomes into several labor market
alternatives. Returns to schooling are estimated for adults aged 25-59 years who are self-
employed as well as those who are working in the public and private sectors. Separate
estimates for men and women are presented for each of the cross section data of 1993,
1997 and 2000.

Some important ﬁndings of this study are the following. Individuals with more
formal schooling are more favorably rewarded and, thus, have higher returns to
schooling. Wage advantage increases in a convex fashion with levels of schooling. Over
years, wage advantages within each level of schooling are, in general, declining. Supply
shifts among individuals with more schooling might serve as one of the possible
explanations for the declining rates.

Parental education captures part of individuals’ family background information as the
inclusion of it in the wage regression reduces estimated coefﬁcients of individuals’

schooling variables. Although the effects of parental schooling on individuals’ returns

are limited, the results reveal that having educated parents is associated with an increase
in earnings relative to having illiterate parents.

Sample selection bias does not turn to be important in these wage regressions.
Selectivity correction terms are mainly signiﬁcant in the self-employment wage
regressions. Likewise, the potential bias from migration seems to be small. Individuals
who were born or are currently living in urban areas do not have a wage advantage that is
signiﬁcantly different from their counterparts who were born or are currently living in
rural areas. Stratifying individuals based on their age reveals that younger individuals
(aged 25-39) who are working in the private sector have a lower rate of return in

comparison to that of the older generation (aged 40-59).

Cepyright by
GREGORIUS DAAN VINCENT PATTINASARANY
2003

To my wife, Ira, and my children, Dani and Vira,
whose love, support and encouragement made this work possible.

ACKNOWLEDGMENTS

It has been a long journey for me to get to this stage of my learning career. I have
experienced many ups and downs during years as a graduate student. But the most
valuable message that I get from this journey is that one should never give up in
obtaining his ultimate goal. There are a lot of people that I should acknowledge for
helping me during this endeavor.

I am indebted to Prof. John A. Strauss, my major advisor, not only for his guidance
and support that he has given for writing this dissertation but also for the friendship that
we have built during the course of time. I am also very grateful for all efforts that he has
done that enabled me to ﬁnally defend my dissertation.

I would like to thank all my committee members: Prof. John Giles, Prof. Carl
Liedholm and Prof. Jeff Biddle for their valuable comments in improving this
dissertation.

I would like to thank the late Prof. Dr. J .L. Tamba; without his initiative I would have
not have the opportunity to pursue Ph.D. degree.

I would like to thank my late father, J .D. Pattinasarany and my late mother, G.M.L.

J ansz-Pattinasarany, whose inspired me a life long love of learning, and my sisters, M.
Sally H.L. Pattinasarany and Fransica M.R. Pattinasarany, for giving me the support and
strength when I really needed it. I would like to thank my father-in-law, Waluyo, and

mother-in-law, Toeti Waluyo, for believing in me and support me along the way.

Finally, I would like to thank my wife, Indera Ratna Irawati Pattinasarany, my son,
Danira Reggie Pattinasarany, and my daughter, Vinira Pritha Pattinasarany, whose love,

support and encouragement made this work possible.

vii

TABLE OF CONTENTS

LIST OF TABLES ............................................................................... x
SECTION 1
INTRODUCTION ............................................................................... 1
SECTION 2
PREVIOUS EMPIRICAL STUDIES ON INDONESIA ................................... 9
2.1. Early Empirical Findings ............................................................. 9
2.2. Empirical Studies of the 19905 ...................................................... 10
2.3. Discussion .............................................................................. 19
SECTION 3
MODEL ........................................................................................... 22
3.1. Sector Choice Model ................................................................. 22
3.2. Wage Function ........................................................................ 26
SECTION 4
DATA .......................................................................................... 28
4.1. IF LS ..................................................................................... 29
4.2. Labor Force Participation for Cross Section Respondents ..................... 33
4.3. Labor Force Participation for Panel Respondents ............................... 37
SECTION 5
SECTOR CHOICE MODEL ................................................................... 40
5.1. Speciﬁcation Tests .................................................................. 41
5.1.1. Hausman-McFadden Tests ................................................. 42
5.1.2. Wald Tests ................................................................... 43
5.3. Factors Affecting Sector Choice ................................................... 44
5.3.1. Own Schooling .............................................................. 45
5.3.2. Parental Schooling ........................................................... 50
5.3.3. Other Variables .............................................................. 51
Appendix Section 5: Sector Choice Model for Panel Respondents .................. 54
SECTION 6
WAGE FUNCTIONS .......................................................................... 58
6.1. Earnings Rate ........................................................................ 60
6.2. Wage Functions: the Effects of Own Schooling ................................. 62
6.2.1. Linear Own Schooling ...................................................... 63
6.2.1. Non-linear Own Schooling ................................................ 65
6.3. Wage Functions: the Effects of Parental Schooling ............................ 71
6.4. Wage Functions: the Effects of Regions of Current Residence ............... 73

viii

6.5. Predicted Rates of Returns ......................................................... 75

6.6. Discussion ............................................................................ 77
6.6.1. Comparison with OLS Results ............................................. 78
Appendix Section 6: Wage Functions for Panel Respondents ........................ 79
SECTION 7
MIGRATION .................................................................................... 83
7.1. Schooling Attainment and Earnings .............................................. 85
7.2. Model .................................................................................. 87
7.3. Discussion ............................................................................ 88
SECTION 8
RETURNS TO SCHOOLIN G BY AGE COHORTS ...................................... 94
8.1. Schooling Attainment and Earnings ............................................... 95
8.2. Model .................................................................................. 96
8.3. Discussion ............................................................................ 97
SECTION 9
RETURNS TO SCHOOLING USING POTENTIAL EXPERIENCE .................. 100
SECTION 10
SUMMARY ...................................................................................... 104
10.1. Empirical Results .................................................................... 106
10.1.1. Sector Choice Model ...................................................... 106
10.1.2. Wage Regressions .......................................................... 109
10.1.3. Migration and Age-Cohort ................................................ 113
10.2. Policy Implications .................................................................. 114
10.3. Future Agenda ....................................................................... 116
BIBLIOGRAPHY ................................................................................ 235

ix

LIST OF TABLES

Table 4.1
Summary of Number of Observations ........................................................ 1 18

Table 4.2
Distribution of Employment by Schooling Attainment, Sector and Gender,
Cross Section Respondents ..................................................................... 119

Table 4.3
Distribution of Employment by Age, Sector and Gender, Cross Section Respondents 120

Table 4.4
Distribution of Employment by Schooling Attainment, Sector and Gender, Panel
Respondents ...................................................................................... 121

Table 4.5
Distribution of Employment by Age, Sector and Gender, Panel Respondents .......... 122

Table 4.6
Transition in Work by Sector and Gender, Panel Respondents ........................... 123

Table 4.7
Transition in Work by Age and Gender, Panel Respondents .............................. 124

Appendix Table 4.1
Distribution of Employment by Schooling Attainment, Sector and Gender,
SUSENAS ........................................................................................ 125

Table 5.1
Hausman-McFadden Tests for Independence of Irrelevant Alternative Property of

Multinomial Sector Choice Model ............................................................ 126

Table 5.2
Wald Tests for Combining Dependent Categories of Multinomial Sector Choice
Model .............................................................................................. 127

Table 5.3A
Multinomial Logit for Sector Choice Model: The Effects of Non-linear Own
Schooling and Parental Schooling (Base Speciﬁcation), IFLSl ........................... 128

Table 5.33
Multinomial Logit for Sector Choice Model: The Effects of Non-linear Own
Schooling and Parental Schooling (Base Speciﬁcation), IF LS2 ........................... 129

Table 5.3C
Multinomial Logit for Sector Choice Model: The Effects of Non-linear Own
Schooling and Parental Schooling (Base Speciﬁcation), IF LS3 ........................... 130

Table 5.4
Multinomial Logit for Sector Choice Model: Summary of the Effects of Own
Schooling ......................................................................................... 131

Table 5.5
Multinomial Logit for Sector Choice Model: Summary of the Effects of Non-linear
Own Schooling and Parental Schooling ...................................................... 132

Appendix Table 5.1A
Multinomial Logit for Sector Choice Model: The Effects of Linear Own Schooling,
IFLS 1 .............................................................................................. 133

Appendix Table 5.1B
Multinomial Logit for Sector Choice Model: The Effects of Linear Own Schooling,
IFLSZ .............................................................................................. 134

Appendix Table 5.1C
Multinomial Logit for Sector Choice Model: The Effects of Linear Own Schooling,
IFLS3 .............................................................................................. 135

Appendix Table 5.2A
Multinomial Logit for Sector Choice Model: The Effects of Non-linear Own
Schooling, IFLSl ................................................................................ 136

Appendix Table 5.2B
Multinomial Logit for Sector Choice Model: The Effects of Non-linear Own
Schooling, IFLSZ ................................................................................ 137

Appendix Table 5.2C
Multinomial Logit for Sector Choice Model: The Effects of Non-linear Own
Schooling, IFLS3 ................................................................................ 138

Appendix Table 5.3A
Multinomial Logit for Sector Choice Model: The Effects of Non-linear Own
Schooling, Parental Schooling and Residency (Full Speciﬁcation), IFLSl .............. 139

Appendix Table 5.3B

Multinomial Logit for Sector Choice Model: The Effects of Non-linear Own
Schooling, Parental Schooling and Residency (Full Speciﬁcation), IFLS2 .............. 141

xi

Appendix Table 5.3C
Multinomial Logit for Sector Choice Model: The Effects of Non-linear Own

Schooling, Parental Schooling and Residency (Full Speciﬁcation), IFLS3 .............

Appendix Table 5.4A
Multinomial Logit for Sector Choice Model: The Effects of Linear Own Schooling,

Panel Respondents, IF LS1 ....................................................................

Appendix Table 5.4B
Multinomial Logit for Sector Choice Model: The Effects of Linear Own Schooling,

Panel Respondents, IFLS2 ....................................................................

Appendix Table 5.4C
Multinorrrial Logit for Sector Choice Model: The Effects of Linear Own Schooling,

Panel Respondents, IF LS3 .....................................................................

Appendix Table 5.5A
Multinomial Logit for Sector Choice Model: The Effects of Non-linear Own

Schooling, Panel Respondents, IFLSl .......................................................

Appendix Table 5.5B
Multinomial Logit for Sector Choice Model: The Effects of Non-linear Own

Schooling, Panel Respondents, IFLS2 .......................................................

Appendix Table 5.5C
Multinomial Logit for Sector Choice Model: The Effects of Non-linear Own

Schooling, Panel Respondents, IFLS3 .......................................................

Appendix Table 5.6A
Multinomial Logit for Sector Choice Model: The Effects of Non-linear Own

Schooling and Parental Schooling (Base Speciﬁcation), Panel Respondents, IFLSl

Appendix Table 5.6B
Multinomial Logit for Sector Choice Model: The Effects of Non-linear Own

Schooling and Parental Schooling (Base Speciﬁcation), Panel Respondents, IFLS2 . ..

Appendix Table 5.6C
Multinomial Logit for Sector Choice Model: The Effects of Non-linear Own

Schooling and Parental Schooling (Base Speciﬁcation), Panel Respondents, IFLS3

Appendix Table 5.7
Primary Duties within Sector of Employment by Gender and Its Average Years of

Schooling .........................................................................................

Appendix Table 5.8

Primary Duties within Sector of Employment by Gender, Panel Respondents .........

xii

. 143

. 145

. 146

. 147

. 148

. 149

.150

151

152

153

154

155

Table 6.1
Hourly Wage within Sector of Employment by Years of Schooling and Sex ........... 156

Table 6.2A
Selectivity Corrected Wage Functions: The Effects of Non-linear Own Schooling,
IFLSl .............................................................................................. 158

Table 6.2B
Selectivity Corrected Wage Functions: The Effects of Non-linear Own Schooling,
IFLS2 .............................................................................................. 159

Table 6.2C
Selectivity Corrected Wage Functions: The Effects of Non-linear Own Schooling,
IFLS3 .............................................................................................. 160

Table 6.3
Selectivity Corrected Wage Functions: Summary of the Effects of Own Schooling 161

Table 6.4A
Selectivity Corrected Wage Functions: The Effects of Non-linear Own Schooling
and Parental Schooling, IF LS1 ................................................................. 162

Table 6.4B
Selectivity Corrected Wage Functions: The Effects of Non-linear Own Schooling
and Parental Schooling, IF LS2 ................................................................. 163

Table 6.4C
Selectivity Corrected Wage Functions: The Effects of Non-linear Own Schooling
and Parental Schooling, IF LS3 ................................................................. 164

Table 6.5
Selectivity Corrected Wage Functions: Summary of the Effects of Non-Linear Own
Schooling and Parental Schooling ............................................................. 165

Table 6.6A

Selectivity Corrected Wage Functions:

The Effects of Non-linear Own Schooling, Parental Schooling and Residency,

IFLS] .............................................................................................. 166

Table 6.6B
Selectivity Corrected Wage Functions:

The Effects of Non-linear Own Schooling, Parental Schooling and Residency,
IFLSZ .............................................................................................. 168

xiii

Table 6.6C

Selectivity Corrected Wage Functions:

The Effects of Non-linear Own Schooling, Parental Schooling and Residency,

IFLS3 .............................................................................................. 170

Table 6.7
Selectivity Corrected Wage Functions: Summary of the Effects of Non-linear Own
Schooling, Parental Schooling and Residency ............................................... 172

Table 6.8
Predicted Returns to Schooling: Speciﬁcations that Include Own Schooling ........... 173

Table 6.9
Predicted Returns to Schooling: Speciﬁcations that Include Own Schooling and
Parental Schooling .............................................................................. 174

Appendix Table 6.1A
Selectivity Corrected Wage Functions: The Effects of Linear Own Schooling,
IFLSl .............................................................................................. 175

Appendix Table 6.1B
Selectivity Corrected Wage Functions: The Effects of Linear Own Schooling,
IFLS2 .............................................................................................. 176

Appendix Table 6.1C
Selectivity Corrected Wage Functions: The Effects of Linear Own Schooling,
IF LS3 .............................................................................................. 177

Appendix Table 6.2

Selectivity Corrected Wage Functions: The Effects of Non-linear Own Schooling,
Parental Schooling, Residency, Household Compositions and Business Assets for
Self-Employed Workers ........................................................................ 178

Appendix Table 6.3A
Selectivity Corrected Wage FunctionsThe Effects of Own Schooling (splines),
IFLSl .............................................................................................. 180

Appendix Table 6.38
Selectivity Corrected Wage F unctionsThe Effects of Own Schooling (splines),
IFLS2 .............................................................................................. 181

Appendix Table 6.3C

Selectivity Corrected Wage FunctionsThe Effects of Own Schooling (splines),
IF LS3 .............................................................................................. 182

xiv

Appendix Table 6.4A
Selectivity Corrected Wage Functions: The Effects of Own Schooling (splines)
and Parental Schooling, IF LSl ................................................................. 183

Appendix Table 6.4B
Selectivity Corrected Wage Functions: The Effects of Own Schooling (splines)
and Parental Schooling, IF L82 ................................................................. 184

Appendix Table 6.4C
Selectivity Corrected Wage Functions: The Effects of Own Schooling (splines)
and Parental Schooling, IFLS3 ................................................................. 185

Appendix Table 6.5A
OLS Wage Functions: The Effects of Non-linear Own Schooling and Parental
Schooling, IF LSl ................................................................................. 186

Appendix Table 6.5B
OLS Wage Functions: The Effects of Non-linear Own Schooling and Parental
Schooling, IFLS2 ................................................................................ 187

Appendix Table 6.5C
OLS Wage Functions: The Effects of Non-linear Own Schooling and Parental
Schooling, IF LS3 ................................................................................ 188

Appendix Table 6.6A
Selectivity Corrected Wage Functions: The Effects of Linear Own Schooling,
Panel Respondents, IF LS1 ..................................................................... 189

Appendix Table 6.6B
Selectivity Corrected Wage Functions: The Effects of Linear Own Schooling,
Panel Respondents, IF LS2 ..................................................................... 190

Appendix Table 6.6C
Selectivity Corrected Wage Functions: The Effects of Linear Own Schooling,
Panel Respondents, IF LS3 ...................................................................... 191

Appendix Table 6.7A
Selectivity Corrected Wage Functions: The Effects of Non-linear Own Schooling,
Panel Respondents, IF LSl ..................................................................... 192

Appendix Table 6.7B

Selectivity Corrected Wage Functions: The Effects of Non-linear Own Schooling,
Panel Respondents, IF LS2 ..................................................................... 193

XV

Appendix Table 6.7C
Selectivity Corrected Wage Functions: The Effects of Non-linear Own Schooling,
Panel Respondents, IFLS3 ..................................................................... 194

Appendix Table 6.8A
Selectivity Corrected Wage Functions: The Effects of Non-linear Own Schooling
and Parental Schooling, Panel Respondents, IF LS1 ........................................ 195

Appendix Table 6.8B
Selectivity Corrected Wage Functions: The Effects of Non-linear Own Schooling
and Parental Schooling, Panel Respondents, IF LS2 ........................................ 196

Appendix Table 6.8C
Selectivity Corrected Wage Functions: The Effects of Non-linear Own Schooling
and Parental Schooling, Panel Respondents, IFLS3 ........................................ 197

Table 7.1A
Distribution of Regional Migration by Years of Schooling, IFLS] ...................... 198

Table 7.1B
Distribution of Regional Migration by Years of Schooling, IFLSZ ...................... 199

Table 7.1C
Distribution of Regional Migration by Years of Schooling, IFLS3 ...................... 200

Table 7.2
Own Schooling Attainment by Region of Birth, Region of Current Residence and
Gender ............................................................................................. 201

Table 7.3
Hourly Wage within Sector of Employment by Region of Birth and Gender ........... 202

Table 7.4
Hourly Wage within Sector of Employment by Region of Current Residence and
Gender ............................................................................................. 203

Table 7.5A
Selectivity Corrected Wage Functions: The Effects of Non-linear Own Schooling
Interacted with Born in Urban, IF LSl ........................................................ 204

Table 7.5B

Selectivity Corrected Wage Functions: The Effects of Non-linear Own Schooling
Interacted with Born in Urban, IF LS2 ........................................................ 205

xvi

 

Table 7.5C
Selectivity Corrected Wage Functions: The Effects of Non-linear Own Schooling
Interacted with Born in Urban, IF LS3 ........................................................ 206

Table 7.6
Selectivity Corrected Wage Functions: Summary of the Effects of Own Schooling
Interacted with Born in Urban ................................................................. 207

Table 7.7A
Selectivity Corrected Wage Functions: The Effects of Non-linear Own Schooling
Interacted with Urban Residence, IF LS1 ..................................................... 208

Table 7.7B
Selectivity Corrected Wage Functions: The Effects of Non-linear Own Schooling
Interacted with Urban Residence, IF L82 ..................................................... 209

Table 7 .7C
Selectivity Corrected Wage Functions: The Effects of Non-linear Own Schooling
Interacted with Urban Residence, IFLS3 ..................................................... 210

Table 7.8
Selectivity Corrected Wage Functions: Summary of the Effects of Own Schooling
Interacted with Urban Residence ............................................................. 21 1

Appendix Table 7.1A
Selectivity Corrected Wage Functions: The Effects of Linear Own Schooling
Interacted with Born in Urban, IFLSl ........................................................ 212

Appendix Table 7.1B
Selectivity Corrected Wage Functions: The Effects of Linear Own Schooling
Interacted with Born in Urban, IFLS2 ........................................................ 213

Appendix Table 7.1C
Selectivity Corrected Wage Functions: The Effects of Linear Own Schooling
Interacted with Born in Urban, IF LS3 ....................................................... 214

Appendix Table 7.2A
Selectivity Corrected Wage Functions: The Effects of Linear Own Schooling
Interacted with Urban Residence, IF LS1 ..................................................... 215

Appendix Table 7 .2B

Selectivity Corrected Wage Functions: The Effects of Linear Own Schooling
Interacted with Urban Residence, IFLSZ ..................................................... 216

xvii

 

Appendix Table 7.2C
Selectivity Corrected Wage Functions: The Effects of Linear Own Schooling
Interacted with Urban Residence, IF LS3 ..................................................... 217

Table 8.1
Own Schooling Attainment by Cohort and Gender ......................................... 218

Table 8.2
Hourly Wage within Sector of Employment by Cohort and Gender ..................... 219

Table 8.3A
Selectivity Corrected Wage Functions: The Effects of Non-linear Own Schooling
Interacted with Young Cohort, IF LS1 ......................................................... 220

Table 8.3B
Selectivity Corrected Wage Functions: The Effects of Non-linear Own Schooling
Interacted with Young Cohort, IF LS2 ......................................................... 221

Table 8.3C
Selectivity Corrected Wage Functions: The Effects of Non-linear Own Schooling
Interacted with Young Cohort, IFLS3 ......................................................... 222

Table 8.4
Selectivity Corrected Wage Functions: Summary of the Effects of Own Schooling
Interacted with Young Cohort ................................................................. 223

Appendix Table 8.1A
Selectivity Corrected Wage Functions: The Effects of Linear Own Schooling
Interacted with Young Cohort, IFLSl ......................................................... 224

Appendix Table 8.1B
Selectivity Corrected Wage Functions: The Effects of Linear Own Schooling
Interacted with Young Cohort, IF LS2 ......................................................... 225

Appendix Table 8.1C
Selectivity Corrected Wage Functions: The Effects of Linear Own Schooling
Interacted with Young Cohort, IF LS3 ......................................................... 226

Table 9.1A
Selectivity Corrected Wage Functions: The Effects of Non-linear Own Schooling
with Age and Potential Experience, IF LS1 ................................................... 227

Table 9.1B

Selectivity Corrected Wage Functions: The Effects of Non-linear Own Schooling
with Age and Potential Experience, IFLS2 ................................................... 228

xviii

Table 9.1C
Selectivity Corrected Wage Functions: The Effects of Non-linear Own Schooling

with Age and Potential Experience, IFLS3 ................................................... 229
Table 9.2

Selectivity Corrected Wage Functions: Summary of The Effects of Own Schooling

with Age and Potential Experience ............................................................ 230

Appendix Table 9.1A
Selectivity Corrected Wage Functions: The Effects of Linear Own Schooling with
Age and Potential Experience, IF LSl ......................................................... 231

Appendix Table 9.1B
Selectivity Corrected Wage Functions: The Effects of Linear Own Schooling with
Age and Potential Experience, IF LS2 ......................................................... 232

Appendix Table 9.1C
Selectivity Corrected Wage Functions: The Effects of Linear Own Schooling with

Age and Potential Experience, IF LS3 ......................................................... 233

Table 10.1

Comparison of Estimates ....................................................................... 234
xix

1. Introduction

Education determines how individuals, communities and nations progress by
improving living standards, enhancing quality of life and providing essential
opportunities for all. In order for nations to survive in a rapidly changing world,
especially to keep up with or ahead of the advancement in technological innovations, it is
essential for a country to have a highly educated and skilled population. In developing
countries, where most of the population still lives in poverty, education can also play a
crucial role in combating poverty.1

Governments in developing countries are still the major players in ﬁnancing the
education sector, especially at the primary and secondary levels. Expenditure on
education is one of the major categories of expenditure in the national budgets of many
developing countries. According to the latest ﬁgures, developing countries spend 3.8
percent of their GNP on public education (UNDP [2000]).2

One obvious question regarding these huge investments in education is what is the
rate of return. Psacharopoulos (1973, 1981, 1985 and 1994) and Psacharopoulos and
Patrinos (2002) have compiled a comprehensive review on private and social rates of

return to investment in education for both developed and developing countries during the

 

' In analyzing changes in the income distribution in Java between the years 1984 and 1990, Cameron
(2000) ﬁnds that poverty reduction was largely associated with increased educational attainment as well as
increases in the incomes of lesser educated workers and income gains to workers outside agriculture.
Increased educational attainment, however, was found to be the largest determinant of the inequality. She
concludes that as Indonesians gain more education and as productivity and hence incomes outside
agriculture increases, poverty is expected to continue to decrease and inequality to increase.

2 This ﬁgure varies among regions from 2.9 percent for East Asia countries to 6.1 percent for Sub-Saharan
Africa countries. As a comparison, the corresponding ﬁgures for OECD countries and the world are 5.0
percent and 4.8 percent, respectively.

last 30 years.3 In general, returns to schooling are positive and tend to be high. Some
patterns regarding the rate of returns emerge from these reviews. First, among the three
levels of education (primary, secondary and tertiary/higher), primary education exhibits
the highest returns of investment. Second, the private and social returns at all levels of
education decline with country per capita income. Third, returns to education are higher
for women than for men. Men have higher returns to primary education while women
experience higher returns to secondary education. Finally, individuals who work in the
private sector enjoy a higher rate of return than those in the public sector.

Empirical work on returns to schooling is, in general, based on Mincer-type (1974)
eamings or wage functions. In principal, this involves OLS regression of the natural
logarithm of earnings (or wage) as the dependent variable with years of schooling and
potential years of experience in the labor market and its square as independent variables.
Early studies on earnings-schooling relationship suffer from two basic problems (see
Card [1999], Schultz [1988] and Strauss and Thomas [1995] for surveys of research).
These problems, which are often dictated by the nature of the available data, are omitted
variable bias and sample selection bias.

Omitted variable bias resulted from omitted measures of ability, family background
and schooling quality. To encounter the problem of ability bias, measures that proxy for

unobserved ability such as IQ and other test results are sometimes included in the wage

 

3 Returns to schooling can be estimated from the private and social points of view. The private rate of
return is calculated by equating (in discounted present value term) a stream of after-tax education beneﬁts
to a stream of educational costs. In addition to the direct costs of education (education fees, book
allowance and other incidental expenses), the costs incurred by the individual also include individuals’
foregone earnings while studying. Social rate of return adds taxes and any net positive social externalities
(on the beneﬁts side of the equation) and augment the public and private subsidies not borne by the
individual (on the costs side). To the extent that education, especially in the primary level, is highly
subsidized, social returns are therefore lower than the private returns.

function.4 The inclusion of proxies for ability has been criticized due to its difﬁculty to
ﬁnd ability measures that are not themselves determined, and thus correlated, with
schooling. Many studies have shown the importance of examining the impact of family
background on earnings. The long tradition in using family background information,
such as parental education and income, is intended to control for unobserved ability or to
proxy human capital investment made by parents not captured by completed years of
schooling. The facts that parents may send their children to better quality schools, or
children who continue to the secondary and higher levels may go to better primary
schools, are examples that parental background measures may also control for school
quality.

Sample selection bias arises because not all individuals in the population report
information about earnings or wages. If the criterion for inclusion in the sample involves
the choice of occupation, labor force participation or migration, then sample selection is
likely to be more important. In practice, estimation of wage functions should incorporate
correction for sample selection by taking into account the fact that the decision to be in
the wage labor market is a choice.

The availability of new and richer data sets as well as more sophisticated estimation

techniques enables researchers to obtain better rate of return estimates.5 Unfortunately,

 

’ Omitting ability or its proxies from the wage function will upward biased the rate of retum estimates, in
most cases by no more than 5-15 percent (Schultz [1988]).

5 Another approach in this literature is the estimation of causal effect of schooling on earnings by using
institutional features of the supply side of the education system as exogenous determinants of schooling
outcomes (see Card [1999] for review). Some examples of supply side variables used in the literature are
variation in individual’s quarter of birth (Angn'st and Krueger [1992]), geographic variation in college
proximity (Card [1993]), changes in legal minimum school-leaving age (Harmon and Walker [1995]), and
major school construction program (Duﬂo [2000] and Clark and Hsieh [2000]). The basic idea of using
these institutional features is to use supply-side shocks to identify demand-side parameters and, thus, places
returns to schooling in a standard “supply and demand” framework.

 

 

 

due to the unavailability of data, studies on the returns to schooling in Indonesia that were
written in the 19903 (Behrrnan and Deolalikar [1991, 1993 and 1995], Deolalikar [1993]
and Duﬂo [2000]) do not explicitly control for family background.6 This implies that we
are not able to learn, for example, the effects of father or mother’s schooling on returns to
schooling.7 As pointed out in Schultz (1988), since more educated parents tend to invest
more in the education of their children, there is no reason why parental schooling should
not be included to examine returns to schooling.8

In this essay I will estimate the private rate of return to schooling using there waves

of Indonesia Family Life Survey (IFLS) data. Detailed information on individuals’ and

 

An interesting ﬁnding of Card (2000) is that the instumental variables (IV) estimates of the returns to
schooling in this literature are typically exceeding the corresponding OLS estimates. One explanation of
this upward bias in the IV estimates is due to the underlying heterogeneity in the returns to education.
Many of the IV estimates that are based on supply-side institutional reforms tend to recover returns to
schooling for a subset of individuals with relatively high return to education. Accessibility of schools, for
example, tends to affect schooling choice of individuals who would otherwise have relatively low
schooling. If the main reason that an individual has low schooling is driven by cost of schooling then IV
estimates based on accessibility of schools will yield an estimate of returns to schooling that is higher than
the marginal returns to schooling in the population. Although institutional feature on the supply-side of
education provides a richer econometric model, the resulted IV estimates should be interpreted with
caution.

6 Bedi and Gaston (2000) is an exception. The papers by Behrrnan and Deolalikar do, however, control for
unobserved household and community heterogeneity effects. Each of the above-mentioned studies
addresses the selectivity bias problem. Deolalikar (1993), in addition to the estimation of returns to
schooling, also estimates schooling enrollment rates. He ﬁnds, among others, that the schooling of head
and his spouse in a male-headed household has some of the strongest effects on enrollment probabilities;
and that the female household head and her spouse do not have any signiﬁcant effects.

7 Parental schooling also affects other aspect of human capital investment on their children like, for
example, health status. Skouﬁas (1999) examines the impact of parental education on the nutritional status
of pre-school school children (age 5 years and under), measured by weight-for-age Z-scores, in Indonesia.
He estimates reduced form for child health status by taking into account unobserved heterogeneity at the
cluster (village); this is done by controlling for the difference in prices and in the quantity and quality of
available medical services. Parental schooling is found to have a signiﬁcantly positive effect on the health
status of children and that these effects vary in urban and rural areas as well as by child gender and child
age. The estimates show that mothers with secondary education have healthier boys than those with lower
levels of education. For girls, the positive health status effects exist only if mother have education above
senior high school level.

8 More educated parents also tend to provide a more favorable learning environment at home with the
implication of lowering the cost of education or increasing the market productivity of attending school for a
given number of year. They may also provide their children with a better social connection so that their
children can get a favorable job.

 

on his/her parents’ characteristics is gathered in the survey. Household level as well as
community and facility level information enriches this survey. The richness of the data
enables the construction of family background information, something that is lacking in
the previous studies, to overcome the omitted variable bias problem. Realizing that most
of the individuals who participate in the labor market are not working for a wage, I will
model the employment outcomes in several labor market alternatives; thus correcting for
selectivity bias problem. The richness of this survey, however, has to be compensated
with the relatively small sample size compared to the data sets used by the above-
mentioned studies.

To be more speciﬁc, I will estimate returns to investment in education by controlling
for parental (father’s and mother’s) schooling. The rates of returns will be estimated for
those who are self-employed as well as those working in the public and private sectors.
Separate estimates for men and women will be presented in this essay since these rates
tend to vary by gender (see Behrrnan and Deolalikar [1993, 1995] for the case of
Indonesia). In addition, I will also utilize information on migration (in particular, the
place of birth and of current residency of the respondents) in the survey to estimate the
effect of migration on rate of return to investment in education (Schultz [1988]).

The utilization of samples covering three periods of time (1993, 1997 and 2000) is
expected to give a better understanding of how returns to schooling investment evolved
over time in Indonesia during the 19905.9 During that span of time, Indonesia has
experience a period of remarkable growth in its economy from 1993 to 1997, a multi

dimensional crisis by that started by late 1997, and a recovery period since mid-1998.

The dynamics of returns to schooling investment during this period is of interest.
Although four years might not be long enough to see the difference in return to schooling,
it is interesting to see whether the rates decreased as per capita expenditure increased and
as the individuals became more educated during the prosperous stage of 1993-1997. The
impact of the economic crisis of 1997-98 has been studied thoroughly using various data
sets, including the IF LS.10 In spite of it, comparison of returns to schooling during this
period is yet to be studied. If due to the crises, a particular sector became more selective
in employing their workers like increasing the minimum level of schooling attainment or
experience with no signiﬁcant increase in compensation structure, returns to schooling in
that sector might fall. These are the issues that will be explored in this paper.

The panel nature of the data is not exploited in this paper, but it is left for ﬁiture work.
However, separate estimates for panel individuals for each survey year are presented in
this work in addition to the estimates for cross-section individuals. The purpose of the
presentation of these panel individual estimates is to provide comparison of the changes
in retruns to schooling between the two sets of individuals.

Some important ﬁndings of this study are the following. Individuals with more
formal schooling are more favorably rewarded and, thus, have higher returns to

schooling. Wage advantage increases in a convex fashion with levels of schooling. Over

 

9 Psacharopoulos and Patrinos (2002) ﬁnd that worldwide average returns to schooling during the 1990
decade have declined by 0.6 percent, while average schooling attainment during the same period have
increased.

w Frankenberg, Thomas and Beegle (1999) look at evidence of the immediate impact of the economic
shocks on a broad array of economic well being measures including children’s schooling and adults’
employment and wages, using IF L82 and IFLSZ+ (conducted on the second half of 1998) data sets. A
more detail study on how individuals and households are affected and how did they respond to the crisis in
relation to schooling of the young is reported in Thomas, Beegle, Frankenberg, Sikoki, Strauss and Teruel
(2001). Another comparison of a range of well being measure three years after the crisis could be found in
Strauss, Beegle, Dwiyanto, Herawati, Pattinasarany, Satriawan, Sikoki, Sukamdi and Witoelar (2002)
based on IFLSZ and IFLS3 data sets.

years, wage advantages within each level of schooling are, in general, declining. Supply
shifts among individuals with more schooling might serve as one of the possible
explanations for the declining rates.

Parental education captures part of individuals’ family background information as the
inclusion of it in the wage regression reduces estimated coefﬁcients of individuals’
schooling variables. Although the effects of parental schooling on individuals’ returns
are limited, the results reveal that having educated parents is associated with an increase
in earnings relative to having illiterate parents.

Sample selection bias does not turn to be important in these wage regressions.
Selectivity correction terms are mainly signiﬁcant in the self-employment wage
regressions. Likewise, the potential bias from migration seems to be small. Individuals
who were born or are currently living in urban areas do not have a wage advantage that is
signiﬁcantly different from their counterparts who were born or are currently living in
rural areas. Stratifying individuals based on their age reveals that younger individuals
(aged 25-39) who are working in the private sector have a lower rate of return in
comparison to that of the older generation (aged 40-59).

This essay is organized as follows. In Section 2, studies on the returns to schooling in
Indonesia are reviewed. Section 3 describes the model and empirical strategy, while
section 4 discusses the data. Presentation of empirical results and discussions of ﬁndings
starts in section 5 with sector choice model, followed by returns to schooling estimates in
Section 6. In Section 7, to control for migration, 1 will present the estimates by taking
into account the importance of place of birth and place of current residence. Since

availability and quality of schooling may differ over time, in Section 8 the returns to

schooling are estimated by incorporating age-cohort effect. In Section 9, returns to
schooling estimates with potential experience (deﬁned as the duration since an individual
ﬁnished/quit school) will be presented as an alternative to using age as approximation of
out-of-school working experience. This essay concludes in section 10 with summary of

ﬁndings and their policy implications.

2. Previous Empirical Studies on Indonesia

This section starts with short reviews of empirical studies on the returns to schooling
in Indonesia that were written prior to 1990. Studies written during the 19903 and
beyond are characterized by richer speciﬁcations as well as more sophisticated
methodologies, in addition to the availability of better quality household survey data. For
that reason, following the review of the earlier empirical ﬁndings, these studies will be

reviewed in more detail.

2.1. Early Empirical Findings

Simanjuntak (1981) provides one of the earliest empirical results on earnings—
schooling relationship in Indonesia. He ﬁnds that, in 1976, the private rates of return for
individual who completed secondary and higher education are 14.5 percent and 19.7
percent, respectively.

Psacharopoulos (1985) compiles returns to schooling estimates, both private and
social rates, on 61 countries. For Indonesia, in 1977, he ﬁnds that the private rates are
25.5 percent for primary school (relative to no schooling) and 15.6 percent for secondary
school (relative to primary school graduate). The corresponding social rates, in 1978, are
21.9 percent for primary, 16.2 percent for secondary and 14.8 percent for higher
education (relative to secondary graduate).

Clark (1983) estimates social rates of returns for secondary schooling in determining
whether secondary education in Indonesia is a good investment. He uses data from a

tracer study for a cohort of senior secondary school graduates covering the ﬁrst two years

after ﬁnishing school in 1976. An individual who graduated from senior secondary
school has a return of 24 percent if he or she can get a job right after graduating. A one-,
two- and three- year waiting period will reduced the rates to 19 percent, 16 percent, and
14 percent, respectively. In addition, individuals who graduate from general secondary
school have a signiﬁcant advantage in the rates of returns compared to those who
graduated from vocational secondary school.

Social rates of return were also estimated in McMahon and Boediono (1992). They
calculate rates for 1982, 1986, 1988 and 1989 for different levels of education starting
from junior secondary level using various rounds of Indonesian Labor Force Survey
(SAKERNAS) data. The average return to general junior secondary education is 14.5
percent, while the corresponding ﬁgure to general senior secondary education is 15.5

percent, although the rates are falling with time.

2.2. Empirical Studies of the 1990s

Six studies on returns to schooling that were written during the 1990s and beyond will
be reviewed next. Our review starts with three papers by Behrrnan and Deolalikar (1991,
1993 and 1995). Deolalikar (1993), Duﬂo (2000), Bedi and Gaston (2000) and Filmer
and Lindauer (2001) will follow the review. The discussion emphasizes on the
methodology used, along with its advantages and disadvantages, in order to build a better
model that will be discussed in the next section.

Behrrnan and Deolalikar (1991, 1993 and 1995) examine the earnings-schooling
relationships for individuals who report wages using the 1986 SAKERNAS data. A

unique feature of their 1991 paper is the adjustment made for the time spent in each level

10

of schooling due to repetition and dropout. They develop a model based on two (very
strong) assumptions since the data do not provide information about individual-speciﬁc
dropout and repetition experience.'1 The ﬁrst assumption, the homogeneity assumption,
states that everyone who enters a given grade is subject to the same average repetition
and dropout rates. Secondly, in the heterogeneity assumption, they assume that students
who enter the next level of schooling have zero values for the repetition and dropout rates
at all lower levels. In addition, students who do not enter the next level have
appropriately adjusted repetition and dropout rates at the lower levels.

Based on these assumptions, they ﬁnd that failure to control for repetition and
dropout rates will signiﬁcantly upward biased the returns to schooling estimates,
especially for the lower levels. For example, at the subprimary level (i.e., for individuals
with 1-3 years of schooling), the rate of returns to schooling is overestimated by 82-114
percent; while for individuals with completed primary schooling, the rate is
overestimated by 38-78 percent. They also divide the data by sex, age, region of
residence and urbanization, and ﬁnd that the biases differ among subsarnples. The biases
are somewhat larger for females, older individuals, and those who live in rural and
relatively remote areas.

Behnnan and Deolalikar (1993 and 1995) analyze whether there are gender
differences on the returns to education in Indonesia. The main feature of these papers is
that they estimate eamings-schooling relationship by controlling, in addition to the time

spent in schooling due to repetition and dropout, for unobserved household and

 

” The standard practice is to assign six years for completed years of schooling variable if an individual
completed primary schooling since there are six grade of primary school, nine years if he or she completed
junior secondary school, etc.

11

 

community heterogeneity via household ﬁxed effects estimation. They estimate wage
rate and hours supplied to the paid labor market equations in their 1993 paper, while the
focus on their 1995 paper is on the wage rate and earnings equations.

Schooling variables in the two papers differ although both attempt to capture non-
linear impact of schooling, the different effects of vocational and general secondary
schooling, and the different effect post-secondary diploma programs and university level
education on the returns to schooling. In their 1993 paper, schooling variables are
represented by completed year of schooling, its square, and its interaction with
dichotomous variables for vocational and diploma schooling. Behrrnan and Deolalikar
(1995), on the other hand, represent schooling with nine dummy variables, namely no
schooling, subprimary, primary completed, vocational junior secondary completed,
general junior secondary completed, vocational senior secondary completed, general
senior secondary completed, post-secondary diploma completed, and university
completed. Potential selectivity bias is addressed by applying Olsen’s least-squares
correction for selectivity bias using household demographic variables, such as the number
of young children and the number of older individuals in the household, to predict
participation.”

With respect to returns to schooling, both papers ﬁnd that estimates of the impact of
schooling on wage rates, on hours supplied to the paid labor market and on earnings that

do not control for unobserved household and community ﬁxed effects (i.e. OLS

 

'2 Based on results from various studies, they claim that the household ﬁxed effects procedure may control
for selectivity bias. The reason behind this observation is as follows. The use of ﬁxed effects procedure,
that controls for unobserved household and community characteristics shared by household members and
included additively in the relation of interest, is intended to avoid omitted variable bias in the schooling
coefﬁcients. Many among these characteristics are also used in the determination of labor market
participation (such as household unearned income and household composition). This implies that ﬁxed
effects estimates should also limit or eliminate selectivity bias.

12

estimates) are substantially upward biased. Therefore, controlling for household and
community ﬁxed effects is likely to control for a substantial share of the relevant
unobserved factors.13

The marginal increases of wage rates and earnings for individuals with post-primary
schooling are greater (in percentage terms) for females than for males relative to those of
primary school graduates. Overall, their estimates do not suggest that females face strong
discrimination in the form of lower rates of return to post-primary schooling investment.
Their results, however, may indicate that private rates of return to schooling investments
in females are higher than are those of males. They also ﬁnd that gender earnings
differentials change more over the life cycle than do the gender wage rates differentials
due to gender differences in hours worked over individual’s life cycle. This implies,
ceteris paribus, that considerably higher wage rates and eamings for males than for
females. Thus, there exists a large wedge between male and female wage rates and a
larger one between male and female eamings.

Deolalikar (1993) estimates school enrollment model for four age groups: 6-11, 12-
14, 15—17, and 18-23 years. These groups roughly correspond to the four levels of
schooling: primary, lower secondary, higher secondary, and tertiary, respectively. He
also estimates the pecuniary returns to schooling (in terms of earnings) for adult men and

women aged above 15 years who have completed their schooling. The data set used in

 

'3 In their 1993 paper, the magnitudes of the biases are calculated. The results indicate that the estimated
biases differ substantially between males and females, between wage rate and hours supplied to the labor
market speciﬁcations, and among schooling levels. The upward bias in the wage speciﬁcation for males
decreases with level of schooling from almost 100 percent for males with 3 years of schooling to about 15
percent for those with 12 years of schooling. For females, returns estimates are biased upward up to 6
years of schooling and the bias become negative afterward. In the hours supplied to the paid labor market
estimates the biases tend to be upward, implying that OLS estimates provide greater estimates than the
ﬁxed effects estimates. For females the magnitude of the bias falls with years of schooling, while for males
no trend was found.

13

this study is the 1987 round of the National Socioeconomic Survey (SUSENAS). As in
Behrrnan and Deolalikar (1995), he uses nine schooling categories and their interaction
with age in wage regressions. The sample selectivity problem is corrected by using
Heckman’s selectivity correction method and by estimating expanded wage function
using OLS.14

The empirical results show signiﬁcant gender differences in the deternrinants of
earnings.15 Returns to schooling are signiﬁcantly lower for men than women at the
secondary and tertiary level. Age-cohort effects do matter in the wage function. Older
cohorts are enjoying a signiﬁcantly higher return to schooling, especially at the secondary
and tertiary levels, than the younger cohorts. For illustration, the rates of return to
primary school graduates for 20-, 40- and 60-year-old men are 6.8 percent, 10.9 percent
and 15.0 percent, respectively, while the corresponding rates for women are, respectively,
6.0 percent, 10.1 and 14.2 percent. The corresponding rates for general higher secondary
graduates are 8.8 percent, 12.9 percent and 17.1 percent for men and 10.5 percent, 14.7
percent and 18.8 percent for women. With respect to the age, the coefﬁcients on age and

age-squared differ signiﬁcantly across gender. For women, the coefﬁcient on both

 

” There is an identiﬁcation problem with using annual earnings as dependent variable in the eamings
ﬁinction. To the extent that annual earnings are the product of hourly wage rate and the number of hours
worked during the year, the wage function should include all the regressors in the market wage and labor
supply equations. In their ﬁrst earnings speciﬁcation, the one with selectivity correction, the correction is
identiﬁed by marital status, household non-labor income and spouse’s age, with the assumption that hours
worked are exogenously determined. The second one, estimated using OLS with no control for sample
selectivity, is an expanded eamings function that includes marital status, non-labor income and spouse’s
age.

Empirical results on the ﬁrst speciﬁcation show that selectivity term is highly signiﬁcant, implying that
sample selectivity is important in inﬂuencing earnings. Schooling coefﬁcients on both speciﬁcations are
signiﬁcant and have the same sign. The magnitude of the schooling coefﬁcients, however, appears to be
larger in the speciﬁcation with selectivity correction. In addition, estimates on the expanded speciﬁcation
show a signiﬁcant gender difference in the primary schooling coefﬁcient. In light of these ﬁndings,
coefﬁcients from speciﬁcation with selectivity bias correction should be interpreted with caution.

l4

variables are not signiﬁcantly different from zero, implying that age does not have
additional effect on earnings other than increasing the returns to schooling. On the other
hand, the coefﬁcient on age for men is signiﬁcantly positive while that on age-squared is
signiﬁcantly negative, indicating that earnings growth is positive and large early in the
life-cycle but falls off rapidly with age, independent of schooling.

Duﬂo (2000) uses a very different approach.16 In this study she examines the
education and earnings trends associated with a major school construction program in
Indonesia, the Sekolah Dasar INPRES program that started in 1973. She links education
and wages information from a large cross-section of men born between 1950 and 1972
from the 1995 intercensal survey of Indonesia (SUPAS) combined with district-level data
on the number of new school built between 1973/74 and 1978/79 in his/her region of
birth. Duﬂo shows that average education attainment for children who entered school
later in the 19703 was higher than those who ﬁnished primary school before 1974. She
also shows that children who lived in districts with greater program intensity, measured
by the target number of new school per primary-school age student in 1971, had, on the
average, higher education attainment. Based on these ﬁndings she uses interactions
between dummy variables indicating age of the individual in 1974 and the intensity of the
program in his region of birth between 1973 and 1978 as exogenous variables and the
instruments in the earnings equation.

OLS estimates show that the rate of return to schooling is 7.8 percent. The point

estimate using 2SLS is 6.75 percent, slightly lower than and signiﬁcantly different ﬁ'om

 

'5 The results discussed here are based on the expanded earnings function after imposing a common set of
coefﬁcient (for men and women) on the age-schooling interacted variables.

'6 This study is an example of the literature that uses changes in the (institutional) supply side of the
educational system as exogenous determinants of schooling outcomes. See Card (2000) for a review.

15

that of OLS. Interacting the year-of-birth dummies with enrollment rate in 1971
increases the estimate to 8.1 percent. Introducing a control for the water and sanitation
program ﬁirther increases the estimate to 10.6 percent. These 2SLS estimates, in contrast
to the general ﬁndings of literature in this genre (see Card [2000]), are not very different
from the OLS estimates.

Another ﬁnding of Duﬁo’s paper is that there is some indication that the return to
schooling is not concave.l7 She also ﬁnds that the estimated marginal returns do not
show signiﬁcant variation for the ﬁrst nine years of schooling. The marginal returns for
the twelfth year (the last year of senior secondary school) and thirteenth year of
schooling, however, is high, implying an indication of ’sheepskin effect’. The retums to
schooling show some variation across regions. In regions with below median population
density, the returns are higher (11.0 percent). The returns are also higher in regions
where the average education level of individuals not exposed to the program is low (12
percent).

Realizing that only 45 percent of the individuals in the sample work for wages and
most of the remaining are self-employed, Duﬂo corrects the possibility of sample
selection bias by using a sample correction procedure and by imputing an income to self-
employed individuals in the sample. The results show that sample selection is not an
important problem in her study.

Bedi and Gaston (2000) investigate the effectiveness of public and private schools at
the secondary level using IFLSl data. The sample used is restricted to individuals who

have 7 to 12 years of schooling, no longer attending schools, and provide information on

_

'7 This result is obtained from OLS estimation using a dummy for each year of education.

16

 

earnings; thus the sample used is selective implying that the generalization of their result
is limited.18 They start their analysis by sorting individuals into four choices of schools:
public, private non-religious, private-Islamic and private-Christian schools. They take
into account the fact that school-sorting process may not be exogenous by modeling it
using multinomial logit. The sorting itself is based on demand and supply considerations
and it reﬂects parental choice and selection criteria determined by school (for example,
minimum test score on primary school examination to enter junior secondary school and
religion). They next estimate earnings equations for each of the four school choices by
including selectivity correction term from the school-sorting model. Estimates of these
earnings equations are then used to decompose earnings differentials between public and
private school graduates.

Since our interest is in the returns to schooling, only results from earnings equations
are discussed here. Variables in the wage function control for individual characteristics,
parental background, ability and school inputs. Individual characteristics include
dummies whether the individual completed junior secondary and senior secondary
education, age (and its square), male dummy, dummy whether Indonesian language is
spoken at home, religion dummies, urban dummy, and region of residence dummies.
Parental background consists of ﬁve parental education dummies. Ability measures are
represented by dummies indicating whether an individual received scholarships while

studying at the secondary school, and whether he or she ever failed a grade during

 

’8 Due to the small sample (N = 1,194), they do not separate between individuals who work for wage and
those who are self-employed and between males and females. For those in the wage sector, earnings are
net wages; while returns to labor and enterprise (net income after expenses) reﬂects earnings in the self-
employment sector. A further investigation on this issue shows that, while there are some differences
between the “pool” and “separate” estimates, the key result is unchanged. The same results obtained from
speciﬁcation checks for pooling among males and females.

17

 

primary school. Three variables capture the school inputs measures; they are the
characteristics of the last school attended by the individuals. These variables are class
size, length of school term, and dummy indicating whether school has dirt ﬂoor (a
measure of school infrastructure).

Schooling variables, in generally, have signiﬁcant positive signs. As we expect, wage
functions estimates will, in addition to the effect of schooling on earnings, reveal a lot of
interesting information such as the effects of parental education, ability measures, and
school inputs to earnings. For example, for individuals who went to public school, those
who won scholarships earn a wage that is 40 percent higher than those who have not, and
those who failed a grade in primary school earn 14 percent less than those who have not.
Parental education, on the other hand, has, on the average, very little effect on eamings.
With regard to school inputs, individuals educated in school with poor infrastructure earn
25-75 percent less than those who went to school with adequate infrastructure do.

Finally, Filmer and Lindauer (2001) study the relationship between public/private
sector compensation levels and schooling attainment using evidence ﬁom SAKERNAS
1998 and SUSENAS 1999. In comparing wage levels, they ﬁnd that, on average,
government (public) sector earnings exceed those of non-govemment (private) sector for
individuals with up to senior secondary level of schooling. For individuals with
schooling attainment beyond senior secondary level, private sector workers are paid more
than their public sector counterparts. Overall, public sector workers earn more than those
working at the private sector with private to public pay ratio of 0.7: 1.

Wage regression speciﬁcations that they use consist of 5 dummy variables for

schooling attainment (primary, junior secondary, senior secondary, some tertiary and

18

university) and their interactions with public sector dummy variable. The determination
to join work force is corrected using Heckrnan selection model with some household
composition variables as identifying instruments. They also show results based on
household ﬁxed effect model.

They ﬁnd no evidence that government is a low pay employer for its average
employee. For the more educated employee, even tough they earn less than they would
in the private sector, the premium is not as large as commonly suspected. Based on these
results they cast doubt on the proposition that low pay is an explanation for government

corruption.

2.3. Discussion

The above literature review shows various ways in the estimation of returns to
schooling that has been done for the case of Indonesia. The ﬁndings can be summarized
as follows. First, the estimated rates of returns are positive but the magnitude varies
depending on the methodology and data set used. An additional year of completed
schooling for men and women with 1-3 years of schooling (sub primary) will increase
wage by 2.4 to 8.0 percent, while for those with 4-6 years of schooling (primary) between
3.3 and 6.7 percent. Among individuals with secondary schooling attainment, estimated
returns rates are 4.2-27.6 percent for lower secondary level (7-9 years of schooling) and
5.1-28.4 percent for upper secondary level (10-12 years of schooling). Beyond secondary
level, an additional year of formal schooling is associated with a 5.4-12.2 increase in
wage. Second, controlling for unobserved household and community effects results in

returns to schooling estimates that are, in general, lower than those produced using OLS.

l9

Third, selectivity sample bias does not seem to signiﬁcantly change the resulted rate of
return estimates, although the sample correction terms are signiﬁcant in some
speciﬁcations. Fourth, there are signiﬁcant gender differences in the returns to schooling
at all levels of schooling with females having the edge on secondary and tertiary levels.
Fifth, using institutional changes in the supply side of education system as determinants
of schooling outcomes is an interesting alternative in studying the returns to schooling.
Finally, the use of family background, child ability and school inputs information enables
researchers to reveal many interesting facts about the effects of these factors on the
estimated rate of return.

The model to be developed in the next section will not use the strategy similar to that
of Duﬂo for the following reasons. First, for the purpose of instrument identiﬁcation, this
procedure requires a large sample size for a population with speciﬁc age range. Second,
there basically two candidates instruments for representing changes in institutional
changes in education policy: (i) the launch of compulsory primary education program for
7-12 year old children (1984) and (ii) the launch of the nine year basic education program
consisting of 6 years of primary education and 3 years of lower secondary education or
equivalent (1994). These two instrumental variables candidates are impossible to
implement because, in addition to small sample size problem, it is too early to evaluate
the two education policy changes with IFLS data collected between 1993 and 2000.19

Controlling for unobserved household and community heterogeneity also requires a
large number of data since, to be included in the sample, one needs observations with at

least two individuals from each household. Besides, the issue for these effects has been

20

thoroughly explored by Behnnan and Deolalikar so that additional gain of knowledge to
be obtained by re-doing this exercise will be limited.

This leaves us with estimating returns to schooling by explicitly addressing the
omitted variable bias and sample selectivity bias problems. The latter is captured by
taking into account that the individual, in making employment decision (not working or
working in a particular sector of employment), maximizes his/her utility. Parental
background, in particular information on parental schooling will be utilized to encounter

omitted variable bias.

 

'9 Notice that children aged 7-12 in 1984, the year when the compulsory primary education began, are of
age 20-25 in 1997 and 23-28 in 2000. Thus, this law only affects a very limited number of individuals in
our 1997 and 2000 data sets.

21

 

3. Model

This essay adopts the human capital investment model of Becker (1964). According
to this model, an individual, in making a human capital investment decision, is assumed
to maximize the discounted present value of future earnings by taking into account the
opportunity cost of time and good spent in obtaining the capital and the rate of interest.
The empirical strategy in estimating returns to schooling investment in this essay consists
of two stages. The ﬁrst is estimating sector choice model in which an individual chooses
his/her path of employment. The rate of return is estimated in the second stage by the
inclusion of appropriate correction terms from the ﬁrst stage. This strategy follows the
two-stage methodology of Lee (1983), which basically is an extension to Heckman’s

(1979) selectivity-correction model to the multinomial logit setting.20

3.1. Sector Choice Model

Selection into sector of employment differs among individuals. Individuals who
choose to be self-employed might have different characteristics from public or private
sector workers. I will explicitly model the individual’s utility maximization process in
making decision to participate the labor market and/or to choose the sector of
employment by the method of Lee (1983). This method is based on estimating a
multinomial logit sector choice model. Multinomial logit is the most appropriate

econometric approach in estimating the probability that an individual will be found in

 

2° In the returns to schooling literature, this approach has been applied in, for example, Strauss and Thomas
(1996) for Brazil, Glick and Sahn (1997) for Guinea, Brown and Sessions (1998) for United Kingdom, and
Murrugarra and Valdivia (1999) for Peru.

22

 

each sector of employment.21 The Multinomial logit model uses a linear function of
individual characteristics (own schooling attainment and age), parental characteristics
(father’s and mother’s schooling attainment), household characteristics (household
composition and assets) and some variables related to residency of the individual. In
addition, to control for the effects of seasonality and price levels between months of
interview, a series of month variables is also included in the model.

There are ﬁve employment paths under consideration-non-participation (the base
category), self-employment, wage employment in the public sector, wage employment in
the private sector and unpaid family worker. IFLS collects six categories of employment
status, namely self-employment without the help of others, self-employment with the
help of household members or temporary/seasonal workers, self-employment with the
help of regular workers, government worker/employee, private worker/employee, and
family workers. The ﬁrst three categories are aggregated into self-employment sector.
Family workers are those who work by helping other household member in order to earn
an income or reap a proﬁt without getting any wage/salary either in money or in kind.22 23
To see whether these divisions are warranted, I will apply statistical testing procedure to
asses the equality of slope parameter vectors associated with each pair of employment
path in this general speciﬁcation (Wald test).

The multinomial logit model assumes that an individual makes his/her decision to

work in a sector of employment on the basis of a latent variable. The underlying latent

 

2' See Maddala ( 1983) and Greene (1992) for reference on multinomial logit.
22 An example for family workers is a child who waits on buyers in his/her parents’ shop.

2° Individuals who works as family workers will not be included in the earnings equation because they do
not report any earnings. The inclusion of family worker in the labor market participation regression as a
separate choice is to learn whether an individual who works as a family worker has different characteristics
from those who are self-employed or working for wage.

23

variable is derived from utility maximization conditional on the choice of each sector of
employment. The indirect utility of individual i to be found in labor market sector j is
speciﬁed as:

Vi]. =,B}X,+aij i=1,...,N j =1,...,M (3-1)
where X .- is a vector of exogenous individual characteristics such as education, age,
household composition, and other factors; and 6}}- is the error component capturing

unobserved variation in preferences not reﬂected in the observed variables. The
individual, facing M employment sector choices, will select alternative j if it gives the
maximum indirect utility:

Vij = max(Vn r", Vij""ViM) (3'2)

The probability that individual i chooses labor market sector j is:

P.. =Pr(V‘.j >V,.,,) for all j¢k

U

= Pr(ﬂj‘Xr + at; > .BIXi + 53k) (3'3)
= Pdﬂin _ﬂliXi > gilt "'51) -
If the 6,]. ’s are independently and identically Gumbel, their differences (5,, — 80.)

have a logistic distribution and the probabilities take the multinomial logit form.24 For

the current application, the choice model for individual i can be deﬁned as:

(4 If ﬂin'I'gmzmax(ﬂin+£r3’ﬂiXi+3i2’ﬂin'I'3imBer'I'gro)
If ﬂiXi'I'giszmaxwixi+5r4aﬂiXi+£rzrﬂin+£n’ﬂer'I'8io)
If ﬂ2Xi+gi2zmax(ﬂ4Xi+8i4’ﬂ3Xi+8i3’pri+8il’ﬂ0Xi+£r'0) (3'4)
if .BiXi‘I’Enzmax(ﬂin+£i4HBin+£r3’ﬂin'I'5i2nBoXi'I'gto)
If .6er+51'0zmax(ﬂ;Xi+5r4n6in+5r3’ﬂiXi'I'512’ﬂin'I'5n)

C4
II
O—‘Nw

 

2’ See Lee (1983) for further details.

24

 

Some method of normalization is needed since, from (3-3), only the differences in
utilities derived from different sectors of employment matter for the determination of the
probabilities. Following standard practice, I will set the parameter vector associated with

non-participation equal to zero. Parameters to be estimated by maximum likelihood are
ﬂ; , ,8; , ,6; , and ,6; , and they are interpreted as the effects of a variable on the utility of

being in sector of employment j relative to the utility from non—participation in the labor

market, the base category.

A potential shortcoming of the multinomial logit model is its dependence on the
independence of irrelevant alternatives (IIA). In principle, the IIA property assumes that
the relative probability of two existing outcomes is not affected by the addition of a third
outcome. For example, say that there are only two different sectors of employment,
nonparticipation and self-employment, and that an individual is equally likely to chose
between the two. Suppose that the individual is given the choice between self-
employrnent and working for wage and assume that he or she views the two as perfect
substitutes. In this case we would expect the probabilities of not working, being self-
employed and working for wage to be 1/2, 1/4 and 1/4 respectively. Multinomial logit,
however, does not treat the probabilities in this way. According to IIA property, it treats
the probabilities as 1/3, 1/3, 1/3 in order to keep the relative probabilities of
nonparticipation and self-employment constant.

Multinomial logit, due to its IIA property, may not be appropriate when there are two
or more alternatives that are close substitutes.25 Nevertheless, multinomial logit has

gained favor in estimating discrete choice model due to its simplicity. In many

25

applications that recognize the fact that an individual making choices, for example in
making decision after graduate ﬁom elementary school (whether to attend general junior
high school or to attend vocational junior high school or to go directly into labor market),

multinomial logit could serve as a useful framework to model the setting.

3.2. Wage Function
Bias for sample selection may arise since it is unlikely that an individual is

randomly assigned to a particular sector of employment. This problem occurs when
unobserved characteristics of an individual affect both the choice selection and earnings
models. If these unobserved characteristics are correlated with the regressors of wage
function, the estimated coefﬁcient of this ﬁmction will be biased. Following the two-
stage method of Lee (1983), self-selectivity problem could be corrected by: (1)
estimating the individual’s self-selected sector of employment choice and (2) using those
results to calculate selectivity-correction term and included that term as a regressor in the
second-stage functions (the wage function) to correct for the potential self-selection bias.

The idea behind this procedure is that the selectivity-correction term for each

individual (2‘).) reﬂects the predicted probability that he or she selects a particular sector

of employment. The inclusion of this term is to represent the importantance of
unobservable characteristics that an individual has on his/her employment path. The

appropriate correction term is deﬁned as (dropping individual subscript for convenience):

A“ = ¢(<D"[P.-,-l)
’ P

0.

(3-5)

 

25 Hausman and McFadden (1984) propose a speciﬁcation test the validity of IIA property (to be explained
in more detail in Section 5).

26

where ¢(.) and (D(.) are the probability density function and cumulative distribution

ﬁrnction of the standard normal; and P,j is the predicted probability that individual i
chooses sector j from the multinomial logit model.

,1. , we will estimate the

Using the estimated selectivity term deﬁned above, 2:
following Mincerian (1974) earnings ﬁrnction:
1n w,j = 7,2,. + gig. + v,j (3-6)
where 1n wy. is the natural logarithm of per hour wages or earnings of individual i in
sector j ; Z ,. is a vector of individual’s human capital characteristics, household-speciﬁc

as well as community and facility information; and vi]. is a zero-mean disturbance terms

capturing the unobserved variation in the wage ﬁrnction.

Another issue to be addressed in estimating wage function that contains selectivity

correction term is identiﬁcation. By the standard identiﬁcation argument, X i should

contain one or more variables that affect employment sector choice but not directly affect
earnings. Schultz (1993) argues that there exists some statistical support to use certain
variables such as non-earned income (e. g. income from assets, public transfer) or assets
as identifying instruments. As IF LS collects information on assets at the household level
(for households whose member engaged in farm or non-farm business) and at the
individual level the value of assets is available for identiﬁcation purpose. The inclusion
of assets variable is an advantage over only using household composition variables as the

studies reviewed in Section 2. In addition, I also use a set of household composition
variables to serve as identifying variables for calculating 21.1.. These variables will be

discussed in detail in the next section.

27

 

4. Data

Analysis of returns to schooling investments in Indonesia utilizes data ﬁom the
Indonesia Family Life Survey (IF LS). In particular, this study uses three waves of IF LS
surveys: IFLSl (ﬁelded in 1993), IFLS2 (1997) and IFLS3 (2000). Individuals who
answered Book IH, the book on adult information, are the object of this study.26 Book III
collects detailed information on a wide-range of issues at individual level for all selected
adults in the household. The primary objective of this essay is to estimate returns to
different levels of schooling for men and women in their working age of 25-59. The
purpose of excluding individuals younger than 25 years is to avoid those whose main
activities are going to school (or being full-time students). Excluding individuals aged 60
or older is to eliminate those who are retired. Thus our sample consists of productive age
individuals so that the issue of having full-time students and retirees in the sample is
avoided.

The number of observations in the ﬁnal sample is presented on Table 4.1. Around
30-40 percent observations are dropped due to age restriction of the study. In the 1993
sample, 4 percent of observations are excluded for incomplete or missing own schooling
information. Most of the individuals in the latter category have their Book III answered

by proxy. Another 4 percent observations are dropped due to incomplete wage

 

2‘ Some information (for example: age, marital status, highest level of school attended and highest grade
completed) is recorded on both householder roster (Book 1, section AR) and Book III. Householder roster
is collected at the household level and administered to the spouse of the head of the household or to the
head of the household. Missing information on Book III could be ﬁlled-up with the corresponding ones
ﬁ'om householder roster. This is not the case with the sample used for this study because Book III is
considered to contain more accurate and reliable information since it is given directly to the individual
him/herself and can thus eliminate memory errors of the householder roster’s respondent. In addition, it is
also not possible to retrieve information from householder roster to be included in the ﬁnal sample since the

28

information.27 In the 1997 and 2000 samples, around 4-5 percent of observations are
dropped due to incompleteness.

This section starts with a brief discussion on IF LS. Description on labor force
participation and its stratiﬁcation by human capital variables (own schooling and age) for
each of the survey will follow.28 29 In addition, similar description on participation for
individuals interviewed in all three surveys (panel respondents) as well as their work
transition between surveys will also be presented. Other data, such as parental schooling,
household composition, business assets, region of birth and of current residence, and
earnings are to be described prior to the presentation of regression results in each

respected sections.

4.1. IFLS

IF LS is a longitudinal survey of individuals, households, and communities in
Indonesia. RAND, Santa Monica, CA administers the ﬁrst and second waves of these
surveys in collaboration with Lembaga Demograﬁ Universitas Indonesia (LD-UI),
J akarta.30 The third wave, IF LS3, is undertaken in cooperation with Pusat Studi

Kependudukan dan Kebij akan Universitas Gajah Mada (PSKK-UGM).

 

roster does not collect essential data for empirical purpose such as work status (self-employment, public
and private sectors, and unpaid family workers) and number of hours worked.

27 Notice that information on wage or income is only collected for individuals who are self-employed or
work at public and private sectors. No observations are excluded based on the availability of wage
information for individuals who are not working or those working as unpaid family worker.

28 Enrployment status includes those who currently are unemployed, searching for work, or temporary out
of work.

29 Throughout this essay, standard errors are adjusted for clustering at the enumeration areas due to IF LS’
survey design (to be discussed in the next subsection).

3° In the second half of 1998, a special survey of IFLS was ﬁelded to study the immediate impact of multi-
dimension crisis that hit Indonesia starting by the end 1997. This survey, IFLSZ+, only covers 25 percent
of the regular IF LS sample.

29

At the individual level, IF LS collects information on socioeconomic and demographic
characteristics of respondents such as detailed information on the history of education,
employment and earnings, migration, and marriage, pregnancy and child bearing. Wide-
ranging health information is also collected in IFLS. This, among others, includes
information on health status of respondents, utilization of health care facilities, tobacco
consumption, health insurance, anthropometrics measures (such as weight, height,
hemoglobin level, blood pressure, and lung capacity), and family planning. In addition,
information on links with non co-resident kin, transfer and borrowing, and individual
assets is also collected. At the household level, IF LS collects extensive information on
consumption, assets (farm business, non-farm business, and household), non-labor
income and household characteristics.

IF LSl is based on a stratiﬁed sampling scheme on provinces and rural-urban areas
within provinces. Inside these strata, enumeration areas (EA) were randomly sampled,
followed by randomly sampled households within EAs. To maximize the representation
of population, culture and socioeconomic diversity of Indonesia, 13 provinces from
Sumatera, Java and other region were selected. These provinces are North Sumatera,
West Sumatera, South Sumatera, Lampung, DKI Jakarta, West Java, Central Java, DI
Yogyakarta, East Java, Bali, West Nusa Tenggara, South Kalimantan and South
Sulawesi. Following this sampling procedure, over 30,000 individuals in 7,224
households from 321 EAs were sampled. Over-sampling occurred in urban EAs and EAs
in smaller provinces to facilitate rural-urban as well as J ava-non-J ava comparisons.

For budgetary reasons, not all household members were interviewed in IFLS 1.

Interviews were conducted with the following household members. First, the household

30

head and his/her spouse. Second, two randomly selected children of the head and spouse
aged 0-14 (interviewed by proxy). Third, an individual aged 50 and above and his/her
spouse, randomly chosen from remaining members. Fourth, for a randomly selected 25
percent of the households, an individual aged 15-49 and his/her spouse, randomly chosen
from remaining members. The sampling rules indicate that as many as six adults in each
household could have been interviewed, but the maximum number of adults actually
interviewed was restricted to four (see Frankenberg and Karoly [1995] for details).

In IFLSZ, ﬁelded between June and November 1997, all IF LSl households are aimed
to be re-surveyed (see Frankenberg and Thomas [2000] for documentation of IF L82).
Around 10—15 percent of households that moved from their original location are
followed. In addition, IFLSZ also added approximately 900 households by tracking
individuals who split-off from their original household. The strategy taken was to track
an entire household or targeted individuals that moved as long as they resided in any of
the IF LS provinces regardless of whether they move within or across province borders.
The targeted individuals from split off households are those who were main respondents
in 1993 (panel respondents —those who answered one or more IFLSl ’3 individual
questionnaires) or those who were born before 1968 (those who were at least 26 years in
1993). To control cost, not all individuals were tracked.

With respect to individuals within household, IFLSZ goal is to interview all
individuals from original households. Among the split-off households, all tracked
individuals were interviewed along with their spouses and any of their biological children

living in the household.

31

The total number of households contacted is 7,629, consisting of 6,752 panel
households and 877 split-off households, while the number of individuals interviewed is
approximately 34,000.

In IF LS3, all original IFLSl households and split-off households of IFLSZ and
IF LSZ+ (see footnote 31) became the targeted households to be found in their original
locations. In addition, households that moved but still resided within IF LS provinces and
Riau were tracked.3 ' Individuals who moved out of their IFLS household were followed
as well. Interviews were conducted with all current members of original (IF LS 1)
households.

Rules for following individuals who moved out of an IFLS household were expanded
in IF LS3 to include: (1) IFLSl main respondents; (2) IF LSl household members who
were born before 1968; (3) individuals born since 1993 in original households; (4)
individuals born after 1988 provided that they were resident in an original households;
(5) an 20% random sample of IFLSl household members who were born between 1968
and 1988.32

Within household, all current members of original 1993 household were interviewed.
For split-off households (whether a split-off from 1997, 1998 or new in 2000), in addition
to the selection rules of IFLS2, any individuals who had lived in a 1993 household plus

their spouses and biological children are also the targets of interview.

 

3' A small number of households in Southeast Sulawesi, Central and East Kalimantan were also followed.
For analysis purpose, these households are reclassiﬁed to the nearby IFLS provinces.

32 See Strauss et.a1. (2002) for further information on comparison on selection of households and
respondents within households among IFLS waves.

32

More than 10,500 households were contacted, 6,768 of which are original households
and 2,648 are new split-off households. Over 43,600 individuals were contacted in
IFLS3. Main ﬁeldwork took place from June through November 2000.

Complementing the individual and household surveys is a comprehensive community
and facility survey. This survey involves interviews with local community leaders
regarding services, physical and economic infrastructure, and history of schools and
health facilities existence. Health service providers and schools are surveyed within each
community. Health facilities surveyed in IFLS come from a probabilistic sample of
facilities that actually serve households in the community. The same goes for school
sampling procedure, where the list of schools comes from the actual schools being used
by households with children who are currently enrolled. In addition, prices of goods and
services at the community level are also collected. This rich information on community
and facility may be matched with individual and household level data for individuals and

households residing in these 321 communities.

4.2. Labor Force Participation for Cross Section Respondents

Non-employment and employment in four occupations are of interest in this study:
non-employment, self—employment, two sectors of wage work (public and private) and
unpaid family work. Individuals, at the time of survey, select themselves into one of
these ﬁve states. The distribution of labor force participation is shown on Table 4.2 (see
All Individuals). The employment rate in our sample ﬂuctuates from 73 percent in 1993

to 71 percent in 1997 to 79 percent in 2000. Corresponding ﬁgures from SUSENAS

33

samples are, respectively, 76, 73 and 73 percent (see Appendix Table 4.1).33 By gender,
the table also shows that 96 percent of men and 55 percent of women engaged in some
type of employment in 1993. In 1997, employment rates slightly decreased to 93 percent
for men and 53 percent for women. In 2000, possibly in response to the economic crisis,
the proportion of women working increased signiﬁcantly to 64 percent while that of men
increased slightly to 94 percent.“ 35
Looking at the dynamics of employment rates more closely at the three sectors for
which rates of retums to schooling are going to be estimated (i.e., self-employment,
public sector and private sector), one can see that the proportion of self—employed men
decreases between 1993-1997 but stays at the same level between 1997-2000; while
proportion of self-employed women stays the same between 1993-1997 and increases
between 1997-2000. Among public sector workers, the fraction of men participating
declined with time, while that of women stayed at below 5 percent. Finally, among
private sector workers, the proportion of both men and women increased between 1993-

2000. Although one can make inferences about transition of work between sectors, 1 will

not discuss this issue until the presentation of results from panel respondents.

 

’3 Estimates on Appendix Table 4.1 are based on comparable characteristics such as individuals selected are
of 25—59 years of age, residing in 13 IFLS provinces and are categorized into similar sector of employment.
For SUSENAS 2000, public and private sector workers are merged into one sector, paid workers, because
the question does not allow this distinction.

3’ Strauss eta]. (2002) found that employment rate for women between 1997 and 2000 increased at the
same magnitude. Nevertheless, SUSENAS data (Appendix Table 4.1) shows that non-employment of
women during the same period fell by 0.5 percent.

’5 It is also interesting to note that most of employed men are working for pay while among around a
quarter is working unpaid at the family business. This proportion, however, change between years.
Whereas in 1993, less than 1 percent of employed men are working unpaid, in 1997 and 2000 this
proportion increased to 2.8 and 3.4 percent, respectively. Among employed women, the corresponding
ﬁgures are, respectively, 28, 20 and 24 percent.

34

Table 4.2 also shows the distribution of labor force participation by schooling
attainment.36 37 In 1993, around 90 percent men with 0-6 years of schooling and 80
percent of men with 7-9 years of schooling are either self-employed or working at the
private sector. This pattern changes with schooling attainment where more than half of
men with beyond secondary level schooling experience work at the public sector.
Among women, around 65 percent with 0-3 years of schooling and 75 percent with 4-9
years of schooling are either self-employed or do not engaged in any work activities.
Even around 55 percent of women with 10-12 years of schooling are still either self-
employed or not working, 26 percent of them work at the public sector. Parallel to the
pattern of men, around 45 percent of women with 13+ years of schooling work at the
public sector, although another 27 percent of them do not work.

The patterns for the 1997 and 2000 sample are comparable to that of 1993 sample. A
notable difference worth mentioning here is that there is a shift in the choice of
employment sector from public sector to private sector among individuals with higher
schooling attainment (10+ years). Whereas in 1993 around 52 percent of men and 46

percent of women with 13+ years of completed schooling working at public sector, in

 

36 Own schooling information is recorded on the highest education level attended and the highest grade
completed at that level. Various schooling measures, such as completed years of schooling, diploma
earned, and schooling dummies, could be constructed from these two variables. For each grade completed
at any level of schooling, years of schooling variable is increased by one. An individual who never
attended school is assigned zero year of schooling, while one who attended the ﬁrst grade of elementary
school but never completed that grade is also assigned zero year of schooling. Completed years of
schooling is constructed as 6 years for individuals whose highest schooling is completed primary, 9 years
for junior secondary, 12 years for senior secondary, 13 years for Diploma-1 (D1) program, 14 years for
Diploma-2 (D2) program, 15 years for Diploma-3 (D3) program and 17 years for any university degree.

37 In IFLSl and IFLSZ, there is no additional information on whether an individual completed Bachelor,
Master, or Doctorate degrees. The maximum years of schooling is thus set to be 17. In IFLS3, separate
categories for Bachelor, Master, or Doctorate degrees are recorded. However, individuals with Master or
Doctorate degrees only make 0.2 percent of total sample. For comparability among survey waves,
maximum completed years of schooling is top coded at 17. All of the speciﬁcations in empirical section do
not distinguish between the levels of schooling achieved beyond secondary due to the small number of data

35

 

2000 these proportion are cut to 36 percent for men and 39 percent for women. The
proportion of private sector men increases from 27 percent in 1993 to 34 percent in 2000,
while the corresponding proportions for women are 19 percent in 1993 and 28 percent in
2000. SUSENAS estimates, however, show that the proportion of workers with tertiary
schooling working at the public and private sector stay at about the same level between
1993 and 2000.

Labor market experience in this study is approximated by age.38 Table 4.3 presents
fractions of labor force participation within each of the four age groups used in this essay
(25-29, 30-39, 40-49 and 50-59 years). One can see from this table, for example, the
fraction of the youngest age group men not engaging in any kind of work ﬂuctuates from
3 percent in 1993 to 13 percent in 1997 and to 8 percent in 2000. One interesting fact
from this table is that non-employment for the oldest individuals shrunk drastically
between 1997 and 2000, from 14 to 10 percent for men and from 49 to 37 percent for
women. This, again, might be interpreted as the contribution of the old individuals in
coping with economic crisis. Furthermore, the fraction of non-employed women aged

40-49 years also decreased from 42 percent in 1997 to 28 percent in 2000.39

 

in this category (only less than 5 percent of individuals in 1993, 6 percent in 1997 and 9 percent in 2000
have post senior secondary level education).

38 In addition to age, IF LS provides information about the year/age an individual started working full-time,
deﬁned as the year/age when working is the primary activity. Unfortunately, there exists inconsistency
between age and the number of year of experience constructed from the information provided, where age is
smaller than experience. Eliminating these inconsistencies reduces the number of observations
signiﬁcantly. In addition, including actual experience in the wage function could yield biased estimates, as
this variable is not exogenous. In section 9, comparison between using age and potential experience as
proxies of labor market experience will be illustrated.

3’ Panel respondents of IFLS2 and IFLS3 shows that 8.1 percent aged 55-64 in 1997 shifted from not
engaging in any kind of work in 1997 to working in 2000. For women, the corresponding number is 15.6
percent (Strauss et.a1. [2002]).

36

 

4.3. Labor Force Participation for Panel Respondents

Panel respondents are deﬁned to be individuals who are in all three surveys provided
that they are of age 25-52 in 1993.40 As previously mentioned, the criteria for selecting
households and individuals changed between surveys. The inclusion of new household
members such as from the split-off households, for example, might be responsible for the
difference in schooling attainment within different waves of the survey. It is for this
reason, performing empirical exercise based on cross section as well as panel respondents
might be fruitful since it will give perspective on the rates of returns to schooling
investment.

Table 4.4 is the equivalent of Table 4.2 for panel respondents. The fraction of overall
non-employed panel men increased between 1993-1997, similar to that of cross section
sample, but stayed at the same level between 1997-2000. Among women, however, the
fraction of non-employed panel respondents fell between 1997-2000, comparable to that
of cross section sample. One can easily see the similarity as well as difference between
employment rates within other sectors. One important difference was found among
public sector workers, where the fraction of individuals with tertiary level of schooling
increased within panel respondents but decreased within cross section individuals. The
fact that there were no major job openings for public servant work during the mid-19903
might partly explain the decrease within cross section sample. Individuals with higher
level of education might decide to stay working at the public sector while those with up

to secondary schooling move to other sectors of employment during the crisis.

 

’0 This means that individuals are 29-56 years old in 1997 and 32-59 in 2000. The number of observations
in the panel sample is 6,519, meaning that there is a 32 percent reduction of total 1993 sample, mostly due
to age restrictions (see bottom portion of Table 4.1). The portion of cross section sample excluded from the

37

It is also interesting to learn the employment behavior of women by age group (see
Table 4.5). One can see a drop in non-employment for the younger women (aged 25-39),
a slight decrease between 1993 and 1997 and a large drop between 1997 and 2000. This
is expected since women of this age, especially those aged 25-29 years, might be having
children and thus will abstain from working activities. But, the fact that this trend could
also be found for older women (aged 40-52), especially a 9 percent decrease in non-
employment between 1997 and 2000, might suggest a crisis effect since these women
might have already stopped childbearing. For older men (aged 40-52), however, the
fraction of non-employment increased as they ages, with a large increase between 1993
and 1997. In contrast, non-employment of men aged 50-59 in the cross section sample
decreased between 1997 and 2000.

Table 4.6 shows work transitions by sector between 1993-1997 and 1997-200 for
panel respondents. Transition measures presented in this table are based on the
employment status of the individual at the time of surveys only; changes in status
between the two points are not taken into consideration. One can observe not only a
large transition for men from private sector to self-employment but also from self-
employment to private sector. During the crisis one can expect the former to take place
but not the latter. Among women, we saw a large transition from non-employment to
self-employment and other working activities, where the percentage changes are during
the 1997-2000. One can also observe transition from private sector to self-employment
and non-employment and vice versa, from non-employment and self-employment to

private sector.

 

panel increases to 47 and 56 percent in 1997 and 2000, respectively, as a result of age restrictions as well as
different criteria used in the selection of individuals interviewed.

38

Another way to look at transition is to look at whether individuals work in both years,
do not work in either year, get job or lose job between years. An individual is
categorized as to 'Get J ob' ('Lose J ob') if he/she works (does not work) in the current
survey while not working (working) during previous interview. Thus, these measures do
not take into account working activities transitions between surveys. Table 4.7 presents
these transitions by age group and gender. For the oldest men, in accordance to previous
results, the fraction of those working in both years decreased (between 1993-1997 and
1997-2000). However, this decrease goes along with an increase in those who get jobs
and a decrease in those who lose jobs. A different pattern was found for women, where
the fraction of those working in both years increased along with a small decrease in those
not working in both years and those who lose jobs as well as an increase in those who get
jobs. This supports the claim that old individuals (especially women) play an important

role coping with economic crisis of 1997-98.

Some of the panel effect might be age-effect not year-effect, but it might not be true
either. Without further analysis, one carmot tell which of age-, cohort- or year-effects
plays an important role in the data (see Deaton [1997] for decomposition analysis).

Patterns of IF LS results (even for panel individuals) do not emerge in SUSENAS (see
Appendix Table 4.1). For example, one cannot see a large drop in non-employment for
women and increase in self-employment for men between 1997 and 2000. In SUSENAS

these changes are only 0.5 percent or less.

39

5. Sector Choice Model

The employment rate in our sample ﬂuctuates from 73 percent in 1993 to 71 percent
in 1997 to 79 percent in 2000."1 The employment rate for men is notably higher than that
of women (93-96 percent for men vs. 53-64 percent for women)"2 Beyond the decision
to participate or not to participate in the employment market, there is also an issue of
what type of employment does an individual choose to participate in. Most of the recent
studies on returns to schooling investments take into account the fact that labor market
participation is not likely to be random. Individuals choosing to participate in the labor
market might have different characteristics, in terms of observable and unobservable
factors that affect wages, from those who decide not to work. Individual characteristics,
for those choosing to work, might also vary among the choices of employment sectors in
a non-random fashion. It is, therefore, necessary to control for the potential selectivity
bias problem in the sample.

In this study I account for this problem by modeling sector choice by the procedure
proposed by Lee (1983) based on estimation of multinomial logit sector choice model.
For that purpose, I assume that individuals in our sample make decisions on whether not
to work, to be self-employed, to work at the public sector, to work at the private sector or

to work at the family business as an unpaid family worker. Our sample shows that men

 

" Recall that our each of our samples consists of adults aged 25-59 years living within 13 IFLS provinces.
Corresponding ﬁgures from SUSENAS samples are, respectively, 76, 73 and 73 percent.

’2 Table 4.2 shows that 96 percent of men and 55 percent of women engaged in some type of employment
in 1993. In 1997, employment rates slightly decreased to 93 percent for men and 53 percent for women. In
2000, probably in response to cope with economic crisis, work participation of women increased
signiﬁcantly to 64 percent while that of men increased slightly to 94 percent. Strauss et.al. (2002) found
that employment rate for women between 1997 and 2000 increased at the same magnitude. Nevertheless,
SUSENAS data (Appendix Table 4.2) shows that nonparticipation of women during the same period fell by
only 0.5 percent.

40

who do not participate in labor force or who are working at family business are
underrepresented. In the sector choice model, these two sectors are combined and served
as the base category in the multinomial logit regressions. Women working at the public
sector are also underrepresented in our sample. Since the interest of this study is to
estimate rates of returns to schooling for men and women who are self-employed or
working at the public and private sectors, pooling women respondents who are working
for wage (i.e., those in the public and private sectors) will not serve the purpose.
Therefore, estimation results for women in the public sector should be interpreted with
caution due to small sample size.

This section will proceed with discussion of speciﬁcation tests for multinomial logit
results, followed by discussion of factors affecting men and women participating in a

particular sector of employment for both cross-section and panel respondents.

5.1. Speciﬁcation Tests

There are many tests that are commonly used to verify multinomial logit
speciﬁcations (see Freese and Long [2000] for further details). Two of them are applied
here: ﬁrst, the Hausman-McFadden (1984) tests to assess the independence of irrelevant
alternatives, and second, Wald tests to determine whether the independent variables
differentiate between any two outcomes. This test is commonly used in determining

whether any two outcomes can be combined.

41

5.1.1. Hausman-McFadden Tests

Hausman and McFadden (1984) proposed a testing procedure to evaluate the
independence of irrelevant alternative (IIA) assumptions of multinorrrial logit regression
speciﬁcations. The test is based on dropping a category from the speciﬁcation and
examining changes in the estimated coefﬁcients. This test provides an indicator on the
appropriateness of multinomial logit speciﬁcation at hand.

The procedure for testing IIA assumptions, in principle, can be explained as follows.

First, estimate the unrestricted model with all J outcomes included; these estimates are
ﬂu . Sample of men, in our multinomial setting, consists of four sectors of employment;
this is the total number of possible outcomes. Second, estimate a restricted version by

eliminating one or more outcome categories; these estimates are [3,, . For example, drop

the sub-sample of men working at the public sector, which leaves us with three sectors of

employment. Third, construct Hausman test statistics:

H”. =0}. —8;)'tV(ii.)—V(ﬁ; >1" ([3, #3:.) (5-1)
where ﬂu is a subset of 3,, after eliminating coefﬁcients not estimated in the restricted
version. H ,, A has an asymptotically chi-square distribution with degrees of freedom

equal to the number of rows in ,3, if IIA is true.

Since the null hypothesis of this test is that the odds of omitting one or more outcome
is independent of other altematives in the sense that it will not change the parameter of

estimates systematically, a signiﬁcant value of H 11.4 indicates that the IIA assumption has
been violated. H m might have a negative value if WA) - 1705,!) is not positive semi-

deﬁnite. However, a negative H ,, A indicates that IIA property holds.

42

Table 5.1 shows the chi-square ( 12 ) value of Hausman-McFadden test statistics

along with its respected p-value. Three speciﬁcations of interest for each gender are
tested. The ﬁrst controls for own schooling that enters in a non-linear fashion in terms of
dummy variables. The second controls for non-linear own and parental schooling
attainment. This speciﬁcation is hereafter referred to as the base specification as
correction factors for wage regression on Section 6 are based upon this speciﬁcation. In
the third speciﬁcation (fall speciﬁcation, hereafter), residency variables are also
controlled for.43 For each speciﬁcation, test statistics are presented for dropping one

particular sector of employment at a time. The results show that most of the H ”A are not

signiﬁcant, implying support for IIA property of the multinomial logit models.44 Almost
half of the test statistics are negative, but these negative values support the existence of
IIA property. These ﬁndings indicate that the use of multinomial logit is appropriate for

modeling sector choice model for both men and women.

5.1.2. Wald Tests
Standard Wald tests are used to determine whether two outcomes can be combined.
The null hypothesis to test, for example, that the coefﬁcients of ‘self employment’ and

‘public sector’ are indistinguishable is:

 

’3 These variables are represented by dummy variables representing province of current residence and their
interactions with whether the individuals reside in urban area.

4’ There are three cases of violation to the IIA property at the base speciﬁcation. Dropping self-employed
men in the 1993 sample, self-employed women in 1997 and private sector men in 2000 will turn H 11.4 to

be signiﬁcant. Controlling for variables in stages, as mentioned in footnote 3, will not give any signiﬁcant
test statistics (results for the ﬁrst three stages are not shown). Thus, these violations are not considered to
be a major issue that can alter the general support for IIA property.

43

H0 : (ﬂl,self|base - ﬂlmubliclbase) = ‘” : (ﬂK,selflbase - ﬂK,publiclbase) : 0 (5-2)

where base is the base category used in the calculation.

For all three speciﬁcations (for both men and women), the Wald test statistics for all
samples, presented on Table 5.2, are highly signiﬁcant. This means that any two sector
of employment in each of these speciﬁcations cannot be combined. These results imply
that the determinants to enter into a particular segment of the labor market are unique so
that, for example, characteristics of individuals choosing to be self-employed are
different from those working at the private sector. In addition, these results conﬁrm that

the divisions of sectors of employment in the sector choice model are warranted.

5.2. Factors Affecting Sector Choice

Estimates for base speciﬁcation of (multinomial logit) sector choice models for men
and women are presented in Table 5.3A, 5.3B and 5.3C.“ A summary of estimates for
own schooling variables is shown on the upper part of Table 5.5. In addition to Table 5.5,
for comparison purposes, a summary of own schooling variables from speciﬁcations in
which own schooling enters as completed years of schooling (or in linear fashion) and
speciﬁcation in which own-schooling are non-linear, both with no control for parental
schooling, are presented on the upper part of Table 5.4. 46

Due to the panel nature of the data, cross section comparison of results among the

three waves of survey should be interpreted with caution. Households as well as

 

’5 Estimation results for the 1993 sample are shown on Table 5.3A, for the 1997 sample on Table 5.3B, and
for the 2000 sample on Table 5.3C. Hereafter, these tables are referred to as Table 5 .3. Likewise are all
the tables that follow.

’6 To obtain pattern from resulted estimates, discussion in this section are based on estimates that are at
least signiﬁcant at the 10 percent level.

individuals are randomly selected within the 1993 wave. In the 1997 and 2000 samples,
however, individuals might not have been random due to selection rules for both
households and individuals within households due (see Section 4). This also applies

when we discuss results based on panel respondents.

5.2.1. Own Schooling

Own schooling plays an important role in allocating men and women to their
employment path. These effects, however, work in different directions depending on the
employment sector of interest. Let’s look ﬁrst at estimates in which schooling attainment
is linear (see Appendix Table 5.1 and its summary, Table 5.4). It is clear that education
appears to be negatively correlated with self-employment for men and women (except for
women in 1997) and with working unpaid at family business for women, all are relative
to non-employment. 47 The negative effect of schooling is stronger for 1993 relative to
2000 for men but is almost at the same level for women. On the contrary, schooling
attainment is positively correlated with employment at the public sector for both men and
women with stronger effect for 2000 than 1993. However, education does not affect men
and women in making decision to work at the private sector relative to non-employment.

Turning to speciﬁcation with non-linear own schooling attainment with no control for
parental schooling (see Appendix Table 5.2 and its summary, Table 5.4) one can see that
level of schooling are important in explaining the effect of schooling in sector choice

model. For self-employed men in 1993, the negative effect of schooling is only true for

 

’7 Recall that the base category for men is the pool of men who are not working and men who are working
at family business, while the base category for women is non-employment. Interpretation of estimated
coefﬁcients of the sector choice model for men and women should always refer to their respected base
categories, even though it is not mentioned in the text.

45

 

those with schooling attainment beyond primary level. By 2000, however, schooling
attainment for men with primary schooling are positively correlated with the probability
of being self-employed (relative to non-employment), while schooling is negatively
correlated with self-employment only for men with tertiary level education.

Among public sector workers, one can also see that the effect of schooling attainment
gets stronger with level of education for both men and women. These effects tend to be
stronger with time, which might support the fact that public sector employment is an
education-intensive sector.

For men, schooling attainment up to secondary level is positively correlated with
private sector employment (but not in 1993). Among women, however, education tends
to have negative effect on employment in the private sector, where the effects are
stronger in 2000 than in 1997. These results reverse the conclusion of speciﬁcation with
linear schooling attainment, where education does not affect employment at the private
sector.

Now let’s look into own schooling estimates of base speciﬁcation in detail (see Table
5.3 and its summary, Table 5.5). In general, the magnitude of the estimates tends to be
smaller in comparison to previous speciﬁcation (the one that does not control for parental
schooling), indicating that parental education does pickup some family background
information of the individuals. The trends of own schooling effects, however, are similar
between the two speciﬁcations.

As was previously mentioned, the magnitude of estimates, which shows the effect of
own schooling on the probability of joining a particular sector, is not linear. For

example, men with sub-primary schooling (4-6 years) have a 162 percent higher chance

46

of working in the public sector versus non-employment relative to men with zero year of
schooling.48 This likelihood increases in a non-linear fashion with schooling attainment,
where odds of men with junior secondary schooling is 285 percent higher while those
with senior secondary schooling is 1,546 percent higher. A similar non-linear trend is
found for public sector women even though signiﬁcant effect of own schooling did not
exist for those with primary school education. The stronger higher odds of joining public
sector as schooling attainment increases between years of survey might indicate that
public sector has become increasingly education-intensive with time.

New insight could be provided when we relate own schooling estimates with the
distribution of primary duties of the individuals. Occupations (primary duties) within
sector of employment along with its average years of schooling are presented in
Appendix Table 5.7. Professional and managerial staffs have the highest average years
of schooling, followed by administration staffs. Those working as an agricultural worker
or as laborer have the lowest schooling attainment.

The negative association between schooling attainment and being self-employed for
men, especially among those who are more able, is an indication that self-employment
sector tend to be a low-technology sector so that the highly educated individuals might
feel counterproductive being self-employed. This is supported by the composition of
occupations within this sector where more than half of self-employed men in our 1993
sample are workers in the agriculture sector and another ﬁfth are sales staffs (see
Appendix Table 5.7). For women, only those who completed 1-3 years of schooling are

more likely to be self-employed. In contrast to self-employed men, half of the self-

 

" The estimated coefficient for public sector men with 4-6 years of schooling is 0.964 (see Table 5.3A).
This implies that the odds for men with sub-primary level of schooling to enter public sector employment is

47

employed women work on sales staffs while another ﬁfth are agriculture workers. Since
working as sales staff requires more schooling than being a farmer, it is possible that the
positive correlation only occurred for those with 1-3 years of schooling with no
association for those with higher schooling attainment. The change in the occupation
compositions among self-employed individuals between surveys, with tendency toward
more education—intensive type of occupation, might explain the positive effect of
schooling for the lower level of schooling attainment and the negative effect for the
higher level in the 1997 and 2000 samples.

Around 30-40 percent of men and 60-70 percent of women working at the public
sector are professional or managerial staffs, while another 25-30 percent of men and 15-
20 percent of women are administrative staffs, two intensive-education types of
occupation. Therefore, it is not surprising to see a positive association between schooling
and the odds of participation in public sector employment.

For women having 4-9 years of schooling is associated with a decrease in the
likelihood of working in the private sector. This tendency holds in all three surveys. The
private sector is dominated by occupations associated with lower level of schooling
attainment such as agricultural worker and laborer. For men, however, there is no clear
pattern between years in terms of schooling attainment and private sector employment.
In 1993, education does not affect men deciding to join the private sector. In 1997
certain levels of schooling (4-6 and 10-12 years) appear to increase private sector
participation. Positive correlation between schooling and private sector employment is

found in 2000 (except for those with 13+ years of schooling). Variables other than

 

em” = 2.62 times (or 162 percent higher than) the odds of their counterparts with no formal schooling.

48

occupational composition might explain the change between years, although one might
notice that the proportion of service staffs, an occupation with relatively high average
years of schooling, almost doubles between 1997 and 2000.

Finally, being an unpaid family worker (which is available for women only) is, as
expected, negatively correlated with education. Around 70-80 percent of women in this
sector work in agriculture. Thus, those with higher schooling attainment are likely to
avoid working at this sector that, in turn, explain the negative effect of education in
participation in (unpaid) family business sector. Schooling effects diminished with time
as evidenced by the non-signiﬁcant effect of schooling in the 2000 sample. Table 4.2
shows a huge increase in family work employment for women with less than tertiary
level of schooling (especially between 1997 and 2000).

Some recent studies in developing countries ﬁnd mixed conclusion on the importance
of education in determining labor participation. For the case of Indonesia, Duﬂo (2000)
ﬁnds that the probability of working for wage for men is affected by schooling.49 Glick
and Sahn (1997), using survey data from Conakry, the capital of Guinea, ﬁnd that more
schooling reduces the probability that men and women become self-employed. They also
ﬁnd that more schooling increases the probability of men to entering public sector
employment but reduces the likelihood of being a wage earner in the private sector. For
women, schooling raises entry probabilities for wage work in both public and private
sectors. Tansel (1999) ﬁnds that, in Turkey, educational attainment increases the
probability of joining public administration, state owned enterprises and the (covered)

private sectors but reduces the likelihood of entering the other employment category. In

 

’9 In her study, Duﬂo makes no distinction between men working in public or private sector.

49

contrast, King (1996) ﬁnds very small or negative effects of years of schooling on total

labor force participation for women in Peru.

5.2.2. Parental Schooling

Father’s education background in the 1993 sample has very limited impact on son’s
and daughter’s employment.50 For women, having a father who completed primary
school is associated with a 25 percent increase in the probability of being self-employed
but a 28 percent decrease in the likelihood of working in the private sector. Even though
mother’s schooling variables are jointly highly signiﬁcant, individual estimates are, in
general, not signiﬁcant so that limited inference could be made from these individual
estimates. In most cases, the direction of association between maternal schooling and the
odds of employment is positive for wage earners and negative for the self-employed.
Having a mother with some elementary schooling increases the likelihood of children’s
private sector employment and of son’s employment at public sector. The relationship
between parental schooling and women working as unpaid family workers is consistent
among surveys. Having parents with some level of schooling is associated with a
decrease in the employment at the family business. This association is expected since
educated parents prefer their children not to work without pay.

In 1997, parental schooling has opposing consequences on son’s and daughter’s
employment. Having a father with at least secondary education reduces employment of

men while having a mother that at least completed primary school increases employment

 

5° Discussion is based on estimates of base speciﬁcation (see Table 5.3). For men, Wald test for the joint
signiﬁcance of father’s schooling on son’s employment is not signiﬁcant for the 2000 sample and only
signiﬁcant at the 10 percent level in 1993.

50

in the private sector.51 For women, having a primary school educated father increases the
chance of working at the self-employment sector while having an educated mother
decreases the odds of being self-employed.

Father’s schooling still has a negative effect on son’s probability of being self-
employed in 2000. However, having a mother with some primary schooling education
decreases men engaging in any working activities. This is also true when the
speciﬁcation is extended to include residency variables.

The inconsistent as well as limited effect of parental schooling on male and female
employment decisions between surveys might be an implication of the limited variation
in parental schooling variables, especially that of mother’s. In all three surveys, around
half of the individuals in the sample have mothers with no formal schooling. The only
consistent effect of parental schooling is in reducing the probability of women working
unpaid in family business sector. In spite of the limited effect of parental schooling,
speciﬁcations that control for own and parental schooling to correct the wage regressions

is of central interest of this study.

5.2.3. Other Variables

Sector participation likelihood of men and women with age follows an inverted U-
shaped proﬁle. In general, younger individuals (aged 25-39) are more likely to be
engaged in any kind of employment activities, while those aged 40+ are less likely to do
so. This trend is true for all sector of employment. There is, however, a noteworthy

change between results from 1993 and 1997 on one hand and from 2000 on the other,

5' The negative effect of father’s schooling on son’s employment holds even after further controlling for
residency variables (full speciﬁcation, see Appendix Table 5.3).

51

especially for men. Estimated coefﬁcients for men aged 40-49 are either signiﬁcantly
negative or not signiﬁcant in the 1993 and 1997 samples. In 2000, however, being 40-49
years old is associated with a 4 percent increase in self-employment work for men and
women and a 10 percent increase in working at the public sector. These increases might
reﬂect the impact of economic crisis of 1997-98 in Indonesia where older people tend to
work as a way to cope with the crisis.

Household composition and business assets variables, intended to identify selection
term for the second stage OLS wage regressions, are jointly signiﬁcant at the 1 percent
level. The presence of adult men (aged 20—49) in the household lowers employment
participation for both men and women. This result, especially for self-employed
individuals might indicate substitution effect for working in this sector. For women,
however, this might indicate women’s role in household work (or an increase in home-
produced goods) and child bearing/rearing responsibilities, as our data show that a
signiﬁcant fraction of educated women do not participate in any work activities. Having
adult women in the household increases a woman probability to work at the private sector
(in 1997 and 2000) but decreases the change of being an unpaid family worker. This
might imply that the existence of adult female in the household might lighten the burden
of household chores and child-care activities for those working for wage but not for those
working without pay.

Having an elderly man (aged 50+) in the household lowers work participation for
men and women except that it increases women participating in family business work
activities. The latter might indicates that the care of elderly man could be combined with

some working activities at the farnily-owned business. The presence of elderly women

52

reduces male employment activities but increases women participation in wage work.
Thus, taking care of the elderly in the household might considerably cut the probability of
employment, especially for men. The same pattern of the effect of household
composition on participation emerges in all three surveys.

Non-wage income (approximated by business assets) plays a fairly signiﬁcant role in
sector choice decision.52 For men, having more assets is associated with higher
probability of being self-employed (in 1993) and lower the chance of private sector
employment (in 1993 and 2000).53 For women, business assets have a positive
association with self- and family business employment as well as working at the public
sector (in 1993). Since both farm- and non-farm business assets are included in the
calculation of business assets, it is possible that self-employed individuals acquiring a
signiﬁcant amount of assets are likely to manage their own business activities and the
same goes for women working at family-own business. The fact that in the private sector
around 70 percent of men and 50-60 percent of women are agriculture workers,
operation/production workers and laborers, a group of individuals that is expected to have
small amount of business assets, might justify this negative effect (see Appendix 5.7).
Notice that for women, business assets have negative effects on private sector

employment even though they are not statistically signiﬁcant.

 

’2 Household business assets variable is constructed as sum of household’s farm business (from section UT)
and non-farm business (section NT) assets. IFLS provides information on values for various type assets
and its ownership percentage by the household, head of household and spouse of head of household. Data
on farm business assets is available if there is at least one householder who, during the past 12 months,
worked in a farm-business but not as a farm worker on other household’s farm. Similarly, data on non-
farrn business assets is recorded if there is at least one householder who worked in a family-owned, non-
farm business or been self-employed in a non-farm business during the past year. F arm-business and non-
farm business assets are collected for the following types of assets: land, house or buildings, vehicles, other
equipment, and other. In addition, hard stem plants and livestock/poultry/ ﬁshpond are also collected for
fann-business assets while supplies/merchandise for non-farm business assets.

53

Sahn and Alderman (1988) ﬁnd that the ownership of paddy land in rural Sri Lanka
reduces labor force participation for both men and women. Signiﬁcant negative effect of
exogenous non-labor income on the participation in the self-employment sector is also
found for women in urban Guinea (Glick and Sahn [1997]). Uneamed income, house and
land ownership are found to signiﬁcantly reduce the likelihood of work for men and
women in Malaysia (Schafgans [2000]).

The inclusion of dummy variables representing province of current residence and its
interaction term with urban region are important in capturing regional differences in the
probability of entering a particular sector of employment (see estimates for full
speciﬁcation, Appendix Table 5.3). Individuals living outside Jakarta, an all-urban area,
have higher probability of being self-employed or working with no pay at family
business. Within province, the chance to work at these sectors is lower in urban areas
compared to rural area. This result veriﬁes our data that the majority of self-employed

individuals or those working as unpaid family worker reside in rural area.

Appendix Section 5: Sector Choice Model for Panel Respondents

Results that are based on panel respondents are interesting. In contrast to the
estimates from cross section samples, estimates from a sample consisting of panel
respondents will present the dynamics of the effect of schooling (as well as other
covariates) in determining individuals’ employment within similar set of individuals. As
was discussed on section 4, the criteria for individual selection and the inclusion of new

household members (such as in the split-off households) might be responsible for the

 

’3 Negative effect of business assets on the probability of working at the private sector also holds in the full
speciﬁcation.

54

difference in schooling attainment within different waves of the survey. In this section as
well as the next one, comparison of estimated coefﬁcients on own schooling between
cross section and panel versions of the sample will also be discussed.

Starting from speciﬁcations with linear own schooling (see Appendix Table 5.4 and
its summary, Table 5.4), one can easily observe that the sign of estimates are similar
between cross-section and panel samples of the surveys. The difference, however, lies in
the magnitude of these estimates. Schooling attainment effects on public sector for men
are higher within panel samples in comparison to those of cross-section, although the
difference shrinks with time.

A more interesting picture could be found from estimates with non-linear own
schooling (see Appendix Table 5.5 and its summary, Table 5.4). Overall, the picture
shows that the trend of association between own schooling and participation, except that
of private sector men, is the same for both cross section and panel individuals. For self-
employrnent sector, positive association is evidenced for individuals with low level of
schooling attainment, that becomes negative for the more educated individuals. A
positive correlation between the two variables of interest is found for those working at
public sector as well as a negative association for private sector women and women
working as family workers. The difference, in terms of trends, is found for men working
for wage at the private sector in 2000. Whereas for the cross section individuals, they are
positively correlated, no signiﬁcant association is found for the panel individuals.

Tuming to the magnitude of estimates, especially for public sector men, one can see
that they are different. Estimates for women who are working for wage also show some

dissimilarity although not as strong as those of public sector men. Since most of the

55

excluded individuals from the cross section data in the 1993 sample are of 53-59 years
old, one can say that individuals aged 25-52 are more likely to work at the public sector
than those of 53-59 years. This might be true since, on average, public sector individuals
retired at the age of 55, although extension for public sector service beyond that age is
possible in some cases. In 1997, the panel respondents are of age 29-56 while cross
section sample also includes those who are 25-28 and 57-59 years as well as new
individuals to the survey. The magnitude of panel sample is higher than those of cross
section, except for those with 7-12 years of schooling where they are comparable. In
2000, as the panel respondent are of age 32-59, estimated coefﬁcients are lower than
those of cross section individuals.

The same conclusion is obtained from the base speciﬁcation (see Appendix Table 5.6
and its summary, Table 5.5). The magnitudes of the estimates tend to be smaller in
comparison to previous speciﬁcation (the one with non-linear own schooling with no
control for parental schooling), indicating that parental schooling variables picked up
some of the family background variations of the individuals as in the cross section
samples.54

It is hard to make deﬁnite statements about the source of the difference between the
magnitude of panel and cross section estimates, especially for the 1997 and 2000

samples. It might be due to aging of the respondents, selection problems between panel

 

5’ Supporting evidence on occupation breakdown for panel respondents for analyzing the negative
(positive) effect of schooling on being self-employed for men (women) as discussed on subsection 5.3.1 is
shown on Appendix Table 5.8. The dynamics in the overall primary duties of panel individuals are not as
large as those of cross section. For example, agriculture workers for cross section individuals are slashed
from 37-40 percent in 1993 to 33 percent in 2000, while for panel individuals the proportion is unchanged
at around 40 percent in both years.

56

and cross section samples, or other reasons. This will be an interesting topic to be

explored in future work.

57

6. Wage Functions

Earnings are deﬁned as the logarithm of hourly income for self-employed workers or
as the logarithm of hourly wage for wage earners (public and private). Three sets of
speciﬁcations will be presented in this section.55 The ﬁrst one, the basic wage function,
includes a set of dummy variables representing schooling attainment, a set of age
variables (in splines), and another set of dummies of month of interview. In each of the
next two speciﬁcations, a set of additional variables is added. They are dummy variables
representing parental (father’s and mother’s) schooling and province of current residence
along with its interaction term with urban region.56 In addition to these variables, each
regression includes control for self-selection into each of the labor market sector
calculated from the multinomial logit sector choice model. Since income for self-
employed individuals is recorded as either net proﬁt or gross income, a dummy variable
indicating whether the income measure is gross income is also included in self-
employrnent speciﬁcations.57 The inclusion of gross income dummy variables only
occurs in the 1993 wage regressions since in IFLS2 and IF LS3 only a tiny portion of self-

employrnent income is collected as gross income.58

 

5’ OLS estimates for each of these speciﬁcations are presented for the three sector of employment of
interest, namely self-employment, public sector, and private sector. These are further stratiﬁed by gender.

5° Due to small cells on some variables, especially for women, some adjustments are unavoidable. These
modiﬁcations are aggregation of own schooling variable for women with 10+ completed years of schooling
and the exclusion of urban interaction terms for women living in some provinces as indicated at the
footnote of respected tables.

57 Net income is deﬁned as gross income subtracted by all business expenses.

58 In the 1993 sample, around 27 percent of self-employment income is recorded as gross income. In the
1997 sample, the proportion decreases to only 2 percent. This proportion is further reduced to 1 percent in
the 2000 sample. The inclusion of gross income dummy variables in 1993 wage regressions and the
exclusion of those from the 1997 and 2000 regressions could serve as an attempt to make regression
comparable between years, especially for returns to schooling for the self-employed.

58

IF LS collects detailed information on primary and secondary jobs for individuals
aged 15 years or older. I only consider primary job information on earnings in this essay
for the following reasons. First, only 43 percent of individuals with more than one job
are working at the same sector of employment in both jobs in the 1993 sample.
Combining, for example, gross income from self-employment (primary job) and net
salary from private sector (secondary job) and divided the result by total hours worked
will not be comparable to either eamings of self-employed workers nor eamings of those
working at the private sector. One solution to overcome this problem is to assign a
weight for each sector of employment of the primary and secondary job; but doing this
will not only complicate the empirical process but also adds additional assumptions on
earnings that are already measured with error. Second, the proportion of individuals with
secondary jobs is roughly 22 percent.59 Thus, ignoring information on secondary jobs is
justiﬁed for consistency of the earnings measure at the expense of sacriﬁcing some facts
of individual’s employment proﬁle.

Descriptions of earnings rates will precede the presentation of estimation results.
Predicted rates of returns and comparison with other studies will conclude this section.
In the Appendix, I will discuss estimation results based on panel respondents. In
addition, I will also compare results based on selectivity corrected estimates and OLS

estimates.

 

’9 In 1997, around 17 percent of working individuals have a second job. While in 2000, the proportion
increases to 25 percent.

59

6.1. Earnings Rate

Information on earnings for individuals who are self-employed (in terms of net proﬁt
or gross income) and those who are working at the public and private sectors (in terms of
net salary/wage) is recorded in last month and last year ﬁgures.60 To protect against
recall error, I will use the former. For the self-employed and wage earners with no
information on past month earnings, last year’s earnings (divided by 12) is used instead.
With regard to hours worked, IF LS collects the number of hours worked the week before
the survey as well as hours worked in a normal week. I use the former in constructing
total hours worked to avoid seasonality problems, especially for self-employed
individuals.61 62 Earnings rate is constructed by dividing monthly earnings by total hours
worked last month.63

Table 6.1 shows the mean and median of earnings rate stratiﬁed by years of
schooling, sector of employment, and sex. Earnings tend to increase with schooling
attainment. As a whole, public sector workers are better compensated than their

counterparts who are self-employed or working at private sector. Eamings disparity

 

6° In 1993, a small fraction (around 7 percent) of net salary/wage data is available in details, where the
amount received is broken down into the money, food, housing, transportation, medical beneﬁts and other
portions. This is ignored here.

6’ Weekly hours worked is restricted to 94 due to the way the data is recorded. Any observations with
weekly hours more than 94 are excluded from the sample. In the 1993 sample, the number of digits in both
measures is two with the possibility that 95 might refer to out of range response, 96 to not applicable, 97 to
refused to answer, 98 to do not know and 99 to missing. Despite of this interpretation, those responses
could also represent the ‘actual’ hours an individual worked per week. There is no way to make sure which
interpretation is the correct one. A hint can be found from time allocation module. This module collects
information on time spent on working to earn wages/salary, farming or doing self-employed activities
during the past week. Most (around 70 percent) of individual working 95-99 hours a week has the
corresponding time allocation on working of either missing or conditional codes. This fact is a support to
the ‘conditional codes’ interpretation. For the 1997 and 2000 samples, weekly hours are also top coded to
94. In any sample, the proportion of weekly hours of more than 94 is 3 percent or less.

62 Hours worked last month = 4.33 "‘ hours worked last week.

60

between gender does not exist among public sector workers. Let’s look more closely at
eamings in each sector of employment.

For the 1993 sample, I present earnings rate of self-employed workers in two
categories, those who provide net proﬁt information and those who answered self-
employment income in terms of gross income. The data shows that the mean of gross
income, overall, is higher than that of net proﬁt, for both men and women. Gross income
is higher for women than for men while net proﬁt is higher (although not signiﬁcant) for
men than for women. In 1997 and 2000, men, on average, have higher net income than
women.

There are some issues on self employment earnings worth mentioning before we
precede with estimating wage functions. Information on self-employment earnings are
collected by asking respondents the net proﬁt gain (net of all business expenses) during
the last month or last year. If the respondents are not able to provide net proﬁt
information then the value gross income during the last month or last year is asked
instead. Acknowledging the difﬁculty in estimating net proﬁt or gross income, especially
in recalling the value of them for the last 12 months (or even last month), information on
hourly eamings rate for the self-employed individuals might be noisy. As an implication,
rates of returns to schooling for the self-employed workers should be interpreted with

caution.

 

‘3 Eamings/wage rate for workers in the agricultural sector (which made up half of self-employed men and
a ﬁfth of self-employed women) are inﬂuenced by seasonality problems. This problem is acknowledged by
the inclusion of a series of dummy variables indicating month of interview.

61

Turning to public sector workers, one can see that men and women are equally
compensated.64 This, among others, might be a reﬂection that wage rates in the public
sector are set by the government. Compensation of women in the private sector,
however, is less than that of men. Although these disparities tend to decrease with
schooling attainment, women, on average, are paid only about 57 percent of that received
by men in 1993 and 68 percent in 1997 and 2000. This might be an indication that
women are discriminated against in the private sector, although this issue will not be

addressed in this study.

6.2. Wage Functions: the Effects of Own Schooling

The presentation of wage function results starts with speciﬁcation that controls for
schooling attainment of individuals and some other common covariates that are going to
be included in subsequent speciﬁcations, such as age (in splines), month of interview and
selection term calculated from multinomial logit sector choice model. Discussion on own
schooling estimates are focused on estimates from speciﬁcations in which schooling
attainment enters as a series of dummy variables.65

Prior to the discussion, to serve as a comparison, I will give a short presentation on
estimates from speciﬁcation with individuals’ completed years of schooling (or
speciﬁcation with linear own schooling attainment). One needs to notice the difference
in interpretation of estimates between the two speciﬁcations. Whereas own schooling in

the linear speciﬁcation refers the average rate of returns for each additional year of

 

6’ The only signiﬁcant difference happens if the number of observation in that particular cell is small, say
less than 25 individuals.

62

schooling regardless of the highest schooling attainment of individuals, each of the
dummy variables estimates representing a range of individuals’ own schooling attainment
refers to the wage advantage/disadvantage of being in the group relative to the that of the
excluded group.

Self-employment work might have different characteristics from that of public or
private sector work. In self-employment work, one deals with managing resources of his
own that might not occur if he works in the public or private sectors. Among others, it
consists of managing machines, workers and other capitals, such as lands and buildings,
needed to conduct the business. In addition, self-employed workers also manage their
own managerial tasks such as book keeping, asset management, marketing and public
relations. One implication of this is that returns to schooling among the self-employed
might be different from those of public or private sector workers. In particular, self-
employment rate of returns might also comprise returns to capital managerial skills,
returns to capital and returns to unpaid family workers. In making returns to schooling
comparison between self-employment work on one hand and public or private sector
work on the other, we need to keep in mind that the two do not share similar

characteristics.

6.2.1. Linear Own Schooling
The most straightforward speciﬁcation in returns to schooling literature is one with
linear schooling attainment. This speciﬁcation follows the one used by Mincer (1974) in

his seminal work in this genre. We are all aware that the downside of using this

 

6’ Within this speciﬁcation, schooling attainment enters as dichotomous variables representing 0-3, 4-6, 7-
9, 10-12, 13+ completed years of schooling. For some groups, as pointed out on notes of the tables,

63

speciﬁcation is that we cannot learn about the variations in returns to schooling between
those who, say, completed primary school and those who completed junior secondary
school.

Estimates for wage function speciﬁcation with linear own schooling are presented on
Appendix Table 6.1, while a summary of own schooling estimates is shown on Table 6.3.
In general, estimated coefﬁcients for completed years of schooling is positive and
statistically signiﬁcant for men and women who are either self-employed or working at
the private sector. Positive estimated coefﬁcients are also found among public sector
workers even though most of them are not statistically signiﬁcant. The exceptions occur
for public sector men in 1993, which have an 8 percent rate of return for each additional
years of schooling and for public sector women in 1997, which have negative (but
insigniﬁcant) rate of return. The positive and insigniﬁcant estimates might indicate wage
compression among public sector workers as was found by Fihner and Lindauer (2001 ).
Estimates based on speciﬁcation with non-linear schooling attainment will provide a
better explanation on this issue.

Men in the self-employment sector, on the average, have a 12.5 percent rate of return
in 1993. This rate decreases to less than 9 percent by 1997 and 2000. Among self-
employed women, average rates of return to one additional year of education also show a
declining trend between surveys although the drop is not so remarkable compared to that
of men. While in 1993 and 1997 the returns rates are comparable around 7.6-7.9 percent,
by 2000 it was cut down to 6 percent. A similar declining trend is also found among men

and women working at the private sector. For men, rates of return declined from 10.8

 

merging one or more age category are unavoidable due to small observations for that particular cells.

64

percent in 1993 to 8.4 in 2000. Among women, however, these rates decreased at a
slower pace, from 12.3 percent in 1993 to 11.4 percent in 2000.

These patterns, in a sense, conﬁrm some of Psacharopoulos ﬁndings as mentioned on
section 1. In particular as per capita income increases, as evidenced by the high level of
economic grth between 1993 and 1997, the rate of return tends to decrease. One can
also see that due to crisis of 1997/98 which caused a slowdown in economic growth of
1997-2000, the decrease in rate of returns during this period is not as much as that of the
1993-1997 period. Another argument for the decreasing rates of return is due to supply

shift of better educated workers.

6.2.2. Non-linear Own Schooling

Estimates for wage function speciﬁcation with non-linear own schooling are
presented on Table 6.2, while a summary of own schooling estimates is shown on Table
6.3. In general, the effects of schooling attainment on eanrings are jointly highly
signiﬁcant. This, however, does not hold for public sector women 1997 and 2000, where
the F -test statistics are not signiﬁcant (p-value = .738 in 1997, and .591 in 2000). For the
public sector in 2000 the F -test has a p-value of .058.

Returns to schooling increase with the level of education in a non-linear fashion.
These wage functions are, in general, convex.66 In 1993, self-employed men with 4-6

years of completed schooling have a 57 percent earnings advantage in comparison to

 

6° Small number of observations prevents me from estimating semi-parametric wage functions that places
no parametric restrictions on earnings-schooling function (see, for example, Strauss and Thomas [1995]
and Lam and Schoeni [1993] for Brazil and Hungerfold and Solon [1987] for the US.)

65

their counterparts with no formal schooling.67 The wage premium is even higher for self-
employed men with junior secondary, senior secondary and tertiary education, where
earnings are, respectively, 2.8, 3.9 and 8.9 higher than that of men with zero year of
schooling. Likewise is the increasing trend in earnings for self-employed women
although the differences in wage advantage by schooling levels are not as high as those
for men.

By 1997, the earnings advantage for self-employed within each level of schooling are
lower than those of 1993. Earnings premia for self-employed men are further reduced by
2000. While self-employed men with junior high education level of schooling enjoy a
178 percent earnings advantage in 1993, by 1997 and 2000 these advantages are cut to
140 and 101 percent, respectively. Among self-employed women, earnings advantages
are also slashed between years of surveys although the magnitudes of the cut are not as
high as those of self-employed men.

Not much can be said with regard to individual estimates on schooling attainment for
public sector workers. In addition to the limited signiﬁcance of the F -test statistics,
estimates for public sector men and women might also suffer from small sample size in
comparison those of self-employment and private sectors.68 This is especially true for
women. Wage advantage for men with upper secondary level and tertiary level of

schooling in 1993 is, respectively, 76 and 151 percent. By 2000, these wage premia

 

67 The coefficient of a dummy variable in semi-log speciﬁcation cannot be interpreted as relative effect (7)
of the variable (b) on the dependent variable (see Halvorsen and Palrnquist [1980]). The appropriate
transformation for this purpose is 7= e” - 1. Own schooling estimate for self-employed men with 4-6 years
of schooling is 453, implying earnings that is e“53 = 1.57 times (or 57 percent higher than) the earnings of
self— employed men with zero year of schooling.

6‘ Since the public sector is an education-intensive sector, only a small proportion in our sanrple has
education attainment with less than completed primary school. For this reason, the excluded category
within each of the public sector regressions is individuals with 0-5 years of completed schooling.

66

slightly increased to 89 and 166 percent. Among women, only estimates in the 1993
sample are signiﬁcant, where wage advantage for those with lower secondary level
schooling is 63 percent, while women with at least upper secondary level education enjoy
a 2.3 times higher wage rate in comparison to those with less than completed primary
schooling education.

Turning to private sector workers, one can observed that wage advantage increases at
an increasing rate within each survey for both men and women. Own schooling estimates
for men experienced a remarkable drop between 1993 and 1997 (except for those with
lower secondary level of schooling). A similar decline, however, is not found between
1997 and 2000, although some small changes occurred during the span of time. Whereas
wage advantage for women with 4-9 years of schooling increased between 1993 and
1997, the reverse is found among women with tertiary schooling.

The overall picture captured from these results is that, within sector of employment,
schooling attainment estimates increases with level of schooling at an increasing rate.
This is true for men and women who are self-employed or working at the private sector.
For public sector workers, however, due to the limited signiﬁcance of the resulted
estimates, not much can be said. These ﬁndings indicate that individuals with more
formal schooling are more favorably rewarded, and that the wage advantage increases in
a non-linear fashion with levels of schooling.

Over the years, earnings/wage advantages within each level of schooling are, in
general, declining. The decreases in these eamings/wage premia are noteworthy between
1993 and 1997, while between 1997 and 2000, the declines are not that large, and in

some cases they are only trivial. Even so, we also saw some small increases in own

67

schooling estimates for a small number of cases. As previously mentioned, one can
argue that the overall economic condition of a country (measured by level of economic
growth) during the span of time might provide an explanation of the decline in wage
advantage between 1993 and 1997 as well as the relatively stable level of it between 1997
and 2000.

Another explanation for this change in rates of returns is an increasing supply of
individuals with more schooling. Our data shows that the average completed years of
schooling for all individuals included in the sample increases by survey. Whereas in
1993, men and women have an average 6.1 and 4.4 formal schooling experience, in 1997
it has increased to 6.9 and 5.2 years. These averages are further increased to 7 .8 and 6.3
years by 2000. 69 Among women with higher secondary and tertiary education in 1997,
14 and 6 percent, respectively, decided to be self-employed; by 2000, these proportions
increased to 21 and 8 percent, respectively (see Table 4.2). Women with similar
schooling qualiﬁcations who worked at the private sector are 20 and 27 percent; while in
2000, the proportions are relatively unchanged: 19 and 28 percent, respectively. This
evidence might partly explain the decrease in earnings advantage between 1997 and 2000
in self-employment sector and the relatively small change of wage advantage in private
sector during the span of time. The same argument does not hold for men, indicating that

there might be some other factor(s) affecting the outcome.

In addition to schooling attainment, our speciﬁcations also include a series of age

variables (in splines), a series of dummy variables indicating month of interview, and a

 

‘9 Increases in schooling attainment also occur within each sector of employment as well as its stratiﬁcation
by sex.

68

selection coefﬁcient. Especially for the 1993 regressions, a dummy variable indicating
whether the earnings of self-employed workers are based on gross income measure is
also included. These variables are discussed next.

F-test for joint signiﬁcance of age variables are at least signiﬁcant at the 10 percent
level, except for private sector women in 2000 (p-value = .321). Our results show that
the slopes of splines are negative for the oldest age group (SO-59 years), except for public
sector women in 2000. For self-employed men in 1993 and 1997 and self-employed
women in 1997, negative estimates could be found as early as being 40-49 years old. It is
worth to mention that age for women might represent experience that is not comparable
to that of men since women might be temporarily out of the work force and engaged in
home production or childcare activities.

A series of month of interview dummy variables, intended mainly to pick up
variation of seasonality in eamings, shows no inﬂuence for private sector employment for
the 1993 and 1997 surveys as the F -tests are not signiﬁcant even at the 10 percent level.
In the 2000 sample, however, these dummy variables are jointly signiﬁcant. The reverse
is true for individuals working at public sector. Among self-employed workers, month of
interview variables are jointly signiﬁcant for the 1993 survey only. Recall that main ﬁeld
work for IFLS] and IF LS2 took place similar time of the year (August-January), while
IFLS3 started as early as June. Thus, the timing of survey has only limited consequence
on earnings.

The selectivity term (,1) also has limited effect on earnings.70 Within this

speciﬁcation, it has a positive impact on earning of self-employed women in 1993 and

 

7° Heckman (1979) suggests that when selection correction terms are not signiﬁcantly different from zero,
one can accept that OLS estimates are consistent as well as preferred because they are more efficient than

69

negative impact on self-employed men (1993), private sector women (1997) and public
sector women (2000).71 The positive sign for self-employed women in 1993 implies that
women in the self-employment sector have higher productivity than average women in
terms of having higher earnings than the eanrings of randomly selected women in the
population. The negative sign implies the reverse.

Gross income dummy variable is only signiﬁcant for women and it has a positive
sign. The result indicates that using gross income as a measure of eamings instead of net
income will increase wage advantage of self-employed women by 39 percent. For men,
however, this variable is not signiﬁcant, indicating that the usage of gross income or net
income does not statistically inﬂuence wage advantage of self-employed men.72

Finally, the model performs best for the public sector. In 1993, for example, it
explains almost 30 percent of the variation in earnings in that sector for men and almost
50 percent for women. The explanatory power of the model is relatively weak for the
self-employment sector for both men and women, where only 4-12 percent of the

variations in earnings are explained. This is expected due to the high measurement errors

in the measurement of earnings.

 

the two-stage estimates. In spite of it, discussion of estimate results are based on the selectivity corrected
version.

7’ The addition of control for parental and provincial covariates, however, changes the signiﬁcance of
selection terms.

72 Extending the speciﬁcation by the inclusion of parental schooling variables does not alter the results for
gross income variables. Further extension of the model by the inclusion of provincial dummy variables
resulted in a positive and signiﬁcant estimates for both self-enrployed men and women.

70

6.3. Wage Functions: the Effects of Parental Schooling

F arnily background information, represented by parental schooling, is added in the
speciﬁcations discussed in this sub-section (see Table 6.4 and its summary, Table 6.5).
Parental schooling enters the model through a set of dummy variables with no schooling
served as the omitted category. F -test for joint signiﬁcance of parental schooling
variables (that of father, mother as well as all parental schooling dummy variables) show
some mixed and sometimes inconsistent results between surveys. While father’s
schooling variables are not jointly signiﬁcant in the men’s wage regressions in 1993, they
are highly signiﬁcant for self-employed men in 1997, and only marginally signiﬁcant for
public sector men in 1997 and self-employed men in 2000. On the other hand, mother’
schooling variables as a whole are not signiﬁcant in affecting earnings for women in all
surveys, except for a marginal signiﬁcant effect for private sector women in 1997 (p-
value = 081). Putting the two sets of parental schooling variables together does not help
the consistency of joint signiﬁcance of these variables.

Before discussing the direct effect of parental schooling on earnings/wage, let us ﬁrst
focus our discussion on the effects of the inclusion of parental schooling variables in the
speciﬁcations. For self-employed and private sector workers, point estimates of own
schooling are, in general, lower in comparison to speciﬁcations that only control for non-
linear own schooling. The magnitude of the cut varies between 1-47 percent.73 These
decreases might due to the positive correlation between own and parental schooling in the
data, so that failure to control for parental schooling might resulted in biased estimates

for own schooling variables. Some increases in point estimates of own schooling are also

 

7’ Percentage change is only calculated for signiﬁcant estimates within the current speciﬁcations.

71

evidenced. These occur for self-employed men with 10+ years of schooling in 1997 and
for men with higher secondary education working at the self-employment sector.74
Among public sector men, the inclusion of parental schooling variables reverses the
signiﬁcance of point estimates from not signiﬁcant to signiﬁcant. In addition, these
parental schooling additions also decrease the magnitude of wage advantage for men with
10+ years of schooling in 1993 but increase the corresponding results in 2000.

Turning to the direct effect of paternal and maternal schooling in wage regressions,
one can easily ﬁnd that all point estimates of parental schooling (excluding missing
categories) that are signiﬁcant have positive signs, implying that having educated parents
is associated with an increase in earnings relative to having illiterate parents. For
example, having a father with some elementary level education is associated with a 21
percent eamings advantage in comparison to having a father with no formal schooling.
Due to the limited number of variables that are signiﬁcant (even at the 10 percent level)
between surveys, we cannot say much about the change in the direct effect of parental
schooling variables across time. The only exception is for self-employed men with a
mother who at least completed primary schooling education. Point estimate for these
men are comparable between the ﬁrst two waves of the survey (.376 in 1993 and .373 in
1997), while slightly dropped to .325 in 2000. Between any two successive surveys,
however, some of the point estimates changes are quite large.

In spite of the not so convincing results of F-test for the joint signiﬁcance of parental
schooling variables and the limited direct effect that can be inferred, the inclusion of

parental schooling is important in explaining point estimates of own schooling and, in

 

7’ The magnitude of the increase in point estimates for these cases is around 4-11 percent.

72

tum, rates of returns to schooling investments. However, it is hard to make any solid
conclusion of whether father’s schooling is more important than mother’s schooling in

inﬂuencing earnings of their children.

6.4. Wage Functions: the Effects of Regions of Current Residence

Speciﬁcations used in the previous sub-section are further extended by the inclusion
of province of current residence along with its interaction terms with urban region (see
Table 6.6 and its summary, Table 6.7). The omitted category for these regional variables
is Jakarta, an all-urban region. The employment market structure as well as wage offers
differs between regions. This is the effect that is expected to be pick-up in this set of
speciﬁcations. F -test results show that these variables are (jointly) very highly
signiﬁcant, except for public sector women in the 1997 sample (p-value = .065).
Adjusted R2 are also improved, in particular for those who are self-employed. The
magnitudes of own schooling estimates are also further reduced relative to those of the
second set of speciﬁcations. All of these results reﬂect the importance of regional
conditions on labor market conditions, the availability of school and its quality in the
calculating rates of returns to schooling.

Own schooling estimates are also affected by the inclusion of provincial variables.
Most of the effect resulted in the declines of point estimates; its magnitude varies
between 1 and 43 percent. For some small number of cases, increases in point estimates
are also evidenced; the magnitude of these increases is small, around 1-4 percent. This
ﬁnding also veriﬁes the importance of including regional differences in estimating rates

of returns to schooling investments.

73

Estimated coefﬁcients for provinces that are statistically signiﬁcant are negative,
implying living in that province, both in rural and urban areas, is associated with eaming
a lower earnings/wage than living in Jakarta. If estimated coefﬁcients for both province
and its urban interaction terms are signiﬁcant, the latter is mostly positive with lower
magnitude than the former. This implies that the wage disadvantage is lower for
individuals living in urban area than those living in rural area of that province.

For self-employed workers, I also present speciﬁcation in which household
composition and household assets variables are included as covariates (see Appendix
Table 6.2).75 The purpose of this exercise is to determine whether household
composition variables can explain self-employment earnings since either unpaid family
worker or temporary workers assist 45 percent of self-employed workers. Possession of
assets might also explain earnings, although one can argue that the accumulation of assets
might be a result of previous earnings, and thus it becomes endogenous. F -test results
show that household composition and assets are jointly signiﬁcant at the 1 percent level.
The estimated coefﬁcient of own schooling variables changes (some increases and some
decreases) by the inclusion of these variables; they are mostly small in magnitude.
Household assets variables for women are signiﬁcant and positive for women, implying
that self-employment eamings for women increase with the amount of business assets
owned by the household. Individual estimates for household composition are only
sporadically signiﬁcant; in general they are positive for having adults in the household.76
The interpretation of this result is that an adult aged 20-49 might help a self-employed

worker in executing his/her work, leading to an increase in earnings.

 

7’ It is possible, however, that identiﬁcation for selectivity terms is not satisﬁed within this speciﬁcation.

76 In addition, the existence of elderly women for self-employed men in 1993 is also signiﬁcant.

74

6.5. Predicted Rates of Returns

Rates of returns to schooling investments can be easily obtained from the
speciﬁcation with linear schooling attainment. In this case, the point estimates of own
schooling variables are the average returns for an additional completed year of schooling.
Within speciﬁcations in which schooling attainment of the individuals enters as a series
of dummy variables representing level of schooling, returns could be found by simply
subtracting the point estimates from adjacent levels of schooling and dividing the result
by the number of years needed to complete that level of schooling.

There is, however, another way to calculate the returns rates that is more reliable. In
addition, the estimated rates could be easily obtained from the regression results without
further calculations. The only modiﬁcation needed within the speciﬁcations of interest is
to replace the dummy variables for own schooling with its splines. Point estimates of
these splines represent the wage advantage for another completed year of schooling
within the range speciﬁed by the splines.

Splines are constructed for those with 0-3, 4-6, 7-9, 10-12 and 13+ years of
schooling.77 Estimates for these speciﬁcations are shown in Appendix Table 6.3 and
Appendix 6.4.78 Predicted returns to schooling estimates are shown for speciﬁcations
that only control for own schooling is shown on Table 6.8. In addition, Table 6.9
presents predicted rates when parental schooling variables are added as controls.

Overall, rate of returns are positive and signiﬁcant for individuals who are self-

employed or working at the private sector, meaning that completing one year of

 

77 To make presentation of results as comparable as possible to speciﬁcation with own schooling dummy
variables, the same category merging strategies are used.

7" The speciﬁcations on Appendix Table 6.3 only control for own schooling, while those of Appendix table
6.4 also control for parental schooling.

75

schooling within any level of education will beneﬁt the individual in terms of increased
wages. Completing one year of schooling for the ﬁrst 3 grades of primary school for
private sector men in 1993, for example, will increase wages by 8.7 percent. Each year
of completed schooling between grade 4 and 6 of elementary will add another 11.4
percent increase in wage. Likewise, the additions to wage if one completes each year of
lower secondary, higher secondary and tertiary level of schooling are, respectively, 7.9,
14.7 and 12.8 percent.

An interesting observation is found for returns among public sector men in 1993.
Returns to an additional year of schooling for individuals with tertiary level schooling
(8.9 percent) are lower than for those with higher secondary level (9.7 percent), although
the difference is small. This is also true for private sector men in 1993 and private sector
women in 1997 and 2000.

Between surveys, one can observe decreases in returns rates for self-employed men,
except for those with the highest schooling attainment. Among private sector men, there
are no clear patterns, some upward and downward movements existed; while among the
most educated, the returns are basically unchanged.

Turning to returns from speciﬁcations that also control for parental schooling, one
can see that the rates decreased by 3-26 percent among self-employed individuals and
those working in the private sector. Among self-employed men in 1997 and 2000, we
saw an increase in return rates.

One important ﬁnding that is related to the government program of 9-year mandatory
schooling is that returns to schooling for those with junior secondary level education are

higher than those who only complete primary school. This result holds for all sector of

76

employment and all waves of the survey, except for private sector men in the 1993

sample.

6.6. Discussion

Our results show that the wage advantage increases with schooling attainment. This
is true for those who are self-employed as well as for those working for wage in the
public and private sectors. Moreover, these ﬁndings are robust to the inclusion of various
sets of covariates. The degree of wage advantage is reduced by the inclusion of these
covariates, implying that parental schooling and regional variables have an effect on
returns to schooling as they pickup some unobservable measures of ability, family
background and schooling quality.

In contrast to worldwide compilation of rates of returns (Psacharopoulos, 1994), in
which primary education exhibits the highest returns among the three levels of education,
our show the ever-increasing pattern of returns with schooling attainment. A similar
conclusion is found among African countries (Schultz, 1998), in particular Ghana and
Burkina Faso, where returns are higher at the secondary and post-secondary school levels
than at primary level. The same is also true for the case of Botswana (Siphambe, 2000).
In Asia, Schafgans (2000) ﬁnds increasing returns to educational attainment among men
and women in Malaysia, and that women have lower returns than men although this
inequality not true among Malay natives. For the case of Turkey, Tansel (1999) ﬁnd that
return rate for state owned enterprise workers and private sector workers are higher

among individuals with middle- and high school level of education than those with

77

primary level education. The same conclusion also found in Taiwan (Liu, Hammit and

Lin, 2000).

6.6.1. Comparison with OLS Results

Given that selectivity correction terms are often not signiﬁcant it will be interesting to
compare estimation results that are based on OLS speciﬁcations that do not control for
selection bias and those that do. In particular, I will compare estimates from
speciﬁcations that control for parental schooling and sample selection (hereafter,
selectivity corrected speciﬁcation; see Table 6.4) and those that control for parental
schooling but not for sample selection (hereafter, OLS speciﬁcation; see Appendix Table
6.5).

Point estimates for own schooling variables among self-employed men, as expected,
are lower in the OLS speciﬁcation in comparison to those of selectivity corrected
speciﬁcation since selection coefﬁcient in the latter is negative. The difference is only
minimal among men with primary level of schooling and gets higher with schooling
attainment, meaning that selection among the self-employed men only affects those who
are more educated. Among self-employed women, estimated own schooling variables
between the two speciﬁcations are basically similar except in the 1993 sample when the
selection term is signiﬁcant and positive. With respect to the latter, own schooling
estimates are higher in the OLS speciﬁcation.

Among private sector workers, schooling attainment effect between the two
speciﬁcations are similar since selection terms in the selectivity corrected speciﬁcations

are not statistically signiﬁcant. Estimates for public sector men show a different trend

78

even though selection terms are not signiﬁcant. Among men, while OLS estimates are
higher than selectivity corrected estimates in the 1993 sample, by 1997 they are
comparable. By 2000, OLS estimates lower than those of selectivity corrected estimates.
The magnitude of selectivity term might be a suspect for these results (-.154 in 1993, .016
in 1997 and .147 in 2000).

In sum, even though the ﬁndings show that selection is not such a big issue in
estimating wage functions, as evidenced by the limited signiﬁcance of selectivity
correction terms, the inclusion of these terms might give better estimates than the

exclusion of them. This is especially true for the self-employed individuals.

Appendix Section 6: Wage Functions for Panel Respondents

As in sector choice model, I will also provide results of some basic regressions of
wage ﬁrnctions using panel respondents. Discussion for panel respondents will be
focused on point estimates of own schooling attainment, detailed estimates are left for the
readers to explore. In particular, the discussion will point out the difference between
estimates of panel- and cross-section respondents. The discussion will cover estimates
for speciﬁcations with linear completed years of schooling (Appendix Table 6.6 and its
summary, Table 6.3), speciﬁcations in which schooling attainment enter as a series of
dummy variables (Appendix Table 6.7 and Table 6.3), and speciﬁcations that control for
the inclusion of parental schooling variables in addition to non-linear schooling
attainment (Appendix Table 6.8 and Table 6.4). In addition, predicted retums to

schooling will also be discussed for speciﬁcations with schooling attainment only (Table

79

6.7) and speciﬁcations that control for parental schooling in addition to own schooling
attainment (Table 6.8).

Within the speciﬁcation with linear own schooling (in terms of completed years of
schooling), point estimates for an additional year of schooling in all sector of
employment in 1993 are somewhat lower in comparison to those from the cross-section.
For example, while self-employed panel men have an average 10.1 percent returns, cross-
section respondents have a slightly higher returns, 12.5 percent. Since most of the
excluded individuals from the cross section data in the 1993 sample are of 53-59 years
old, it is possible that these older individuals drive-up returns within the cross-section
sample due to, among other things, advantage in working experience. We also see that
for private sector workers and self-employed men, schooling returns based on panel
respondents are higher than those of cross section for both 1997 and 2000 sample. These
differences are, however, small. Among panel men, point estimates are similar within
each sector of employment between years of survey, indicating that returns to schooling
are basically unchanged over time.

Similar trends in returns to schooling are found ﬁom speciﬁcations in which
schooling attainment is non-linear. For example, each of the statistically signiﬁcant
estimates for self-employed panel men is lower than their corresponding cross-section
estimates in 1993, while the reverse is true in 1997 and 2000. Likewise is the case for
private men, where point estimates in 1993 are lower for the panel respondents while
they are higher in 1997 and 2000. One important observation worth mentioning here is

that the difference between individual estimates from cross-section and panel samples is

80

not that big so that, at least within this speciﬁcation, returns rates for cross-section and
panel respondents are close one to another.

Extending the speciﬁcation with parental schooling does not alter the trends from
speciﬁcations that only control for own schooling attainment. As expected, magnitude of
these estimates is mostly lowered. Thus, parental schooling affected retums rates for
cross-section and panel samples. Comparing the magnitude own schooling estimates
between cross-section and panel samples, one can learn that returns rates are lower for
panel individuals, especially for self-employed workers in 1993.

Predicted rates of returns gives a more interesting picture. Whereas returns for self-
employed panel men are lower in 1993, they are higher by 2000. A closer look at the
most educated group of self-employed men shows that there exists a different trend.
Whereas in the cross-section sample, the returns increased between 1993 and 1997 (from
18.9 to 20.9 percent for an additional completed years of tertiary schooling), by 2000 it
decreased to 13.2 percent. Among panel sample, however, retru'ns rates decreased
between 1993 and 1997 (from 24.3 to 14.9 percent), and increased to 23.8 percent by
2000. It is hard to give a precise answer to this phenomenon. One possible answer
would be that due to the prosperous economic condition in Indonesia during 1993-1997,
there might be a technological progress or innovation (say, in the agriculture sector) that
affected earnings of self-employed workers. Since the cross section sample in 1997
consists of younger individuals with possibly more recent knowledge of the technology,
one might expect that returns for the most educated group in the cross-section sample
increased while that of panel sample decreased since have yet to catch-up with the new

technology. Economic crisis might caused that technology to be not applicable due to,

81

among others, cost considerations, and the old agricultural technology are once again
broadly used. This might resulted in the increase of returns for panel individuals while
that of cross-section are adversely affected.

Adding parental schooling variables to the speciﬁcations does not change the trends
of returns rates. In general, predicted rates are lowered by the inclusion of these family
background information.

As in sector choice model, evidence on hand is not enough to pinpoint the source of
difference between the magnitude of panel and cross section estimates, especially for the
1997 and 2000 samples. Aging of the respondents and selection problems between panel
and cross section samples might be responsible for these differences. But, there might be

other factors causing this result that future work may explore.

82

7. Migration

It is widely known that individuals with more human capital, say those who are better
educated, are more likely to migrate than those with less. In low-income countries, net
migration tends to take place from the rural to the urban sector. Higher real wages in
urban areas and inadequate opportunities for educated individuals from rural areas will
attract these individuals to migrate to urban areas. In addition, migration of individuals
from the rural to urban areas might reﬂect their attempt diversify risk against adverse
condition that might affect agricultural work and production in the rural areas such as
drought, ﬂooding and ﬁre. The question posted in this section is whether individuals who
self-select themselves to migrate from rural to urban do actually possess a higher rate of
returns to schooling investments than those who do not. In other words, the exercise
undertaken in this section is intended to discover whether migration selection is a
potential problem in calculating rates of returns.

To see migration proﬁles and their relation to individual’s schooling attainment in our
samples, I divide the data into 6 regions of birth and 6 regions of current residency.
These regions are Sumatera, Java and other islands, each are further stratiﬁed by rural
and urban area. Table 7.1 shows the distribution of migration between regions by
schooling attainment. 79 It is clear that the more educated individuals are more likely to
migrate. In 1993, among individuals with no formal schooling, 79 percent of those born
in rural Surnatera are currently living in the same region and only 20 percent moved to

urban Sumatera, the corresponding ﬁgures for individuals born in rural Java are 71 and

 

79 Notice that migration that took place between birth and time of survey is not taken into consideration in
these tables.

83

18 percent, while for other islands are 85 and 14 percent. Compare these ﬁgures to those
of individuals with beyond primary school schooling attainment. Around 43 percent of
individuals born in rural Surnatera migrated to urban Surnatera and another 11 percent to
urban Java, leaving 46 percent staying in the region where they were born. In Java, six-
tenths of individuals born in nrral regions are currently living in urban regions of Java.
Similar patterns are found for the 1997 and 2000 samples. Looking at the migration
proﬁle of all individuals, we can see that the numbers between years are not that far off.
There exists, however, more out migration among individuals born in rural Bali, West
Nusa Tenggara, South Kalimantan and South Sulawesi (provinces included in the other
islands region) in 2000 than in 1993 or 1997. While among individuals born in the urban
areas of Sumatera and Java, there is more migration to the rural areas in 2000 compared
to that of 1993 and 1997. The latter might reﬂect the 1997 crisis effect: as factories in the
urban areas are closed and workers are laid-off, migration from urban to rural areas are
accentuated. Frankenberg, Smith and Thomas (2002) ﬁnd that the bottom quarter of
households (ranked by their per capita expenditure, PCE) in urban areas lost household
members during the crisis while urban households with above median PCE gained
members. Household size was expanding in the rural areas with the smallest expansion
for the poorest rural households.

Schultz (1988) argues that when migration occurs, the interpretation of proportional
shift in wage function associated with an additional year of schooling as the private rate
of returns to schooling investment might no longer hold since several working
assumptions are violated. In other words, migration selection might raise a problem in

the calculation of rates of returns to schooling. He suggests estimating rate of return by

84

 

 

splitting the sample with respect to individual’s place of schooling or place of birth and
by individual’s place of current residence. Following his suggestion, in this section I will
divide individuals into two groups, the ﬁrst consists of those born in rural area while the
second of those born in urban. I will also split up individuals based on whether they are
currently residing in a rural or urban region. The resulting estimates are expected to
indicate differences in returns to schooling that are due to migration patterns of
individuals in the samples.80

This sections starts with a description of schooling attainment and eamings with
respect to place of birth and of current residency. A short explanation of the speciﬁcation
to be used for empirical purpose will follow. Discussion of estimated results will

conclude the section.

7.1. Schooling Attainment and Earnings

Average years of schooling of individuals stratiﬁed by their region of birth and by
current residence are presented in Table 7.2. In general, individuals born in urban areas
have more formal schooling than their counterparts born in rural areas. Our data shows
that average completed years of schooling increased between surveys. It also shows that
the gap between individuals born in urban and those born in rural areas in terms of years
of formal schooling completed slightly increased between surveys, from 2.7 years in
1993 to 3.1 years in 2000. Similar pattems of schooling attainment are found among

individuals with respect to their region of current residence.

 

8° There is no study on returns to schooling for the case of Indonesia that explicitly taking into account the
importance of migration by splitting up the data by region of birth and region of current residency. Some
recent studies in this genre include Schultz and Mwabu (1998) for South Africa and Duraisamy (2000) for
India.

85

 

Average completed years of schooling among individuals born in rural areas is higher
than those currently residing in rural; this is also true for urban born individuals in
comparison to those of urban residence. Migration of the (mostly) more educated
individuals from rural to urban might drive down schooling attainment of the remaining
rural residence. Since there might also be individuals with low schooling attainment who
migrated from rural to urban, the average years of schooling of urban residence is pushed
down.

Moving to hourly wages, there are no statistically signiﬁcant differences between
rural- and urban born public sector workers (Table 7.3) and between rural- and urban
residence public sector workers (Table 7.4). The only exception exists in the 1993
sample, where urban residence public sector workers have, on average, a marginally
signiﬁcant higher hourly wage than those of rural residence. The similarity of wage rates
between regions of birth as well as of residence might be due to the fact that wages in this
sector are set by the govemment which in principle, does not differentiate between
regions.

Among self-employed and private sector workers, however, urban-rural differences in
average hourly wages are signiﬁcant both by birth and by residence. With respect to
place of birth, the ratio of wages between urban and rural declined between 1993 and
2000. 81 Whereas within the self-employed the ratio decreased from 1.8 in 1993 to 1.4 in
2000, the wage ratio decline among private sector workers is not as large, from 1.7 in
1993 to 1.5 in 2000. A similar downturn in the urban-rural wage ratio occurred between

surveys occurred for workers in the self-employment and private sectors, stratiﬁed by

86

 

their region of residence. For self—employed workers the ratio fell from 2 in 1993 to 1.4
in 2000, while that of private sector workers decrease from 2.4 to 1.5 during the same
period. It interesting to see that the wage rate disparity between urban and rural
individuals by birth and by residence is reduced with time in spite of the slightly increase
in disparity of schooling attainment (in terms of absolute difference of average years of
schooling).

Another interesting observation worth mentioning here is that the average wage rates
are lower among rural residents than among rural born individuals. This is also true
among urban individuals although the difference is not as large as those of rural

individuals except for the 2000 sample, where they are similar.

7.2. Model

The speciﬁcations used in estimating the effects of migration on returns to schooling
are easier than those of section 6. In particular, no parental schooling information enters
the speciﬁcation. Parental schooling for individuals born in rural areas, as expected, is
highly concentrated on the no-schooling category, higher than when all individuals are
included in the sample.82 Parental schooling variables will be excluded on that ground as
they have very limited effect in the wage regressions (see section 6). Out-of-school
experience (approximated by age) enters the model in quadratic form instead of splines.

In some categories, especially public sector, the number of observations is very small;

 

8' Notice that wage rates are presented in nominal terms so that no direct comparison of these rates between
surveys could be presented. Urban-rural wage rates ratios along with their comparison between surveys are
shown instead.

82 Individual living in rural area whose father has no formal schooling makes up 70-78 percent of the data,
while those living in urban area 52-58 percent. The corresponding ﬁgures for unschooled mother are 56-63
percent for rural area and 39-42 percent for urban area.

87

 

 

therefore, no province variables are included in the speciﬁcation to conserve degree of
freedom.

To be more speciﬁc, all individuals, regardless of their region of birth, are pooled in
the regressions instead of running separate regression for individuals born in rural and
those born in urban. I include interaction terms between all covariates and a dummy
variable indicating whether the individual was born in an urban area. One advantage of
using this speciﬁcation is that it allows one to statistically check whether the estimated
coefﬁcients are indeed different between regions of birth of the individuals. In addition,
using this speciﬁcation will increase the number of observations in comparison to the
separate speciﬁcations of rural- or urban born individuals. A similar set of regressions is

also run for individuals with respect to their region of current residency.

7.3. Discussion

The discussion of regression results consists of two parts; the ﬁrst one reports
estimates for individuals based on birthplace and the second based on residence.
Estimated coefﬁcients of wage functions for individuals born in rural and urban regions
are presented on Appendix Table 7.1 for speciﬁcations in which schooling attainment
variables enter in linear fashion as completed years of schooling. The corresponding
estimates with non-linear schooling attainment are shown on Table 7.5. Summary of
own schooling point estimates for these speciﬁcations are found on Table 7.6.

The linear model results show that holding everything else constant, individuals born
in urban areas do not have a higher retum to additional completed year of schooling.

Point estimates for years of schooling interacted with the bom-in-urban dummy variable

88

 

are, in general, not statistically signiﬁcant. Exceptions are for self-employed women in
the 1993 sample and for private sector men and women in the 2000 sample, although the
latter is only marginally signiﬁcant. Point estimates among these groups are positive
indicating that being born in an urban area is associated with additional returns to
schooling investments. Among self-employed women in the 1993 sample, the rate of
return for an additional year of schooling is 5.4 percent for those born in rural areas while
an additional 6.3 percentage points existed for urban born women. Likewise is the
additional 2.3 percentage points returns for an another year of formal schooling enjoyed
by private sector workers men (in the 2000 sample) on top of the 7.2 percent returns
among those born in rural. The rate of returns decreased over the years for self-employed
and private sector men, while for women they remained practically unchanged.83

Point estimates for own schooling variables among rural born individuals within the
non-linear speciﬁcations are positive and increasing with higher levels of education
attained. Interaction variables are mostly not signiﬁcant, implying that no wage premia
exist for being born in urban region in comparison to rural born individuals. Even though
not signiﬁcant, these interaction variables are mostly negative. A negative interaction
term indicates that the wage premium for individuals born in urban areas are lower than
that of rural born individuals. The only occurrence where both schooling attainment and
its interaction term are statistically signiﬁcant at 5 percent or better is found for private
sector men with lower secondary level of schooling in the 1993 sample. This group of

men enjoys a 134 percent wage advantage over those who never completed a single year

 

83 Point estimates for public sector workers are, in general, not statistically signiﬁcant for both
speciﬁcations in which schooling attainment variables enter in a linear- and non-linear fashion. These are
also true for regression results in which individuals are stratiﬁed by their region of residence. Small

89

 

of formal schooling, while being born in urban area decreases the wage premium to only
66 percent. The ﬁnding that region of birth does not affect the wage advantage in general
is supported by F -test statistics for joint signiﬁcance of interaction of schooling and age
variables that are mostly not signiﬁcant.

Over years, these schooling estimates decreased among men and women. For self-
employed women with at least high secondary schooling attainment, the earnings
advantage increased from 94 percent relative to unschooled women in the 1993 sample to
120 percent in 2000.

Tuming to speciﬁcations for individuals that take into account region of residence,
sirrrilar ﬁndings emerges. Similar in the sense that the interaction terms are in general not
signiﬁcant so that returns rates for investments in education are not statistically different
between individuals residing in rural and urban settings.

Linear speciﬁcation estimates show that a year of completed schooling gives 10-11
percent returns for self-employed men and 7-8 percent for private sector men (see
Appendix Table 7.2 with summary on Table 7.8). Among self-employed women,
especially in the 1993 and 1997 samples, living in an urban area is associated with more
than doubled returns rate than those living in rural. While private sector women living in
urban area in our 1993 sample are getting an extra 4.3 percentage points returns in
comparison to their counterparts living in rural area.

Non-linear speciﬁcations provide limited evidence of the advantage or disadvantage
in terms of eanrings for living in urban area (see Table 7.7 with summary on Table 7.8).

Living in rural area for the most educated private sector men in the 1993 sample is

 

observations might be responsible for these results. For this reason I will not discuss results for public
sector workers.

90

associated with a wage premia that is three times as much as those with zero year of
schooling; while living in urban areas signiﬁcantly increases wage advantage among this
group. Likewise is the case for the most educated women working at self-employment
sector in the 1993 sample and private sector women with lower secondary schooling level
of education in the 1997 sample. For these groups of women, no wage premia for living
in rural area in comparison to those of unschooled women, while living in urban is
associated with a wage advantage of, respectively, 210 and 166 percent.

Experience, proxied by age and age-squared, plays a limited role in explaining wage
difference for self-employed and private sector workers for both sets of speciﬁcations.
The sign of these age variables, provided that they are signiﬁcant, are positive while
those of age-squared are negative, implying the increasing effect (at a decreasing rate) of
experience in the wage functions. Selection coefﬁcients, as in Section 6, have only

limited effect on wage of the workers.

In sum, information on the region of birth (whether an individual was born in rural or
urban area) and on the region of current residence (whether an individual resides in rural
or urban area) plays a limited role in determining wage and, in turn, the rates of returns to
schooling investment. Notice that stratifying individuals based on region of birth is
intended to capture, among other things, the variance of schooling availability and quality
between rural and urban area in determining income. The second set of regressions, that
stratiﬁed the individuals based on their region of residence, is aimed at explaining the
difference of labor market conditions and wage structure in determining earnings. Our

results show that migration selection is not a potential problem in calculating returns to

91

schooling. The bias in the estimated returns is minimal if one does not include variables
that represent migration behavior of individuals for the case of Indonesia within the time
period considered. Of course this conclusion might change with time, as new datasets are
available.

One reason for the limited role of migration variables might due to the deﬁnition of
urban and rural region, which was taken from the BPS (the Central Bureau of Statistics),
is imprecise. For example, the high population densities of the urban and rural areas are
causing the difference between the two areas unclear. This problem affects the region of
current residence. In addition, since information on urban-rural region of birth is
collected based on recollection of individuals, it might also be possible that the
respondents did not provide accurate response with regard to their region of birth.

Some of current studies exploring migration issues in the context of developing
countries only stratify individuals based on their residency since information on place of
birth might not be available. Duraisamy (2000) ﬁnds, for the case of India, that returns to
per year of schooling are higher in rural than in urban areas for primary and secondary
levels of schooling as well as those with technical diploma/certiﬁcate. The results hold
for both men and women.84 Schultz and Mwabu (1998) ﬁnd that among African natives
in South Aﬁica, returns are higher in rural than in urban areas.85 They identiﬁed that
limited access of Africans to secondary and higher education as well as to working
opportunities under the Apartheid government as the explanation of the results. Both of

these studies, however, obtained separate estimates for individuals living in urban or rural

 

8" In this study, however, no distinction between the sectors of employment was made.
85 The study estimates returns rate for Africans, colored, Indian and white groups of individuals.

92

areas. No direct comparison on returns rates between the two groups is presented as was

done in this study.

93

8. Returns to Schooling by Age Cohorts

Our sample consists of individuals aged 25-59 years in each of the three waves
considered. This implies that respondents were born between 1934 and 1978 for the
IFLSl sample, 1938-1982 for the IFLS2 sample and 1941-1985 for the IFLS3 sample.
During this span of time, Indonesia went from movements to gain her freedom to the
implementation of a series of comprehensive development programs (called the ﬁve-year
development plan, Repelita). Schooling availability during this period changed
drastically. During the Dutch-Indies administration, school admission was only awarded
to children whose parents worked for or had close relations with the administration. In
1950, around 5 million children were enrolled in primary schools while another 6 million
children were not (Oey-Gardiner, 2000). The number of enrolled student increased ever
after. In 1973 the government started a major school construction program especially at
the primary level in 1973 (see Duﬂo [2000]). The goal of this program was to achieve
universal enrolhnent during the fourth Repelita of 1984-1989. School quality, in addition
to schooling availability, also improved during this span of time. Improvement in the
availability of schoolbooks as well as improvement in the quality of teachers, especially
at the primary level, testiﬁed to this enhancement in school quality. In addition, there
also exists a national-based curriculum for each level of schooling that is revised on a
regular basis, which might also affect the quality of students across time.

In light of the progress in schooling availability and quality, different cohorts in our
sample may have gone through schooling of different quality. One way to account for

the disparity of schooling quality in estimating rate of return to schooling investment is to

94

re-estimate the rates separately for different cohorts.86 87 For that purpose, in this section
I will split individuals in our sample into two age cohorts — a young cohort (individuals
aged 25-39) and an old cohort (40-65).88 One problem in aggregating individuals over
such a wide-range of age is to assume that individuals within a cohort have the same
schooling and labor market conditions. For empirical purpose, it is better to have a
smaller class interval of age grouping but small sample (especially for public sector
workers) prevents me from splitting up the data into more than two age cohort groups.
Section 8 starts with a description of schooling attainment and earnings among these
two age cohorts groups. It is followed by a brief explanation of the speciﬁcations used

and concludes with discussion of wage regression results.

8.1. Schooling Attainment and Earnings

Table 8.1 shows schooling attainment (in terms of years of schooling) for the young
and old cohorts by sex and sector of employment. The table shows that individuals
belonging to the young cohort are more educated than their old cohort counterparts.
Average years of completed schooling for each cohort increases with time. However, the
years of formal schooling disparity between cohorts is also increasing, indicating that

schooling attainment of the younger individuals increases at a faster rate than that of the

 

86 In the United States, this literature focuses on the effects on earnings of the baby-boomers and other
generation as they entered labor market and over time. Welch (1979), for example, shows that as the large
cohort of baby boom generation went into employment, their entry-level market wages were depressed
although the effects diminished over time.

87 For developing countries setting, recent studies include Schultz and Mwabu (1998) for South Africa,
Baraka (1999) and Vere (2002) for Taiwan, Duraisamy (2000) for India, and Aromolaran (2002) for
Nigeria.

88 Average age of IFLS] sample is 40.5 years for men and 39.6 years for women; while for IF LS2 sample,
these averages are 39.6 and 39.4 years, respectively. Men in IFLS3 sample is, on average, a year younger
(38.5 years) than those in IFLSZ sample; while the corresponding women are half year younger (38.8
years).

95

older. By 1993, the disparity between cohorts is 1.4 years with young individuals having,
on average, 5.3 years of schooling while their older counterparts 4.4 years. These
averages increased to 8.1 and 5.5 years by 2000, causing the disparity to increase to 2.6
years. Major school construction started to show some results in increasing schooling
attainment of individuals in the 2000 sample. Extension to the criteria used in
interviewing respondents (see Section 4) might also responsible for the increase given
that the average age of individuals decreases with time of survey (see footnote 88) and
that younger individuals are expected to have more formal schooling.

Hourly earnings between young and old cohorts who are self-employed are not
signiﬁcantly different. Average earnings are, however, slightly higher for the younger
individuals. Among private sector workers, wage rate are at parity in the 1993 and 1997
samples. Younger workers are paid 10 percent higher wage than older individuals by
2000. As expected, among public sector workers, it is the older individuals that are
favorably paid in comparison to their younger counterparts. Between years, older
individuals within this sector are paid a third more than what is earned by the young
cohort. Given the fact that earnings for public sector worker are set by the government,
the earnings advantage of the old cohort might reﬂect their tenure as public sector

servant.

8.2. Model
Similar speciﬁcations as those of Section 7 are utilized to study the effect of cohorts
in estimating returns rates. Whereas the model in previous section interacts all covariates

with a dummy variable indicating whether the individual were born in urban area or are

96

currently living in urban area, in this section the interaction dummy variable indicates

whether the individual belongs young cohort or is of age 25-39 years at time of survey.89

8.3. Discussion

Point estimates of covariates of wage ﬁrnctions for young and old cohorts are
presented on Appendix Table 8.1 for speciﬁcation in which schooling attainment
variables enter in linear fashion as completed years of schooling. Corresponding
estimates with non-linear schooling attainment are shown on Table 8.2. Summary of
own schooling point estimates for these speciﬁcations are found on Table 8.3.90

The linear model results show that belonging to the young cohort among private
sector men is associated with a lower return to an addition year of school completed. In
1993, older individuals within this group have returns rates of 13.2 percent while that of
younger individuals is 3.9 percentage points lower. Likewise are the returns in the 1997
and 2000 surveys that are, respectively, 10.3 and 10.2 percent for older individuals,
representing a 2.7 and 2.4 percentage points advantage over the younger individuals.
Thus, between years, the difference between retums of old and young cohorts decreased
from 3.9 percentage points in 1993 to 2.4 percentage points by 2000.

Among self-employed workers, younger men in the 1993 are unfavorably rewarded in

terms of returns rates. Whereas older individuals have a return of 15.6 percent for each

 

89 Similar to speciﬁcation on Section 7, some aggregation of schooling attainment (0-6 years for men and 0-
9 years for women) is unavoidable due to the small number of observations within these categories,
especially for those belonging to the young cohort.

9° For the similar reasons mentioned on Section 7, I will not discuss results for public sector workers.
Within the linear speciﬁcations, returns for old cohort women in the 1997 sample is 14.6 percent, while that
of younger cohort is 28.2 percent lower, with both point estimates are statistically signiﬁcant. The latter
means that returns rate for younger women is negative. A closer look at the number of observations used
for this regression is quite small, 309, in which 196 belonging to old cohort and the remaining, 113, to
young cohort.

97

year of formal schooling, returns for young cohort are 4.1 percentage points lower (p-
value = .054). Beyond these groups, returns rates are at parity between the two cohorts.

Turning to the non-linear schooling attainment speciﬁcations, point estimates for own
schooling are mostly positive and increasing with level of education for both men and
women. Interaction terms of schooling attainment with young cohort dummies are
negative and mostly signiﬁcant for private sector men. Among private sector women,
these interaction terms are mostly negative but the degree of signiﬁcance is limited.
Interaction terms among self-employed workers are not signiﬁcant (except for men in
1993) but their magnitudes are also mostly negative. These negative interaction terms
imply that the young cohort tends to have lower wage premia in comparison to the old
one. For example, in the 1997 sample, private sector older men with higher secondary
schooling have wage advantage of 3.5 times those of unschooled men, while belonging to
young cohort reduced this advantage to only 1.5 times.

Following Schultz and Mwabu (1998), interpretation of the negative interaction terms
is that the supply of younger workers relative to its derived demand may be higher than
the supply of older workers relative to its derived demand in the private sector. Data in
Table 8.1, as previously mentioned, supports the ﬁnding. Average years of schooling of
young cohort are higher than that of old cohort, while hourly earnings for the two groups
are (mostly) at parity. These patterns would suggest higher returns to schooling for older
men working at private sector.

Between 1993 and 1997, the magnitude of interaction terms among private sector
men tends to decrease, while the reverse is true between 1997 and 2000. Among private

sector women with at least higher secondary level of education, the disparity of the wage

98

advantage between cohorts increased between 1993 and 1997 and decreased between

1997 and 2000, in contrast to that of private sector men.

In sum, stratifying the samples by individuals’ age and comparing their returns rates
show that the younger individuals working at private sector have lower wage advantage,
and thus, rates of return in their schooling investments in comparison to those of the old
generation. These differences are more pronounced among men than women indicating
that wage advantage is more evenly spread between cohorts among women than among
men. Among the self-employed, however, no statistically signiﬁcant differences were
found.

Recent studies on returns to schooling investments between cohorts in the context of
developing countries show mixed results. Duraisamy (2000) ﬁnds that the returns to
primary, middle and secondary schooling in India are lower for the younger cohorts (i.e.,
the 15-29 and 30-44 years old individuals) than for the oldest cohort (45-65 years).
However, for higher secondary, college and technical diploma the opposite is true.
Aromolaran (2002), for the case of Nigeria, ﬁnds that wage returns to an additional year
of post secondary schooling are signiﬁcantly higher for younger workers (25-34 years)
than for older workers (45-64 years). De Brauw and Rozelle (2002) ﬁnd that the returns
to another year of formal schooling for off-fann workers in rural China is 9.3 percent for

individuals younger that 35 years and 3.4 percent those aged 35 and older.

99

9. Returns to Schooling using Potential Experience

So far, labor market experience (or out-of-school experience) in all empirical
exercises for estimating returns to investment in schooling is approximated using age of
individuals. In developing countries, as dictated by the data, it is common to use age as
an approximation of out-of-school experience while in developed countries experience is
used instead. Mincer (1979) notes that even though age may represent a depreciation
factor in the wage function it is not a good measure of accumulated post-school
investments, and that the latter are better represented by actual experience in the labor
market.

There are some variations to the approximation of labor market experience. One of
them, which is commonly applied, is by constructing experience variable from age minus
completed years of schooling and six, the latter comes from the assumption that
individuals start school at the age of 6. This approach is not utilized in this essay due to
grade repetition and dropout rate, which is quite signiﬁcant for developing countries,
including Indonesia.91 In this section, labor market experience is approximated by
‘potential experience’ constructed by reducing the individual’s age by the age at which
he/she ﬁnished or quit school. IFLS is probably the only survey in Indonesia that collects
information on age an individual ﬁnished or quit school which enables one to construct
the potential experience variable. It is expected that the use of potential experience will
take into account the issue of grade repetition which, as noted by Behrrnan and

Deolalikar (1991), is important in explaining returns to schooling. Information on

100

individual’s schooling history is only collected for those aged 15-49 years so that age at
which one ﬁnished or quit school is not available for those aged 50-59 years. For this
reason, return rates are calculated for individuals aged 25-49 instead of 25-59 as was in
previous sections.

This measure is, of course, not a perfect representation of ‘true’ labor market
experience, especially among women. For example, potential experience still suffers
from the issue of individuals who temporarily quit school and working for a certain
period of time due to economic hardship. Among women who are in the middle of
having children and taking care of them, the way potential experience is calculated might
for sure not accurately representing labor market experience. The purpose of doing this
exercise is to learn whether returns to schooling are sensitive to the use of different
measure of labor market experience.

Speciﬁcations utilized to describe these differences are simpliﬁed versions of the
ones use in Section 6. In particular, age and potential experience enter the speciﬁcation
in quadratic terms. In addition, no controls for parental schooling and province of
residence are included in the wage regressions. Own schooling enters as number of years
of schooling completed (linear) and as a set of dummy variables representing level of
schooling (non-linear). Point estimates for the former are presented in Appendix Table
9.1 and for the latter in Table 9.1. Summary of schooling attainment covariates are
shown on Table 9.2. As previously mentioned, in order to make comparison between
returns estimates with age and potential experience as labor market experience,

individuals with incomplete information on potential experience are excluded from the

 

9' As noted on Section 2, Behrman and Deolalikar (1991) ﬁnd that failure to control for repetition and
dropout rates will signiﬁcantly upward bias the returns to schooling estimates, especially for the lower

101

regressions. Thus, samples in this section are relatively younger than those of Section
6.92

Starting from the linear speciﬁcation, one can see that an additional year of formal
schooling will result in consistently higher rates within speciﬁcation with potential
experience than that with age.93 The differences are, however, not large; the largest
difference existed for public sector men in the 1993 sample, where returns using potential
experience are 3.7 percent higher than that using age (12.8 vs. 9.2 percent). The
difference in point estimates among self-employed and private sector men stays at the
same level between years of surveys. Corresponding differences among women ﬂuctuate
with time.

Tunring to speciﬁcations in which schooling attaimnent variables are non-linear, one
can easily observe that wage advantages among self-employed and private sector men are
higher in the potential experience speciﬁcation than in that using age.94 The magnitudes
of these differences are, in general, small. Among women, however, some of the
corresponding differences are positive (i.e., higher) and some are negative (i.e., lower).
Again, these differences are small in terms of magnitude. Private sector women in 2000
have lower wage advantage if potential experience is used as proxy for labor market

experience that if age is use instead.

 

levels.

92 As an aside, one can make comparison between the ‘full’ sample (i.e., consisting of individuals aged 25-
59) and sample of individuals aged 25-49 (see Appendix Table 6.1 and Appendix Table 9.1 for linear
speciﬁcation, or Table 6.2 and Table 9.1 for non-linear speciﬁcation, or its summary, Table 6.3 and Table
9.2). Excluding individuals aged 50-59 does not change point estimates of own schooling by much, for
both linear and non-linear speciﬁcations. Especially for self-employed women, an additional year of
schooling gives basically identical rates of returns.

93 The difference in own schooling estimates between the two speciﬁcations is reported only if both point
estimates are statistically signiﬁcant at 5 percent or better.

102

A higher point estimate in the potential experience speciﬁcation than in the age
speciﬁcation means that using age as proxy for labor market experience might
downwardly bias returns to schooling. A lower point estimates indicates the reverse.
Our results show that returns for self-employed and private sector men are downward

biased if age is use as approximation of labor market experience.

 

9’ It is not easy to make inference for public sector men since point estimates are only sporadically
signiﬁcant.

103

10. Summary

This study estimates private rates of return to schooling investment in Indonesia using
three waves of Indonesia Family Life Survey (IF LS): 1993, 1997 and 2000. Two features
that distinguish the study from similar studies in this genre for the case of Indonesia are
the following. First, it starts ﬁom the assumption that individual maximizes his/her
utility in making decision not to participate in the labor force, to be self-employed, to
work for wage in the public sector, to work for wage in private sector, or to take part in
the unpaid family-owned business. This feature is implemented by modeling a sector
choice model using multinomial logit, following Lee (1983). Second, it takes into
account the fact that better educated individuals, especially those from rural areas, are
more likely to migrate to urban areas since returns to schooling may be higher in the
urban areas. Following Schultz (1988) observations are divided on the basis of whether
the individual was born in rural-urban regions and on the basis of whether the individual
currently residing in rural-urban regions.

While the ﬁrst feature is intended to resolve the problem of sample selection bias, the
second one is meant, to some extent, to overcome omitted variable bias from omitting
cost of living (and thus real wages) variation between rural-urban regions or from
variation in school quality where the individual obtain his/her formal education. In
addition, omitted variable bias resulted from omitted measures of ability and family
background is also taken into account by the inclusion of parental schooling inforrrration
in the wage speciﬁcation. Parental schooling attainment is rarely included in estimating

returns to schooling in Indonesia because of lack of data.

104

This study adopts the human capital investment model of Becker (1964). According
to this model, an individual makes human capital investment decisions by maximizing
the discounted present value of future earnings with respect to years of schooling subject
to the opportunity cost of time and goods spent (in acquiring such capital) and the rate of
interest. The wage function used to estimate the rate of return is a Mincerian (1974)
function with some modiﬁcations. To be speciﬁc, an individual’s schooling attainment
enters the speciﬁcation through a set of dummy variables representing a range of
completed years of schooling. Parental background is represented by a set of father’s and
mother’s schooling attainment dummy variables. Experience, approximated by age of
the individuals, comes in splines. In addition, a selectivity error term to correct for
sample selection bias is also included in the speciﬁcation. This error term is calculated
from the sector choice model, which serves as the ﬁrst leg of the two-stage multinomial
logit-ordinary least squares procedure of Lee (1983). Rettu'ns to schooling are estimated
based on adult men and women in their working age of 25-59. Separate estimates of
returns to schooling are calculated for men and women to get insight into factors
affecting employment participation decisions as well as the determinant of rettu'ns to
schooling by gender and to learn whether there are differences in returns to schooling by
gender.95

This section starts with summarizing empirical results for both sector choice model
and wage regressions. It follows by explaining some policy implications from empirical

ﬁndings. Agenda for future works concludes the section.

 

95 Schultz (2001) provides explanations why returns to schooling estimates are different by gender, while
Alderman and King (1998) explore explanation of gender differences in parental investment in schooling,

105

10.1. Empirical Results

10.1.1. Sector Choice Model

Sector choice estimates show the importance of education in allocating men and
women to their employment paths and that the effect of schooling attainment is not
linear.96 Schooling effects, however, work in different directions depending on the
employment sector of interest. Among self-employed men in the 1993 sample, a
negative effect of schooling is evidenced for those with at least lower secondary level
education, and these negative effects get stronger with higher level of schooling. By
2000, however, schooling attainment for men with primary schooling is positively
correlated with the probability of being self-employed with no signiﬁcant effect of
schooling attainment beyond the primary level. Among self-employed women, a positive
association between schooling and the odds of joining the sector is found only for those
with primary level education; these patterns hold for all three waves of survey.
Therefore, those who are more able in terms of their education avoid being self-
employed, while those with less formal schooling prefer self employment work. The
effect of schooling attainment for public sector employment gets stronger with level of
education for both men and women. The fact that these effects tend to be stronger with
time might indicate that public sector employment has become increasingly education-

intensive with time.

 

which has inrplications to the disparity of return rates between sexes.

96 Results are based on ‘base speciﬁcation’ estimates. Base speciﬁcation refers to speciﬁcation that
includes non-linear individuals’ own schooling attainment, parental schooling, a set of household
compositions variables and business assets. Selectivity terms that are included in the wage regressions are
calculated based on these base speciﬁcations.

106

Among men working in the private sector, we see no association between schooling
and employment within the sector in 1993. By 2000, however, positive and signiﬁcant
associations emerged for those with up to secondary level education. Among private
sector women, negative associations were found for those with 4-9 years of completed
schooling. The fact that private sector employment is dominated by occupations
associated with lower levels of schooling attainment such as paid agricultural workers
might explain the negative effect of schooling attainment on private sector employment
for women. Being an unpaid family worker among women is, as expected, negatively
correlated with education. Around 70-80 percent of women in this sector work are
engaged in agriculture work. Thus, those with higher schooling attainment are likely to
avoid working at this sector and which, in turn, explains the negative effect of education
in participation in (unpaid) family business sector.

Parental schooling covariates have, in general, limited effects on employment
decisions of children. Not much can be said about the direct effect of father’s or
mother’s schooling. For example, in the 1993 sample, having a father who completed
primary school is associated with a 25 percent increase in the probability of being self-
employed but a 28 percent decrease in the likelihood of working in the private sector.
These associations, however, do not hold across surveys. In some cases, parental
education has a counterintuitive effect on employment path of children. For example, in
the 1997 sample, having a father with at least secondary schooling is associated with a
decrease in the probability of any working activity for men. The only solid association
across years is found among women working at the family business sector where effect of

parental father’s and mother’s schooling are negative. This is expected since educated

107

parents prefer their children not to work unpaid. Limited variation in parental schooling
variables, especially that of mother’s, might be responsible for the unusual results. In
spite of these shortcomings, parental schooling variables are jointly signiﬁcant. In
addition, the inclusion of these variables captured some of the family background
information in making the employment decision as evidenced by the decrease in the
magnitude of own schooling point estimates.

Sector participation probabilities of men and women follows an inverted U-shaped
proﬁle with respect to age, with younger individuals (aged 25-39) more likely to
participate in any kind of employment activities compared to their older counterparts
(aged 40-59). Household composition and business asset variables, intended to identify
selection term for the second stage OLS wage regressions, are jointly signiﬁcant at the 1
percent level. The presence of adult men (aged 20-49) in the household lowers
employment participation for both men and women. This, among self-employed men
might indicate substitution effect for working in this sector, while among women the
ﬁnding might indicate women’s role in household work (or an increase in home-produced
goods) and child bearing/rearing responsibilities. Having an adult woman in the
household increases a woman probability to work at the private sector (in 1997 and
2000), implying that the existence of adult female in the household might lighten the
burden of household chores and child-care activities. The presence of an elderly woman
(aged 50+) reduces men employment activities but increases women participation in
wage work. Having an elderly man (aged 50+) in the household lowers work
participation for men and women except that it increases women participating in family

business work activities. The latter might indicates that the care of elderly man could be

108

combined with some working activities at the family-owned business. With respect to
business assets, men having more assets are associated with higher probability of being
self-employed (in 1993) but a lower the chance of private sector employment (in 1993
and 2000). Among women, business assets have positive association with self- and
family business employment as well as working at the public sector (in 1993).
Individuals living outside Jakarta, an all-urban area, have higher probability of being
self-employed or working with no pay at family business. Within province, the chance to

work at these sectors is lower in urban area compared to those living in rural area.

10.1.2. Wage Regressions

Our results show that the wage advantage increases with schooling attainment. This
is true for those who are self-employed as well as for those working for wages in the
public and private sectors. Moreover, these ﬁndings are robust to the inclusion of various
sets of covariates. The degree of wage premia are reduced by the inclusion of these
covariates, implying that parental schooling and regional variables have an effect on
returns to schooling as they pick-up some unobservable measures of ability, family
background and schooling quality.

In 1993, a self-employed men with 4-6 years of completed schooling has a 44 percent
earnings advantage in comparison to their counterparts with no formal schooling. The
wage premiums are even higher for self-employed men with junior secondary, senior
secondary and tertiary education, where eamings are, respectively, 2.5, 3.5 and 7.6 higher
than that of men with zero years of schooling. There is an increasing trend in earnings

for self-employed women although the differences in wage advantage by schooling levels

109

are not as high as those for men. By 1997, the earnings advantage for the self-employed
within each level of schooling are lower than those of 1993. Earnings premia for self-
employed men are further reduced by 2000. Among self-employed women, earnings
advantages are also slashed between years of surveys although the magnitudes of the cut
are not as high as for self-employed men.

Schooling attainment estimates for public sector workers are positive for men with at
least upper secondary level education, while among women these estimates are in general
not signiﬁcant. In addition to the limited signiﬁcance of the F -test statistics, estimates for
public sector women (as well as men) might also suffer from small sample sizes in
comparison those of self-employment and private sectors.

Wage premia among private sector workers increase with level of schooling at an
increasing rate within each survey for both men and women. Own schooling estimates
for men experienced a remarkable drop between 1993 and 1997 (except for those with
lower secondary level of schooling). A similar decline, however, is not found between
1997 and 2000, although some small changes occurred during that span of time.

The overall picture captured from these wage regressions indicates that individuals
with more formal schooling are more favorably rewarded, and that the wage advantage
increases in a convex (non-linear) fashion with levels of schooling. Over years,
eamings/wage advantages within each level of schooling are, in general, declining. The
decreases in these earnings/wage premia are noteworthy between 1993 and 1997, while
between 1997 and 2000, the declines are not as large, and in some cases they are only
trivial. One can argue that the overall economic condition of a country (measured by

level of economic growth) during the span of time might provide an explanation of the

110

decline in wage advantage between 1993 and 1997 as well as the relatively stable level of
it between 1997 and 2000. Another explanation for this change in rates of returns is due
to supply shift in individuals with more schooling. The (overall) average completed
years of schooling in the 1993 sample is 6.1 years for men and 4.4 years for women; by
2000, the corresponding numbers are 7.8 and 6.3 years, respectively.

Turning to estimated rate of returns to investment in schooling, our results show that
returns are positive and signiﬁcant for individuals who are self-employed or working at
the private sector, meaning that completing one year of schooling within any level of
education will beneﬁt the individual in terms of increase earnings/wages (refer to Table
6.9 to obtain the magnitude of these rates among levels of schooling, employment sectors
as well as years of survey). Completing one year of schooling for the ﬁrst 3 grades of
primary school for private sector men in 1993, for example, will increase wage by 8.6
percent. Each year of completed schooling between grade 4 and 6 of elementary will add
another 10.8 percent increase in wage. Likewise, the additions to increase in wage if
he/she completes each year of lower secondary, higher secondary and tertiary level of
schooling are, respectively, 7.4, 13.4 and 11.7 percent. Between surveys, one can
observe decreases in returns for self-employed men, except for those with the highest
schooling attainment. Among private sector men, there are no clear patterns, some
upward and downward movements existed; while among the most educated, the retums
are basically unchanged.

One important ﬁnding that is related to the government program of 9-year mandatory
schooling program is that returns to schooling for those with junior secondary level

education is higher than those who only complete primary school. This result holds for

111

all sectors of employment and all waves of the survey with one exception: for private
sector men in the 1993 sample.

With respect to age, our results show that the slopes of splines are negative for the
oldest age group (50-59 years), except for public sector women in 2000. A series of
month of interview dummy variables, intended mainly to pick up variation of seasonality

in earnings, has only limited consequences on earnings. The selectivity term (It) also

has limited effect on earnings.

F -tests for the joint signiﬁcance of parental schooling variables show some mixed and
sometimes inconsistent results between surveys. While father’s schooling variables are
not jointly signiﬁcant in the men’s wage regressions in 1993, they are highly signiﬁcant
for self-employed men in 1997, and only marginally signiﬁcant for public sector men in
1997 and self-employed men in 2000. Mother’ schooling variables as a whole are not
signiﬁcant in affecting earnings for women in all surveys except for a marginal
signiﬁcant effect for private sector women in 1997. Point estimates (or direct effect) of
paternal and maternal schooling in wage regressions shows that all statistically signiﬁcant
results are positive, implying that having educated parents is associated with an increase
in eamings relative to having illiterate parents. However, the existence of these
statistically signiﬁcant estimates is sporadic. In spite of the not so convincing results of
F-test for the joint signiﬁcance of parental schooling variables and the limited direct
effect that can be inferred, the inclusion of parental schooling is important in explaining
point estimates of own schooling and, in turn, rates of return to schooling investments.

Cost of living differences and difference in employment market structure might also

affect wage premia. F-test results show that region dummy variables are (jointly) very

112

highly signiﬁcant. Results show that living in a province outside Jakarta, both in rural
and urban areas, is associated with lower wage than living in Jakarta. In addition, results
also ﬁnd that wage disadvantage is lower for individuals living in urban area than those
living in rural area of that province.

One ﬁnal point worth mentioning here is a comparison between the results of this
study and that of Filmer and Lindauer (2001) as shown in Table 10.1. In order to make
results comparable among studies, wage premia from this study are weighted by the
number of individuals (men and women) in each level of schooling. OLS estimates of
this study refer to results from speciﬁcation with no control for parental schooling and
sample selection. The ﬁndings show that individuals with junior high (or lower
secondary, 7-9 years of schooling) is more favorably rewarded than that of Filmer and
Lindauer, while wage premia for individuals with schooling attainment beyond lower
secondary are comparable among studies. Adding parental schooling information and
selectivity correction terms show that the difference in estimated own schooling

attainment between the two studies are not that far off.

10.1.3. Migration and Age Cohort

The effect of migration in this study is captured by stratifying individuals based on
region of birth and of current residence. The former is intended to capture, among others,
the variance of schooling availability and quality between rural and urban areas in
determining income, while the other is aimed at explaining the difference of labor market
conditions and cost of living in determining earnings. Our results show that neither

region of birth nor region of current residence play an important role in determining

113

wages and, in turn, the rates of return to schooling investment. The ﬁnding veriﬁes that
migration selection is not a potential problem in calculating returns to schooling or at
least that the bias in the estimated returns are minimal if one does not include variables
that represent migration behavior of individuals.

Since individuals of different cohorts in our sample may have gone through schooling
of different quality as well as availability, individuals are stratiﬁed into two age cohorts —-
young cohort (individuals aged 25-39) and old cohort (40-65) to account for these
disparities. Our regression results show that the younger individuals working in the
private sector have a lower wage advantage, and thus, rates of team in their schooling
investments in comparison to those of the old generation. These differences are more in
evidence among men than women indicating that the wage advantage is more evenly
spread between cohorts among women than among men. No statistical signiﬁcant
differences between cohorts were found among men and women working in the self-
employment sector. Interpretation of the negative interaction terms is that the supply of
younger workers relative to its derived demand may be higher than the supply of older

workers relative to its derived demand in the private sector.

10.2. Policy Implications

There are a number of ways in which analysis of rates of returns in schooling
investments are useful for policy making. On one hand, social returns may provide an
indication of which level of schooling the government should invest in most. For
example, if the returns to primary education are signiﬁcantly higher than those of

secondary education, then policy makers, which in the case of developing countries are

114

the governments, can allocate more resources to primary education. On the other hand,
returns to schooling can also help the government in seeking the most appropriate
education policy that is consistent with the development of human capital of its country.
For example, if return rates are low for a certain level of education, then analysis of
returns to schooling might provide an explanation of the causes of the low returns.
Likewise if the returns across years are declining, then an explanation of the source of the
decline might be very useful for policy making.

Empirical ﬁndings illustrate that returns to schooling investments are, in general,
positive and increasing with level of education. In relation to the 9-year mandatory
schooling program, our results show that returns for attending a year of secondary level
education is high, especially for women. The result might serve as a motivation for the
central government and/or local authorities to make comprehensive plan in administering
the operation of education at the secondary level, especially, the junior secondary level.
The plan, in addition to provision of schooling infrastructures, should also deal with the
preparation of curriculum, books and other instructional equipments as well as human
resource planning for the teachers at this level of schooling. Moreover, special attention
should be given to overcome the urban-rural gap and rich-poor gap in the enrollment
rates at the junior secondary school level.97 An indication of variations of returns within
province as well as its rural-urban differences as shown from wage regressions might
serve as valuable information for the government.

Following conventional wisdom, returns might fall as Indonesia’s economy developes

and Indonesians, on average, gain more education. For example, if the growth of supply

115

of secondary level graduates in the employment market is ahead of its demand, the high
returns for secondary level education might not persist. It is also possible that at some
point in the ﬁrture tertiary education graduates might drive secondary level graduates to
lower paying job, causing increase in unemployment. Therefore, not only should the
government make an accurate plan of junior secondary schooling system as dictated by
the commitment to implement 9-year compulsory schooling program, but it also has to

plan employment creation for these future graduates.

10.3. Future Agenda

From an empirical point of view, estimates of returns to schooling in this study have
some shortcomings. First, the small number of observations is causing low predictive
power for some estimates. As an example, it is difﬁcult to make clear-cut statements at
the disaggregated level for those working in the public sector. Second, the nature of
parental schooling data that has not much variation is also causing the same problem,
especially in the 1993 wave of IF LS. Third, this study compares cross section individuals
between years of survey instead of exploiting the panel nature of IFLS data, even though
some comparison of results using panel individuals between years are shown for sector
choice and wage regressions.

The ﬁrst path for future agenda is to redo this exercise using more sophisticated panel
data econometric tools. This gives the opportunity of actually learn behavioral change
within individuals. Decomposition techniques of factors affecting the decline in returns

across years might add richness to our knowledge as well as provide valuable input for

 

97 Oey-Gardiner (2000) ﬁnds that in 1999 the urban-rural gap in age-speciﬁc enrollment rates is high
among junior secondary age (88 percent in urban vs. 74 percent in rural) and that the gap between the rich

116

 

policy makers. Realizing that individuals might migrate due to better work opportunities
or schooling availability and quality gives rise to the prospect of endogenizing migration
in estimating schooling returns.

Although the focus of this study is to learn the effects of family background and
migration on the estimation of returns to schooling, I realized that there is more that can
be done with the rich nature of the data. The ﬁrst extension is to use the national ﬁnal
examination (EBTANAS) scores to proxy for ability. Off course not all individuals
interviewed have an EBTANAS score which will signiﬁcantly reduce the number of
observations. Since IF LS also collects information on non-co resident kin, it is possible
to get the effect of siblings, for example, to enrich our speciﬁcation. Another possibility
is to link with information from community and facility survey of IFLS to get

information on schooling quality.98

 

 

and poor is disturbing (93 percent for riches quintile vs. 66 percent for the poorest quintile).
9" See Betts (1999) for a review on the literature on the returns to quality of education.

117

Table 4.1
Summary of Number of Observations

 

 

 

Observations Percentage
Dropped Remained of Book III
Respondents
aged 25-59
IFLSI
Individuals answering Book III 14,418 -
Observations dropped due to:
- Age restrictions (age<25 or age>59) 3,887 10,531 100.0
- Incorrrplete schooling attainment a) 457 10,074 95.7
- Incomplete sector of employment 11 10,063 95.6
- Wage missing 395 9,668 91.8
- Wage outlier (wage< 10 or wage>20,000) 38 9,630 91.4
IFLSZ
Individuals answering Book III 21,562 -
Observations dropped due to:
- Age restrictions (age<25 or age>59) 8,665 12,897 100.0
- Incomplete schooling attainment 65 12,832 99.5
- Incomplete sector of employment 1 12,831 99.5
- Wage missing 586 12,245 94.9
- Wage outlier (wage<10 or wage>20,000) 55 12,190 94.5
IFLS3
Individuals answering Book III 26,731 -
Observations dropped due to:
- Age restrictions (age<25 or age>59) 11,190 15,541 100.0
- Incomplete schooling attainment 8 15,533 99.9
- Incomplete sector of employment 5 15,528 99.9
- Wage missing 559 14,969 96.3
- Wage outlier (wage<10 or wage>40,000) 108 14,861 95.6
Panel Respondents b)
IFLS] 3,111 6,519 67.7
IFLSZ 5,671 6,519 53.5
IFLS3 8,342 6,519 43 .9

 

Source: IFLS], IFLS2 and IFLS3.

a) 411 respondents have their Book III answered by proxy.

b) Percentages on the last colomn are relative to ﬁnal observations of respected surveys.

118

 

Table 4.2

Distribution of Employment by Schooling Attainment, Sector and Gender
Cross Sectlon Respondents

 

Non-emp.
Men Women

Self Emp.
Men Women

Public Sector
Men Women

Prlvate Sector

Men Women

Family Worker

Men Women

Observations
Men Women

 

IFLSl
0 year

1-3 years
4-6 years
7-9 years
10-12 years

13+ years

All Individuals

1 FLSZ
0 year

1-3 years
4-6 years
7-9 years
10-12 years

13+ years

All Individuals

IFLS3
0 year

1-3 years
4-6 years
7-9 years
10—1 2 years

13+ years

All Individuals

4.2 43.3
(0.88) (2.12)
3.3 39.1
(0.65) (2.13)
3.4 49.8
(0.55) (1.87)
5.2 53.9
(0.93) (2.51)
3.6 42.2
(0.69) (2.36)
4.4 27.3
(1.27) (4.00)

3.8 44.9
(0.32) (1.38)

7.0 46.4
(1.05) (2.17)
6.7 43.6
(0.84) (2.16)
6.1 49.3
(0.62) (1.76)
8.2 58.6
(0.99) (2.15)
6.6 46.3
(0.74) (1.78)
1 1.2 23.4
(1.53) (2.64)

7.1 47.1
(0.42) (1.24)

7.4 30.3
(1.13) (1.81)
3.9 28.8
(0.67) (1.69)
3.9 36.1
(0.43) (1.33)
5.8 47.7
(0.71) (1.82)
6.1 42.8
(0.60) (1.46)
8.4 22.7
(1.17) (1.87)

5.5 35.8
(0.33) (0.93)

60.6
(2.95)
63.7
(2.48)

23.3
(1.53)
27.2
(1.59)
58.5 26.3
(1.95) (1.30)
45.1 23.4
(2.65) (1.96)
22.4 15.7
(1.81) (1.66)
14.8 6.3
(2.44) (2.16)

49.3 23.8
(1.69) (0.88)

53.4 23.4
(2.65) (1.66)
55.5 27.4
(2.22) (1.80)
49.4 26.6
(1.63) (1.27)
39.0 19.3
(2.11) (1.61)
24.5 14.1
(1.55) (1.29)
15.3 6.0
(1.77) (1.45)

41.3 22.5
(1.30) (0.85)

56.5 28.4
(2.66) (1.75)
57.1 32.2
(2.22) (1.55)
52.4 29.4
(1.55) (1.21)
41.9 27.4
(1.79) (1.54)
27.9 21.2
(1.39) (1.11)
20.8 8.1
(1.67) (1.20)

42.3 26.5
(1.16) (0.77)

2.3 0.6
(0.67) (0.26)
1.7 0.9
(0.46) (0.50)
5.9 0.8
(0.83) (0.37)
12.9 2.7
(1.91) (0.75)
36.2 26.0
(2.28) (2.14)
52.2 45.5
(3.04) (4.46)

13.4 4.7
(0.96) (0.49)

2.7 0.4
(0.63) (0.19)
1.3 0.6
(0.51) (0.39)
3.9 0.5
(0.55) (0.28)
9.3 2.5
(1.25) (0.68)
24.7 15.6
(1.53) (1.32)
40.8 42.5
(2.63) (2.96)

11.5 4.7
(0.68) (0.40)

0.9 0.3
(0.39) (0.16)
1.0 0.5
(0.34) (0.24)
2.6 0.4
(0.41) (0.16)
5.5 1.2
(0.79) (0.38)
17.6 8.7
(1.07) (0.86)
35.5 38.6
(1.96) (2.28)

9.8 4.6
(0.52) (0.31)

31.7 13.7
(2.73) (1.29)
30.1 12.5
(2.33) (1.23)
31.5 8.7
(1.79) (0.89)
35.9 8.4
(2.33) (1.24)
37.3 12.7
(2.36) (1.60)
27.4 18.9
(2.73) (2.94)

32.5 11.5
(1.38) (0.70)

33.0 15.7
(2.64) (1.32)
34.4 14.7
(2.24) (1.40)
37.7 12.1
(1.62) (0.97)
41.1 11.4
(2.04) (1 .27)
42.0 19.7
(1.91) (1.49)
31.1 27.1
(2.48) (2.62)

37.4 15.1
(1.18) (0.74)

31.4 18.8
(2.75) (1.63)
35.4 16.8
(2.18) (1.38)
37.8 16.9
(1.55) (1.11)
43.2 12.2
(1.76) (1.19)
45.0 19.4
(1.50) (1.14)
33.8 28.4
(1.92) (2.05)

39.2 17.8
(1.09) (0.75)

1.2 19.1
(0.51) (2.02)
1.3 20.3
(0.38) (2.11)
0.7 14.3
(0.25) (1.47)
0.9 11.7
(0.38) (2.04)
0.6 3.5
(0.28) (1.07)
1.1 2.1
(0.64) (1.19)

0.9 15.2
(0.15) (1.29)

3.9 14.1
(0.89) (1.61)
2.2 13.7
(0.53) (1.56)
3.0 11.5
(0.44) (1.17)
2.4 8.4
(0.63) (1.45)
2.2 4.4
(0.54) (0.86)
1.5 1.0
(0.56) (0.57)

2.6 10.7
(0.31) (0.90)

3.9 22.2
(0.90) (1.92)
2.5 21.7
(0.56) (1.82)
3.3 17.2
(0.46) (1.28)
3.6 11.5
(0.63) (1.30)
3.4 7.8
(0.47) (0.90)
1.5 2.1
(0.44) (0.61)

3.2 15.3
(0.26) (0.91)

568 1.493
788 1,089
1,376 1,573
576 514
697 543
270 143
4,275 5,355
698 1,626
777 1,079
1,747 1,974
752 695
1,170 910
463 299
5,607 6.583
570 1,271
865 1,156
2,099 2,439
1.101 983
1,774 1,339
736 528
7,145 7,716

 

Source: lFLSl, IFLSZ and IFLS3.
Estimates are in percentage. Standard errors (in parentheses) are robust to clustering at the community level.

119

 

._o>o_ bEsEEOQ 05 an war—83.0 9 5.58 0.5 $805.85: :5 Echo Emu—Em .owﬁcoobq 5 Ba moans—sum
Ami—n: :5 NWT: .59.: U3.50m

 

 

 

an: an: 35.: 8...: :3: :5: Gt: Gm: 5.: 32:

Sn. :3 a: 3 m: «.8 Z 3. 3m 2m QR ed 88:98
Ga: 808 Go: 5.: :98 86.: am: so: an: and:

23 83 6.2 3 «.2 3m 3 «.3 3H 6.3 QR 2 as???
so: and: so: at: 82: 83: 5.: an: 32.: :2:

36:6. $3 3; on no. ”.3 3 Na an 2:. 3m 34 as» 3.8
5.: 3.: ANN: so: 83: $3: §.: 5.: 5.: 5.8

83 23 we 3 6.8 cam E on w: Sm Ev E 8&32

«3.:
6N: and: $2: 9%.: 34.8 §.: 30: A8: 3w: E:

:3 83 n: 3 ma 3” Z n: 6.2 0% ma. n2 as; 3-8
3.: the Go: 9.0: 96.8 :2: an: E: 85: 5.3

ms; 82 6.2 I «.2 mom we «.2 NR we. 2:. 3 e831;
so: Ewe $6.: an: an: 36.8 as: an: at: 5.8

«Nam 23 2: 2 we 2:. .3 5.: SN 3% m3 3 £832;
EV: 82: 5.: 3.: $2: 32: a”: 2.6.: 5.: as:

:3 ea 3 2 3m 3:. E 3. oi Sm mm 4.2 88:02

8.:—
§: 3.2: Ammo: so: 53: so: 54.: Ga: Ga: 36.8

£2 we; «.2 Z 3 2m 2 02 EN can 34 3 88:98
5.: :2: 25.: Go: 5.8 5.: an: 8.: :6: 8...:

82 2.2 N: no 5.: QR 3, 5.2 08 a? v.8 ma engage
an: $3: $3: an: 5.: SN: A2: A8: 30: and:

$6.. 42.. ”.2 no 6.: 6.2 we 2: .2 at. 4.3. 3 88:68
9;: 2.6.8 3.: 85: $98 $1: AR: 85: am: 5.8

no 8m ”.2 3 2: $4 3“ 3 4.2 <3. 2“ mm $65-2

5...:
.5535 :02 Eon—53 =02 EOE—GB =02 Hon-33 :02 EOE—o; no: Evan—:3 no:
8:33.336 .8333 3.8."...— ..38m 895.:— ..83m 33.... «amaze—aim 20m ionic—9:01.82

 

3:00:38: accuom 89.0
.6250 E:— ueuoom Jw< .3 vague—18m no 5:39.55

n... 933,—.

120

Table 4.4

Distribution of Employment by Schooling Attainment, Sector and Gender

Panel Respondents

 

Non-emp.
Men Women

Self Emp.
Men Women

Public Sector
Men Women

Private Sector
Men Women

Family Worker

Men Women

Observations
Men Women

 

IFLS]
0 year

1-3 years
4-6 years
7-9 years
10-12 years

13+ years

All Individuals

IFLSZ
0 year

1-3 years
4-6 years
7-9 years
10-12 years

13+ years

All Individuals

1 FLS3
0 year

1-3 years
4-6 years
7-9 years
10-12 years

13+ years

All Individuals

2.5 43.3
(0.85) (2.61 )
1.9 40.3
(0.60) (2.38)
1.4 49.2
(0.38) (2.01)
1.3 53.3
(0.59) (2.83)
2.3 38.3
(0.65) (2.53)
2.5 27.1
(1.24) (4.47)

1.8 44.6
(0.24) (1.50)

4.7 41.5
(1.13) (2.51)
3.4 39.3
(0.83) (2.42)
3.0 47.3
(0.59) (2.03)
4.4 56.7
(1.22) (2.86)
3.8 41.4
(0.98) (2.71)
1.7 22.1
(0.94) (3.83)

3.5 43.8
(0.41 ) (1.48)

5.3 27.5
(1.15) (2.04)
2.8 27.0
(0.75) (1.96)
2.9 33.7
(0.54) (1.73)
4.6 46.6
(1.09) (2.77)
4.6 38.1
(0.97) (2.76)
4.9 13.8
(1 .82) (2.95)

3.8 32.1
(0.41) (1.24)

61.2
(3.56)
61.5
(2.88)
57.7
(2.23)

20.9
(1.97)
24.9
(1.73)
25.8
(1.43)
44.7 22.9
(3.11) (2.19)
21.7 14.1
(2.09) (1.69)
11.9 2.1
(2.63) (1.45)

48.2 22.3
(1.83) (0.96)

57.7
(3.34)
57.4
(2.82)
56.3
(2.11)

25.4
(2.03)
28.2
(2.18)
29.7
(1.59)
44.3 23.5
(3.25) (2.30)
26.3 13.7
(2.34) (1.93)
14.0 3.9
(2.52) (1.73)

47.3 25.2
(1.67) (1.08)

63.0
(3.25)
63.2
(2.64)
61.7
(2.19)

30.4
(2.23)
32.4
(1.79)
33.6
(1.59)
48.4 29.9
(3.10) (2.36)
30.6 24.1
(2.30) (2.35)
14.6 7.2
(2.49) (2.00)

51.6 30.2
(1.60) (1.05)

1.5 0.8
(0.69) (0.35)
1.2 1.0
(0.47) (0.56)
6.1 1.0
(0.98) (0.49)
14.6 3.1
(2.30) (0.87)
41.6 30.1
(2.68) (2.54)
56.3 51.0
(3.99) (5.24)

14.9 5.6
(1.12) (0.60)

3.1 0.3
(0.91) (0.18)
1.3 0.9
(0.60) (0.60)
5.5 0.7
(0.91) (0.45)
13.7 3.4
(2.14) (1.01)
38.7 25.5
(2.49) (2.44)
59.2 55.9
(3.69) (4.56)

14.6 5.3
(1.00) (0.55)

1.6 0.4
(0.74) (0.22)
1.0 0.7
(0.45) (0.29)
3.5 0.7
(0.65) (0.31 )
10.4 1.5
(1.76) (0.62)
32.3 17.1
(2.27) (2.1 1)
62.0 67.8
(3.46) (4.07)

13.0 5.2
(0.85) (0.46)

33.9 15.3
(3.44) (1.66)
34.0 13.1
(2.72) (1.50)
34.0 9.2
(2.13) (1.04)
38.9 9.4
(2.84) (1.54)
34.0 13.3
(2.65) (1.89)
27.5 17.7
(3.46) (3.73)

34.3 12.1
(1.56) (0.82)

33.0 17.9
(3.19) (1.72)
36.3 17.2
(2.80) (1.86)
34.6 10.8
(1.94) (1 .07)
36.6 7.7
(2.88) (1.36)
31.1 14.0
(2.53) (2.06)
25.1 17.3
(3.43) (3.25)

33.7 14.1
(1.41) (0.89)

29.2 18.5
(3.26) (1.89)
31.1 15.9
(2.54) (1 .73)
30.1 14.6
(1.98) (1.33)
34.5 7.6
(2.97) (1.37)
31.3 13.0
(2.26) (1 .73)
18.5 11.2
(2.70) (2.46)

30.1 14.6
(1.36) (0.93)

0.9 19.7
(0.52) (2.40)
1.4 20.7
(0.50) (2.26)
0.8 14.9
(0.28) (1.53)
0.5 11.4
(0.37) (2.33)
0.4 4.2
(0.29) (1.39)
1.9 2.1
(1 .07) (1 .45)

0.9 15.4
(0.18) (1.38)

1.6 14.9
(0.70) (2.06)
1.5 14.5
(0.62) (1.86)
0.7 11.5
(0.26) (1.35)
1.1 8.7
(0.54) (1.79)
0.2 5.4
(0.21) (1.41)
0.0 0.8

- (0.78)

0.9 11.7
(0.21) (1.10)

1.0 23.2
(0.53) (2.09)
1.8 24.1
(0.65) (2.06)
1.8 17.4
(0.43) (1.68)
2.1 14.4
(0.73) (2.10)
1.3 7.8
(0.51) (1.64)
0.0 0.0

1.5 18.0
(0.23) (1.19)

325 852
517 794
929 1,190
378 385
488 405
160 96
2,797 3,722
385 973
465 699
926 1,172
366 379
476 372
179 127
2,797 3,722
305 799
495 748
938 1,242
374 395
480 386
205 152
2,797 3,722

 

Source: IFLSI, IFLSZ and IFLS3.
Estimates are in percentage. Standard errors (in parentheses) are robust to clustering at the community level.

121

 

._o>o_ bEsEEOo of an mew—82:0 8 6:29 2m $0855qu EV macho Eats—am .owﬁcoob: E 98 8883mm .39 E ow<

amt—u: :5 NW2”: .54.”: U3.59m

 

 

 

an: 5.: $6.: am: at: $_.: en: ﬁn: $22.: 5.:
3 SN :2 3 2: 2m 3 o5 mom 3m 2m 2: as; ~28
so: 5.: an: 8:: $2: 5.: so: 22.: 28.: $6.:
23 5 N2 2 n2 :3 an 4.2 3m 3% 2m 2 es» 3.9.
GM: 82: am: $52: :2: 5.: am: am: $1: :2:
$3 an; E. 2 6.2 can 3 02 in now ”am 3 ea» 3.2
E: 62: so: am: 3:: $6.: $5: $22.: 2.: $2:
m2 m2. of 3 3. 3m 2. 2.: SN «.2. v.2 2.. .68» Sam
8...:
5.: $2: an: 86.: $2: Go: ﬁn: $5.: $6.: 5.:
m: SN 2: no 2. SN 2 $6 mam 6.8 6.2. :2 28» ~28
$6.: $2: 6:: 5.: $2: an: 5.: 2.: 28.: $2:
ma 5 w: o._ 6.2 3.... 3. 6.2 E can 4:. an as: 21:.
SN: E: $_.: am: :2: ANN: 2m: am: 6:: $2:
83 SS 62 S n: in S a: :2 n2. ”.9. 3 as» 3-8
Gm: $2: :6: Ga: :2: 5.: 86.: 5:: 3:: $2:
an 3. 3 mo 3. N2. 3 a: $2 n$ mom 3 as» 3-2
«2...:
2.: $2: $6.: as: so: an: 2.: 25.: so: 2.:
m2 SN 62 2 2: EN 2 n2 2% 6.2 a? 3 28» 60%
G»: 82: 2:: 3:: 8w: $6.: 8:: am: so: 62:
23 E 6: mo 5: EN 3. 62 5a in $3 3 as» 21:.
2.: 82: Go: Go: :3: 2m: 8:: 5.: as: $2:
86.. 23 N: no 0: 3m 2 3.. SN >3. 2... 2 48» 92;
co: 9:: 8...: $6.: 32: $6.: an: 86.: 22.: $2:
3 m2. 3.. 2 m: as. S. 2: 6.2 0.3. 2% 2 as; 3-8
.3...—
EOEPQ’ no: EUR-c?» no: nun—03 =02 EMS—c?» =0: EOE—35 no: EOE—c?» =02
2.323.830 hos—.55 bans: ..33m 39:...— ..38m 83:.— Eoﬁﬁdiﬁm zom EoE?EE?:e Z

 

Sac—Ecumom .25.—
..ouuoO can .825 .ow< .3 EoEmeEEH .3 53.5329

m6. 021%

122

.thﬁ .1... 55:? KEN u so: “a: w: H 23.222: =< .6: 2283030
3.5.28 8 Ewt 05 co saga—:50 :o 338 938 2 23:8 to. 2: co 33> Q

63...: 23 NWT: .54": u359w

 

 

 

9:2 :.2 :.3 ~.m N.:m :2 Q8. m._ ::m :.m. :.E w.m :.::_ :.:_ SN 2: Yam :.:. .88.
h.— _ m: m: :.: EN :.N :.: :: :: :.: 5.: :.: :8 5m m: :.: a; :_ 3x83 4:23".
:3 N2 ms m: YN Wm 5mm v: 5:” :.: 5.: ~._ Wmm :.: mi :.: Wm M: 3.8m 29E:
On :.: m: v.4. :: m: 3.— :: n2 :.: :.: v: m: :: :.: :6 v: m: 3.8m 0:25
~23 1m 2 N: Em: m.m mew :.: :.m m: :3 Q. 53” NN van m: m.:~ ed Evie—9:: tom
:5. :.: 3. N: m.w :3 Wm N: :._ :.: N._ :_ Wow. :.v m.m :: N.m m3 .:o§o_qE?:oZ
An :::N 2:. 3:—
:.::_ 5.: :3 m6 ~13 ma». :.::_ :.: 5mm :.: m5. Wm :.::_ :.N. Wmm m: :18. Wow 38,—.
«.2 :6 :.: :.: m.m win :.: :: m: :.: m: :: :a in :.: :.: Na :.m 3:83 >25:
:2 5.: :.: N: 2. :.N mém :: m.m~ m._ :.w N._ :.:o. v: :3 :.: Nd NN 830m Sgt:
:.m :.: :.: we :.: N: :.: :.: N._ of :.: v: :.: :.: m: N: m: m: 3.8m 6:25
2.3 :.N Z :: 22 md 83 m: N: m: :.:” m._ Ymm m._ m: m: 5.3 :.m Ego—:8“: tom
:.3. m.m :.v N: :.h N.:m w: :.: :.: :.: :.: w: New :2 Wm :: m6 m5 Evie—QEoéoz
Q 33 23 ma:—

23... 33.83 8.3:.— u:::.— dEm .93 23,—. 3335 3.2:...— u.3=m dam— ...Eo 23,—. 3:33 8.3....— o..::.— .:5m .93

£5... =3 .22 2.5.... . =5 .82 £5...— 25 .82

.3535 .32 2.3222.— =<

 

8.32—2.33: .23.—
..oeaoU 2:. 330m .3 3.33 5 5223:.
:.v 33:.

123

.NN...m u :0an ..:...N u :03. m:.md n $0323.05 ..< 3.. 0.830530 .m::. E 0&<

050.55.... $5.505

mats: $50.83. waste? .0: 0....3 >033 2.050 0... a. 9.83 .0: 000... 8.53 050.0: .. $0.. 000.... .20.. .00. o. 00 63.83.00 0. 33.0.3.0... c< 0:80.. 5.. :8. 000.
30.5.0.5 ..o 0:... 0... .0 :..—00:00 088 :. .093.an was: 35.0.2.0... 0... 550...? 00.5.0008 .. $0.55 .0 0:... 0... .0 3.0.0 .2295 :0 .0000: m. 0.53 :. 5.5.0.5....
0.5.8.8 0. ...wt 0... :o €2.52? .0 .8000 0.30. 0. 0.0.0. ...0. 0.... co 30> A0

48...... .05 mmqm. .54.... N00.2%

 

 

 

 

:.m :.:. :3 Wm... :.N ...: ... ma: ﬁn m.~. Q3 :.:: .08...
..:. :.. «SN :.3 N: :.: :.: w.:.. m: v... . ...w. :.mm 0.00.. Nm-:m
cs «..:. m.mN v.:v Om in w: mé: ad 0... . w.~. :.:: 0.00.. :73
w... 5w. :2 n. .m .... 5.. m: ..:: Nd v... wd. ...:N. 0.00.. :m-:m
...: :.mm 5.5 Q? :N ..N :.: ~.m: :.: Nd. :.: .1: 0.00.. :~-mN
a 88 E; 32
:.m. v.3 Non w. .v :.m m. m: «6: ..: :.: m... ...3. .08...
..w. N6. ER :.wm ..:. :.. :.. ..:: :.3 ...: :.:. :.wm 0.00.. Nw-:m
..m. n... :...N :.3 Wm m. ...: m.n: ..: ...: ..m. ..:: 0.00.. 91::
m... :.3 m.:~ v.3 :.m :.: m: :.:: v... w: :.:. :.:: 0.80.. :m-:m
:.m. w.:. ...:m wdm md :N m: «6: ...: v... :.3 :.mm 0.00.. :~-nm
.0 82 2: 82

.00.. .30... .30..

0.30.. ...2. 0.30.. ...2. 0.00.. ...2.
...... 08.. ...... .00 ..0....0 ... ..c.. 08.. ...... .00 50....0 :. ...... 000.. ...... .00 00....0 ...
...—9: 07. ... ...—03 ...—0.3 oz 5 ...—03 ...—9: 02 ... ...—03
5:33 :02 0.2—0.2.0... ..<
3:030:03. .05...

00.500 ...—0 0w< .3 ...—03 a. 5.5.3....

h... 03:...

124

Appendix Table 4.1
Distribution of Employment by Schooling Attainment, Sector and Gender
SUSENAS

 

Non-emp. Self Emp. Public Sector Private Sector Family Worker Observations
Men Women Men Women Men Women Men Women Men Women Men Women

 

Susenas 1993
0 year

1-3 years
4-6 years
7-9 years
10-12 years

13+ years

All individuals

Susenas 1997
0 year

1-3 years
4-6 years
7-9 years
10-12 years

13+ years

All Individuals

Susenas 2000 a)
0 year

1-3 years
4-6 years
7-9 years
10-12 years

13+ years

All Individuals

4.7 38.3 73.0 25.5 0.4 0.2 18.0 11.8 3.9 24.2 12,624 30,127
(0.19) (0.28) (0.40) (0.25) (0.06) (0.02) (0.34) (0.19) (0.17) (0.25)
2.4 38.9 71.2 25.5 0.8 0.2 22.3 10.5 3.3 24.9 20,912 26,628
(0.11) (0.30) (0.31) (0.27) (0.06) (0.03) (0.29) (0.19) (0.12) (0.26)
2.5 45.7 65.2 23.2 3.2 0.3 24.9 8.2 4.3 22.7 44,071 42,541
(0.07) (0.24) (0.23) (0.20) (0.08) (0.03) (0.21) (0.13) (0.10) (0.20)
4.6 57.5 50.5 21.1 10.5 1.8 30.3 6.2 4.1 13.5 15,057 11,735
(0.17) (0.46) (0.41) (0.38) (0.25) (0.12) (0.37) (0.22) (0.16) (0.31)
6.1 45.9 27.5 11.0 32.6 25.4 30.4 11.4 3.5 6.4 20,369 13,036
(0.17) (0.44) (0.31) (0.27) (0.33) (0.38) (0.32) (0.28) (0.13) (0.21)
14.1 33.8 11.3 5.4 45.6 37.6 26.7 20.5 2.3 2.7 6,388 3,346
(0.44) (0.82) (0.40) (0.39) (0.62) (0.84) (0.55) (0.70) (0.19) (0.28)
4.2 43.3 55.9 22.3 10.7 39 25.4 10.0 3.8 20.5 119,421 127,413

(0.06) (0.14) (0.14) (0.12) (0.09) (0.05) (0.13) (0.08) (0.06) (0.11)

6.3 42.9 70.7 23.3 0.7 0.2 18.4 10.6 4.0 23.0 9,280 22,510
(0.25) (0.33) (0.47) (0.28) (0.09) (0.03) (0.40) (0.21) (0.20) (0.28)
3.2 42.9 71.0 23.5 0.9 0.3 22.0 10.3 3.0 22.9 16,370 22,283
(0.14) (0.33) (0.35) (0.28) (0.07) (0.04) (0.32) (0.20) (0.13) (0.28)
3.1 49.6 64.0 21.3 2.8 0.4 26.3 8.4 3.9 20.3 45,566 47,341
(0.08) (0.23) (0.22) (0.19) (0.08) (0.03) (0.21) (0.13) (0.09) (O.l8)
5.3 61.1 51.3 18.5 8.4 1.1 31.0 6.6 4.0 12.8 18,106 14,841
(0.17) (0.40) (0.37) (0.32) (0.21) (0.08) (0.34) (0.20) (0.15) (0.27)
6.7 51.5 30.3 11.1 26.6 18.6 32.9 11.8 3.5 7.0 26,034 18,199
(0.16) (0.37) (0.28) (0.23) (0.27) (0.29) (0.29) (0.24) (0.11) (0.19)
12.1 32.7 12.4 4.9 46.0 39.3 27.8 20.5 1.7 2.6 8,095 5,048
(0.36) (0.66) (0.37) (0.30) (0.55) (0.69) (0.50) (0.57) (0.14) (0.22)
5.0 48.2 53.1 19.7 11.0 4.5 27.3 9.9 3.6 17.8 123,451 130,222
(0.06) (0.14) (0.14) (0.11) (0.09) (0.06) (0.13) (0.08) (0.05) (0.11)
7.3 41.4 70.0 23.3 19.3 12.2 3.4 23.2 8,455 18,771
(0.28) (0.36) (0.50) (0.31) (0.43) (0.24) (0.20) (0.31)
4.1 41.9 69.8 23.9 23.4 11.3 2.7 22.9 14,308 19,901
(0.17) (0.35) (0.38) (0.30) (0.35) (0.22) (0.14) (0.30)
3.9 48.4 64.0 21.3 28.1 9.7 4.0 20.6 43,812 47,289
(0.09) (0.23) (0.23) (0.19) (0.21) (0.14) (0.09) (0.19)
6.4 58.4 52.7 18.9 36.7 9.2 4.1 13.6 19,351 16,578
(0.18) (0.38) (0.36) (0.30) (0.35) (0.22) (0.14) (0.27)
9.0 53.2 34.9 12.7 52.6 26.4 3.4 7.6 28,439 20,650
(0.17) (0.35) (0.28) (0.23) (0.30) (0.31) (0.11) (0.18)
12.5 32.6 17.7 6.2 67.4 58.0 2.4 3.2 9,023 6,180
(0.35) (0.60) (0.40) (0.31) (0.49) (0.63) (0.16) (0.22)
6.3 47.7 53.2 19.6 36.8 15.2 3.6 17.5 123,388 129,369
(0.07) (0.14) (0.14) (0.11) (0.14) (0.10) (0.05) (0.11)

 

Source: Susenas 1993, 1997, and 2000.
a) Public and Private sector workers are merged into one category.
Estimates are in percentage.

125

 

Table 5.1
Hausman-McFadden Tests

for Independence of Irrelevant Alternative Property

of Multinomial Sector Choice Model

 

Speciﬁcation that controls for

Non-linear Own and

Non-linear Own

 

 

Non-linear Own , Schooling, Parental
Omitted Category Schooling Parental Schooling Schooling and
(base) Residency (full)
X2 p -value 12 p -value X2 p -value
IFLSl
Men
Self Employment -15.74 - -15.63 - -287.15 -
Public Sector -2.4O - 65.94 0.093 -1.46 -
Private Sector 51.10 0.076 -8.92 - -7.32 -
Women
Self Employment 19.58 1.000 3.34 1.000 1.52 1.000
Public Sector 1.16 1.000 0.31 1.000 3.70 1.000
Private Sector 4.14 1.000 -10.94 - 22.32 1.000
Family Worker -6.47 - -1.51 - 1.13 1.000
IFLSZ
Men
Self Employment 1.50 1.000 12.84 1.000 278.58 0.000
Public Sector 0.23 1.000 1.90 1.000 2.98 1.000
Private Sector 4.71 1.000 -2.63 ~ -56.11 -
Women
Self Employment -3.99 - 10.79 1.000 -0.42 -
Public Sector 0.01 1.000 0.79 1.000 0.17 1.000
Private Sector -6.54 - -0.68 - 3.15 1.000
Family Worker 6.03 1.000 86.61 0.169 0.80 1.000
IFLS3
Men
Self Employment -16.92 - -54.00 - 7.13 1.000
Public Sector 3.57 1.000 -20.72 - -12.77 -
Private Sector 7.49 1.000 -29.33 - -27.45 -
Women
Self Employment 1.75 1.000 -17.91 - -0.26 -
Public Sector 0.05 1.000 0.60 1.000 -0.26 -
Private Sector 4.28 1.000 -O.40 - -20.44 -
Family Worker 136.23 0.000 -0.51 - 4.31 1.000

 

Source: IFLSl, IFLSZ and IFLS3.
A negative value of x2 statistics indicates that Independence of Irrelevant Alternative property

holds (see text for details).

Table 5.2

Wald Tests for Combining Dependent Categories

of Multinomial Sector Choice Model

 

Speciﬁcation that controls for

Non-linear Own and

Non-linear Own

 

 

Non-linear Own Schooling, Parental
Omitted C3198”? Schooling Parental Schooling Schooling and
(base) Residency (full)
12 p -value X: p -value 12 p -value
IFLS]
Men
Self-Public 530.34 0.000 545.83 0.000 666.35 0.000
Self-Private 181.03 0.000 206.80 0.000 455.28 0.000
Self-Non-emp/Family Worker 161.01 0.000 179.04 0.000 274.97 0.000
Public-Private 276.05 0.000 286.21 0.000 409.01 0.000
Public-Non-empJFamily Worker 224.80 0.000 239.60 0.000 348.56 0. 000
Private-Non-empJFamily Worker 176.46 0.000 191.30 0.000 285.38 0. 000
Women
Self-Public 375.46 0.000 428.29 0.000 590.19 0.000
Self-Private 88.26 0.000 103.23 0.000 178.62 0.000
Self-Family Worker 136.10 0.000 163.99 0.000 334. 48 0.000
Self-Non-emp. 149.71 0.000 164.62 0.000 400 94 0.000
Public-Private 169.15 0.000 226.28 0.000 393.50 0.000
Public-Family Worker 306.01 0.000 401.76 0.000 466.56 0.000
Public-Non-emp. 378.36 0.000 454.33 0.000 664.12 0.000
Private-Family Worker 117. 37 0.000 141 9.4 0.000 333.56 0.000
Private-Non-emp. 88. 76 0.000 100. 70 0.000 225.16 0.000
Family Worker-Non-cmp. 140 85 0.000 170. 05 0.000 393.21 0.000
IFLSZ
Men
Self-Public 588.39 0.000 639.95 0.000 841.89 0.000
Self-Private 291.07 0.000 343.17 0.000 616.29 0.000
Self-Non-empJFamily Worker 410.65 0.000 452.38 0.000 557.32 0.000
Public-Private 497.93 0.000 522.67 0.000 680.12 0.000
Public-NonoempJFamily Worker 519.03 0. 000 545.18 0.000 633. 73 0.000
Private-Non-en'pJFamily Worker 366.06 0. 000 440. 27 0.000 579. 55 0.000
Women
Self-Public 290.53 0.000 358 33 0.000 505.92 0.000
Self-Private 252. 76 0.000 268.08 0.000 389.13 0.000
Self-Family Worker 82 92 0.000 104 42 0.000 284.64 0.000
Self-Non-emp. 207.38 0.000 224.73 0.000 442.44 0.000
Public-Private 267.61 0.000 312.98 0.000 474.76 0.000
Public-Family Worker 277.73 0.000 332.22 0.000 632.85 0.000
Public-Non-emp. 388.57 0.000 434. 42 0.000 629.02 0.000
Private-Family Worker 150. 05 0.000 179. 00 0.000 372.32 0.000
Private-Non-emp. 138.74 0. 000 146. 83 0.000 334.39 0.000
Family Worker-Non-emp. 152.53 0 000 164.54 0.000 568. 95 0.000
IFLS3
Men
Self-Public 602.17 0.000 61 1.74 0.000 765.46 0.000
Self-Private 347.1 1 0.000 385.20 0.000 560.70 0.000
Self-Non-empJFamiiy Worker 458.49 0.000 494.19 0.000 617.90 0.000
Public-Private 506.61 0.000 542.41 0.000 718. 22 0.000
Public-Non-empJFamily Worker 512.41 0.000 536. 69 0. 000 654. 42 0.000
Private-Non-emp/Famiiy Worker 299.24 0.000 302.19 0 000 382. 61 0.000
Women
Self-Public 303 61 0.000 308. 53 0.000 502.61 0.000
Self-Private 250. 57 0.000 287. 76 0.000 423.90 0.000
Self-Family Worker 133.51 0.000 161.52 0.000 343.42 0.000
Self-Non-emp. 266.52 0.000 276.52 0.000 437.77 0.000
Public-Private 316.42 0.000 353.36 0.000 479.08 0.000
Public-Family Worker 415.70 0.000 433.30 0.000 540.10 0.000
Public-Non-emp. 381.44 0.000 394.23 0.000 588. 59 0.000
Private-Family Worker 183.28 0.000 215.12 0.000 455. 84 0.000
Private-Non-emp. 143.49 0.000 164.75 0.000 353.64 0.000
Family Worker-Non-ernp. 213.65 0.000 239.30 0.000 555.97 0.000

 

Source: IFLS], lFLSZ and IFLS3.

127

Table 5.3A

Multinomial Logit for Sector Choice Model:

The Effects of Non-linear Own Schooling and Parental Schooling (Base Speciﬁcation)

 

 

 

 

iFLSl
Men Women
Sci? Public Private Seﬁ Public Private Family
Emp. Sector Sector Emp. Sector Sector Worker
Own Schooling
1-3 years 0.109 -0.249 -0.059 0.355 ”‘ 0.419 -0.054 0.167
(0.40) (0.51) (0.20) (2.94) (i. 13) (0.37) (1.33)
4-6 years 0.021 0.964 ” -0.008 0.101 -0.011 -0.673 ”‘ -0.339 "
(0.08) (2.25) (0.03) (0.84) (0.03) (4.45) (2.07)
7-9 years -0.593 ‘ 1.347 ”‘ -0.254 -0.049 1.106 ’ -0.784 “‘ -0.501 ”
(1.93) (2.91) (0.81) (0.29) (1.83) (3.72) (2.14)
10-12 years -i.043 '" 2.801 ”‘ -0.024 —0.163 3.917 “" 0.023 -1.340 ”‘
(3.14) (6.00) (0.07) (0.89) (6.38) (0.11) (3.62)
13+ years -1.843 ‘” 2.846 “‘ -0.488
(4.42) (5.36) (1.27)
Father's Schooling
Some Elementary 0.0001 -0.273 -0. 144 0.118 0.459 -0.237 0.042
(0.00) (0.96) (0.53) (1.00) (1.47) (1.40) (0.29)
Completed Elementary 0312 -0.304 -0.357 0.224 ‘ 0.387 0.329 ‘ —0.113
(1.13) (1.03) (1.26) (1.74) (1.13) (1.79) (0.69)
Secondary/'i‘ertiary -0.256 -0.466 -0.364 -0.111 -0.002 -0.034 -0.639 “
(0.73) (1.25) (1.02) (0.52) (0.01) (0.16) (2.08)
Missing ~0.476 ' -0.933 “' -0.284 -0002 0.042 0.225 -0.293 ‘
(1.81) (2.91) (1.10) (0.02) (0.13) (1.34) (1.74)
Mother's Schooling
Some Elementary 0.284 0.860 ‘" 0.627 ” -0.137 0.021 0.422 " 0.014
(0.94) (2.65) (2.02) (1.07) (0.08) (2.38) (0.09)
Completed Elementary/ 0127 0.191 0.312 -0.231 0.273 0.109 -0.374 ‘
Secondary/Tertiary (0.39) (0.57) (0.97) (1.60) (1.14) (0.60) (1.86)
Missing -0.213 0.093 -0. 130 -0.112 0.147 -0.014 0.254
(0.81) (0.31) (0.50) (0.85) (0.60) (0.07) (1.64)
Age (spline)
25-29 years 0.094 0.240 " 0.134 0.063 0.451 “‘ 0.019 0.018
(0.85) (1.76) (1.18) (1.26) (3.96) (0.34) (0.33)
30-39 years 0.091 ” 0.141 ‘“ 0.062 0.072 ‘" 0.041 0.028 0.004
(2.29) (3.34) (1.53) (4.55) (1.52) (1.42) (0.20)
40-49 years -0.068 ‘ -0.039 -0.102 ”‘ 0.004 -0.039 -0.005 0.019
(1.78) (0.93) (2.62) (0.22) (1.18) (0.21) (1.01)
50-59 years -0.144 ‘” -0.204 “" -0.193 "‘ -0.052 ” -0.154 " -0.108 "‘ -0.097 ”‘
(3.79) (4.04) (4.60) (2.43) (2.27) (3. 15) (3.54)
H“ Composition and Business Assets
ii men, aged 20-49 -0.476 ”‘ -0.276 “ -0.381 ’” -0.353 ”‘ -0.331 “‘ -0.365 ”‘ -0.063
(4.17) (2.07) (3.17) (5.30) (2.97) (3.98) (1.01)
8 women. aged 20-49 0.073 0.279 0.1 11 -0.030 -0. 147 0.036 -0.454 ”‘
(0.46) (1.53) (0.67) (0.43) (1 . 10) (0.40) (4.56)
ii men. aged 50+ 0597 “ .0448 0338 -0.532 “‘ -0.301 -0.513 “‘ 0.363 ‘°‘
(2.22) (1.37) (1.22) (4.86) (1.32) (3.69) (3.80)
if women. aged 50+ -0.204 -0.123 .0135 0.220 “ 0.606 ‘“ 0.221 -0.406 '“
(1.30) (0.63) (0.82) (2.07) (2.82) (1.50) (2.75)
Business Assets(million) 0.512 ”‘ 0.213 -0.957 ” 0.057 ‘ 0.071 " -0.523 0.096 ”
(2.79) (1.04) (1.96) (1.69) (1.68) (1.48) (2.53)
Constant 0.660 -8.089 “ -1.010 -2.710 " -17.495 “‘ -1.312 -0.946
(0.22) (2.12) (0.33) (1 .86) (5.48) (0.83) (0.59)
Wald test
Own Schooling 433.34 235.40
(0.000) (0.000)
Fathefs Schooling 19.50 28.33
(0.077) (0.029)
Mother‘s Schooling 22.34 24.73
(0.008) (0.016)
Parental Schooling 44.13 51.10
(0.002) (0.005)
Age 115.74 110.57
(0.000) (0.000)
HH Composition 33.33 117.54
(0.001) (0.000)
HH Composition and Assets 41.97 124.65
(0.000) (0.000)
Pseudo 8’ 0.168 0.099
Observations 4.275 5,355
Source: 1FLSi.

Base category for men is nonparticipation and family worker sectors, and for women is nonparticipation. Month of interview dummy
variables are included in the regressions but are not reported. Omitted category for own and parental schooling is no schooling. Adjustments
are made due to small cell size for some variables for women: 10+ years for own schooling. Standard errors are robust to clustering at the
community level and heteroskedasticity. Absolute t-statistics are in parentheses. Signiﬁcant at the 1%(""). 5°/o(”) and 10%(‘) indicated.
p -values for Wald test are in parentheses.

128

 

Table 5.33

Multinomial Logit for Sector Choice Model:

The Effects of Non-linear Own Schooling and Parental Schooling (Base Speciﬁcation)

 

 

 

 

li-‘LSZ
Men _ _ Women
SeiF Public Private Self Public IrFP vate Family
Emp. Sector Sector Emp. Sector Sector Worker
Own Schooling
1-3 years 0.305 ‘ -0.549 0.237 0.311 ” 0.241 -0.127 0.067
( 1.67) (1.24) (1.16) (2.59) (0.46) (0.95) (0.42)
4-6 years 0.353 ” 0.686 ” 0.374 ‘ 0.222 ‘ 0.074 -0.567 ”‘ -0.136
(2.02) (2.21) (1.90) (1.89) (0.17) (4.36) (0.92)
7-9 years 0.078 1.484 ”’ 0.315 -0.155 1.559 ‘” -0.869 ”’ -0.468 "
(0.37) (4.32) (1.37) (0.98) (2.85) (5.17) (2.23)
10-12 years 0.143 3.103 “‘ 0.710 “‘ -0.056 4.348 “‘ —0.031 -O.670 ‘“
(0.73) (10.02) (3.25) (0.34) (7.54) (0.20) (2.65)
13+ years 0654 " 3.198 “‘ 0.080
(2.35) (9.44) (0.29)
F ather's Schooling
Some Elementary -0285 -0.158 -0.087 0.193 ‘ 0.451 0.130 -0.025
(1.59) (0.61) (0.50) (1.74) (1.49) (1.03) (0.17)
Completed Elementary -0231 0.054 -0.056 0.256 “ 0.364 0.162 -0.226
(1.16) (0.22) (0.27) (2.12) (1.23) (1.14) (1.18)
Secondary/Tertiary .0740 "” -0.519 ‘ -0.517 “ 0.211 0.166 0.338 “ -0.787 “‘
(3.14) (1.85) (2.20) (1.25) (0.53) (1.98) (3.09)
Missing -O.475 ‘ -0.536 -0.077 0.108 -0.400 0.116 -O.316
(1.93) (1.63) (0.33) (0.78) (1.03) (0.68) (1.44)
Mother's Schooling
Some Elementary -0.042 0.238 -0029 0252 " -0.023 -0.084 0.068
(0.25) (1 . 10) (0.18) (2.23) (0.09) (0.62) (0.43)
Completed Elementary/ 0.092 0.438 ’ 0.467 ” -0.36i ‘” 0.387 ‘ 0.075 -0.219
Secondary/Tertiary (0.44) (1.69) (2.27) (3.03) (1.81) (0.52) (1 .24)
Missing 0.136 0.240 0.445 ‘ ~0.359 “ 0.265 0.038 0.160
(0.52) (0.71) (1.81) (2.53) (0.67) (0.22) (0.70)
Age (spline)
25-29 years 0.336 ”‘ 0.518 ‘” 0.262 "‘ 0.068 0.225 ” -0.051 -0.066
(5.89) (4.96) (5.03) (1.38) (2.09) (1.22) (1.00)
30-39 years 0.089 ”‘ 0.216 "‘ 0.055 " 0.093 ‘” 0.225 ”‘ 0.010 0.075 “‘
(3.54) (7.43) (2.19) (6.52) (8.71) (0.66) (3.69)
40-49 years 0.020 0.023 -0.026 .0004 -0.124 “‘ -0.032 ‘ -0.040 “
(0.75) (0.74) (1.00) (0.27) (3.60) (1.88) (2.23)
50-59 years -0.140 ‘” ~0.226 “‘ -0.177 ‘” -0.042 " 0.008 -0.114 “‘ -0.065 “
(4.88) (5.97) (6.21) (2.21) (0.15) (4.05) (2.38)
Hi! Composition and Business Assets
# men. aged 20-49 -0.313 "” -0.292 ‘" -0.228 ”‘ -0.228 "‘ -0.344 "‘ -0.176 ‘" -0.192 ""
(5.22) (3.67) (3.89) (4.85) (4.23) (3.41) (3.52)
it women. aged 20-49 -0.055 0.004 -0.019 «0.069 -0.144 0.079 ‘ -0.l8i ““
(0.81) (0.05) (0.28) (1.40) (1.52) (1.68) (2.69)
it men, aged 50+ 40.718 ‘” -0.655 “‘ .0601 ”‘ -0.301 "‘ -0.184 -0.104 0.379 ”"
(5.67) (3.83) (4.99) (3.83) (1.10) (1.13) (3.94)
# women. aged 50+ -0.464 “‘ -0.408 ”‘ -0.231 ” 0.124 0.277 ' 0.305 “‘ 0.046
(4.35) (3.34) (2.31) (1.52) (1.83) (3.59) (0.37)
Business Assets (million) 0.015 -0.022 -0.111 0.025 ‘ -0.015 -0.028 0.031 ‘
(1.02) (0.63) (1 .44) (1.70) (0.68) (0.82) (1.92)
Constant -7.419 ”" ~16.949 ”‘ -5.S77 ‘" -3036 " -12.479 ‘” 0.493 0.544
(4.67) (5.68) (3.88) (2.15) (4.08) (0.42) (0.30)
Wald test
Own Schooling 441.35 208.70
(0.000) (0.000)
Father’s Schooling 22.96 35.83
(0.028) (0.003)
Mother's Schooling 23.19 30.64
(0.006) (0.002)
Parental Schooling 88.75 75.59
(0.000) (0.000)
Age 413.17 309.71
(0.000) (0.000)
HH Composition 136.37 1 10.43
(0.000) (0.000)
HH Composition and Assets 139.64 117.45
(0.000) (0.000)
Pseudo R ’ 0.146 0.091
Observations 5,607 6,583
Source: lFLSZ.

Base category for men is nonparticipation and family worker sectors, and for women is nonparticipation. Month of interview dummy
variables are included in the regressions but are not reported. Omitted category for own and parental schooling is no schooling. Adjustments
are made due to small cell size for some variables for women: 10+ years for own schooling. Standard errors are robust to clustering at the
community level and heteroskedasticity. Absolute r-statistics are in parentheses. Signiﬁcant at the 1%(‘“), 5%(“) and 10°/o(‘) indicated.
p -values for Wald test are in parentheses.

129

 

Table 5.3C
Multinomial Logit for Sector Choice Model:
The Effects of Non-linear Own Schooling and Parental Schooling (Base Specification)

 

 

 

 

 

il-‘LSJ
_ _Men _ _ Women_
Self Public Private Self Public Private Family
Emp. Sector Sector Emp. Sector Sector Worker
Own Schooling
1-3 years 0.596 ” 0.772 0.644 “ 0.196 ‘ 0.494 -0.174 0.012
(3.00) (1.28) (3.03) (1.70) (0.88) (1.37) (0.08)
4-6 years 0.532 ” 1.793 ‘” 0.606 " -0.008 0.072 -0.437 ‘” .0362 ”
(3.16) (3.62) (3.00) (0.07) (0.11) (3.53) (2.69)
7-9 years 0.244 2.579 ‘” 0.506 “ -0.220 1.178 ” -1.083 ‘” -0.908 ‘"
(1.30) (5.14) (2.46) (1.61) (2.18) (7.03) (4.88)
10-12 years 0.069 3.954 ‘" 0.487 " .0330 4.376 ”‘ -0.345 ” -1.047 ”"
(0.36) (7.94) (2.32) (1.60) (8.17) (2.40) (5.04)
13+ years 0178 4.797 ”‘ 0.302
(0.74) (9.28) (1.22)
Father's Schooling
Some Elementary 0.175 0.178 0.100 0.157 0.286 0.168 0.138
(1.02) (0.77) (0.58) (1.47) (1.08) (1.28) (1.16)
Completed Elementary 0.041 0.1 13 0.074 0.074 0.240 0143 0286 "
(0.26) (0.56) (0.48) (0.78) (i .14) (1.26) (2.49)
Secondary/Tertiary -0.377 “ -0.205 -0.150 -0.115 0.008 0.081 -0.518 “
(1.99) (0.89) (0.82) (0.92) (0.03) (0.58) (2.69)
Missing -0.203 -0.599 ” -0.138 0.121 -0.477 0.139 -0.032
(1.07) (2.24) (0.69) (1.14) (1.48) (1.02) (0.21)
Mother's Schooling
Some Elementary -0.494 " -0.426 " -0.361 ” 0.049 0.024 0.003 .0001
(3.26) (2. 15) (2.43) (0.45) (0.12) (0.03) (0.00)
Completed Elementary/ -0.263 ‘ -0.151 0.121 -0.147 0.027 0.179 0264 "
Secondary/Tertiary (1.75) (0.82) (0.82) (1.53) (0.15) (1.61) (2.00)
Missing -0.119 -O.295 0.109 -0.428 ”‘ -0.578 -0.177 -0.379 ”
(0.56) (1.01) (0.51) (3.69) (1.64) (1.23) (2.28)
Age (spline)
25-29 years 0.221 ‘” 0.354 ‘” 0.141 ” 0.156 ”‘ 0.106 -0.011 0.088 ‘
(4.70) (3.97) (3.16) (3.67) (1.10) (0.31) (1.91)
30-39 years 0.039 ‘ 0.161 "‘ 0.010 0.067 ‘” 0.271 ”‘ 0.034 “ 0.015
(1.85) (6.25) (0.50) (5.12) (10.24) (2.34) (0.89)
40-49 years 0.040 ‘ 0.097 ”' 0.004 0.036 " -0.067 ” -0.040 " 0.001
( 1.76) (3.67) (0.17) (2.68) (2.23) (2.28) (0.05)
50-59 years -0.101 ”‘ -0.224 “‘ -0.174 ”‘ -0.075 ‘“ -0.050 .0099 ‘” «0.069 ”
(3.64) (5.55) (5.64) (4.04) (1.05) (4.09) (3.14)
HB Composition and Business Assets
# men, aged 20-49 .0273 ‘" -0.334 ”' -0.172 ’” -0.187 ‘” -0.264 ”" -0.152 "‘ -0.104 "
(6.01) (5.16) (3.66) (5.21) (3.52) (3.94) (2.55)
# women. aged 2049 -0.057 -0.051 -0.017 0.003 0.017 0.083 “ -0.156 “
(1.08) (0.79) (0.33) (0.09) (0.25) (2.16) (3.33)
ii men, aged 50+ 0692 “‘ -0.682 ‘” -0.491 “‘ -0.234 ” -0.107 -0.016 0.351 ”'
(5.98) (4.72) (4.48) (3.21) (0.66) (0.20) (4.60)
# women, aged 50+ -0.422 ‘” -0.399 ” -0.325 " 0.051 -0.030 0.217 ” -0.067
(4.70) (3.40) (3.46) (0.75) (0.22) (2.98) (0.79)
Business Assets(million) 0.024 -0.0001 -0.095 ‘ 0.010 ’ 0.016 -0.017 0.015 “‘
(1.52) (0.01) (1.71) (1.84) (1.48) (1.02) (2.48)
Wald test
Own Schooling 502.89 316.13
(0.000) (0.000)
Father's Schooling 18.21 42.95
(0.109) (0.000)
Mother’s Schooling 32.60 30.86
(0.000) (0.002)
Parental Schooling 72.49 92.83
(0.000) (0.000)
Age 403.05 370.83
(0.000) (0.000)
Hl-l Composition 210.80 110.60
(0.000) (0.000)
HH Composition and Assets 215.52 115.53
(0.000) (0.000)
Pseudo R ’ 0.138 0.084
Observations 7,145 7,716
Source: iFLS3.

Base category for men is nonparticipation and family worker sectors, and for women is nonparticipation. Month of interview dummy
variables are included in the regressions but are not reported. Omitted category for own and parental schooling is no schooling. Adj ustrnents
are made due to small cell size for some variables for women: 10+ years for own schooling. Standard errors are robust to clustering at the
community level and heteroskedasticity. Absolute t-statistics are in parentheses. Signiﬁcant at the 1%(‘”), 5°/o(”) and 10°/o(‘) indicated. p-
values for Wald test are in parentheses.

130

Multinomial Logit for Sector Choice Model:
Summary of the Effects of Own Schooling

Table 5.4

 

 

 

Men Women
Self Public Private Self Public Private Family

Employment Sector Sector Employment Sector Sector Worker
A. Cross Section Sample
Linear Own Schooling
IFLSl ~0.114 "" 0.265 "‘ -0.015 -0.027 " 0.411 "'" -0.024 -0.109 ‘"
iFLSZ -0.042 “‘ 0.297 "‘ 0.022 -0.017 0.469 “‘ 0.009 0074 ‘"
lFLS3 -0.062 ”" 0.306 ”‘ 0.000 0.032 ‘" 0.467 "'" -0.008 -0.113 “‘
Non-linear Own Schooling
IFLS]
1-3 years 0.061 -0.302 -0.088 0.363 ‘” 0.498 -0.060 0.165
4-6 years 0019 0.968 " -0.015 0.111 0.156 -0.686 ”" -0.391 ”
7-9 years -0.657 “' 1.351 '“” -0.250 -0.064 1.319 ” -0.788 “" -0.656 ‘”
10-12 years a) —1.135 "‘ 2.805 "“ -0.027 -0.250 4.117 ”‘ 0.059 -1.681 *"
13+ years -1.965 ”' 2.831 ‘" -0.472
[FLSZ
1-3 years 0.253 0.548 0.230 0.303 "" 0.319 -0.102 0.029
4-6 years 0.288 " 0.748 "" 0.389 "‘ 0.216 “ 0.246 -0.509 ”‘ -0.234
7-9 years -0.036 1.564 “" 0.317 -0.177 1.826 ”" -0.758 '” -0.665 ”"
10-12 years a) -0.021 3.174 ”" 0.705 "" -0.102 4.678 ”" 0.160 -l.052 ‘“
13+ years 0896 ”‘ 3.240 "" 0.062
lFLS3
1-3 years 0.566 ” 0.735 0.633 ” 0.189 0.498 -0.167 -0.024
4-6 years 0.499 ” 1.793 “" 0.617 " 0.001 0.163 0420 ” -0.443 ”
7-9 years 0.155 2.547 '” 0.515 " -0.246 "' 1.276 “‘ -1.041 ‘" -1.090 ""
10-12 years a) -0.246 3.883 “" 0.501 ” -0.328 ” 4.498 ““ -0.240 " -1.429 ”"
13+ years 0474 ” 4.688 ‘“ 0.325
B. Panel Sample
Linear Own Schooling
IFLS] -0.079 ” 0.341 "‘ 0.020 -0.027 " 0.435 *“ -0.018 -0.104 ‘"
lFL82 -0.023 0.351 ”‘" 0.045 -0.046 "‘" 0.470 ”‘ -0.048 ”" -0.082 ”“
lFLS3 -0.095 “" 0.327 "* -0.024 —0.055 "‘ 0.483 "" -0.073 "" -0.126 "'"
Non-linear Own Schooling
iFLSi
1-3 years 0.024 -0.210 0.050 0.335 ” 0.263 -0.104 0.146
4-6 years 0.367 1.843 ""' 0.471 0.191 0.084 -0.655 ‘" -0.339 ‘
7-9 years 0.244 2.831 "“ 0.806 0.016 1.201 ” -0.660 "" -0.704 ‘"
10-12 years a) —0.679 3.957 "" 0.451 -0.271 4.072 ‘” 0.170 -1.456 ‘”
13+ years -2.027 ”“ 3.569 “”" -0.297
IFLSZ
1-3 years 0.252 -0.667 0.358 0.229 1.044 ‘ 0.011 0.053
4-6years 0.499 " 1.058 “ 0.603 "" 0.089 0.655 -0.669 ”‘ -0.345 "
7-9 years -0.206 1.523 ”" 0.242 -0.310 ‘ 2.172 ”" -1.199 ‘” -0.786 “"
10-12 years a) -0.263 3.188 '" 0.515 -0.604 ‘” 4.981 "" -0.059 -0.996 ”‘
13+ years 0.028 4.505 ”" 1.378 "
IFLS3
1-3 years 0.243 -0.254 0.317 0.073 0.593 -0.256 -0.029
4-6 years 0.163 0.927 ‘ 0.314 -0.106 0.479 -0.559 " -0.564 ""
7-9 years 0361 1.789 " 0.162 -0.555 “ 0.930 -1.527 "" -1.087 ‘"
10-12 years a) -0.861 ” 2.957 "" -0.011 -0.608 " 4.428 ”‘ -0.570 “ -1.589 ‘"
13+ years -1.334 " 3.992 "“ -0.138

 

Source: Based on estimates of Appendix 5.1A, 5.lB, 5.1C, Appendix Table 5.2A, 5.28, 5.2C, Appendix Table 5.4A, 5.4B, 5.4C, and

Appendix Table 5.5A, 5.58, 5.5C.
a) 10+ years for women.

131

Multinomial Logit for Sector Choice Model:

Table 5.5

Summary of the Effects of Non-linear Own Schooling and Parental Schooling

 

 

 

Men Women
Self Public Private Self Public Private Family

Employment Sector Sector Employment Sector Sector Worker
A. Cross Section Sample
iFLSl
1-3 years 0.109 -0.249 -0.059 0.355 "" 0.419 -0.054 0.167
4-6 years 0.021 0.964 " -0.008 0.101 0.011 .0673 ”' -0.339 ”
7-9 years -0.593 ‘ 1.347 ‘" -0.254 -0.049 1.106 " -0.784 ”" -0.501 ”
10-12 years a) -1.043 "" 2.801 ‘” -0.024 -0.163 3.917 '"" 0.023 -l.340 ""
13+ years -1.843 ”’ 2.846 ”" -0.488
IFLSZ
1-3 years 0.305 ’ -0.549 0.237 0.311 “‘ 0.241 -0.127 0.067
4-6 years 0.353 " 0.686 ” 0.374 ' 0.222 ‘ 0.074 -O.567 "‘ -0.136
7-9 years 0.078 1.484 “" 0.315 -0.155 1.559 ”" -0.869 ‘" 0468 "
10-12 years a) 0.143 3.103 ‘" 0.710 “" -0.056 4.348 ”" -0.031 -0.670 ‘"
13+ years -0.654 " 3.198 ‘“ 0.080
lFLS3
1-3 years 0.596 ” 0.772 0.644 ” 0.196 ‘ 0.494 -0.174 0.012
4-6 years 0.532 " 1.793 “‘ 0.606 “' -0.008 0.072 -0.437 "‘ -0.362
7-9 years 0.244 2.579 ”‘ 0.506 " -0.220 1.178 “ -1.083 "‘ -0.908
10-12 years a) -0.069 3.954 “' 0.487 “ -0.230 4.376 "‘ -0.345 " -1.047
13+ years 0178 4.797 ‘" 0.302
B. Panel Sample
lFLSi
1-3 years -0.029 -0.264 -0.042 0.318 " 0.247 -0.103 0.125
4-6 years 0.318 1.733 "" 0.374 0.174 0.025 0.653 ”‘ -0.284
7-9 years 0.205 2.695 ”" 0.672 0.025 1.138 " —0.691 "" -0.564 "
10-12 years a) -O.687 3.833 ”" 0.331 -0.211 4.006 "" 0.071 -1.152 ”"
13+ years -2.103 "‘ 3.422 "‘ -0.521
lFLSZ
1-3 years 0.317 -0.602 0.381 0.236 0.911 0.002 0.101
4-6 years 0.524 * 1.023 " 0.556 " 0.098 0.439 -0.694 ”‘ -0.202
7-9 years 0120 1.498 "" 0.219 -0.274 1.915 "" -1.233 '" -0.530 "
10-12 years a) -0.133 3.184 ”‘ 0.528 -0.522 " 4.734 "‘ -0.115 -0.562 "
13+years 0.084 4.411 "” 1.285 "'
"71.83
1-3 years 0.178 -0.326 0.234 0.089 0.638 -0.265 0.029
4-6 years 0129 0.862 0.215 -0.110 0.468 -0.592 ” -0.450 “
7-9 years 0362 1.719 " 0.037 -0.544 ” 0.947 -1.566 "" «0.842 "‘
10-12 years a) -0.747 " 2.958 ‘” -0.116 -0.576 " 4.486 ”’ ~0.620 " -1.133 ”‘
13+ years -1.168 ” 4.001 "‘ -0.254

 

Source: Based on estimates of Table 5.3A, 5.3B, 5.3C, and Appendix Table 5.6A, 5.6B, 5.6C.

a) 10+ years for women.

132

 

Appendix Table 5.1A
Multinomial Logit for Sector Choice Model:
The Effects of Linear Own Schooling

 

 

 

 

 

iFLSl
Men Women
Self Public Private Self Public Private Family
Employment Sector Sector Employment Sector Sector Worker
Own Schooling -0.114 '” 0.265 ”‘ -0.015 -0.027 " 0.411 "‘ -0.024 -0.109 ”‘
(5.57) (9.68) (0.70) (2.28) (6.95) (1.36) (4.98)
Age (spline)
25-29 years 0.092 0.252 ' 0.136 0.068 0.414 “' 0.001 0.020
(0.85) (1.80) (1.23) (1.35) (3.62) (0.02) (0.35)
30-39 years 0.100 “ 0.140 ‘” 0.061 0.071 “‘ 0.040 0.028 0.005
(2.54) (3.40) (1.53) (4.57) (1.55) (1.40) (0.28)
40-49 years -0.070 ‘ -0.038 -0.103 “" 0.001 -0.024 0.001 0.013
(1.81) (0.93) (2.65) (0.06) (0.81) (0.04) (0.71)
50-59 years -0.142 ”‘ -0.203 “" -0.l91 ”‘ -0.057 ”‘ -0.128 “ -0.097 "‘ -0.107 ""
(3.74) (4.01) (4.58) (2.71) (1.98) (2.88) (3.84)
HH Composition and Business Assets
# men, aged 20-49 -0.470 "' -0.298 " -0.383 ”‘ -0.337 ‘” -O.349 '“ ~0.391 ”' -0.056
(4.23) (2.31) (3.23) (5.13) (3.32) (4.20) (0.91)
# women, aged 20—49 0.054 0.259 0.099 -0.036 -0.211 0.032 -0.444 ”"
(0.34) (1.44) (0.59) (0.51) (1 .57) (0.34) (4.44)
# men, aged 50+ -0.673 " -0.570 ‘ -0.374 -0.522 “" -0.347 -0.522 “" 0.350 ‘”
(2.51) (1.72) (1.34) (4.78) (1.62) (3.80) (3.62)
# women, aged 50+ -0.246 -0.186 -0.179 0.187 ‘ 0.518 ”‘ 0.190 -0.341 ”
(1.60) (1.01) (1.10) (1.90) (2.68) (1.47) (2.41)
Business Assets (million) 0.535 ”‘ 0.251 -0.976 " 0.056 0.076 -0.533 0.091 "
(2.86) (1.20) (1.94) (1.63) (1 .64) (1.47) (2.32)
Month of Interview
October 0158 0.162 0.240 -0.128 0.393 0.034 -0.5 88
(0.50) (0.45) (0.82) (0.61) (0.98) (0.16) (1.36)
November -0.172 0.501 0.211 0.366 “ 0.924 “ 0.364 ‘ -0.197
(0.57) (1.57) (0.73) (1.74) (2. 17) (1.66) (0.52)
December 0.103 0.157 0.042 0.345 0.474 0.211 0.669 ‘
(0.35) (0.53) (0.14) (1.62) (1.18) (0.98) (1.75)
January 0.390 0.885 ” 0.487 -0.190 0.106 -0.361 0.100
(1.04) (2.27) (1.46) (0.71) (0.23) (1.20) (0.21 )
Constant 0.845 -9.067 ” -1.108 -2.614 ‘ -17.574 ‘” -1.019 -0.876
(0.28) (2.33) (0.37) (1.81) (5.44) (0.64) (0.55)
Wald test
Age 119.86 103.28
(0.000) (0.000)
H11 Composition 36.58 1 14.44
(0.000) (0.000)
HH Composition and Assets 44.52 120.71
(0.000) (0.000)
Month of Interview 20.43 49.07
(0.059) (0.000)
Pseudo 8’ 0.161 0.087
Observations 4,275 5,355
Source: lFLSl.

Base category for men is nonparticipation and family worker sectors, and for women is nonparticipation. Omitted category for month of interview
is August/September. Standard errors are robust to clustering at the community level and heteroskedasticity. Absolute t-statistics are in
parentheses. Signiﬁcant at the l°/o(“‘). 5%(“) and 10°/o(') indicated. p -va1ues for Wald test are in parentheses.

133

Appendix Table 5.13
Multinomial Logit for Sector Choice Model:

The Effects of Linear Own Schooling

 

 

 

 

iFLSZ
Men Women
Self Public Private Self Public Private Family
Employment Sector Sector Employment Sector Sector Worker
Own Schooling -0.042 "‘ 0.297 ”‘ 0.022 -0.017 0.469 ‘” 0.009 -0.074 “‘
(3.10) (14.75) (1.52) (1.45) (9.79) (0.73) (4.43)
Age (spline)
25-29 years 0.326 ‘“ 0.553 “‘ 0.264 “‘ 0.070 0.284 ” -0.033 -0.068
(5.76) (5.20) (5.11) (1.44) (2.56) (0.81) (1.04)
30-39 years 0.087 ‘“ 0.206 “‘ 0.052 ” 0.096 ‘” 0.214 "‘ 0.006 0.076 ”‘
(3.48) (7.01) (2.1 1) (6.68) (8.38) (0.43) (3.71)
4049 years 0.026 0.024 -0.025 -0.009 -0.111 ‘“ -0.032 ‘ -0.038 “
(0.98) (0.78) (0.99) (0.65) (3.57) (1.89) (2.11)
50-59 years -0.136 ‘“ -0.219 '“ -0.169 ”' -0.050 ”‘ 0.019 -0.101 ‘“ —0.069 "
(4.80) (5.89) (6.01) (2.68) (0.35) (3.56) (2.54)
Hii Composition and Business Assets
ii men, aged 20-49 .0314 ”‘ -0273 ”‘ -0.212 '” -0.223 ‘" -0.321 “‘ ~0.l86 ’” -0.l99 ”‘
(5.22) (3.50) (3.61) (4.77) (4.17) (3.58) (3.64)
# women, aged 20-49 -0.078 0.016 -0.024 .0077 .0207 “ 0.105 ” ~0.195 ‘”
(1.12) (0.20) (0.35) (1 .57) (2.40) (2.26) (2.92)
ii men, aged 50+ -0.724 "‘ -0.685 “‘ -0.607 ‘” 0283 ”‘ .0218 -0.072 0.385 ”‘
(5.71) (4.00) (5.03) (3 .63) (1.27) (0.80) (3.97)
# women, aged 50+ -0.487 ‘” -0.444 ‘“ -0.29l ”‘ 0.136 ‘ 0.172 0.328 ”‘ -0.049
(4.59) (3.59) (2.92) (1.67) (1.1 1) (4.02) (0.40)
Business Assets (million) 0.014 -0.027 -0.121 0.024 " -0.021 -0.030 0.030 "
(0.97) (0.84) (1 .50) (1.70) (0.89) (0.86) (1 .86)
Month of interview
October 0.059 0.060 -0.236 0.205 0.480 ‘ 0.117 0.247
(0.31) (0.23) (1.19) (1.29) (1.85) (0.63) (0.80)
November 0.125 0.343 -0.089 0.081 0.805 “‘ 0.116 0.269
(0.65) (1.23) (0.44) (0.51) (2.62) (0.58) (0.83)
December 0017 -0.074 0.016 0.153 0.296 0.201 0.342
(0.08) (0.26) (0.08) (0.78) (1.1 1) (0.99) (0.99)
January -0.295 0.009 -0.083 -0.550 ‘" -0.049 0.068 -0.235
(1 . 1 1) (0.03) (0.32) (2.85) (0.15) (0.34) (0.69)
Constant -6.936 “‘ -18.683 ‘” ~5.358 ”‘ -2.935 “ -15.489 ”‘ -0.242 0.613
(4.45) (6.16) (3.77) (2.1 l) (4.87) (0.21) (0.34)
Wald test
Age 407.81 289.49
(0.000) (0.000)
Hl-l Composition 149.21 126.63
(0.000) (0.000)
1111 Composition and Assets 154.34 133.66
(0.000) (0.000)
Month of interview 16.60 31.44
(0.165) (0.012)
Pseudo R’ 0.134 0.083
Observations 5.607 6,583
Source: lFLSZ.

Base category for men is nonparticipation and family worker sectors. and for women is nonparticipation. Omitted category for month of interview
is August/September. Standard errors are robust to clustering at the community level and heteroskedasticity. Absolute t-statistics are in

parentheses Signiﬁcant at the 1%(‘”), 5°/o(“) and 10%(‘) indicated. p-vaiues for Wald test are in parentheses.

134

Appendix Table 5.1C
Multinomial Logit for Sector Choice Model:

The Effects of Linear Own Schooling

 

 

 

 

lFLS3
Men Women
Self Public Private Self Public Private Family
Employment Sector Sector Employment Sector Sector Worker
Own Schooling -0.062 ‘” 0.306 “‘ 0.000 -0.032 “‘ 0.467 '” -0.008 -0.113 ‘”
(4.90) (17.06) (0.03) (3.52) (14.25) (0.75) (8.22)
Age (spline)
25-29 years 0218 ’" 0.358 “‘ 0.141 “' 0.148 “‘ 0.148 0.001 0.076 ‘
(4.67) (3.95) (3.12) (3.53) (1.42) (0.04) (1.68)
30-39 years 0.042 “ 0.162 ‘” 0.011 0.070 “" 0.274 ‘” 0.035 ‘° 0.016
(2.03) (6.28) (0.55) (5.27) (9.88) (2.43) (0.98)
40.49 years 0.047 “ 0.102 ”‘ 0.007 0.033 “ -0.045 -0.040 “ 0.000
(2.07) (3.86) (0.31) (2.47) (1 .61) (2.34) (0.00)
50-59 years -0.100 ‘” -0.226 ‘“ -0.l76 ”‘ -0.083 "" -0.068 -0.089 ”" -0.075 “
(3.65) (5.78) (5.73) (4.53) (1.38) (3.73) (3.37)
HH Composition and Business Assets
8 men, aged 20-49 0274 ‘“ -0.338 '" -0.166 ” -0.185 “' -0.200 " -0.160 ”" -0.107 “
(5.85) (5.14) (3.42) (5.19) (2.59) (4.15) (2.63)
#1 women, aged 20-49 -0.073 -0.076 -0.026 -0.003 -0.041 0.093 ” -0.168 “"
(1.39) (1.17) (0.49) (0.10) (0.56) (2.49) (3.57)
ii men, aged 50+ -0.708 ”‘ -0.707 ‘“ -0.496 ‘” ~0.226 " 0.126 0.001 0.358 “‘
(6.14) (4.86) (4.48) (3.10) (0.73) (0.02) (4.71)
11 women. aged 50+ -0.431 ”‘ -0.407 “ -0.344 '" 0.076 -0.048 0.242 ” -0.034
(4.76) (3.40) (3.61 ) (1 .13) (0.32) (3.34) (0.40)
Business Assets (million) 0.022 ‘ -0.005 .0099 ‘ 0.010 ‘ 0.010 -0.017 0.015 ”
(1.68) (0.33) (1.71) (1.93) (0.90) (1.01) (2.66)
Month of interview
October -0.023 -0.184 -0.042 0.326 “ 0.466 0.386 ” 0.230
(0.15) (0.76) (0.24) (2.38) (1.57) (2.50) (1.04)
November 0.006 -0.229 -0.124 0.404 ” 0.363 0.277 ‘ 0.369
(0.04) (0.97) (0.73) (3.04) (1.41) (1.77) (1 .61)
December -0.095 -0.115 0.139 0.434 ” 0.252 0.271 0.286
(0.56) (0.48) (0.70) (2.87) (0.87) (1.50) (1.17)
January 0.047 -0.313 0.397 " 0.246 ' -0.273 0.432 “ 0.317
(0.24) (1.00) (1.86) (1 .66) (0.84) (2.64) (1.31)
Constant -3.413 “ -13.086 “‘ -1.618 -4.820 ‘” -12.520 ”" -0.984 -2.420 '
(2.67) (5.10) (1.30) (4.12) (4.17) (1.01) (1.91)
Wald test
Age 413.46 331.29
(0.000) (0.000)
1111 Composition 222.38 1 16.43
(0.000) (0.000)
HH Composition and Assets 227.89 121.69
(0.000) (0.000)
Month of interview 20.22 26.83
(0.063) (0.043)
Pseudo R ’ 0.130 0.083
Observations 7,145 7,716
Source: lFLS3.

Base category for men is nonparticipation and family worker sectors. and for women is nonparticipation. Omitted category for month of interview
is June/July. Standard errors are robust to clustering at the community level and heteroskedasticity. Absolute r-statistics are in parentheses.

Signiﬁcant at the 1°/o(‘”), 5%(") and 10%(') indicated. p-values for Wald test are in parentheses.

135

Appendix Table 5.2A
Multinomial Logit for Sector Choice Model:

The Effects of Non-linear Own Schooling

 

 

 

 

lFLSi
Men Women
Self Public Private Self Public Private Family
Employment Sector Sector Employment Sector Sector Worker
Own Schooling
1-3 years 0.061 -0.302 -0.088 0.363 ‘” 0.498 -0.060 0.165
(0.23) (0.63) (0.31) (2.98) (1.38) (0.41) (1.25)
4-6 years -0.019 0.968 ” -0.015 0.111 0.156 -0.686 ‘” -0.391 "
(0.07) (2.34) (0.05) (0.93) (0.39) (4.61) (2.36)
7-9 years -0.657 “' 1.351 ‘” -0.250 -0.064 1.319 ” -0.788 “" -0.656 ‘”
(2.25) (3.03) (0.83) (0.40) (2.43) (3.86) (2.82)
10—12 years -1 135 "‘ 2.805 ‘“ -0.027 -0.250 4.117 ‘” 0.059 -l.681 ”‘
(3.76) (6.39) (0.09) (1.46) (7.68) (0.32) (4.73)
13+ years -1.965 ”‘ 2.831 ”' -0.472
(4.85) (5.58) (1.26)
Age (spline)
25-29 years 0.097 0.260 " 0.138 0.062 0.453 ‘“ 0.014 0.014
(0.88) (1 .91) (1.22) (1.22) (3.95) (0.24) (0.25)
30-39 years 0.094 " 0.137 "‘ 0.062 0.073 '" 0.036 0.028 0.007
(2.39) (3.29) (1.53) (4.66) (1.37) (1.41) (0.40)
40-49 years -0.068 ‘ -0.036 -0.103 “‘ 0.005 -0.038 -0.005 0.016
(1 .77) (0.87) (2.66) (0.32) (1.18) (0.20) (0.89)
50-59 years -0.140 “‘ -0.199 ”" -0.191 "” -0.052 “ -0.148 " -0.106 “‘ -0.102 ”‘
(3.68) (3.98) (4.59) (2.42) (2.20) (3.12) (3.73)
HH Composition and Business Assets
# men, aged 20-49 -0.476 ”‘ -0.284 " -0.380 ’” -0.351 ’" -0.322 ‘” -0.368 “‘ -0.071
(4.28) (2.18) (3.22) (5.29) (2.86) (3.99) (1.14)
# women. aged 20—49 0.039 0.248 0.090 -0.036 -0.152 0.026 -0.439 ‘”
(0.25) (1.40) (0.55) (0.52) (1.12) (0.29) (4.42)
# men. aged 50+ -0.653 ” -0.574 ‘ -0.373 -0.532 "’ -0.324 -0.519 ”‘ 0.347 ”"'
(2.46) (1 .78) (1.35) (4.86) (1.45) (3.75) (3.58)
# women. aged 50+ 0240 -0.147 0.177 0.192 ‘ 0.577 ‘" 0.183 -0.336 “
(1.58) (0.80) (1.09) (1.95) (3.15) (1.42) (2.40)
Business Assets (million) 0.534 ”‘ 0.239 -0.958 ' 0.057 ‘ 0.071 ‘ 0528 0.096 ”
(2.85) (1.15) (1.95) (1.67) (1.70) (1.49) (2.51)
Month of interview
October -0.158 0.164 0.226 -0.101 0.270 0.023 -0.564
(0.50) (0.46) (0.77) (0.48) (0.61) (0.1 1) (1.29)
November ~0.183 0.503 0.207 0.380 ‘ 0.764 0.355 ‘ 0185
(0.61) (1.57) (0.71) (1.81) (1.56) (1.65) (0.48)
December 0.085 0.141 0.018 0.363 ’ 0.323 0.187 0.682 ‘
(0.29) (0.47) (0.06) (1 .72) (0.74) (0.86) (1 .77)
January 0.415 0.941 ” 0.492 -0.139 -0.037 0407 0.148
(1 . 1 1) (2.47) (1.48) (0.52) (0.07) (1 .32) (0.31)
Constant 0.438 -8.676 ” -1.128 -2.684 ‘ -17.335 “‘ -1.184 -0.901
(0.15) (2.27) (0.3 7) (1.84) ( 5.40) (0.74) (0.56)
Wald test
Own Schooling 524.67 273.06
(0.000) (0.000)
Age 119.09 113.76
(0.000) (0.000)
HH Composition 37.08 113.29
(0.000) (0.000)
HI-l Composition and Assets 45.49 121.96
(0.000) (0.000)
Month of Interview 21.07 46.24
(0.049) (0.000)
Pseudo R ’ 0.163 0.095
Observations 4,275 5,355
Source: IFLSI.

Base category for men is nonparticipation and family worker sectors, and for women is nonparticipation. Omitted category for own schooling is

no schooling and for month of interview is August/September. Standard errors are robust to clustering at the community level and

heteroskedasticity. Absolute r-statistics are in parentheses. Signiﬁcant at the 1%(“‘). 5%(“) and 10%(") indicated. p-values for Wald test are

in parentheses.

136

Appendix Table 5.28
Multinomial Logit for Sector Choice Model:

The Effects of Non-linear Own Schooling

 

 

 

 

lFLSI
Men Women
Self Public Private Self Public Private Family
Employment Sector Sector Employment Sector Sector Worker
Own Schooling
1-3 years 0.253 -0.548 0.230 0.303 “ 0.319 -0.102 0.029
(1.38) (1.22) (1.12) (2.45) (0.60) (0.75) (0.18)
4-6 years 0.288 ‘ 0.748 " 0.389 " 0.216 ‘ 0.246 -0.509 ”" -0.234
(1 .66) (2.45) (1.98) (1 .78) (0.59) (3.86) (1.43)
7-9 years -0.036 1.564 “‘ 0.317 -0.177 1.826 ”‘ -0.758 ‘" -0.665 ‘”
(0.18) (4.72) (1.41) (1.12) (3.69) (4.57) (3.18)
10-12 years -0.021 3.174 "‘ 0.705 ‘“ -0.102 4.678 ”' 0.160 -1.052 ‘”
(0.11) (10.69) (3.28) (0.62) (9.40) (1.1 1) (4.30)
13+ years -0.896 "‘ 3.240 ‘" 0.062
(3.37) (10.13) (0.23)
Age (spline)
25-29 years 0330 ‘" 0.520 ‘“ 0.258 "‘ 0.071 0.227 ” -0.052 -0.066
(5.79) (4.98) (4.97) (1.44) (2.14) (1.25) (1.01)
30-39 years 0.092 ‘” 0.213 ”‘ 0.059 " 0.094 ‘” 0.219 ‘” 0.010 0.075 ‘”
(3.66) (7.34) (2.34) (6.58) (8.78) (0.71) (3.64)
40-49 years 0.022 0.020 -0.027 -0.005 -0.134 “‘ -0.034 "' -0.036 "
(0.85) (0.66) (1.04) (0.37) (3.92) (2.04) (2.04)
50-59 years 0136 ‘” -0.224 ”“ -0.171 "‘ -0.045 “ 0.012 -0.111 "“ -0.068 “
(4.78) (5.93) (5.98) (2.41) (0.22) (3.92) (2.48)
iii-l Composition and Business Assets
ii men, aged 2049 -0.325 ”‘ -0.291 ”‘ -0.224 ‘” -0.229 ”' -0.331 ‘” —0.170 “" -0.201 ‘"
(5.36) (3.65) (3.80) (4.88) (3.98) (3.28) (3.66)
111 women. aged 20-49 -0.067 0.000 -0.017 -0.075 -0.138 0.088 ‘ -0. I90 ‘”
(0.98) 0.00 (0.26) (1.52) (1.48) (1.85) (2.86)
ii men. aged 50+ -0.720 “‘" -0.670 “‘ -0.607 "' -0.286 ‘” -0.159 -0.101 0.389 ‘”
(5 .68) (3 .94) (5 .04) (3.68) (0.96) (1.10) (4.02)
it women. aged 50+ 0460 ‘“ -0.416 “‘ -0.267 ”’ 0.140 ‘ 0.223 0.294 ”’ -0.042
(4.32) (3.40) (2.68) (1.75) (1.49) (3.48) (0.35)
Business Assets (million) 0.017 -0.023 -0.118 0.025 ‘ -0.014 -0.030 0.030 ‘
(1.12) (0.61) (1.46) (1.71) (0.66) (0.83) (1.85)
Month of interview
October 0.051 0.056 -0.243 0.208 0.437 ‘ 0.129 0.249
(0.27) (0.21) (1.22) (1.31) (1.94) (0.69) (0.80)
November 0.093 0.337 -0.110 0.072 0.652 ” 0.129 0.266
(0.49) (1.21) (0.54) (0.46) (2.1 1) (0.64) (0.82)
December -0.028 -0.036 0.013 0.175 0.342 0.210 0.359
(0.13) (0.13) (0.06) (0.88) (1.47) (1.03) (1.03)
January -0.250 0.111 -0.040 .0530 ‘" -0.023 0.015 -0.206
(0.94) (0.33) (0.15) (2.74) (0.08) (0.08) (0.60)
Constant -7.431 ‘” ~17.015 ‘” ~5.416 “‘ -3.120 ” -12.401 '” 0.564 0.492
(4.69) (5.71) (3.77) (2.23) (4.11) (0.48) (0.27)
Wald test
Own Schooling 518.71 264.31
(0.000) (0000)
Age 423.95 313.13
(0.000) (0.000)
HH Composition 145.98 109.80
(0.000) (0.000)
1111 Composition and Assets 149.83 117.95
(0.000) (0.000)
Month of interview 15.70 26.76
(0.205) (0.044)
Pseudo R ’ 0.139 0.087
Observations 5.607 6.583
Source: "-132.

Base category for men is nonparticipation and family worker sectors, and for women is nonparticipation. Omitted category for own schooling is

no schooling and for month of interview is August/September. Standard errors are robust to clustering at the community level and

heteroskedasticity. Absolute r-statistics are in parentheses Significant at the 1%(‘"), 5°/o(”) and 10%(‘) indicated. p-values for Wald test are

in parentheses.

137

Appendix Table 5.2C
Multinomial Logit for Sector Choice Model:
The Effects of Non-linear Own Schooling

 

 

 

 

iFLS3
Men Women
Self Public Private Self Public Private Family
Employment Sector Sector Employment Sector Sector Worker
Own Schooling
1-3 years 0.566 " 0.735 0.633 ” 0.189 0.498 ~0.167 -0.024
(2.85) (1 .24) (3.00) (1 .64) (0.89) (1.28) (0.17)
4-6 years 0.499 " 1.793 ”‘ 0.617 ” 0.001 0.163 -0.420 " -0.443 “
(2.96) (3.67) (3.08) (0.01) (0.25) (3.36) (3.16)
7-9 years 0.155 2.547 *" 0.515 ” -0.246 ‘ 1.276 ” -1.041 "‘ -l.090 ”‘
(0.83) (5.15) (2.52) (1.86) (2.44) (6.76) (5.84)
1012 years 0246 3.883 ”‘ 0.501 " -0.328 " 4.498 "“ -0.240 ‘ -1.429 ‘“
(1.33) (8.01) (2.46) (2.43) (8.71) (1.73) (7.29)
13+ years -0.474 " 4.688 “" 0.325
(2.07) (9.45) (1.41)
Age (spline)
25-29 years 0.228 ‘“ 0.359 ”" 0.145 ” 0.151 "" 0.106 -0.013 0.082 ‘
(4.87) (4.03) (3.24) (3.57) (1.10) (0.36) (1.78)
30-39 years 0.043 " 0.164 ‘” 0.012 0.067 ’" 0.270 "’ 0.033 ” 0.015
(2.10) (6.42) (0.61) (5.10) (10.19) (2.26) (0.91)
40-49 years 0.043 ‘ 0.096 ”" 0.005 0.034 " -0.078 ” -0.041 “ 0.000
(1.90) (3 .65) (0.23) (2.58) (2.64) (2.40) (0.01)
50-59 years 0098 ”‘ -0.225 ‘” -0.174 ”‘ -0.079 "‘ -0.048 0.099 ”‘ -0.071 "
(3.53) (5.71) (5.65) (4.23) (1.02) (4.10) (3.21)
ill-l Composition and Business Assets
18 men, aged 20-49 -0.280 ‘” -0.335 ‘” -0.169 "‘ -0.|89 '” -0.256 “ -0.153 ”‘ -0.110 "'
(6.02) (5.16) (3.53) (5.30) (3.44) (3.97) (2.71)
11 women. aged 2049 -0.063 -0.054 -0.018 0.001 0.012 0.088 ” -0.163 “‘
(1.20) (0.84) (0.34) (0.02) (0.17) (2.30) (3.50)
# men. aged 50+ -0.703 ‘" -0.681 ‘” -0.494 ‘“ -0.233 ” -0.084 -0.013 0.352 ""
(6.10) (4.72) (4.47) (3.19) (0.51) (0.17) (4.62)
it women, aged 50+ 0424 ”" -0.402 ” -0.338 ”" 0.078 -0.023 0.221 ” -0.035
(4.70) (3.40) (3.57) (1.16) (0.17) (3.04) (0.41)
Business Assets (million) 0.023 0.000 -0.098 ‘ 0.010 " 0.017 ’ -0.018 0.015 ”
(1.61) (0.02) (1.72) (1.89) (1.65) (1.03) (2.55)
Month of interview
October -0.011 -0.138 -0.033 0.331 ” 0.404 0.397 " 0.238
(0.07) (0.57) (0.19) (2.41) (1.48) (2.57) (1.08)
November 0.012 .0220 -0.125 0.411 ” 0.404 ' 0.295 ‘ 0.379 ‘
(0.08) (0.93) (0.74) (3.08) (1 .72) (1 .86) (1 .66)
December -0.104 -0.096 0.130 0.437 “ 0.253 0.288 0.292
(0.61 ) (0.40) (0.66) (2.89) (0.97) (1.61) (1.20)
January 0.045 -0.282 0.390 ‘ 0.249 ‘ -0.069 0.433 " 0.319
(0.23) (0.91) (1 .82) (1.68) (0.24) (2.62) (1 .32)
Constant -4.319 " -13.265 ‘” -2.255 ‘ -5.018 ‘“ -9.432 ” 0.264 -2.662 ”
(3.36) (5.23) (1.81) (4.23) (3.43) (0.27) (2.07)
Wald test
Own Schooling 606.58 366.01
(0.000) (0.000)
Age 428.71 380.66
(0.000) (0.000)
1111 Composition 222.09 1 13.66
(0.000) (0.000)
1111 Composition and Assets 226.65 1 18.88
(0.000) (0.000)
Month of interview 19.97 25.00
(0.068) (0.070)
Pseudo R ’ 0.133 0.080
Observations 7,145 7,716
Source: iFLS3.

Base category for men is nonparticipation and family worker sectors, and for women is nonparticipation. Omitted category for own schooling is
no schooling and for month of interview is June/July. Standard errors are robust to clustering at the community level and heteroskedasticity.
Absolute t-statistics are in parentheses. Si gniﬁcant at the l%(""), 5°/o(“) and 10%(‘) indicated. p -values for Wald test are in parentheses.

138

Appendix Table 5.3A
Multinomial Logit for Sector Choice Model:

The Effects of Non-linear Own Schooling, Parental Schooling and Residency (Full Specification)

 

 

 

 

IFLS]
Men Women
Self Public Private Self Public Private Family
Employment Sector Sector Employment Sector Sector Worker
Own Schooling
1-3 years 0.104 -0.328 -0.140 0.402 ”‘ 0.535 .0026 0.137
(0.39) (0.67) (0.50) (3.45) (1.50) (0.18) (1.16)
4-6 years 0.134 0.928 ” -0.124 0.185 0.190 -0.688 “‘ -0.249 ‘
(0.48) (2.14) (0.44) (1.57) (0.39) (4.62) (1.74)
7-9 years -0.366 1.353 ”' -0.383 0.016 1.322 ‘ -0.817 ”‘ -0.406 '
(1.15) (2.84) (1.19) (0.10) (1.74) (3.93) (1.92)
10-12 years -0.699 " 2.912 ’” -0.028 -0.020 4.281 ”‘ -0.029 -1.111 ”‘
(2.08) (5.97) (0.08) (0.10) (5.62) (0.13) (3.28)
13+ years -1.384 '" 3.004 ”‘ -0.509
(3.28) (5.47) (1.31)
Father's Schooling
Some Elementary -0.159 -0.339 -0.258 0.039 0.501 -0.261 -0.142
(0.59) (1.15) (0.92) (0.32) (1.59) (1.57) (0.93)
Completed Elementary -0.382 -0.366 -0.472 ‘ 0.288 “ 0.598 -0.314 ‘ 0.044
(1.36) (1.26) (1.65) (2.08) (1.62) (1.69) (0.26)
Secondary/Tertiary -0.3 16 -0.579 -0.5 19 -0.068 0.245 0.023 -0.493
(0.87) (1.54) (1.43) (0.32) (0.68) (0.10) (1.51)
Missing -0.523 “ -0.940 ‘” -0.385 -0.016 0.166 -0248 -0.304 "
(1.97) (3.02) (1.49) (0.13) (0.51) (1.45) (1.66)
Mother's Schooling
Some Elementary 0.274 0.830 “ 0.615 ” 0131 -0.014 0.442 " -0.147
(0.88) (2.51) (1.97) (0.98) (0.06) (2.50) (0.85)
Completed Elementary/ 0015 0.208 0.280 -0.135 0.335 0.145 -0.249
Secondary/Tertiary (0.04) (0.62) (0.86) (0.90) (1 .28) (0.78) (1 . 10)
Missing -0.244 0.010 -0.159 -0.072 0.096 0.035 0.148
(0.93) (0.03) (0.62) (0.55) (0.37) (0.19) (0.82)
Age (spline)
25-29 years 0.119 0.242 ‘ 0.118 0.063 0.530 ‘” 0.018 0.026
(1.04) (1.73) (1.00) (1.26) (4.46) (0.31) (0.44)
30-39 years 0.103 " 0.153 ‘" 0.059 0.081 "‘ 0.054 ‘ 0.027 0.022
(2.53) (3.53) (1.43) (5.04) (1.84) (1.34) (1.23)
40-49 years 0068 ‘ -0.044 -0.097 “ 0.005 -0.035 0.000 0.003
(1.78) (1.07) (2.49) (0.30) (1.01) (0.01) (0.17)
50-59 years 0145 ‘" -0.199 “" -0.197 "‘ -0.046 "" -0.132 ‘ -0.108 “‘ -0.085 ’"
(3.71) (3.90) (4.65) (2.02) (1.86) (3.12) (2.72)
H“ Composition and Business Assets
ii men. aged 2049 -0.395 "‘ -0.234 ‘ -0.418 ”‘ -0.327 ‘” -0.335 ‘” -0.349 ”‘ 0.080
(3.40) (1.72) (3.42) (4.73) (2.79) (3.85) (1.04)
# women, aged 20-49 0.109 0.312 ‘ 0.112 0.027 0.016 0.075 -0.243 "
(0.68) (1 .73) (0.68) (0.39) (0.12) (0.83) (2.25)
ii men, aged 50+ -0.572 " -0.423 -0.440 —0.587 “" -0.389 ’ -0.587 ”‘ 0.441 '"
(2.06) (1.30) (1.58) (5.41) (1.75) (4.23) (3.89)
# women, aged 50+ 0.182 .0166 -0.181 0.224 " 0.571 ‘” 0.178 -0.276 ‘
(1.13) (0.86) (1.09) (2.04) (2.70) (1.21) (1.79)
Business Assets (million) 0.456 ”‘ 0.156 -0.734 " 0.040 0.076 ' -0.484 0.105 "
(2.79) (0.83) (1.83) (1.35) (1.67) (1.47) (2.58)
Province of Residence
North Summers 2.892 ‘" 1.054 0.282 1.445 ‘“ 1.270 ” 0.165 4.139 ‘”
(4.54) (1.34) (0.34) (3.01) (2.10) (0.31) (6.52)
North Sumatera " Urban -1.816 ‘” -0.093 -0.106 -0.940 ‘ 0.647 -0.486 -2.153 ”‘
(2.69) (0.09) (0.12) (1.92) (0.58) (0.76) (3.52)
West Surnatera 1.775 ’” 0.972 0.355 1.126 '“ 1.545 ”‘ 0.749 “ 3.300 ‘”
(3.75) (1.36) (0.67) (3.55) (2.88) (2.56) (6.86)
West Surnatera " Urban -1.656 " -0.023 0.097 -0.314 1.034 -0.090 -2.299 ‘°
(1.98) (0.02) (0.13) (0.88) (1.49) (0.24) (2.35)
South Surnatera 2.263 ‘” 0.557 -0.206 0.875 “ 0.910 -0.807 " 3.236 ”‘
(3.81) (0.83) (0.39) (2.36) (0.71) (1.71) (6.17)
South Surnatera " Urban «1.371 ‘ 0.634 0.181 -1.203 "‘ -0.822 0.348 -3.191 "‘
(1.81) (0.78) (0.26) (2.88) (0.63) (0.53) (4.46)
(continued)

139

Appendix Table 5.3A (continued)

Multinomial Logit for Sector Choice Model:

The Effects of Linear Own Schooling, Parental Schooling and Residency (Full Specification)

 

 

 

 

IFLSI
Men Wow
Self Public Private Self Public Private Family
Employment Sector Sector Employment Sector Sector Worker
Lampung 2.114 “‘ -0141 -0.884 ' 0.964 ”‘ 0.794 -0.451 4.377 ‘”
(4.61) (0.20) ( 1.78) (3.54) (0.90) (1.26) (7.16)
Lampung ‘ Urban -l.011 “' -0.607 0.557
(2.44) (0.84) (0.95)
West Java 2.266 ”‘ 1.612 ” 0.688 0.087 1.343 “ -0.289 2.563 ”’
(4.19) (2.42) (1.21) (0.32) (2.01) (0.92) (5.62)
West Java ‘ Urban -1.803 “' -0.948 -0.920 —0.187 —0.313 0.047 -1.630 "‘
(3.41) (1.61) (1.62) (0.68) (0.49) (0.16) (2.84)
Central Java 1.315 ‘f’ 0.988 ” -0.001 0.934 "' 1.812 ‘” 0.321 2.369 ”‘
(3.78) (2.59) (0.00) (3.58) (3.87) (1.20) (5.70)
Central Java ‘ Urban -0.149 0.186 0.532 -0.005 -1.082 ' 0.112 -0.805 “
(0.28) (0.30) (1.05) (0.02) (1.87) (0.32) (2.30)
Yogyakarta 1.072 " 1.093 ” -0.407 2.696 "‘ 3.360 ”‘ 1.929 ‘“ 4.633 ”'
(2.06) (2.03) (0.54) (8.19) (7.00) (3.10) (7.55)
Yogyakarta ‘ Urban -0.767 0.745 0.489 -0.990 “ -2.223 “‘ -0.617 -2.715 ‘”
(1.27) (1.16) (0.60) (2.46) (4.20) (0.87) (4.18)
East Java 0.984 " -0.161 0006 0.621 ” 1.071 ' 0.624 ” 1.507 ‘”
(2.57) (0.29) (0.01) (2.58) (1.66) (2.06) (3.83)
East Java ‘ Urban -0.424 0.371 «0.118 -0.137 —0.260 -0.300 -0.946 ”
(1 .01) (0.59) (0.26) (0.57) (0.42) (0.85) (2.24)
Bali 0.779 ' 0.352 -1.474 ‘“ 1.096 “‘ 1.898 "‘ 0.060 -0.451
(1.85) (1.03) (2.90) (4.14) (3.83) (0.15) (0.68)
Bali ‘ Urban 1.401 2.646 “ 2.873 “‘
(1.30) (2.06) (2.62)
WestNusa Tenggara 1.054 “ 0.892 ” -0.553 1.453 ”‘ 2.742 “‘ 0.328 3.274 ”"
(2.51) (2.33) (1.32) (5.41) (5.06) (1.41) (6.75)
West Nusa Tenggara " Urban -0.895 -0.236 -0.613 0.318 0.186 0.083 -1.336 ”
(1.17) (0.27) (0.86) (0.78) (0.20) (0.16) (2.23)
South Kalimantan 2.577 “ 2.221 ‘” —0.201 1.782 "' 2.337 ”‘ -0.318 3.685 ”‘
(2.53) (2.75) (0.17) (5.93) (3.88) (0.57) (7.52)
South Kalimantan ‘ Urban -0.780 -0.666 0.160 -0.733 “ -0.221 0.120 -3.442 ‘”
(0.62) (0.78) (0.12) (2.13) (0.34) (0.19) (2.88)
South Sulawesi 1.176 ” 0.639 -1.537 “ -0.059 1.935 ‘“ -1.521 ”‘ 1.654 ”‘
(2.45) (1 .04) (2.36) (0.20) (4.52) (4.31) (3.52)
South Sulawesi ‘ Urban -0.823 -0.586 0.343 0.281 -1.291 ‘ 0.499 -0.790
(1.31) (0.69) (0.48) (0.64) (1 .92) (0.74) (1.34)
Constant -1.712 -8.983 " -0.206 -3.164 ” -21.6l3 “‘ -1.027 -4.768 ‘”
(0.55) (2.30) (0.06) (2.15) (6.45) (0.64) (2.69)
Wald test
Own Schooling 360.67 234.14
(0.000) (0.000)
Father's Schooling 17.88 27.59
(0.1 19) (0.035)
Mother's Schooling 17.87 20.50
(0.037) (0.058)
Parental Schooling 37.09 55.42
(0.016) (0.002)
Age 126.93 123.83
(0.000) (0.000)
iii-i Composition 26.13 99.19
(0.010) (0.000)
iii-I Composition and Assets 36.74 114.66
(0.001) (0.000)
Residency 528.13 1 123.49
(0.000) (0.000)
Pseudo R ’ 0.222 0.174
Observations 4,275 5,355
Source: "-181.

Base category for men is nonparticipation and family worker sectors. and for women is nonparticipation. Month of interview dummy variables are
included in the regressions but are not reported. Omitted category for own and parental schooling is no schooling and for province of residence is
Jakarta. Adjustments are made due to small cell size for some variables for women: 10+ years for own schooling and no urban interaction terms
for the province of Lampung and Bali. Standard errors are robust to clustering at the community level and heteroskedasticity. Absolute :-
statistics are in parentheses. Signiﬁcant at the 1%(“‘), 5%(”) and 10°/o(‘) indicated. p -values for Wald test are in parentheses.

140

Appendix Table 5.33
Multinomial Logit for Sector Choice Model:
The Effects of Non-linear Own Schooling, Parental Schooling and Residency (Full Specification)

 

 

 

 

 

lFL82
Men Women
Self Public Private Self Public Private Family
Employment Sector Sector Employment Sector Sector Worker
Own Schooling
1-3 years 0.286 -0.617 0.121 0.370 ‘” 0.286 .0070 0.168
(1.53) (1.37) (0.58) (3.10) (0.57) (0.50) (1.08)
4-6 years 0.395 “ 0.658 ” 0.212 0.283 “ 0.154 -0.601 ‘” 0.062
(2.15) (2.04) (1.05) (2.52) (0.35) (4.51) (0.49)
7-9 years 0.266 1.601 ‘“ 0.174 -0.119 1.620 " -0.955 ‘” -0.264
(1.16) (4.35) (0.73) (0.75) (2.56) (5.60) (1.21)
10-12 years 0.317 3.141 ‘" 0.500 “ 0.007 4.499 “" -0.110 -0.378 ‘
(1.50) (9.47) (2.21) (0.04) (6.93) (0.69) (1.75)
13+ years 0436 3.233 ”‘ -0.183
(1.48) (8.79) (0.64)
Father's Schooling
Some Elementary -0.309 ‘ -0.146 -0. 196 0.127 0.438 0.023 -0.146
(1.69) (0.55) (1.09) (1.16) (1.45) (0.18) (1.07)
Completed Elementary -0.217 0.022 -0.202 0.227 ‘ 0.449 -0.026 -0.257
(1.09) (0.09) (0.98) (1.95) (1.53) (0.19) (1.38)
Secondary/1' ertiary -0.627 ‘“ -0.528 ‘ -0.658 ”‘ 0.242 0.270 0.189 -0.769 ‘”
(2.61) (1.85) (2.77) (1.44) (0.89) (1.12) (2.98)
Missing -0.420 ‘ -0.520 -0.180 0.085 -0.367 0.041 -0.345
(1.73) (1.55) (0.78) (0.62) (0.96) (0.24) (1.55)
Mother's Schooling
Some Elementary 0.006 0.298 -0.037 -0.189 -0.080 -0.023 0.052
(0.04) (1.43) (0.22) (1 .63) (0.29) (0.17) (0.32)
Completed Elementary/ 0.171 0.589 " 0.392 ‘ -0.193 0.478 ” 0.175 0.115
Secondaryfiertiary (0.78) (2.31) (1.90) (1.58) (2.21) (1.19) (0.64)
Missing 0.185 0.445 0.435 ‘ -0.270 ‘ 0.309 0.173 0.201
(0.71) (1.27) (1.76) (1.83) (0.86) (0.95) (0.88)
Age (spline)
25-29 years 0.347 '" 0.526 ‘" 0.264 ‘” 0.088 ’ 0.217 ” -0.055 -0.033
(6.02) (4.88) (4.99) (1.72) (1.99) (1 .33) (0.48)
30—39 years 0.097 ‘” 0.222 ”" 0.043 ' 0.101 ’” 0.245 ”‘ 0.007 0.098 “‘
(3.85) (7.68) (1.70) (6.84) (9.38) (0.45) (4.84)
40-49 years 0.021 0.026 -0.022 -0.003 -0.123 '” -0.029 ‘ -0.041 ”
(0.77) (0.84) (0.84) (0.22) (3.59) (1.69) (2.20)
50-59 years -0.133 '” .0226 “‘ -0.181 ”‘ -0.032 " 0.024 -0.111 ”' -0.058 “
(4.51) (5.78) (6.25) (1 .65) (0.40) (3.85) (2.00)
[iii Composition and Business Assets
# men, aged 20-49 0257 “‘ -0.267 ‘” -0.251 ‘” -0.184 ‘” «0.321 ‘“ -0.162 ”‘ -0.110 ‘
(4.32) (3.37) (4.31) (4.10) (3.88) (3.23) (1.86)
# women. aged 2049 -0.013 0.026 -0.034 -0.018 -0.088 0.098 “ -0.045
(0.18) (0.31) (0.49) (0.37) (0.78) (2.09) (0.68)
ii men. aged 50+ -0.733 ‘“ -0.654 “‘ -0.651 ‘" -0.356 ”‘ -0.184 -0.177 ‘ 0.309 ‘“
(5.57) (3.74) (5.37) (4.35) (1.09) (1.89) (3.13)
# women. aged 50+ -0.484 ‘” -0.456 ”‘ -0.295 ‘” 0.086 0.213 0.236 '” 0.027
(4.50) (3.55) (2.93) (1.01) (1.42) (2.70) (0.22)
Business Assets (million) 0.005 -0.031 -0.086 0.015 -0.021 -0.025 0.026 '
(0.35) (0.91 ) (1.47) (1.09) (0.93) (0.96) (1.83)
Province of Residence
North Surnatera 1.417 ‘” 0.326 -0.942 ” 1.154 '" 1.402 “‘ -0.686 3.256 “"
(4.82) (0.66) (2.45) (3.29) (2.99) (1.27) (5.53)
North Sumatera " Urban -0.894 ‘" 0.452 0.901 " ~1.018 “‘ 0.042 0.193 -3.452 ”‘
(2.68) (0.60) (2.07) (2.65) (0.04) (0.33) (4.74)
West Sumatera 1.256 “‘ 1.089 ” 0.117 0.921 ‘” 1.364 ‘” 0.082 1.762 ‘"
(4.57) (2.48) (0.35) (3.14) (2.92) (0.20) (4.00)
West Surnatera ‘ Urban 0.073 0.818 0.533 .0277 0.604 0.243 -0.898
(0.17) (1 .24) (1.38) (0.68) (1 .08) (0.56) (0.86)
South Sumatera 1.146 ‘“ 0.601 -0.718 ‘ 0.328 0.378 -1.076 ‘” 2.210 “‘
(2.83) (1.10) (1.69) (0.96) (0.53) (3.35) (4.72)
South Sumatera ‘ Urban -0.596 0.372 0.600 -0.411 -0.691 0.184 -1.855 ”‘
(0.83) (0.66) (1.05) (0.93) (0.78) (0.41) (2.70)
(continued)

141

 

Appendix Table 5.31! (continued)

Multinomial Logit for Sector Choice Model:

The Effects of Linear Own Schooling, Parental Schooling and Residency (Full Specification)

 

 

 

 

IFLSZ
Men Women J
Self Public Private Self Public Private Family
Employment Sector Sector Employment Sector Sector Worker
Lampung 1.334 "‘ -0.706 -1.011 “‘ 0.507 ” 1.116 ’ -1.177 '" 2.667 “‘
(4.06) (1.29) (3.29) (2.08) (1 .94) (3.46) (6.94)
Lampung ‘ Urban -0.252 1.132 1.033 ”‘
(0.54) (1.41) (2.60)
West Java 1.421 "‘ 1.388 "‘ 0.186 -0.057 0.309 -0.935 ‘” -0.065
(5.17) (3.60) (0.66) (0.21) (0.56) (4.54) (0.16)
West Java ‘ Urban -0.662 “ 0.004 0.530 ‘ 0.043 0.674 0.817 ‘“ -0.552
(2.02) (0.01) (1.72) (0.18) (1 .35) (3.75) (1.26)
Central Java 1.447 ‘“ 1.452 ”‘ 0.237 1.697 ‘" 2.206 “‘ 0.551 “ 2.731 ‘“
(4.68) (3.85) (0.80) (6.12) (4.79) (2.01) (6.36)
Central Java ‘ Urban -0.095 -0.027 0.444 -0.538 “ —0.928 ‘ 0.067 -0.841 ”
(0.25) (0.06) (1.38) (2.02) (1.84) (0.21) (2.25)
Yogyakarta 1.255 “‘ 1.291 ‘ 0.001 2.660 ”' 2.290 ‘” 1.365 ” 3.913 ‘”
(3.08) (1.92) (0.00) (6.80) (4.15) (2.56) (9.08)
Yogyakarta " Urban -0.154 0.168 0.968 -1.306 ‘" -0.847 -0.357 -2.144 ”"
(0.35) (0.24) (1.38) (3.20) (1.39) (0.62) (5.15)
East Java 1.116 ‘” 0.439 -0054 0.249 0.973 ” 0.154 1.378 "“
(4.24) (1 .00) (0.17) (1.06) (2.14) (0.56) (3.23)
East Java " Urban 0.504 1.307 " 1.146 ‘" 0.084 0.070 0.194 -0.307
(1.53) (2.38) (2.76) (0.38) (0.17) (0.68) (0.72)
Bali 2.177 ”‘ 2.455 “‘ 0.210 1.964 ”‘ 2.288 '"" 0.178 2.904 ‘”
(6.54) (5. 10) (0.64) (5.75) (3.80) (0.69) (5.55)
Bali ‘ Urban -0.369 0.334 0.675 -0.509 -0.349 0.391 -1.889 ”‘
(1.01) (0.47) (1.22) (1.35) (0.38) (1.12) (3.50)
West Nusa Tenggara 1.625 “' 1.540 f” -0.253 1.264 ‘” 2.050 "" 0.276 3.256 “"
(4.45) (3.89) (0.57) (4.44) (5.04) (0.78) (7.94)
West Nusa Tenggara ‘ Urban -0.836 0.500 -0.088 -0.041 0.168 -0.298 -1.677 “
(1 .44) (0.83) (0.17) (0.09) (0.24) (0.53) (2.37)
South Kalimantan 1.457 f” 1.893 "‘ -0.606 1.044 “‘ 1.787 ‘" -0.176 2.578 ‘"
(4.34) (3.78) (1.21) (3.93) (3.43) (0.43) (5.20)
South Kalimantan " Urban -0.546 -0.365 0.219
(0.84) (0.50) (0.34)
South Sulawesi 1.021 ‘” 1.451 ”‘ -1.341 ’“ -0.191 0.911 ” -1.644 ”‘ -1.997 ‘f‘
(3.54) (4.15) (3.32) (0.65) (2.38) (5.61) (2.94)
South Sulawesi ’ Urban -0.285 -0.068 0.791 "
(0.77) (0.15) ( 1.68)
Constant -9.049 ”’ -18.473 ”‘ -5.239 ‘" -4.009 ‘" -13.534 ”" 1.023 -2.067
(5.56) (5.98) (3.60) (2.75) (4.19) (0.86) (1.06)
Wald test
Own Schooling 390.70 230.27
(0.000) (0.000)
Father's Schooling 19.47 34.75
(0.078) (0.004)
Mother’s Schooling 17.67 26.29
(0.039) (0.010)
Parental Schooling 64.89 66.15
(0.000) (0.000)
Age 464.23 381.63
(0.000) (0.000)
HH Composition 123.57 83.28
(0.000) (0.000)
1111 Composition and Assets 128.14 92.54
(0.000) (0.000)
Residency 569.69 919.63
(0.000) (0.000)
Pseudo R 2 0.188 0.158
Observations 5,607 6,583
Source: iFLSZ.

Base category for men is nonparticipation and family worker sectors, and for women is nonparticipation. Omitted category for own and parental
schooling is no schooling, for province of residence is Jakarta, and for month of interview is August/September. Adjustments are made due to
small cell size for some variables for women: 10+ years for own schooling and no urban interaction terms for the province of iampung, South
Kalimantan and South Sulawesi. Standard errors are robust to clustering at the community level and heteroskedasticity. Absolute r-statistics are
in parentheses. Signiﬁcant at the 1%(‘”), 5%(”) and 10%(‘) indicated. p -va1ues for Wald test are in parentheses.

142

Multinomial Logit for Sector Choice Model:

Appendix Table 53C

The Effects of Non-linear Own Schooling, Parental Schooling and Residency (Full Specification)

 

 

 

 

iFLS3
Men Women
Self Public Private Self Public Private Family
Employment Sector Sector Employment Sector Sector Worker
Own Schooling
1-3 years 0.628 " 0.845 0.571 " 0.224 ‘ 0.613 -0.I85 0.034
(3.09) (1 .3 8) (2.65) (1.93) (1 .09) (1.42) (0.25)
4-6 years 0.616 “ 1.896 ”‘ 0.496 “ 0.048 0.229 -0.505 “‘ -0.224 '
(3.48) (3.76) (2.38) (0.42) (0.34) (4.04) (1 .73)
7-9 years 0.377 ‘ 2.797 ‘” 0.348 -0.135 1.473 “‘ -1.168 ”° -0.647 ”‘
(1.93) (5.50) (1 .61) (0.99) (2.64) (7.45) (3.76)
10-12 years 0.110 4.148 ‘” 0.268 -0.093 4.729 ”" -0.438 ” -0.627 "
(0.55) (8.20) (1.19) (0.65) (8.53) (2.89) (3.38)
13+ years 0.045 4.978 ‘” 0.044
(0.18) (9.48) (0.17)
Father's Schooling
Some Elementary 0.128 0.263 0.072 0.074 0.284 0.123 0.044
(0.73) (1 .08) (0.42) (0.70) (1.05) (0.94) (0.37)
Completed Elementary 0.083 0.166 0.034 0.060 0.299 -0.228 "‘ -0.257 ”
(0.53) (0.80) (0.22) (0.61) (1.37) (1.97) (2.17)
Secondary/Tertiary -0.287 -0.144 -0.237 -0.102 0.185 0.0004 -0.426 ”
(1.46) (0.60) (1.26) (0.79) (0.77) (0.00) (2.16)
Missing -0.159 -0.461 " -0.162 0.128 -0.407 0.051 -0.004
(0.81) (1.67) (0.79) (1.23) (1.22) (0.38) (0.02)
Mother's Schooling
Some Elementary -0.471 “ -0.396 " -0.323 “ 0.022 0.091 0.018 0.084
(3.09) (1 .97) (2.13) (0.20) (0.43) (0.15) (0.62)
Completed Elementary/ -0.150 0.019 0.136 -0.065 0.132 0.168 0.147
Secondary/Tertiary (0.98) (0.1 1) (0.89) (0.67) (0.69) (1.47) (1.1 1)
Missing -0.058 -0.223 0.151 -0.444 “‘ -0.530 -0.172 -0.454 "
(0.27) (0.77) (0.71) (3.64) (1.47) (1.19) (2.66)
Age (spline)
25-29 years 0.236 "‘ 0.348 “" 0.137 “ 0.171 “‘ 0.127 -0.021 0.114 “
(4.92) (3.82) (3.04) (4.00) (1.22) (0.59) (2.34)
30-39 years 0.044 ” 0.172 ‘” 0.003 0.065 ’“ 0.288 ‘" 0.030 “ 0.022
(2.08) (6.60) (0.16) (5.06) (10.46) (2.05) (1 .29)
40-49 years 0.039 ‘ 0.098 ‘” 0.001 0.043 " -0.065 ” -0.037 ” 0.014
(1.71) (3.61) (0.06) (3.16) (2.08) (2.10) (0.81)
50-59 years -0.097 ‘“ -0.225 “‘ -0.174 "‘ -0.075 “' -0.044 -0.106 “" .0074 “
(3.49) (5.62) (5.55) (4.02) (0.92) (4.33) (3.38)
iii! Composition and Business Assets
ii men, aged 20-49 0250 "' -0.313 ‘” -0.189 "‘ -0.153 "' -0.224 ” -0.136 "’ -0.041
(5.15) (4.93) (3.97) (4.24) (2.76) (3.58) (0.93)
it women, aged 20-49 .0037 -0.041 -0.017 0.042 0.092 0.090 ” 0.047
(0.68) (0.61 ) (0.32) (1.17) (1.30) (2.17) (0.96)
ii men, aged 50+ -0.699 “" -0.664 ‘” -0.497 “‘ -0.293 “‘ -0.168 —0.056 0.263 ”
(6.09) (4.47 ) (4.49) (3.96) (1 .03) (0.71) (3.27)
if women, aged 50+ 0422 "‘ -0.437 "‘ -0.338 ‘” 0.048 -0.018 0.227 “ 0.001
(4.67) (3.67) (3.62) (0.71) (0.13) (3.00) (0.01)
Business Assets (million) 0.021 ‘ -0.004 -0.071 0.007 0.014 -0.013 0.015 ”
(1.67) (0.28) (1.49) (1.53) (1.24) (1.12) (2.47)
Province of Residence
North Sumatera 0.904 ” 0.431 -0.368 1.142 “ 2.284 ” 0.105 2.736 “‘
(2.60) (0.56) (0.77) (3.14) (3.43) (0.24) (6.48)
North Summera ‘ Urban -0.479 -0.076 0.139 -0.899 “ -l.589 “ -1.123 “ -2.450 “
(1.25) (0.10) (0.28) (2.29) (2.22) (2.47) (3.33)
West Sumtera 0.482 ' 0.588 ‘ -0.572 ‘ 0.876 " 1.568 ” -0.596 ' 1.321 ‘"
(1.79) (1.71) (1.93) (2.72) (3.26) (1.71) (3.58)
West Sumntera ' Urban -0.408 0.565 0.706 -0.354 0.552 0.259 -0.481
(1 .00) (1 .00) (1 .56) (1 .04) (0.96) (0.64) (1.09)
South Summers 0.725 ‘ 0.025 -0.954 ‘ 0.830 “ 0.494 -0.960 ” 2.627 “‘
(1.77) (0.04) (1.76) (2.78) (0.65) (2.73) (7.68)
South Surnatera ‘ Urban -0.336 0.721 1.428 “ ~0.791 ‘ -0.420 0.468 -1.735 ’”
(0.60) (1.02) (2.29) (1.95) (0.51) (0.99) (4.02)
(continued)

143

 

Appendix Table 5.3C (continued)

Multinomial Logit for Sector Choice Model:

The Effects of Linear Own Schooling, Parental Schooling and Residency (Full Specification)

 

 

 

 

 

iFLS3
Men Womsn
Self Public Private Self Public Private Family
Employment Sector Sector Employment Sector Sector Worker
Lampung 1.701 ‘” 0.688 0.007 0.997 ” 1.439 -0.866 “ 2.872 "‘
(4.43) (1.31) (0.01) (3.43) (1.64) (2.22) (7.29)
Lampung " Urban -1.150 ‘ 0.246 -0.023 -0.558 0.189 0.627 -2.059 “"
(1.80) (0.32) (0.03) (1.35) (0.15) (1.13) (4.35)
West Java 0.391 0.812 ” -0.396 -0.112 1.418 ” -0.542 ” 0.960 “
(1.51) (2.26) (1.52) (0.54) (2.96) (2.81) (3.19)
West Java ‘ Urban -0.549 “ -0.274 0.383 -0.208 -0.429 0.272 -0.884 ”
(2.02) (0.75) (1.43) (0.88) (0.98) (1.30) (3.00)
Central Java 0.627 “ 1.179 ” -0.261 1.168 ‘" 2.226 ‘“ 0.035 2.115 “"
(2.47) (3 .42) (0.85) (4.43) (5.79) (0.13) (6.46)
Central Java ‘ Urban -0.114 -0.443 0.389 -0.328 -0.979 ” 0.139 -1.220 "‘
(0.34) (1 . 17) (1.02) (1.26) (2.26) (0.41) (3.73)
Yogyakarta 0.273 0.538 -0.527 1.687 ‘” 2.890 ‘” 0.446 3.180 “"
(0.87) (0.92) (0.89) (5.78) (6.28) (0.96) (6.43)
Yogyakarta ' Urban 0.154 0.436 1.012 ’ -0.608 ‘ -1.478 ” 0.213 -1.590 “
(0.40) (0.71) (1 .65) (1.90) (3.02) (0.43) (3.21)
East Java 0.488 " 0.062 .0288 0.596 “ 1.589 ”‘ 0.230 1.914 ”‘
(2.24) (0.17) (1.09) (3.37) (3.71 ) (1 .27) (6.61)
East Java " Urban -0.406 f 0.310 0.259 -0.495 “ -0.765 ‘ -0.307 -1.592 "‘
(1 .68) (0.72) (0.88) (2.36) (1.74) (1.43) (4.01)
Bali 1.031 “ 1.676 "‘ 0400 0.874 ”‘ 1.645 ” -1.227 “ 0.841 ”
(2.76) (3.79) (1.15) (3.94) (3.16) (3.32) (2.65)
Bali ‘ Urban -0.026 0.752 0.698 -0254 0.583 1.122 ” -0.755
(0.05) (1.17) (1.42) (0.75) (0.67) (2.58) (1.15)
West Nusa Tenggara 0.891 ”‘ 1.372 “" -0.196 0.461 ‘ 2.277 ‘" -0.369 1.821 ‘”
(4.07) (4.26) (0.57) (1.77) (6.00) (1 .33) (6.59)
West Nusa Tenggara " Urban -0.457 -0.151 -0.088 0.272 -0.467 0170 -1.024 ”‘
(1 .57) (0.37) (0.22) (0.98) (0.77) (0.36) (3.93)
South Kalimantan 0.932 “ 1.722 ”" -0.757 ‘ 0.900 ‘” 2.079 ” -1.284 “ 2.601 ‘”
(2.57) (3.56) (1.96) (4.29) (3 .02) (3.07) (9.45)
South Kalimantan ‘ Urban -0.796 ‘ -0.499 0.002 -0.437 -0.376 0.718 -2.445 “‘
(1.71) (0.91) (0.00) (1.33) (0.51) (1.38) (4.89)
South Sulawesi 0.509 ” 0.995 " -1.853 "‘ -0.390 1.216 -1.840 “' 0.829 “
(2.12) (2.76) (4.39) (0.97) (1 .64) (6.08) (2.12)
South Sulawesi ‘ Urban -0.019 0.711 1.439 " 0.011 -0.224 0.379 -0.715 “
(0.06) (1 .49) (2.97) (0.03) (0.30) (1.15) (1.71)
Constant -5.015 "" -13.903 ‘“ -1.531 -5.814 ‘" -11.719 "‘ 0.391 -5.056 ‘"
(3 .74) (5.36) (1.21) (4.89) (4.00) (0.39) (3.67)
Wald test
Own Schooling 461.82 291.91
(0.000) (0.000)
Father's Schooling 11.67 36.02
(0.473) (0.003)
Mother's Schooling 24.05 23.44
(0.004) (0.024)
Parental Schooling 44.73 71.68
(0.002) (0.000)
Age 443.67 428.46
(0.000) (0.000)
HH Composition 172.05 79.21
(0.000) (0.000)
HH Composition and Assets 176.21 87.33
(0.000) (0.000)
Residency 331.55 829.93
(0.000) (0.000)
Pseudo R’ 0.169 0.132
Observations 7,145 7,716
Source: IFLS3.

Base category for men is nonparticipation and family worker sectors, and for women is nonparticipation. Omitted category for own and parental
schooling is no schooling, for province of residence is Jakarta, and for month of interview is lune/July. Adjustments are made due to small cell

size for some variables for women: 10+ years for own schooling. Standard errors are robust to clustering at the community level and

heteroskedasticity. Absolute r—statistics are in parentheses. Significant at the l%(‘”), 5°/o(“) and 10°/o(‘) indicated. p-values for Wald test are

144

Appendix Table 5.4A
Multinomial Logit for Sector Choice Model:
The Effects of Linear Own Schooling

 

 

 

 

 

Panel Respondents, iFLSl
Men Won_i_e_g
Self Public Private Self Public Private Family
Employment Sector Sector Employment Sector Sector Worker
Own Schooling -0.079 “' 0.341 '” 0.020 -0.027 “ 0.435 ”‘ -0.018 -0.104 ‘”
(2.40) (8.48) (0.59) (2.03) (6.96) (0.89) (4.54)
Age in 1993 (spline)
25-29 years 0.084 0.242 0.143 0.064 0.438 '” 0.062 0.014
(0.67) (1.47) (1.10) (1.11) (3.58) (0.97) (0.24)
30-39 years 0.147 ‘” 0.203 ‘” 0.114 “' 0.080 “" 0.033 0.025 0.013
(2.79) (3.72) (2.13) (4.59) (1.12) (1.16) (0.66)
40-49 years -0.122 ” -0.115 " -0.172 "‘ -0.011 -0.066 0.008 0.028
(2.03) (1.80) (2.88) (0.49) (1.52) (0.29) (1 .24)
50-52 years 0.189 0.417 0.323 0.023 0.160 -0.226 -0.283 "
(0.73) (1.50) (1 .27) (0.22) (0.68) (1 .47) (2.23)
Hi! Composition and Business Assets
# men. aged 20-49 ~0.708 ”‘ -0.309 -0.537 "‘ -0.366 ‘" ~0.308 "' -0.449 ”" -0.064
(3.87) (1 .45) (2.85) (4.40) (2.21) (3.97) (0.83)
11 women, aged 20-49 -0.039 0.264 0.072 0.059 -0. 160 0.011 -0.409 “"
(0.1 1) (0.68) (0.19) (0.63) (1.00) (0.09) (3.02)
# men. aged 50+ -1.210 '” -1.526 “‘ -0.968 ‘f‘ -0.460 “" -0.294 -0.518 “‘ 0.174
(3.71) (3.99) (2.85) (3.19) (1.13) (3.29) (1.31)
if women. aged 50+ -0.057 0.018 0.029 0.111 0.399 " 0.247 ‘ -0.326 ‘
(0.20) (0.06) (0.10) (0.88) (1 .77) (1 .72) (1.76)
Business Assets (million) 0.377 -0.163 -1.101 ‘ 0.051 -0.066 .0416 0.080 "
(1.59) (0.63) (1.78) (1.22) (0.95) (1.15) (1.79)
Month of interview
October -0.2 59 -0043 0.100 -0. 192 0.283 -0.107 -0.708
(0.57) (0.08) (0.24) (0.76) (0.66) (0.41) (1.50)
November 0196 0.456 0.109 0.428 ’ 0.871 ' 0.281 -0.397
(0.44) (0.96) (0.27) (1.69) (1 .92) (1 .05) (0.95)
December -0.179 -0.108 0.423 0.478 ‘ 0.340 0.070 0.508
(0.42) (0.24) (1 .07) (1 .86) (0.82) (0.27) (1.23)
January 0.594 1.040 " 0.428 -0.298 -0.084 -0.420 -0.028
(1.05) (1 .77) (0.80) (0.90) (0.16) (1.26) (0.06)
Constant 1.442 -8.965 ” -1.009 -2.710 -18.l81 '” -2.599 -0.604
(0.42) (1 .98) (0.28) ( 1.63) (5.24) (1 .43) (0.37)
Wald test
Age 79.86 67.33
(0.000) (0.000)
H11 Composition 49.69 57.27
(0.000) (0.000)
1111 Composition and Assets 58.41 65.03
(0.000) (0.000)
Month of Interview 22.61 51.61
(0.031) (0.000)
Pseudo R ’ 0.172 0.092
Observations 2.797 3,722
Source: iFLSl.

Base category for men is nonparticipation and family worker sectors, and for women is nonparticipation. Omitted category for month of interview
is August/September. Standard errors are robust to clustering at the community level and heteroskedasticity. Absolute r-statistics are in
parentheses Signiﬁcant at the l%(”‘), 5°/u(”) and 10%(‘) indicated. p-values for Wald test are in parentheses.

145

Appendix Table 5.4B
M ultlnomial Logit for Sector Choice Model:
The Effects of Linear Own Schooling

 

 

 

 

 

 

Panel Respondents, iFL82
Men Women
Self Public Private Self Public Private Family
Employment Sector Sector Employment Sector Sector Worker
Own Schooling -0.023 0.351 ‘“ 0.045 -0.046 ‘” 0.470 ”‘ -0.048 “‘ -0.082 “‘
(0.89) (10.85) (1 .60) (3.35) (6.91) (2.62) (3.82)
Age in 1993 (spline)
25-29 years 0089 0.154 0.025 0.066 0.495 ‘" 0.102 ‘ 0.157 "
(0.62) (0.95) (0.18) (1.25) (3.68) (1.93) (2.12)
30—39 years 0.067 0.086 ‘ 0.002 0.052 ‘” 0.019 -0.003 0.027
(1.45) (1.76) (0.04) (3.27) (0.57) (0.14) (1.36)
40-49 years -0.065 -0.027 -0.084 0055 ” -0.074 -0.042 0060 '
(1.17) (0.43) (1.44) (2.49) (1.65) (1.53) (1.95)
50-52 years -0.253 -0.587 ” -0.246 0.087 0.157 -0.046 -0.033
(1.45) (2.34) (1 .32) (0.89) (0.66) (0.36) (0.23)
HH Composition and Business Assets
# men, aged 20-49 0308 ”" -0.284 " -0.275 ” -0.239 ‘” -0.311 “ -0.268 "" -0.211 ‘"
(2.79) (2.03) (2.43) (4.00) (2.55) (3.58) (3.19)
# women. aged 2049 -0.204 0.031 —0. 168 -0.006 -0.247 ‘ 0.008 -0.229 “‘
(1 .41) (0.20) (1 . 12) (0.10) (1.84) (0.10) (2.62)
# men. aged 50+ -0.604 ” -0.814 ”“ .0496 ‘ -0.212 ‘ -0.061 -0.283 ” 0.388 ‘“
(2.17) (2.62) (1.78) (1.81) (0.26) (2.06) (3.13)
# women, aged 50+ 0179 -0.020 -0.112 0.216 ” 0.051 0.265 ” -0.183
(0.81 ) (0.08) (0.50) (1.99) (0.23) (2.20) (1 .05)
Business Assets (million) -0.007 -0.205 "‘ -1.215 “"' 0.046 -0.004 -0.352 0.055
(0.31) (3.07) (3.43) (1.26) (0.13) (1.38) (1.39)
Month of interview
October 0.348 0.303 0.093 0.177 0.604 " 0.292 0.094
(1 .02) (0.77) (0.29) (1.03) (2.02) (1.27) (0.29)
November 0.452 0.711 ‘ 0.299 0.103 1.025 “ 0.110 0.010
(1 .29) (1.67) (0.91) (0.59) (2.56) (0.44) (0.03)
December 0386 0.359 0.422 0.310 0.487 0.389 0.344
(0.89) (0.73) (1.01) (1.40) (1.43) (1.41) (0.91)
January -1.390 ‘" -0.481 0.971 ” -0.713 ” 0.522 0.205 0.482
(3.20) (0.98) (2.31) (2.42) (1 . 19) (0.68) (1 .03)
Constant 5.579 -6.041 2.339 -2.299 -20.234 ‘" -3.517 “ -5.187 ”
(1.38) (1.33) (0.60) (1.55) (5.54) (2.37) (2.32)
Wald test
Age 73.44 56.10
(0.000) (0.000)
HH Composition 22.18 76.26
(0.036) (0.000)
H11 Composition and Assets 42.34 82.95
(0.000) (0.000)
Month of interview 28.39 23.12
(0.005) (0.111)
Pseudo R ’ 0.159 0.084
Observations 2,797 3,722
Source: lFLSZ.

Base category for men is nonparticipation and family worker sectors. and for women is nonparticipation. Omitted category for month of interview
is August/September. Standard errors are robust to clustering at the comtrnunity level and heteroskedasticity. Absolute r-statistics are in
parentheses Signiﬁcant at the 1%(“"). 5°/o(“) and 10%(‘) indicated. p-values for Wald test are in parentheses.

146

Appendix Table 5.4C
Multinomial Logit for Sector Choice Model:
The Effects of Linear Own Schooling
Panel Respondents, iFLS3

 

 

 

 

 

Men Women
Self Public Private Self Public Private Family
Employment Sector Sector Employment Sector Sector Worker
Own Schooling -0.095 ‘” 0.327 ‘” -0.024 -0.055 ‘" 0.483 ”‘ o0.073 ‘” -0.126 “"
(4.16) (10.89) (0.98) (4.28) (10.03) (4.38) (6.73)
Age in 1993 (spline)
25-29 years -0.065 0.113 0.071 0.040 0.397 " 0.156 “ 0.074
(0.50) (0.71) (0.53) (0.86) (3.00) (2.57) (1.25)
30—39 years 0.101 “ 0.157 “ 0.050 0.057 ” 0.059 ‘ 0.011 -0.009
(2.30) (3.20) (1 . 14) (3.14) (1.71) (0.49) (0.44)
40-49 years 0152 ” -0.157 "‘ -0.194 "' -0.058 “‘ ~0.095 ‘ -0.078 ” -0.026
(3.05) (2.49) (3.67) (2.55) (1.85) (2.41) (1.03)
5052 years 0238 -1.233 “' -0.153 -0.035 0.037 -0.149 -0.171
(1.54) (3.46) (0.89) (0.37) (0.16) (0.96) (1.57)
HH Composition and Business Assets
ii men. aged 20-49 -0.230 ” -0.206 ‘ -0.229 “ .0182 ‘“ -0.362 ” .0193 “ .0091 ‘
(2.80) (1.93) (2.80) (3.98) (2.93) (3.06) (1.71)
if women, aged 20-49 -0.233 ” -0.202 -0.172 -0.053 -0.286 “ -0.163 “ -0.261 """
(2.17) (1 .63) (1 .49) (1 .04) (2.27) (2.24) (4.01)
# men, aged 50+ -0.251 -0.257 -0.115 -0.348 “ -0.527 ” -0.351 “ 0.410 “'
(0.94) (0.78) (0.40) (3 .37) (2.31) (2.65) (3.87)
# women, aged 50+ -0.197 -0.103 -0.219 0.108 0.094 0.133 -0.190
(1.18) (0.51) (1 .20) (0.97) (0.43) (0.98) (1 .42)
Business Assets (million) 0.190 “ 0.145 ‘ -0.231 ‘ 0.019 " -0.019 -0.128 0.026 "
(2.26) (1.68) (1.76) (1.68) (0.78) (1.57) (2.01)
Month of Interview
October 0.067 -0.148 0.036 0.478 ” 0.666 “ 0.518 ” 0.239
(0.25) (0.45) (0.13) (2.80) (2.07) (2.51) (0.98)
November «0.093 -0.378 -0.166 0.619 '“ 0.728 ” 0.412 ‘ 0.469 "
(0.33) (1.10) (0.56) (3.69) (2.43) (1.84) (1.78)
December 0.027 -0.142 0.145 0.795 ”‘ 0.692 ‘ 0.491 ' 0.456
(0.08) (0.39) (0.46) (4.11) (1.96) (1.88) (1.57)
January -0.164 -0.465 0.122 0.352 ' 0.403 0.370 0.198
(0.44) (0.94) (0.32) (1.75) (0.88) (1 .41) (0.68)
Constant 5.337 -4.963 1.031 -1.219 -17.344 ”" -4.452 “ -1.938
(1.45) (1.12) (0.28) (0.92) (4.60) (2.65) (1.16)
Wald test
Age 94.52 61.20
(0.000) (0.000)
1111 Composition 22.22 105.50
(0.035) (0.000)
HH Composition and Assets 45.87 1 18.36
(0.000) (0.000)
Month oflnterview 5.52 27.14
(0.938) (0.040)
Pseudo R ’ 0.170 0.094
Observations 2,797 3,722
Source: iFLS3.

Base category for men is nonparticipation and family worker sectors. and for women is nonparticipation. Omitted category for month of interview
is June/July. Standard errors are robust to clustering at the community level and heteroskedasticity. Absolute r-statistics are in parentheses.
Signiﬁcant at the 1°/o(“‘), 5%(”) and 10°/o(‘) indicated. p-values for Wald test are in parentheses.

147

Appendix Table 5.5A
Multinomial Logit for Sector Choice Model:

The Effects of Non-linear Own Schooling

 

 

 

 

Panel Respondents, iFLSi
Men Wom_gn
Self Public Private Self Public Private Family
Employment Sector Sector Employment Sector Sector Worker
Own Schooling
1-3 years 0.024 -0.210 0.050 0.335 ” 0.263 -0.104 0.146
(0.06) (0.29) (0.13) (2.16) (0.62) (0.59) (0.93)
4-6 years 0.367 1.843 “' 0.471 0.191 0.084 0655 "‘ -0.339 ‘
(0.95) (3.05) (1.22) (1.26) (0.19) (3.78) (1 .91)
7-9 years 0.244 2.831 “' 0.806 0.016 1.201 “ -0.660 ”‘ -0.704 ”‘
(0.46) (3.99) (1 .50) (0.08) (2.30) (2.78) (2.67)
10-12 years -0.679 3.957 ‘" 0.451 -0.271 4.072 "‘ 0.170 -1.456 ‘”
(1.49) (6.13) (1.02) (1.37) (8.27) (0.81) (3.80)
13+ years -2.027 ”' 3.569 ““ -0.297
(3.79) (5.52) (0.57)
Age in 1993 (spline)
25-29 years 0.101 0.268 ‘ 0.156 0.058 0.515 ‘“ 0.078 0.007
(0.78) (1 .67) (1 .16) (0.99) (4.06) (1 .23) (0.11)
30-39 years 0.135 ” 0.197 ‘” 0.108 ‘ 0.082 "' 0.022 0.023 0.016
(2.49) (3.51) (1 .96) (4.71) (0.73) (1 .08) (0.81)
40-49 years 0114 ‘ -0.112 ‘ -0.166 ‘” -0.006 .0080 ‘ 0.004 0.031
(1.92) (1.75) (2.81) (0.27) (1.69) (0.14) (1.35)
50-52 years 0.128 0.425 0.290 0.045 0.116 -0.253 -0.270 “
(0.49) (1.53) (1 . 13) (0.43) (0.47) (1.60) (2.12)
1111 Composition and Business Assets
# men, aged 2049 -0.727 ”‘ -0321 -0.561 ‘" -0.377 ”‘ -0.279 ‘ -0.429 ”' -0.072
(4.03) (1 .55) (3.02) (4.49) (1.90) (3.77) (0.92)
1: women, aged 20-49 -0.053 0.254 0.058 0.062 -0.086 0009 0.408 ”‘
(0.16) (0.70) (0.17) (0.66) (0.52) (0.08) (3.01)
ii men, aged 50+ -1.120 ‘" -i.532 "‘ -0.936 ‘“ -0.461 ‘“ -0.269 -0.513 ’“ 0.181
(3.49) (4.15) (2.82) (3.19) (1.02) (3.23) (1.35)
# women, aged 50+ -0.086 0.050 0.013 0.104 0.479 ” 0.256 " -0.326 "
(0.31) (0.16) (0.05) (0.83) (2.18) (1 .76) (1 .77)
Business Assets (million) 0.405 " -0.135 -1.065 " 0.053 -0.056 -0.417 0.083 ‘
(1.69) (0.52) (1.74) (1.28) (0.86) (1.19) (1.88)
Month of interview
October -0.294 -0.1 18 0.040 -0.170 0.058 -0. 133 ~0.687
(0.64) (0.23) (0.10) (0.67) (0.13) (0.51) (1 .44)
November -0.262 0.394 0.052 0.439 " 0.637 0.253 -0.388
(0.59) (0.84) (0.13) (1 .74) (1 .24) (0.96) (0.92)
December -0.218 -0.132 -0.464 0.491 ‘ 0.188 0.037 0.514
(0.51) (0.31) (1.17) (1 .92) (0.43) (0.14) (1 .24)
January 0.614 1.061 ‘ 0.407 -0.244 -0.249 -0.493 0.024
(1 .09) (1.85) (0.77) (0.74) (0.47) (1.45) (0.05)
Constant 0.659 -9.313 ” -1.444 .2803 ‘ -18.692 ‘“ -2.828 -0.598
(0.19) (2.09) (0.39) (1 .66) (5.21) (1 .55) (0.36)
Wald test
Own Schooling 374.43 241.75
(0.000) (0.000)
Age 79.58 76.40
(0.000) (0.000)
HH Composition 51.86 57.49
(0.000) (0.000)
Iii-i Composition and Assets 60.65 65.69
(0.000) (0.000)
Month of Interview 23.45 47.44
(0.024) (0.000)
Pseudo R’ 0.177 0.101
Observations 2,797 3,722
Source: iFLSl.

Base category for men is nonparticipation and family worker sectors, and for women is nonparticipation. Omitted category for own schooling is

no schooling and for month of interview is August/September. Standard errors are robust to clustering at the community level and

heteroskedasticity. Absolute r-statistics are in parentheses. Signiﬁcant at the 1%("‘), 5%(") and 10%(') indicated. p-values for Wald test are

in parentheses

148

 

Appendix Table 5.5B
Multinomial Logit for Sector Choice Model:
The Effects of Non-linear Own Schooling

 

 

 

 

 

Panel Respondents, iFLSZ
Men Wogn
Self Public Private Self Public Private Family
Employment Sector Sector Employment Sector Sector Worker
Own Schooling
1-3 years 0.252 -0.667 0.358 0.229 1.044 " 0.011 0.053
(0.80) (1 .09) (1 . 1 1) (1 .55) (1 .70) (0.07) (0.26)
4.6 years 0.499 ‘ 1.058 ” 0.603 “ 0.089 0.655 -0.669 ‘“ -0.345 ‘
(1 .75) (2.40) (1 .97) (0.68) (1 .24) (4.21) (1.82)
7-9 years 0206 1.523 ”' 0.242 -0.310 " 2.172 ‘” -l.199 ”" -0.786 ‘”
(0.57) (3.17) (0.68) (1.65) (3.65) (5.20) (3.20)
10-12 years -0.263 3.188 ‘“ 0.515 -0.604 “‘ 4.981 "" -0.059 -0.996 “"
(0.75) (6.76) (1.33) (2.81) (8.21) (0.30) (3.08)
13+ years 0.028 4.505 "" 1.378 "
(0.04) (6.28) (1.98)
Age in 1993 (spline)
25-29 years -0.098 0.144 0.011 0.058 0.581 "" 0.115 “ 0.153 ”
(0.69) (0.91) (0.08) (1.08) (4.29) (2.16) (2.07)
30-39 years 0.064 0.091 ‘ 0.002 0.054 “" 0.013 -0.004 0.029
(1 .40) (1 .89) (0.05) (3.37) (0.40) (0.22) (1 .42)
40-49 years 0064 -0.032 -0.084 -0.052 ” -0.093 ‘ -0.042 -0.059 ‘
(1.16) (0.51) (1.44) (2.34) (1.88) (1.50) (1.93)
50—52 years 0235 -0.583 “ -0.228 0.102 0.182 -0.053 -0.024
(1 .35) (2.23) (1 .22) (1 .05) (0.69) (0.40) (0.17)
1111 Composition and Business Assets
11 men, aged 20-49 0301 ”‘ -0.276 ‘ -0.267 " -0.247 ”" ~0.336 ‘” -0.254 ‘"” -0.212 ’”
(2.74) (1.96) (2.37) (4.16) (2.85) (3.32) (3.19)
# women. aged 20-49 0190 0.044 -0.160 -0.002 -0.217 -0.018 -0.229 ‘"
(1 .34) (0.29) (l . 10) (0.03) (1.59) (0.25) (2.63)
ii men, aged 50+ -0.633 ” -0.826 ”‘ -0.523 ‘ -0.214 ‘ -0.115 -0.276 " 0.391 ‘”
(2.29) (2.64) (1 .88) (1.82) (0.49) (1.98) (3.17)
11 women. aged 50+ -0.186 -0.017 -0.124 0.215 “ 0.109 0.229 ‘ -0.188
(0.83) (0.07) (0.55) (1.99) (0.50) (1 .84) (1 .08)
Business Assets (million) -0.004 -0.215 ‘” -1.214 ”‘ 0.047 -0.016 -0.365 0.056
(0.17) (3.21) (3.48) (1.27) (0.50) (1.41) (1.38)
Month of interview
October 0.353 0.344 0.087 0.171 0.648 " 0.307 0.093
(1.06) (0.89) (0.28) (1.00) (2.40) (1 .34) (0.28)
November 0.442 0.770 ‘ 0.305 0.087 0.950 "‘ 0.119 0.002
(1 .26) (1.81) (0.93) (0.50) (2.36) (0.47) (0.01)
December 0.406 0.473 0.441 0.329 0.562 ' 0.405 0.357
(0.94) (0.97) (1 .06) (1.47) (1 .85) (1 .47) (0.93)
January -i.389 ‘” -0.419 -0.991 ” -0.688 ” 0.579 0.179 -0.459
(3.15) (0.86) (2.33) (2.33) (1.47) (0.59) (0.98)
Constant 5.588 -4.877 2.557 -2.270 -21.480 ‘” -3.766 ” -5.149 "
(1.39) (1.09) (0.66) (1.51) (5.73) (2.51) (2.28)
Wald test
Own Schooling 397.46 193.92
(0.000) (0.000)
Age 74.04 63.35
(0.000) (0.000)
iii-i Composition 21.51 75.22
(0.043) (0.000)
HH Composition and Assets 42.76 84.18
(0.000) (0.000)
Month of interview 27.70 22.36
(0.006) (0.132)
Pseudo R ’ 0.164 0.092
Observations 2,797 3,722
Source: iFLSZ.

Base category for men is nonparticipation and family worker sectors. and for women is nonparticipation. Omitted category for own schooling is
no schooling and for month of interview is August/September. Standard errors are robust to clustering at the community level and
heteroskedasticity. Absolute t-statistics are in parentheses. Signiﬁcant at the 1%(‘”), 5%(") and 10%(‘) indicated. p-values for Wald test are
in parentheses

149

Appendix Table 5.5C
Multinomial Logit for Sector Choice Model:

The Effects of Non-linear Own Schooling

 

 

 

 

Panel Respondents, iFLSZ
Men Wom_t_en
Self Public Private Self Public Private Family
Employment Sector Sector Employment Sector Sector Worker
Own Schooling
1-3 years 0.243 -0.254 0.317 0.073 0.593 0256 -0.029
(0.75) (0.35) (0.91) (0.52) (0.92) (1.54) (0.19)
4-6 years 0.163 0.927 ‘ 0.314 -0.106 0.479 -0.559 “ 0564 ”"
(0.60) (1 .67) (1 .03) (0.76) (0.66) (3.34) (3.62)
7-9 years .0361 1.789 “ 0.162 -0.555 " 0.930 -1.527 ‘” -1.087 ‘"
(1 . 14) (3.13) (0.46) (3.23) (1 .30) (6.41) (4.91)
10-12 years -0.861 " 2.957 ”" -0.011 -0.608 " 4.428 "“ -0.570 ” -1.589 '“
(2.67) (5.30) (0.03) (3.22) (7.36) (2.84) (5.68)
13+ years -1.334 ” 3.992 "‘ -0.138
(3.11) (6.56) (0.32)
Age in 1993 (spline)
25-29 years 0062 0.115 0.072 0.035 0.487 ”‘ 0.160 ” 0.070
(0.47) (0.74) (0.54) (0.75) (3.79) (2.65) (1.18)
30-39 years 0.094 “ 0.153 ” 0.046 0.059 “ 0.053 0.008 -0.007
(2.14) (3.12) (1.05) (3.22) (1 .59) (0.3 8) (0.36)
40-49 years -0.145 “ -0.159 " -0.189 “‘ -0.057 ” -0.116 ” -0.080 “ -0.026
(2.90) (2.51) (3.58) (2.47) (2.39) (2.44) (1 .01)
50-52 years -0.243 -1.249 ” -0.159 .0025 0.134 -0.153 -0.162
(1.52) (3.37) (0.90) (0.26) (0.53) (0.96) (1.50)
HH Composition and Business Assets
ii men, aged 2049 -0.238 " -0.197 ‘ -0.234 " -0.188 “‘ -0.421 “ -0.182 “ -0.097 "
(2.87) (1.81) (2.82) (4.07) (3.32) (2.87) (1 .82)
# women, aged 20-49 -0.222 “ -0.192 -0.166 -0.048 -0.247 ” -0.178 “ -0.254 ‘“
(2.09) (1.58) (1 .45) (0.93) (2.03) (2.46) (3.93)
ii men, aged 50+ -0.274 -0.221 -0.124 -0.351 ” -0.606 ” -0.332 '” 0.408 '"
(1.03) (0.66) (0.43) (3 .3 8) (2.69) (2.49) (3.86)
it women, aged 50+ -0.197 -0.133 -0.220 0.107 0.133 0.106 -0.186
(1.18) (0.67) (1 .22) (0.96) (0.63) (0.78) (1 .38)
Business Assets (million) 0.188 " 0.146 ‘ -0.229 ’ 0.019 -0.009 .0132 0.026 ‘
(2.29) (1.72) (1.75) (1.64) (0.46) (1.64) (1.95)
Month of interview
October 0.069 -0.136 0.033 0.483 ” 0.705 “ 0.530 " 0.246
(0.26) (0.41) (0.12) (2.84) (2.22) (2.57) (1.01)
November -0.065 -0.383 -0.156 0.629 ”‘ 0.798 ” 0.433 " 0.474 ‘
(0.23) (1.10) (0.53) (3.76) (2.62) (1.91) (1.80)
December 0.022 -0.168 0.137 0.801 ‘” 0.561 0.518 ” 0.461
(0.07) (0.46) (0.43) (4.15) (1 .64) (1.99) (1 .59)
January 0180 -O.427 0.100 0.363 ‘ 0.450 0.396 0.212
(0.48) (0.85) (0.26) (1.81) (1.07) (1.51) (0.72)
Constant 4.833 -4.106 0.687 -1.189 -17.846 "" -4.422 " -1.911
(1.31) (0.94) (0.18) (0.89) (4.83) (2.63) (1.15)
Wald test
Own Schooling 411.15 270.83
(0.000) (0.000)
Age 94.52 66.14
(0.000) (0.000)
HH Composition 22.67 106.08
(0.031) (0.000)
11H Composition and Assets 45.60 117.17
(0.000) (0.000)
Month of interview 5.42 28.82
(0.942) (0.025)
Pseudo R’ 0.173 0.091
Observations 2,797 3,722
Source: iFLS3.

Base category for men is nonparticipation and family worker sectors, and for women is nonparticipation. Omitted category for own schooling is

no schooling and for month of interview is August/September. Standard errors are robust to clustering at the community level and

heteroskedasticity. Absolute r-statistics are in parentheses. Signiﬁcant at the l%("“), 5%(") and 10%(‘) indicated. p-values for Wald test are

in parentheses.

150

The Effects of Non-linear Own Schooling and Parental Schooling (Base Speciﬁcation)

Appendix Table 5.6A

Multinomial Logit for Sector Choice Model:

 

 

 

 

 

Panel Respondents, iFLSl
_ Mg _ Women
Self Public Private Self Public Private Family
Employment Sector Sector Employment Sector Sector Worker
Own Schooling
1-3 years 0029 «0.264 -0.042 0.318 " 0.247 -0.103 0.125
(0.07) (0.36) (0.10) (2.08) (0.58) (0.58) (0.84)
4-6 years 0.318 1.733 “' 0.374 0.174 0.025 -0.653 “‘ -0.284
(0.77) (2.78) (0.91) (1.17) (0.05) (3.74) (1.62)
7-9 years 0.205 2.695 ‘” 0.672 0.025 1.138 ‘ -0.691 “‘ -0.564 ”
(0.36) (3.66) (1.18) (0.12) (1 .90) (2.84) (2.11)
10-12 years -0.687 3.833 ”" 0.331 -0.211 4.006 “' 0.071 -1.152 ‘"
(1 .27) (5.42) (0.64) (1 .01) (6.72) (0.31) (2.84)
13+ years -2.103 ”‘ 3.422 '" -O.521
(3.68) (5.04) (0.94)
F ather's Schooling
Some Elementary 0.387 -0.110 0.259 0.104 0.112 -0.325 ' 0.098
(0.91) (0.24) (0.61) (0.79) (0.35) (1.71 ) (0.61)
Completed Elementary 0678 ‘ -0.697 " -0.719 ' 0.255 0.054 -0.273 -0.258
(1.79) (1.74) (1.91) (1.63) (0.16) (1.36) (1.33)
Secondary/1‘ ertiary -0.662 -0.842 -0.808 -0.014 -0.184 0.120 -0.306
(1.15) (1.47) (1.44) (0.06) (0.50) (0.50) (0.91)
Missing 0.160 -0.574 0.347 0.056 -0.044 -0.132 -0.210
(0.36) (1.16) (0. 78) (0.38) (0.11) (0.67) (1.03)
Mother's Schooling
Some Elementary 0.066 1.025 “ 0.489 -0.103 0.031 0.517 ‘” 0.091
(0.15) (2.23) (1.11) (0.69) (0.12) (2.64) (0.55)
Completed Elementary/ 0.894 ‘ 1.180 ” 1.380 ‘" -0.273 0.269 0.092 ~0.465 ”
Secondary/Tertiary (1.85) (2.53) (2.93) (1.59) (1.08) (0.44) (2.18)
Missing -0.575 0.081 -0.517 -0.048 0.087 0.046 0.301
(1.32) (0.18) (1 .23) (0.32) (0.29) (0.22) (1 .59)
Age in 1993 (spline)
25-29 years 0.087 0.231 0.138 0.059 0.508 “‘ 0.085 0.012
(0.70) (1.46) (1.06) (1.02) (4.03) (1.33) (0.20)
30-39 years 0.133 “‘ 0.206 ‘” 0.112 ” 0.081 "‘ 0.026 0.023 0.011
(2.50) (3.68) (2.08) (4.56) (0.82) (1.08) (0.59)
40-49 years -0.115 ‘ -0.113 ‘ -0.168 '“ -0.006 —0.079 ' 0.007 0.033
(1.91) (1.75) (2.81) (0.29) (1.66) (0.22) (1.42)
50-52 years 0.192 0.450 0.379 0.049 0.092 -0.270 ‘ -0.230 ‘
(0.70) (1 .57) (1 .41) (0.47) (0.37) (1.69) (1.80)
“H Composition and Business Assets
ii men, aged 20.49 -0.730 “‘ -0.286 -0.562 ”‘ -0.379 ‘” -0.289 “' -0.424 “" -0.057
(4.00) (1.38) (3.01) (4.52) (2.01) (3.73) (0.74)
11 women, aged 20-49 -0.032 0.256 0.075 0.065 -0.075 -0.019 -0.432 “‘
(0.10) (0.72) (0.22) (0.68) (0.45) (0.16) (3.19)
ii men, aged 50+ -1.208 ‘" -1.461 “" -1.050 "' -0.458 ‘” -0.269 -0.514 ‘” 0.203
(3.32) (3.52) (2.88) (3.15) (1.00) (3.23) (1.55)
it women. aged 50+ 0.111 0.143 0.221 0.110 0.523 " 0.269 -0.444 ”
(0.32) (0.39) (0.65) (0.80) (1.98) (1 .61) (2.27)
Business Assets (million) 0.436 ' -0.101 -1.020 ' 0.057 0.061 -0.405 0.089 “
(1.73) (0.37) (1.66) (1.38) (0.95) (1.18) (2.00)
Constant 1.114 -8.414 ‘ -1.036 -2.841 ' ~18.591 ‘" -3.0|1 ‘ -0.675
(0.33) (1.90) (0.29) (1.69) (5.22) (1 .66) (0.41 )
Wald test
Own Schooling 314.28 193.08
(0.000) (0.000)
Father‘s Schooling 27.35 21.86
(0.007) (0.148)
Mother's Schooling 46.85 25.16
(0.000) (0.014)
Parental Schooling 62.73 42.61
(0.000) (0.038)
Age 79.43 74.86
(0.000) (0.000)
1111 Composition 45.36 59.82
(0.000) (0.000)
HH Composition and Assets 54.68 68.79
(0.000) (0.000)
Month of Interview 24.41 47.61
(0.018) (0.000)
Pseudo R ’ 0.186 0.105
Observations 2,797 3,722
Source: iFLSl.

Base category for men is nonparticipation and family worker sectors, and for women is nonparticipation. Month of interview dummy variables are included
in the regressions but are not reported Omitted category for own schooling is no schooling. Standard errors are robust to clustering at the community level
and heteroskedasticity. Absolute r-statistics are in parentheses. Signiﬁcant at the 1%("”), 5%(") and 10%(‘) indicated. p -values for Wald test are in

parentheses.

151

Appendix Table 5.6B
Multinomial Logit for Sector Choice Model:
The Effects of Non-linear Own Schooling and Parental Schooling (Base Speciﬁcation)

 

 

 

 

 

Panel Respondents, iFLSZ
_ Me_n 4 Women _
Self Public Private Self Public Private Family
Employment Sector Sector Employment Sector Sector Worker
Own Schooling
1-3 years 0.317 -0.602 0.381 0.236 0.911 0.002 0.101
(1.06) (1.01) (1.20) (1.62) (1.49) (0.01) (0.52)
4-6 years 0.524 ‘ 1.023 “ 0.556 " 0.098 0.439 -0.694 ‘” -0.202
(1.86) (2.37) (1.83) (0.75) (0.80) (4.34) (1.14)
7-9 years —0.120 1.498 “" 0.219 —0.274 1.915 ”‘ -l.233 “‘ -0.530 “
(0.32) (3.06) (0.58) (1.43) (3.00) (5.21) (2.14)
10-12 years -0.l33 3.184 ”‘ 0.528 -0.522 ” 4.734 ”" -0.i 15 -0.562 '
(0.39) (6.82) (1.38) (2.34) (7.10) (0.53) (1.75)
13+ years 0.084 4.411 ‘” 1.285 ‘
(0.12) (6.34) (1.87)
Father's Schooling
Some Elementary -0.381 -0.638 -0.270 0.399 ‘" 0.689 " 0.380 ” -0.029
(1.23) (1.60) (0.82) (2.86) (2.15) (2.29) (0.14)
Completed Elementary 0244 ~0.173 -0.101 0.226 0.371 0.064 -0.384 ‘
(0.53) (0.35) (0.22) (1.55) (1.03) (0.33) (1.67)
Secondaryfi‘em'ary -l.307 “‘ -l.459 ‘“ -l.185 “' 0.202 -0.015 0.245 -0.834 ”
(2.62) (2.76) (2.37) (0.95) (0.04) (1.03) (2.56)
Missing -0.209 -0.728 -0.042 0.131 -0.184 0.473 “ -0.530 "
(0.40) ( 1.23) (0.08) (0.75) (0.39) (2.05) (i .79)
Mother's Schooling
Some Elementary 0.221 0.645 ‘ 0.274 -0.396 ”' 0.081 0.359 “ -0.238
(0.63) (1.69) (0.77) (2.81 ) (0.27) (2.00) (1.18)
Completed Elementary/ 1.017 " 1.393 ” 1.313 “‘ -0.417 ”‘ 0.360 —0.023 0175
Secondary/Tertiary (2.05) (2.57) (2.66) (2.91) (1.37) (0.11) (0.76)
Missing 0132 0.133 0.198 -0.323 ‘ 0.402 -0.288 0.341
(0.27) (0.24) (0.41) (1.77) (0.86) (1.20) (1.12)
Age in 1993 (spline)
25-29 years -0.113 0.122 0.00001 0.055 0.563 '" 0.110 ” 0.158 “
(0.79) (0.77) (0.00) (1.01) (4.19) (2.05) (2.13)
30-39 years 0.069 0.105 ” 0.008 0.053 "‘ 0.028 -0.006 0.025
(1.51) (2.16) (0.18) (3.29) (0.81) (0.29) (1.24)
40-49 years -0.061 -0.035 ~0.078 ~0.050 “ -0.098 ‘ -0.041 -0.062 ”
(1.11) (0.55) (1.35) (2.23) (1.87) (1.45) (2.02)
50-52 years -0.228 -O.539 “ -0.234 0.110 0.188 -0.043 -0.024
(1.31) (2.02) (1.26) (1.15) (0.68) (0.33) (0.16)
“H Composition and Business Assets
11 men, aged 20-49 0298 ‘” —0.283 ” —0.268 “ -0.247 ‘” -0.343 “‘ -0.252 ‘” -0.206 ”‘
(2.73) (2.03) (2.38) (4.15) (2.97) (3.27) (3.12)
# women, aged 2049 -0.214 0.035 -0.187 -0.0002 -0.211 0.023 -0.223 ”
(1.51) (0.23) (1.30) (0.00) (1.53) (0.32) (2.55)
ii men, aged 50+ -0.645 ” -0.803 " -0.54l " -0.224 ‘ -0.105 -0.279 ” 0.379 ‘“
(2.35) (2.55) (1.97) (1.90) (0.45) (1.99) (3.08)
111 women, aged 50+ 0204 -0.012 -0.106 0.195 ‘ 0.137 0.221 ' -0.203
(0.92) (0.05) (0.48) (1 .80) (0.63) (1 .76) (1.14)
Business Assets (million) -0.003 -0.208 ‘” -i.l72 ”‘ 0.047 -0.018 -0.352 0.057
(0.14) (3.21) (3.38) (1.32) (0.57) (1.40) (1.41)
Constant 6.046 -4.281 2.764 —2.195 -2l.274 ’“ ~3.693 " ~5.150 ”
(1.50) (0.95) (0.72) (1.45) (5.72) (2.46) (2.27)
Wald test
Own Schooling 331.25 165.61
(0.000) (0.000)
Father's Schooling 17.38 37.18
(0.136) (0.002)
Mother’s Schooling 15.22 27.93
(0.085) (0.006)
Parental Schooling 43.78 63.23
(0.002) (0.000)
Age 72.54 63.84
(0.000) (0.000)
HH Composition 22.26 74.35
(0.035) (0.000)
HH Composition and Assets 42.03 82.12
(0.000) (0.000)
Month of interview 27.87 22.17
(0.006) (0.138)
Pseudo R ’ 0.171 0.098
Observations 2,797 3,722
Source: lFLSZ.

Base category for men is norparticipation and family worker sectors, and for women is norparticipation. Month of interview dummy variables are included
in the regressions but are not reported. Omitted category for own schooling is no schooling. Standard errors are robust to clustering at the community level
and heteroskedasticity. Absolute r-statistics are in parentheses. Signiﬁcant at the l%(‘“), 5%(") and 10%(‘) indicated. p -values for Wald test are in

parentheses.

152

Appendix Table 5.6C
Multinomial Logit for Sector Choice Model:
The Effects of Non-linear Own Schooling and Parental Schooling (Base Speciﬁcation)
Panel Respondents, iFLS3

 

 

 

 

 

 

Me_n _ Women 1
Self Public Private Self Public Private Family
Employment Sector Sector Employment Sector Sector Worker
Own Schooling
1-3 years 0.178 0326 0.234 0.089 0.638 -0.265 0.029
(0.54) (0.44) (0.67) (0.63) (0.98) (1.62) (0.19)
4-6 years 0.129 0.862 0.215 -0.110 0.468 «0.592 "' -0.450 ”
(0.47) (1.52) (0.70) (0.77) (0.64) (3.43) (2.86)
7-9 years -0.362 1.719 ” 0.037 -0.544 ” 0.947 -1.566 ‘” -0.842 ”‘
(1.14) (2.99) (0.10) (3.06) (1.31) (6.63) (3.67)
10-12 years -0.747 ” 2.958 ”‘ -0.116 -0.576 ” 4.486 ”‘ ~0.620 “ -1.133 ‘”
(2.21) (5.10) (0.32) (2.88) (7.47) (2.78) (3.76)
13+ years -1.168 “" 4.001 ”‘ -0.254
(2.57) (6.29) (0.57)
Father's Schooling
Some Elementary 1.172 ” 1.065 ” 1.134 ” 0.023 -0.090 -0.051 -0.l20
(2.80) (2.40) (2.66) (0.16) (0.27) (0.27) (0.69)
Completed Elementary 0.182 0.255 0.361 0.106 0.110 -0.185 -0314 s
(0.66) (0.80) (1.27) (0.78) (0.39) (1.06) (1.90)
Secondary/Tertiary -0.168 0.075 0.174 0.001 -0.367 -0.220 -0.860 ”
(0.45) (0.19) (0.47) (0.01) (1.12) (0.96) (2.74)
Missing -0.174 ~O.274 -0.034 0.003 -0.614 0.046 -0.153
(0.55) (0. 72) (0.11) (0.02) (1 .50) (0.23) (0.80)
Mother's Schooling
Some Elementary 0396 -0.305 -0.445 0.078 0.187 0.338 ‘ 0.085
(1.19) (0.81) (1.32) (0.52) (0.66) (1.78) (0.45)
Completed Elementary! -0.559 " 0.381 -0.117 0.169 0.067 0.280 -0.334 ‘
Secondary/Tertiary (1.92) (1.22) (0.39) (1.20) (0.24) (1.53) (1.89)
Missing -0.353 -0.184 0.017 -0.425 “ -1.025 " 0.161 -0.339
(0.95) (0.43) (0.05) (2.94) (1.96) (0.82) (1.61)
Age in 1993 (spline)
25-29 years -0.047 0.119 0.080 0.039 0.478 ‘“ 0.161 “ 0.079
(0.36) (0.76) (0.59) (0.84) (3.71 ) (2.66) (1.34)
30-39 years 0.093 ” 0.154 " 0.047 0.060 ” 0.059 ‘ 0.010 -0.009
(2.11) (3.16) (1.08) (3.26) (1 .67) (0.48) (0.45)
40-49 years -0.l40 ” -0.158 “ -0.190 ‘” -0.053 “ -0.117 " -0.078 " -0.023
(2.76) (2.45) (3.54) (2.29) (2.38) (2.39) (0.91)
50-52 years -0.242 -1.237 " -0.137 -0.026 0.142 -0.157 -0.169
(1.47) (3.29) (0.75) (0.27) (0.57) (0.99) (1.56)
1111 Composition and Business Assets
# men. aged 20.49 -0.233 "' -O.197 ‘ -O.24l ” -0.188 '” -0.427 " -0.180 " -0.097 ‘
(2.76) (1.81) (2.84) (4.07) (3.29) (2.84) (1.80)
8 women, aged 20-49 ~0.217 ” -0.186 -0.159 -0.049 -0.232 ' -0.l74 ” -0.252 ”‘
(2.07) ( 1.55) ( 1.40) (0.96) (1.91) (2.41) (3.86)
# men, aged 50+ -0.300 -0.230 -0.121 -0.351 " -0.625 “' ~0.33l "' 0.405 ‘”
(1.12) (0.69) (0.42) (3.38) (2.81) (2.49) (3.80)
8 women. aged 50+ 0224 -0.156 ~0.220 0.080 0.121 0.100 -0.216
(1.35) (0.78) (1.22) (0.71) (0.57) (0.73) (1.62)
Business Assets (million) 0.184 ” 0.142 ‘ -0.219 ‘ 0.019 -0.012 -0.129 0.027 "
(2.28) (1.70) (1.72) (1.59) (0.54) (1.63) (1.89)
Constant 4.461 -4.174 0.406 -1.258 -17.522 '” -4.479 ” -2.035
(1.20) (0.95) (0.11) (0.95) (4.73) (2.65) (1.23)
Wald test
Own Schooling 364.42 268.22
(0.000) (0.000)
Father’s Schooling 14.53 21.38
(0.268) (0.164)
Mother‘s Schooling 15.62 23.70
(0.075) (0.022)
Parental Schooling 48.09 53.45
(0.001) (0.003)
Age 91.45 66.39
(0.000) (0.000)
HH Composition 23.75 104.61
(0.022) (0.000)
HH Composition and Assets 45.31 1 15.79
(0.000) (0.000)
Month of interview 4.94 29.11
(0.960) (0.023)
Pseudo R ’ 0.180 0097
Observations 2,797 3,722
Source: iFLS3.

Base category for men is nonparticipation and family worker sectors, and for women is nonparticipation. Month of interview dummy variables are included
in the regressions but are not reported. Omitted category for own schooling is no schooling. Standard errors are robust to clustering at the community level
and heteroskedasticity. Absolute r-statistics are in parentheses. Signiﬁcant at the 1°/o("‘). S%(”) and 10%(") indicated. p -values for Wald test are in

parentheses.

153

 

.mMAE can NW4": .51—...: H3.50m

 

 

2: ad 2: md o._ od od no ma mm mo 06 mo 3050
we m6 we ad 5.2 Wm EN as m.m~ 9N ad ad md 2233
vs v.0 as 0.: a.: me we m.w_ 0mm v; ed 0: me..— BxEB cocoavoaﬁcozﬁoao
we Wm N.m 93 man wNn man can m.w~ men oé Na. mom c8103 one—nothx
ms ..c Na ma. v.2 ms a.m_ 5%.. 0.2 Em v.2 can «w team mauve—om
:. .6 «mm o._N Nam— md— ah. ha On «an ~.~ wsm 5.2 .mﬁm EoEoSooihoBm
«.2 Om. 0.2 Ym in m.o Yo we we m.m_ N.m~ No We .cSm cocabm_EEe<
0.2 a.m_ v.2 mg no No a.o w.w Nd v. G odm ma m: tam _~towa:a2\_~:o_mmomo£

amt—n:
a.a n: ma ma _._ o6 od _.o v6 9N ms .6 ed 82:0
Wm Nam ah 5m w.: _._ ON 06. 9mm 9o ON ad ch 38085
no men ..n a.m_ _.NN _.m We mam mNm ed as 92 WE coo—83 coneseoiEocﬁoaO
a.m ad we m.~m adm 5.2. m.~w 9mm a.o_ we as v.2“ Ngm 8x33 one—sothx
no rev ed m.m we... v; o.o ..2 we m._ v.2 9N : tﬂm moorCom
a.n ad ms 93 _.m_ o.m_ mm 9w 2: a._ Wm Ewm m.m~ beam EoEoSooiBoﬁm
m.: _.N_ 0: mé N6 _.o o.o 9w ad _.om aam _.o m.o .Cﬁm cosmbﬂEEv<
5:2 0&— mNH o.a m6 v.0 od 5w . we ode m._m m._ _._ tﬁm 35352220883:—

«mu—..:
od— o.N_ a.a 0.0 m6 0.0 Cd 0.0 Cd v.0 5m o.o o.o E050
ow m.m in _.m We. o._ _.m 9v MEN wd NM me mm £233
ad ad we w.~_ _.a_ 5v o.w_ Fem mom o._ ma Y: 0.3 .5453 cozunvok—EOSSQHO
m.m «am w.m 5am :.m Now a.mm asm N. 5 «am wh mam awn .8103 B§3tw<
Om Em ad ow 9w mm ..m 9.2 ma 9m ca ms o.~ tﬁm 822%
we ad ad odm mm.— a.a Qw— vé Om 0.0 _._ a._m VAN tﬂm EoEoSooktmo—mm
m: m.~_ a.o_ MN _.m _.0 od 5m o6 NS 5.3 _.o Nd beam cocmmeE—é
_.N_ _.N_ Na. o.w ..a We o6 mm ms 0.3 «.3 o._ m._ team Etowacavzﬁcoﬁmomoi

5dr..—

.3e.—. 5:33 :02 5:53 :92 5:83 :02 .3535 :02 5:33 .32 5:33 .52

 

19:32.5 be an> .M><

2.3336:— =< coo—.83 :.:—a..— .880m 8.3....— aeuoom 33:.— .aoﬁzsaﬁm =om

 

”532.5 ..e 33> 09:22 a: E... .3230 .3 Eon—.333”.— ue ..33m 55...» more: .03-Eu.—
E 2...... 52.2.3.

154

 

.mmAE 28 $45 .51—..: meadow

 

Nd d. _ dd dd dd ad md Wm dd _.d
#6 m._ _ a. ~ a.a d6 5mm N4. #6 Ev 56
dd w.m_ wd m.a d.w~ wéN nd mg.“ md~ m.m_
mam _.am dds. a.am adm wdm 5w vow QNN 2%
NC w.a mé Ndm wdm m. : Em WE dém dd
de 0: a.a m.m~ dd Wm _.N d._ Yam dd—
d.N 5v dd dd _.m dd «.3 d.NN md dd
md ad dd dd wé in ﬁnd wdm ad ad
Nd m._ dd dd Nd md m._ wd dd _.d
a.m d. 2 v; Nd. dd ﬁvm d._ m._ mm d.m
N2 th dd dd daN ﬁmm md m6 dd— 5.3
0mm wdm 0dr Nah ddm mdN Nam m.a m.- mém
dd Nd _._ dd <2 Nd md wd— _.N wd
hdm m.m_ 9: bd— md dd d4 N; mdw a._N
mm Nd dd dd dd d6. v.3 m.a~ _.d md
dd ad Nd dd md wd VS m.mm m._ d._
dd dd dd dd dd dd md ..v dd dd
van a.d_ N; NV mm a.am md «am 54... ..m
0.3 ddm 5v Nam mam a. _ m d. _ a.a wd_ Num—
bdm wdm v.5 Ném mmm d._N _.a Wm _._N wdm
m6 dd Wm dd dd. vd YN d.d_ Nam N.N
w.m~ Q: _.a 92 d6. Wm dd md ddm m.a~
wd Nd dd dd «6 9m Q: 5mm _.d _.d
_.a _.a Nd dd dd wd wdd v. :.. d. _ ad

€050

3233

50:03 :ocoadea—Eouﬁono
poo—53 BEEPCM<

.tSm mootcom

beam HEEouagaﬁmo—am

team :odmbm_:_Ed<

team 35355206835

mm:—

$050
8233

3x83 :oto:d9&:¢dﬁo:0
nod—:03 05:35"

.cSm moogm

tSm “EEoSothoﬁm
ocﬁm :onmbm_:_Ed<

.cSm Etowacmzdacowmmodoi
amt—..:

£050
”5.593

8x53 :Otuzguﬁgdﬁono
Coo—.83 one—sage.

beam 8333

tﬂm HEEoSooémoEm
team :Odabm_:_Ed<

tam .mtowaﬁgm:o_mm£oi

55:

 

5:83 :92 5:33 :02 5:33 :02 5:53 :02 5:33 :05

 

£2.2sz =< coo—.83 22:5...— ..33m 39;...— aeuoom 9.3.5 EoﬁmeEEm ...ew

 

«Endgame: .23.—
..od50 .3 «:oﬁzeaﬁm be ..33m :..—:.3 85:9 $88....—
3 2.3. 52.2.3.

155

 

 

 

E
3023.33

.5 m3 2 we an 8» 882530
as” as. at an 3: on; E :2. 2e 5:52
as as as 33 Si and as as as 85 seesaw

:... :m 3“ 83 as. 3.. «S; 9: S: an: :82 as» 3.
32 SN e 2 8a 5. 883630
a: an ace 8 08 as; E on. own 8:82
at as as d :s a5 a5 a. 3 do: as 85 28.5.6.

... E an and RN R: we? a- 93 So so: as» 3
RN on b 2 8n ma 868530
8a on” on EN 83 ma; 2. a; 8m 5:82
as as as a5 a3 a5 and as as h9cm Essen

... an :m 9.» a :3 one; 3” c2 3a :82 go» o

«WE.
Be an; an 3.“ SN 28 83 an»... 863530
2a a: New 8- 33 _ :... an- N _ e an 3 8m an .382
as as as aw: so: as and 33 a: as as as Bum 235m

:.. «on as. MS ea- $3 52 .. SN- E: a; en 23 on” :82 38
ea 22 8m an Na 8 S 9: 88.0530
2 N8 23 a- $3 om... at- So a9 ca «3 a: :28:
and and 8m: 83 av: 35 83 a3 :5 dos and ac: 85 285m

. sea 3.? as; a- 5a M: 5 new 82 E; on- «a: a} :8: care
9 SN 2 E : 8 2: a: 868530
:m an no on a? 83 ..: ﬁe Rm 8 a? av :28:
3.: :s as aed 3N: ac: awad a3: awd 5.3 5.: as: 85 28:3

... an a .3 .. 8.. as an. ca..- cam 83 an ma ”2; :82 as» a-”
E m? a a 8 SN am an 268530
ta ta as. a? ma. max 9:- s n a an on a? 566.2
:s as as am: as: an: ac: am : as at as 3s 5am 235%

:... Sm N2 o8 :.. men an R... :.. as. 08.. o3 a as ewe :82 no...» es
em. SN 2 2 me of Ra a: 883630
3; :~ Em men an 8e 2 5 E o 3a «3 5622
as as as a _ s at a _ s as: 5 d as as as as 85 235m

..: a: an as. . ca 08 oz; New 3; New 2 9... 2m :82 as:ﬂ 2
SN 2: o 2 we 2: 2a RN 823530
3 as. am _ :. ea 23 5 3a 3 m _- 5 cam :28:
as as as svs aos and dead do: SE S d at as 85 285m

..: am. :a es. ..: m? 23.. ﬂ ”5 v: 3. mm one :82 .8;

GE:

3555: 5:33 5.2 355::- 5:33 52 355:5 5:33 52 355:5 5:33 52
1.353 803:1.— 1. 3.05 0:31.— ?Eouci—wmecd :.3—ﬂims— =ow 3.3“.05 3033.13.”— :5

 

ﬁe a.a-ah.

5m :5 “...—00.3w .3 25> .3 25333:.m .3 .353 ...—:..: uuu3 :..—3:

156

.3885 Secs 93 asam 53.x; 2: a 383mm .32: 3.5858 2: a 9.5.3.0 9 2.52 ea :95 “523m
.33 Ba N3": :3": 688m

 

 

 

2.2 53 an N2 53a 23 28.2530
to N8 33 a2. ace 2? «6 8a 33 5:82
as as as ans Es aw: ass am: 2 _ s 85 263m

... 8m 85 3a 2 £3 33 ..: a: Sea :3. =8: .33
o; :.3 an 2.“ am So 88288
9a 2 _.a 33 o a _ 3. a .3 «8 3; 83 :28:
£5 a _ s :m : 85 a5 3a: aos ass 2 3 Sum 28.5w

am 83 a?“ 3 sin 23 an :3. 85 :82 esaé
as at. s 8 a8 Ev 28.2830
8m 33 $3 $2 83 83 can 82 83 .582
avs am: 5: aas aos is as cos ans 85 28:3

..: .8 was. 35 . a: Ea Sam ..: 8o; 82 :2 :82 e3» 3
2.. ME a R a: a8; 35530
o2 o: 83 max 33 £3 a? N8 «:2 .582
a. s as as ass :ms is aos 5.: a3 85 283m

... So 2 3 ME; ... as: sa; 8; ..: $0 $2 $3 :82 as» 3.
3. com o a NR 3v 882530
o: 23 8: 3% can 3a.. .3 ea RS 582
am: as an: Es :m 3 5s 5s aw: ass 8cm 323

... OS Ra 23 So a ms: : 8“ 23 2.3 :82 e8» 3
an a: v n Em «an 283030
5 a? 25 .86 as; :3 8. am 08 5.82
as as aos aas 83. : ats Ass :8 am: sum 226%

... an a: on? . 8%: :3 Si SM 25 caoa =82 5» o

85:
30 33 8m 3 as: 23 283030
Sq 3v Ea ma- 33 3: ms av sec 5.8.2
as as :s an s an s as as as as Sum 53m

... :é Ea 22 mm- a.a mama _. 5 83 $2 :82 .33
o8 20 3a a: s: mmm 28333
o £3 32 a? «a.a 23 a a: ca; :28:
av: A a: as aw: a. s as ass :3 am: 55 28:3

2 :3 ma; 8 $3 35 9a. #5 nova :82 38:2
ab 2; t 2 v2 mom 823530
a _ m mam 8 2 3 :3 m :2 SN 05 v8 5.32
:4: ac: as ams a _ s 6v: aas aos an: 55 235m

... 2w ME 83 an. aaoa «a... 2:- as. 25 :82 3% a-a

355:5 5:33 52 355:5 5:33 52 355:5 5:33 52 355:5 5:33 52
I335 8.3a Jo3om 9:3“— aﬁouiﬂeuv ...oElnoJQEH 20m 9.35.395 ...»Emodmﬁm ...om

 

5w 1.3 3:235 .3 25> .3 3083.43”.— .3 330m 55..» 033 233:
Geezgs 3 ~35

157

Table 6.2A

Selectivity Corrected Wage Functions:

The Effects of Non-linear Own Schooling

 

 

 

 

"-151
Men Women
Self Public Private Self Public Private
Employment Sector Sector Employment Sector Sector
Own Schooling
1-3 years 0.104 0.108 0.180 0.067
(1.18) (1.24) (1.50) (0.65)
4-6 years 0.453 ""‘ 0.501 "“ 0.509 *" 0.301 "
(5.04) (6.59) (4.32) (2.48)
7-9 years 1.024 ”" 0.157 0.781 ""‘ 0.739 "" 0.490 " 0.731 ""
(8.65) (1.15) (9.20) (4.68) (1.83) (4.58)
10-12 years 1.368 "" 0.563 “" 1.177 "“ 0.952 "" 1.453 ‘”" 1.707 ""‘
(8.48) (2.75) (13.32) (4.99) (3.11) (13.03)
13+ years 2.190 "" 0.921 "“ 1.793 ""
(7.63) (3.90) (13.05)
Age (spline)
25-29 years 0.042 0.003 0.074 "" 0.092 ‘ 0.118 0.102 ”
(1.07) (0.05) (2.71) (1.76) (1.61) (2.53)
30-39 years 0.011 0.044 "" 0.005 0.020 0.030 ” —0.010
(0.91) (3.50) (0.59) (1.25) (2.40) (0.68)
40-49 years -0.022 "" 0.019 0.014 -0.020 0.031 ‘“" 0.0005
(2.19) (1.46) (1 .61) (1.51) (2.06) (0.03)
50-59 years 0.014 -0.006 -0.055 "‘ 0.014 -0.011 -0.027
(0.94) (0.30) (2.66) (0.62) (0.32) (1 .04)
Month of interview
October 0.299 ” -0.139 0.118 0.371 " -0.083 0.041
(2.06) (0.93) (0.95) (2.53) (0.46) (0.28)
November 0.424 ‘" -0.251 0.144 0.403 “" -0.266 ‘ -0.128
(3.26) (1 .60) (1.19) (2.67) (1.78) (0.94)
December -0.094 -0.061 0.087 0.131 -0.141 -0. 147
(0.73) (0.42) (0.69) (0.88) (0.96) (1 .04)
January -0.031 0.069 0.259 " 0.090 0.175 -0.064
(0.18) (0.39) (1.67) (0.39) (1.03) (0.30)
Gross Income 0.076 0.331 ‘“
(1.10) (3.21)
Selection Coefﬁcient -0.6l4 ”" -0.101 0.077 0.862 "” 0.033 -0.055
(3.61) (0.57) (1 .05) (3.09) (0.12) (0.38)
Constant 4.449 "“" 6.474 "" 3.331 "* 1.295 2.465 2.512 "
(4.07) (3.65) (4.27) (0.82) (0.90) (2.21)
F-test
Own Schooling 25.27 9.63 62.71 1 1.98 5.25 52.81
(0.000) (0.000) (0.000) (0.000) (0.006) (0.000)
Age 2.15 7.73 5.81 2.78 14.37 2.22
(0.075) (0.000) (0.000) (0.027) (0.000) (0.068)
Month of interview 7.44 2.07 0.80 2.97 2.04 0.91
(0.000) (0.086) (0.526) (0.020) (0.093) (0.456)
Adjusted R 2 0.1 16 0.297 0.265 0.089 0.483 0.304
Root MSE 1 . 172 0.665 0.821 1.227 0.585 0.905
Observations 2,107 574 1,391 1,272 252 616
Source: IFLSl.

Omitted category for own schooling is no schooling (0-6 years for public sector workers) and for month of interview is
August/September. Adjustments are made due to small cell size for some variables for women: 10+ years for own
schooling. Selection coefficients are calculated based on Base Speciﬁcation estimates. Standard errors are robust to
clustering at the community level and heteroskedasticity. Absolute t-statistics are in parentheses. Signiﬁcant at the
l°/o(“"), 5°/o(”) and 10%(‘) indicated. p -values for F -test are in parentheses.

158

Table 6.28

Selectivity Corrected Wage Functions:

The Effects of Non-linear Own Schooling

 

 

 

 

lFLSZ
Men Women
Self Public Private Self Public Private
Employment Sector Sector Employment Sector Sector
Own Schooling
1-3 years 0.268 "" 0.113 0.191 " 0.122
(2.88) (1.25) (1.71) (1.01)
4-6 years 0.452 ‘" 0.351 ”‘ 0.270 ”" 0.430 ”"
(5.43) (4.68) (2.64) (3.85)
7-9 years 0.875 ”‘ 0.013 0.772 "" 0.693 *" 0.126 0.865 "‘
(8.59) (0.08) (8.90) (4.91) (0.50) (5.44)
10-12 years 1.079 "" 0.192 0.912 “" 1.033 "‘ -0.012 1.464 ""
(8.36) (0.76) (1 1.13) (6.66) (0.03) (12.56)
13+ years 1.914 "" 0.414 1.433 *“
(9.01) (1.31) (12.97)
Age (spline)
25-29 years -0.004 0.026 0.045 ‘“" -0.044 0.018 0.082 "W
(0.09) (0.30) (2.30) (0.83) (0.23) (2.72)
30-39 years 0.017 “ 0.010 0.022 """" 0.042 ""' 0.015 0.022 "
(1 .67) (0.58) (2.98) (2.68) (0.69) (1.77)
40-49 years -0.018 "' 0.026 "" 0.000 -0.024 " 0.027 0.019
(1 .83) (2.67) (0.05) (2.05) (1 .47) (1 .27)
50-59 years -0.009 0.006 -0.027 ‘ 0.011 0.012 -0.049 "
(0.74) (0.33) (1 .78) (0.74) (0.44) (1 .90)
Month of interview
October 0137 -O.178 " -0.041 -0.182 -0.247 “ 0.100
(1.24) (2.04) (0.53) (1.53) (2.12) (0.84)
November -0.211 “ -O.l84 ‘ -0.033 -0.091 -0.395 "’ 0.045
(1.92) (1.87) (0.45) (0.80) (2.72) (0.37)
December 0.038 -0.058 0.045 0.175 -O.151 0.010
(0.34) (0.61) (0.55) (1.19) (1.11) (0.08)
January-April -0.066 0.006 0.155 "' 0.271 0.061 0.334 ”
(0.47) (0.06) (1 .80) (1.18) (0.44) (2.04)
Selection Coefﬁcient -0.162 -0.313 -0.065 -0.074 -0.405 -0.796 "W
(0.84) (1 .41) (0.54) (0.29) (1.65) (3.09)
Constant 6.257 "* 6.874 " 4.914 ”" 7.119 *" 7.662 "'" 4.296 "‘
(4.60) (2.43) (8.37) (4.42) (2.70) (5.41)
F-test
Own Schooling 23.22 2.63 59.02 13.97 0.30 49.22
(0.000) (0.050) (0.000) (0.000) (0.738) (0.000)
Age 2.40 5.98 10.05 1.95 4.32 6.55
(0.050) (0.000) (0.000) (0.101) (0.002) (0.000)
Month of interview 1.86 2.13 1.76 2.62 3.63 1.42
(0.1 16) (0.077) (0.137) (0.035) (0.007) (0.228)
Adjusted R 2 0.102 0.219 0.196 0.063 0.203 0.298
Root MSE 1.158 0.642 0.820 1.232 0.598 0.950
Observations 2,3 1 8 645 2,099 1,480 309 992
Source: lFLSZ.

Omitted category for own schooling is no schooling (0-6 years for public sector workers) and for month of interview is
August/September. Adjustments are made due to small cell size for some variables for women: 10+ years for own
schooling. Selection coefﬁcients are calculated based on Base Speciﬁcation estimates. Standard errors are robust to
clustering at the community level and heteroskedasticity. Absolute l-statistics are in parentheses. Signiﬁcant at the
1%(""), 5%(") and 10°/o("‘) indicated. p -va1ues for F -test are in parentheses.

159

Table 6.2C
Selectivity Corrected Wage Functions:

The Effects of Non-linear Own Schooling

 

 

 

IFLS3
Men Women
Self Public Private Self Public Private
Employment Sector Sector Employment Sector Sector
Own Schooling
1-3 years 0.215 " 0.191 “ 0.260 " 0.335 ’”
(2.06) (1 .98) (2.40) (3.31)
46 years 0.390 ‘" 0.387 "" 0.314 ”“ 0.419 ""'
(4.21) (4.76) (3.32) (4.12)
7-9 years 0.698 ”"‘ 0.209 0.646 '" 0.535 "" 0.331 0.867 ""
(6.42) (1 .26) (7.42) (4.83) (1.03) (5.68)
10-12 years 1.053 "”" 0.637 “ 0.915 """' 0.824 *" 0.355 1.492 “"
(8.54) (2.22) (10.69) (6.69) (0.61) (13.79)
13+ years 1.652 “" 0.978 ” 1.424 ““"
(10.19) (2.49) (14.06)
Age (spline)
25-29 years 0.035 0.070 0.005 0.097 " 0.062 0.015
(1.22) (1 .06) (0.32) (2.01) (0.60) (0.55)
30-39 years 0.009 0.043 " 0.018 "" 0.017 -0.044 0.013
(0.99) (2.05) (2.95) (1.32) (1.18) (1.24)
40-49 years 0.006 0.025 "' 0.007 0.005 0.028 " 0.004
(0.67) (2.24) (0.90) (0.46) (1 .70) (0.32)
50-59 years -0.031 ‘" 0.015 -0.032 " -0.019 0.051 " -0.013
(2.73) (0.64) (2.54) (1.28) (2.15) (0.59)
Month of interview
August 0171 " 0.010 0.016 -0.113 0.065 -0.032
(2.26) (0.10) (0.30) (1.05) (0.48) (0.30)
September -0. 136 -0.062 0.013 -0. 106 -0.027 -0.046
(1 .54) (0.69) (0.25) (0.96) (0.21) (0.40)
October -0.097 0.004 0.135 “ -0.098 0.150 0.013
(1 .01) (0.04) (2.44) (0.88) (0.96) (0.1 1)
November-January 0.022 0.150 0.223 "" -0.074 0.064 0.200 "
(0.22) (1 .28) (3.78) (0.57) (0.43) (1.75)
Selection Coefﬁcient -0.487 "" -0.035 0.034 0.363 -0.729 " -0.267
(3.02) (0.13) (0.44) (1 .24) (2.26) (1 .02)
Constant 6.249 “* 5.186 ” 6.448 ‘" 3.328 ” 7.376 " 6.009 ""'
(7.20) (2.30) (14.89) (2.11) (2.31) (8.52)
F -test
Own Schooling 24.61 2.51 76.49 12.87 0.53 56.60
(0.000) (0.05 8) (0.000) (0.000) (0.591) (0.000)
Age 3.16 5.46 6.99 2.46 3.81 1.17
(0.014) (0.000) (0.000) (0.045) (0.005) (0.321)
Month of interview 1.88 1.37 5.17 0.32 0.55 2.27
(0.1 1 1) (0.242) (0.000) (0.866) (0.697) (0.060)
Adjusted R 2 0.075 0.249 0.181 0.042 0.299 0.263
Root MSE 1.152 0.682 0.810 1.242 0.673 0.946
Observations 3,023 702 2,801 2,047 352 1,376

 

Source: IFLS3.

Omitted category for own schooling is no schooling (0-6 years for public sector workers) and for month of interview is
June/July. Adjustments are made due to small cell size for some variables for women: 10+ years for own schooling.
Selection coefﬁcients are calculated based on Base Speciﬁcation estimates. Standard errors are robust to clustering at
the community level and heteroskedasticity. Absolute t-statistics are in parentheses. Signiﬁcant at the 1%("”),
5%("”") and lO°/o(‘) indicated. p -values for F -test are in parentheses.

160

 

 

Table 6.3
Selectivity Corrected Wage Functions:
Summary of the Effects of Own Schooling

 

 

 

Men Women
Self Public Private Self Public Private

Employment Sector Sector Employment Sector Sector
A. Cross Section Sample
Linear Own Schooling
lFLSl 0.125 "" 0.080 ""'* 0.108 "" 0.079 "“ 0.081 0.123 ‘”
IFLSZ 0.089 “" 0.039 0.087 “"' 0.076 "'* -0.019 0.116 *"
IFLS3 0.087 "‘ 0.054 0.084 “" 0.060 “* 0.044 0.114 "'""
Non-linear Own Schooling
lFLSl
1-3 years a) 0.104 0.108 0.180 0.067
4-6 years 0.453 "“" 0.501 "" 0.509 "* 0.301 "
7-9 years 1.024 "* 0.157 0.781 "‘ 0.739 ‘" 0.490 ‘ 0.731 “*
10-12 years b) 1.368 ""‘ 0.563 "" 1.177 ”" 0.952 "" 1.453 “* 1.707 ”*
13+ years 2.190 "" 0.921 "" 1.793 ""
IFLSZ
l-3 years a) 0.268 """ 0.113 0.191 ‘ 0.122
4-6 years 0.452 ""‘ 0.351 *“ 0.270 "“" 0.430 "'"
7-9 years 0.875 ""‘ 0.013 0.772 *" 0.693 ”" 0.126 0.865 “"
10-12 years b) 1.079 "'" 0.192 0.912 "’" 1.033 "" -0.012 1.464 *"
13+ years 1.914 "" 0.414 1.433 ""
IFLS3
1-3 years a) 0.215 "”' 0.191 *" 0.260 " 0.335 "*
4-6 years 0.390 ”" 0.387 “"“ 0.314 "”“ 0.419 *"
7-9 years 0.698 "" 0.209 0.646 ”" 0.535 "* 0.331 0.867 "*
10-12 years b) 1.053 ”"' 0.637 " 0.915 "" 0.824 ‘”" 0.355 1.492 "*
13+ years 1.652 ”‘* 0.978 " 1.424 "'"
B. Panel Sample
Linear Own Schooling
lFLSl 0.101 ”" 0.072 ” 0.092 "* 0.054 *” 0.052 0.116 ”"
IFLSZ 0.109 "* 0.038 0.088 "”" 0.081 *" 0.013 0.115 ““
IFLS3 0.097 "“ 0.080 " 0.089 "'" 0.073 "" 0.125 "" 0.095 *"
Non-linear Own Schooling
IFLS]
1-3 years a) 0.092 0.120 0.049 0.172
4-6 years 0.296 ""‘ 0.387 "‘ 0.355 " 0.294 “
7-9 years 0.821 "'" 0.251 0.680 "" 0.543 *" 0.291 0.691 *"
10-12 years b) 1.124 "" 0.816 “" 1.012 "W 0.497 " 1.142 ” 1.652 "'"
13+ years 1.925 *” 1.103 ”" 1.523 ""‘
IFLSZ
1-3 years a) 0.368 ”" 0.120 0.134 0.075
4-6 years 0.405 **"‘ 0.407 ““ 0.235 " 0.232 "
7-9 years 0.964 ‘"” 0.010 0.876 "'" 0.774 "" 0.192 0.463 """
10-12 years b) 1.391 “'" 0.558 "‘ 0.961 "" 1.153 ”" 0.501 1.687 ""
13+ years 2.079 "i' 0.796 '” 1.332 ""
IFLS3
1-3 years a) 0.288 " 0.350 " 0.128 0.291 '"‘
4.6 years 0.411 ““ 0.427 ”'" 0.214 " 0.278 "
7-9 years 0.765 ”" 0.172 0.864 '" 0.490 "'" 0.343 0.654 "‘
10-12 years b) 1.277 ‘” 0.793 " 1.074 "" 1.113 ”‘ 1.773 "" 1.366 ""
13+ years 2.217 *"'"‘ 1.081 "' 1.595 ""

 

Source: Based on estimates of Appendix Table 6.1A, 6.1B, 6.1C, Table 6.2A, 6.28, 6.2C, Appendix Table 6.6A, 6.6B,

6.6C, and Appendix Table 6.7A, 6.7B, 6.7C.

a) Omitted category for public sector workers is 0-6 years.

b) 10+ years for women.

161

 

Table 6.4A
Selectivity Corrected Wage Functions:
The Effects of Non-linear Own Schooling and Parental Schooling

 

 

 

lFLSi
Men Wom_e_n
Self Public Private Self Public Private
Employment Sector Sector Employment Sector Sector
Own Schooling
1-3 years 0044 0.100 0.139 0.047
(0.50) (1.11) (1.16) (0.43)
4-6 years 0.361 ‘" 0.478 ”“ 0.413 ”’ 0.243 ‘
(4.06) (6. 10) (3.46) (1.95)
7-9 years 0.922 ‘” 0.140 0.743 “‘ 0.584 '“ 0.450 0.634 ”"
(7.86) (1.02) (8.23) (3.58) (1.50) (3.76)
10-12 years 1.259 ‘" 0.497 ” 1.096 ‘" 0.718 "" 1.160 ‘ 1.507 ‘"
(7.87) (2.32) (11.32) (3.67) (1 .82) (8.47)
13+ years 2.026 ”" 0.826 “‘ 1.677 ‘"
(7.17) (3.30) (11.56)
F ather's Schooling
Some Elementary 0.191 " 0.111 0.006 0.105 -0.040 0.104
(2.32) (1 . 1 1) (0.08) (0.84) (0.24) (0.89)
Completed Elementary 0.170 ‘ 0.006 0.051 0.159 -0.237 0.193
(1.68) (0.06) (0.63) (1.35) (1 .54) (1.45)
Secondaryffeniary 0.208 -0.028 0.158 0.264 -0.006 0.232
(1.26) (0.26) (1.64) (1.38) (0.04) (1.12)
Missing 0.089 0.051 -0.023 -0.071 -0.084 0.124
(0.98) (0.40) (0.24) (0.62) (0.49) (1.04)
Mother's Schooling
Some Elementary 0.189 “ —0.021 0.021 0.148 0.111 0.026
(2.01 ) (0.23) (0.28) (1.18) (1 .04) (0.21)
Completed Elementary/ 0.376 "‘ 0.164 "' 0.102 0.281 " 0.152 0.123
Secondary/Tertiary (3.42) (1 .98) (1 .24) (1 .97) (1 .25) (0.82)
Missing 0.180 ” 0.089 0.073 0.098 0.080 -0.076
(2.29) (0.85) (0.85) (0.90) (0.50) (0.60)
Age (spline)
25-29 years 0.043 0.000 0.074 ‘" 0.081 0.084 0.104 “‘
(1.07) (0.01) (2.72) (1.53) (0.92) (2.53)
30-39 years 0.014 0.041 ”" 0.005 0.027 0.035 " -0.008
(1.24) (3.33) (0.69) (1.63) (2.53) (0.55)
40-49 years 0020 ‘ 0.020 0.014 -0.020 0.031 ” 0.003
(1.94) (1.51) (1.57) (1.46) (2.08) (0.17)
50-59 years 0.014 -0.004 -0.055 "“ 0.013 0.009 -0.028
(0.99) (0.19) (2.63) (0.61) (0.27) (1 .12)
Gross income 0.091 0.325 ‘”
(1.33) (3.20)
Selection Coefﬁcient —0.760 “" -0.154 0.105 0.845 ““ -0.120 -0.052
(4.19) (0.85) (1.27) (2.82) (0.29) (0.33)
Constant 4.352 ”“ 6.608 ”" 3.273 '” 1.509 3.847 2.411 "‘
(3.93) (3.73) (4.22) (0.94) (1 .03) (2.06)
F-test
Own Schooling 21.58 7.21 44.84 6.49 1.65 20.45
(0.000) (0.000) (0.000) (0.000) (0.196) (0.000)
Father's Schooling 1.64 0.67 1.01 1.04 2.22 0.71
(0.165) (0.614) (0.400) (0.387) (0.070) (0.589)
Mother's Schooling 5.07 1.88 0.65 1.44 0.75 0.41
(0.002) (0.134) (0.585) (0.2 32) (0.521) (0.744)
Parental Schooling 5.38 l .08 1.33 1.82 1.59 0.98
(0.000) (0.378) (0.236) (0.083) (0.143) (0.443)
Age 2.11 7.20 6.01 2.76 13.54 2.32
(0.080) (0.000) (0.000) (0.028) (0.000) (0.058)
Adjusted R 2 0.130 0.297 0.266 0.095 0.484 0.303
Root MSE 1.162 0.665 0.820 1.223 0.584 0.906
Observations 2,107 574 1,391 1,272 252 616

 

Source: lFLSl.

Month of interview dummy variables are included in the regressions but are not reported. Omitted category for own
schooling is no schooling (0-6 years for public sector workers) and for parental schooling is no schooling. Adjustments are
made due to small cell size for some variables for women: 10+ years for own schooling. Selection coefﬁcients are
calculated based on Base Speciﬁcation estimates. Standard errors are robust to clustering at the community level and
heteroskedasticity. Absolute t-statistics are in parentheses. Signiﬁcant at the 1%(‘”), 5%(“) and 10%(‘) indicated. p-
values for F -test are in parentheses.

162

 

 

Table 6.43
Selectivity Corrected Wage Functions:
The Effects of Non-linear Own Schooling and Parental Schoollng

 

 

 

 

IFLSZ
Men Women
Self Public Private Self Public Private
Employment Sector Sector Employment Sector Sector
Own Schoollng
1-3 years 0.171 ‘ 0.106 0.194 "' 0.028
(1.87) (1.18) (1.65) (0.23)
4-6 years 0.354 "‘ 0.323 "" 0.230 “ 0.194
(4.26) (4.33) (2.05) (1.53)
7-9 years 0.841 ‘” 0.169 0.699 "*" 0.544 "" 0.114 0.461 "
(8.56) (1 .23) (7.99) (3.69) (0.38) (2.36)
10-12 years 1.173 ""' 0.546 “‘ 0.793 ‘" 0.770 "" 0132 1.170 ”"
(8.99) (2.67) (9.46) (4.60) (0.20) (8.56)
13+ years 2.121 “" 0.871 "‘ 1.220 ‘“”
(10.16) (3.47) (10.31)
Father's Schoollng
Some Elementary 0.280 *" -0.042 0.052 0.165 " -0.114 0.024
(3.03) (0.43) (0.79) (1 .70) (0.68) (0.19)
Completed Elementary 0.179 " 0.033 0.097 0.314 '" -0.135 0.097
(1.80) (0.34) (1 .49) (2.70) (0.95) (0.80)
Secondary/1‘ ertiary 0.494 "‘ -0.148 0.219 ‘“ 0.505 ”" -0.070 0.369 “
(3 .59) (1 .30) (2.69) (2.82) (0.47) (2.40)
Missing 0.270 " .0142 0.053 0.115 -0.208 0.047
(2.1 1) (0.72) (0.62) (0.83) (0.91) (0.39)
Mother's Schoollng
Some Elementary 0.064 0.059 0.101 " 0.073 0.113 0.009
(0.70) (0.65) (1 .70) (0.59) (1.01) (0.08)
Completed Elementary/ 0.373 "" 0.162 " 0.207 ‘” -0.084 0.062 0.228 ”
Secondary/1‘ ertiary (3.72) (2.13) (3.41) (0.59) (0.49) (2.28)
Missing 0.048 0.057 0.130 0189 0.270 0.127
(0.40) (0.29) (1.58) (1.17) (0.91) (1.03)
Age (spline)
25-29 years 0062 0.069 0.049 " -0.031 0.013 0.055 ‘
(1 .44) (0.83) (2.54) (0.57) (0.15) (1.74)
30-39 years 0.010 0.030 ‘ 0.019 ”" 0.051 "" 0.006 0.008
(0.96) (1 .92) (2.61) (3.12) (0.20) (0.63)
40-49 years 0022 ” 0.031 “" —0.003 -0.021 " 0.033 0.019
(2.25) (3.15) (0.33) (1.81) (1.34) (1.26)
50-59 years 0.001 -0.007 -0.027 ‘ 0.012 0.009 -0.076 “‘
(0.06) (0.40) (1 .80) (0.80) (0.31) (2.82)
Selection Coefﬁcient -0.922 “" 0.016 0.150 0.181 -0.491 -0.074
(4.46) (0.09) (1.15) (0.58) (1 .27) (0.25)
Constant 8.443 "" 4.718 " 4.523 ”" 6.259 ‘" 8.126 ” 4.061 “"'
(6.52) (1 .79) (7.85) (3.66) (2.26) (5. 10)
F-test
Own Schooling 27.55 6.48 37.45 6.50 0.39 30.30
(0.000) (0.000) (0.000) (0.000) (0.675) (0.000)
Father's Schooling 4.02 2.08 1.86 2.54 0.49 2.32
(0.003) (0.083) (0.1 15) (0.040) (0.747) (0.057)
Mother’s Schooling 5.24 1.54 3.99 1.01 0.50 2.27
(0.001) (0.204) (0.008) (0.388) (0.685) (0.081)
Parental Schooling 6.96 1.92 5.56 2.63 0.39 3.83
(0.000) (0.066) (0.000) (0.012) (0.908) (0.001)
Age 2.13 7.72 8.67 2.48 3.56 4.60
(0.076) (0.000) (0.000) (0.044) (0.008) (0.001)
Adjusted R 2 0.121 0.223 0.207 0.070 0.190 0.306
Root MSE 1.146 0.641 0.814 1.227 0.603 0.945
Observations 2,3 1 8 645 2,099 1 .480 309 992
Source: IFLSZ.

Month of interview dummy variables are included in the regressions but are not reported. Omitted category for own
schooling is no schooling (0-6 years for public sector workers) and for parental schooling is no schooling. Adjustments are
made due to small cell size for some variables for women: 10+ years for own schooling. Selection coefﬁcients are
calculated based on Base Speciﬁcation estimates. Standard errors are robust to clustering at the community level and
heteroskedasticity. Absolute t-statistics are in parentheses. Signiﬁth at the l%(“‘), 5%(") and 10%(") indicated. p -
values for F -test are in parentheses.

163

 

Table 6.4C
Selectivity Corrected Wage Functions:
The Effects of Non-linear Own Schoollng and Parental Schooling

 

 

 

IFLSJ
Men Women
Self Public Private Self Public Private
Employment Sector Sector Employment Sector Sector
Own Schooling
1-3 years 0.186 ‘ 0.193 “ 0.219 ” 0.280 ""
(1.82) (2.01) (1.99) (2.64)
4-6 years 0.354 ‘” 0.377 “" 0.246 " 0.334 ‘"
(3.92) (4.66) (2.59) (3.19)
7-9 years 0.677 "" 0.286 0.618 ”" 0.435 ”"' 0.168 0.590 “‘
(6.36) (1 .58) (7.13) (3.77) (0.50) (3.49)
10-12 years 1.086 ‘” 0.821 “ 0.847 "”‘ 0.692 ”" -0.107 1.205 ‘“
(8.62) (2.38) (9.83) (5.08) (0.14) (9.53)
13+ years 1.644 "‘ 1.218 "" 1.275 ”"
(9.96) (2.61) (12.09)
F ather's Schooling
Some Elementary -0.072 -0.053 -0.024 0.052 -0.049 -0.031
(1.07) (0.53) (0.42) (0.51) (0.26) (0.34)
Completed Elementary 0.032 0.002 0.025 0.089 -0.007 0.025
(0.46) (0.03) (0.50) (1.06) (0.04) (0.27)
Secondary/Tertiary 0.145 0.083 0.117 ” 0.246 ” 0.168 0.460 “"
(1.52) (0.85) (2.04) (2.06) (0.94) (4.40)
Missing —0.158 ‘ -0.056 -0.061 0.120 0.276 0.081
(1.92) (0.35) (0.92) (1.20) (1.06) (0.86)
Mother's Schooling
Some Elementary 0.160 "' 0.001 0.030 0.034 0.047 -0.079
(2.24) (0.01) (0.57) (0.34) (0.45) (0.91)
Completed Elementary/ 0.325 ”‘ 0027 0.103 ""' 0.130 0.013 0.061
Secondaryf Tertiary (4.31) (0.34) (2.1 1) (1.40) (0.12) (0.65)
Missing 0.298 ”" -0.195 0.122 0.100 -0.192 «0.109
(3.09) (1.15) (1 .63) (0.88) (0.88) (1 .06)
Age (spline)
25-29 years 0.021 0.087 0.006 0.082 " 0.051 -0.005
(0.70) (1 .24) (0.37) (1.65) (0.48) (0.20)
30-39 years 0.008 0.056 " 0.017 “" 0.013 -0.068 0.011
(0.85) (2.33) (2.71) (0.99) (1.41) (1.07)
40-49 years 0.005 0.030 ” 0.006 0.004 0.038 ' -0.004
(0.57) (2.38) (0.72) (0.29) (1 .84) (0.32)
50-59 years 0030 "‘ 0.008 -0.035 ”‘ -0.017 0.052 ” -0.025
(2.60) (0.31) (2.75) (1.10) (2.25) (1.16)
Selection Coefﬁcient -0.899 ”‘ 0.147 0.148 0.128 -0.962 “ 0.315
(4.65) (0.45) (1.52) (0.38) (2.15) (1.10)
Constant 6.936 “" 4.186 " 6.302 ”‘ 4.048 " 8.509 “' 5.805 ”"
(7.70) (1.66) (14.30) (2.40) (2.27) (8.39)
F-test
Own Schooling 23.91 2.75 50.82 7.34 0.32 28.12
(0.000) (0.043) (0.000) (0.000) (0.730) (0.000)
Father's Schooling 2.14 0.55 1.86 1.21 1.05 7.71
(0.075) (0.700) (0.1 15) (0.306) (0.384) (0.000)
Mother’s Schooling 7.59 0.53 1.93 0.78 0.46 1.11
(0.000) (0.663) (0.123) (0.503) (0.709) (0.343)
Parental Schooling 5.41 0.63 2.87 1.65 0.90 5.03
(0.000) (0.728) (0.006) (0.1 19) (0.507) (0.000)
Age 2.39 5.94 6.60 1.48 4.16 0.91
(0.050) (0.000) (0.000) (0.206) (0.003) (0.458)
Adjusted R 2 0.086 0.247 0.185 0.044 0.297 0.277
Root MSE 1.146 0.682 0.808 1.240 0.674 0.937
Observations 3,023 702 2,801 2,047 352 1,376

 

Source: 1F LS3.

Month of interview dummy variables are included in the regressions but are not reported. Omitted category for own
schooling is no schooling (0-6 years for public sector workers) and for parental schooling is no schooling. Adjustments are
made due to small cell size for some variables for women: 10+ years for own schooling. Selection coefﬁcients are
calculated based on Base Speciﬁcation estimates. Standard errors are robust to clustering at the community level and
heteroskedasticity. Absolute t-statistics are in parentheses. Signiﬁcant at the l%("‘), S%(”) and 10%(‘) indicated. p -
values for F -test are in parentheses.

164

 

 

Table 6.5
Selectivity Corrected Wage Functions:
Summary of the Effects of Non-Linear Own Schooling and Parental Schooling

 

 

 

Men Women
Self Public Private Self Public Private

Employment Sector Sector Employment Sector Sector
A. Cross Section Sample
IFLS]
1-3 years a) 0.044 0.100 0.139 0.047
4-6 years 0.361 ‘” 0.478 “" 0.413 "“" 0.243 "
7-9 years 0.922 *" 0.140 0.743 *" 0.584 ""' 0.450 0.634 "‘
10—12 years b) 1.259 "" 0.497 *" 1.096 "" 0.718 ““ 1.160 " 1.507 ""
13+ years 2.026 "“ 0.826 "* 1.677 "”"
IF L82
1-3 years a) 0.171 " 0.106 0.194 "' 0.028
4-6 years 0.354 "”" 0.323 "'" 0.230 " 0.194
7-9 years 0.841 ‘“" 0.169 0.699 "* 0.544 *"”" 0.114 0.461 "‘"'
10-12 years b) 1.173 """' 0.546 "‘ 0.793 *" 0.770 "* -0.132 1.170 *"
13+ years 2.121 ”" 0.871 "" 1.220 ”"'
1F LS3
1-3 years a) 0.186 ‘ 0.193 "”" 0.219 "”" 0.280 ""
4-6 years 0.354 "‘" 0.377 “" 0.246 "' 0.334 *"
7-9 years 0.677 ‘” 0.286 0.618 *” 0.435 “'“ 0.168 0.590 "*
10-12 years b) 1.086 "" 0.821 "”' 0.847 "”” 0.692 "“" -0.107 1.205 ""
13+ years 1.644 "“ 1.218 "" 1.275 "*
3. Panel Sample
iFLSl
1-3 years a) 0.019 0.137 0.020 0.164
4-6 years 0.196 "' 0.385 ""' 0.247 0.239 “
7-9 years 0.685 *” 0.287 0.670 ""” 0.360 "' 0.358 0.585 "”""‘
10-12 years b) 0.979 ‘" 0.902 " 0.969 """" 0.258 1.188 1.440 “"
13+ years 1.745 ”" 1.196 "'" 1.452 "'"
IFLSZ
1-3 years a) 0.303 "’ 0.112 0.116 0.069
4-6 years 0.309 "" 0.353 "" 0.158 0.111
7-9 years 0.846 ""‘ 0.214 0.771 ""' 0.628 "" 0.405 0.226
10-12 years b) 1.289 "" 1.028 "”" 0.776 '1‘” 0.921 “" 1.237 "‘ 1.332 ‘"
13+ years 2.038 "" 1.385 "”" 1.097 ”"
iFLS3
1-3 years a) 0.228 "' 0.345 “ 0.105 0.305 "
4-6 years 0.324 *" 0.399 "" 0.167 0.252 "‘
7-9 years 0.696 "" 0.199 0.810 ”"‘ 0.428 ‘“ 0.277 0.490 "'
10-12 years b) 1.281 *" 0.834 " 0.968 "'" 1.084 "" 1.544 "" 1.140 """"
13+ years 2.302 "“ 1.153 “ 1.406 “"

 

Source: Based on estimates of Table 6.4A, 6.4B, 6.4C and Appendix Table 6.8A, 6.88, 6.8C.

a) Omitted category for public sector workers is 0-6 years.

b) 10+ years for women.

165

 

 

Table 6.6A
Selectivity Corrected Wage Functions:
The Effects of Non-linear Own Schooling, Parental Schoollng and Residency

 

 

 

 

IFLS]
Men Women
Self Public Private Self Public Private
Employment Sector Sector Employment Sector Sector
Own Schooling
1-3 years 0.017 —0.026 0.053 -0.026
(0.21 ) (0.33) (0.48) (0.25)
4-6 years 0.270 ”‘ 0.271 "" 0.377 ”" 0.109
(3.25) (3.91) (3.36) (0.97)
7-9 years 0.760 "‘ 0.139 0.441 "'“ 0.510 “‘ 0.125 0.423 ”
(7.09) (1 .05) (5.19) (3.1 1) (0.43) (2.23)
10-12 years 1.095 "‘" 0.504 " 0.776 "" 0.645 "" 0.393 1.189 ”"
(7.18) (2.58) (8.99) (3.82) (0.69) (6.85)
13+ years 1.797 "" 0.850 ‘” 1.327 "“"
(6.43) (3.83) (10.74)
Father's Schoollng
Some Elementary 0.169 " 0.093 -0.027 0.113 -0.118 -0.013
(2.20) (0.94) (0.42) (0.99) (0.71) (0.1 1)
Completed Elementary 0.158 " 0.023 -0.048 0.062 -0.290 ‘ 0.157
(1.69) (0.22) (0.65) (0.54) (1 .76) (1 .12)
Secondary/Tertiary 0.020 —0.006 0.007 0.285 " -0.005 0.166
(0.13) (0.05) (0.07) (1 .68) (0.03) (0.86)
Missing 0.047 0.015 -0.054 -0.07 8 -0.017 0.074
(0.55) (0.12) (0.62) (0.71) (0.11) (0.63)
Mother's Schooling
Some Elementary 0.180 " -0.074 0.062 0.213 " 0.145 0.102
(1.98) (0.83) (0.86) (1 .72) (1 .26) (0.82)
Completed Elementary/ 0.314 "“ 0.047 0.096 0.201 0.141 0.073
Secondary/Tertiary (2.95) (0.52) (1 .32) (1 .49) (1 .08) (0.50)
Missing 0.199 ‘“ 0.071 0.061 0.166 0.005 -0.064
(2.68) (0.74) (0.76) (1 .62) (0.03) (0.53)
Age (spline)
25-29 years 0.031 —0.005 0.064 “‘ 0.065 0.013 0.095 '"'
(0.78) (0.09) (2.63) (1 .35) (0.14) (2.26)
30-39 years 0.008 0.042 "' -0.003 0.009 0.026 "‘ -0.012
(0.69) (3.37) (0.42) (0.57) (1 .91) (0.88)
40-49 years 0019 “ 0.015 0.010 -0.014 0.033 “ 0.009
(2.05) (1.18) (1.26) (1 .04) (2.14) (0.60)
50-59 years 0.012 -0.006 -0.040 “ 0.007 0.011 -0.045 ‘
(0.86) (0.29) (2.13) (0.30) (0.32) (1.76)
Province of Residence
North Surnatera -1.263 "‘ -0.060 -0.567 " -1.381 "‘ -0.100 -0.469 ”“
(5.86) (0.45) (2.53) (5.07) (0.37) (2.88)
North Surnatera " Urban 1.107 "‘ -0.521 "‘ 0.109 0.999 "" -0.247 0.021
(4.50) (2.73) (0.47) (3.23) (0.84) (0.12)
West Surnatera -0.686 “" -0.030 -0.590 ‘” -0.998 "" -0.002 -0.593 ""
(3.62) (0.18) (5.04) (2.84) (0.01) (2.76)
West Surnatera “ Urban 0.024 0.090 0.102 0.393 0.040 -0.166
(0.11) (0.43) (0.82) (0.95) (0.23) (0.54)
South Surnatera -1.381 “" -0.413 -1.002 “" -l.406 "" -0.612 ” -0.209
(6.69) (1 .27) (3.62) (6.22) (2.08) (0.54)
South Sumtera "' Urban 0.865 ‘” 0.666 " 0.768 “ 1.037 "‘ 0.457 -0.173
(4.19) (1 .98) (2.61) (3.27) (1 .63) (0.43)
Lampung —1.714 "'" 0.206 -1.422 ‘" -1.175 ”" -0.826 ”" -0.765 “'
(9.09) (1 .09) (7.37) (4.26) (2.66) (2.07)
Lampung ‘ Urban 0.716 ”‘ -1.325 ‘" 0.497 ”
(2.88) (6.68) (2.05)
West Java -1.373 ‘” —0.202 -0.998 “" -0.885 ““ -0.472 " -0.983 "‘
(6.45) (0.85) (5.35) (4.34) (2.26) (4.51)
West Java " Urban 0.684 ""' -0.063 0.550 "" 0.472 ” 0.269 0.454 "
(3.90) (0.25) (3.37) (2.50) (1 .56) (2.12)
(continued)

166

 

Table 6.6A (continued)
Selectivity Corrected Wage Functions:
The Effects of Non-linear Own Schooling, Parental Schoollng and Residency

 

 

 

lFLSl
Men Women
Self Public Private Self Public Private
Employment Sector Sector Employment Sector Sector
Central Java -1.250 ‘“ -0.097 -0.935 "‘ -1.353 "" -0.022 -0.936 ‘"
(8.57) (0.51) (8.63) (8.29) (0.11) (4.80)
Central Java ‘ Urban 0.523 ‘” -0.290 0.177 0.999 ”‘ -0.302 -0.027
(3.66) (1 .44) (1 .32) (6.25) (1 .32) (0.13)
Yogyakarta -1.811 ‘" -0.361 "" -0.916 ‘” -1.292 '” -0.064 -1.175 ‘“
(8.34) (3.37) (5.39) (3.91) (0.26) (3.65)
Yogyakarta ' Urban 1.157 ”‘ 0.055 0.084 0.786 "‘ -0.132 0.617 ‘
(5.52) (0.34) (0.49) (2.27) (0.39) (1 .73)
East Java 0906 ‘” -0.625 ” -1.105 ”‘ -1.064 “" -0.119 -0.829 ""
(6.04) (2.54) (11.41) (5.87) (0.64) (5.22)
East Java " Urban 0.273 ” 0.414 ‘ 0.292 ” 0.229 -0.064 -0.108
(2.01) (1.66) (2.36) (1.02) (0.23) (0.77)
Bali 0611 ”‘ -0.284 ” .0565 ‘ -0.059 -0.363 " -0.422 ”
(4.07) (2.27) (1.93) (0.32) (1 .70) (2.48)
Bali " Urban —0.077 0.039 -0.015
(0.58) (0.26) (0.05)
West Nusa Tenggara -1.269 ”‘ ~0.336 “ -1.151 ”‘ -0.791 "" 0.027 —O.978 ‘”
(7.04) (2.54) (6.73) (3.37) (0.14) (3.86)
West Nusa Tenggara ' Urba 0.925 "" 0.175 0.249 0.065 0.493 ‘ -0.401
(4.23) (0.74) (1 .03) (0.30) (1 .74) (1 .64)
South Kalimantan -0.888 ”" -0.187 ‘ -0.581 ‘” -0.609 “' -0.864 ’ -0.399
(4.74) (1.90) (3.47) (2.56) (1 .70) (1.37)
South Kalimantan ‘ Urban 0.480 ” 0.104 -0.106 0.045 0.421 -0.471
(2.22) (0.59) (0.58) (0.19) (0.79) (1.18)
South Sulawesi -l.382 “'" -0.433 “" -1.343 ”" -1.229 ‘" 0.401 ‘ -1.547 "
(5.67) (3.27) (5.31) (5.35) (1.72) (2.17)
South Sulawesi * Urban 0.390 0.157 0.100 0.174 -0.420 " 1.065
(1 .27) (0.81) (0.36) (0.67) (1.67) (1 .48)
Cross income 0.237 ”‘ 0.499 ”‘
(3.90) (5.33)
Selection Coefﬁcient -0.934 '“ -0.142 0.196 “ 0.381 -0.474 0.036
(5.27) (0.88) (2.10) (1.37) (1.33) (0.24)
Constant 6.201 ‘” 7.072 "" 4.472 ”' 3.569 " 7.444 ” 3.435 ”‘
(5.51) (4.1 1) (6.40) (2.41) (2.09) (2.74)
F-test
Own Schooling 17.31 9.31 38.75 7.25 0.25 15.91
(0.000) (0.000) (0.000) (0.000) (0.779) (0.000)
Father's Schooling 1.60 0.36 0.26 1.09 1.88 0.46
(0.174) (0.840) (0.901) (0.360) (0.117) (0.764)
Mother's Schooling 4.74 0.79 0.70 1.68 1.06 0.48
(0.003) (0.502) (0.556) (0.171) (0.367) (0.699)
Parental Schooling 4.52 0.45 0.45 1.87 1.41 0.61
(0.000) (0.871) (0.870) (0.074) (0.206) (0.749)
Age 1.52 6.23 3.52 1.25 8.93 2.33
(0.195) (0.000) (0.008) (0.288) (0.000) (0.057)
Residency 10.86 6.06 13.03 7.33 1.98 3.82
(0.000) (0.000) (0.000) (0.000) (0.010) (0.000)
Adjusted R 2 0.224 0.330 0.397 0.188 0.506 0.363
Root MSE 1.097 0.649 0.743 1.159 0.571 0.866
Observations 2,107 574 1,391 1,272 252 616

 

Source: lFLS 1.

Month of interview dummy variables are included in the regressions but are not reported. Omitted category for own
schooling is no schooling (0-6 years for public sector workers), for parental schooling is no schooling and for province of
residence is Jakarta. Adjustments are made due to small cell size for some variables for women: 10+ years for own
schooling and no urban interaction terms for the province of Lampung and Bali. Selection coefﬁcients are calculated based
on Base Speciﬁcation estimates. Standard errors are robust to clustering at the community level and heteroskedasticity.
Absolute r-statistics are in parentheses. Signiﬁcant at the 1%('"), 5%(”) and 10%(‘) indicated. p-values for F-test are

in parentheses.

167

 

 

Table 6.63
Selectivity Corrected Wage Functions:
The Effects of Non-linear Own Schooling, Parental Schoollng and Residency

 

 

 

 

lFLSZ
Men Women
Self Public Private Self Public Private
Employment Sector Sector Employment Sector Sector
Own Schooling
1-3 years 0.169 " 0.056 0.176 -0.097
(2.03) (0.66) (1.58) (0.83)
4-6 years 0.316 "" 0.224 "‘ 0.211 ” 0.004
(4.17) (3.16) (2.00) (0.03)
7-9 years 0.770 “" 0.108 0.500 ‘" 0.486 “‘“ 0.007 0.087
(8.45) (0.78) (5.96) (3.44) (0.03) (0.45)
10-12 years 1.162 "“’ 0.446 ” 0.576 "" 0.736 "'“ -0.424 0.891 ‘"
(9.63) (2.16) (7.24) (4.54) (0.66) (6.87)
13+ years 2.171 “" 0.742 ”" 1.007 ""
(10.61) (2.88) (9.23)
Father's Schooling
Some Elementary 0.274 ‘" 0.001 0.069 0.094 —0.078 0.064
(3.14) (0.00) (1 .13) (0.98) (0.47) (0.50)
Completed Elementary 0.191 “ 0.062 0.045 0.234 " -0.191 0.012
(2.00) (0.61) (0.73) (2.01) (1.16) (0.10)
Secondary/'1' ertiary 0.354 ‘”" -0.095 0.131 " 0.367 ‘“' -0.074 0.295 "
(2.56) (0.79) (1.71) (2.07) (0.44) (1.96)
Missing 0.209 "‘ -0.125 0.042 0.013 -0.185 0.002
(1.78) (0.59) (0.51) (0.10) (0.70) (0.01)
Mother's Schooling
Some Elementary 0.011 -0.014 0.056 0.059 0.090 -0.042
(0.13) (0.15) (0.93) (0.48) (0.68) (0.39)
Completed Elementary/ 0.297 ‘“ 0.099 0.175 "" 0.125 0.037 0.189 ‘
Secondary/Tertiary (3.09) (1.28) (2.95) (0.91) (0.28) (1 .92)
Missing 0.054 0.038 0.040 -0. 180 0.244 0.094
(0.47) (0.18) (0.51) (1.13) (0.72) (0.75)
Age (spline)
25-29 years 0090 ""' 0.060 0.041 " -0.036 0.007 0.059 "
(2.29) (0.70) (2.23) (0.68) (0.08) (1 .89)
30-39 years 0.001 0.026 0.017 " 0.038 " 0004 0004
(0.1 1) (1 .57) (2.34) (2.29) (0.12) (0.34)
40-49 years 0023 “ 0.035 ‘" -0.011 -0.024 ” 0.034 0.019
(2.53) (3.43) (1 .26) (2.06) (1 .29) (1 .25)
50-59 years 0001 -0.008 -0.018 0.017 0.007 -0.077 ""
(0.1 1) (0.40) (1 .16) (1.18) (0.22) (2.95)
Province of Residence
North Surnatera -0.792 *” 0.034 —0.473 "W 0617 “ -0.035 -0.432 ‘
(5.13) (0.22) (3.82) (2.53) (0.14) (1 .82)
North Sumatera " Urban 0.758 "" -0.431 "" 0.269 ‘ 0.278 -0.524 "’ 0.253
(4.62) (2.38) (1 .87) (1 .13) (2.12) (0.94)
West Sumatera -0.374 0.011 -0.306 "" -0.639 “ -0.703 ” -0.346
(1 .23) (0.06) (3.32) (2.27) (2.25) (1.39)
West Sumatera "' Urban 0.210 -0.002 -0.055 0.287 0.516 "‘ -0.107
(0.65) (0.01) (0.56) (1 .00) (1 .88) (0.35)
South Surnatera -l.181 "'“ —0.041 -0.494 ”" -1.127 ”‘ -0.289 -0.894 ”"
(8.65) (0.15) (3.46) (4.42) (0.25) (4.18)
South Surnatera "' Urban 0.833 ‘" 0.242 -0.028 1.266 ”‘ 0.194 0.741 “
(3.19) (0.87) (0.14) (4.10) (0.17) (2.36)
Lampung -1.294 "" -0.022 -0.829 "" -0.722 “ -0.256 -1.139 ‘”
(11.02) (0.13) (4.77) (2.31) (0.87) (5.96)
Lampung ‘ Urban 1.290 """' -0.090 0.931 “‘
(5.33) (0.16) (3.93)
West Java -0.566 ‘" -0.016 -0.568 ““ -0.742 "" —0.210 -0.774 ‘"
(5.24) (0.09) (6.28) (3.27) (0.90) (3.70)
West Java " Urban 0.377 "’ -0.111 0.242 " 0.492 ” -0.209 0.402 "
(3.00) (0.66) (2.55) (2.42) (1 .16) (1 .97)
(continued)

168

 

 

Table 6.68 (continued)
Selectivity Corrected Wage Functions:
The Effects of Non-linear Own Schooling, Parental Schoollng and Residency

 

 

 

 

iFLSZ
Men Women
Self Public Private Self Public Private
Employment Sector Sector Employment Sector Sector
Central Java -1.084 "‘" -0.149 -1.021 “" -1.033 ‘" -O.459 " -1.331 ‘”
(8.87) (1.00) (8.83) (4.94) (2.35) (7.03)
Central Java " Urban 0.530 ‘” -0.167 0.412 ”‘ 0.578 ‘” 0.118 0.611 ”
(3.44) (1 . 15) (3.05) (3.30) (0.77) (2.59)
Yogyakarta -1.242 ”" -0.094 -0.678 " -l.179 ”’ -0.522 ” -0.707 ”‘
(6.66) (0.67) (2.52) (5.82) (2.26) (3.03)
Yogyakarta “ Urban 0.566 ” 0.006 -0.092 0.579 ”" 0.253 0.208
(2.48) (0.06) (0.33) (3.72) (1 .01) (0.80)
East Java -0.737 "" -0.012 -0.773 “‘ -0.628 ‘" -0.339 " -0.779 "“
(6.00) (0.05) (10.07) (2.92) (2.01) (6.89)
East Java " Urban -0.050 -0.210 0.189 " -0.004 -0.212 -0.0004
(0.33) (0.85) (2.06) (0.02) (1 .62) (0.00)
Bali -0.683 ‘“ -0.054 0.663 ‘“ -O.343 -0.525 -0.616 “'"
(5.84) (0.50) (4.74) (1 .64) (1 .62) (3.68)
Bali ' Urban 0.136 0.065 0.189 -O.250 -0.034 0.236
(1.01) (0.38) (1.07) (1.41) (0.11) (1.20)
West Nusa Tenggara -1.000 "‘ 0.053 -0.773 ‘“ -0.686 ‘" -0.313 -0.997 “"
(4.66) (0.44) (5.52) (2.93) (1.51) (6.82)
West Nusa Tenggara " Urba 0.879 ”’ -0.393 ” -0.235 0.036 -0.011 0.351
(3.72) (2.08) (1 .42) (0.18) (0.07) (1.24)
South Kalimantan -1.079 ’" 0.148 -0.578 ”" -0.357 -0.447 ‘ -0.480 ""
(6.32) (1.18) (2.96) (1.40) (1.88) (2.71)
South Kalimantan ‘ Urban 0.918 ‘" -0.312 " 0.193
(4.08) (1.85) (0.73)
South Sulawesi -1.048 "" -0.419 “ -0.822 “" -1.213 "‘ -O.377 " -0.947 "W
(3.22) (2.51) (3.40) (5.62) (2.01) (2.67)
South Sulawesi " Urban 0.363 0.386 ‘ 0.099
(1.09) (1.81) (0.36)
Selection Coefficient -1.242 "" -0.058 0.224 “ -0.062 -0.603 0.138
(6.23) (0.30) (1 .74) (0.20) (1 .59) (0.44)
Constant 10.323 ”‘ 5.318 " 5.331 ‘" 7.526 ‘“ 9.167 ”“ 4.413 ”‘
(8.77) (1 .97) (9.74) (4.41) (2.67) (5.57)
F -test
Own Schooling 29.96 4.94 28.76 6.30 0.61 25.61
(0.000) (0.002) (0.000) (0.000) (0.545) (0.000)
Father‘s Schooling 3.02 1.62 0.89 1.45 0.64 2.43
(0.018) (0.169) (0.468) (0.218) (0.636) (0.048)
Mother's Schooling 3.55 0.76 3.05 1.07 0.32 2.07
(0.015) (0.515) (0.028) (0.360) (0.81 1) (0.104)
Parental Schooling 5.26 1.33 3.41 1.82 0.38 2.74
(0.000) (0.235) (0.001) (0.082) (0.91 1) (0.009)
Age 4.89 7.53 6.71 1.61 2.46 4.14
(0.001) (0.000) (0.000) (0.170) (0.047) (0.003)
Residency 12.62 2.07 10.98 6.27 1.55 6.32
(0.000) (0.003) (0.000) (0.000) (0.065) (0.000)
Adjusted R2 0.195 0.236 0.286 0.118 0.197 0.372
Root MSE 1.097 0.635 0.772 1.195 0.600 0.899
Observations 2,3 1 8 645 2,099 1,480 309 992
Source: IFLSZ.

Month of interview dununy variables are included in the regressions but are not reported. Omitted category for own
schooling is no schooling (0—6 years for public sector workers), for parental schooling is no schooling and for province of
residence is Jakarta. Adjustments are rmde due to small cell size for some variables for women: 10+ years for own
schooling and no urban interaction terms for the province of Lampung, South Kalimantan and South Sulawesi. Selection
coefﬁcients are calculated based on Base Speciﬁcation estimates. Standard errors are robust to clustering at the community
level and heteroskedasticity. Absolute t-statistics are in parentheses. Signiﬁcant at the 1%(”"), 5%(”) and 10%(‘)
indicated. p -vaiues for F -test are in parentheses.

169

 

 

Table 6.6C
Selectivity Corrected Wage Functions:
The Effects of Non-linear Own Schooling, Parental Schooling and Residency

 

 

 

 

 

iFLS3
Men Women
Self Public Private Self Public Private
Employment Sector Sector Employment Sector Sector
Own Schooling
1-3 years 0.159 ‘ 0.172 " 0.193 " 0.235 ”
(1.66) (1.78) (1.74) (2.29)
4-6 years 0.337 ”‘ 0.312 ”‘ 0.233 ” 0.297 ‘"
(3.98) (3.79) (2.40) (2.85)
7-9 years 0660 '“ 0.220 0.538 ”" 0.369 ‘“ 0.397 0.504 "‘
(6.61) (1 .23) (6.17) (3.19) (1 .08) (3.08)
10-12 years 1.108 ”" 0.722 ” 0.762 ‘" 0.669 ”‘ 0.021 1.137 ”‘
(9.02) (2.16) (8.72) (5.03) (0.02) (8.53)
13+ years 1.653 ”' 1.099 "" 1.171 ”‘
(10.16) (2.43) (10.86)
Father's Schooling
Some Elementary 0038 0.003 0.002 0.068 -0.1 13 —0.027
(0.59) (0.03) (0.04) (0.65) (0.62) (0.30)
C ompieted Elementary 0.063 0.018 0.020 0.057 -0.008 -0.007
(0.93) (0.22) (0.41) (0.66) (0.04) (0.08)
Secondary/Tertiary 0.137 0.096 0.092 0.180 0.251 0.377 ""
(1.46) (1.00) (1.61) (1.51) (1.38) (3.60)
Missing -0.138 ‘ -0.079 -0.074 0.092 0.301 0.039
(1.70) (0.49) (1.12) (0.91) (1.12) (0.41)
Mother's Schooling
Some Eiernentary 0.165 " -0.038 0.002 0.040 -0.057 -0.067
(2.25) (0.40) (0.05) (0.40) (0.52) (0.82)
Completed Elementary/ 0.274 ‘“” -0.073 0.068 0.138 -0.051 0.050
Secondaryffertiary (3.71 ) (0.87) (1 .35) (1 .53) (0.43) (0.57)
Missing 0.272 ""‘ —0.264 0.096 0.123 -0.347 -0.i35
(2.91) (1.57) (1.26) (1.10) (1.59) (1.31)
Age (spline)
25-29 years -0.007 0.108 0.006 0.068 0.049 -0.003
(0.24) (1.61) (0.39) (1.33) (0.43) (0.10)
30-39 years 0.006 0.046 " 0.016 ” 0.008 -0.071 0.015
(0.70) (2.02) (2.41) (0.62) (1.37) (1 .42)
40-49 years 0.004 0.030 " 0.003 -0.004 0.035 -0.007
(0.47) (2.47) (0.34) (0.37) (1 .62) (0.51)
50-59 years -0.031 “‘" 0.015 -0.032 "" -0.007 0.059 ”‘ -0.025
(2.78) (0.58) (2.44) (0.45) (2.62) (1.14)
Province of Residence
North Surnatera -0.699 “" -0.348 -0.302 "“ -0.755 ” 0.233 -0.347 "
(2.84) (1.62) (2.73) (2.29) (0.74) (1.69)
North Sumatera ‘ Urban 0.654 ‘“' 0.316 0.126 0.438 -0.462 0.126
(2.55) (1.40) (1.12) (1.28) (1.20) (0.57)
West Surnatera -0.140 0.153 -0.073 -0.833 “" 0.219 -0.737 "
(1.02) (0.75) (0.56) (4.83) (1.16) (2.18)
West Sumatera "' Urban 0.312 0.270 0.054 0.294 -0.123 0.217
(1 .63) (0.91) (0.26) (1 .26) (0.71) (0.58)
South Surnatera -0.606 ”" -0.103 -0.164 -0.543 ”" -0.315 ' -0.315
(3.65) (0.44) (1.28) (2.69) (1.67) (1.23)
South Surnatera " Urban 0.818 "" 0.239 -0.041 0.668 “‘ 0.049 -0.114
(3.10) (0.80) (0.24) (2.86) (0.24) (0.31)
Lampung -0.504 ‘" -0.926 '"' -0.357 ‘” -0.524 " -0.977 0.599 ”"
(3.46) (2.04) (2.98) (2.52) (1.11) (3.31)
Lampung " Urban 0.606 ”" 0.984 "" 0.345 ‘ 0.680 ‘ 1.568 “ -0.091
(2.86) (2.08) (1 .71) (1 .73) (1 .79) (0.38)
West Java —0.242 " -0.150 —0.224 ‘" -0.281 -0.096 -0.173
(1 .88) (0.94) (3.28) (1 .36) (0.47) (1.40)
West Java "‘ Urban 0.211 0.073 0.122 ‘ 0.137 -0.083 -0.104
(1 .45) (0.62) (1 .68) (0.70) (0.45) (0.74)
(continued)

170

 

 

Table 6.6C (continued)
Selectivity Corrected Wage Functions:
The Effects of Non-linear Own Schooling, Parental Schooling and Residency

 

 

 

iFLS3
Men Women
Self Public Private Self Public Private
Employment Sector Sector Employment Sector Sector
Central Java -0.589 "m -0.090 -0.377 *" -0.754 "“ 0.018 -0.624 "“
(4.82) (0.49) (3 .53) (4.66) (0.12) (3 .34)
Central Java " Urban 0.290 ” -0.241 -0.027 0.342 ” -0.206 0.131
(2. 10) (1 .50) (0.24) (2.47) (1 .08) (0.60)
Yogyakarta -0.920 ‘" -0.070 -0.536 "“ -0.952 ""‘ -0.216 -1.242 ”"
(5.77) (0.44) (4.18) (3.81) (0.85) (4.12)
Yogyakarta " Urban 0.380 " -0.090 -0.130 0.468 "‘ -0.206 0.506
(2.18) (0.64) (1 .00) (1 .75) (0.76) (1 .54)
East Java -0.255 ” -0.114 -0.384 "“ -0.477 "" 0.077 —0.521 “"
(2.42) (0.65) (5.35) (2.67) (0.38) (4.60)
East Java "' Urban -0.050 0.148 0.059 0.010 -0.137 -0.012
(0.46) (0.96) (0.73) (0.06) (0.70) (0.09)
Bali -0.667 ”‘ 0.005 -0.341 """‘ -0.445 * 0.066 -0.660 "
(4.15) (0.04) (2.69) (1.84) (0.28) (2.36)
Bali "' Urban 0.470 "" -0.249 " 0.066 0.052 -0.308 0.324
(2.97) (2.03) (0.45) (0.23) (1.39) (1.12)
West Nusa Tenggara -O.489 ‘" -0.262 -0.420 "" -0.447 “ -0.467 -0.781 "““
(3.19) (1 .33) (4.99) (2.23) (1 .47) (6.30)
West Nusa Tenggara "‘ Urba 0.343 0.087 -0.122 0.343 ” 0.203 0.075
(1 .32) (0.45) (0.67) (2.03) (0.68) (0.58)
South Kalimantan -0.253 0.185 -0.156 —0.188 0.376 0.228
(1.59) (1.05) (1.04) (1.02) (1.43) (0.88)
South Kalimantan “ Urban 0.334 -0.241 —0.026 0.090 -0.619 "‘ -0.719 ‘”"
(1 .41) (1 .50) (0.14) (0.50) (2.49) (2.25)
South Sulawesi -0.304 " ~0.176 -0.366 “ -0.672 " -0.319 -0.972 "‘""
(1 .70) (0.62) (2.00) (2.15) (0.39) (3.27)
South Sulawesi " Urban 0.228 -0.164 0.086 -0.044 -0.020 0.308
(1 . i 1) (0.60) (0.42) (0.14) (0.02) (0.57)
Selection Coefﬁcient -1.139 ”" 0.082 0.168 -0.054 -0.976 "‘ 0.361
(5.85) (0.26) (1 .28) (0.16) (2.05) (1 .45)
Constant 8.170 "'" 3.927 "' 6.592 "" 5.047 ‘" 8.668 “ 6.028 “"
(8.89) (1 .67) (14.80) (2.83) (2.17) (8.85)
F -test
Own Schooling 25.97 2.62 41.27 6.71 0.94 24.08
(0.000) (0.051) (0.000) (0.000) (0.3 90) (0.000)
Father’s Schooling 1.70 0.53 1.36 0.65 2.03 6.09
(0.148) (0.712) (0.245) (0.624) (0.092) (0.000)
Mother's Schooling 6.06 1.02 0.95 1.02 0.87 1.23
(0.000) (0.384) (0.414) (0.385) (0.459) (0.298)
Parental Schooling 4.43 0.84 1.54 1.25 1.81 4.06
(0.000) (0.552) (0.152) (0.271) (0.087) (0.000)
Age 2.22 6.35 5.24 1.02 3.59 1.27
(0.065) (0.000) (0.000) (0.397) (0.007) (0.280)
Residency 4.67 1.79 4.70 3.59 2.74 4.45
(0.000) (0.014) (0.000) (0.000) (0.000) (0.000)
Adjusted R 2 0.1 15 0.269 0.21 1 0.069 0.312 0.318
Root MSE 1.127 0.672 0.795 1.224 0.667 0.910
Observations 3 ,023 702 2,801 2,047 352 1 ,3 76

 

Source: IFLS3.

Month of interview dummy variables are included in the regressions but are not reported. Omitted category for own
schooling is no schooling (0-6 years for public sector workers), for parental schooling is no schooling and for province of
residence is Jakarta. Adjustments are made due to small cell size for some variables for women: 10+ years for own
schooling. Selection coefﬁcients are calculated based on Base Speciﬁcation estimates. Standard errors are robust to
clustering at the community level and heteroskedasticity. Absolute t-statistics are in parentheses. Signiﬁcant at the

1%("”"), 5%(”) and 10°/o("') indicated. p -va1ues for F -test are in parentheses.

171

 

 

Table 6.7

Selectivity Corrected Wage Functions:
Summary of the Effects of N on-linear Own Schooling, Parental Schooling and Residency

 

 

 

Men Women
Self Public Private Self Public Private
Employment Sector Sector Employment Sector Sector
iFLSl
1-3 years a) 0.017 -0.026 0.053 -0.026
4-6 years 0.270 "* 0.271 "" 0.377 ""‘ 0.109
7-9 years 0.760 "* 0.139 0.441 "“" 0.510 W“ 0.125 0.423 "
10-12 years b) 1.095 "" 0.504 ** 0.776 ""' 0.645 "* 0.393 1.189 ‘"
13+ years 1.797 “* 0.850 "* 1.327 *“
lFLSZ
1-3 years a) 0.169 " 0.056 0.176 -0.097
4-6 years 0.316 **"' 0.224 “* 0.211 " 0.004
7-9 years 0.770 "“ 0.108 0.500 *** 0.486 **"‘ 0.007 0.087
10-12 years b) 1.162 "" 0.446 ** 0.576 "* 0.736 "‘" -0.424 0.891 **"‘
13+ years 2.171 "" 0.742 *** 1.007 *"
1FL83
1-3 years a) 0.159 " 0.172 * 0.193 "' 0.235 "
4.6 years 0.337 *” 0.312 '1‘" 0.233 " 0.297 "*
7-9 years 0.660 *" 0.220 0.538 "* 0.369 ""‘ 0.397 0.504 "*
10-12 years b) 1.108 W" 0.722 ** 0.762 "* 0.669 *“ 0.021 1.137 "”"
13+ years 1.653 "" 1.099 " 1.171 *"

 

Source: Based on estimates of Table 6.6A, 6.6B, 6.6C.
a) Omitted category for public sector workers is 0-6 years.
b) 10+ years for women.

172

 

 

J

Table 6.8

Predicted Returns to Schoollng:

Speciﬁcations that include Own Schooling

 

 

 

Men Women
Self Public Private Self Public Private
Employment Sector Sector Employment Sector Sector
A. Cross Section Sample
lFLSl
1-3 years a) 6.3 “ -1.4 8.7 " 6.6 2.7 3.7
4-6 years 11.3 "" 11.4 “" 12.0 "" 7.3
7-9 years 20.1 ‘” 7.8 " 7.9 ”" 1.3 20.3 23.5
10-12 years b) 11.7 " 9.7 "”' 14.7 "”“ 12.4 "““ 8.7 "‘ 15.3
13+ years 18.9 ‘ 8.9 “" 12.8 ”"
iFLSZ
1-3 years a) 5.3 0.3 5.2 5.3 -6.7 10.5 *‘
4-6 years 8.3 *” 8.0 ”“ 4.3 4.1
7-9 years 14.0 "* 3.5 13.6 “‘ 9.7 " 10.1 18.5 "“
10-12 years b) 6.6 "‘ 5.0 4.3 ‘ 16.0 "* 2.9 11.3 "*
13+ years 20.9 ""‘ 6.3 “‘" 12.7 ""
iFLS3
1-3 years a) 6.9 "”" -7.8 9.5 "" 8.1 “ -7.2 13.4 ”*
4-6 years 6.0 ” 4.0 1.3 0.2
7-9 years 9.7 “" 9.2 8.0 ”"' 7.6 "* 37.2 "' 16.5 ”"‘
10-12 years b) 12.6 “" 4.0 9.4 “" 9.4 ” 3.5 13.5 ""‘
13+ years 13.2 "‘ 5.3 ” 12.5 ""
B. Panel Sample
lFLSl
1-3 years a) 4.6 -8.7 10.3 "“" 2.2 3.6 7.1
4-6 years 6.7 " 6.2 "‘ 11.9 ” 2.1
7-9 years 19.4 ""‘ 13.0 “ 8.9 ""' 1.8 10.9 21.4 ""
10-12 years b) 8.3 10.8 " 12.1 "" 3.2 7.0 17.3 ”"'
13+ years 24.3 “ 5.9 ” 10.5 "”"
lFLSZ
1-3 years a) 5.7 -4.3 6.6 4.5 -9.2 9.2
4-6 years 6.7 "‘ 9.4 ”" 3.4 -0.4
7-9 years 18.4 "* 6.7 14.9 ‘" 17.3 ”" 18.5 20.7 ‘"
10-12 years b) 13.3 " 13.5 "' 2.5 13.8 2.3 19.3 “"
13+ years 14.9 "' 7.1 ""‘ 8.4 ""
IFLS3
1-3 years a) 11.2 "”" -8.7 10.8 " 6.0 -5.3 11.9 ”
4-6 years 2.7 2.8 1.7 -0.6
7-9 years 10.7 ‘“" 10.3 12.8 "" 9.3 "' 41.8 " 15.2 "
10-12 years b) 18.2 W" 8.3 8.3 *" 18.4 "" 10.9 ‘ 15.9 ”"
13+ years 23.8 “‘ 3.1 12.9 ”"

 

Source: Based on estimates of Appendix Table 6.3A, 6.3B, 6.3C. Detailed estimates for panel sample are not shown.

a) 1-6 years for public sector workers.
b) 10+ years for women.

173

 

’.

 

Table 6.9

Predicted Returns to Schooling:
Specifications that include Own Schooling and Parental Schooling

 

 

 

Men Women
Self Public Private Self Public Private
Employment Sector Sector Employment Sector Sector
A. Cross Section Sample
iFLSl
1-3 years a) 4.2 -2.3 8.6 "”" 5.2 2.4 2.4
4-6 years 10.7 “"“ 10.8 "" 10.0 *"”" 6.5
7-9 years 19.5 "* 7.1 7.4 ""‘ -0.1 7.5 21.6 "*
10-12 years b) 11.3 " 7.8 13.4 ""‘ 9.4 0.1 13.0 ""
13+ years 17.8 " 8.0 *“ 11.7 "*
lFLSZ
1-3 years a) 0.5 1.9 4.3 5.9 -6.9 7.2
4-6 years 10.3 "* 7.8 *" 2.1 1.4
7-9 years 15.6 "* 7.9 11.7 "‘ 6.0 9.9 14.7 "“"
10-12 years b) 11.3 "" 11.0 " 2.8 12.6 “ 1.8 12.2 "‘"
13+ years 22.8 "" 8.1 **" 10.5 "*
iFLS3
1-3 years a) 5.2 -7.9 9.3 “" 6.9 " -7.3 12.5 ""'
4-6 years 6.8 *"‘ 3.8 0.5 -0.8
7-9 years 9.7 *" 9.3 7.4 *" 6.4 " 16.6 12.2 "“
10-12 years b) 14.5 "* 3.6 8.2 "" 8.4 " -10.7 12.9 "“
13+ years 12.3 *" 4.9 10.8 “"
B. Panel Sample
lFLSl
1-3 years a) 2.3 -10.7 11.0 "”""‘ 0.9 3.6 6.3
4-6 years 5.9 " 5.3 9.3 "”" 0.7
7-9 years 18.3 ""‘ 11.3 * 8.6 *" 0.1 1.4 19.2 "'*"'
10-12 years b) 7.7 8.4 11.2 ”* -O.4 -0.1 14.9 "“"
13+ years 24.4 *"' 4.9 "‘ 9.6 ’"‘
lFLSZ
1-3 years a) 3.2 -2.0 5.2 3.3 -9.2 8.1
4-6 years 6.0 8.8 "" 2.1 -3.3
7-9 years 17.4 *" 12.3 " 12.8 "‘"‘ 14.7 "”" 22.5 18.2 "*
10-12 years b) 14.4 "* 21.4 *** 0.2 10.3 5.2 15.7 "”"
13+ years 16.8 " 9.4 *" 7.1 **
iFLS3
1-3 years a) 8.3 " -9.9 10.9 " 5.4 -5.9 11.7 "
4-6 years 3.1 1.5 0.9 -2.0
7-9 years 11.1 "* 10.3 12.3 "“ 9.1 “ 35.2 * 11.3
10-12 years b) 20.3 *** 5.8 6.4 " 19.8 "'" 4.7 14.7 ""
13+ years 26.7 "" 2.7 11.2 "

 

Source: Based on estimates of Appendix Table 6.4A, 6.4B, 6.4C. Detailed estimates for panel sample are not shown.

a) 16 years for public sector workers.

b) 10+ years for women.

174

 

 

Appendix Table 6.1A
Selectivity Corrected Wage Functions:
The Effects of Linear Own Schooling

 

 

 

IFLS]
Men Women
Self Public Private Self Public Private
Employment Sector Sector Employment Sector Sector
Years of schooling 0.125 "* 0.080 "”’ 0.108 W" 0.079 "" 0.081 0.123 W"
(10.58) (3.18) (16.82) (6.62) (1.63) (12.27)
Age 0.024 0.078 * 0.096 "”" 0.085 " 0.051 0.061 "
(0.92) (1 .75) (3.64) (2.07) (0.78) (1 .73)
Age2(x 102) -0031 0066 p.114 m -0.099 " -o.033 -o.o75 *
(1 .00) (1.31) (3.39) (2.02) (0.41) (1 .70)
Month of interview
October 0.323 ** -0.147 0.116 0.386 *" 0.026 0.035
(2.20) (0.89) (0.94) (2.65) (0.15) (0.22)
November 0.423 "* -0.259 0.127 0.407 "" -0.214 -0.127
(3.28) (1 .49) (1.06) (2.74) (1.31) (0.86)
December -0.082 .0100 0.070 0.132 0.008 -0.191
(0.64) (0.64) (0.55) (0.90) (0.06) (1 .26)
January -0.002 0.026 0.225 0.094 0.317 "‘ -0.140
(0.01) (0.13) (1 .47) (0.40) (1 .78) (0.61)
Gross Income 0.077 0.329 ""
(1.11) (3.16)
Selection Coefﬁcient -0.596 “* -0.114 0.089 0.777 "* -0.270 «0.017
(3 .63) (0.60) (1 .28) (3. 10) (0.97) (0.12)
Constant 5.077 "" 4.580 *" 3.493 "" 2.429 " 5.284 “ 4.027 "”"
(8.71) (3.42) (6.63) (2.30) (2.48) (5.28)
F -test
Age 0.70 14.66 8.52 2.20 16.29 1.50
(0.499) (0.000) (0.000) (0.1 13) (0.000) (0.225)
Month of interview 7.39 1.53 0.64 3.1 l 2.37 1.03
(0.000) (0.195) (0.634) (0.016) (0.056) (0.393)
Adjusted R 2 0.1 14 0.291 0.263 0.088 0.475 0.273
Root MSE 1. 173 0.668 0.822 1.228 0.589 0.925
Observations 2,107 574 1,391 1,272 252 616

 

Source: IFLS 1 .

Omitted category for month of interview is August/September. Selection coefﬁcients are calculated based on Base
Speciﬁcation estimates in which dummy variables for own schooling are replaced by years of schooling. Standard
errors are robust to clustering at the community level and heteroskedasticity. Absolute t-statistics are in parentheses.
Signiﬁcant at the 1%(*“), 5%(") and 10%(") indicated. p -va1ues for F -test are in parentheses.

175

f

 

 

Appendix Table 6.1B
Selectivity Corrected Wage Functions:
The Effects of Linear Own Schooling

 

 

 

lFLSZ
Men Women
Self Public Private Self Public Private
Employment Sector Sector Employment Sector Sector

Years of schooling 0.089 "" 0.039 0.087 "‘“ 0.076 "" -0.019 0.116 ‘"

(9.17) (1 .40) (16.70) (7.42) (0.52) (14.27)
Age 0.072 *“ 0.026 0.110 ”" 0.069 " -0.001 0.129 ”*

(2.76) (0.42) (5.96) (1 .79) (0.01) (4.43)
Age2(x10'2) -0.088 m -0.010 p.127 m -o.o75 : 0.023 -0.146 m

(2.92) (0.14) (5.43) (1 .73) (0.29) (4.08)
Month of interview
October 0128 -0.168 " -0.043 -0.189 -0.253 *"‘ 0.114

(1 .13) (1 .84) (0.54) (1 .56) (2.24) (0.97)
November -0.206 " -0.190 " -0.041 -0.097 -0.418 W" 0.060

(1 .86) (1 .97) (0.55) (0.82) (2.90) (0.51)
December 0.046 -0.086 0.043 0.145 -O.157 0.045

(0.40) (0.91) (0.52) (0.98) (1.23) (0.37)
January-April -0.038 0.009 0.160 " 0.260 0.067 0.404 "

(0.27) (0.09) (1 .85) (1 . 13) (0.50) (2.46)
Selection Coefﬁcient 0.009 -0.297 -0.015 0.019 -0.473 “ -0.526 "

(0.05) (1 .39) (0.13) (0.08) (2.18) (2.46)
Constant 4.676 "'* 6.668 "" 3.988 "* 4.455 "" 8.357 ‘“ 3.607 """

(7.41) (3.51) (10.79) (4.33) (4.08) (6.57)
F-test
Age 4.87 11.04 21.48 1.74 5.84 10.96

(0.008) (0.000) (0.000) (0. 177) (0.003) (0.000)
Month of interview 1.84 1.95 2.01 2.30 3.77 1.86
(0.121) (0.102) (0.091) (0.059) (0.006) (0.1 16)

Adjusted R 2 0.096 0.218 0.194 0.059 0.241 0.283
Root MSE 1.162 0.643 0.821 1.234 0.583 0.960
Observations 2,3 1 8 645 2,099 1 ,480 309 992

 

Source: iFLS2.

Omitted category for month of interview is August/September. Selection coefﬁcients are calculated based on Base
Speciﬁcation estimates in which dummy variables for own schooling are replaced by years of schooling. Standard
errors are robust to clustering at the community level and heteroskedasticity. Absolute r-statistics are in parentheses.
Signiﬁcant at the 1%(*"), 5%(‘"") and 10%(*) indicated. p -va1ues for F -test are in parentheses.

176

 

 

Appendix Table 6.1C
Selectivity Corrected Wage Functions:
The Effects of Linear Own Schooling

 

 

 

iFLS3
Men Women
Self Public Private Self Public Private
Employment Sector Sector Employment Sector Sector
Years of schooling 0.087 ‘** 0.054 0.084 ‘“ 0.060 “* 0.044 0.114 *"
(10.50) (1.53) (18.88) (6.88) (0.94) (16.54)
Age 0.084 "* 0.037 0.064 ‘“" 0.127 "” -0.068 0.064 "
(3.69) (0.52) (4.18) (3.45) (0.76) (2.52)
Age2(x 10'?) -0.098 m -0.016 -o.o73 m -o.r4o m 0.101 .0.070 n
(3.60) (0.21) (3.74) (3.36) (1 .00) (2.19)
Month of interview
August -0.170 “ 0.003 0.022 -0.095 0.123 -0.037
(2.24) (0.04) (0.40) (0.87) (0.92) (0.35)
September -0. 1 19 -0.061 0.004 -0.082 0.001 -0.064
(1.35) (0.72) (0.08) (0.74) (0.01) (0.57)
October -0.104 -0.014 0.130 " -0.073 0.219 0.001
(1 . 10) (0.15) (2.36) (0.64) (1 .34) (0.01)
November-January 0.025 0.162 0.241 "" -0.069 0.063 0.198 "'
(0.24) (1 .46) (4.08) (0.53) (0.37) (1 .80)
Selection Coefﬁcient -0.340 ” -0.278 0.116 0.525 ‘ -0.375 -0.162
(2.20) (1.09) (1.37) (1.85) (1.51) (0.82)
Constant 5.464 ”* 6.727 "" 5.251 "* 3.333 "" 9.165 ‘" 4.959 ""‘
(10.77) (3.00) (17.96) (3.11) (3.34) (8.68)
F -test
Age 6.90 8.86 1 1.77 6.16 7.33 4.64
(0.001) (0.000) (0.000) (0.002) (0.001) (0.010)
Month of interview 1.80 1.61 5.84 0.21 0.98 2.73
(0.127) (0.171) (0.000) (0.936) (0.420) (0.029)
Adjusted R 2 0.072 0.244 0.183 0.044 0.269 0.279
Root MSE 1.154 0.683 0.809 1.240 0.688 0.936
Observations 3,023 702 2,801 2,047 352 1 ,3 76

 

Source: 1F LS3.

Omitted category for month of interview is June/July. Selection coefficients are calculated based on Base Speciﬁcation
estimates in which dummy variables for own schooling are replaced by years of schooling. Standard errors are robust
to clustering at the community level and heteroskedasticity. Absolute t-statistics are in parentheses. Signiﬁcant at the
1°/o(""'), 5°/o("”") and 10%(‘) indicated. p -values for F -test are in parentheses.

177

 

Appendix Table 6.2
Selectivity Corrected Wage Functions:
The Effects of Non-linear Own Schooling, Parental Schooling, Residency,
Household Compositions and Business Assets for Self-Employed Workers

 

 

 

 

Men Women
1993 1997 2000 1993 1997 2000
Own Schooling
1-3 years 0.001 0.132 0.145 0.007 0.272 0.164
(0.01) (1 .56) (1 .49) (0.05) (1 .58) (1 .26)
4-6 years 0.264 ”" 0.297 ”'" 0.329 “"‘ 0.324 "'" 0.324 "' 0.201 "'
(3.22) (3.95) (3.87) (2.14) (1 .80) (1 .79)
7-9 years 0.805 "'” 0.802 "'" 0.672 "" 0.443 " 0.485 ”" 0.324 “
(7.43) (7.36) (6.39) (2.56) (3.48) (2.54)
10-12 years 1.251 W" 1.299 "'* 1.168 ‘” 0.594 "" 0.578 " 0.636 "*
(7.55) (6.56) (7.34) (2.69) (2.57) (4.01)
13+ years 1.998 ‘" 2.345 ”* 1.708 ‘"
(6.41) (7.28) (7.53)
Father's Schooling
Some Elementary 0.145 "' 0.319 *"‘" -0.041 0.112 0.139 0.065
(1 .94) (3.23) (0.63) (0.93) (1 .19) (0.60)
Completed Elementary 0.148 0.237 " 0.066 0.042 0.312 " 0.028
(1 .60) (2.28) (0.98) (0.27) (2.05) (0.28)
Secondary/Tertiary 0.004 0.421 "" 0.151 0.234 0.384 " 0.140
(0.03) (2.88) (1 .59) (1 .43) (1 .87) (1.19)
Missing 0.059 0.245 ‘ -0.144 "‘ -0.043 0.042 0.081
(0.69) (1 .91) (1 .79) (0.37) (0.30) (0.75)
Mother's Schooling
Some Elementary 0.217 ‘”" 0.030 0.182 “ 0.211 -0.010 0.031
(2.38) (0.35) (2.35) (1 .52) (0.07) (0.30)
Completed Elementary/ 0.351 "‘""' 0.355 "'” 0.299 """'" 0.211 -0.265 0.155
Secondary/Tertiary (3.22) (3.29) (3.64) (1 .41) (1 .29) (1 .53)
Missing 0.205 ”" 0.138 0.292 ‘" 0.097 -0.310 0.141
(2.81) (1.17) (3. 10) (0.79) (1 .49) (1 .02)
Age (spline)
25-29 years 0.043 -0.108 "”" -0.016 0.052 -0.018 0.063
(1 .09) (2.37) (0.46) (1 .02) (0.31) (0.96)
30.39 years -0.001 -0.003 0.006 0.006 0.058 " 0.002
(0.07) (0.28) (0.70) (0.22) (1 .71) (0.1 1)
40-49 years -0.012 -0.034 "" -0.004 .0.018 -0.020 -0.011
(1 . 10) (3.24) (0.41) (1 .24) (1 .44) (0.55)
50-59 years 0.031 " 0.008 -0.037 ”" 0.002 0.027 0.002
(2.16) (0.62) (2.72) (0.07) (1 .62) (0.12)
(continued)

178

Appendix Table 6.2 (continued)
Selectivity Corrected Wage Functions:
The Effects of Non-linear Own Schooling, Parental Schooling, Residency,
Household Compositions and Business Assets for Self-Employed Workers

 

 

 

Men Women
1993 1997 2000 1993 1997 2000
Bus. assets and HH compositions
Business assets (million) -0.020 0.015 0.001 0.142 "‘" 0.046 "" 0.015 *"”'
(1.23) (1 .02) (0.11) (3.86) (3.59) (2.81)
# men, aged 2049 0.261 *“ 0.110 ”"‘ 0.016 0.071 -0.039 0.012
(5.43) (2.86) (0.46) (0.66) (0.50) (0.25)
# women, aged 20-49 0.047 0.160 "" 0.099 "" 0.098 0.013 0.072 ”
(0.84) (5.15) (3.39) (1 .64) (0.27) (2.43)
# men, aged 50+ 0.003 -0.043 0.040 0.019 -0.033 0.014
(0.04) (0.57) (0.59) (0.09) (0.20) (0.12)
# women, aged 50+ 0033 0.218 ”‘ 0.070 0.142 -0.113 -0.003
(0.55) (3.07) (1 .39) (1 .29) (1 .60) (0.04)
Gross income 0.234 ‘" 0.504 "*
(3.87) (5.41)
Selection Coefﬁcient -1.178 "" -1.767 """' -1.374 '“‘”" 0.297 0.694 -0.175
(5.72) (3.94) (4.02) (0.35) (0.71) (0.23)
Constant 5.651 ""‘ 10.844 "* 8.411 "* 3.864 "' 5.824 *" 5.280 "‘
(5.02) (7.21) (7.54) (1.76) (2.18) (1.91)
F -test
Own Schooling 17.69 14.05 13.14 5.97 5.09 5.40
(0.000) (0.000) (0.000) (0.000) (0.001) (0.000)
Father's Schooling 1.31 3.19 1.90 0.76 1.25 0.48
(0.265) (0.014) (0.1 10) (0.554) (0.289) (0.749)
Mother's Schooling 5.75 3.73 5.81 0.99 1.51 0.85
(0.001) (0.01 1) (0.001) (0.399) (0.213) (0.465)
Parental Schooling 5.01 4.21 3.93 1.57 1.53 1.07
(0.000) (0.000) (0.000) (0.144) (0.156) (0.384)
Age 1.63 3.32 2.16 0.93 2.1 l 0.94
(0.167) (0.01 1) (0.072) (0.445) (0.079) (0.443)
Residency 10.39 1 1.77 4.43 7.29 6.07 3.29
(0.000) (0.000) (0.000) (0.000) (0.000) (0.000)
Bus. assets and HH composi 6.58 10.96 3.62 4.92 4.91 3.30
(0.000) (0.000) (0.003) (0.000) (0.000) (0.006)
Adjusted R 2 0.237 0.217 0.1 19 0.204 0.135 0.078
Root MSE 1.088 1.082 1.125 1.147 1.184 1.219
Observations 2,107 2,318 3,023 1,272 1,480 2,047

 

Source: IFLSl, IFLSZ and IFLS3.
Province of residence along with its interaction with urban dununy variables and month of interview dummy variables are
included in the regressions but are not reported. Omitted category for own schooling is no schooling (0-6 years for public
sector workers) and for parental schooling is no schooling. Adjustments are made due to small cell size for some variables
for women: 10+ years for own schooling. Selection coefﬁcients are calculated based on Base Speciﬁcation estimates.
Standard errors are robust to clustering at the community level and heteroskedasticity. Absolute t-statistics are in

parentheses. Signiﬁcant at the l%(""), 5%(") and 10%(‘) indicated. p -va1ues for F -test are in parentheses.

179

Appendix Table 6.3A
Selectivity Corrected Wage Functions:
The Effects of Own Schooling (splines)

 

 

 

 

lFLSl
Men Women
Self Public Private Self Public Private
Employment Sector Sector Employment Sector Sector
Own Schooling (splines)
1-3 years a) 0.063 ‘ -0.014 0.087 *" 0.066 0.027 0.037
(1.92) (0.33) (2.59) (1.39) (0.51) (0.89)
4-6 years 0.113 "”" 0.114 "* 0.120 *" 0.073
(4.21) (3.62) (3.21) (1.52)
7-9 years 0.201 *” 0.078 * 0.079 "" 0.013 0.203 0.235 *“
(6.14) (1 .72) (3.06) (0.30) (1 .59) (3.94)
10-12 years 0.117 ““' 0.097 " 0.147 ‘1'" 0.124 *"‘ 0.087 " 0.153 ""
(2.51) (2.06) (5.38) (2.12) (1.93) (4.03)
13+ years 0.189 ‘ 0.089 ”* 0.128 *"
(1.87) (4.64) (3.53)
Age (spline)
25-29 years 0.047 -0.007 0.074 "'" 0.102 " 0.071 0.104 "
(1 .22) (0.11) (2.68) (1.91) (1.08) (2.56)
30-39 years 0.010 0.040 “" 0.006 0.020 0.023 ‘ -0.008
(0.88) (3.16) (0.73) (1.26) (1.75) (0.54)
40-49 years -0.023 *" 0.020 0.014 -0.022 0.037 ” 0.000
(2.23) (1.65) (1.61) (1.64) (2.25) (0.01)
50-59 years 0.014 -0.005 -0.054 " 0.012 -0.007 -0.026
(0.96) (0.26) (2.60) (0.56) (0.18) (1 .01)
Month of interview
October 0.320 “ -0.127 0.121 0.392 ”* 0.041 0.061
(2.21) (0.83) (0.98) (2.69) (0.22) (0.41)
November 0.429 *" -0.243 0.140 0.413 ”* -0.175 -0.125
(3.31) (1.52) (1.15) (2.72) (1.06) (0.91)
December -0.092 -0.058 0.081 0.144 0.018 -0. 150
(0.72) (0.40) (0.64) (0.96) (0.12) (1 .06)
January 0019 0.066 0.244 0.091 0.324 " -0.083
(0.11) (0.36) (1 .58) (0.39) (1.76) (0.39)
Gross Income 0.074 0.323 "*
(1.07) (3.13)
Selection Coefﬁcient -0.637 "'“ -0.196 0.080 0.810 "'“ 0.207 -0.109
(3.78) (1.14) (1 .10) (2.88) (0.76) (0.78)
Constant 4.309 "" 6.989 "‘" 3.282 "“” 1.129 4.311 " 2.549 ‘”"
(4.00) (3.94) (4. 13) (0.70) (1 .69) (2.24)
F -test
Own Schooling 26.92 6.23 58.74 12.55 1.54 54.49
(0.000) (0.000) (0.000) (0.000) (0.208) (0.000)
Age 2.32 7.85 (0.000) 3.15 14.57 2.39
(0.057) (0.000) (0.000) (0.015) (0.000) (0.052)
Month of interview 7.62 1.91 0.73 3.12 2.28 1.06
(0.000) (0.1 10) (0.573) (0.016) (0.064) (0.380)
Adjusted R 2 0.1 17 0.305 0.265 0.091 0.482 0.296
Root MSE 1.171 0.661 0.821 1.226 0.585 0.910
Observations 2,107 574 1,391 1,272 252 616

 

Source: IFLS l.

a) 1-6 years for public sector workers.

Omitted category for month of interview is August/September. Adjustments are made due to small cell size for some
variables for women: 10+ years for own schooling. Selection coefﬁcients are calculated based on Base Speciﬁcation
estimates in which dummy variables for own schooling are replaced its splines. Standard errors are robust to clustering
at the community level and heteroskedasticity. Absolute t—statistics are in parentheses. Signiﬁcant at the l°/o("“""'),
5%(") and 10%(*) indicated. p -va1ues for F -test are in parentheses.

180

Appendix Table 6.3B
Selectivity Corrected Wage Functions:
The Effects of Own Schooling (splines)

 

 

 

 

IFLSZ
Men Women
Self Public Private Self Public Private
Employment Sector Sector Employment Sector Sector
Own Schoollng (splines)
1-3 years a) 0.053 0.003 0.052 0.053 -0.067 0.105 "
(1 .49) (0.10) (1 .62) (1 .24) (1.14) (2.45)
4-6 years 0.083 "" 0.080 “* 0.043 0.041
(2.64) (3.07) (1 .07) (0.95)
7-9 years 0.140 "" 0.035 0.136 “"' 0.097 ” 0.101 0.185 ""
(4.51) (0.70) (6.01) (2.13) (0.83) (5.01)
10-12 years 0.066 ‘ 0.050 0.043 "' 0.160 "‘ 0.029 0.113 ""'
(1 .68) (1.02) (1.92) (3.37) (0.70) (4.32)
13+ years 0.209 ”"‘ 0.063 ““ 0.127 "'"
(4.65) (3.31) (5.91)
Age (spline)
25-29 years -0.002 0.030 0.045 " -0.037 0.062 0.072 "'*
(0.05) (0.35) (2.28) (0.70) (0.84) (2.40)
30-39 years 0.019 " 0.012 0.023 "‘ 0.042 "* 0.021 0.024 "
(1.78) (0.73) (3.12) (2.83) (1.03) (1.94)
40-49 years -0.017 " 0.026 "* 0.000 0024 " 0.021 0.018
(1 .73) (2.68) (0.02) (2.04) (1.31) (1 .20)
50-59 years -0.011 0.005 -0.028 ‘ 0.010 0.025 -0.054 "'
(0.90) (0.26) (1 .84) (0.67) (0.94) (2. 14)
Month of interview
October -0.137 -0.171 ‘ -0.044 -0.192 -0.227 " 0.108
(1.22) (1.93) (0.56) (1.61) (2.00) (0.94)
November -0.204 " -0.183 " -0.039 -0.092 -0.354 "‘ 0.070
(1.83) (1 .92) (0.53) (0.80) (2.53) (0.59)
December 0.045 -0.081 0.035 0.154 -0.144 0.027
(0.40) (0.86) (0.43) (1.04) (1.12) (0.22)
January-April -0.060 -0.006 0.148 ‘ 0.210 0.068 0.360 ”
(0.43) (0.06) (1.71) (0.93) (0.52) (2.25)
Selection Coefﬁcient -0.147 -0.293 -0.051 -0.096 -0.285 -0.718 ""'
(0.76) (1 .52) (0.44) (0.41) (1 .43) (3.09)
Constant 6.277 *" 6.666 ” 4.907 “" 6.984 """ 6.139 ” 4.441 ""'
(4.60) (2.46) (8.32) (4.36) (2.46) (5.64)
F -test
Own Schooling 21.99 2.79 57.37 15.57 0.56 51.36
(0.000) (0.027) (0.000) (0.000) (0.645) (0.000)
Age 2.52 5.89 10.38 2.11 4.62 6.99
(0.041) (0.000) (0.000) (0.079) (0.001) (0.000)
Month of interview 1.81 1.82 1.71 2.35 3.16 1.58
(0.127) (0.124) (0.146) (0.054) (0.015) (0.179)
Adjusted R ’ o. 100 0.221 0.195 0.065 0.245 0.306
Root MSE 1.160 0.641 0.820 1.231 0.582 0.945
Observations 2,3 1 8 645 2,099 1,480 309 992

 

Source: IFLSZ.

a) 1-6 years for public sector workers.
Omitted category for month of interview is August/September. Adjustments are made due to small cell size for some
variables for women: 10+ years for own schooling. Selection coefficients are calculated based on Base Speciﬁcation
estimates in which dummy variables for own schooling are replaced its splines. Standard errors are robust to clustering

at the community level and heteroskedasticity. Absolute t-statistics are in parentheses. Signiﬁcant at the 1%("’),

5°/o(“") and 10°/o("‘) indicated. p -va1ues for F -test are in parentheses.

181

Appendix Table 6.3C

Selectivity Corrected Wage Functions:
The Effects of Own Schooling (splines)

 

 

 

IFLS3
Men Women
Self Public Private Self Public Private
Employment Sector Sector Employment Sector Sector
Own Schooling (splines)
1-3 years a) 0.069 " -0.078 0.095 ”" 0.081 " —0.072 0.134 ""
(2.12) (1 . l 1) (2.89) (2.00) (0.61) (3.39)
4-6 years 0.060 ” 0.040 0.013 -0.002
(2. 10) (1.60) (0.35) (0.04)
7-9 years 0.097 "* 0.092 0.080 "‘ 0.076 ”“ 0.372 "' 0.165 ”"
(3.91 ) (1 .60) (4.66) (2.70) (1 .97) (4.67)
10-12 years 0.126 “" 0.040 0.094 "”" 0.094 " 0.035 0.135 ""‘
(4.16) (0.57) (5.80) (2.53) (0.68) (6.04)
13+ years 0.132 "““ 0.053 " 0.125 ‘"
(4.72) (1 .98) (7.74)
Age (spline)
25-29 years 0.034 0.038 0.003 0.091 ‘ 0.060 0.016
(1.17) (0.60) (0.21) (1 .83) (0.61) (0.61)
30-39 years 0.010 0.022 0.018 "" 0.019 0.012 0.013
(1 .13) (1 .00) (3.03) (1.42) (0.42) (1.31)
40-49 years 0.005 0.016 0.008 0.004 0.006 0.003
(0.62) (1 .55) (0.96) (0.35) (0.45) (0.24)
50-59 years 0032 "‘ 0.033 -0.032 " -0.020 0.059 " -0.022
(2.73) (1 .60) (2.52) (1.31) (2.39) (1 .06)
Month of interview
August -0.173 " 0.003 0.014 -0.121 0.102 -0.014
(2.30) (0.03) (0.26) (1.12) (0.74) (0.13)
September -0. 132 -0.059 0.008 -0.1 16 0.020 -0.047
(1.50) (0.68) (0.16) (1.04) (0.15) (0.42)
October -0.094 0.002 0.129 " -0.113 0.217 0.001
(0.98) (0.03) (2.34) (1 .00) (1.40) (0.01)
November-January 0.026 0.193 "' 0.222 ““ -0.090 0.043 0.208 ‘
(0.26) (1.71) (3.79) (0.68) (0.26) (1 .88)
Selection Coefﬁcient -0.501 ”" -0.370 0.035 0.258 -0.258 -0.053
(3.07) (1.39) (0.46) (0.84) (1.13) (0.26)
Constant 6.333 “* 7.425 "" 6.500 "" 3.696 "”" 5.869 " 5.711 ""‘
(7.32) (3.23) (15.28) (2.27) (1.90) (8.09)
F -test
Own Schooling 24.07 2.16 79.60 13.74 2.22 68.36
(0.000) (0.073) (0.000) (0.000) (0.087) (0.000)
Age 3.24 4.21 7.16 2.18 4.10 1.49
(0.012) (0.002) (0.000) (0.071) (0.003) (0.205)
Month of interview 1.94 2.00 5.26 0.37 0.73 2.53
(0.102) (0.093) (0.000) (0.832) (0.570) (0.040)
Adjusted R 2 0.074 0.255 0.187 0.045 0.300 0.287
Root MSE 1.153 0.678 0.807 1.240 0.673 0.930
Observations 3,023 702 2,801 2,047 352 1 ,3 76

 

Source: IF LS3.

a) 1-6 years for public sector workers.
Omitted category for month of interview is June/July. Adjustments are made due to small cell size for some variables
for women: 10+ years for own schooling. Selection coefﬁcients are calculated based on Base Speciﬁcation estimates
in which dummy variables for own schooling are replaced its splines. Standard errors are robust to clustering at the

community level and heteroskedasticity. Absolute r-statistics are in parentheses. Signiﬁcant at the 1°/o("'"), 5%(")
and 10%(*) indicated. p -values for F -test are in parentheses.

182

 

i

Appendix Table 6.4A
Selectivity Corrected Wage Functions:
The Effects of Own Schoollng (splines) and Parental Schoollng

 

 

 

 

lFLSl
Meir Wor_r_ien
Self Public Private Self Public Private
Employment Sector Sector Employment Sector Sector
Own Schoollng (splines)
1-3 years a) 0.042 -0.023 0.086 " 0.052 0.024 0.024
(1.31) (0.55) (2.50) (1.11) (0.47) (0.55)
4-6 years 0.107 “" 0.108 ”‘ 0.100 ‘“ 0.065
(3.91) (3.45) (2.72) (1.31)
7-9 years 0.195 ”" 0.071 0.074 “' -0.001 0.075 0.216 ”‘
(6.05) (1.54) (2.80) (0.02) (0.48) (3.58)
10-12 years 0.113 ” 0.078 0.134 ”‘ 0.094 0.001 0.130 ‘”
(2.46) (1.61) (4.85) (1.55) (0.02) (3.16)
13+ years 0.178 ‘ 0.080 ”‘ 0.117 ‘”
(1.73) (3.83) (3.22)
F ather's Schooling
Some Elementary 0.190 ” 0.135 -0.009 0.128 -0.189 0.118
(2.30) (1.40) (0.12) (1.04) (1 .25) (0.97)
Completed Elementary 0.156 0.015 0.030 0.166 -0.361 ” 0.238 "
(1.55) (0.16) (0.36) (1.40) (2.42) (1.73)
Secondary/1" ertiary 0.204 -0.020 0.162 ' 0.252 ~0.055 0.276
(1.22) (0.18) (1.67) (1.27) (0.34) (1.38)
Missing 0.089 0.071 -0.017 -0.054 -0.118 0.133
(0.98) (0.57) (0.19) (0.47) (0.65) (1.10)
Mother's Schooling
Some Elementary 0.185 " -0.039 0.019 0.142 0.182 ' 0.021
(1.98) (0.42) (0.26) (1.14) (1 .73) (0.16)
Completed Elementary/ 0.361 “‘ 0.155 " 0.099 0.257 ‘ 0.182 0.107
Secondary/Tertiary (3.31) (1.93) (1.17) (1.83) (1.59) (0.70)
Missing 0.174 “ 0.092 0.058 0.092 0.021 -0.077
(2.21) (0.87) (0.68) (0.86) (0.12) (0.59)
Age (spline)
25-29 years 0.047 -0.011 0.074 “" 0.090 ‘ -0.006 0.107 “
( 1.20) (0.19) (2.68) (1.67) (0.08) (2.60)
30-39 years 0.014 0.036 ”‘ 0.006 0.027 0.027 ‘ -0.007
( 1.24) (2.92) (0.80) (1 .65) (1 .97) (0.44)
40-49 years -0.020 " 0.021 0.014 -0.022 0.040 " 0.003
(1.97) (1.65) (1.57) (1.57) (2.42) (0.17)
50-59 years 0.014 0.003 -0.054 ” 0.012 0.002 -0.027
( 1.00) (0.13) (2.57) (0.55) (0.06) (1.09)
Gross income 0.088 0.318 ‘“
(1.29) (3.13)
Selection Coefﬁcient -0.760 ‘” -0.271 0.110 0.827 ‘” -0.625 -0.120
(4.29) (1 .49) (1 .32) (2.75) (1 .59) (0.77)
Constant 4.214 ‘" 7.236 ‘“ 3.234 "‘ 1.304 7.796 " 2.434 ”
(3.86) (4.09) (4.09) (0.80) (2.25) (2.08)
F -test
Own Schooling 22.88 4.32 41.21 6.96 0.48 20.05
(0.000) (0.002) (0.000) (0.000) (0.693) (0.000)
Father's Schooling 1.56 0.92 1.09 0.98 3.15 0.94
(0.185) (0.452) (0.363) (0.420) (0.016) (0.442)
Mother’s Schooling 4.80 1.92 0.54 1.27 1.74 0.34
(0.003) (0.127) (0.653) (0.285) (0.162) (0.794)
Parental Schooling 5.17 1.21 1.23 1.68 2.08 1.15
(0.000) (0.296) (0.286) (0.1 15) (0.050) (0.333)
Age 2.28 7.15 5.97 3.07 12.84 2.60
(0.061) (0.000) (0.000) (0.017) (0.000) (0.037)
Adjusted R 3 0.130 0.306 0.265 0.095 0.489 0.296
Root MSE 1.162 0.661 0.821 1.223 0.581 0.910
Observations 2,107 574 1,391 1,272 252 616
Source: lFLSl.

a) 1-6 years for public sector workers.
Month of interview dummy variables are included in the regressions but are not reported. Omitted category for parental schooling

is no schooling. Adjustments are made due to small cell size for some variables for women: 10+ years for own schooling.
Selection coefﬁcients are calculated based on Base Speciﬁcation estimates in which dummy variables for own schooling are
replaced its splines. Standard errors are robust to clustering at the community level and heteroskedasticity. Absolute r-statistics
are in parentheses. Signiﬁcant at the l°/o(‘“). 5%(”) and 10°/o(’) indicated. p -values for F -test are in parentheses.

183

 

Appendix Table 6.4B
Selectivity Corrected Wage Functions:
The Effects of Own Schooling (splines) and Parental Schoollng

 

 

 

 

iFLSZ
Men Women
Self Public Private Self Public Private
Employment Sector Sector Employment Sector Sector
Own Schooling (splines)
1-3 years a) 0.005 0.019 0.043 0.059 «0.069 0.072
(0.15) (0.55) (1.36) (1.35) (1.14) (1.62)
4-6 years 0.103 """ 0.078 ”" 0.021 0.014
(3.24) (3.00) (0.52) (0.31)
7-9 years 0.156 "‘ 0.079 0.117 "" 0.060 0.099 0.147 “"
(5.14) (1.65) (5.20) (1.26) (0.67) (3.96)
10-12 years 0.113 "" 0.110 " 0.028 0.126 ” 0.018 0.122 ”‘
(2.88) (2.37) (1.30) (2.58) (0.33) (4.67)
13+ years 0.228 ”‘ 0.081 *" 0.105 “‘"
(5.36) (4.68) (4.70)
Father's Schooling
Some Elementary 0.295 “" -0.043 0.057 0.172 " -0.064 0.009
(3 .20) (0.44) (0.86) (1.77) (0.42) (0.07)
Completed Elementary 0.194 ‘ 0.011 0.100 0.322 ‘“ -0.106 0.104
(1.95) (0.1 1) (1 .52) (2.75) (0.77) (0.87)
Secondary/'1" ertiary 0.510 "‘ -0.175 0.221 ”" 0.488 ”" -0.071 0.307 “
(3.69) (1 .53) (2.71) (2.70) (0.47) (2.09)
Missing 0.286 ” -0.141 0.052 0.114 -0.207 0.016
(2.23) (0.73) (0.61) (0.82) (0.92) (0.13)
Mother's Schooling
Some Elementary 0.069 0.061 0.108 ‘ 0.079 0.112 0.030
(0.75) (0.68) (1.81) (0.65) (0.98) (0.26)
Completed Elementary/ 0.354 “" 0.180 " 0.210 ”" -0.091 0.082 0.202 ”
Secondary/Tertiary (3.56) (2 .33) (3.51) (0.65) (0.74) (2 .02)
Missing 0.029 0.057 0.132 -0.196 0.294 0.121
(0.24) (0.29) (1.62) (1.21) (1.07) (0.97)
Age (spline)
25-29 years 0059 0.069 0.049 ” -0.022 0.057 0.056 ‘
(1 .39) (0.85) (2.51) (0.40) (0.71) (1 .85)
30-39 years 0.012 0.031 " 0.020 “" 0.054 ”’ 0.016 0.015
(1.12) (2.11) (2.73) (3.40) (0.66) (1 .21)
40-49 years 0021 ” 0.030 ”‘ -0.003 -0.021 ’ 0.025 0.020
(2.12) (3.07) (0.30) (1 .82) (1.36) (1.29)
50-59 years —0.001 0.008 -0.028 ‘ 0.011 0.023 -0.071 ”"
(0.1 1) (0.42) (1.84) (0.75) (0.84) (2.74)
Selection Coefficient -0.897 ‘“ 0.003 0.161 0.225 -0.336 -0.275
(4.37) (0.02) (1.26) (0.76) (1 .24) (1 .19)
Constant 8.413 “‘ 4.640 ‘ 4.521 ‘” 5.955 ‘" 6.462 ” 4.262 ‘”
(6.54) (1.81) (7.78) (3.51) (2.19) (5.41)
F -test
Own Schooling 26.46 6.19 36.39 7.30 0.44 29.01
(0.000) (0.000) (0.000) (0.000) (0.724) (0.000)
Father’s Schooling 4.34 1.92 1.89 2.49 0.32 1.61
(0.002) (0.107) (0.1 1 1) (0.043) (0.862) (0.172)
Mother's Schooling 4.95 1.86 4.26 1.13 0.53 1.69
(0.002) (0.136) (0.006) (0.338) (0.664) (0.169)
Parental Schooling 7.02 1.74 5.80 2.52 0.32 2.96
(0.000) (0.099) (0.000) (0.015) (0.946) (0.005)
Age 2.01 7.71 8.88 2.94 3.83 5.34
(0.093) (0.000) (0.000) (0.021) (0.005) (0.000)
Adjusted R 2 0.118 0.225 0.207 0.073 0.232 0.312
Root MSE 1.148 0.640 0.814 1.226 0.587 0.940
Observations 2,3 1 8 645 2,099 1,480 309 992
Source: IFLSZ.

a) 1-6 years for public sector workers.

Month of interview dummy variables are included in the regressions but are not reported. Omitted category for parental schooling
is no schooling. Adjustments are made due to small cell size for some variables for women: 10+ years for own schooling.
Selection coefﬁcients are calculated based on Base Speciﬁcation estimates in which dummy variables for own schooling are
replaced its splines. Standard errors are robust to clustering at the community level and heteroskedasticity. Absolute r-statistics
are in parentheses. Signiﬁcant at the 1%(“"), 5%(“) and 10%(') indicated. p -values for F -test are in parentheses.

184

Appendix Table 6.4C
Selectivity Corrected Wage Functions:

The Effects of Own Schooling (splines) and Parental Schooling

 

 

 

 

lFLS3
Men Women
Self Public Private Self Public Private
Employment Sector Sector Employment Sector Sector
Own Schooling (splines)
1-3 years a) 0.052 -0.079 0.093 "W 0.069 " -0.073 0.125 "‘
(1 .62) (1 .04) (2.83) (1 .67) (0.61) (3.07)
4-6 years 0.068 ” 0.038 0.005 -0.008
(2.42) (1 .48) (0.15) (0.25)
7-9 years 0.097 ”‘ 0.093 0.074 "" 0.064 " 0.166 0.122 ""
(3.89) (1.51) (4.34) (2.22) (0.82) (3.33)
10-12 years 0.145 ‘” 0.036 0.082 “" 0.084 " -0.107 0.129 ”‘
(4.70) (0.44) (4.94) (2.13) (1.25) (5.88)
13+ years 0.123 ”" 0.049 0.108 ""
(4.21) (1 .62) (6.42)
Father's Schooling
Some Elementary 0068 -0.071 -0.022 0.055 -0.017 -0.045
(1 .00) (0.71) (0.38) (0.54) (0.09) (0.49)
Completed Elementary 0.031 -0.020 0.025 0.086 0.01 1 0.031
(0.44) (0.23) (0.50) (1 .03) (0.06) (0.34)
Secondary/Tertiary 0.155 0.073 0.110 "' 0.201 ’ 0.277 0.358 ”"
(1.62) (0.72) (1.92) (1.69) (1.46) (3.51)
Missing -0.157 “ 0.028 -0.060 0.119 0.202 0.083
(1.91) (0.17) (0.91) (1.19) (0.69) (0.89)
Mother's Schooling
Some Elementary 0.163 ” -0.001 0.027 0.042 0.036 -0.060
(2.27) (0.01) (0.51) (0.41) (0.36) (0.71)
Completed Elementary/ 0.323 ‘” -0.001 0.090 ‘ 0.125 0.082 0.019
Secondaryffertiary (4.31) (0.02) (1 .86) (1 .36) (0.69) (0.22)
Missing 0.308 "‘“ -0.137 0.120 0.106 -0.197 -0.105
(3.18) (0.79) (1.60) (0.92) (0.77) (1.03)
Age (spline)
25-29 years 0.021 0.040 0.004 0.082 0.023 0.004
(0.70) (0.60) (0.27) (1.61) (0.23) (0.14)
30-39 years 0.009 0.023 0.017 "'" 0.015 -0.052 0.012
(0.99) (0.94) (2.81) (1.17) (1.25) (1.20)
40-49 years 0.004 0.017 0.007 0.003 0.022 0001
(0.52) (1.38) (0.80) (0.29) (1.41) (0.05)
50-59 years 0030 “ 0.034 -0.034 "" -0.019 0.070 “" —0.029
(2.60) (1.52) (2.71) (1.22) (2.67) (1.35)
Selection Coefficient -0.906 "‘ -0.364 0.136 0.118 -0.897 “ 0.228
(4.65) (1.14) (1 .43) (0.34) (2.20) (1 .04)
Constant 6.980 "‘ 7.357 “" 6.368 ”" 4.090 " 9.261 “' 5.685 "‘"
(7.77) (2.81) (14.71) (2.35) (2.56) (8.22)
F -test
Own Schooling 23.01 2.12 54.25 7.36 2.04 38.79
(0.000) (0.077) (0.000) (0.000) (0.109) (0.000)
Father's Schooling 2.18 0.59 1.65 0.90 1.26 5.22
(0.069) (0.669) (0.160) (0.466) (0.288) (0.000)
Mother's Schooling 7.74 0.23 1.62 0.77 0.60 0.64
(0.000) (0.879) (0.182) (0.513) (0.615) (0.588)
Parental Schooling 5.54 0.53 2.42 1.41 1.20 3.36
(0.000) (0.808) (0.018) (0.197) (0.301) (0.002)
Age 2.48 4.26 6.74 1.54 4.60 1.28
(0.043) (0.002) (0.000) (0.189) (0.001) (0.275)
Adjusted R 2 0.085 0.252 0.190 0.046 0.304 0.295
Root MSE 1.146 0.680 0.806 1.239 0.671 0.926
Observations 3,023 702 2,801 2,047 352 1,376

 

Source: iFLS3.

a) 1-6 years for public sector workers.

Month of interview dummy variables are included in the regressions but are not reported. Omitted category for parental schooling

is no schooling. Adjustments are made due to small cell size for some variables for women: 10+ years for own schooling.
Selection coefﬁcients are calculated based on Base Speciﬁcation estimates in which dummy variables for own schooling are
replaced its splines. Standard errors are robust to clustering at the community level and heteroskedasticity. Absolute t-statistics

are in parentheses. Signiﬁcant at the 1%(‘"), 5%(“) and lO°/o(") indicated. p -values for F -test are in parentheses.

185

f

 

Appendix Table 6.5A
OLS Wage Functions:
The Effects of Non-linear Own Schoollng and Parental Schooling

 

 

 

 

lFLSl
Men Women
Self Public Private Self Public Private
Employment Sector Sector Employment Sector Sector
Own Schoollng
1-3 years 0.093 0.108 0.050 0.041
(1.05) (1 .21) (0.44) (0.39)
4-6 years 0.364 ”" 0.484 ”‘ 0.317 “”' 0.227 "
(4.01) (6.18) (2.80) (2.01)
7-9 years 0.789 '” 0.207 ‘ 0.743 "‘ 0.552 “‘ 0.498 “' 0.618 ‘”
(6.94) (1.73) (8.22) (3.39) (2.29) (3.84)
10-12 years 0.811 ‘“ 0.665 ”‘ 1.100 "" 0.897 ‘“ 1.351 ‘“ 1.498 ‘”
(6.58) (6.74) (11.33) (4.95) (9.23) (8.66)
13+ years 1.485 ”‘ 1.025 ”‘ 1.700 ‘”
(6. 10) (9.48) (11.81)
Father's Schooling
Some Elementary 0.235 ‘” 0.104 -0.003 0.064 «0.018 0.098
(2.85) (1.06) (0.05) (0.53) (0.12) (0.85)
Completed Elementary 0.156 0.008 0.056 0.057 -O.217 0.184
(1 .55) (0.08) (0.69) (0.52) (1 .47) (1 .43)
Secondary/Tertiary 0.254 -0.032 0.158 0.292 -0.007 0.236
(1.56) (0.29) (1.63) (1.55) (0.05) (1.13)
Missing 0.023 0.015 -0.033 .0068 -0.089 0.120
(0.26) (0.12) (0.37) (0.58) (0.52) (1.00)
Mother's Schooling
Some Elementary 0.093 —0.005 0.013 0.226 ‘ 0.109 0.036
(1.03) (0.06) (0.17) (1.84) (1.02) (0.31)
Completed Elementary/ 0.300 ”‘ 0.163 ‘ 0.087 0.344 " 0.163 0.125
Secondary/Tertiary (2.78) (1 .94) (1 .08) (2.44) (1.54) (0.84)
Missing 0.152 ‘ 0.104 0.073 0.117 0.105 -0.075
(1.95) (1.02) (0.85) (1.06) (0.93) (0.59)
Age (spline)
25-29 years 0.029 0.010 0.072 "” 0.060 0.107 ” 0.103 “
(0.75) (0.20) (2.67) (1.13) (2.26) (2.50)
30-39 years 0.026 “ 0.046 ‘“ 0.007 0.007 0.036 ”‘ -0.008
(2 .40) (4.21) (0.94) (0.48) (2.68) (0.54)
40-49 years -0.020 ” 0.024 "' 0.015 ‘ -0.019 0.030 ” 0.002
(2.01) (1.97) (1.73) (1.37) (2.18) (0.15)
50-59 years 0.014 -0.013 -0.053 "" 0.010 -0.012 -0.030
(0.95) (0.65) (2 .53) (0.44) (0.38) (1.20)
Gross income 0.079 0.315 “‘
(1.13) (3.09)
Constant 4.269 "" 5.913 "‘ 3.414 ‘" 3.416 "' 2.816 " 2.360 “
(3.89) (3.86) (4.47) (2.28) (2.13) (2.05)
F -test
Own Schooling 20.24 34.17 45.03 8.32 58.40 20.79
(0.000) (0.000) (0.000) (0.000) (0.000) (0.000)
Father's Schooling 2.58 0.62 1.06 0.83 1.81 0.68
(0.038) (0.647) (0.376) (0.506) (0.130) (0.606)
Mother's Schooling 3.32 1.72 0.55 2.45 0.84 0.43
(0.020) (0.164) (0.650) (0.064) (0.474) (0.732)
Parental Schooling 4.46 1.02 1.23 2.08 1.23 0.97
(0.000) (0.416) (0.284) (0.046) (0.290) (0.451)
Age 3.42 15.65 6.49 1.25 16.82 2.36
(0.009) (0.000) (0.000) (0.289) (0.000) (0.055)
Adjusted R 3 0.1 17 0.297 0.265 0.088 0.486 0.304
Root MSE 1.171 0.665 0.820 1.228 0.583 0.905
Observations 2,107 574 1,391 1,272 252 616
Source: lFLSl.

Month of interview dummy variables are included in the regressions but are not reported. Omitted category for own
schooling is no schooling (0-6 years for public sector workers) and for parental schooling is no schooling. Adjustments are
made due to small cell size for some variables for women: 10+ years for own schooling. Standard errors are robust to
clustering at the community level and heteroskedasticity. Absolute t-statistics are in parentheses. Signiﬁcant at the
l°/o(""), 5°/o("") and lO%(") indicated. p -va1ues for F -test are in parentheses.

186

Appendix Table 6.58

 

 

 

 

OLS Wage Functions:
The Effects of Non-linear Own Schooling and Parental Schoollng
IFLSZ
Men Women
Self Public Private Self Public Private
Employment Sector Sector Employment Sector Sector
Own Schooling
1-3 years 0.218 " 0.106 0.171 0.019
(2.44) (1.18) (1.59) (0.16)
4-6 years 0.363 ”t 0.319 “" 0.202 ” 0.174 "
(4.37) (4.28) (2.07) (1 .67)
7-9 years 0.729 *” 0.162 0.692 ”"' 0.546 ”'" 0.403 0.432 ‘"
(7.50) (1 .42) (7.96) (3.70) (1 .60) (2.80)
10-12 years 0.837 "" 0.528 "‘ 0.791 "" 0.802 ”" 0.706 ‘” 1.162 “"
(8.18) (5.72) (9.48) (5.03) (2.94) (8.74)
13+ years 1.551 ”" 0.849 "" 1.251 ‘“
(8.85) (9.69) (10.59)
Father's Schooling
Some Elementary 0.226 " -0.042 0.041 0.152 -0.028 0.026
(2.51) (0.44) (0.63) (1 .62) (0.20) (0.21)
Completed Elementary 0.114 0.032 0.091 0.294 ‘"' -0.070 0.101
(1.14) (0.34) (1 .40) (2.60) (0.53) (0.86)
Secondary/Tertiary 0.357 *" -0.149 0.211 "* 0.484 "" -0.066 0.382 “"
(2.64) (1 .32) (2.59) (2.76) (0.44) (2.76)
Missing 0.145 -0.140 0.031 0.101 -0.283 0.053
(1.15) (0.71) (0.37) (0.72) (1.30) (0.46)
Mother's Schooling
Some Elementary 0.019 0.058 0.104 ‘ 0.092 0.109 0.009
(0.21) (0.66) (1 .76) (0.79) (0.96) (0.08)
Completed Elementary/ 0.258 ‘" 0.161 ‘”“ 0.188 "‘" -0.051 0.145 0.233 "
Secondary/Tertiary (2.64) (2.13) (3.20) (0.40) (1 .33) (2.29)
Missing -0.027 0.057 0.109 -0.153 0.337 0.130
(0.23) (0.29) (1.36) (1 .01) (1 .16) (1 .05)
Age (spline)
25-29 years 0.006 0.067 0.046 “ -0.038 0.064 0.052 "
(0.16) (0.86) (2.40) (0.72) (0.92) (1 .82)
30-39 years 0.020 ‘ 0.029 ”“ 0.022 ‘"” 0.046 "'” 0.048 “" 0.007
(1.90) (2.86) (3.33) (3.53) (2.97) (0.61)
40-49 years 0012 0.031 “"' 0.000 —0.021 " 0.007 0.019
(1 .23) (3.21 ) (0.04) (1 .80) (0.53) (1 .22)
50-59 years 0009 -0.007 -0.024 0.011 0.025 -0.078 "“
(0.73) (0.38) (1 .64) (0.76) (0.95) (3.06)
Constant 5.722 “‘ 4.823 “ 4.743 ""” 6.758 "" 4.878 "' 4.046 ‘"
(5.17) (2.14) (8.72) (4.46) (2.45) (5.08)
F -test
Own Schooling 27.09 42.05 37.33 7.80 5.57 33.61
(0.000) (0.000) (0.000) (0.000) (0.004) (0.000)
Father's Schooling 2.57 2.04 1.80 2.46 0.47 3.08
(0.037) (0.089) (0.129) (0.045) (0.757) (0.016)
Mother's Schooling 3.03 1.54 3.50 0.88 0.70 2.26
(0.029) (0.205) (0.015) (0.449) (0.555) (0.082)
Parental Schooling 4.25 2.03 5.14 2.58 0.52 5.14
(0.000) (0.051) (0.000) (0.013) (0.817) (0.000)
Age 1.96 16.85 10.51 3.38 6.25 5.04
(0.100) (0.000) (0.000) (0.010) (0.000) (0.001)
Adjusted R2 0.113 0.224 0.207 0.071 0.186 0.307
Root MSE 1.151 0.640 0.814 1.227 0.604 0.944
Observations 2,31 8 645 2,099 l ,480 309 992
Source: IFLSZ.

Month of interview dummy variables are included in the regressions but are not reported. Omitted category for own
schooling is no schooling (0-6 years for public sector workers) and for parental schooling is no schooling. Adjustments are
made due to small cell size for some variables for women: 10+ years for own schooling. Standard errors are robust to
clustering at the community level and heteroskedasticity. Absolute t-statistics are in parentheses. Signiﬁcant at the
1%("‘), 5%(”) and 10%(") indicated. p -va1ues for F otest are in parentheses.

187

Appendix Table 6.5C

 

 

 

OLS Wage Functions:
The Effects of Non-linear Own Schooling and Parental Schooling
lFLS3
Men Women
Self Public Private Self Public Private
Employment Sector Sector Employment Sector Sector
Own Schooling
1-3 years 0.213 ” 0.186 ‘ 0.209 ” 0.320 ‘"
(2.05) (1.93) (2.03) (3.28)
4-6 years 0.359 ”" 0.373 ‘" 0.237 ”" 0.383 “‘
(3.92) (4.61) (2.64) (4.04)
7-9 years 0.577 "‘ 0.217 0.608 ‘” 0.425 ”" 0.602 ‘”" 0.710 ”‘
(5.61) (1.37) (7.04) (3.82) (2.02) (5.59)
10-12 years 0.770 ‘” 0.654 “" 0.843 ”"' 0.711 “" 1.518 ”" 1.245 ”‘
(7.79) (5.00) (9.80) (5.48) (5.73) (10.06)
13+ years 1.237 ”‘ 0.992 ”‘ 1.302 ”‘
(9.58) (7.44) (12.36)
Father's Schooling
Some Elementary 0054 -0.057 -0.024 0.048 0.065 -0.041
(0.80) (0.57) (0.41) (0.47) (0.36) (0.45)
Completed Elementary 0.013 -0.004 0.023 0.080 0.126 0.043
(0.19) (0.04) (0.45) (0.99) (0.74) (0.47)
Secondary/Tertiary 0.056 0.080 0.110 " 0.248 " 0.220 0.430 “"
(0.61) (0.83) (1 .93) (2.08) (1 .23) (4.14)
Missing -0.183 “ -0.030 -0.066 0.114 0.077 0.060
(2.26) (0.21) (1.01) (1.14) (0.32) (0.65)
Mother's Schooling
Some Elementary 0.079 0.003 0.029 0.033 0.027 -0.078
(1.14) (0.03) (0.56) (0.33) (0.25) (0.90)
Completed Elementary/ 0.209 ‘” -0.018 0.082 ‘ 0.139 0.014 0.029
Secondary/Tertiary (2.82) (0.22) (1.73) (1.54) (0.12) (0.34)
Missing 0.231 ” -0.176 0.109 0.117 -0.334 -0.105
(2.46) (1.02) (1.47) (1.09) (1.63) (1.03)
Age (spline)
25-29 years 0.067 ” 0.076 0.005 0.074 " 0.092 0.004
(2.45) (1.19) (0.37) (1.66) (0.88) (0.17)
30-39 years 0.016 ‘ 0.046 “" 0.020 '” 0.011 0.038 ‘” 0.012
(1 .71) (4.53) (3.48) (0.88) (2.79) (1 .18)
40-49 years 0.009 0.027 ”" 0.008 0.002 0.004 0.001
(1 .10) (2.91 ) (0.98) (0.16) (0.29) (0.09)
50-59 years 0027 ” 0.015 -0.031 " -0.015 0.040 " -0.019
(2.42) (0.86) (2.49) (1 .05) (1 .78) (0.89)
Constant 4.918 ‘“ 4.944 ”" 6.435 ‘" 4.473 "‘ 3.716 5.948 ‘“
(6.45) (2.74) (15.15) (3.56) (1.24) (8.66)
F -test
Own Schooling 28.14 34.52 52.07 8.14 19.25 28.23
(0.000) (0.000) (0.000) (0.000) (0.000) (0.000)
Father's Schooling 1.64 0.53 1.76 1.21 0.56 6.84
(0.162) (0.714) (0.134) (0.307) (0.691) (0.000)
Mother’s Schooling 3.85 0.41 1.39 1.08 1.18 0.88
(0.010) (0.748) (0.244) (0.359) (0.317) (0.451)
Parental Schooling 3.07 0.59 2.53 1.87 1.02 4.80
(0.003) (0.768) (0.014) (0.072) (0.418) (0.000)
Age 7.57 17.16 9.41 2.50 6.47 0.92
(0.000) (0.000) (0.000) (0.041) (0.000) (0.452)
Adjusted R2 0.077 0.248 0.184 0.045 0.289 0.277
Root MSE 1.151 0.682 0.808 1.240 0.678 0.937
Observations 3,023 702 2,801 2,047 352 1,376

 

Source: lFLS3.

Month of interview dummy variables are included in the regressions but are not reported. Omitted category for own
schooling is no schooling (0-6 years for public sector workers) and for parental schooling is no schooling. Adjustments are
made due to small cell size for some variables for women: 10+ years for own schooling. Standard errors are robust to
clustering at the community level and heteroskedasticity. Absolute t-statistics are in parentheses. Signiﬁcant at the

1%(‘"), 5%(") and 10%(") indicated. p -values for F -test are in parentheses.

188

Appendix Table 6.6A
Selectivity Corrected Wage Functions:
The Effects of Linear Own Schooling

 

 

 

Panel Respondents, lFLSl
Men Women
Self Public Private Self Public Private
Employment Sector Sector Employment Sector Sector
Years ofschooling 0.101 “"‘ 0.072 '“' 0.092 *""" 0.054 ""' 0.052 0.116 ”“
(7.45) (2.31) (11.96) (3.64) (0.95) (10.33)
Age 0.096 " 0.109 "' 0.050 0.041 0.088 0.097 *
(2.19) (1.77) (1.39) (0.73) (1.16) (1.70)
Age2(x 10") -0135 .. -0107 -0051 -0042 .0077 -0125
(2.40) (1 .37) (1 .07) (0.60) (0.79) (1 .64)
Month of interview
October 0.302 " -0.142 0.154 0.305 "‘ 0.045 0.054
(1 .98) (0.70) (1 .27) (1.73) (0.24) (0.31)
November 0.470 "" -0.262 0.190 0.364 “ -0.204 -0.117
(3.62) (1 .27) (1.63) (2.01) (1 .34) (0.73)
December 0047 -0.077 0.010 0.108 0.046 -0.294 "'
(0.36) (0.40) (0.08) (0.60) (0.34) (1 .80)
January 0059 0.009 0.188 0.196 0.315 -0.157
(0.33) (0.04) (1.13) (0.64) (1 .57) (0.64)
Gross income 0.026 0.250 "‘
(0.32) (2.11)
Selection Coefﬁcient -0.581 "" -0.116 0.090 0.506 -0.423 0.099
(3.34) (0.54) (1.19) (1.48) (1 .43) (0.55)
Constant 3.982 “"' 4.014 “" 4.350 *" 3.744 ""' 5.031 ** 3.213 *"
(4.67) (2.66) (6.44) (2.68) (2.19) (2.98)
F -test
Age 4.43 6.23 5.38 0.73 20.1 1 1.64
(0.013) (0.002) (0.005) (0.485) (0.000) (0.197)
Month of interview 6.47 1.45 1.53 1.75 1.77 2.01
(0.000) (0.219) (0. 194) (0.139) (0.140) (0.094)
Adjusted R :’ 0.086 0.233 0.218 0.038 0.463 0.274
Root MSE 1.145 0.686 0.769 1.207 0.591 0.877
Observations 1,347 416 959 830 210 450

 

Source: IF LSl .

Omitted category for month of interview is August/September. Selection coefﬁcients are calculated based on Base
Speciﬁcation estimates for panel respondents in which dummy variables for own schooling are replaced by years of
schooling. Standard errors are robust to clustering at the community level and heteroskedasticity. Absolute :-

statistics are in parentheses. Signiﬁcant at the 1%("'"""), 5%(*"‘) and 10%("') indicated. p-values for F -test are in

parentheses.

189

Appendix Table 6.68
Selectivity Corrected Wage Functions:
The Effects of Linear Own Schooling

 

 

 

Panel Respondents, iFLSZ
Men Women
Self Public Private Self Public Private
Employment Sector Sector Employment Sector Sector
Years ofschooling 0.109 *“ 0.038 0.088 "* 0.081 *" -0.013 0.115 ""
(1 1.36) (0.94) (12.96) (6.16) (0.30) (9.07)
Age 0.007 -0.017 0.088 " 0.096 "' 0.036 0.051
(0.16) (0.24) (2.21) (1 .72) (0.44) (0.90)
Age2(x10'2) -0022 0.046 -0.116 n -0137 r -0034 -0.058
(0.38) (0.54) (2.22) (1.91) (0.32) (0.76)
Month of interview
October -0.226 * -0.123 0.003 -0.198 -0.355 "”" 0.094
(1.74) (1 .05) (0.03) (1 .44) (2.06) (0.66)
November ~0.211 "‘ -0.124 -0.003 -0.127 -0.457 *" 0.068
(1.69) (0.99) (0.03) (0.94) (2.19) (0.44)
December 0.1 12 0.024 0.144 0.173 -0.223 0.077
(0.82) (0.20) (1 .44) (1 .02) (1 .37) (0.54)
January-April 0.201 0.038 0.249 *" -0.171 -0.053 0.428 “
(0.67) (0.26) (2.13) (0.48) (0.27) (2.06)
Selection Coefﬁcient -0.872 W" -0.201 0.040 -0.646 ** -0.441 "‘ -0.069
(5.58) (0.70) (0.31) (2.16) (1.74) (0.46)
Constant 6.730 *" 7.377 *** 4.571 "* 5.222 *" 7.754 "" 4.602 *"
(7.69) (3.71 ) (6.17) (4.26) (3.46) (4.36)
F -test
Age 2.84 7.48 2.46 4.03 1.80 1.06
(0.060) (0.001) (0.087) (0.019) (0.170) (0.349)
Month of interview 2.99 1.16 1.97 1.61 2.51 1.10
(0.019) (0.331) (0.099) (0.172) (0.045) (0.359)
Adjusted R 2 0.082 0.162 0.185 0.059 0.267 0.194
Root MSE 1.162 0.634 0.831 1.194 0.481 1.029
Observations 1,322 409 943 937 196 524

 

Source: lFLSZ.

Omitted category for month of interview is August/September. Selection coefﬁcients are calculated based on Base
Speciﬁcation estimates for panel respondents in which dummy variables for own schooling are replaced by years of
schooling. Standard errors are robust to clustering at the community level and heteroskedasticity. Absolute :-

statistics are in parentheses. Signiﬁcant at the 1°/o("”""), 5°/o(""") and 10°/o("') indicated. p-values for F -test are in

parentheses.

190

Appendix Table 6.6C
Selectivity Corrected Wage Functions:
The Effects of Linear Own Schooling

 

 

 

Panel Respondents, lFLS3
Men Women
Self Public Private Self Public Private
Employment Sector Sector Employment Sector Sector

Years of schooling 0.097 "““ 0.080 "‘ 0.089 ‘”" 0.073 *" 0.125 " 0.095 ""

(9.38) (1 .90) (1 1.70) (6.36) (2.40) (7.24)
Age 0.054 0.155 "' 0.115 "" 0.057 -0.108 0.120 **

(1.29) (1.90) (2.80) (1.11) (1.09) (2.14)
Age2 (x 102) -0.080 -0.169 .0149 ""' -0.081 0.165 -0.168 "

(1.47) (1.61) (2.78) (1.19) (1.25) (2.24)
Month of interview
August 0264 ‘”" -0.001 0.083 -0.275 "' 0.060 0.018

(2.85) (0.01) (0.76) (1.97) (0.43) (0.13)
September 0158 -0.114 -0.032 -0.318 " -0.083 -0.094

(1 .38) (0.93) (0.31) (2.34) (0.65) (0.62)
October -0.234 " -0.029 0.102 -0.272 " 0.078 0.012

(2.05) (0.23) (1 .00) (1 .92) (0.36) (0.07)
November-January -0.060 0.078 0.069 0.017 0.089 0.320 "‘

(0.40) (0.56) (0.57) (0.10) (0.53) (2.01)
Selection Coefficient -0.560 "’" 0.022 -0.093 -0.370 -0.073 0.561 *"

(3.85) (0.08) (1 . 15) (1 .23) (0.28) (3.23)
Constant 6.514 “1" 4.080 "' 4.717 "" 6.268 "" 8.530 "" 3.248 "*

(8.12) (1 .92) (6.17) (5.36) (3.55) (2.98)
F-test
Age 2.26 8.65 3.92 1.03 3.49 2.84

(0.106) (0.000) (0.021) (0.358) (0.033) (0.060)
Month of interview 2.41 0.78 0.74 2.17 0.65 1.93
(0.049) (0.539) (0.563) (0.072) (0.630) (0.107)

Adjusted R 2 0.067 0.188 0.184 0.051 0.395 0.196
Root MSE 1.169 0.647 0.813 1.209 0.626 0.983
Observations 1 ,442 364 842 1 ,1 24 192 545

 

Source: iFLS3.

Omitted category for month of interview is J une/J uly. Selection coefﬁcients are calculated based on Base

Speciﬁcation estimates for panel respondents in which dununy variables for own schooling are replaced by years of
schooling. Standard errors are robust to clustering at the community level and heteroskedasticity. Absolute :-
statistics are in parentheses. Signiﬁcant at the 1%(*"'*), 5%(‘”') and 10%("‘) indicated. p -values for F -tcst are in

parentheses.

191

Appendix Table 6.7A
Selectivity Corrected Wage Functions:
The Effects of Non-linear Own Schooling

 

 

 

 

Panel Respondents, lFLSl
Men Women
Self Public Private Self Public Private
Employment Sector Sector Employment Sector Sector
Own Schooling
1-3 years 0.092 0.120 0.049 0.172
(0.83) (1 .24) (0.31) (1 .40)
4-6 years 0.296 "" 0.387 ”" 0.355 ” 0.294 ”
(2.73) (4.76) (2.31) (2.23)
7-9 years 0.821 *" 0.251 0.680 "" 0.543 "" 0.291 0.691 ”'"
(5.79) (1 .49) (7.36) (2.96) (1 .10) (4.00)
10-12 years 1.124 *" 0.816 "”" 1.012 "" 0.497 "' 1.142 ‘"" 1.652 ‘”
(5.88) (3.07) (10.24) (1 .94) (2.14) (11.22)
13+ years 1.925 ”" 1.103 "" 1.523 *”
(5.50) (3.71) (8.77)
Age in 1993 (spline)
25-29 years 0.030 0.043 0.073 " 0.064 0.123 0.128 **"‘
(0.64) (0.67) (2.49) (1 .12) (1 .58) (2.97)
30-39 years 0.005 0.050 "“" 0.002 0.014 0.044 ”" -0.012
(0.43) (3.27) (0.28) (0.74) (3.44) (0.75)
4049 years 0022 0.029 "' 0.019 -0.023 0.013 0.002
(1.60) (1 .94) (1 .64) (1.19) (0.61) (0.10)
50-52 years 0057 -0.123 -0.055 0.182 ” 0.015 -0.066
(0.84) (1.35) (0.74) (2.01) (0.16) (0.53)
Month of interview
October 0.285 " -0.117 0.157 0.270 -0.133 0.073
(1 .88) (0.60) (1 .32) (1.50) (0.71) (0.44)
November 0.481 "" -0.202 0.215 0.377 " -0.322 " -0.123
(3.72) (1 .03) (1 .87) (1 .92) (2.23) (0.81)
December -0.053 -0.017 0.034 0.1 18 -0. 149 -0.247
(0.42) (0.09) (0.28) (0.62) (0.97) (1 .59)
January -0.099 0.109 0.210 0.168 0.129 -0.115
(0.56) (0.52) (1 .27) (0.56) (0.64) (0.50)
Gross income 0.037 0.247 “'
(0.45) (2.09)
Selection Coefﬁcient -0.625 *” 0.111 0.087 0.612 -0.101 0.041
(3.61) (0.51) (1.12) (1.52) (0.30) (0.23)
Constant 4.947 "'" 4.772 " 3.390 "* 2.613 2.708 1.606
(3.84) (2.34) (4.06) (1 .47) (0.89) (1.30)
F -test
Own Schooling 12.91 6.47 29.60 4.88 2.47 38.48
(0.000) (0.000) (0.000) (0.001) (0.089) (0.000)
Age 2.19 5.41 4.14 1.50 13.83 2.71
(0.070) (0.000) (0.003) (0.203) (0.000) (0.032)
Month of interview 6.79 1.70 1.63 1.59 2.03 1.81
(0.000) (0.151) (0.168) (0.178) (0.094) (0.129)
Adjusted R 2 0.084 0.260 0.215 0.037 0.466 0.307
Root MSE 1.146 0.674 0.770 1.207 0.589 0.857
Observations 1,347 416 959 830 210 450
Source: lFLSl.

Omitted category for own schooling is no schooling (0-6 years for public sector workers) and for month of interview is
August/September. Adjustments are made due to small cell size for some variables for women: 10+ years for own
schooling. Selection coefﬁcients are calculated based on Base Speciﬁcation estimates for panel respondents. Standard
errors are robust to clustering at the community level and heteroskedasticity. Absolute t-statistics are in parentheses.
Signiﬁcant at the l%("'"), 5°/o(‘”") and 10%(‘) indicated. p -values for F -test are in parentheses.

192

r

 

 

Appendix Table 6.7B
Selectivity Corrected Wage Functions:
The Effects of Non-linear Own Schooling

 

 

 

 

Panel Respondents, IFLSZ
Men Women
Self Public Private Self Public Private
Employment Sector Sector Employment Sector Sector
Own Schooling
1-3 years 0.368 "'" 0.120 0.134 0.075
(3.18) (1.11) (1.05) (0.48)
4-6 years 0.405 "" 0.407 "" 0.235 ” 0.232 "'
(3.81) (4.30) (2.12) (1.69)
7-9 years 0.964 "" 0.010 0.876 "" 0.774 "‘" 0.192 0.463 ”
(7.68) (0.06) (7.81) (4.47) (0.72) (2.24)
10-12 years 1.391 ”" 0.558 " 0.961 "" 1.153 "" 0.501 1.687 ‘"
(9.99) (1 .80) (8.50) (5.29) (1 .21) (1 1.47)
13+ years 2.079 "" 0.796 " 1.332 ‘"
(7.29) (2.06) (9.51)
Age in 1993 (spline)
25-29 years 0.048 -0.001 0.044 0.064 0.049 -0.015
(1.15) (0.02) (1.33) (1.43) (0.71) (0.29)
30-39 years -0.010 0.025 ‘ 0.004 -0.010 0.026 "' 0.019
(0.79) (1 .67) (0.37) (0.60) (1 .93) (1 .22)
40-49 years -0029 " 0.035 "“ -0.004 -0.015 -0.013 0.002
(1 .92) (2.93) (0.25) (0.80) (0.69) (0.08)
50-52 years 0.122 “ -0.134 -0.086 -0.097 0.039 -0.091
(1 .72) (1 .43) (0.99) (1 .10) (0.43) (0.73)
Month of interview
October -0.227 " -0.104 -0.002 -0.176 -0.263 0.090
(1 .79) (0.94) (0.02) (1 .28) (1 .56) (0.63)
November -0.212 ' -0.060 0002 «0.115 -0.320 ‘ 0.112
(1 .72) (0.47) (0.02) (0.87) (1 .71) (0.70)
December 0.101 0.1 1 1 0.146 0.185 -0.146 0.063
(0.75) (1 .00) (1 .45) (1 .09) (0.88) (0.45)
January-April 0.151 0.098 0.257 " -0.164 0.008 0.410 "
(0.51) (0.69) (2.14) (0.45) (0.04) (1 .99)
Selection Coefficient -0.917 ”" 0.074 0.035 -0.702 " -0.168 -0.171
(6.01) (0.29) (0.25) (2.30) (0.73) (1 .24)
Constant 5.449 *" 6.816 ""' 4.947 "‘ 5.191 "" 6.046 '"' 6.240 ‘"
(4.71) (4.12) (5.30) (3.84) (2.49) (4.24)
F -test
Own Schooling 29.49 3.25 33.96 11.39 0.80 44.97
(0.000) (0.023) (0.000) (0.000) (0.453) (0.000)
Age 2.38 5.27 1.27 2.64 1.98 0.71
(0.052) (0.000) (0.282) (0.034) (0.100) (0.587)
Month of interview 2.76 1.82 2.03 1.55 2.09 1.01
(0.028) (0.126) (0.090) (0.186) (0.086) (0.403)
Adjusted R 2 0.088 0.182 0.180 0.061 0.276 0.232
Root MSE 1.158 0.626 0.833 1.192 0.479 1.004
Observations 1 ,322 409 943 937 196 524
Source: IFLSZ.

Omitted category for own schooling is no schooling (0-6 years for public sector workers) and for month of interview is
August/September. Adjustments are made due to small cell size for some variables for women: 10+ years for own
schooling. Selection coefﬁcients are calculated based on Base Speciﬁcation estimates for panel respondents. Standard
errors are robust to clustering at the commrmity level and heteroskedasticity. Absolute t-statistics are in parentheses.
Signiﬁcant at the 1%(“"), 5%(""') and 10%(‘) indicated. p -values for F -test are in parentheses.

193

 

 

 

Appendix Table 6.7C
Selectivity Corrected Wage Functions:
The Effects of Non-linear Own Schooling

 

 

 

 

Panel Respondents, IFLS3
Men Women
Self Public Private Self Public Private
Employment Sector Sector Employment Sector Sector
Own Schooling
1-3 years 0.288 " 0.350 " 0.128 0.291 ‘“"
(2.10) (2.44) (1.17) (2.40)
4-6 years 0.411 "" 0.427 *""' 0.214 ‘”" 0.278 ”
(3.42) (3.34) (2.08) (2.21)
7-9 years 0.765 *""' 0.172 0.864 "”" 0.490 "“ 0.343 0.654 “"
(5.24) (0.67) (6.28) (3.38) (0.88) (2.69)
10-12 years 1.277 "' 0.793 " 1.074 "" 1.113 "* 1.773 "" 1.366 ""'
(7.65) (1 .83) (8.06) (6.06) (4.05) (7.50)
13+ years 2.217 "'""' 1.081 " 1.595 ”"
(8.30) (1.82) (7.78)
Age in 1993 (spline)
25-29 years 0.024 0.113 0.012 0.072 "‘ -0.028 0.085 ‘
(0.61) (1.60) (0.28) (1.77) (0.51) (1.72)
30-39 years 0.008 0.040 " 0.022 " -0.022 0.017 -0.021
(0.55) (2.39) (1.97) (1.59) (1.26) (1.40)
40-49 years -0.040 “ 0.004 -0.015 0.010 0.014 0.028
(2.32) (0.23) (1 .07) (0.53) (0.59) (1 .25)
50-52 years 0.106 0.076 -0.056 -0.131 0.097 -0.384 "
(1 .33) (0.38) (0.84) (1.51) (0.55) (2.61)
Month of interview
August -0.265 "”'“" 0.004 0.071 -0.329 ” 0.126 0.049
(2.88) (0.03) (0.63) (2.36) (0.90) (0.36)
September -0.198 ‘ -0.116 -0.031 -0.380 *" 0.005 -0.075
(1 .74) (0.90) (0.29) (2.77) (0.04) (0.48)
October -0.236 "' -0.024 0.101 -0.362 " 0.002 -0.002
(2.07) (0.19) (0.96) (2.57) (0.01) (0.01)
November-January -0.061 0.122 0.055 -0.011 0.068 0.389 ‘”"
(0.42) (0.83) (0.45) (0.06) (0.46) (2.41)
Selection Coefﬁcient -0.727 ‘" 0.132 -0.168 " -0.754 ‘”" 0.013 0.572 ""'
(4.51) (0.36) (1 .93) (2.53) (0.06) (2.98)
Constant 6.838 “" 4.019 "' 6.452 “" 5.856 "" 7.444 ”" 3.022 "”"
(6.46) (1 .73) (5.18) (4.72) (3.75) (2.13)
F -test
Own Schooling 19.44 1.79 30.40 10.15 10.12 14.61
(0.000) (0.149) (0.000) (0.000) (0.000) (0.000)
Age 1.87 9.19 2.11 2.08 1.26 2.95
(0.1 16) (0.000) (0.079) (0.083) (0.287) (0.021)
Month of interview 2.48 1.09 0.63 3.08 0.34 2.57
(0.043) (0.360) (0.642) (0.016) (0.850) (0.039)
Adjusted R 2 0.072 0.210 0.180 0.049 0.456 0.202
Root MSE 1.165 0.638 0.815 1.210 0.593 0.979
Observations 1 ,442 364 842 1 ,124 192 545
Source: iFLS3.

Omitted category for own schooling is no schooling (0-6 years for public sector workers) and for month of interview is
June/July. Adjustments are made due to small cell size for some variables for women: 10+ years for own schooling.
Selection coefficients are calculated based on Base Speciﬁcation estimates for panel respondents. Standard errors are
robust to clustering at the community level and heteroskedasticity. Absolute t-statistics are in parentheses. Signiﬁcant at
the 1%(""""), 5%(“) and 10%(‘) indicated. p -values for F -test are in parentheses.

194

9

 

 

Appendix Table 6.8A
Selectivity Corrected Wage Functions:
The Effects of Non-linear Own Schooling and Parental Schoollng
Panel Respondents, [F L81

 

 

 

 

Menﬁ Women
Self Public Private Self Public Private
Employment Sector Sector Employment Sector Sector
Own Schooling
1-3 years 0.019 0.137 0.020 0.164
(0.17) (1.34) (0.12) (1.28)
4-6 years 0.196 " 0.385 ”" 0.247 0.239 "
(1.79) (4.47) (1.58) (1.72)
7-9 years 0.685 ‘” 0.287 0.670 ”" 0.360 “ 0.358 0.585 ”’
(4.79) (1 .49) (6.67) (1.86) (1.17) (2.98)
10.12 years 0.979 ‘” 0.902 ” 0.969 “‘ 0.258 1.188 1.440 ""
(5.17) (2.49) (8.57) (1 .01) (1.57) (7.22)
13+ years 1.745 ”’ 1.196 ”" 1.452 ‘”
(5.02) (2.86) (8.16)
Father's Schooling
Some Elementary 0.261 " -0.001 -0.053 0.082 0.038 0.100
(2.61) (0.01) (0.64) (0.57) (0.23) (0.80)
Completed Elementary 0.163 0.009 -0.028 0.152 -0.206 0.386 ”"
(1.35) (0.08) (0.28) (1 .04) (1 .26) (2.74)
Secondary/Tertiary 0.220 -0.105 0.028 0.294 -0.061 0.251
(1.06) (0.85) (0.23) (1.22) (0.36) (1.09)
Missing 0.008 -0.121 —0.1 15 -0.l98 -0.134 0.240 ‘
(0.07) (0.65) (1.02) (1.41) (0.68) (1.78)
Mother's Schooling
Some Elementary 0.148 0.063 0.012 0.183 0.074 -0.013
(1.34) (0.51) (0.13) (1.22) (0.72) (0.10)
Completed Elementary/ 0.454 W” 0.164 ‘ 0.147 0.291 ‘ 0.186 0.113
Secondary/Tertiary (3.61) (1.76) (1.42) (1.71) (1 .37) (0.71)
Missing 0.151 0.176 0.127 0.068 0.123 -0.061
(1.53) (1.24) (1.28) (0.54) (0.71) (0.48)
Age in 1993 (spline)
25-29 years 0.031 0.048 0.073 “ 0.052 0.120 0.127 '”
(0.66) (0.75) (2.47) (0.89) (1.13) (2.83)
3039 years 0.010 0.051 ”‘ 0.002 0.023 0.050 "" -0.009
(0.77) (3.07) (0.28) (1.08) (3.71) (0.54)
40-49 years -0.019 0.032 ” 0.021 " -0.023 0.009 0.006
(1.43) (2.11) (1.76) (1.19) (0.41) (0.28)
50-52 years 0041 -0.130 -0.059 0.177 “ 0.018 -0.066
(0.58) (1 .46) (0.79) (1 .99) (0.16) (0.54)
Gross income 0.051 0.235 ""
(0.63) (1.99)
Selection Coefﬁcient -0.730 "‘ 0.190 0.101 0.595 -0.057 -0.036
(4.03) (0.61) (1.14) (1.28) (0.12) (0.18)
Constant 4.824 ’" 4.335 ‘ 3.372 ‘” 2.890 2.634 1.662
(3.70) (1.91) (4.03) (1.57) (0.61) (1.27)
F -test
Own Schooling 10.29 3.88 22.74 2.08 1.23 15.86
(0.000) (0.010) (0.000) (0.084) (0.296) (0.000)
Father's Schooling 2.16 0.57 0.47 1.44 0.80 2.30
(0.074) (0.688) (0.755) (0.221) (0.525) (0.060)
Mother's Schooling 4.52 1.46 1.12 1.19 0.69 0.30
(0.004) (0.228) (0.340) (0.312) (0.560) (0.825)
Parental Schooling 5.04 0.84 0.99 1.57 0.59 1.78
(0.000) (0.554) (0.440) (0.143) (0.759) (0.094)
Age 1.63 4.48 4.10 1.44 12.78 2.83
(0.166) (0.002) (0.003) (0.221) (0.000) (0.026)
Adjusted R -‘ 0.104 0.256 0.214 0.047 0.459 0.313
Root MSE 1.134 0.676 0.770 1.201 0.593 0.853
Observations 1,347 416 959 830 210 450
Source: IF LSl.

Month of interview dununy variables are included in the regressions but are not reported. Omitted category for own schooling
is no schooling (0-6 years for public sector workers) and for parental schooling is no schooling. Adjustments are made due to
small cell size for some variables for women: 10+ years for own schooling. Selection coefﬁcients are calculated based on Base
Speciﬁcation estimates for panel respondents. Standard errors are robust to clustering at the community level and
heteroskedasticity. Absolute («statistics are in parentheses. Signiﬁcant at the 1°/o(‘"), 5°/o(”) and 10%(‘) indicated. p -
values for F -test are in parentheses.

195

Appendix Table 6.8B
Selectivity Corrected Wage Functions:
The Effects of Non-linear Own Schooling and Parental Schooling

 

 

 

 

Panel Respondents, lFLSZ
Me_n Women
Self Public Private Self Public Private
Employment Sector Sector Employment Sector Sector
Own Schooling
1-3 years 0.303 ”‘ 0.112 0.116 0.069
(2.61) (1 .04) (0.87) (0.46)
4-6 years 0.309 '” 0.353 "‘ 0.158 0.111
(2.82) (3.72) (1 .26) (0.80)
7-9 years 0.846 '” 0.214 0.771 "" 0.628 “" 0.405 0.226
(6.53) (1 .27) (6.74) (3.47) (1 .12) (1 .00)
10-12 years 1.289 ‘” 1.028 “" 0.776 "" 0.921 "" 1.237 " 1.332 "‘
(9.30) (3.15) (6.23) (3.89) (1.79) (7.29)
13+ years 2.038 ”‘ 1.385 ”" 1.097 “"
(7.00) (3.40) (7.03)
Father's Schooling
Some Elementary 0.229 " 0.008 0.004 0.093 0.222 0.058
(2.19) (0.07) (0.04) (0.69) (1 .26) (0.31)
Completed Elementary 0.107 0.063 0.041 0.317 "‘ 0.149 0.082
(0.84) (0.66) (0.43) (2.26) (0.94) (0.48)
Secondary/Tertiary 0.458 " -0.164 0.184 0.430 ‘ 0.035 0.433 "
(2.44) (1 .44) (1 .36) (1.96) (0.22) (2.22)
Missing «0.003 -0.508 " 0.160 0.073 0.083 0.193
(0.02) (2.22) (1 .13) (0.45) (0.47) (1 .16)
Mother's Schooling
Some Elementary 0.088 0.121 0.184 " 0.150 0.178 -0.002
(0.80) (1.18) (1.96) (0.85) (1.49) (0.01)
Completed Elementary/ 0.328 “ 0.240 ‘”‘ 0.308 "‘“ -0.050 0.166 0.266 ”
Secondary/Tertiary (2.50) (2.55) (3.40) (0.28) (1.19) (2.12)
Missing 0.171 0.349 0.080 -0.069 0.008 -0.041
(1 .19) (1.65) (0.61) (0.37) (0.03) (0.27)
Age in 1993 (spline)
25-29 years 0.059 0.021 0.045 0.064 0.147 -0.021
(1.44) (0.44) (1.37) (1.39) (1.58) (0.40)
30-39 years 0012 0.044 ‘" 0.001 -0.004 0.033 ” 0.020
(0.89) (2.78) (0.13) (0.21) (2.22) (1 .25)
40-49 years 0024 0.034 "' -0.005 -0.015 -0.026 0.005
(1.58) (2.97) (0.31) (0.75) (1.16) (0.23)
50-52 years 0.126 "' -0.163 " -0.077 -0.091 0.079 -0.084
(1 .76) (1 .72) (0.91) (1 .00) (0.83) (0.68)
Selection Coefﬁcient -1.134 ‘” 0.492 ‘ 0.200 -0.638 0.289 -0.007
(7.16) (1 .74) (1.44) (1 .62) (0.78) (0.04)
Constant 5.201 ‘” 5.114 ”‘ 4.646 "" 5.006 "‘ 1.671 6.094 “’
(4.64) (2.89) (4.91) (3.57) (0.46) (4.07)
F -test
Own Schooling 25.93 5.46 18.06 5.74 2.02 19.84
(0.000) (0.001) (0.000) (0.000) (0.137) (0.000)
Father's Schooling 2.54 3.17 0.84 1.58 0.71 1.99
(0.040) (0.015) (0.500) (0.180) (0.587) (0.097)
Mother's Schooling 2.12 2.62 4.02 0.58 1.16 1.92
(0.097) (0.051) (0.008) (0.630) (0.327) (0.128)
Parental Schooling 3.54 2.46 3.46 1.42 l .14 1 .87
(0.001) (0.019) (0.001) (0.196) (0.345) (0.076)
Age 1.90 8.30 1.22 1.96 2.62 0.70
(0.1 10) (0.000) (0.303) (0.101) (0.038) (0.591)
Adjusted R’ 0.104 0.207 0.196 0.065 0.268 0.238
Root MSE 1.148 0.617 0.825 1.190 0.481 1.000
Observations 1 ,322 409 943 937 196 524
Source: iFLSZ.

Month of interview dummy variables are included in the regressions but are not reported. Omitted category for own schooling
is no schooling (0-6 years for public sector workers) and for parental schooling is no schooling. Adjustments are made due to
small cell size for some variables for women: 10+ years for own schooling. Selection coefﬁcients are calculated based on Base

Speciﬁcation estimates for panel respondents. Standard errors are robust to clustering at the community level and

heteroskedasticity. Absolute r-statistics are in parentheses. Signiﬁcant at the 1%(""'), 5%(”) and 10%("') indicated. p-
values for F -test are in parentheses.

196

Appendix Table 6.8C
Selectivity Corrected Wage Functions:
The Effects of Non-linear Own Schooling and Parental Schooling

 

 

 

 

Panel Respondents, lFLS3
Mgr Wom_en
Self Public Private Self Public Private
Employment Sector Sector Employment Sector Sector
Own Schoollng
1-3 years 0228 " 0.345 ” 0.105 0.305 ”
(1 .66) (2.40) (0.94) (2.34)
4-6 years 0.324 ”‘ 0.399 ”" 0.167 0.252 ‘
(2.73) (3.12) (1.61) (1.89)
7-9 years 0.696 ”‘ 0.199 0.810 ”" 0.428 “" 0.277 0.490 "
(4.74) (0.83) (5.97) (2.80) (0.69) (1.86)
10-12 years 1.281 ”" 0.834 ” 0.968 ”‘ 1.084 ‘" 1.544 “" 1.140 ”‘
(7.58) (2.05) (7.22) (5.34) (2.83) (5.19)
13+ years 2.302 ”‘ 1.153 " 1.406 "‘“
(8.76) (2.06) (6.80)
Father's Schooling
Some Elementary 0.027 -0.035 -0.081 0.062 -0.101 -0.133
(0.27) (0.30) (0.80) (0.42) (0.54) (0.92)
Completed Elementary 0.261 ”" -0.076 0.063 0.034 -0.006 0.150
(2.75) (0.75) (0.65) (0.30) (0.03) (1 . 10)
Secondary/Tertiary 0.526 ‘“ -0.109 0.164 -0.044 0.081 0.316
(3.19) (0.97) (1.45) (0.24) (0.42) (1 .54)
Missing -0.051 0.050 -0.128 0.003 0.123 0.111
(0.46) (0.38) (1 .36) (0.03) (0.47) (0.80)
Mother's Schoollng
Some Elementary 0.080 0.179 " 0.017 -0.061 0.052 0.048
(0.79) (1 .69) (0.16) (0.43) (0.44) (0.33)
Completed Elementary/ 0.250 ” 0.154 0.147 ‘ 0.223 ‘ 0.076 0.107
Secondary/1” ertiary (2.29) (1 .46) (1 .65) (1 .70) (0.69) (0.71)
Missing 0.463 "" -0.318 ‘ 0.208 ‘”" 0.172 -0.103 -0.275 "
(3.52) (1.76) (1.98) (1.15) (0.34) (1.89)
Age in 1993 (spline)
25-29 years 0.039 0.1 1 1 0.008 0.069 -0.059 0.078
(1 .00) (1 .62) (0.18) (1.65) (0.86) (1 .54)
30-39 years 0.000 0.043 " 0.021 ‘ -0.021 0.019 -0.016
(0.00) (2.54) (1 .93) (1 .48) (1.17) (1 .07)
40-49 years -0.036 ” 0.002 -0.019 0.010 0.023 0.029
(2.15) (0.14) (1.30) (0.50) (0.84) (1.31)
50-52 years 0.118 0.090 -0.039 -0.132 0.057 -0.395 ""
(1 .50) (0.46) (0.57) (1 .49) (0.33) (2.67)
Selection Coefﬁcient -1.i61 ‘” 0.192 -0.053 -0.908 *” -0.094 0.585 ‘“”
(6.56) (0.57) (0.69) (2.64) (0.30) (2.73)
Constant 6.625 “‘ 3.917 ‘ 6.446 ‘" 6.112 “‘ 8.602 "‘ 3.161 '"'
(6.21) (1 .76) (5.23) (4.89) (3.29) (2.13)
F-test
Own Schooling 20.54 1.85 21.89 7.33 4.60 6.95
(0.000) (0.139) (0.000) (0.000) (0.012) (0.000)
Father's Schooling 3.64 0.34 1.71 0.10 0.53 1.31
(0.006) (0.851) (0.146) (0.981) (0.716) (0.267)
Mother's Schooling 4.99 2.87 1.91 1.48 0.26 1.54
(0.002) (0.037) (0.128) (0.219) (0.855) (0.205)
Parental Schooling 6.04 1.33 2.48 0.80 0.55 1.63
(0.000) (0.238) (0.017) (0.588) (0.796) (0.128)
Age 2.01 8.52 1.87 1.83 1.32 2.64
(0.092) (0.000) (0. 1 14) (0.123) (0.266) (0.034)
Adjusted R 2 0.092 0.219 0.186 0.048 0.442 0.209
Root MSE 1.153 0.635 0.812 1.210 0.601 0.975
Observations l ,442 364 842 l ,1 24 192 545
Source: iFLS3.

Month of interview dununy variables are included in the regressions but are not reported. Omitted category for own schooling
is no schooling (0-6 years for public sector workers) and for parental schooling is no schooling. Adj ustments are made due to
small cell size for some variables for women: 10+ years for own schooling. Selection coefﬁcients are calculated based on Base
Speciﬁcation estimates for panel respondents. Standard errors are robust to clustering at the community level and
heteroskedasticity. Absolute r-statistics are in parentheses. Signiﬁcant at the l%("‘"), 5%(") and 10%(‘) indicated. p-
values for F -test are in parentheses.

197

Table 7.1A
Distribution of Regional Migration by Years of Schooling

 

 

 

 

 

 

lFLSl
Region of Current Residency
Region of Birth Sumatera Java Other Islands
Rural Urban Rural Urban Rural Urban
0 year (N=1,889)
Surnatera - Rural 78.9 19.5 0.0 1.6 0.0 0.0
Surnatera - Urban 8.7 78.3 0.0 8.7 4.3 0.0
Java - Rural 8.8 0.9 71.0 17.9 0.9 0.5
Java - Urban 0.6 1.2 16.6 80.4 0.6 0.6
Other Islands - Rural 0.0 0.0 0.2 0.7 84.8 14.3
Other Islands - Urban 0.0 0.0 0.0 0.0 39.5 60.5
1-5 years (N =2,806)
Surnatera - Rural 75.6 23.6 0.0 0.9 0.0 0.0
Surnatera - Urban 21.7 76.7 0.8 0.8 0.0 0.0
Java - Rural 9.7 1.3 62.4 24.9 1.3 0.4
Java - Urban 1.2 1.5 9.5 85.5 1.2 0.9
Other Islands - Rural 0.2 0.0 0.0 1.1 75.7 23.0
Other Islands - Urban 0.0 0.0 0.0 3.1 36.7 60.2
6 years (N=1,921)
Surnatera - Rural 63.9 31.9 0.4 3.8 0.0 0.0
Surnatera - Urban 22.7 71.6 0.0 5.7 0.0 0.0
Java - Rural 4.8 1.3 58.9 32.6 1.2 1.1
Java - Urban 0.0 0.6 8.9 88.8 0.6 1.2
Other Islands - Rural 0.0 0.0 0.0 2.7 66.5 30.8
Other Islands - Urban 0.0 0.0 0.0 4.5 24.7 70.8
7+ years (N =2,699)
Sumatera - Rural 45.7 42.5 0.2 11.3 0.2 0.0
Surnatera - Urban 10.2 68.2 0.0 21.2 0.4 0.0
Java - Rural 3.1 2.3 33.6 57.5 2.4 1.2
Java - Urban 0.8 1.5 3.1 91.2 0.8 2.7
Other Islands - Rural 0.0 0.7 0.0 2.6 60.7 36.0
Other Islands - Urban 0.5 0.5 0.0 14.1 24.5 60.4
All Individuals (N =9,3 1 5)
Surnatera - Rural 63.3 31.4 0.2 5.1 0.1 0.0
Surnatera - Urban 15.0 71.3 0.2 13.1 0.4 0.0
Java - Rural 7.2 1.4 58.5 30.9 1.4 0.7
Java - Urban 0.7 1.3 7.0 88.4 0.8 1.8
Other Islands - Rural 0.1 0.1 0.1 1.5 74.2 24.0
Other Islands - Urban 0.2 0.2 0.0 7.1 30.2 62.2

 

Source: IFLS 1.

Estimates are in percentage relative to region of birth.

198

Table 7.1B
Distribution of Regional Migration by Years of Schooling

 

 

 

 

IFLSZ
Region of Current Residency
Region of Birth Surnatera Java Other Islands
Rural Urban Rural Urban Rural Urban
0 year (N =2,152)
Surnatera - Rural 78.4 19.4 0.7 1.5 0.0 0.0
Surnatera - Urban 15.2 72.7 0.0 9.1 3.0 0.0
Java - Rural 8.4 0.9 72.5 17.2 0.9 0.2
Java - Urban 1.7 1.7 15.7 80.9 0.0 0.0
Other Islands - Rural 0.2 0.0 0.2 0.4 84.6 14.7
Other Islands - Urban 0.0 0.0 0.0 0.0 51.0 49.0
1-5 years (N =2,855)
Surnatera - Rural 78.2 20.2 0.2 1.3 0.0 0.0
Sumatera - Urban 18.6 79.6 0.9 0.9 0.0 0.0
Java - Rural 7.6 1.5 63.5 26.1 1.0 0.4
Java - Urban 0.9 1.3 10.9 85.3 0.6 0.9
Other Islands - Rural 0.2 0.0 0.0 1.0 75.5 23.2
Other Islands - Urban 0.0 0.0 0.0 2.3 37.5 60.2
6 years (N =2,450)
Surnatera - Rural 69.6 28.0 0.0 2.5 0.0 0.0
Surnatera - Urban 21.2 76.8 0.0 2.0 0.0 0.0
Java - Rural 3.6 1.0 63.1 30.8 1.2 0.3
Java - Urban 0.5 0.7 11.8 85.3 1.0 0.7
Other Islands - Rural 0.0 0.0 0.0 1.4 70.9 27.7
Other islands - Urban 0.0 1.0 0.0 1.9 30.5 66.7
7+ years (N =3,994)
Surnatera - Rural 54.1 37.4 0.4 7.9 0.2 0.0
Surnatera - Urban 10.0 72.4 1.1 15.7 0.3 0.5
Java - Rural 2.4 1.4 37.1 56.7 1.4 0.9
Java - Urban 0.6 1.5 7.8 88.1 0.7 1.3
Other islands - Rural 0.2 0.7 0.0 1.9 66.4 30.8
Other Islands - Urban 0.4 0.4 0.0 8.8 24.4 66.1
All Individuals (N =1 1,451)
Surnatera - Rural 67.2 28.3 0.3 4.1 0.1 0.0
Surnatera - Urban 13.7 74.4 0.8 10.4 0.3 0.3
Java - Rural 5.5 1.2 59.1 32.6 1.1 0.4
Java - Urban 0.7 1.3 9.8 86.5 0.7 1.0
Other Islands - Rural 0.2 0.2 0.1 1.1 75.3 23.2
Other Islands - Urban 0.2 0.3 0.0 4.9 32.4 62.3

 

Source: IFLS2.

Estimates are in percentage relative to region of birth.

199

Table 7.1C
Distribution of Regional Migration by Years of Schooling

 

 

 

 

IFLS3
Mion of Current Residency
Region of Birth Sumatera Java Other Islands
Rural Urban Rural Urban Rural Urban
0 year (N=1,838)
Surnatera - Rural 80.3 18.8 0.9 0.0 0.0 0.0
Surnatera - Urban 28.6 61.9 0.0 9.5 0.0 0.0
Java - Rural 6.1 0.8 72.7 19.7 0.4 0.2
Java - Urban 2.7 0.7 42.6 53.4 0.7 0.0
Other Islands - Rural 0.2 0.0 0.2 0.7 75.6 23.4
Other Islands - Urban 0.0 0.0 0.0 2.9 22.9 74.3
1-5 years (N =3,262)
Surnatera - Rural 78.5 20.4 0.2 0.8 0.2 0.0
Sumatera - Urban 33.0 64.3 0.0 2.6 0.0 0.0
Java - Rural 8.5 1.5 60.5 27.7 1.1 0.6
Java — Urban 1.5 0.5 24.9 71.2 0.8 1.0
Other Islands - Rural 0.2 0.0 0.0 1.0 73.7 25.1
Other Islands - Urban 0.0 0.0 0.0 2.1 20.6 77.3
6 years (N =3,282)
Surnatera - Rural 71.2 26.7 0.5 1.6 0.0 0.0
Surnatera - Urban 27.4 69.0 0.0 3.5 0.0 0.0
Java - Rural 3.7 0.6 60.4 33.5 1.2 0.6
Java - Urban 0.9 1.1 24.5 71.8 0.9 0.9
Other Islands - Rural 0.2 0.0 0.0 0.9 65.1 33.8
Other Islands - Urban 0.0 0.0 0.0 1.1 17.2 81.6
7+ years (N =6,448)
Surnatera - Rural 51.9 38.6 0.4 8.8 0.0 0.2
Surnatera - Urban 15.4 66.3 1.6 16.4 0.0 0.3
Java - Rural 2.0 1.5 38.4 55.7 1.2 1.2
Java - Urban 0.5 1.3 10.3 86.4 0.4 1.1
Other Islands - Rural 0.0 0.0 0.4 1.7 58.6 39.4
Other Islands - Urban 0.3 0.3 0.5 9.4 18.9 70.6
All Individuals (N =14,830)
Surnatera - Rural 64.8 30.0 0.4 4.6 0.1 0.1
Sumatera - Urban 19.8 66.3 1.1 12.6 0.0 0.2
Java - Rural 4.7 1.1 55.3 37.0 1.1 0.7
Java - Urban 0.8 1.1 16.3 80.2 0.5 1.0
Other Islands - Rural 0.1 0.0 0.2 1.1 67.7 30.9
Other Islands - Urban 0.2 0.2 0.3 6.7 19.2 73.5

 

Source: IFLS3.

Estimates are in percentage relative to region of birth.

200

Table 7.2
Own Schooling Attainment by Region of Birth,
Region of Current Residence and Gender

 

 

 

. Region of
Region of Birth Current Residence
Rural Urban Rural Urban
IFLS]
Men Mean 5.4 8.1 4.7 7.9
Standard Error (0.15) (0.19) (0.17) (0.19)
Observations 2,993 1,148 2,178 1,963
Women Mean 3.7 6.4 3.1 6.0
Standard Error (0.14) (0.19) (0.15) (0.19)
Observations 3,731 1,470 2,751 2,450
All Individuals Mean 4.5 7.2 3.8 6.8
Standard Error (0.14) (0.17) (0.15) (0.18)
Observations 6,724 2,618 4,929 4,413
IFLSZ
Men Mean 6.1 8.9 5.4 8.5
Standard Error (0.15) (0.17) (0.16) (0.17)
Observations 3,666 1,464 2,717 2,413
Women Mean 4.3 7.3 3.7 6.8
Standard Error (0.14) (0.17) (0.15) (0.17)
Observations 4,503 1,850 3,376 2,977
All Individuals Mean 5.1 8.0 4.5 7.6
Standard Error (0.13) (0.15) (0.14) (0.16)
Observations 8,169 3,314 6,093 5,390
IFLS3
Men Mean 6.9 9.9 6.2 9.4
Standard Error (0.13) (0.16) (0.14) (0.16)
Observations 5,087 2,052 3,599 3,540
Women Mean 5.3 8.6 4.7 7.9
Standard Error (0.13) (0.15) (0.15) (0.16)
Observations 5,474 2,221 3,895 3,800
All Individuals Mean 6.1 9.2 5.4 8.6
Standard Error (0.12) (0.14) (0.14) (0.15)
Observations 10,561 4,273 7,494 7,340

 

Source: lFLSl, IFLSZ and IFLS3.
Means are in years. Standard errors are robust to clustering at the community level.

201

 

 

Table 7.3
Hourly Wage within Sector of Employment by Region of Birth and Gender

 

 

 

 

 

 

Self Employment Public Sector Private Sector
Rural Urban Difference Rural Urban Difference Rural Urban Difference

lFLSl

Men Mean 669 1,198 529 ‘" 1,789 1,903 114 693 1,137 444 “‘
Standard Error (39) (100) (103) (102) (140) (157) (41) (77) (77)
Median 330 577 -247 1,424 1,320 104 451 726 -274
Observations 1.659 386 343 219 874 475

Women Mean 621 1,119 499 "" 1,737 1,990 253 380 749 369 ‘"
Standard Error (45) (101) (110) (188) (210) (229) (33) (81) (83)
Median 275 495 -220 1,376 1,650 -274 231 385 -154 .
Observations 919 323 140 107 417 186 «1

A11 Individuals Mean 652 1,162 510 *” 1,774 1,932 158 592 1,028 436 “‘
Standard Error (32) (77) (81) (108) (132) (138) (34) (67) (66)
Median 299 541 -242 1 .419 1,443 ~25 361 650 -289
Observations 2,578 709 483 326 1 ,291 661

iFLSZ

Men Mean 1,196 1,948 752 ‘" 2,395 2,387 -8 1,146 1,736 590 ""
Standard Error (54) (137) (145) (118) (133) (173) (47) (84) (90)
Median 594 990 -396 1,938 1,976 -38 770 1,155 -385
Observations 1 ,764 435 370 236 1 ,2 1 8 646

Women Mean 1,033 1,719 686 “" 2,260 2,544 284 731 1,243 511 "‘
Standard Error (59) (179) (185) (152) (148) (194) (54) (96) (106)
Median 462 742 -280 2,045 2,309 -265 3 89 770 -38 1
Observations 1 ,109 347 164 135 621 337

All individuals Mean 1,133 1,847 713 ”" 2,353 2,444 91 1,006 1,567 561 “‘
Standard Error (42) (121) (125) (105) (105) (134) (41) (72) (76)
Median 545 855 -3 10 1 .997 2,075 -78 642 1 ,069 -428
Observations 2,873 782 534 371 1,839 983

IFLS3

Men Mean 2,926 4,141 1,215 "" 5,085 5,325 240 2,191 3,062 870 ”"
Standard Error (119) (241) (262) (195) (376) (421) (76) (140) (151)
Median 1 .443 2,100 -656 4,330 3,923 407 1,443 1,925 -481
Observations 2,458 563 467 234 1,766 1,033

Women Mean 2,169 3,141 972 ”" 5,138 5,140 2 1,257 2,449 1,192 ""
Standard Error (119) (278) (294) (303) (283) (413) (69) (182) (191)
Median 952 1,359 -406 4,491 4,505 -14 770 1,443 -674
Observations 1,552 491 218 132 861 513

All individuals Mean 2,633 3,675 1,042 ”* 5,102 5,258 156 1,885 2,858 973 ‘"
Standard Error (94) (195) (206) (174) (276) (318) (62) (130) (136)
Median 1,237 1,704 -467 4,380 4,266 1 15 1,237 1,767 -530
Observations 4,010 1,054 685 366 2,627 1,546

 

Source: lFLSl, iFLSZ and IFLS3.
Means are in current (nominal) Rupiahs. Standard errors are robust to clustering at the community level. Signiﬁcant at the
i%("”), 5%(") and 10%(‘) indicated.

202

Table 7.4
Hourly Wage within Sector of Employment by Region of Current Residence and Gender

 

 

 

Self Employment Public Sector Private Sector
Rural Urban Difference Rural Urban Difference Rural Urban Difference

lFLSl

Men Mean 578 1,207 629 “" 1,588 1,943 354 ” 473 1,075 602 ”“
Standard Error (39) (80) (89) (97) (123) (157) (29) (64) (70)
Median 278 594 -316 1,296 1,441 -145 362 693 -331
Observations 1 .424 62 1 1 73 389 506 843

Women Mean 535 1,018 483 "‘ 1,735 1,890 155 288 675 387 ‘"
Standard Error (53) (66) (85) (164) (218) (273) (24) (64) (68)
Median 220 493 -273 1,414 1,506 -92 220 330 -1 10
Observations 688 554 70 177 283 320

All individuals Mean 564 1,118 554 *” 1,631 1,926 296 * 407 965 559 "“
Standard Error (35) (56) (66) (85) (135) (160) (23) (58) (62)
Median 264 541 -277 1,315 1,449 -134 289 601 -313
Observations 2,1 12 1,175 243 566 789 1,163

IFLSZ

Men Mean 1,097 1,894 798 "" 2,426 2,372 -55 941 1,624 683 "“
Standard Error (58) (104) (119) (136) (119) (181) (44) (72) (84)
Median 544 1 ,026 -482 2,084 1 ,925 159 660 1 ,069 -409
Observations 1 ,51 5 684 224 382 745 1 ,1 i 9

Women Mean 857 1,679 822 "" 2,328 2,413 85 556 1,171 615 ""
Standard Error (52) (130) (140) (160) (152) (220) (44) (80) (91)
Median 412 693 -2 80 2,071 2,199 -129 346 722 -375
Observations 854 602 87 2 12 404 554

All individuals Mean 1,010 1,793 783 “" 2,399 2,386 -12 805 1,474 668 ""‘
Standard Error (43) (93) (102) (110) (111) (156) (37) (62) (72)
Median 495 855 -3 60 2,079 1 .997 81 541 962 -421
Observations 2,369 1,286 3 1 1 594 1,149 1,673

IFLS3

Men Mean 2,730 3,930 1,200 "" 4,984 5,265 281 1,989 2,851 862 “"
Standard Error (128) (195) (234) (243) (250) (350) (87) (112) (141)
Median 1,375 1,980 -605 4,505 4,141 363 1,320 1,833 -513
Observations l .958 1 ,063 250 45 1 1,099 1 ,700

Women Mean 1,963 2,934 970 "“"" 5,562 4,929 -633 1,119 2,091 972 *“
Standard Error (136) (199) (241) (501) (207) (540) (82) (131) (154)
Median 828 1,237 -409 4,648 4,401 247 693 1,203 -510
Observations 1,1 1 8 925 1 16 234 550 824

All individuals Mean 2,451 3,466 1,015 "'"‘ 5,167 5,150 -17 1,699 2,603 904 “"
Standard Error (103) (162) (192) (245) (194) (313) (72) (102) (124)
Median 1,150 1,650 -499 4,593 4,254 338 1,155 1,650 -495
Observations 3 .076 1 .988 366 685 1 ,649 2,524

 

Source: lFLSl , IFLSZ and iFLS3.

Means are in current (nominal) Rupiahs. Standard errors are robust to clustering at the community level. Signi ﬁcant at the
1%("'"), 5°/o(") and 10%(‘) indicated.

203

.885..an e. 2e asi 5.. use: a .38? 8&2 e5 :.:.sm ..::s. 2:. a .8586 8855.3 5 2e 85:53 e333... 5333.852 e5 .2....
$5.888 05 S autumn—u 9 .958 2a Echo vice—am .noﬁﬁzmo 5.2359on 8.5 no v9.3 38.3.8 95 aeosEoOu 5.82% 9:338 :30 3.: E3» +2 ”cos—o? .8: moist-a, 9:3. 8.: on: :3 :55
2 on: own:— Ba Eco—Emis‘ Anus—53 389. 9.3:: 3.. who» 0.8 3:038 0: ﬂ a.a—09.3. 6.3 3.. bemoan 32:5 60:80.. 8: 2a 59 823052 05 E coca—2: 2a 833...; .355: 3333:. .3 5:02

 

 

204

 

 

 

:3": uSeem
8e as New. $2 «en «a: 223830
48... :3 32 e23 ”med 2: mm: .8:
:.3 83 2 _.o 83 23 a _ .e e m 33.3
83.8 88.9 29.8 38.8 88.: 82.8
33 «ed. 26 .3 m; 2.: em<
88.8 38.8 88.8 88.8 88.8 88.:
an: as w; a.a 8.“ :.: méeeﬁm 56
83.8 22.8 38.8 3%.: 35.8 32.: 28.82.2235
2.... e? a: 3.. 9.. 8d ewe. e5 «5.88m :30
:34
5:8 5.: an: as: so: 5.: $2: 89: 5.8 an: an: 3.:
E3. ... 23 ”SN- ... e. E : 23 . Ed on... ... as; 83. ... 25 55 ... £3 5928
at: 5.8 28.: 5.: an: an: 22: 3a.: 2.2: 5.: E: 5.:
23 m5? zed .. SS. ..: an..- : 23 :3 e S 33. Sod . :.:. ... as... 28680 5:83
3e: 2: a: 2 ..: sen:
. we; : $3 :3. «86 2:85 new
:.:: 5.: 83: 5.8 as: an: 3.8 :..: $3: an: 8.: 32:
2o? ego. 35¢. 155 Ned . 82.. need : .23. as? .86. Bed. 485. f: are?
5.8 5.: 33: and: Se: EN: :3: :.3: 8:: an: 25.8 5.8
23 $3 ~85 ego .2.? . :.:. Sod : SE $2: ES. 83. ~86 ew<
8”: Ex: :.3: E: 32: 8...:
mmmdt nee ON»; $2.0. a: wood ommdt one EON Emu» +M_
:3: 33: an: and: an: an: as: 83: 5.: an: as: 32.:
Nocd one See.— hw_._ 025 Need as #006 Sth a: 9:.— thd: an Nhod hmndt a: mom; memo» N70—
809 5.: 8.: an: 3:: so: 32: and: an: an:
52.0 one cameo Omod e: 206 as med- see and weed 026 N261 one god Emu» $4.
398 as: an: as: 5.8 A8,: Go: an:
20.? an mde 2N6. one wovd ”mod. a: mend thd. a: an6 ﬁne» or?
an. : a: .8 5.: a». : an: AR. : 8:: 3.8
$3 83:. : Bed. _. News 83. :3 85. 23. e8» 3
a.a—oesom :30
93.5 92...: 33.5 Inf: n21: 33.5
... Eon ease—=80 ... 58 38530 ... Een ...».uEeeu ... Eon ...suEeeu ... ES. .5355 ... Ben ...esEeeu
330m 82::— cSuom v.3...— EoEMeEEm 20m .895 89;: .83 33.:— uathsmEm :5
5:33 :02
5...:

..:-.5 ... Eon 5.3 1899.3:— un=eeaom HBO anus—1.52 ue 825m 2:.
uncansm one? 3.99.30 agate—om
(mg. 03:.

 

assessed 5 as no: a: 82:. a .3865 2.3.2 as raga 5:3... 2: a 28:85 88558 5 2e 8:333 2289‘ £83323»; as .32
gen—5:3 05 3 wee—89.0 8 338 8a Echo :5..de .8385» 5:35“.on can so v9.3 noun—3.8 2a $56508 cacao—om $5.858 Eso 3.. so» +2 “EEO? 8.. 933.5, 089. no: on.» :8 =3.»
8 one oven. 98 35:53.81: Agate? 888 23:: 8.. £3» v.8 958:8 o: 2 mew—8:8 :30 5.. bemoan. 82:5 60:88 8: 8n :5 829.052 05 ﬁ 603.2: 0.3 8.? P25: 39385 :0 5:22

 

 

 

 

 

~25 85cm
and as on} 48.. e8 83 223.580
92. $2. «2.. :2. 32. mm: mm: 53.
83 ~25 m8... :3 on? 2 S 8:23.?
28.: 38.: :92: 88.: 88.: 88.:
8a ,2” «2. N2. 25 5a em<
88.: $32: 88.2: 882: 88.: 88.:
3.3 «2: :3 8.3 8.“ 8.2 9:623 .56
$2.: 32:: 8.2: 5.2: $8.: $2: 2:3 88825
e2 2... :... N2. ”an e: em... as 3.85m .56
as»:
$2: 22.: 8:: $~.: $2: 3...: $2: an: 8:: $~.: at: 2.2:
52. :.. 83 $3. ... as: ea... :.. 83 :3 :.. 83 3.8. 83 22 ... 82. .588
$2: :2: G2: 23.: an: 22: $9: 82: 82: c 2: :n: so:
5.? and «”2. . $3. and. $2: : $3 «.3? New? 82. :3. :.. 32:. 222.80 8:83
$2: an: 22: $2: $2: an: 22: $2: $2: 82: 32: G2:
and t. $2.2- 3.2- 8.3.: ~23. $2.- a .... ... mad- :3. 221 we: .2..- f: are}. 5
82: as: $2: $2: 2 Z: :1: $2: $3: $3: :2: $2: 2:: N
82.. ... 2 ..o :3 SS. :2: weed “8.9 ... 3:. e32: Sod Sod. 88.9 uu<
32: $2.: $2: Rm: :2: as:
3.3. :.. £2 .98. ~22. $3 :... 22 3% +2
a2: 5.: 3:: am: 82: $0: $2: $3: 28.: SN: 82: SN:
23. ... RE 82. 83. ~25 :.. ”who 83. ... 2.3 $2.. $3 $3 ... 32 use.A 2-2
3:: 82: $2.: $2: $2: 3:: $2: 5.2: so: 5.:
. Q2: 83 e 22.. ... 9.3 :3. ... 22. SS. 83 23. :.. $8.. $92 3
$2: :2: 2.8.: an: 82: an: 23.: 3°:
82:. ”:.o 3.2.. .. 82: :3. ... 82. . ewe? ..: $2. 3&3.
$2: Ex: 2;: 23.: $2: 5.: 5.: 5.:
82:. 82. . coed. : 83 ”3o. ._ 83 22:. ... :2: 28.2 2
«38:8 :5
if: 53.5 555 535 555 3E:
... Bee .3355 ... Es. 339580 ... Bee .8355 ... Eem axon—89 ... E8 .83:er ... ....en .3355
.83 see... 58m 23.... assesses”.— ..em seam 35... Seen 3......— easisdsm ..em
nose? no:
8...:

use: 9. ...—em 5.: 9899.8..— un:ee__om 55 anus—1.32 be scuba 2:.
"2332:...— oun3 $899.30 3.2.90.5
mmﬁ 93:.

 

8855...... ... .5 .8: 8.. 8...? m 38...... 5.2. ...... .....8 ..:...x... ...... ... 28.-:88 $8553.. ... 2.. 8.8.3.3 05.8.2 5.833.222. ...... .26.

b.5EEoo u... .a mctoﬁiu 2 $450.. 2a Eoto vac—am ...-088.89 :2...quon Sam no 083 003.323 2a 3:0..qu3 5.80.3 $5.85... 5.6 8.. ...-80» +2 .8803 3.. 85a? 2:9- 5.. on; :3 :55
o. on: 03:. Ba $=o§3€< .Aﬂufc? 388 9.32. 8.. 28» 0.8 9585... o: m. 2:338 :16 5.. bemoan 83.80 .votoau. .0: Bn .2. 8:28.859. 0.... ... cog—2.. a.a 853:3 .9556 3o.>.o.=. ..o 552

 

 

 

 

 

.828. ”858
:2 :2 83.8 888.8 .2 .88 Seats-no
:8... 82. $8.. .82. 82. S. .. mm: 83.
22. ... 2. 83... 8... ... .32. 3.2. .- .. 8.8.8...
882: .52: 88.8 :82: 882: 82.:
m2. .8... o: .8... 8.: a... um...
82:: 82.... .82: 88.: 8.2: 88.8
888 8.. 8.8 3.8 :1 8.8. M.....8..om .50
.38.: 82:: 882: 882: 8:2: 882: 8:3 88.832...
8.. $8 8.. 88.. 2... 82. um... ...:w 85.8.38 .56
:8-..
82: 8 . ..: 82: 88.0 .8. c 82. 82: 8:8 .. 82: 88. .. 82: 82...
oawd. .5. -w.w moo. o: 0320. Smﬁ- ..: mama $3.0. o: mind Nomi. :06 omod a: .3... 389.50
8:8. :2: 8...: 88.8. 82: 82: 88... 8...: 82: 8...: 8:: 8;.
: 8... 82.. 82.9 : 88..- 3.2. 88... :2. 22. .82. MS...- nv2w :.. 82.. 20.2-=89 5.80.08
82: 82: 82: 88.8. 88... 8.... 82: 88.8. 82: 82: 82: A. .8.
83 88.8- 22. 3 28.2 . :88.- 808- $2.- ... 8.3- 8.3.8.- 888- $2. : 8...? ...... 5%?
82: 82: 82: 88... .8... 88... 82: 8.8. :2: 82: 82: 88...
88.? :2. 22.. . 82.- . 8.... 82. 82. ... .82. 8:... :2. .8? . 82. a»...
82. :2: 82: 82. 82: 82:
8.2. ... ... ... 82. 88... 82...- ... 83... e8» +8.
8.... 82... 8...: :2: 82: 8.8. 82: 88.8. 82: 88... 82: 82:
3.2. ... 88.. 8:..- 838 as... ..: 22. 82.- ... :8... .52. 8.2. 82...- ... .8. may. 8.-.:
82: 82: 82: 88.0 82: 82: 82: 82: : ...: 82:
:82.- .: 8...: 8.... ... 22. £2.- .: 3...... 882. 8.... .8? :.. 82. e8» 8-8
82: 88.9 82: 88.8. 82: 88.... 82: 88.3
wwodu ooo vwmd ogd o9. ObNd .36. no. Novd OnNd. ... 9:6 38% 0.?
82: 88.8. 82: 82.8. 82: 8.8. 88... 88.8.
2.2: ... 882. .8... : 82. .....o- : 8.2. 832.. .. 82... e8» 8-.
85.8.3 55
92:: 92...: 32...: a.a-5 :21: cans:
... Eon ..:.uESu ... 9.8 2.0.9525 ... Es. .....eESu ... Eon ..::an ... ER. .5355 ... Eon 2.0.0580
.893 8.2.... .325 9.3.... 305.935”.— ..om .Soom 8!»...— .330m 83.:— ...oE.~o_&Em 20m
5835 no:
8......

:..—.5 ... ...—cm 5.3 3.3.8.: u==ooaum :30 23.—:-....2 ..: 83.5 0.:-

"nastasm can? 688th 33.8—om

Um... «Zn-F

206

Table 7.6

Selectivity Corrected Wage Functions:
Summary of the Effects of Own Schooling interacted with Born in Urban

 

 

 

 

Self Employment Public Sector Private Sector
Coefﬁcient Born in Coefﬁcient Born in Coefﬁcient Born in
Urban Urban Urban

A. Men
Linear Own Schooling
lFLSl 0.127 "* -0.009 0.105 *“ -0.053 0.107 “"‘ -0.015
IFLSZ 0.102 '" 0.036 0.195 -0.135 0.083 *“ -0.008
IFLS3 0.097 "“ 0.013 0.109 “ -0.107 0.072 “'" 0.023 "””
Non-linear Own Schooling
lFLSl
1-3 years a) 0.070 -0. 180 0.131 -0.107
4-6 years 0.429 "* -0.237 0.503 *“ -0.038
7-9 years 0.921 "" -0.102 0.109 0.068 0.852 ‘" —0.348 "“
10-12 years 1.505 "" -0.537 0.672 " -0.372 1.166 ""' -0.201
13+ years 2.061 "" -0.336 0.968 ”* -0.199 1.829 ""' -0.335
IFLSZ
1-3 years a) 0.334 "" -0.313 0.166 " -0.058
4-6 years 0.559 "‘" -0.386 0.400 "* -0.111
7-9 years 1.004 *" -0.170 0.207 -0.262 0.815 "" -0.221
10-12 years 1.304 *“ 0.163 0.748 0.653 0.923 *" -0.160
13+ years 2.327 *" 0.245 0.992 -0.451 1.385 *" -0.139
lFLS3
1-3 years a) 0.238 “ -O.445 0.218 “ -O.161
4-6 years 0.410 "" —0.256 0.402 ”‘ -O.141
7-9 years 0.699 "“ -0.031 0.154 0.325 0.644 “" 0076
10-12 years 1.211 "" -O.235 0.612 0.474 0.910 ‘" -0.082
13+ years 1.845 *" -0.235 0.935 0.681 1.117 "" 0.318
B. Women
Linear Own Schooling
lFLSl 0.054 "“" 0.063 0.034 0.094 0.108 ”" 0.014
IFLSZ 0.057 "" 0.020 -0.046 0.025 0.101 "* 0.015
IFLS3 0.051 *" 0.019 -0.173 -0.046 0.095 "“ 0.024 *
Non-linear Own Schooling
1F LS]
1-3 years a) 0.242 "‘ -0.612 -0.020 0.381
4-6 years 0.408 "" -0.216 0.283 "" -0.018
7-9 years 0.613 “" 0.030 0.596 ‘”" 0.187
10+ years 0.664 “ 0.602 0.166 1.187 1.647 ""‘ 0.062
IFLSZ
1-3 years a) 0.303 " -O.460 0.009 -0.006
4-6 years 0.292 " -0.244 0.178 -0.096
7-9 years 0.745 "" -0.528 0.207 0.572 "'
10+ years 0.758 "‘ 0.252 -0.680 0.606 1.423 "“ -0.123
lFLS3
1-3 years a) 0.252 " 0.001 0.329 "" -0.075
4-6 years 0.270 "" 0.090 0.384 '" -0.088
7-9 years 0.433 "“ 0.199 0.751 "" -O.250
10+ years 0.786 "* 0.086 -0.408 -1.097 1.275 "‘ 0.244

 

Source: Based on estimates of Appendix Table 7.1A, 7. 1 B, 7.1C and Table 7.5A, 7.58, 7.5C.
a) Omitted category for public sector men is 0-6 years, for women 0-9 years.

207

88558; E 98 anvk 8.. mos—a? R 8836:. Cake— 98 Aeavxn 6.2.5: 05 8 ~585ch 385588 E 93 8:883; 2203“ 5.8583888... 9.8 .08.
88:53 2.. a war—89:0 2 8.58 8a .988 88855 88:58 823E009.“ Sam 8 883 80830.8 08 3.8.qu8 .8838 $5828 :30 8.. v.80» +2 EoEoB 8.. 8335, 089. 8.“ 08m :8 :95
3 28 89: 98 35.5384“ A838? 888 23:9 8.. Enos 0.8 958.8» 8: mm 958:3. Eso 8.. Damage 88:80 83892 8: 08 :5 838082 05 E nous—oi 2a .8385, .3288 3088:: .8 5.82

 

 

 

 

 

.55: “858
m8 S~ ~43 8.3 ~8 9.3 2833.3
83 2.3 $2 8:. was on: mm: 83.
2 3 8:8 33 8 3 33 m: .o .. z 8.3?
5 _.8 88.8 $8.8 88.8 88.8 :88
3.. 3.8 we... own an £~ om<
$8.8 $3.8 88.8 88.8 88.8 88.8
3.2 8..~ «3 $2 .3 :.: M5.8.3 :30
33.8 83.8 $8.8 $8.8 $1.8 :28 E 228835
a; _~.o 8.. 8 .~ «2 o~._ 0? as ”532$ :30
..Si
338 828 $08 A88 no.8 3~8 A38 808 :58 A3,: and 2m:
. $3. 3. 8..“ :2- : 83 52 : as; 88 ... $3. 33 as.“ 82 ... 8m.“ 3528
5.8 888 888 8a: 38 Ga: §.: 5.: 5.8 $~8 A-.: 8.8
: «£8 $3. 82. 88. «:5. 83.: ~38 . 3.3 ~88 .82.? ~98 ... -~._- 20.5880 8:83
Eu: 2 <8 :58 $8
. :.3 .. 83 .. 33 29° 2:85 9.80
an: 5.8 5.8 5.8 :...8 A88 388 :28 ~58 398 at: A-.:
3.? ~88 Rod. 83 :2. 88.0. ~88 .. 8 ..o. 53.. ~88 $3 3.3.. f: 8.8...
5.: $68 5.8 :...8 5.8 95.8 5.8 :98 .88 358 8%: 58.8
~88 83 ~88 _~o.8 898 owed ~88 : 85.8 82.. ES 88.? ago om<
5.8 888 A88 8.... 8.8 88.8
3 Rod ..: 8... ~36. .83 35.8. ... :3 5% +2
A38 398 5.8 398 5.8 398 5.: 8.8 3.8 :88 5.8 $8
9.3 ..: $2 33 ~88 :.. ~2._ ~58 83 ... and 2 3 23 SS. ... 8:... Ex ~78
:08 5.8 5.8 3nd :...8 3.8 328 5.: 898 5.8
RE. .. :3 ~38 .. -§ ~88. ... :.3 :38 $3 $3 :.. a; an.» i
8.8 an: 5.8 5.8 5.: 28.8 $~8 and
83. 23 39o .. 83 . ~03 ... 83 Rod. :.. 33 e8» 3.
$08 a~8 A88 58 80.: :88 3.8 5.8
88.8 88¢ Sod. -_ .8 Ed ~38 88. ~38 an; 3
5.8.3 5.5
8828»— ouaoEuoz 02.038: 3.838: 3.823.: 3838:
55.: 2.0.9580 52:: ...suESu :35 30.2880 :25 .3880 :35 2.3880 :35 339580
8.3m 8.2...— 880m 9.3...— EoEmoEEm :5 880m 8'2...— 885 9.3.:— aao—SneEEm 20m
8:83 8:
5H,:

3.82%»— :uf: ...—t .8828..— u==eeaum EKG anus--82 .8 88cm 2:.

8.8.8.8... can? 8899.80 5.280—om

«SH sank.

208

38.855. ... a. as... s. 8...? a 838...... .88. 85 .8...“ ..:...x. 28 a 8.88.5.8 38.85.... ... 2.. 38838.. 328.... 8.83.3856: 85 .25.

8.8888 2.. .8 8.5.8.8 o. .88.. 8 Echo 88.38% 889.83 :o..8....uonm 3.5 .8 88.... 38.3.3 98 8.5.8503 8.80.3 8589.3. .58 8.. 80» +2 8803 8.. 838...... 088 8.. us.» :8 =8...»
8 3.. 0.8... 88 9§§q8< A889.» 888 9.3.... 8.. E8» one. m5..8..8 o: 8.. 8.89.8 :38 8.. bemoan BEEO 8389. 8: 28 5.. 88.8082 0... ... 80.5.9... 9... 838...... >88... 33.5.... ..o 5.8.).

 

 

 

 

 

.881358
88 8~ 88.. . $8.. 88 8.8 82.5880
888 .88 .-.. 888 3.88 8.... mm: .83.
8.8 -~8 888 888 .-8 8.8 3.8.8.8..
..888 8.8.8 .388 .8888 .888 .8888
R... 8.. 8.. 88 «.8 :8 o~<
.8888 .828 .8888 .8888 .8888 .8888
8.8 8.. 88... 88.8 8... ~8.~. 8:868 .50
..888 ..-.8 .828 .288 .888 .388 8:3 2885....
~.... 8.. 888 8.8 .88 R8 8.2 85 8:888 .56
.83.
.88 .88 .8... .88 ..88 .~..8 .~..8 .~..8 .88 ...... .88 .~..8:
1.8 ... 2.... 8~.~.- ... 8.n.- 88.. .. 8.... 388 ... 8:8 822. 8.8 88..- ... :3 .525
..:... .88 .8... .88 .~..8 .. ..: ..~.: .88 .38 .88 .38 .88
888. 8.8 . ..~8.. : 8-.~- .. 888. 8.8 3~8 ~88 ~..~8 8.8. 3~8 ..: 8.....- 8.0.2.8880 8.60.8
.88 .88... .88... .88... .8~.8 .8... .~..8 .88 .~..8 .88 .1... .8”...
838- . 8.88.- ~88. . 88.8 ~88 8.8.8- 8%.. ... 8.8.- v-8.. E8. .8..- .88... .88. 8%?
..~.: ...... Gm... .~..8 .28 ..:... .88 .~..8 .88 .....8 ...... .8...
888 888 8.8 . 83.8. 28.8 888 8.8.8 ... 8.8 88.8 888 2.88 . 888. ow<
.88 .88 .~..8 .88 .88 .88
.~..8 :.. 8... 8. .8 88 .38. ... ER .52. +2
.88 .~..8 ...... .8... .8... .v~.~. .88 .888 .88 .888 .888 .88
888. ... 8~.. . .~..8 . 8...- 888 : ~88 8888. ... 8.88 «8.8 8.8 ~. .8. :.. 88... e3. ~.-8.
.88 .2.: .....8 .888 .38 8.8 .~..8 ....8 .88 .888
... .88 888. -~8 .. .88 8:8 ... 888 888 8:8 888. ... .88 59. 8-8
.88 .88 .88 .8... .888 .~..8 .~..8 .88
8.8. .28 888. . ~v~8 8~.8. ... .88 8.8. ..: .88 an... 3.
.88 .88 .88 .88 ..~.8 .8... .888 .88
..:... 828. .8~8. .. 888 888. n. .8 8.8. : 8~8 :8. 8-.
«£858 5.5
3.828.. 8.8.8.3.. 3.838% 3.86.8.— 3.81.8..— 3.80.81
.25 .....oESu 53.: 8.39880 ..:..5 .3828 82.5 30.9880 3...: =59.qu .385 .838er
8.8m 8.»... 880m 3......— ...u...8......m ..um 8.3m 39...; 8.3m 9.3.... 88.3.8... ...om
38.83 5.2
8...:

3.838.. 8...: 5.3 .88-8.... "3.82.5 :30 80.5.2.2 ... 88...; 2....

3.8.8.8... and; $88.80 b.2825

ES 0......

209

385—.28 5 2a 683* 5.. mo:_a>.R 63365 73°:— 95 Ancvoxh A253... 05 a 2.855% 63055.8: 5 8a 3338?. 2283‘ .Eutmavoxmobﬁ: :5 _o>o_
gasp—Eco 05 a war—88.9 2 “mane.— Ba ﬂoto vécﬁm .86—Emu 5:859on 345 :0 v3.3 88.3.8 03 “2.3qu3 5.30.8 $5.838 .56 8.. E3» +2 E253 8.. mosntg 088 .8 37. :3 :88
S 3: 0:9: 08 3=oEHSE< .Aﬂofok 889. 23:: 8.. who» :.:: 2:033. 0: ﬂ a.a—8:3. :30 5‘. banana @0250 60:82 8: 08 SD 229.832 05 E vows—95 98 853:3 ASE—6 Bo._t8=_ :0 5:02

 

 

 

 

 

$5: ”853
:3 own 93 82 .2. :3 823283
3.2 Sod 22 8g :3... mm _ ._ mm: 63.
$3 23 m8... 3:. :3 25¢ .. x was?
39.8 88.8 38.8 88.8 $8.8 $8.8
33 :.m w: 8.” 22 Ed ...?V
88.: can: 68.: 88.9 $8.: 88.:
ME :3 can :..: :2 2 ..: 9:623 :30
:38 $8.8 83.8 28.: Gone :8: 2:3 message.
25 2 ...” :.N 2.” who 8.. om< v5 2:33 56
.87k
32: A3,: 5.: and § .: an: 32: SI : 3.8 5. : an: em:
_ww.o a: moné . 23.07 .0. WNTFN mOwN- :3 62:. mmmdu o9. th...“ wme- 87m hood. 0.. Eng. Edam—50
$2: 5.8 an. : 3a.: 5.8 23.8 34.: A3,: 5.8 898 a: .8 A3:
:.3. SI: ”3.. : £2- 33 £3. : o3? ... ~35 Ed xx; 23 ... 9.3. 229580 882%
so: an. : so. : so: 2 m. : as: A3: 3. : 3.8 an: 93.: SN. :
a8.“ 83- . 30$. ... v2.2. 22 _- 3:- $3- . 33. 3.2- $3.. 23- :3. f: 5%?
32: 5.: 3.: an: an: 5.8 2%.: 5.: :3: and: a»: an:
Rod. 08¢ . $2 : :99 we; we: :55 mac 2 3 :3 25¢ 53 g2
22.: :3: :.:: 8:: E: 33:
:3 ... ES 33 no... : 32:. ... 92 3% +2
3.8 so: 5.: 3m: 3:: an: 3de :2: :.3: a”: 5.: 3.:
83 :.. 82 82 £3- .33. ... £2. 83. ... 82. MS... I: . $2.. ... 22 was 2.2
3...: 8.: 33: ab: A8,: 3%.: are :2: ES: 5.:
22. 83 v2; . :3 83. ... $2 :3 E; :3. ... m3... 98» i
33: so: so: 30: SN: 5.: 3%: 3w:
awed wONd ommd. o. acmd EN? o: 091: WONG. o: owmd g 3
so: so: so: :9: :2: 2c: 5.: 3.:
:3 $3 23. 23 2 ..o. .. ”MS 33. . 33 e8» 2
«5.8-om :5
3.3231 8.829:— 3538... 3.338% 3.328.: 3.338:
535 2.35.89 .35 2.25.89 .35 ...qu30 :35 .8355 555 2.9.9580 Sf: :.....uEuou
3.90m 3.2.5 ..83m 83...— .auﬁwoiam :5 .335 85:...— ..23m 3......— uaoanEEm :5
5:33 .32
mm:—

ouau—xuoz :.....D 5..» 39:8:— ua=¢oaom :30 ..:-...-..oz .3 88:”.— 2E.

"2.2.2::— ou!‘ $899.30 5.2.8.5

05.6 03:.

210

Table 7.8

Selectivity Corrected Wage Functions:
Summary of the Effects of Own Schooling Interacted with Urban Residence

 

 

 

 

Self Employment Public Sector Private Sector
Coefficient Urban Coefficient Urban Coefﬁcient Urban
Residence Residence Residence

A. Men
Linear Own Schooling
lFLSl 0.116 "" 0.008 0.197 " -0.147 0.078 *" 0.022 ‘
IFLSZ 0.108 "" 0.003 0.295 "' -0.226 0.071 *" 0.006
IFLS3 0.102 *” -0.012 0.081 -0.017 0.072 ”" 0.012
N on-linear Own Schooling
lFLSl
1—3 years a) 0.042 -0.050 -0.042 0.273
4-6 years 0.358 *" -0.055 0.305 ‘" 0.262 “
7-9 years 0.757 "“ 0.169 0.368 -0.391 0.741 “"‘ -0.068
10-12 years 1.418 “" -0.262 0.537 -0.115 0.823 “" 0.303
13+ years 2.088 *“ -0.044 0.801 -0.042 1.107 "" 0.627 “
IFLSZ
1-3 years a) 0.267 " -0.104 0.115 -0.045
4-6 years 0.481 “" -0.105 0.357 "" -0.120
7-9 years 0.937 "‘ -0.079 0.116 0.009 0.705 ”“ -0.116
10-12 years 1.396 "" -0.112 0.518 -0.024 0.819 "" -0.090
13+ years 2.477 ”* -0.141 0.745 0.110 1.103 ‘" 0.125
[F LS3
1-3 years a) 0.204 " -0.257 0.238 ** -O.115
4-6 years 0.389 "" -0.265 0.456 “'" -0.214
7-9 years 0.628 "" -0.014 0.077 0.311 0.583 "‘ 0.003
10-12 years 1.292 *" -O.461 " 0.814 0.033 0.946 "" 0156
13+ years 2.143 "" -0.743 " 1.163 0.145 1.018 "'" 0.313
B. Women
Linear Own Schooling
lFLSl 0.037 ” 0.048 *"‘ 0.159 "* -0.103 0.074 ‘" 0.043 "
lFLSZ 0.034 " 0.044 ‘”" -0.224 0.231 0.082 *" 0.027
IFLS3 0.043 “" 0.013 -0.204 0.128 0.091 *" 0.023
N on-linear Own Schooling
lF LS]
1-3 years a) 0.123 -0.020 0.039 -0.076
4-6 years 0.318 " 0.044 0.212 -0.110
7-9 years 0.622 " -0.043 0.611 " -0.127
10+ years 0.018 1.132 ““ 0.052 0.736 1.341 "'" 0.253
lFLSZ
1-3 years a) 0.276 " -0.201 -0.154 0.174
4-6 years 0.242 * -0.098 0.151 -0.147
7-9 years 0.381 " 0.222 -0.306 0.978 ‘"
10+ years 0.622 " 0.376 -3.125 "' 3.021 ‘ 1.298 "" -0.059
IFLS3
1-3 years a) 0.232 -0.210 0.149 0.221
4-6 years 0.306 "”" -0.326 0.268 0.080
7-9 years 0.271 * 0.064 0.340 0.420
10+ years 0.686 "" -0.041 -1.686 1.707 1.250 ‘" 0.202

 

Source: Based on estimates of Appendix Table 7.2A, 7.28, 7.2C and Table 7.7A, 7.7B, 7.7C.
a) Omitted category for public sector men is 0-6 years, for women 0-9 years.

211

 

 

880553.. ... 2.. «.2....

...... mos—«>- & 633...... ......b. E... .....e\..n 53...... 95 ... Enemiwﬁ 80852.. E P... 8.8588. 828.2 .....ozmuvoxmgﬁn v5. _o>o_ b.5653 o... ... wctoﬁio ... .95.: 93 89:0 gm 8...—99.3.
.... v.32. 3 teen—a2 8.. wczoesum Es... ...... 8.2.5.. Ev £023 ... 38:58 52359on 08m .... .88.. 38.3.3 98 85.0503 canoe—um ...onESQombgmac. m. 3015.... .... 5:9: ...... bemoan. EEO

 

 

 

 

 

...-E .858
8.. SN 8%. 3n. 8. 8...” accents...
N8... .8... ....N. 8.... 8.... 8 . .. mm: 83.
8n... ...n... m. .... ....N... 8N... ..:... N x 38......
88.8 ...88 .~88 .81.... .28.... ...88
8... ...... ..2 8.. 8.. ...... 5.5.5... 5.82
.88.... .88.... .82.. .89... ...88 .82..
8... 8... N... .2 ...... ...... um...
...88 .21... ...88 ...88 ...88 ........8
8.8 8... ......N 8...... 8... 8.3 M8.8.8 .50
.28.... ......8 .85.... .82.. .32.. .85.... £5. 288.2...
...... 8... $8 8... 8.. 2... am... 2.. 828.3- .50
:2-..
....8 .88 .88 .28 .3... .8... .....8 .8... .8... .88 ..N.... ..:...
8...- ... 8N... 83- ... 8.3 . $3. . 82 8.... ... «88 82. ... .8... ....N... ... 82 .5326
.38 .8... .8... .8... .88 ......N. .8... .3... .X... .....8 .8... .88
...-N... 8...... 8m... . 8.....- : N8..- ... :..... 88... 8.... 8.....- .8... 8n... ... .8... .5682... 5.8.8
.....N. ......N. .2.... ...m...
: .8... : :.N... .8... 8.... 258.320
.88 .8... .88 .8... .88 .3... .8... .88 .88 .....8 .8... .~..8
....N... 88.. .8...- . ....N... .8... «2... E... 8.... c8...- 58... ...-.....- mbo... 52.:
......8 .88 .8... .8... .88 .88 .88 .88 .8... .....8 .....8 .2....
8.... 8....- . a8...- 8N... ... 8.....- : ..8... .8... 8...... 8.....- ....o... 8....- 88...- 82.85
...v... .38 .88 .8... .28 .....m. .38 .8... .88 .~..8 .~..8 .88
.8... 8....- 8.....- :.....- .: m8...- .: 8n... 8...- ...N... ......- 83. : «8...- ... 8.... .8582
...... .....8 ......8 .88 .88 ......N. .88 ...m... .88 .88 .38 .3...
SN... 3...... .8... .m.... : 8.....- :. 8.... .2.... 8N... 8.... 8...... ........ . 8N... .8300
32:02-— .—° 5:62
.88 .8... .88 .8... .8... .88 .~..8 ..:.N. .2.... .88 .....8 .....8
8.....- R.....- .8..- .8... ..:... : ... .... 8.... : 8.....- .88 8...... m......- .8...- ...... 5.9.2
.88 .8... .88 .....8 .8... .8... .88 .38 .88 .2... .~..8 ......8
v8... 3...... 8.... 8.... 8...- . 8.... 8...... : m8... 3...... ..8... 8.... 8.... a?
......8 .6... .88 .88 .88 .28 ...N... .32. .8... .88 ......8 .88
1.... ... 8.... 2...... v8... ... n8... ... v8... 2...... ... 8.... m8...- :. 8.... 8.....- :. 8.... 8.8.3... 88>
nun-5 ..:—.5 :.3-5 5...: ......5 62...:
... E...— 236580 ... 55.. .=«..ESU ... Eon 839.880 ... Eon 39.9580 ... E...— ...o.oEo..u ... E...— 2.9.9589
Snow ..:... 3.99m 2......— =§=8=Em ...m .83 32.... .398 9.3.... 38.39.... .3
5:33 52
.3...

5...: ... ...—on 5.3 9899.8:— ue=3gow :30 .32.: .... 83:“.— 2:.

8:235...— ouaB “.38-.30 5.280—om
<.... ...: 5.5.3...

212

«805:2... ... 0.3 82. k

.... mun—a? Q 638...... .83... v5. .....xh ..:...x; o... .a .58:in ...-895:2... ... 98 8.53.8? . 828.7. .b.o..m~voxmo.o.o: «.5 .96. €5.55... 2.. .a N5.8.2.... o. ...-:..... pa 8...... vacuum 85.09.3-
.... 28.. .3 coon—...... o... 9:333. .56 8.. 838.5, .9556 so...) ... 888:8 22305925 095 :.. 083 Rania—8 0.8 85.05.30 3.82% con—5.5.33. a. 30.5.... .... 5:9: 3.. .9388 38:5

 

 

 

 

 

..:—.5 ... :.3: a...» $899.02.— 9502.5 5:0 :2...— u.. 33:”.— 2...

...—e325..— ouu.‘ $88.55 5.280—om
a... .....F 5.2....

8...... .858
.... ..N 88.. 8.... 8.. ....N Essie
3.... 8.... «8.. :..... 3...... .2.. mm: .8...
8N... .8... .8... n . N... .NN... .... .. .. a 8.3.3.
..88 ..88 .88.... ...N.... .82.. ......8
8.. 8.. ...... e... 8.. 8.. 3.33:... ......2
3.8.... .N....... .N . .8 ........8 88.8 .... ..8
....m .... ..N.. 8.... .... 8.. a...
......8 .8... ...88 ......8 88.8 88.8
8...... .N... ...... 8...: ...... 8.... .....Sﬁm .50
......8 .8... ......8 .82.. ...88 .82.. 9.5. 82.8.2...
.... ..N... .2. 8... RN 8... 8.. ...... 8:8... .56
:3...
.8... ...... .38 .....N. ...N... .N..N. ...... ..N... .8... .88 ...... .2....
.8. ... N8.N 8..N- : .~..8 .8... : N8. 8.... :.. 8.... 3.... .....- ....N. ..: N2... .528
.88 ...... ...-.8 .8... .8... ..:... .....N. .....8 .N..... .88 ...... ....N.
..N... 83 8.... 8...... NE...- .8... ... .8... ......- o.....- .8... ......v .. :..... ...-.6880 8.8.8
.88 ......N. .88 .8. .. ... .... .88 .88 .88 .. ..8 .88 .88 .... ....
.8...- : 8.... .8... E... 8.... ...... 8....- .. Na... .8... 8.... .8... :..... :.3-.55.
......N. ..N.N. .8... ...N... ...... .8... ......8 .8... .....8 .88 .88 .. ..8
.. 8...... .. .8... . 8...? 8...... SN..- 8N... 8....- 8... 8.... 8...? 8.... 3.... .358...
.....8 .. .... ...... .8... .2.... .....8 .8... .88 .88 ...... ...-.8 .....N.
2 ....- .2... 8n...- 8N...- .88 .8...- 8 ..9 N8... N2... N2..- E... : .88 .8882
.88 .2... ...... .88 ...-.8 .8... .88 .8... .8... .8... ...... .8...
3...... 8.... .88 8.8 .88 ..:...- 8... .8...- . 8...... . .8...- 8...... 8....- .828
tats... .. 5...:
...... ...N... ...-.8 ......8 .8... .8... ....8 .8... .....8 .N..... .....8 .8...
...N.. :.. 3.....- .8... 8.... 3.....- ......- .3. ... 8...- 8.... N88- 8.... ......- ...... ...»...
..:-.8 ......N. .8... .8... ....8 ...... ...... 8...... .8... .....8 .88 ...-.8
.8...- .: N. .... :..... 8.....- ........ .8... 8.....- .: N. .... 8....- EN... .8...- 8... a»...
.....8 .8... ...N... .8... .88 .2.... ...-.8 ..N.... .8... .8... .3... ..N...
2.... ... :..... .8... 8......- ..8... ... .8... 8.....- .: 8.... 8....- 8... .8... ... 8.... 8.8.3... 38>
3...: 5...: 3...: 8...: :..... 5......
... ...... ..:..ESU ... ...... ......ESU ... ...... 23.8.80 ... ...... ......ESU .. ......— ...o..=.2.u ... E... .5352...
.893 8:...— .83m 32.5 .agmsaﬁm .....m .335 8.2...— .88w 8......— EoEaSQEm :5
=33? :02
8...:

213

 

 

6.9.9555. ...

9.... 83k .... .2.—«... a 628...... $.32 ..:. 5.53% ..:...x; o... ... 2.8555 .momoﬁcaaa ... 0... «9.8.28... 9.2.03... .b.9..m«voxmo.o§. ...... _9>9_ $5.558 9... ... 95.9.3.9 o. .25... 9... 2...... €3.85
9.5.9.9. .... 23» .3 v3.39. 9... 9599.9. :30 8.. 833...... ..:—...... :9...) ... 3.3.3.9 5.895995 9...... .... v9.3 3.33.8 9.8 35.2....98 ......929m 5.59:... a. 39.29.... .... 5.8:. .0.— bowoﬁ. @9280

 

 

 

 

 

 

 

...... .82...
...... .... ......N .....N .... .N.... 22.383
...... 8.... .... 8... ...... 3... mm: ......
..N... N . .... 3...... ...... ...... ...... .. z 32......
..8... 3...... .82.. 88.... 8...... 88...
8N ...N ...... ...... ...... N... 3.52.... 5.82
......o. :8... 2.8.... :8... .....o... 38....
N... ...... .... ...... a... .... a...
88.... ..:... 88.... 88.... 3...... 8......
8.3.. .... 8.... 8...: .... ...... 8.8.... .50
3...... .39... ......o. .58.... ......o. .3... 2.....2....2.......
...... ...... ...N .... 8.. .N... a... ...... 2:8... .50
28-...
.2.... .3... .N..... ...... ...... ..:... ...... ...... ...... ...... ...... .8...
37¢ 999 3.0.? mwmdu 999 oohNN Mmhd- :... bmvé N36: 9: 2nd mm—é OVTV 5:6 9: eomé :.3ch
.8... .8... ...... ...... ...... .8... ...... ...... ...... .2.... ...... ......
...N... ...... ...... : .....- 8.... ...... . 3... .8... .8? 8.... 8.... ... ...... 22.88.. 8.8....
...... .....N. ...... .8... .3... .3... .8... .8... .3... ...... .8... 8....
...... ..: ...... ...... 3.... ...... 2...... .8... ... ...... ...... N..... ...... .8... b25...£§>oz
.NN... ...... ...... ...... .N.... ...... .3... ...... ...... ..:... ...... .8...
...... 2...... .N.... ...... 8...... :....v ..:... E... ..:... .8... ..:... 3:... 3......0
.3... .8... .3... ...... .3... .8... ...... .8... .8... ...... .. .... ......
...... 8...? ... ...... . .....w 8... .2.? :..... .8... 8...... .....o. ..8... 3.... .3288.
.2.... ...... ...... ...... ...... .2.... ..N... ...... ..:... .2.... 9.... .....N.
...... 8...? . N3... 8.... 8... 8.... ...... :...9 ...... .8... 2...... .. ...... .292
its... ... 5.3...
...... ...... .8... ...N... ...... ...... ...... ...o... .3... .3... 8...... ......
...... . 3...- .....- : 2.... : 2.....- . .8..- ...... ... :..... ...... 8.... .8... .. .8..- ..-... ......<
.8... .8... ...... .8... ...... ...... ...... ...... .3... ...... .8... ....N.
3...... . 8.... ...... : ...... . 8... . 3...... .8... ... 8.... ..o... 2...... .8... .. o8... ......
...... ...... .. ..N.... ..:... .3... .. . ... ...... 8...... ...... ...... ...... ......
_. 38... ... ...... 3...... .2.... ..:... :.. ...... :.. .8... ..: N8... 8.... . ...... ...... ... 8.... 2.8.8.32...
3...... ...—...: 5...... 8...: :..... =3...
... ......— 2595990 ... ...—3— 2.9.95er ... :.3.— .9395990 ... 9.9m 39.95990 ... ...—on 39.95990 a. E...— 2.9—95990
.88. 3.2... .88. 3...... 2.2.2.2.. =9. .88. 3.3.... .38. ..:... Eggs... ....
98.33 :92
......

99...: ... ...—on 5.». 3.99.8:— 9.=8..9m ..BO .92.: .... 590.5 2:.

...—3.9.5..— 9ua3 6899....0 ...—.599.“
...... :..... gaze...

214

882253: 5 Ba .98-k

é 88?. a 828.65 5.88. 85 :.3.“ fix; 28 .8 2:35:85 885838 5 ea 8.58%; 2285 868383522. 85 .32 £5,258 2.. .. 8:38: 2 .982 2. e85 853m 8:888
..o £89» an wood—n2 88 wE—oonum 6.6 .8 83895 EBB: 523 5 385.58 con—853% 8.8m no woman 3.8—:28 98 3:3qu8 cocoa—om .baEBnumbmsw:< 8 33385 ‘8 5:9,: .8: bow33 32:5

 

 

 

 

 

..md. ”8588
88 $~ ~83 $9. 88 $9~ 288858
888 ~88 88 _ ._ .88 .88 82 ._ mm: .8:
8898 88.8 328 8:8 £~8 ~28 N x was?
: 2.8 :88 83.8 82.8 32.8 8898
898 8.. v9. 38 8.. R... 305258 .882
32.8 8898 A898 598 8898 $8.8
«2 89: 88 89". $8 83 om<
8898 E98 8898 8898 A898 8898
:.3. -... 3.8. 88: 8.“ “~88 8:858 E50
:88 :88 5:8 :88 88.8 8898 2E3 288.35
«3 $8 5.: :.~ 8: 8~._ &< 85 8:888 .56
:34
38.: 83: 88.8 89: 5.8 an: :98 :88 39: A88 88.: 5.:
.. 8;. ... 888 888 .88 ~38 : £88 58 ... :98 «88 ~88. 29. ... 8.9“ .8328
$9: :98 :88 :98 398 :88 898 :9: $9: $9: 898 8.8
: ~58 888 888- 8888- .88. 828 _:8. . :~8 . 297 8:8 898 ... 8...- 629880 8.82%
$8.: 89: :98 $98
. 888 : 838 : 83 8898, 8.85 as:
398 $9: $9: 8.8 :9: 5.8 5.8 398 898 5.8 5.8 898
~28 ~88- 3~8 388 8:8. 888 888 3.8 828 8888- 898 $88. :25.
A898 5.8 :98 8~.8 :38 G9: 898 :8: 898 :.~.8 :88 A28
388. .2.? 88.8 888 88. .88 S :8- 8~8 8888 888. .38- ~88 .3888
5.8 88.8 :98 5.8 :98 5.: 89: $9: :.~.8 398 39: $9.:
88~8 8:8- 8:8. 88.? 8:8- 8~8 8~8. 398 8:8 888. . 88.8. ... 8R8 3:532
:28 3~8 £98 : _.8 38.8 89: 89: 33: 398 :98 :98 :8:
888 $88 888- ~88 88. :88 «£8- «28 88 ~_~8- 888 . :98 2990
30353:— .—9 5:0:
39: 898 £98 39: 5.8 :9: 3:8 898 :8: an: :4: 88.8
-_8. 888- 2 .8 888- 288 888- 888 .. 8:8. 828 3.8- 888 388. f: 8%?
an: 398 :88 39: :88 :88 88.8 89: 3.: $9: $9: $98
888 288 888- 8.8 888- 288 888. : 888 828. E8 888. 888 uw<
398 398 3.: an: .2.: Ed 89: 898 39: am: 898 :2:
9. mad :9 3.0.6 Me_d. 0.. 026 Q.- wowed .9 SEA: 0 NN¢.Q 0... ”No.9 5‘89: 99 hO—d £56 .0. o_ —.O mam—8:350 ﬂao>
8:928: 8528.: 3:928»— 35281 85288— 8.338”—
..35 2.3525 :35 323:8: 535 30.8880 if: .5358: 535 2.39580 535 229539
.325 8.3....— hoaoom 33...— EtheEEH ...om .330» 8.3....— uSuow 33.5 ...!FASAEH :3
5:33 :0:
EH:

8538: 55.5 5.3 98228:— u==8=um :36 .32.: :8 803.5 05.

3:285...— ouu? :83..ch malty—om

<2. .3: 885...?

215

 

 

88055.3 E 0.8 «hawk

5. 829.. a. 8.8.8. ....8. .8 ..:..x... ..:...x. 2.. a 28585 $0828 ... a. 8......53 05.8.... 6.2588232 ...a .26. 3.5.558 2.. a 88.9.0 2 .88. ....a not» 28:3 8.39.3.
.... £8» .3 @0832 2a 9:028 .55 8.. 838.8. >883. 523 ... 838.58 5.305025 85m .8 v9.3 83:28 98 8.8.0508 cacao—om ..BEo.aom>m:m=< m. 30.38.... .3 5.5:. 3.. bemoan. v3.50

 

 

 

 

 

 

 

.8... ”858
8... 88 8t. 8.... 8.. 8..” 2838....
m8... 8“... 88.. 8.... 8...... :.. .. mm: .8..
88... RN... .8... 88... ..8... 8 . ... n .. 8.3.3
.88.... $8.... .21... .8.... .8.... .83..
8.. 8.~ 8.. :1 a... m. .N 3232...... 5.52
.28.... .28.... .82.. 88.... 88.... .88....
8.. 88 8.. 8.8 8.... 8.. 9....
- 38.... 88.... 88.... 88.... 88....
8.8 8... 3...: 3.8. 8... 8.3. 8:8..8 .50
.88.... 28.... .8.... .82.. .. . ..o. .88.... 95o. 288.2...
8... 8... 8.. ...... 8.~ 82 om< ..5 8.39.8 56
.8....
... . .... .2.... .. . ... .8.”. .8.... .12. . . n... .8.... .8... .2.... .8. .. .8....
a8... ... «8.. .8...- : 88.. ..8.~ : 82 2.... ... 8.... 82. 88.. c8..- ... 8: .588
.8.... .8... .2... .8... .8... .8... .3... .2.... .2... .2... .8.... .8....
8...... .8... 2n. . m8..- . m8... ..8... ..:... :..... «8... Sn. $2. ... 8...- 8.qu8 8.80.8
.8.... .8... .2.... .8.... .8.... .2.... .8... .28 .8.... .8.... .~..... .8....
8...... .. .8... 3...... 82. 8.... 8.... 2...... : .8... 8.... 3...... N8... 8.... .8855.
.. .... .8... .8... .8.... .... . .. .. n... .8. .. .N. . .. .3... .8.... .8.... .8....
82.. RN... 8...... 8.... 8...? n8... 2...... 8.... o. .... 8.... 2...... 8.... 8.88..
.8.... .8.... .2.... .8.... .8... .8.... .8... .3... .8.... .8.... .8.... ......
8.... ..8... 8...... 82.. «8.9 8.... 8.... m8... ... .... 82.... 8.... . .8.... 8.882
.3... .8... .8.... .2.... .8... .2.... .8.... .3... .8... .8... ..:... ...m...
.8.... 88... 88.... .8.... 8a? 8...... 8.... 8...... 8m... . 8m... .8.? 8.... .3290
its... ... ...-..:
.8.... ......N. .8.... .2... .3... .8... .8.... ..:. ...... .8... .8... .3...
23- : 88...- .88- ~88 1.... 83.- a... ... 8.....- .8.... 8...“- 83.8- 83 ...... 5.8...
.3... .8... ...... .. .. .... .2.... ...... .. .8.... .8.: .8.... ...”... .8... .8...
.8... . «8... $2. 88.... 8...... N8... .8.... ... 8.... 88.... 88... . m8... m8... om...
.8... .88. .8... .8... .8.~. ...3. .8.... .88. .8... .8... .2.... ...E.
8.... :.. N8... .8... v8... : 3...... : «8... 8.... ... .8... 88.... . a... 8.... ..: 8.... 8.32.38 28>
ouaoutom 3:023: 3:938: ego—snou— ooaoEuom 3.338%
if: ...quer 5...... 2.0.9580 5...... 335.89 535 ..::an 5...: 2.0.880 52.5 2.0.9580
.....uom 3u>tm ..Soom 3......— aaﬂahoaﬁm ..om ..Soom Sgt.— ..Soom 9.3...— .aoE»2mﬁm =5
:.:—3.5 no:
8...:

3538.: ..:-.5 5.3 $83.32.— uu=ooaum :30 .35.— ... 590.5— of.

...—23......— ouuB 6898.30 32.3—om

..E o:..: €52...

216

6063.55 5

98 7.8- k 8.. 3365- Q 6286... ALXA: :5. A33.“ A353... 05 .6 585.5% 63055.3 ... o... 33:86-. 3235?. 563338.22. :5 35. b_§EEOu 05 .6 9.5.93 2 3.50.. 9.... Echo vgm
.w:=oo:8 .3 E8» .3 “.86—no. 86 2:85... :30 .8 836...? >862: 523 E 60.6838 :23quon 3.6m :o :33 van—=23 2.. 3:2qu8 cargo—om $3225.. 8 303.8... ..o 5.5:. 8.. bemoan. “SEED

 

 

 

 

 

.8.... H858
«2.. 6mm 9.6.~ 663 .8. .86 82.2860
3.6 56 RN. 6666 £66 ~v... mm: .8..
3.~6 ~86 $66 66. 6 Q~6 ~86 .. x 8.8.6:.
338 $3.8 32.8 866.8 88.8 .-6.8
~6.~ .66 $6 65. n~.. 6~.~ 5.55:6 58.2
26.8 866.8 888 8668 2668 $88
3.. 6; mn~ 66... ~66 ~6.~ om...
8668 38.8 8668 8668 .-~.8 8668
:... .. 3.6 m3. 3...: 3.. 66.: 65.86% :30
$3.8 $2.8 $2.8 .~.68 338 :28 2:3 26.882...
66.. 6.... 66.. $6 .66 26 om< .5 M3:623 .50
68.-.;
.608 .668 .8. : $8 ..:. : 3.8 :6. : 82: A. ..8 a: . : .608 .38
“$6 ..: E... 68.6.- : :....8 . 663.. ... .666 6%.? ... 6.66 ~66? .666 636. ... -..6 .525
8.8 5.8 :68 .8.: an: .38 8...: .88 8.8 .28 .~..8 .2.:
686. :66 m. ... .. $6..- 68.. 23.8 . ~36. ..: 6666 :.66. 3.6. 6666 ... 56. 220880 5:83
.-8 3...: 8.8 .28 8...: :0: 8...: .2.: :..: an: .668 868
$66. 82. :66 66.6 636 626. 6:6. ... 3M6 .~m6. ~86 : .666 ...6. b257§§>oz
.~..8 $68 :...8 .88 an: an: 86.8 .62. .8.: 58 $68 368
686 866 66.6 SS. «26 ~28 66.6. ... ~.~6 B~6. 2.6 ~:6 66.8 .325
8.8 .668 .668 368 .2.: 5.: 8...: 5.: .38 3~.: 2.68 26.:
2 ..6 2.6. ~v~6 626. §~6 . 33. 626 2 .6. : 6:6. 6:6 366 6:6. 383%
$68 .~m8 .~..8 5.8 26.8 3...: 368 :68 3...: 3.: .28 $68
62.8 6666 $66 .86. 66.6 .56. .86 ~86. E6. 3.6 «~66 .. ~86. .992
38:02- uc £232
868 5.: .8.: 5.: 3n: .~m8 5.8 36.: .~m8 $6.8 .~v8 86.:
...... . 6666. .636- _. 2...? 86.2. $2- 638 . 68...“- :36- 268 663- . 6666- .66. 5.9.2
.38 an: .8.: ..:: .66.: .608 :n: 9.6.: :68 :~8 :68 5.:
26.? ~86 6S6 . £66. . ~26 -66 8.66 266 $66 ~86 686 9.66 o~<
av: .62.: .668 2 ..: 5.8 .~..8 an: E6: 2.. .8 $68 .38 .28
~86 :.. .666 -_6 .63. 266 ... $66 2.6 ..: ~86 :66. :86 ~56. ... ~86 8:868... 33>
02.026.3— 3.323: 3:028: 3:038: 3:028: 3:028:
=35 .....uESu 52.5 23.880 SE: ..:.QESU ﬁe: 39:52.9 52:: 30.9589 5...: 329580
.393 3.2...— ..83m 33...— ...uEhoEEm ...»m .393 8.2.5 33% 33.:— EoENoEEm :5
5:33 :92
8:.

3.323% 33.5 5.? $3938.: "...—25% ...:O .32.: uo nook—n.— 2:.
3:385...— owug 3.99.30 b.233um

US. .3: €863.

217

Table 8.1
Own Schooling Attainment by Cohort and Gender

 

 

 

Young Old
lFLSl
Men Mean 6.5 5.7
Standard Error (0.18) (0.18)
Observations 2,1 14 2,161
Women Mean 5.3 3.3
Standard Error (0.16) (0.16)
Observations 2,868 2,487
All Individuals Mean 5.8 4.4
Standard Error (0.16) (0.17)
Observations 4,982 4,648
IFLSZ
Men Mean 7.6 6.1
Standard Error (0.16) (0.16)
Observations 3,019 2,588
Women Mean 6.3 3.8
Standard Error (0.15) (0.16)
Observations 3,641 2,942
All Individuals Mean 6.8 4.9
Standard Error (0.15) (0.15)
Observations 6,660 5,530
IFLS3
Men Mean 8.7 6.5
Standard Error (0.14) (0.15)
Observations 4,1 12 3,033
Women Mean 7.5 4.6
Standard Error (0.14) (0.15)
Observations 4,394 3,322
All Individuals Mean 8.1 5.5
Standard Error (0.13) (0.15)
Observations 8,506 6,355

 

Source: lFLSl, IFLSZ and IF LS3.
Young cohort consists of individuals aged 25-39 years, while old cohort 40-

59 years. Means are in years. Standard errors are robust to clustering at

the connnunity level.

218

Table 8.2

Hourly Wage within Sector of Employment by Cohort and Gender

 

 

 

Self Employment Public Sector Private Sector
Young Old Difference Young Old Difference Young Old Difference

lFLSl

Men Mean 762 782 20 1,466 2.136 670 “‘ 802 930 128 "
Standard Error (53) (54) (67) (88) (129) (142) (44) (79) (75)
Median 394 343 51 1,127 1,660 -533 550 525 25
Observations 952 1.155 256 318 843 548

Women Mean 848 683 -165 " 1.716 2.164 448 " 525 438 -87
Standard Error (63) (57) (78) (175) (199) (203) (47) (47) (55)
Median 370 275 95 1,267 1,886 -619 275 231 44
Observations 601 67 1 169 83 3 53 263

All Individuals Mean 796 746 -50 1.565 2,142 576 "" 721 771 50
Standard Error (43) (43) (52) (98) (128) (133) (39) (63) (56)
Median 385 322 63 1,182 1,732 -550 479 412 67
Observations 1,553 1,826 425 401 1.196 81 1

IFLSZ

Men Mean 1.378 1,311 -67 1.975 2.702 727 "" 1.320 1,452 132
Standard Error (75) (67) (91) (104) (122) (155) (47) (95) (95)
Median 714 619 96 1.650 2,307 -657 924 902 22
Observations 1 .038 1.280 275 370 1 .379 720

Women Mean 1,205 1,202 -3 2.153 2,902 750 "" 1,008 761 -247 "'"
Standard Error (85) (86) (l 10) (125) (209) (205) (69) (71) (86)
Median 495 495 0 1,983 2.566 -583 606 370 237
Observations 675 805 196 1 13 655 337

A11 Individuals Mean 1.310 1,269 -41 2.049 2,749 700 ““ 1,220 1,232 12
Standard Error (60) (55) (68) (87) (111) (120) (45) (78) (72)
Median 619 577 41 1,777 2,309 -533 825 674 151
Observations 1 .713 2.085 471 483 2.034 1,057

IFLSJ

Men Mean 3,110 3.191 81 4,153 5,901 1,748 ”“ 2,495 2,548 53
Standard Error (137) (154) (188) (226) (264) (351) (84) (143) (152)
Median 1.650 1.567 82 3.514 4,797 -1.282 1,650 1,540 1 10
Observations 1 .493 1.530 297 405 l .929 872

Women Mean 2.531 2,289 -242 4,699 5.651 952 ” 1.927 1.262 -665 ”"
Standard Error (183) (133) (207) (277) (329) (420) (109) (108) (127)
Median 1,006 990 16 3.936 5.164 -1.228 1,155 655 500
Observations 956 1 .091 194 1 58 9 12 464

All Individuals Mean 2,884 2.816 -68 4,369 5.831 1.462 ‘” 2.313 2,101 -212 ‘
Standard Error (115) (117) (140) (196) (218) (289) (75) (114) (113)
Median 1.375 1.283 92 3.793 4,833 -1.040 1.443 1,227 217
Observations 2,449 2.621 491 563 2.841 1.336

 

Source: lFLSl, IFLSZ and IFLS3.

Means are in current (nominal) Rupiahs. Standard errors are robust to clustering at the community level. Signiﬁcant at the
1%(”"). 5%(") and 10%(‘) indicated.

219

38:303.. ... o... .8.-.. 3.. 83.9.Q 383...... 5.13:. :.3 ......xa 5.3.9... 2.. ... 33.....«5 3853.2. ... 8a 85.3%-. 85.83. 53.333338: :.3 .26.
b.5558 o... 3 «atoms? 8 .33. 9:. Echo 53.55 335.50 332.3on Sam .3 woman. 33.3.8 a.a 3:0.6Eooo 3.30.3 «5.02.8. :30 3.. 9.3.. .5. 50.33 3.. 3.3.8.. 959. 3.. on... :3 :55
o. 3.. 0.3... p... $=§€< .53....» .033. 9.....3 3.. v.8» 3.... «5.09.8 a: m. «...—09.3. 53 3.. .6388 .3550 30:80.. .2. 2a ..:. 93.9852 9... ... 3.3.9... 2.. 833.5. .955... 39.38... .3 5.3.2

 

 

 

 

 

.3... ”85cm
2.. a. Q... :..... in 8... 22333.3
3.... an... a... :.3 .8... a: .. mm: .8..
8m... 5.... .8... EN... 83 a. .c .. z 85.8...
38.8 38.8 $2.8 33.8 88.8 .22:
:... 5.. 2... .3 av. v... om<
$8.8 32.8 88.8 88.8 88.8 88.:
~23 8.. .3 $3 5.. 3.2 .5893 :30
38.8 838 .33. 88.8 32.8 .22: 2:2 Eocene...
a... :..c S... S... 3.. 2 ._ a... .5 5.8.8 .56
.8.-..
.2.: .38 .2.: 3...: so... .8.: 82: $2: .~..8 :..: .2.: :..:
£2- .3... 8; 8.: 2.... Rm... 3“... n8... .8.... .. .8... 83.. ..: «S... 3588
82: .8.: .~..8 3.: 6...: .3: .8.: $3. 8...: :..: 8...... 8...:
3.6. 2...: 2N6 $529 a 0M..- 2... M.N.. SUN... n. ..O. 2: Ono...» :3 55.6. 3—6 :3 nowdr 2.065000 cog—0w
£98 .3... G2: 5.:
Nu... .. EN... .8... as... 8.8:. new
.2.: so... .8.: 2...: .39 .8.: .38 no: .8.: .33 ..:: 3.:
3...? o. _ .4 .8.... $3. 8.... . _ ..o. .8.... $3. _ 8.9 as... 82.. ... _ .c f: 5 .09.
.3: .~..8 :2: .8.: Ed. 82: .8.: .3: :2: and. .3: .2.:
SN... 2 2. So... 3.... 95.9 o. _ .o m :3. w _ N... .8... 8.3. 83 a _ .o. ow<
33: .2.: 82: .8.: .2.: 8...:
2: mac..- 0: wovd :3 .c... 3 OFVO 22.6. a... WVBN mag ...m.
a... : .3: .8.: .~..8 :.3. .3: .8.: 5... : and .2.: .2.: .2.:
... 8v...- eoo SON vavd Nov... ode a: 3mm... ..: hood. :3 NNW. on mwmd mNNd .3 omcdi :.. Vnw. may» N—uo.
:.:: .3: .8.: an: .3: 8...: 8...: .2.: .3: 8...:
83. ... .3... .324 :.. 2...... $3. ... 0:3 Rm... 23. an... ... 3... e8. ..-.
82: 3...: $2: .3: 3...: .2.: .8.... 3...:
$2. 8.... Ed. ..: $3 :3. ... 83 28.... ... .3... e8. 3
.2.: an: .8.: .~..8 .8.: 3...: El: .8.:
.8... ..:... 38.0 2 ...: ... £3. ... $2 to... 23 an... m.
«5.95.5 ..BO
:35 :35 :25 :25 :35 :25
«53> ...quer «53> 30.9530 «53> 23.9...er «53> 39.9....er «53> .33....er «53> 50.2....30
3.3m 0.5.... 3.25 9.3.... 305.335“.— ..om 3.3m 8:...— .330m 3......— ..35ho.m5m ...»m
3.33 .32
3......

.3an «53> ......» 3.3.3.... «532.3 556 .32....an 3 $8.... 2....
3.3.3.3.. 53$ Baotou 5.2.3.5
<n.« 0......

220

.§2§ ... a. .84 a. 822.. a 028...... ....x.... ...... .....x. .......\... 2.. 3 .5355... .8852... ... a. 8.5.3.. . 05.8.... ....ocauoxmeus; .05 .26.

b.5553 0... .0 «53.8.0 0. .030. 0... v3.3 0305.5 8.5500 :0...ao....00..m 08m ..0 008.. 00.0.3.8 0... 050.0508 3.30..“ «5.09.3. .53 3.. 0.30» +... 50303 3.. 830...; 050... 3.. on... ..00 ....50
0. 0.... 0.005 03 0505.02.34. .5050... 3.00.. 0.3.... 3.. 030» 0.... «5.00:0... 0.. 0. «5.09.0... .50 3.. E0830 00550 00.300. 3.. 0.... ..:. 50.0850. 0... ... 00.5.0... 0... 830...... 5.5.... 30.30.... ..0 5.33.

 

 

 

 

 

.8... ”838
N8 8.. 8.... 33 m3. ...... 82.5030
.3... 8.... mm... :..... m3... .... .. mm: .8..
3.... EN... .8... SN... ...... ... .... .. x 023......
$8.... .30.... .82.. .08.... .8.... .80....
N... ...... n... 2.. no. 0... um<
88.... $3.... .08.... .08.... 28.... .08....
3.... 9.... w . ... a... ...... :..... 05.860 .56
.25.... En... $2.... .22.. .32.. .03.... 2E». gnaw—ca.
S. x... 2.. M... .... 2.... a... .05 058.3 E.5
.8....
...... .32... .3... .8... .2.... .3... .8.... ...... ...... .2.... .9... .8...
as... we... 3...... : 8...... NS..- ... 0...... 2...... ...... ...... ...... :....- ... .8... .528
.8.... ...... ...... .P... .8.... .3... ..:... .2.... .3... ...“... ...... .2...
3.... .3... 3.... . ..:... .2.... 3.... .8.... 3.... an... 8.... ...... ... 5...- 5.9880 .8083
an... .....N. .2... .3... .3... .2.... ...... ...... .2.... ...... .00.... ......N.
.2.... : 8...... 08.3- 2...... .8.... ...... .3..- 2.....- .2... 3.2..- 3.....- : 0.3. ..-... 5.03.
.8.... ..:... .Nv... ...... .2.... .3... ..:... .8... .2.... .3... .8... .2...
:..... : 3m... .2... $0... .8.... .2.? .8... ...... .2.... 0...... N2... .. «...... a...
.....N. am... .2.... .8.... .8. : ......
00 mend. 0: cm... vamd‘ :0 .cad bond. :0 wwwd «39A +m.
...... .00.... .9... .2.... an... .8.... .2... .3... ...... .8.: .3... ..:.
.00 5%...- 000 how. and. .36: NOW? 000 v. ... 00 mend. 0: on... and. 000 one... ONvd. 000 03.. Each N76.
.3... .3... .8.... ..:... .3... .8.... .3... ..:... ...... ......
3...... ... $2. 8.... ... :..... . 3...... ... ..:... ...... ..:... ......v ... :... 08.. ....
.5... .....N. .8... .8... .8.... ......m. .00... .3...
0...... :.. 8m... :..... :.. can... ..:... ... :..... 3...? ... 3...... 03.0...
...... ..:... ...... .8.: ...... ...... .3... ......
2.... «.2... .2.... . E... 9...... SN... .3... : on... 08.. m.
0.58.00 :5
.330 03.30 .330 .3100 .330 .3500
«53> 50.05000 «53> 30.05000 «53> 30.0—C000 «53> 30.05000 «53> 50.05000 «53> 50.05000
.880 3.3.... .280 U:..... 2.8.35.0 =8 3.80 3:... 8.00m 0.3.... 38.55.. .....m
..0503 00—).
.8....

.330 «53> ...... 00.003...— «....00..0m :30 30....-37. ..0 300...... 0.:

“50.35.... 03?» 00.00.30 3.2.00.3

and 030,—.

221

 

 

 

88583 5 ea 53-..: 3. 82?: .382 £82 93 A::.\.m .A...:.\._ 95 a 28:35 3.2328 2 93 8829332031.: .bsnauoxmesﬁ :8 .32

$5563 05 «a wctoﬁiu o. 7.39 95 E25 Eng—um $295.3 cargo—58am can no woman v0.2.5.8 03 85.0593 canoe—om $5.85» .56 he ago» +2 E0803 3.. 8335, 2:9. 8.. on? :3 :55.
8 on: 03:. 93 8:25? .€< Are—.53 388 9:25 5.: 33» 0.8 wczoosum c: v: mc=oosuw .56 3.: bemoan voEEO 68.892 8: 8a :5 228938 05 5 vows—2: 2a 853:3 .9556 3:325 ..o 5:22

 

 

 

 

 

.84: 85cm
23 an S3 83 «2. 82 22553
3:3 «.8: 3.2 now: at; E ._ mm: 83.
83 83 a8: a. .o 33 38... .. z 8533
83.8 38.8 83.8 $8.8 28.8 88.8
3.. SN ”2 :.m a.” a3 &<
$8.8 $5.8 88.8 88.8 22.8 88.8
at: .3 «.3 5d. 2.. 2 .2 9:323 55
88.8 E38 25.8 88.8 8%.: 82.8 25. 225325
:2 9:. :8 «an a; 3: a? 93 méooﬁm 56
.§....
5.: A38 5.: 5.: 5.: am: 5.: 3:: 2:: 8:8 3:9 5.:
. 83 2:? 83. ..: 93.2 369 . £3 . 83. $3 . ~22: 25.. 2.3. ... $3 .5328
:2: at: $2: Go: 5.8 8:: E: 5.: 5.: 3...: at: 8::
28 $3. 83 : 8.... $3 :.3. . $3 $3. . :2- 23 3:: ... “8... .5688: 8:83
so: as: 83: A8: at: 3:8 25.: 5.: 5w: 3.: 5.8 80.8
_. 8:3 . SEN- 23m- .. 83m $3... 83- .. so: . 23.2- 8:; $2: 3%- 3:- f: 5%?
co: 8:: :38 A8,: 3:: 33: A3: A8: 5.: 3:: an: an:
: :.3. . 3.3 33 . Ste. 23 as: . 33. ..:.o 81.. £3 «86 v86 2.3.
3...: 53: Ex: 5.: an: 5.:
.5. avad. :5 SIN ONw._.. ..: ”2..— wbvd. .9. WMNN Emu» +2
§.: :0: 3:: :2: $2: 62: an: A2: $0: 5.: :3: an:
: coed. ... g: :3? :3. wood. ... 83 ES. ... $3 E..- : 22 91.? ..: 3.2 38» :-.:
E: 82: so: 3:: 5.: an: 8:: 39: $98 an:
a bomdu 2: m2; w. To 2:. owvd a. 03.6. ..: mad 2%.? ... wwvd $86 «an Nine Emu» @A.
:2: Gm: 398 at: 3:: an: 33: Ga:
no.6. ...: mwmd hN-d .3 —0N.o 22.? o: amvd 9:6 9:. ammd Each 9.?
2a.: 5.: no.8 5.: 5.: 5.: ES: Go:
«2.9 ..: :2 ZS . 83 $3. : 23 «2.9 .. 83 28» 3
"...—2.5m :BO
:35 :25 :25 :25 :25 :25
9:2; 30.9530 uu=o> «exam—.30 u==o> 2.9.9530 «53> 2.89580 2.25 2.03530 ”:.:; 2.0.9530
330m 89:...— ucuuom 9.3...— Eu—amcE—cm :5 Stem 8.2...— .330m 83...— EuEEAEm 20m
5:33 :02
nm4m_

:.:—cu «53> ...—t 3328.: 9502.3 5.5 sues—.32 :o 88:“... 2:.
8.3.3.5..— uua? 6893.30 b.3828

UM.” a.a-ah.

222

Table 8.4
Selectivity Corrected Wage Functions:

Summary of the Effects of Own Schooling Interacted with Young Cohort

 

 

 

 

Self Employment Public Sector Private Sector
Coefﬁcient Young Coefﬁcient Young Coefficient Young
Cohort Cohort Cohort

A. Men
Linear Own Schooling
lFLSl 0.156 ‘" -0.041 "‘ 0.050 " 0.058 0.132 ""‘ -0.039 ”*
IFLSZ 0.124 “" -0.023 0.080 ”‘ -0.068 0.103 ‘" -0.027 ‘”"
IFLS3 0.092 "" 0.017 0.102 "' -0.080 0.102 "‘ -0.024 ”‘
Non-linear Own Schooling
IF LS]
1-3 years a) 0.119 -0.077 0.347 "'" -0.486 *"
4-6 years 0.498 "‘ -0.096 0.580 "" -0.211
7-9 years 1.239 "" -O.359 0.019 0.327 0.876 “‘ -0.243
10-12 years 1.834 ‘" -0.639 " 0.225 0.845 " 1.522 "”" -0.607 ““
13+ years 2.745 "“ -0.731 0.470 “ 1.101 “" 2.468 "" -1.063 ""
IFLSZ
1-3 years a) 0.259 " -0.051 0.207 -O.169
4—6 years 0.488 ”" -0.089 0.411 '" -0.110
7-9 years 1.111 ““ -0.281 0.070 0.175 0.949 "" -0.304 "
10-12 years 1.646 "" -0.420 0.670 ”‘ -0.554 1.136 "" -0.343 ”
13+ years 2.885 “* -0.707 0.941 ‘" -0.594 1.730 ”'" -0.503 "
lFLS3
1-3 years a) 0.266 " -0.194 0.315 " -0.254
4-6 years 0.339 “‘ 0.143 0.439 ‘” -0.135
7-9 years 0.772 “" 0.009 0.488 “ -O.477 0.915 "‘ -O.449 "
10-12 years 1.344 “" -0.140 1.270 " -1.171 1.067 "" 0277
13+ years 2.235 "“ -0.478 1.768 *" -1.520 2.141 “" -0.949 ‘"
B. Women
Linear Own Schooling
lFLSl 0.084 "" -0.008 0.154 -O.156 0.132 "" -0.017
IFLSZ 0.072 "‘ -0.003 0.146 " -0.282 “" 0.128 "" -0.017
IFLS3 0.060 ”“ -0.002 -0.148 0.098 0.121 "" -0.011
Non-linear Own Schooling
lFLSl
1-3 years a) 0.115 0.0004 —0.101 0.268
4—6 years 0.548 "‘ -0.l77 0.206 0.149
7-9 years 0.880 ‘" -0.298 0.958 "" -O.309
10+ years 0.834 ““ 0.280 0.407 0.494 2.019 ‘" -0.400 ’
IFLSZ
1-3 years a) 0.273 ‘ -0.163 0.122 -0.122
4-6 years 0.390 "‘ -0.221 0.500 "" -0.334
7-9 years 0.587 ‘" 0.165 0.764 "" -0.146
10+ years 1.114 ‘" -0.262 -0.141 -0.312 1.867 ‘" -0.547 "*
lFLS3
1-3 years a) 0.209 "' 0.127 0.374 *" -0.194
4-6 years 0.261 " 0.127 0.385 ‘“" -O.105
7-9 years 0.489 "" 0.118 1.133 "’ -0.567 ‘
10+ years 0.920 "" -0.095 -0.251 -0.181 1.822 "" -0.496 "

 

Source: Based on estimates of Appendix Table 8.1A, 8.18, 8. 1C and Table 8.3A, 8.3B, 8.3C.
a) Omitted category for public sector men is 0-6 years, for women 0-9 years.

223

 

 

dogging 5 8a «37%
...... mos—a? m 633...... 9.3....— vca A3...\em ..:...x. 05 8 585.8% .momﬂzcﬂaa E 98 3.535.... 838.7. 56.588328"... 9.... .26. £55.53 05 .a wctuﬁa... 2 3.58 P... 895 23.8% 8...—8:8
.0 may» .3 noun—...: o... 9:823. :30 8.. 332:5 bug... 523 E 3.358 5.8959on Sam .8 v9.3 v3.2.6.3 2.. $5.qu3 cocoa—om .SDESQowbmsm:< m. 30.28:. .... 5.5.: 8.. 303.8 voEEO

 

 

 

 

 

.3... 858
e... RN «2.. an. «R ......N sonata...
2.... Nam... 42.. m . m... 8.... 8 . .. mm: .8..
MR... .5... 2...... ..R... SN... 2.... .. m 8.3.3
.82.. ..:.8 .2 ..8 ..:... 32.8 88.8
2.. 3.. 8.. ...... ...... ....n 3222... ... 55.2
32.8 .288 .26.... ......8 .48.... 82.8
8.. .....N .....N ....N 22 mm. .9.
88.8 .32.. 88.8 88.8 ........8 ........8
3.3 2.. o. . . N 8.3. N . ... .....8 3.8.3 :30
.088 82.8 .83.... .888 .82.. .898 9E2 25323...
3.. ..2 3... a... :.. ...N ow... ..5 3.8.3 .50
.8....
.2.... .8.... ...m... .38 .2.... .2.... .8.... .98 .2.... .38 ..m... .28
Ea- N8. :..... 8...? 3.... So... 32. an... .52”- t .8... ~02..- ... 48... .52....
an... .....8 ...m. .. .N . .... .Nn. .. .38 .m . .... .....8 ......8 ......8 .38 .38
3&8 mom... 2...... 3.... 3...... ... S... .8... 5...... .. 8.... ..: one... ..8... ... 2...... 22.880 8.583
.8.... .2.... .2.... ...N...
..:... . SN... 9...... 2...... 8.8:. as...
.28 .38 .38 .2.... .2.... ......8 5.8 ......8 .38 .. ..8 ......8 .38
e. .... m2... 8 . ... ... . ... m8... ... .... N... ... 8 . ... 3...... 2.... n. . ... 2...... .55.
......8 ......8 .38 ......8 ......8 .. .... ...... .. .....8 an... .5 . .... .~..8 .38
2...... 3...... 3...... .8... «2.... N3... . 8m... .8.... ..:... 2...... :.. o9... : 8m? .8588
......8 .8.... .38 ......8 ...... .. ......N. .R. .. .8.... .8.... ...N. .. .38 .... ..
n2... 2...... 8m... 2...... an? : new... . 8m... «4...... 3...... :.3. ... NS... .2... 55982
.~..8 ......8 .98 .2.... .....8 .8... ...... .....8 .2.... ......8 .38 .38
8.... 8...... .2.... 8.... 3...... _. .8.... . N2... .8.... 8...... .8.... ... .8... m8... .325
.30—>58:— 56 5:02
...m. .. .38 .58 .2. .. ......8 .38 .38 .3. .. .8.... an... .8... .8. ..
$4.... .2.... 8.... $2.. 3...... 8...... 3...? wow? :..... 2...... SN... ...... ...... 5%....
.3... ...... ......8 ...N. .. .....8 an... ......8 ...N. .. ...~.8 .8.... .8... ...... ..
RN... .. .... $2.. .2... .8... 8.... .8... 2.... 3...... .8.... on... 3.... ow<
......8 .2.... 5.... ...m... .2.... ......8 .38 $3.. ...... ......N. .8... .8....
Sod. ..: N26 026: #26 wood. 2: vwod e: omod. 2.... N26 wmcd .5 Chad 9 $96. :3 026 wE—occumuo Emo>
:35 :25 :25 :35 :35 :25
«53> ...vaer ”55> 2.0.3.:er ”:.:; :.quer ”53> ...quer ":.:; 2.0.2.:er unse> :.quer
..eauow 3!»...— »33m 33:.— .cthEQEH ...om .893 8.3....— ueuuow 9.3-i gage—ASH.— .3
5:83 :02
5......

22.00 95.; 5.3 .8838..— ua=eoaom :30 .32.: .... 33.5 2:.
”32.25..— oua>a v0.8.2.0 3.2.90.5
<3 9...: €5.22

224

 

 

 

.30-0500:... 0. 0... .00.. k
8. 3...:- a 8.8.2.. ......8 9a ......m ..:...x. 2.. ... €85.90 .8855... ... .5 games.- . 338.... 0.238.022... ...... .25. 0.5.5.8 2.. a 9.029.. o. ...-2.2 pa 5:... 23:30 9.32.3
..0 £00.. .3 0000.00. 0... 00:00:00 :30 .00 8.00:? .955... 00...? 0.. 00.08000 00.300680 80m :0 000-3 00.0.0200 2.. 0.5.0.0000 00,500.00 ..00E0.000>80=< m. 30.30.... ..0 5:0... .00 0.0023 000.8

 

 

 

 

 

 

 

00...... ”00.000
~00 000 000.. 000.N 000 0. 0.N 9.0003030
N000 0000 00m. ...00 5.0.0 00... 002 .000.
.000 00~0 000.0 ~0~0 00N0 00.0 ..z 028.03.
88.8 .088 .30.... ......8 .808 .01....
00.. 00.... 00.. 3.0 .N. Na. 303-5.5.0 5:02
.0 . 00. .0000. .00. 0. 800.0. 2 .00. 3000.
.....m ...... ...... 02. 0...... 03 um...
.80.... .08.... .80.... .08.... .08.... .08....
0N0... 0N0 ..00N ...~v. 00... 0000 00:00:00 :30
.00 0 0. . 0 000. .00. .0. .0000. .0000. .0000. A“...-.0. 800080....
2.. ..: ..m. .....N 2... N»... on... ...... 3:850 .50
.8.-K
.2... .008 .00.... .00... .008 ......N. .8.... ...... .2.... ...»... ...... .00....
0.0.0 0.0.0- 000.0 000... 000.0- 3.. 000.0 00~0 000.0 8.0 000.0 000.0. .3 000... .5800
.00 0. 3.0. .000. .000. .0N0. .000. .mm. .. .1... ...-0. .000. ..00. .000.
00.0. 000.0 .3. 002- .30 00.0 000.0 000.0. . N000 3.0.0. .000 000.0 .3 ~000- .=0_0E000 00000.00
:00. .0... .. .000. ..00. .000. .0. .0. .00. .. .0n0. .20. .0N0. :00. .20.
0-0. . 000.0 00.0. 00 ..0 030 0000 00....0 000.0. 000.0. 3.00 ...... 000.0. .t0<-b§§0
.00. .. ..0. .. .000. .000. .000. .000. .000. 2 0. . . .0~.0. .000. .000. .000.
._ 000.0. . 0000 00.0. .000. 00.0 00.0 0.20. 00.0 030 00.0. 0000. 000.0 30.5000.
.38 .8.... ......N. .8... .00.... .8.... ..m... .008 ......8 .3... .~..8 ......
0. ..0 ..000. 3 ~000- 0000 000.0 00.0. 02.0. 0000 000.0. ...-.0. .000. 00.0. 50.00.62
.2... .00... ...-.8 .00.... .0... .F... :..... .2.... ...-.8 ...... .2... ...N...
2N0. ~30 00.0. 0.00. 030 . 030. 020 3.0- 00.0- 3.0. 000.0. .20. 000.00
303.0... .0 500.).
.000. .20. £00. .000. ...-0.... .000. .00.... .02. .000. .3... .000. .00...
000.0~ .- .08.... 00.00- 00~.0 .20 0000. 000.0- 00.0.- 03... 00......- 0000~- .000. .00. 5.000.
...... .....N. :00. .000. 3.0. .000. .0N0. .00... .000. :0... .0... .00...
00.....0. 3 000.0 00.0 000.0. 000.0 5.0. 000.0 ..N..0 00.0. 00.0 00~0 . 00.0. 00¢.
.0. . .. ..:-.0. .080. .00.... .0 . .0. .000. :00. 30.. .. .000. .000. .3. .0 .00....
Cod. :0 an... 0: Nde. .5 ov_.o Moo-o. :0 NFC-o :- «Red: 000 med woo-O.- 000 Owe-o MNod. 000 #26 M53050”... «30>
22.00 :0000 0.0—.00 t0__00 :0000 0.0-.00
u0=0> 000.0E000 U000> «00.05000 ”53> 30.05000 uaa0> «00.05000 0000> 30.05000 "03> 30.05000
.8000 0.02.5 00.000 0.3...— .08000350 :00 00.000 0.02.; 00.000 0.30.— ...0808050 :00
00.003 002
00......

20:00 «03> 5...» 00.0000...— ua=00=00 Etc .000..— 00 0.00-cm 0.:
300.85...— 0003 08000-30 00.2.0800
...... ......- 505......

225

 

 

83.0553: 5
38 507k .8 mos—8.»- a 68365 ALQoo. v5 Aonvobn .A:Le\o_ 2.: .8 Enociwﬁ 868553: E 88 8:35? 8 329K? 565383820: :5 _o>o_ bis—5:8 95 3 95893 2 .392 98 Echo vgm
.mE—ooﬁm .3 who» .3 nous—no.— 88 3:028 :30 3.. 838.53 3.5:: 553 E 38:58 80.5350on 88m .8 v3.3 v0.8.3.8 Ea 856E000 cage—um .38—<05... m. 3035:: .3 5:9: 3.. bomoﬁo voEEO

 

 

 

 

 

885: 952
82.. ~n~ :8.~ .89~ ~88 m~9~ 825980
$88 ~82: ~$~._ 898 .888 82.. mm: 83.
828 ~8~8 $88 8: .8 828 888 .. a 8.3.3
$.98 $2.8 32.8 8898 :898 $898
~$.~ 2.. 28 one 89. ~9~ 32:25.8 5.52
38.8 $898 : _ _ .8 $898 $.98 8898
83 8~.~ a: :.v 83 -.~ o~<
8898 :898 8898 8898 32.8 3898
9%: 38 :._~ 5.2: 89. ~98“ 828.3 .50
$88 83.8 32.8 $898 $8.8 $2.8 3:3 228.25
8.. 88.8 $8 38 898 :8 :2 :3 82822 .50
:....-k
:9: $98 $9: :9: an: 39: $8.: : 2: $9: 898 $98 $9:
: 8:8 3 _.~- 292. : $892 ~38. . .8...“ . 23. 2.2 83 838 29.- ... $88 .5385
398 $9: :98 :.9: $88 $98 8.8 $98 $9: :98 898 $9:
828. 5.8 8898 . 8;..- 838 58 298 38 ~28. 3:8 2~8. : 838. 322.895 8283
$98 $9: $.98 $9: :9: $~.: $98 an: 38.: 898 898 $~.:
898 828 8:9 828 8898 8:8. 58 ~28 . ~98 $98. .. 8.8 8898. 82572582
$98 698 8c. : : _.8 :98 :38 : ..: :98 :98 :3. : 398 8~.~:
$28 8899 $98 -88 888 828- 82.8 : ~28 : 898 . .899 z 8898 : ~8~8. 3825
$98 5 .8 5.8 $98 $~. : an. : $98 $98 $98 G~. : $98 89 :
888. $88 8899 S98- ~88 8.8. $98 888. 82.8 ~28. 5.98 2.8. .2582
398 5.8 $98 :18 $9: 88.: 898 $98 :98 3~.8 :9: 38.8
:88. 8898 £98 :98 . -~8 . 828. 298. $98 $98 :98. . -~8 ... .898. :23
33:8... .8 ...-02
:.9: :3: 88.8 :.9: $9: 398 8.8 $2: $9: 5.8 898 :.98
: 828$ : 85.2- $88.89 . 2M8» 83:1 ~89? .. 8.9: . :98- E9? 32.- : .8- $898- $8 8%?
$98 39 : a9 : a9 : 5.. : 898 $8. : $9 : Gm. : $98 $98 $98
.. 88.8. . 8:8 :98 . 838. ~28 ~88 _. ~28. 2.8 828. E8 388 3.98 om<
$8.8 $~.8 $98 $98 : _ .8 $9: 5.8 A~9~: $98 89: $88 $98
:98. ... .~..8 8898 8:8. ~88. ... 8898 ... «~99 ... ~88 8:98. . ~88 :98 ... ~88 2:828:33;
:25 :25 :25 :25 :25 :25
:.:; 8389530 2.25 39.9530 :25 30.0530 uni; 30.9—=er uni; 30.9580 uaao> 30.9—=30
3.9% 85:...— hoauom 9.3.:— eathoEEm ...om ..83w 8:...— hoeuom 9.3.5 «agasmﬁm— :3
5:33 :02
85:

22.80 u==o> 5.3 18920:: 3.32-um 53° ..:-3 ..o 33:”.— 2:.
”2.2.2:...— ouaB v0.3.5.5 3.2.8.25
88 .25 «.223.

226

.momoﬁcea 5 2a 68- k 5.. 32a? Q ...2365 $9.32 vca A...o\om .93....)— 05 .u .caocEwE .momoﬁcﬂaa E 8.. 8.5.36; 3283. .bﬁummvomeuE: v5. .26. b.5868

95 ... wctoﬁio o. .958 8a ﬁche Eggsm 6895.8 cocaocmuoam 08m .8 v9.3 8...—:23 8a 8:3qu3 cocoa—ow .mE—oozom :30 8.. Ego» +2 ”5895 8.. 832.5, «88 8.. on... :8 :36 S
0.6 03:. 8a 3:252. €< Amber?» .308 232. 8.. 23% 0.8 3:85... o: a. 3:038 :39 8.. bemoan 38.8 6859: 8: 2a :5 30.89%“: 2.. E nous—2.. 2a 333:? 3:56 Burr—3:. ..o 5:02

 

 

 

.54.... H8.25m
3.. EN v3 new 2N «mm 8... .3 mm...— 8... .3 mm»... 22.96030
vow... mam... mam. mowd mom... mom. vowd owed we... 3%... owed a! ._ mm: .03—
22. m3... 086 ~26 :.3. Rod «mm... 32. w. ..c mmmd and 2 ..o «m vo.m:_.v<
32.... 28.... :3... $2.. 88.... $3.... 88.... 88.... 29.... 88.... 38.... 68....
mm. o: m~.m v... at: 3d mm... 3.3 3.9 :3. 8.0. mm. &<
88.... 88.... 88.... 88.... as... 88.... 88.... 88.... 88.... 88.... 88.... 88....
3.3 and. «.2: 3.3. wo.~ on.» and». owe. 3.2.. 3.3. 3.3 Noam. wczcosow E50
883%
and. 33.. 5.... 33. so... 82.. :02. 32.. 33.. :3. G3. 5....
2:. Vhaé ... CONS .5. $35 ...: vwvé 9 hmmd Om_._ 8:. n— ..n ..:. wmmd 5...... wand ...... wocé Q... m3.m ....- wom.v 53950
32.. $3.. .8... 62.. A5,... :0... 5.... 23.. $3.. $2.. .8... $3.
$3.... .9. 22.. _. $2. 82.. mac. 2.2. 2.3.. :56. .23 $06. 956 . cemd ..: Sod. 253580 cocoa—om
:.3. G3. 22.. $2..
2... End ...... bond 36.... 08.? 2:35 v.85
.3... 5.8 83. 32.. 2 2.. 5A. 33. Q _ . _. $2.. 5.. _. 3o. .. an. ..
Rod. .. om; : mid. God; 08.0. .1. $2.. 3 35¢. 02:. mmod. god. v2.6. god. Awe. x.nom<
8m. _. :N. _. $2.. 82.. 32.. ER. A . Na. 52.. so. _. 8w. _. 5.. _. a... _.
33. .mod. 3 02.6 owed owed .. :.:. ..: 33. N5... 83. _. Rod e 2:... :.3. «wt.
:..... _. 6mm. :.. .... an. _. at»... 83.
.2.... th._ 9.... NNm._ ..:... VOWN 2.... can; ..:. 3W- 2... QNNN munch +2
:3. an... 32. 35.2. £2. 83.. 80.2. $2.. 23.. :2... G3. Es.
:3 omw; ..:. ~86 u: we; :3 Nae; .9 $9.. :... coaé 9: _ON._ ..:. hmbd «2' mmm; .5. «ho: :... mmwd ..:... 8N; may» N76—
5... 5... .30 33.. an... 83. :2. :.:. 8...... ES. :4... so...
:5 Omwd ahmd 8: :.od :... Nahd and a: _Nh.o «an whhd » OhNd to. chg .2.. cacd o_N.o u: _¢w.c mush ouh
:.:. and a3. :.3”. $2. .8.... $3.. :2.
..:: Ohmd ... cmmd :3 hmmd ...... mmvd .8.... oavd ...: 2nd 9: G36 :... mwmd mac» 0:?
63.. an... .8.... :N. _. .8.... Ge... 32.. no...
.22. ~22. mmod ow; ocod mid 3.3. 2.2. £3» a.
«...—conga :5
380m .33 65‘..— .Soom ..ewuom ...—am ..Suow 532% ...—um .33 .33 ..:—.m—
oaa>tm 9.3.:— =om 8.2...— u=a=m =ow 85:...— xzsm =5 89:...— 33.5 ...om
35:256. ..:—.8.:— 5? don—w ou< 5.: 5.395925 ouaotoem ..:-.39.— 53 .uoam mu< 5.3 nocauEuoﬂm

 

5:33

52

 

ad":

35...me .5388— ean ou< 5.3 u=._oo..um ..BO uses—.-:oz ..: atop—m 2:.
3332::— uuag $893.30 3.2325
«:5 93a...

227

.goﬁcoan 5 8a 683* 8.. 8:.«>.& 683%.: FVXA: Ea Fuvﬁh .Auunyx; 0:. 8 Eager—Em dome—.583 5 8a 833.3. g 8283‘ 53953883: c5 .26. £555.80
95 S mar—83.0 3 $30.. 8a ﬁg @3985 338:8 caﬁoﬁooaw 08m :0 v9.3 toga—:23 8a ﬂake—boon cocoa—ow $5.858 .56 .8 B3» +3 ”co—=03 5.. 83%.? 8:8 3.. 93m :3 :95. 2
3c ova: 2: 85:3: ...v< Ampere? 369. 9.3.6 Bu E3» 0.8 9:853. 0: mm m=:oo:8 :30 .8 bemoan“. 3280 68.88 8: 2a :5 22823.. 05 E nova—us 98 «:..—out!» gnu 33235 .3 5:02

 

 

 

 

 

.34“: 52:8
Sm Eu 1 2 5* m2 1 ... at; 8n 52 $5 8m 55 823388
:.3 22. on: 8.3 83 RS 33 «$5 a... a; so... :2 mm: 83.
«Rd 3:. ~86 £3 83 v86 3 _ .c 3 _ .c 385 on _ .c 32. £2. N x was?
ASN.8 38.8 85.8 65.8 c 8.8 52: $8.8 $8.8 $2.8 88.8 38.8 82.8
3.. 3m «.2. $3 .3. 8._ 3.2 3.“ 23 own. 26 a: &<
$8.8 $3.8 88.8 88.8 38.8 88.8 88.8 88.8 88.8 $8.8 $8.8 88.8
8.2 mg no.» 8.? 8d 2 .2 2.9. can 8.: :.3 a? 3.: wéooﬁm 55

38¢
33: 5.6: 3.8 35$ was Ed 329 62: 33: 3nd and 3.3
:0 thd 9!. Ocmd :5 mmmd :5 and 95 thﬁ .2.. GOT? 9: N86 2... $05 .39 wow...“ :9 owv.v ..: ones :3. 2%.? Edam—.50
8_.: ate A8,: $93 an: Anne 5.: 3d :2: 2 _. c 5. s c as
83. ..: £3. 33. ... ”3°. :3. 83. 3:6. .. 33. .35 $3. 53. 33. 506580 8:83
35.8 328 $98 $4.: 3.8 5.: 5.8 23V 82: and 82: :9:
«86. Red. «85. 83. «85 «2.9 .. Rod. : 83 8.3. : 82.. c8... ”8.? f: 5N»?
8”: an: 5.8 an: £98 at: £93 an: 80.8 Sad 83: 5.:
”:3 83. 83 . 83 89° :3 ..: «.85 23. ~86 :... 32. 88... . 89o ow<
£3 : and $5 5. _ c 6.... : A3:
...: 2.0.— 2:. owed u!- aowé ..: mom; 9 036 9!. $9— 3+2
$33 5.8 82; 33: a3: 5.3 $93 5.: 55 A86 38.: $03
:... New; MONO. 2:. two.— .i... wmmé wcod 9.3. m36 :i awod b Nmmd ..: w_o._ 2: wNwd wMNd :5 mmod mung SA:
:93 :.3: and £93 82: :..: 5.3 :2: 5.2 eat :2: A38
9...... :56 ENG. .2.... mmhd a: mend Owed oi. coed .50 02.6 25¢ 23 wmwd 9.: and :56 ...: 356 $30» a.a.
$0: 398 and 5.. s :.3; 3a.: 3.9 33¢
3N6 u.- med 2: ommd s. 2N6 a... ommd 2.5 vad ...: Ode a: va6 Each 0.?
8:: 5.: 808 30: SN: .36 $3 and
:3. . ammo Sod £3 2 ..o ... 32 £2. :.. £3 3% 3
2583 :6

.33 .33 dam .383 .33 ...—am usuom .89“ ..:—E .388 .39“ ...EN

3:...— 235. .3 3.2... 9.3.... :.m 39:.— xzﬂ ..om 32... 33.... :8
8...:&£ .338...— 5... 8am We. 5.: 2.2.5.83 angina 32.3.... 5.: 8% gum. ..:: 3.358%

 

 

non—33

NS“:

:02

3:0...27m ..:—.8...— uaa uu< 5.3 ”...—3..“ as ..:-5.2.2 ..o 33:”.— 2:.
”zeta: «an? 38:5 £3.83
56 93:.

228

down—.883 E 93 633% 3‘. 333. m .voﬁoﬁﬁ 9.3.6— 93 A.Lo\om .Aooavﬁ; 05 «a :.3:in 68055.2. 5 0.3 3.538?» 22342 5.0533883... van .30. 5.5.58

05 E weighs—u 9 #452 2a E26 @8938 .wSmEcmo cocaoEooam 8am no 883 8223.8 0.3 356503 count—ow .wE—oosom :30 3.. £8» +2 “5:83 8.. mothg 088 5.. uni :3 :95 o.
26 035 8a ﬂcocbmzqnt‘ Aﬁcio? 888 9:95 he.“ E3» 0.8 9:858 o: a 3:028 .56 .8 bemoan 32:5 6859: 5: 8a :5 229.9%“: 95 E vows—2: 8a 333:; .9556 30.385 .3 5:32

 

 

 

 

.84.: 528m
23 :m 83 23 :m 83 23 can 93 23 «an 23 2232330
3.2. 82. 82 $3 83 «2.. 8: 39¢ 2“: .53 89° on... mm: 83.
$2 33 ES 83 83 856 at... 2 .3 «So 2:. $3 88... ,. a Big
83.8 :28 38.8 ASN.8 508 $58 88.8 88.8 63.8 38.8 28.8 32.8
Rd «E and of 83 .3 ~12 8.: mg w: m; 22 uw<
88.8 85.8 88.8 88.8 83.8 88.8 88.8 38.8 88.8 88.8 28.8 88.8
3.2 m2. “:.: :.mm .3 8.: 3% mg on: 2.3 m3. 9.: w=__8._uw .50
:34
A86 8 33 8 a: .8 50.8 :98 5.8 $3. 3 8 _ . _ 8 838 an. _ 8 898 A38
..: NO-d .2.. 03.5 2: $36 I: 086 5...... NE...— qo VMNM 9: 870 :9 wm_.m :... NONK .9... E _.O 2. 2:6 2.... hmhd E3300
A38 A88 3.: SN: and an: 58 A38 :98 3.8 A88 A88
32. : m3? 8; 83. : £3. 83 89o .. $3. :.. 83. ~88 08.8 :.. mono. EosEoou 8:83
618 :1: 5.8 8E8 $8 :3. 8 and E38 898 G _ .8 :38 £08
23 $2. 23. Rod. 83 . RS. : v8.8 3. 8:. Sod 89¢. 83.. 23. f: 5“»?
E8 5.. 8 3m. : so. : :38 A88 3&8 838 $08 $08 5.: 58
8a.? . v8.8 088 $3 :3. : 2 ..o ..: v85 ... ~23. v85 :3 23. Rod ow<
$3: :88 :.3: and: and A38
... as: ... 83 ..: Na: ... :I 3. :.3 :.. 83 2.8» +2
:38 5.8 5.8 :.3: 58 E8 83: :88 3.8 $28 and $.18
:5 _mm._ «99 oo_._ 2.... mg; 9:. :3.— __M.o :3 wad 2:. NNOA .05 Mde .... 02.— oco :od .2.. 0.0.0 0: Kb.— Eavz N72
328 an: 338 30.8 388 A88 A88 A3,: 5.8 5.8 :0: A88
... 83. 83 ... £2. 2.. .93 .38 ..: v35 ... 23 £3 ... $3 :.. 83. 3.8 :.. «So as: a-»
308 $98 $8 and :38 El; 308 $28
a. :md :5 vmvd :... 08.6 955 206 9:. ONYO 92.. Nmnd 9: movd ..:. th6 Ban» 01v
2 ”.8 5.8 5.8 5.8 5. 8 A88 3. 8 5. 8
a: 5:6 9: hmmd o: wand 00.. mzud 0 mid ..:. 3N6 0 N26 N26 Eng m-—
u._=8._um :5
380m .33 ...EM .338 .3ou ...Em .335 .893 ...-am .303 .825 ...—an.—
.EE 9.3.... ...m 3.2... 225 ...m 3.3:.— u........ .3 8.5... 9.3.... .3
3.3:qu ..:—.89— 5... 00mm ﬁx: Eggnog—.925 02.0.:me 3.5:..— .E.» 6% ou< 5.3 33359on
5:33 .32

 

mm:—

uoaotuem ..:—.82— ....n 02 5.3 u..=2=_om 53¢ hues—.-:cz .3 898,—”.— 2:.
"2.335...— ou-B 5835.30 3.281%

U; 03:.

229

Table 9.2
Selectivity Corrected Wage Functions:
Summary of The Effects of Own Schooling with Age and Potential Experience

 

Speciﬁcation with A e

 

Spec. with Potential Experience

 

 

Self Public Private Self Public Private
Emp. Sector Sector Emp. Sector Sector
A. Men
Linear Own Schooling
lFLSl 0.120 "‘ 0.092 “* 0.099 ”'" 0.126 “" 0.128 *" 0.113 ”"‘
lFLSZ 0.076 "’" 0.032 0.082 ”‘ 0.084 "* 0.039 0.099 "“"
lFLS3 0.093 “" 0.064 * 0.084 ""‘ 0.100 "‘" 0.046 0.099 "‘"
Non-linear Own Schooling
lFLSl
1-3 years a) 0.101 0.039 0.178 0.069
4-6 years 0.385 "* 0.449 ”" 0.513 "* 0.496 ‘"
7-9 years 0.891 “* 0.219 0.696 "”” 1.042 “* 0.279 * 0.778 "“
10-12 years 1.299 "" 0.852 *" 1.057 "’" 1.355 ""' 0.757 "* 1.201 "“”
13+ years 2.224 "" 1.244 "" 1.590 "" 2.364 "" 1.322 "" 1.872 ""'
IFLSZ
1-3 years a) 0.338 ""‘ 0.088 0.365 ‘“ 0.118
4-6 years 0.452 ”“ 0.320 ‘" 0.482 "" 0.356 “*
7-9 years 0.804 "" 0.031 0.701 ‘" 0.858 "" 0.045 0.776 ""
10-12 years 0.932 "" 0.238 0.828 "" 1.018 “" 0.332 "‘ 0.989 ""
13+ years 1.664 "* 0.516 * 1.365 "”" 1.809 "" 0.646 “" 1.677 "“
IFLS3
1-3 years a) 0.172 0.182 "' 0.244 " 0.188 "
4-6 years 0.427 *" 0.405 "" 0.522 "‘ 0.429 "‘
7-9 years 0.674 "* 0.255 0.660 *" 0.769 "'" 0.256 0.715 ""
10-12 years 1.077 "" 0.616 ” 0.911 *" 1.136 ‘" 0.543 "”' 1.022 ""
13+ years 1.667 "" 0.991 "‘" 1.417 "" 1.852 ""‘ 0.750 “"‘ 1.678 ”"
B. Women
Linear Own Schooling
lFLSl 0.077 "" 0.044 0.125 "" 0.089 "" 0.112 0.137 ""
[F L82 0.074 "" -0.006 0.109 “" 0.082 ‘“ 0.005 0.124 ""
IFLS3 0.059 ”" 0.052 0.113 "" 0.083 "" 0.046 0.121 ""
Non-linear Own Schooling
1F [.81
1-3 years a) 0.180 0.085 0.092 0.091
4-6 years 0.429 "" 0.357 "" 0.356 "“" 0.370 *"
7-9 years 0.721 "" 0.361 0.792 W" 0.671 “* 0.279 0.836 *"
10+ years 0.996 ‘” 1.059 " 1.692 ‘" 1.125 "" 0.961 ""' 1.830 ""
[F L82
1-3 years a) 0.186 0.052 0.236 "' 0.014
4-6 years 0.213 " 0.320 "‘" 0.256 ” 0.204
7-9 years 0.696 "* 0.056 0.768 "" 0.735 *" -0.213 0.604 ""
10+ years 0.993 “" 0.008 1.358 "'" 1.064 "* -0.203 1.462 "*
lFLS3
1-3 years a) 0.343 "" 0.365 "" 0.327 "" 0.317 ""
4-6 years 0.412 ""‘ 0.409 "* 0.454 ‘” 0.311 “
7-9 years 0.634 ”* 0.244 0.861 "" 0.716 "" 0.497 0.666 ""
10+ years 0.832 "" 0.311 1.461 "‘ 1.068 "‘ 1.160 ““ 1.331 ""

 

Source: Based on estimates of Appendix Table 9.1A, 9.13, 9.1C and Table 9.1A, 9.18, 9.1C.
a) Omitted category for public sector men is 0-6 years, for women 0-9 years.

230

 

.momoﬁcea 5 Ba .mo...k .8
mos—8,- & 62365 9.88.: vca Taveém A5388..— o... ... Enemiwﬁ .momoﬁcﬂan 5 2a 88566-. 223.5. 58.830828”... E... .26.. 555.58 05 ... mctomau o. $.30. .5 Echo 25.3% 85.853.
.8 who» .3 88599. 2.. mczoozom .56 ...... 835?» >556 so...) 5 3.85.8 905.3583 8...... .8 03...... 00.23.80 2.. 9.8.5280 cacao—om ..onEuiombmzwa< 5 35385.... 5.5.... 8.. bemoan 83.5.5

 

 

 

8:. 35cm
83. 2~ e8 8.. 2~ $8 8... .3. m9... 8... .3. 8.... 22.2880
~88 ~88 89. ~ 5... 89.. ~o~.. 89.. ~88 on... 8.8 $8... ~m . .. mm: .8..
£3 89.... a8... 22. ~38 89¢ ~m~8 88 ... .... $.98 -~.o 2 .... .. 9 823.3
38.8 .998 .898 .398 $88 .898 .898 .898 .898 .398 $28 .898
8... ~n.~ .... 8.8 .9~ $~.. 8.... 8.~ 8s 8.... m: 3.... 32:35.... 58.2
$2.8 $8.8 :88 $38 .898 .298 .898 .898 .~..8 .898 88.8 .898
a: :.m .2 a... 8.: 8... 8... ~92 on... 8... 8.: o~.. 8..
.8....
.~..8: 898 $98 .88 .8.: .28 .-.m~: $98: .8.: 898 $98 an:
:5 Mai .59 Nvmé 8: 3M6 a!- vrmf a v2.” mmmd ...: mad e09 Eva .2: med .8. 2.5.». 2:. cm”? a... N@_.v 2.8950
:98 .88 698 :98 .2.: .2.: .~..8 G98 G98 .-.8 .88 .2.:
~88 8~8 : ~88 888 82.. : 82. 89° 8.... .... .88 89.. m2... ..: 888 222880 888.8
898 698 398 .88
.. 3m... 3 .8... ~88 $98 8.8:. 955
G98 :~.~: $_.: .~..8 .8.: $9: .8.: 398 8~8 $~.: $98 :~.8
8:8 : 8m... 8.... 8.8 . 8m... .8... 82. 29.. $98 89.. 898 $98 5:5.
:9: .28 :98 .88 .~..8 .88 .88 .88 ..:.8 .88 .8.: 8.8
8.8 .88 .~..8 .28 888 m. .... .8... .28 .88 ~88 8.8 .88 .3508:
$98 .8.: 89: .88 .8.: $9: .8.: :98 5.8 .8.: :9: an:
n. .8 .28 . 82 8.8 2.8 . 8m... 8.... : 3.8 ..: 88... .2... . 888 .... 82. .8882
898 .m _ .8 .2. : : _.8 a _ .8 .8. : :98 5.8 .~..8 .898 898 $98
298 88... .8... .88 ~88 £3 2 ..o .~..8 : 8m... 2 .... ~28 .. 8m... 3880
303.8... he 5.3—2
.88 .89: 308 898 .88 $9: :9: $9: 898 .8.: 88.8 .8.:
898 . 8.... : 828 :98 ~28 ..: 8~8 $.98 . ~88 ~88 898 8.98 398 ..-... 5%....
.~..8 .88 .88 $98 .38 898 .8.: .28 .38 .8.: .8.: :9:
29.. ~88 ... m8... 3.9.. 398 :.. 33 ... 98... 89c 89.. . 88... 89.. $8... a?
3.8 .8.: 398 .8.~: .88 :98 $3: .88 :38 39.: .8.: .~..8
...: 59.6 N. _.o a... owed ...... mid 35.0 a: shod 2: M. _.o 2: mad ..:. ON—d a: oood e: NERO 9: ONTG EEOC—#950 ”80>
3.8m 3.8m .95 .898 3.8m ......m 38w 3.9.8 ...s... Snow 3.98 .8.—....
32... #3.... ...m 3.3.... 3b.... :8 3.2... 9.13.... ...m 3.3.... 9.2.... =8
unnotomum 135:5 5.3 60.5 uu.< ..:? necquuom—m 3.5.2:”.— .azu80m 5.3 .925 9? 5.3 new-9509.5

 

99.—:.3

ouuotonum 3.5....— u... ou< ..:? MES...“ :36 .32.: he 83:“.— on...

54.:

38.8.5... 8.3 .5255 £2.83

<2. 9...... 5.2.2.3.

:02

231

 

88058....— 5 Ba .8.- k .0.
2.5.2.- ... 828...... ....8. .2... .....8 2:89.. 2.. .. ...8888 $8882.... ... p... 882.5... 88.884. 8.283.222. ...a .22 8.58:8 2.. ... 885...... o. .382 8.. 28.... 28:28 82.8....
.3 ﬂag .3 woos—av. Ba $509.3. E50 3.. 8035.? .955: no.5: 5 3.85.3 5:350on 8mm ..0 :83 3.3330 93 8.5.0508 gag—om ..onEquwbmzwzt. w. 303.85.... 5.8... ..0.. 8.03.3 @8580

 

 

 

 

 

.82.... 858
8.. 88 ... ... 8.. .8 ... ... 88.. 88 89. 88.. 8~m 8.... 2858.0
888 $.88 89. 888 £98 89. 898 888 8 ... 888 888 mm . .. mm: .8..
88 8 .8 8898 88 .88 ~88 8 . 8 8 .8 .88 8.8 ~28 ~88 N z 8.28...
$2.8 : .98 :28 8.2.8 :88 :88 3.98 82.8 $898 .898 .~..8 $88
.9. .9.” 8.. 89. .88 89. 89». 89. 8.. 8.~ 8.. .9. 32:2... ... 5.5.2
:..98 :88 .898 .88 .898 $8.8 .888 :88 .898 .898 .888 .898
8.... .....~ mm... :8 m... 89. m9... 898 m9. ...8. 8... 8.... 08..
.8....
.88: .898 .898 .89... $98 $9: :98. .89.: .898: :~.8 :98 :98
a: @Nwﬁ .5. maﬁa to. omod 9.: ammé on 000.5 0 GONM or. mwhd no. @006 9:. mmhﬁ :3 —ON.V a: cnmﬁ 5.9 89? ...—52.50
:98 .8.: .~98 .88 .8.: .898 5.8 .89: :98 .88 .8.: .898
: .898 . 83.8 3.8 :.. 898 . 398 ~88 .898 . .898 8.8 ~88. 898 8.8 2.22.880 8.8208
.88 .898 .-.: .88 .898 .898 :98 .898 .88 .88 .898 .898
.. 8898 898 8.98 .. 898 898 $98 .. 29.. 8.8 .88 : 88 8.8 898 .....<.b§§.
.898 $9: :98 .898 $9: .898 .88 .28 :98 5.8 :98 .898
888 8.8 8.8 898 8.8 8.8 888 8.98 888 89.. ~88 888 $8.58:
.898 .8.: .898 .898 $.98 .88 .88 .89: .8.: $98 .898 .88.:
$88 .. -$8 $88 .88 ... .88 ~88 898 8. .8 8:8. $98 .88 E8. 588262
.898 .898 .8.: .38 .898 .8.: .898 .8.: .8.: .898 .8.: $9:
888 : 888 8:8 ~88 .. 8~8 828 888 8.8. -.8 888 ... .8 .~..8 .3290
325... ... ...-..:.
$98 .8 . .8 : ..8 .8 .. : .88 .8... : .~9 : .88 .898 .88. .88 $9:
898 8.88. 888. 898 888 . .~..8 3.98 :.. 8.8 8.98 .3. ... ..8 888 . :.8 ...8. 8.8}.
.89: .8.: :98 .88.: $8 .89: :98 .88 .898 :98 .....8 .~..8
omcd 29° :90 vwod :od .- om_.c I: 92.6 :9 Owed. N56 2:. mo—d :od. .8. god om<
.~..8 .~..8 $98 .89.: .~..8 .-.8 .898: :9: .-.8 .88: $~.: $98
...: v26 mead 9.: Nmod :.. 2:6 08.0: 5: 3.06 :5 896 256 '2' wwod a: Nwod chd I... owed mum—8:380 an>
.888 .888 2...... .898 Stow 8...... .88» Stem ......m .398 .888 2......
3.3.... 2.8.... ...m 8.2... 2...... ...m 3.3.... 2...... ...m 82...... 2...... ...m
coautomxgaaom 5.3 6on ﬂ< 5...» 95358.5 8:8..me 32.8...— a...» .8.—m ou< 5.3 sauna—8am

 

...—:33

«3.:
3:82.55.— 7—389. u... 92 a...» u-._8..um ..:O .30....— 89 83c
8.332.: own? 9899.30 3.2.00.5
....8 «...: 888......

:02

Noah

232

 

«805283 5

2a 32. k be 822,. Q .3306... $3:— cca A:¥\°m .Aacbﬁ; 05 S «sacchwﬁ .33553: cm 03 3.53.3; 838:2 .bBUmavomeBo: v5 .96— bEsEEou 05 3 9.5.3.0 9 .399 93 Eat» Eavcﬂm
.w=__oo:8.~o may» .3 noon—a9. ﬂu wE_oo:um .56 5.. 33%? aEEsu 523 5 mega”. :oznuEuonm 345 no coma: v0.2.5.3 Ba 356508 cocoa—ow 53.35:. m. 32385.8 5:9: 5.. bcwoﬁo REE—O

 

 

 

.mmH: ”088m
23 :m 83 23 :m «.8: 23 gen 20” 23 a8 $3 eoﬁtano
50° :3 «2.. 886 was NS 80c 33 mm... 80° 02... RI mm: 63.
:3 8.3 85¢ 23 $3 355 a _ .o .83 N8... 5 _ .o 23 3:. N a 33:2
83.8 32.8 88.8 38.8 $9.8 $8.8 88.8 38.8 82.8 88.8 $2.8 $1.8
m0~ N0. 13 $.N 3o 35 :3 8d 2... :2 3.. :.. 325:: :o 582
$2.8 3.2.8 388 98.8 $08 $8.8 88.8 $8.8 $8.8 88.8 38.8 28.8
3: 2.. 5m «on :.o 2.0 Fm. 5.0 an: 2:. an m: ow<
33:“
Ed: :.3: :58 3.8 $0: :88 :03: 90m: :.38 :3: 5.8 808
..:. N56 0: mwoﬁ ..:. mo_.n .2.. 3:6 9 0:..0 2.. amwd :5 1:6 :0 3nd ..:. m0—S ...... $56 8: owoﬁ .99 an6 Edam—.00
:38 80: A8,: 208 $~.: 3:8 at: 308 808 at: 5.8 and
23. 20c. . 30° $3. $3. : 23 RS .. 58. :.. 80¢. 33 5.8 2.. £28. 629880 c.2803
$0: 308 5.8 :0: 5.8 :28 8.8 5.8 E: :58 :3: 5.:
. 2.2. mm; 23.. $2. $3. no.8 ..: ~30 : 80c 33 :.. and . 83 £3 runawzaegoz
:58 :38 Am _ .8 :58 :6. c 3.8 :08 :38 5.8 3.8 $08 3 _ .8
wood : 20° «:3 v8.8 33 ”8o ... 23 33 ~85 ... 8:. 85:. 2o... .3890
5.8 so. : 5.8 :08 :08 : ..8 5.8 :08 :38 5.8 :.:8 :58
35.8 83 08.8 53. ES. :3. ”3o Sod 23. Red «8.8 .8.? h3.5....3m
:28 E. : :88 $08 Gm. : :98 9.08 :58 :.0 8 $08 :08 a0 8
08.9 . 3.3 83. one? :2. :56 «.85 83 2 ..o. v88 v85 20? §w=<
3838... be 5.32
5.8 :38 ac. : :m. : 5.8 :5. : Go. 3 :08 8:8 3N8 £08 :.:8
:3. 83 33. «V5? 08o. . SS. Sod. ... $8 83 8:8 ms? :.3. f: 5%?
Ge. : :08 :5. : :3. : 58 £8 :28 5.8 $08 :38 :28 :5. z
:3 25¢. . 08¢ we: 82. z 23 ..: 18¢ ..: 39¢. 83. Sod v85 «XE $3
50. a :2: :58 3.0: :5: 5.8 :.3: :3: GS : 60:8 5.: :.3:
2.5 _N_.c owed a: 93.0 9: m_ _.o Nmod o: owed 9!. god 356 :.. 87¢ ..:. vwod Q wood a: mood 95253.3 33>
330m .825 .QEH 3.9% .825 ...Em .895 .33“ in.“ 380m .83 .aEm
33:.— 3....... .3 3.3:.— o........ .3 3.2... 9.3.... .3 35... 33.... ...m
35.33%. image...— 5; .uomm ou< 5.3 sauna—90am 02.9.3me :.:-.80.— .E? 60.5 03 azkﬂouauﬁuoam

 

.8535

mg“:

:02

anotaxm 3.8....— E... uu< 5.: ”5.83m 55 :25 he :85. 2F
3552...“. van? 039830 5.23%
o; .3: 532:?

233

 

 

 

Table 10.1
Comparison of Estimates
SAKERNAS 1998 IFLS 1997
Filmer & Lindauer OLS estimates Selectivity corrected
(2001) w/parental schooling

Junior high 0.53 0.646 0.561
Senior high 0.82 0.880 0.670
Some tertiary 1.16 1.099 1.022
University 1.26
Observations 27,759 4,045 4,045

 

Estimates of SAKERNAS 1998 are from Table 2 of Fihner & Lindauer (2001). IF LS estimates are
average of estimates weighted by number of observations of men and women working within the
public and private sectors in each level of schooling. OLS estimates refer to estimates of speciﬁcation
with no controls for parental schooling and selectivity term

234

BIBLIOGRAPHY

Angrist, Joshua D. and Krueger, Alan B. (1992). The Effect of Age at School Entry on
Educational Attainment: An Application of Instrumental Variables with Moments
from Two Samples. Journal of the American Statistical Association: 87(418), 328-
336.

Aromolaran, Adebayo B. (2002). Private Wage Returns to Schooling in Nigeria: 1996-
1999. Discussion Paper: No. 849. Economic Growth Center, Yale University, New
Haven, Connecticut.

Baraka, Jessica L. (1999). Returns to Education in Taiwan: A Cross-section and Cohort
Analysis. Mimeo. Research Program in Development Studies, Princeton
University.

Behnnan, Jere R. and Deolalikar, Anil B. (1991). School Repetition, Dropouts, and the
Rates of Return to Schooling: The Case of Indonesia. Oxford Bulletin of
Economics and Statistics: 53(4), 467-480.

Behrrnan, Jere R. and Deolalikar, Anil B. (1993). Unobserved Household and
Community Heterogeneity and the Labor Market Impact of Schooling: A Case
Study for Indonesia. Economic Development and Cultural gauge: 41, 461-488.

Behrrnan, Jere R. and Deolalikar, Anil B. (1995). Are There Differential Returns to
Schooling by Gender? The Case of Indonesian Labour Markets. Oxford Bulletin of
Economics and Statistics: 57(1), 97-117.

Bedi, Arjun S. and Garg, Ashish (2000). The Effectiveness of Private versus Public
Schools: The Case of Indonesia. J oumiof Development Economics: 61, 463-494.

Bedi, Arjun S. and Gaston, Noel (1999). Using Variation in Schooling Availability to
Estimate Educational Returns for Honduras. Economics of Education Review: 18,
1 07—1 1 6.

Brown, Sarah and Sessions, John G. (1998). Education, Employment Status and

Earnings: A Comparative Test of the Strong Screening Hypothesis. Scottish J oumg
of Political Economy: 45(5), 586-591.

Cameron, Lisa A. (2000). Poverty and Inequality in Java: Examining the Impact of the
Changing Age, Educational and Industrial Structure. Journ_al of Development
Economics: 62, 149—180.

Card, David. (1993). Using Geographic Variation in College Proximity to Estimate the
Return to Schooling. Paper No. 4483, National Bureau of Economic Research,
Cambridge, MA,

235

Card, David (1999). The Causal Effect of Education on Earnings. In Ashenfelter, Orley
and Card, David (eds), Handbook of Labor Economics Volume 3A. Amsterdam:
Elsevier.

Clark, David H. (1983). How Secondary School Graduates Perform in the Labor
Market: A Study of Indonesia. Mimeo, The World Bank, Washington, DC.

Clark, Diana E. and Hsieh, Chang-Tai (2000). Schooling and Labor Market Impact of
the 1968 Nine-Year Education Program in Taiwan. Mimeo, Department of
Economics, Princeton University.

de Brauw, Alan and Rozelle, Scott (2002). Reconciling the Returns to Education in
Rural China. Mimeo, Department of Agricultural and Resource Economics,
University of California, Davis.

Deaton, Angus (1997). The Analysis of Household Surveys: A Microeconometric
Approach to Development Policy. Baltimore: The Johns Hopkins University Press
for The World Bank.

Deolalikar, Anil B. (1993). Gender Differences in the Returns to Schooling and in
School Enrollment Rates in Indonesia. The Journal of Human Resources: 28(4),
899-932.

Duflo, Esther (2000). Schooling and Labor Market Consequences of School
Construction in Indonesia: Evidence from an Unusual Policy Experiment. Mimeo,
Department of Economics, Massachusetts Institute of Technology.

Duraisamy, P. (2000). Changes in Returns to Education in India, 1983-94: By Gender,
Age-Cohort and Location. Discussion Paper: No. 815. Economic Growth Center,
Yale University, New Haven, Connecticut.

Filmer, Deon and Lindauer, David L. (2001). Does Indonesia Have a ‘Low Pay’ Civil
Service? Bulletin of Indonesian Economic Studies: 37(2), 189-205.

F rankenberg, Elizabeth and Karoly, Lynn (1995). The 1993 Indonesian Family Life
Survey: Overview and Field Report. Publication No. DRU-1195/1-NICHD/AID,
RAND, Santa Monica, CA.

Frankenberg, Elizabeth and Thomas, Duncan (2000). The Indonesia Family Life
Survey (IFLS): Study Design and Results from Wives 1 and 2. Publication No.
DRU-2238/Volume 1/NIA/NICHD, RAND, Santa Monica, CA.

Frankenberg, Elizabeth; Thomas, Duncan and Beegle Kathleen (1999). The Real Costs
of Indonesia’s Economic Crisis: Preliminary Findings from the Indonesia Family
Life Surveys. Wogcing Payer Series: No. 99-04, Labor and Population Program,
RAND, Santa Monica, CA.

236

Freese, Jeremy and Long, Scott J. (2000). Tests for the Multinomial Logit Model.
Stag Technical Bulletin: STE-58, Stat Corp.

Glick, Peter and Sahn, David E. (1997). Gender and Education Impacts on
Employment and Earnings in a Developing Country: The Case of Guinea.
Economic Develoment and Cultml Change: 45(4), 793—823.

Greene, William H. (1992). Econometric Analysis, 2"d edition. New York: Macmillan
Publishing Company.

Halvorsen R. and Palrnquist, R. (1980). The Interpretation of Dummy Variables in
Semi-logarithmic Equations. American Economic Review: 70, 474-475.

Harmon, Colm and Walker, Ian (1995). Estimates of the Economic Return to Schooling
for The United Kingdom. American Economic Review: 85(5), 1278-86.

Hausman, J. and McFadden, D. (1984). A Speciﬁcation Test for the Multinomial Logit
Model. Econometrica: 52, 1219-1240.

Heckman, James J. (1979). Sample Selection Bias as a Speciﬁcation Error.
Econometrica: 47(1), 153-162.

Hungerfold, Thomas and Solon, Gary (1987). Sheepskin Effects in the Returns to
Education. Review of Economics and Siitistics: 69, 175-177.

King, Elizabeth M. (1996). Education, Work and Earnings of Peruvian Women.
Economics of Education Review: 15(3), 213-230.

Lam, David and Schoeni, Robert F. (1993). Effects of Family Background on Earnings
and Returns to Schooling: Evidence from Brazil. Journal of Political Economy:
101(4), 710-740.

Lee, Lung-Fei (1983). Generalized Econometric Models with Selectivity.
Econometrica: 51(2), 507-512.

Liu, J in-Tan; Hammit, James K. and Lin, Cyongchiou J eng (2000). Family Background
and Returns to Schooling in Taiwan. Economics of Education Review: 19, 113-
125.

Maddala, G. (1983). Limited Dependent and Oualiiitive Variables in Econometrics.
New York: Cambridge University Press.

McMahon, Walter W. and Boediono (1992). Universal Basic Education: An Overall
Strategy of Investment Priorities for Economic Growth. Economics of Edlﬂtlon
Review: 11(2), 137-151.

237

Mincer, Jacob. (1974). Schooling, Experience, and Earnings. New York: Columbia
University Press for The National Bureau of Economic Research.

Mincer, Jacob (1993). Studies in Human Capital: Collected Essays of Jacob Mincer,
Vol. 1. Brookﬁeld, Vermont: Edward Elgar Publishing Ltd.

Oey-Gardinaer, Mayling (2000). Schooling in a Decentralised Indonesia: New
Approaches to Access and Decision Making. Bulletin of Indonesian Economic
Studies: 36(3), 127-134.

Psacharopoulos, George (1973). Returns to Education: An Intematiogl Comparison.
Amsterdam: Elsevier.

Psacharopoulos, George (1981). Returns to Education: An Updated International
Comparison. Comparative Education: 17(3): 321-341.

Psacharopoulos, George (1985). Returns to Education: A Further International Update
and Implications. The Journal of Human Resources: 20(4), 583-604.

Psacharopoulos, George (1994). Returns to Investment in Education: A Global Update.
World Development: 22(9), 1325-1343.

Psacharopoulos, George and Patrinos, Harry A. (2002). Returns to Investment in
Education: A Further Update. Policy Research WorkingLPapir: No. 2881. The
World Bank, Washington, DC.

Sahn, David E. and Alderman, Harold (1988). The Effects of Human Capital on
Wages, and the Determinants of Labor Supply in a Developing Country. Journal of
Development Economics: 29, 157-183.

Schafgans, Marcia M.A. (2000). Gender Wage Differences in Malaysia: Parametric
and Semiparametric Estimation. Journal of Development Economics: 63(2), 351-
378.

Schultz, T. Paul (1988). Educational Investment and Returns. In Chenery, H. and
Srinivasan, T.N. (eds), Handbook of Development Economics. Vol. 1. Amsterdam:
North Holland Publishing Company.

Schultz, T. Paul and Mwabu, Germano (1998). Wage Premia for Education and
Location, by Gender and Race in South Africa. Discussion Paper: No. 785.
Economic Growth Center, Yale University, New Haven, Connecticut.

Simanjuntak, Payaman J. (1981). The Market for Educated Labor in Indonesiaa Some
Policy Approach. Dissertation, Boston University.

238

Siphambe, Happy K. (2000). Rates of Return to Education in Botswana. Economics of
Education Review: 19(3), 291-300.

Skouﬁas, Emmanuel (1999). Parental Education and Child Nutrition in Indonesia.
Bulletin of Indonesian Economic Studies: 35(1), 99-119.

Strauss, John and Thomas, Duncan (1994). Wages, Schooling and Background:
Investments in Men and Women in Urban Brazil. In Birdsall, N.; Bruns, B. and
Sabot R., Opportunity Foregone: Education,Growth and Inequality in Brazil.
Washington, DC: The World Bank.

Strauss, John and Thomas, Duncan (1995). Human Resources: Empirical Modelling of
Household and Family Decisions. In Srinivasan, TN. and Behrrnan, Jere (eds),
Handbook of Development Economics. Vol. 3. Amsterdam: North-Holland
Publishing Company.

Strauss, John; Beegle, Kathleen; Dwiyanto, Agus; Herawati, Yulia; Pattinasarany,
Daan; Satriawan, Elan; Sikoki, Bondan; Sukamdi and Witoelar, Firman (2002).
Indonesian Living Standards Three Years After the Crisis: Evidence from the
Indonesian Family Life Survey. RAND, Santa Monica, CA and Center for
Population and Policy Studies, Gadjah Mada University, Yogyakarta.

Tansel, Aysit (1999). Public-Private Employment Choice, Wage Differentials and
Gender in Turkey. Discussion Paper: No. 797. Economic Growth Center, Yale
University, New Haven, Connecticut.

Thomas, Duncan; Beegle, Kathleen, Frankenberg, Elizabeth; Sikoki, Bondan and
Strauss, John (2001). Education in a Crisis. Mimeo, Department of Economics,
UCLA.

Murrugarra, Edmundo and Valdivia, Martin (1999). The Returns to Health for Peruvian
Urban Adults: Differentials Across Genders, the Life-Cycle and the Wage
Distribution. Mimeo, Inter-American Development Bank, Washington, DC.

Vere, James P. (2002). Returns to Schooling of Increasingly-Educated Cohorts in
Taiwan. Mimeo, School of Economics and Finance, University of Hong Kong.

239

  
  

«...-...“: . ..:s-ru-u -mun-(«00‘”Net-leeﬁfsv3m 'T'."H$-t-

..- . --.-rﬁ.-.

u ...-...-_v..--ogou_n5(iul
., ,, .

. .

     

   

11111111111111»