1
IEW Institute for Empirical Research in Economics 1lt
Quantitative Economics of Education
Winter 20102011
Instructor Prof Suumlssmuth Time 1115 ndash 1245
Location SR 7 Day Wed
MSc Economics
2
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 2lt
Outline Part I
Introduction and motivation
I Investment in Human Capital
1 Investment decision and life-cycle model Theory
2 Investment decision and life-cycle model Empirics
3 Views on education besides classical HC Theory I
4 Views on education besides classical HC Theory II
Midterm Exam
3
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 3lt
I2 Investment decision and life-cycle model Empirics
or how to validate implications from HC-Theory models
The classic approach OLS specification of HC income function
hypotheses (Mincer 1974)
4
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 4lt
Estimates for 3 different countries
obviously
ldquoMincer Equationrdquo
Source Johnes (1993)
5
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 5lt
ldquoMincer Equationrdquo
Source Mincer (1974)
original AEPs from Mincerrsquos seminal book
6
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 6lt
How to calculate the upper turning point of the income function
Differentiate with respect to years worked
In the Johnesrsquo example above it lies between 105 and 225 years that is
the peak is reached clearly before retirement age
ldquoMincer Equationrdquo
7
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 7lt
Ad hoc 2 practical remarks
- If X is not available (but S is) help yourself with X = age ndash S ndash 6
- Itrsquos all conditional estimates example university degrees
estimating returns makes only sense in case of completing
= so called Heckman-Critique (James J Heckman Nobel 2000)
ldquoMincer Equationrdquo
8
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 8lt
I21 Age-earnings profiles and Mincer-Regressions
2 principal methods to calculate rates of return on education
(a) Algebraic method
(b) Mincer-Regression
Starting point internal rate of return (IRR)
Algebraic method
9
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 9lt
Objective Calculation of ROI for educational investment of level x
Period of education starts now and takes S years
(working life ends in T years)
x = 1 eg primary education (completed typ compulsory school)
x = 2 eg secondary education (completed secondary school)
x = 3 eg tertiary education (university degree)
K-12
Algebraic method
10
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 10lt
Definition
The IRR is that value r that fulfills the following equation
where
Yx = income for an x-level of education
C = costs of education
t = time index (year)
Algebraic method
11
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 11lt
Special case
cost of education = foregone earnings (exclusively)
Algebraic method
12
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 12lt
private ROEs exclusively consider individual costs of education
social ROEs imply costs for the society as a whole
(financing education through taxes)
typically
social ROE lt private ROE
Note We can sharply discriminate private from social ROEs
Algebraic method
13
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 13lt
Central drawbacks of this method
Data intense
Information on AEPs for all levels of education = required
Query-based AEPs for individuals and small groups = often erratic
Smoothing of such AEPs = (frequently) required
ldquoTypicalrdquo (private) ROE based on algebraic calculation 10
Quantitatively returns on physical capital
Algebraic method
14
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 14lt
2-tier approach
1 Construct AEPs for specific groups of the population
2 Compute IRR on the base of these AEPs
A rare example of the use of the algebraic methods =
the works by Giora Hanoch (Jo Human Resources 1967)
Her seminal paper is entitled
ldquoAn Economic Analysis of Earnings and Schoolingrdquo
Algebraic method
15
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 15lt
Earnings by age and years of schooling (Hanoch 1967)
Whites North Whites South
Algebraic method
16
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 16lt
w
Yt = annual income for t years of education
Ct = total cost of investment in t-th year of education
rt = return
Assumption
Mincer-Regressions
A simple way to develop a testable Mincerian function
17
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 17lt
Back-tracing to period 0 (1st non-compulsory year of schooling) gives
Applying natural logs (ln) and the standard approximation
we get
Mincer-Regressions
18
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 18lt
For an identical return over all years of education this implies
Considering the impact of experience on income we finally end up w
the legendary Mincerian specification
Mincer-Regressions
19
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 19lt
Interpretation of (semi-log- or log-lin-specification)
Increasing S by 1 yr increases ln(Y) by
NB
therefore if ln(Y) increases by this = an increase of Y by
Mincer-Regressions
20
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 20lt
A slightly more sophisticated version of a Mincerian income function
Mincer-Regressions
w S1 S2 and S3 denoting dummy (01) variables for which
S1 = 1 for a level of education of at least first grade
S2 = 1 for a level of education of at least a high school grade
S3 = 1 for a level of education of at least a college degree
21
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 21lt
auf
Example
An individual who completed high school but not college is characterized by
S1 = 1 S2 = 1 and S3 = 0
Let Nx = yrs of education to attain level of education x
the private ROE on
primary education then is given by r1 = bN1
secondary education then is given by r2 = cN2
college education then is r3 = dN3
Mincer-Regressions
22
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 22lt
Advantages Mincerian vs algebraic method
less data-intense
Completely based on individual data
(instead of groupedaggregated data)
Disadvantages
Primarily suited to compute private ROEs
Completely based on individual data
(instead of groupedaggregated data)
Mincer-Regressions
23
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 23lt
We should keep 3 caveats in mindhellip
1 Education = (partially) a consumption good
- fun-factor bandwagon (business admin fad) political motivation
- eg Angrist and Krueger (JASA 1992)
run for universities during Vietnam War
9 increase in students enrolled for 1965-69
lsquoVietnam Era Draft Lotteryrsquo
1970-73 Prob(study) 5 higher for small norsquos
Mincer-Regressions
24
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 24lt
and alsohellip
2 incomplete labor markets wages may not reflect productivity
3 lsquoSignallingrsquo and lsquoScreeningrsquo Mincerian approach
Mincer-Regressions
Returns to Investment in Education in global perspective
Psacharopoulos and Patrinos (EE 2004)
25
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 25lt
ROEs w increasing level of education
ROEs w increasing level of development
private ROE gt social ROE
MincerndashRegressions
3 major styilized facts (international perspective)
26
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 26lt
MincerndashRegressions
27
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 27lt
Overall women benefit more from education (in terms of returns) but
MincerndashRegressions
28
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 28lt
Major problems of global comparative studies
Sample selection and stratification
1 Developing countries bias towards cities major companies etc
2 Civil servants soldiers etc sample etc
(wages market determined)
Methodology
questionable really same estimation strategy for all countries
MincerndashRegressions
29
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 29lt
Some further critical aspects
Control variables = problematic in MincerndashRegressions (eg occupation)
Functional form linear quadratic in parts linear log
Sheepskin effects (lsquomilestonesrsquo after 12 and 16 yrs)
Selection models
(labor market activity = endogenous decision)
MincerndashRegressions
30
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 30lt
Some sample findings (German Labor Force Survey 2004)
Experience-Income profiles (university male)
MincerndashRegressions
31
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 31lt
Some sample findings (German Labor Force Survey 2004)
Experience-Income profiles (university female)
MincerndashRegressions
32
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 32lt
Some sample findings (German Labor Force Survey 2004)
High school graduates wo further vocational training (control group)
MincerndashRegressions
33
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 33lt
Some sample findings (German Labor Force Survey 2004)
High school graduates wo further vocational training (control group)
MincerndashRegressions
34
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 34lt
Some sample findings (German Labor Force Survey 2004)
High school graduates wo further vocational training (control group)
MincerndashRegressions
35
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 35lt
Some sample findings (German Labor Force Survey 2004)
High school graduates wo further vocational training (control group)
MincerndashRegressions
2
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 2lt
Outline Part I
Introduction and motivation
I Investment in Human Capital
1 Investment decision and life-cycle model Theory
2 Investment decision and life-cycle model Empirics
3 Views on education besides classical HC Theory I
4 Views on education besides classical HC Theory II
Midterm Exam
3
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 3lt
I2 Investment decision and life-cycle model Empirics
or how to validate implications from HC-Theory models
The classic approach OLS specification of HC income function
hypotheses (Mincer 1974)
4
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 4lt
Estimates for 3 different countries
obviously
ldquoMincer Equationrdquo
Source Johnes (1993)
5
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 5lt
ldquoMincer Equationrdquo
Source Mincer (1974)
original AEPs from Mincerrsquos seminal book
6
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 6lt
How to calculate the upper turning point of the income function
Differentiate with respect to years worked
In the Johnesrsquo example above it lies between 105 and 225 years that is
the peak is reached clearly before retirement age
ldquoMincer Equationrdquo
7
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 7lt
Ad hoc 2 practical remarks
- If X is not available (but S is) help yourself with X = age ndash S ndash 6
- Itrsquos all conditional estimates example university degrees
estimating returns makes only sense in case of completing
= so called Heckman-Critique (James J Heckman Nobel 2000)
ldquoMincer Equationrdquo
8
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 8lt
I21 Age-earnings profiles and Mincer-Regressions
2 principal methods to calculate rates of return on education
(a) Algebraic method
(b) Mincer-Regression
Starting point internal rate of return (IRR)
Algebraic method
9
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 9lt
Objective Calculation of ROI for educational investment of level x
Period of education starts now and takes S years
(working life ends in T years)
x = 1 eg primary education (completed typ compulsory school)
x = 2 eg secondary education (completed secondary school)
x = 3 eg tertiary education (university degree)
K-12
Algebraic method
10
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 10lt
Definition
The IRR is that value r that fulfills the following equation
where
Yx = income for an x-level of education
C = costs of education
t = time index (year)
Algebraic method
11
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 11lt
Special case
cost of education = foregone earnings (exclusively)
Algebraic method
12
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 12lt
private ROEs exclusively consider individual costs of education
social ROEs imply costs for the society as a whole
(financing education through taxes)
typically
social ROE lt private ROE
Note We can sharply discriminate private from social ROEs
Algebraic method
13
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 13lt
Central drawbacks of this method
Data intense
Information on AEPs for all levels of education = required
Query-based AEPs for individuals and small groups = often erratic
Smoothing of such AEPs = (frequently) required
ldquoTypicalrdquo (private) ROE based on algebraic calculation 10
Quantitatively returns on physical capital
Algebraic method
14
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 14lt
2-tier approach
1 Construct AEPs for specific groups of the population
2 Compute IRR on the base of these AEPs
A rare example of the use of the algebraic methods =
the works by Giora Hanoch (Jo Human Resources 1967)
Her seminal paper is entitled
ldquoAn Economic Analysis of Earnings and Schoolingrdquo
Algebraic method
15
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 15lt
Earnings by age and years of schooling (Hanoch 1967)
Whites North Whites South
Algebraic method
16
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 16lt
w
Yt = annual income for t years of education
Ct = total cost of investment in t-th year of education
rt = return
Assumption
Mincer-Regressions
A simple way to develop a testable Mincerian function
17
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 17lt
Back-tracing to period 0 (1st non-compulsory year of schooling) gives
Applying natural logs (ln) and the standard approximation
we get
Mincer-Regressions
18
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 18lt
For an identical return over all years of education this implies
Considering the impact of experience on income we finally end up w
the legendary Mincerian specification
Mincer-Regressions
19
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 19lt
Interpretation of (semi-log- or log-lin-specification)
Increasing S by 1 yr increases ln(Y) by
NB
therefore if ln(Y) increases by this = an increase of Y by
Mincer-Regressions
20
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 20lt
A slightly more sophisticated version of a Mincerian income function
Mincer-Regressions
w S1 S2 and S3 denoting dummy (01) variables for which
S1 = 1 for a level of education of at least first grade
S2 = 1 for a level of education of at least a high school grade
S3 = 1 for a level of education of at least a college degree
21
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 21lt
auf
Example
An individual who completed high school but not college is characterized by
S1 = 1 S2 = 1 and S3 = 0
Let Nx = yrs of education to attain level of education x
the private ROE on
primary education then is given by r1 = bN1
secondary education then is given by r2 = cN2
college education then is r3 = dN3
Mincer-Regressions
22
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 22lt
Advantages Mincerian vs algebraic method
less data-intense
Completely based on individual data
(instead of groupedaggregated data)
Disadvantages
Primarily suited to compute private ROEs
Completely based on individual data
(instead of groupedaggregated data)
Mincer-Regressions
23
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 23lt
We should keep 3 caveats in mindhellip
1 Education = (partially) a consumption good
- fun-factor bandwagon (business admin fad) political motivation
- eg Angrist and Krueger (JASA 1992)
run for universities during Vietnam War
9 increase in students enrolled for 1965-69
lsquoVietnam Era Draft Lotteryrsquo
1970-73 Prob(study) 5 higher for small norsquos
Mincer-Regressions
24
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 24lt
and alsohellip
2 incomplete labor markets wages may not reflect productivity
3 lsquoSignallingrsquo and lsquoScreeningrsquo Mincerian approach
Mincer-Regressions
Returns to Investment in Education in global perspective
Psacharopoulos and Patrinos (EE 2004)
25
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 25lt
ROEs w increasing level of education
ROEs w increasing level of development
private ROE gt social ROE
MincerndashRegressions
3 major styilized facts (international perspective)
26
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 26lt
MincerndashRegressions
27
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 27lt
Overall women benefit more from education (in terms of returns) but
MincerndashRegressions
28
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 28lt
Major problems of global comparative studies
Sample selection and stratification
1 Developing countries bias towards cities major companies etc
2 Civil servants soldiers etc sample etc
(wages market determined)
Methodology
questionable really same estimation strategy for all countries
MincerndashRegressions
29
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 29lt
Some further critical aspects
Control variables = problematic in MincerndashRegressions (eg occupation)
Functional form linear quadratic in parts linear log
Sheepskin effects (lsquomilestonesrsquo after 12 and 16 yrs)
Selection models
(labor market activity = endogenous decision)
MincerndashRegressions
30
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 30lt
Some sample findings (German Labor Force Survey 2004)
Experience-Income profiles (university male)
MincerndashRegressions
31
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 31lt
Some sample findings (German Labor Force Survey 2004)
Experience-Income profiles (university female)
MincerndashRegressions
32
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 32lt
Some sample findings (German Labor Force Survey 2004)
High school graduates wo further vocational training (control group)
MincerndashRegressions
33
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 33lt
Some sample findings (German Labor Force Survey 2004)
High school graduates wo further vocational training (control group)
MincerndashRegressions
34
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 34lt
Some sample findings (German Labor Force Survey 2004)
High school graduates wo further vocational training (control group)
MincerndashRegressions
35
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 35lt
Some sample findings (German Labor Force Survey 2004)
High school graduates wo further vocational training (control group)
MincerndashRegressions
3
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 3lt
I2 Investment decision and life-cycle model Empirics
or how to validate implications from HC-Theory models
The classic approach OLS specification of HC income function
hypotheses (Mincer 1974)
4
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 4lt
Estimates for 3 different countries
obviously
ldquoMincer Equationrdquo
Source Johnes (1993)
5
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 5lt
ldquoMincer Equationrdquo
Source Mincer (1974)
original AEPs from Mincerrsquos seminal book
6
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 6lt
How to calculate the upper turning point of the income function
Differentiate with respect to years worked
In the Johnesrsquo example above it lies between 105 and 225 years that is
the peak is reached clearly before retirement age
ldquoMincer Equationrdquo
7
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 7lt
Ad hoc 2 practical remarks
- If X is not available (but S is) help yourself with X = age ndash S ndash 6
- Itrsquos all conditional estimates example university degrees
estimating returns makes only sense in case of completing
= so called Heckman-Critique (James J Heckman Nobel 2000)
ldquoMincer Equationrdquo
8
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 8lt
I21 Age-earnings profiles and Mincer-Regressions
2 principal methods to calculate rates of return on education
(a) Algebraic method
(b) Mincer-Regression
Starting point internal rate of return (IRR)
Algebraic method
9
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 9lt
Objective Calculation of ROI for educational investment of level x
Period of education starts now and takes S years
(working life ends in T years)
x = 1 eg primary education (completed typ compulsory school)
x = 2 eg secondary education (completed secondary school)
x = 3 eg tertiary education (university degree)
K-12
Algebraic method
10
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 10lt
Definition
The IRR is that value r that fulfills the following equation
where
Yx = income for an x-level of education
C = costs of education
t = time index (year)
Algebraic method
11
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 11lt
Special case
cost of education = foregone earnings (exclusively)
Algebraic method
12
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 12lt
private ROEs exclusively consider individual costs of education
social ROEs imply costs for the society as a whole
(financing education through taxes)
typically
social ROE lt private ROE
Note We can sharply discriminate private from social ROEs
Algebraic method
13
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 13lt
Central drawbacks of this method
Data intense
Information on AEPs for all levels of education = required
Query-based AEPs for individuals and small groups = often erratic
Smoothing of such AEPs = (frequently) required
ldquoTypicalrdquo (private) ROE based on algebraic calculation 10
Quantitatively returns on physical capital
Algebraic method
14
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 14lt
2-tier approach
1 Construct AEPs for specific groups of the population
2 Compute IRR on the base of these AEPs
A rare example of the use of the algebraic methods =
the works by Giora Hanoch (Jo Human Resources 1967)
Her seminal paper is entitled
ldquoAn Economic Analysis of Earnings and Schoolingrdquo
Algebraic method
15
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 15lt
Earnings by age and years of schooling (Hanoch 1967)
Whites North Whites South
Algebraic method
16
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 16lt
w
Yt = annual income for t years of education
Ct = total cost of investment in t-th year of education
rt = return
Assumption
Mincer-Regressions
A simple way to develop a testable Mincerian function
17
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 17lt
Back-tracing to period 0 (1st non-compulsory year of schooling) gives
Applying natural logs (ln) and the standard approximation
we get
Mincer-Regressions
18
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 18lt
For an identical return over all years of education this implies
Considering the impact of experience on income we finally end up w
the legendary Mincerian specification
Mincer-Regressions
19
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 19lt
Interpretation of (semi-log- or log-lin-specification)
Increasing S by 1 yr increases ln(Y) by
NB
therefore if ln(Y) increases by this = an increase of Y by
Mincer-Regressions
20
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 20lt
A slightly more sophisticated version of a Mincerian income function
Mincer-Regressions
w S1 S2 and S3 denoting dummy (01) variables for which
S1 = 1 for a level of education of at least first grade
S2 = 1 for a level of education of at least a high school grade
S3 = 1 for a level of education of at least a college degree
21
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 21lt
auf
Example
An individual who completed high school but not college is characterized by
S1 = 1 S2 = 1 and S3 = 0
Let Nx = yrs of education to attain level of education x
the private ROE on
primary education then is given by r1 = bN1
secondary education then is given by r2 = cN2
college education then is r3 = dN3
Mincer-Regressions
22
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 22lt
Advantages Mincerian vs algebraic method
less data-intense
Completely based on individual data
(instead of groupedaggregated data)
Disadvantages
Primarily suited to compute private ROEs
Completely based on individual data
(instead of groupedaggregated data)
Mincer-Regressions
23
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 23lt
We should keep 3 caveats in mindhellip
1 Education = (partially) a consumption good
- fun-factor bandwagon (business admin fad) political motivation
- eg Angrist and Krueger (JASA 1992)
run for universities during Vietnam War
9 increase in students enrolled for 1965-69
lsquoVietnam Era Draft Lotteryrsquo
1970-73 Prob(study) 5 higher for small norsquos
Mincer-Regressions
24
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 24lt
and alsohellip
2 incomplete labor markets wages may not reflect productivity
3 lsquoSignallingrsquo and lsquoScreeningrsquo Mincerian approach
Mincer-Regressions
Returns to Investment in Education in global perspective
Psacharopoulos and Patrinos (EE 2004)
25
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 25lt
ROEs w increasing level of education
ROEs w increasing level of development
private ROE gt social ROE
MincerndashRegressions
3 major styilized facts (international perspective)
26
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 26lt
MincerndashRegressions
27
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 27lt
Overall women benefit more from education (in terms of returns) but
MincerndashRegressions
28
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 28lt
Major problems of global comparative studies
Sample selection and stratification
1 Developing countries bias towards cities major companies etc
2 Civil servants soldiers etc sample etc
(wages market determined)
Methodology
questionable really same estimation strategy for all countries
MincerndashRegressions
29
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 29lt
Some further critical aspects
Control variables = problematic in MincerndashRegressions (eg occupation)
Functional form linear quadratic in parts linear log
Sheepskin effects (lsquomilestonesrsquo after 12 and 16 yrs)
Selection models
(labor market activity = endogenous decision)
MincerndashRegressions
30
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 30lt
Some sample findings (German Labor Force Survey 2004)
Experience-Income profiles (university male)
MincerndashRegressions
31
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 31lt
Some sample findings (German Labor Force Survey 2004)
Experience-Income profiles (university female)
MincerndashRegressions
32
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 32lt
Some sample findings (German Labor Force Survey 2004)
High school graduates wo further vocational training (control group)
MincerndashRegressions
33
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 33lt
Some sample findings (German Labor Force Survey 2004)
High school graduates wo further vocational training (control group)
MincerndashRegressions
34
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 34lt
Some sample findings (German Labor Force Survey 2004)
High school graduates wo further vocational training (control group)
MincerndashRegressions
35
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 35lt
Some sample findings (German Labor Force Survey 2004)
High school graduates wo further vocational training (control group)
MincerndashRegressions
4
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 4lt
Estimates for 3 different countries
obviously
ldquoMincer Equationrdquo
Source Johnes (1993)
5
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 5lt
ldquoMincer Equationrdquo
Source Mincer (1974)
original AEPs from Mincerrsquos seminal book
6
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 6lt
How to calculate the upper turning point of the income function
Differentiate with respect to years worked
In the Johnesrsquo example above it lies between 105 and 225 years that is
the peak is reached clearly before retirement age
ldquoMincer Equationrdquo
7
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 7lt
Ad hoc 2 practical remarks
- If X is not available (but S is) help yourself with X = age ndash S ndash 6
- Itrsquos all conditional estimates example university degrees
estimating returns makes only sense in case of completing
= so called Heckman-Critique (James J Heckman Nobel 2000)
ldquoMincer Equationrdquo
8
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 8lt
I21 Age-earnings profiles and Mincer-Regressions
2 principal methods to calculate rates of return on education
(a) Algebraic method
(b) Mincer-Regression
Starting point internal rate of return (IRR)
Algebraic method
9
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 9lt
Objective Calculation of ROI for educational investment of level x
Period of education starts now and takes S years
(working life ends in T years)
x = 1 eg primary education (completed typ compulsory school)
x = 2 eg secondary education (completed secondary school)
x = 3 eg tertiary education (university degree)
K-12
Algebraic method
10
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 10lt
Definition
The IRR is that value r that fulfills the following equation
where
Yx = income for an x-level of education
C = costs of education
t = time index (year)
Algebraic method
11
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 11lt
Special case
cost of education = foregone earnings (exclusively)
Algebraic method
12
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 12lt
private ROEs exclusively consider individual costs of education
social ROEs imply costs for the society as a whole
(financing education through taxes)
typically
social ROE lt private ROE
Note We can sharply discriminate private from social ROEs
Algebraic method
13
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 13lt
Central drawbacks of this method
Data intense
Information on AEPs for all levels of education = required
Query-based AEPs for individuals and small groups = often erratic
Smoothing of such AEPs = (frequently) required
ldquoTypicalrdquo (private) ROE based on algebraic calculation 10
Quantitatively returns on physical capital
Algebraic method
14
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 14lt
2-tier approach
1 Construct AEPs for specific groups of the population
2 Compute IRR on the base of these AEPs
A rare example of the use of the algebraic methods =
the works by Giora Hanoch (Jo Human Resources 1967)
Her seminal paper is entitled
ldquoAn Economic Analysis of Earnings and Schoolingrdquo
Algebraic method
15
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 15lt
Earnings by age and years of schooling (Hanoch 1967)
Whites North Whites South
Algebraic method
16
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 16lt
w
Yt = annual income for t years of education
Ct = total cost of investment in t-th year of education
rt = return
Assumption
Mincer-Regressions
A simple way to develop a testable Mincerian function
17
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 17lt
Back-tracing to period 0 (1st non-compulsory year of schooling) gives
Applying natural logs (ln) and the standard approximation
we get
Mincer-Regressions
18
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 18lt
For an identical return over all years of education this implies
Considering the impact of experience on income we finally end up w
the legendary Mincerian specification
Mincer-Regressions
19
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 19lt
Interpretation of (semi-log- or log-lin-specification)
Increasing S by 1 yr increases ln(Y) by
NB
therefore if ln(Y) increases by this = an increase of Y by
Mincer-Regressions
20
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 20lt
A slightly more sophisticated version of a Mincerian income function
Mincer-Regressions
w S1 S2 and S3 denoting dummy (01) variables for which
S1 = 1 for a level of education of at least first grade
S2 = 1 for a level of education of at least a high school grade
S3 = 1 for a level of education of at least a college degree
21
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 21lt
auf
Example
An individual who completed high school but not college is characterized by
S1 = 1 S2 = 1 and S3 = 0
Let Nx = yrs of education to attain level of education x
the private ROE on
primary education then is given by r1 = bN1
secondary education then is given by r2 = cN2
college education then is r3 = dN3
Mincer-Regressions
22
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 22lt
Advantages Mincerian vs algebraic method
less data-intense
Completely based on individual data
(instead of groupedaggregated data)
Disadvantages
Primarily suited to compute private ROEs
Completely based on individual data
(instead of groupedaggregated data)
Mincer-Regressions
23
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 23lt
We should keep 3 caveats in mindhellip
1 Education = (partially) a consumption good
- fun-factor bandwagon (business admin fad) political motivation
- eg Angrist and Krueger (JASA 1992)
run for universities during Vietnam War
9 increase in students enrolled for 1965-69
lsquoVietnam Era Draft Lotteryrsquo
1970-73 Prob(study) 5 higher for small norsquos
Mincer-Regressions
24
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 24lt
and alsohellip
2 incomplete labor markets wages may not reflect productivity
3 lsquoSignallingrsquo and lsquoScreeningrsquo Mincerian approach
Mincer-Regressions
Returns to Investment in Education in global perspective
Psacharopoulos and Patrinos (EE 2004)
25
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 25lt
ROEs w increasing level of education
ROEs w increasing level of development
private ROE gt social ROE
MincerndashRegressions
3 major styilized facts (international perspective)
26
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 26lt
MincerndashRegressions
27
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 27lt
Overall women benefit more from education (in terms of returns) but
MincerndashRegressions
28
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 28lt
Major problems of global comparative studies
Sample selection and stratification
1 Developing countries bias towards cities major companies etc
2 Civil servants soldiers etc sample etc
(wages market determined)
Methodology
questionable really same estimation strategy for all countries
MincerndashRegressions
29
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 29lt
Some further critical aspects
Control variables = problematic in MincerndashRegressions (eg occupation)
Functional form linear quadratic in parts linear log
Sheepskin effects (lsquomilestonesrsquo after 12 and 16 yrs)
Selection models
(labor market activity = endogenous decision)
MincerndashRegressions
30
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 30lt
Some sample findings (German Labor Force Survey 2004)
Experience-Income profiles (university male)
MincerndashRegressions
31
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 31lt
Some sample findings (German Labor Force Survey 2004)
Experience-Income profiles (university female)
MincerndashRegressions
32
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 32lt
Some sample findings (German Labor Force Survey 2004)
High school graduates wo further vocational training (control group)
MincerndashRegressions
33
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 33lt
Some sample findings (German Labor Force Survey 2004)
High school graduates wo further vocational training (control group)
MincerndashRegressions
34
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 34lt
Some sample findings (German Labor Force Survey 2004)
High school graduates wo further vocational training (control group)
MincerndashRegressions
35
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 35lt
Some sample findings (German Labor Force Survey 2004)
High school graduates wo further vocational training (control group)
MincerndashRegressions
5
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 5lt
ldquoMincer Equationrdquo
Source Mincer (1974)
original AEPs from Mincerrsquos seminal book
6
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 6lt
How to calculate the upper turning point of the income function
Differentiate with respect to years worked
In the Johnesrsquo example above it lies between 105 and 225 years that is
the peak is reached clearly before retirement age
ldquoMincer Equationrdquo
7
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 7lt
Ad hoc 2 practical remarks
- If X is not available (but S is) help yourself with X = age ndash S ndash 6
- Itrsquos all conditional estimates example university degrees
estimating returns makes only sense in case of completing
= so called Heckman-Critique (James J Heckman Nobel 2000)
ldquoMincer Equationrdquo
8
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 8lt
I21 Age-earnings profiles and Mincer-Regressions
2 principal methods to calculate rates of return on education
(a) Algebraic method
(b) Mincer-Regression
Starting point internal rate of return (IRR)
Algebraic method
9
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 9lt
Objective Calculation of ROI for educational investment of level x
Period of education starts now and takes S years
(working life ends in T years)
x = 1 eg primary education (completed typ compulsory school)
x = 2 eg secondary education (completed secondary school)
x = 3 eg tertiary education (university degree)
K-12
Algebraic method
10
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 10lt
Definition
The IRR is that value r that fulfills the following equation
where
Yx = income for an x-level of education
C = costs of education
t = time index (year)
Algebraic method
11
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 11lt
Special case
cost of education = foregone earnings (exclusively)
Algebraic method
12
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 12lt
private ROEs exclusively consider individual costs of education
social ROEs imply costs for the society as a whole
(financing education through taxes)
typically
social ROE lt private ROE
Note We can sharply discriminate private from social ROEs
Algebraic method
13
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 13lt
Central drawbacks of this method
Data intense
Information on AEPs for all levels of education = required
Query-based AEPs for individuals and small groups = often erratic
Smoothing of such AEPs = (frequently) required
ldquoTypicalrdquo (private) ROE based on algebraic calculation 10
Quantitatively returns on physical capital
Algebraic method
14
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 14lt
2-tier approach
1 Construct AEPs for specific groups of the population
2 Compute IRR on the base of these AEPs
A rare example of the use of the algebraic methods =
the works by Giora Hanoch (Jo Human Resources 1967)
Her seminal paper is entitled
ldquoAn Economic Analysis of Earnings and Schoolingrdquo
Algebraic method
15
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 15lt
Earnings by age and years of schooling (Hanoch 1967)
Whites North Whites South
Algebraic method
16
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 16lt
w
Yt = annual income for t years of education
Ct = total cost of investment in t-th year of education
rt = return
Assumption
Mincer-Regressions
A simple way to develop a testable Mincerian function
17
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 17lt
Back-tracing to period 0 (1st non-compulsory year of schooling) gives
Applying natural logs (ln) and the standard approximation
we get
Mincer-Regressions
18
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 18lt
For an identical return over all years of education this implies
Considering the impact of experience on income we finally end up w
the legendary Mincerian specification
Mincer-Regressions
19
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 19lt
Interpretation of (semi-log- or log-lin-specification)
Increasing S by 1 yr increases ln(Y) by
NB
therefore if ln(Y) increases by this = an increase of Y by
Mincer-Regressions
20
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 20lt
A slightly more sophisticated version of a Mincerian income function
Mincer-Regressions
w S1 S2 and S3 denoting dummy (01) variables for which
S1 = 1 for a level of education of at least first grade
S2 = 1 for a level of education of at least a high school grade
S3 = 1 for a level of education of at least a college degree
21
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 21lt
auf
Example
An individual who completed high school but not college is characterized by
S1 = 1 S2 = 1 and S3 = 0
Let Nx = yrs of education to attain level of education x
the private ROE on
primary education then is given by r1 = bN1
secondary education then is given by r2 = cN2
college education then is r3 = dN3
Mincer-Regressions
22
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 22lt
Advantages Mincerian vs algebraic method
less data-intense
Completely based on individual data
(instead of groupedaggregated data)
Disadvantages
Primarily suited to compute private ROEs
Completely based on individual data
(instead of groupedaggregated data)
Mincer-Regressions
23
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 23lt
We should keep 3 caveats in mindhellip
1 Education = (partially) a consumption good
- fun-factor bandwagon (business admin fad) political motivation
- eg Angrist and Krueger (JASA 1992)
run for universities during Vietnam War
9 increase in students enrolled for 1965-69
lsquoVietnam Era Draft Lotteryrsquo
1970-73 Prob(study) 5 higher for small norsquos
Mincer-Regressions
24
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 24lt
and alsohellip
2 incomplete labor markets wages may not reflect productivity
3 lsquoSignallingrsquo and lsquoScreeningrsquo Mincerian approach
Mincer-Regressions
Returns to Investment in Education in global perspective
Psacharopoulos and Patrinos (EE 2004)
25
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 25lt
ROEs w increasing level of education
ROEs w increasing level of development
private ROE gt social ROE
MincerndashRegressions
3 major styilized facts (international perspective)
26
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 26lt
MincerndashRegressions
27
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 27lt
Overall women benefit more from education (in terms of returns) but
MincerndashRegressions
28
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 28lt
Major problems of global comparative studies
Sample selection and stratification
1 Developing countries bias towards cities major companies etc
2 Civil servants soldiers etc sample etc
(wages market determined)
Methodology
questionable really same estimation strategy for all countries
MincerndashRegressions
29
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 29lt
Some further critical aspects
Control variables = problematic in MincerndashRegressions (eg occupation)
Functional form linear quadratic in parts linear log
Sheepskin effects (lsquomilestonesrsquo after 12 and 16 yrs)
Selection models
(labor market activity = endogenous decision)
MincerndashRegressions
30
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 30lt
Some sample findings (German Labor Force Survey 2004)
Experience-Income profiles (university male)
MincerndashRegressions
31
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 31lt
Some sample findings (German Labor Force Survey 2004)
Experience-Income profiles (university female)
MincerndashRegressions
32
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 32lt
Some sample findings (German Labor Force Survey 2004)
High school graduates wo further vocational training (control group)
MincerndashRegressions
33
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 33lt
Some sample findings (German Labor Force Survey 2004)
High school graduates wo further vocational training (control group)
MincerndashRegressions
34
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 34lt
Some sample findings (German Labor Force Survey 2004)
High school graduates wo further vocational training (control group)
MincerndashRegressions
35
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 35lt
Some sample findings (German Labor Force Survey 2004)
High school graduates wo further vocational training (control group)
MincerndashRegressions
6
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 6lt
How to calculate the upper turning point of the income function
Differentiate with respect to years worked
In the Johnesrsquo example above it lies between 105 and 225 years that is
the peak is reached clearly before retirement age
ldquoMincer Equationrdquo
7
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 7lt
Ad hoc 2 practical remarks
- If X is not available (but S is) help yourself with X = age ndash S ndash 6
- Itrsquos all conditional estimates example university degrees
estimating returns makes only sense in case of completing
= so called Heckman-Critique (James J Heckman Nobel 2000)
ldquoMincer Equationrdquo
8
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 8lt
I21 Age-earnings profiles and Mincer-Regressions
2 principal methods to calculate rates of return on education
(a) Algebraic method
(b) Mincer-Regression
Starting point internal rate of return (IRR)
Algebraic method
9
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 9lt
Objective Calculation of ROI for educational investment of level x
Period of education starts now and takes S years
(working life ends in T years)
x = 1 eg primary education (completed typ compulsory school)
x = 2 eg secondary education (completed secondary school)
x = 3 eg tertiary education (university degree)
K-12
Algebraic method
10
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 10lt
Definition
The IRR is that value r that fulfills the following equation
where
Yx = income for an x-level of education
C = costs of education
t = time index (year)
Algebraic method
11
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 11lt
Special case
cost of education = foregone earnings (exclusively)
Algebraic method
12
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 12lt
private ROEs exclusively consider individual costs of education
social ROEs imply costs for the society as a whole
(financing education through taxes)
typically
social ROE lt private ROE
Note We can sharply discriminate private from social ROEs
Algebraic method
13
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 13lt
Central drawbacks of this method
Data intense
Information on AEPs for all levels of education = required
Query-based AEPs for individuals and small groups = often erratic
Smoothing of such AEPs = (frequently) required
ldquoTypicalrdquo (private) ROE based on algebraic calculation 10
Quantitatively returns on physical capital
Algebraic method
14
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 14lt
2-tier approach
1 Construct AEPs for specific groups of the population
2 Compute IRR on the base of these AEPs
A rare example of the use of the algebraic methods =
the works by Giora Hanoch (Jo Human Resources 1967)
Her seminal paper is entitled
ldquoAn Economic Analysis of Earnings and Schoolingrdquo
Algebraic method
15
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 15lt
Earnings by age and years of schooling (Hanoch 1967)
Whites North Whites South
Algebraic method
16
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 16lt
w
Yt = annual income for t years of education
Ct = total cost of investment in t-th year of education
rt = return
Assumption
Mincer-Regressions
A simple way to develop a testable Mincerian function
17
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 17lt
Back-tracing to period 0 (1st non-compulsory year of schooling) gives
Applying natural logs (ln) and the standard approximation
we get
Mincer-Regressions
18
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 18lt
For an identical return over all years of education this implies
Considering the impact of experience on income we finally end up w
the legendary Mincerian specification
Mincer-Regressions
19
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 19lt
Interpretation of (semi-log- or log-lin-specification)
Increasing S by 1 yr increases ln(Y) by
NB
therefore if ln(Y) increases by this = an increase of Y by
Mincer-Regressions
20
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 20lt
A slightly more sophisticated version of a Mincerian income function
Mincer-Regressions
w S1 S2 and S3 denoting dummy (01) variables for which
S1 = 1 for a level of education of at least first grade
S2 = 1 for a level of education of at least a high school grade
S3 = 1 for a level of education of at least a college degree
21
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 21lt
auf
Example
An individual who completed high school but not college is characterized by
S1 = 1 S2 = 1 and S3 = 0
Let Nx = yrs of education to attain level of education x
the private ROE on
primary education then is given by r1 = bN1
secondary education then is given by r2 = cN2
college education then is r3 = dN3
Mincer-Regressions
22
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 22lt
Advantages Mincerian vs algebraic method
less data-intense
Completely based on individual data
(instead of groupedaggregated data)
Disadvantages
Primarily suited to compute private ROEs
Completely based on individual data
(instead of groupedaggregated data)
Mincer-Regressions
23
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 23lt
We should keep 3 caveats in mindhellip
1 Education = (partially) a consumption good
- fun-factor bandwagon (business admin fad) political motivation
- eg Angrist and Krueger (JASA 1992)
run for universities during Vietnam War
9 increase in students enrolled for 1965-69
lsquoVietnam Era Draft Lotteryrsquo
1970-73 Prob(study) 5 higher for small norsquos
Mincer-Regressions
24
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 24lt
and alsohellip
2 incomplete labor markets wages may not reflect productivity
3 lsquoSignallingrsquo and lsquoScreeningrsquo Mincerian approach
Mincer-Regressions
Returns to Investment in Education in global perspective
Psacharopoulos and Patrinos (EE 2004)
25
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 25lt
ROEs w increasing level of education
ROEs w increasing level of development
private ROE gt social ROE
MincerndashRegressions
3 major styilized facts (international perspective)
26
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 26lt
MincerndashRegressions
27
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 27lt
Overall women benefit more from education (in terms of returns) but
MincerndashRegressions
28
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 28lt
Major problems of global comparative studies
Sample selection and stratification
1 Developing countries bias towards cities major companies etc
2 Civil servants soldiers etc sample etc
(wages market determined)
Methodology
questionable really same estimation strategy for all countries
MincerndashRegressions
29
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 29lt
Some further critical aspects
Control variables = problematic in MincerndashRegressions (eg occupation)
Functional form linear quadratic in parts linear log
Sheepskin effects (lsquomilestonesrsquo after 12 and 16 yrs)
Selection models
(labor market activity = endogenous decision)
MincerndashRegressions
30
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 30lt
Some sample findings (German Labor Force Survey 2004)
Experience-Income profiles (university male)
MincerndashRegressions
31
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 31lt
Some sample findings (German Labor Force Survey 2004)
Experience-Income profiles (university female)
MincerndashRegressions
32
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 32lt
Some sample findings (German Labor Force Survey 2004)
High school graduates wo further vocational training (control group)
MincerndashRegressions
33
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 33lt
Some sample findings (German Labor Force Survey 2004)
High school graduates wo further vocational training (control group)
MincerndashRegressions
34
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 34lt
Some sample findings (German Labor Force Survey 2004)
High school graduates wo further vocational training (control group)
MincerndashRegressions
35
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 35lt
Some sample findings (German Labor Force Survey 2004)
High school graduates wo further vocational training (control group)
MincerndashRegressions
7
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 7lt
Ad hoc 2 practical remarks
- If X is not available (but S is) help yourself with X = age ndash S ndash 6
- Itrsquos all conditional estimates example university degrees
estimating returns makes only sense in case of completing
= so called Heckman-Critique (James J Heckman Nobel 2000)
ldquoMincer Equationrdquo
8
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 8lt
I21 Age-earnings profiles and Mincer-Regressions
2 principal methods to calculate rates of return on education
(a) Algebraic method
(b) Mincer-Regression
Starting point internal rate of return (IRR)
Algebraic method
9
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 9lt
Objective Calculation of ROI for educational investment of level x
Period of education starts now and takes S years
(working life ends in T years)
x = 1 eg primary education (completed typ compulsory school)
x = 2 eg secondary education (completed secondary school)
x = 3 eg tertiary education (university degree)
K-12
Algebraic method
10
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 10lt
Definition
The IRR is that value r that fulfills the following equation
where
Yx = income for an x-level of education
C = costs of education
t = time index (year)
Algebraic method
11
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 11lt
Special case
cost of education = foregone earnings (exclusively)
Algebraic method
12
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 12lt
private ROEs exclusively consider individual costs of education
social ROEs imply costs for the society as a whole
(financing education through taxes)
typically
social ROE lt private ROE
Note We can sharply discriminate private from social ROEs
Algebraic method
13
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 13lt
Central drawbacks of this method
Data intense
Information on AEPs for all levels of education = required
Query-based AEPs for individuals and small groups = often erratic
Smoothing of such AEPs = (frequently) required
ldquoTypicalrdquo (private) ROE based on algebraic calculation 10
Quantitatively returns on physical capital
Algebraic method
14
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 14lt
2-tier approach
1 Construct AEPs for specific groups of the population
2 Compute IRR on the base of these AEPs
A rare example of the use of the algebraic methods =
the works by Giora Hanoch (Jo Human Resources 1967)
Her seminal paper is entitled
ldquoAn Economic Analysis of Earnings and Schoolingrdquo
Algebraic method
15
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 15lt
Earnings by age and years of schooling (Hanoch 1967)
Whites North Whites South
Algebraic method
16
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 16lt
w
Yt = annual income for t years of education
Ct = total cost of investment in t-th year of education
rt = return
Assumption
Mincer-Regressions
A simple way to develop a testable Mincerian function
17
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 17lt
Back-tracing to period 0 (1st non-compulsory year of schooling) gives
Applying natural logs (ln) and the standard approximation
we get
Mincer-Regressions
18
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 18lt
For an identical return over all years of education this implies
Considering the impact of experience on income we finally end up w
the legendary Mincerian specification
Mincer-Regressions
19
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 19lt
Interpretation of (semi-log- or log-lin-specification)
Increasing S by 1 yr increases ln(Y) by
NB
therefore if ln(Y) increases by this = an increase of Y by
Mincer-Regressions
20
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 20lt
A slightly more sophisticated version of a Mincerian income function
Mincer-Regressions
w S1 S2 and S3 denoting dummy (01) variables for which
S1 = 1 for a level of education of at least first grade
S2 = 1 for a level of education of at least a high school grade
S3 = 1 for a level of education of at least a college degree
21
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 21lt
auf
Example
An individual who completed high school but not college is characterized by
S1 = 1 S2 = 1 and S3 = 0
Let Nx = yrs of education to attain level of education x
the private ROE on
primary education then is given by r1 = bN1
secondary education then is given by r2 = cN2
college education then is r3 = dN3
Mincer-Regressions
22
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 22lt
Advantages Mincerian vs algebraic method
less data-intense
Completely based on individual data
(instead of groupedaggregated data)
Disadvantages
Primarily suited to compute private ROEs
Completely based on individual data
(instead of groupedaggregated data)
Mincer-Regressions
23
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 23lt
We should keep 3 caveats in mindhellip
1 Education = (partially) a consumption good
- fun-factor bandwagon (business admin fad) political motivation
- eg Angrist and Krueger (JASA 1992)
run for universities during Vietnam War
9 increase in students enrolled for 1965-69
lsquoVietnam Era Draft Lotteryrsquo
1970-73 Prob(study) 5 higher for small norsquos
Mincer-Regressions
24
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 24lt
and alsohellip
2 incomplete labor markets wages may not reflect productivity
3 lsquoSignallingrsquo and lsquoScreeningrsquo Mincerian approach
Mincer-Regressions
Returns to Investment in Education in global perspective
Psacharopoulos and Patrinos (EE 2004)
25
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 25lt
ROEs w increasing level of education
ROEs w increasing level of development
private ROE gt social ROE
MincerndashRegressions
3 major styilized facts (international perspective)
26
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 26lt
MincerndashRegressions
27
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 27lt
Overall women benefit more from education (in terms of returns) but
MincerndashRegressions
28
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 28lt
Major problems of global comparative studies
Sample selection and stratification
1 Developing countries bias towards cities major companies etc
2 Civil servants soldiers etc sample etc
(wages market determined)
Methodology
questionable really same estimation strategy for all countries
MincerndashRegressions
29
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 29lt
Some further critical aspects
Control variables = problematic in MincerndashRegressions (eg occupation)
Functional form linear quadratic in parts linear log
Sheepskin effects (lsquomilestonesrsquo after 12 and 16 yrs)
Selection models
(labor market activity = endogenous decision)
MincerndashRegressions
30
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 30lt
Some sample findings (German Labor Force Survey 2004)
Experience-Income profiles (university male)
MincerndashRegressions
31
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 31lt
Some sample findings (German Labor Force Survey 2004)
Experience-Income profiles (university female)
MincerndashRegressions
32
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 32lt
Some sample findings (German Labor Force Survey 2004)
High school graduates wo further vocational training (control group)
MincerndashRegressions
33
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 33lt
Some sample findings (German Labor Force Survey 2004)
High school graduates wo further vocational training (control group)
MincerndashRegressions
34
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 34lt
Some sample findings (German Labor Force Survey 2004)
High school graduates wo further vocational training (control group)
MincerndashRegressions
35
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 35lt
Some sample findings (German Labor Force Survey 2004)
High school graduates wo further vocational training (control group)
MincerndashRegressions
8
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 8lt
I21 Age-earnings profiles and Mincer-Regressions
2 principal methods to calculate rates of return on education
(a) Algebraic method
(b) Mincer-Regression
Starting point internal rate of return (IRR)
Algebraic method
9
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 9lt
Objective Calculation of ROI for educational investment of level x
Period of education starts now and takes S years
(working life ends in T years)
x = 1 eg primary education (completed typ compulsory school)
x = 2 eg secondary education (completed secondary school)
x = 3 eg tertiary education (university degree)
K-12
Algebraic method
10
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 10lt
Definition
The IRR is that value r that fulfills the following equation
where
Yx = income for an x-level of education
C = costs of education
t = time index (year)
Algebraic method
11
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 11lt
Special case
cost of education = foregone earnings (exclusively)
Algebraic method
12
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 12lt
private ROEs exclusively consider individual costs of education
social ROEs imply costs for the society as a whole
(financing education through taxes)
typically
social ROE lt private ROE
Note We can sharply discriminate private from social ROEs
Algebraic method
13
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 13lt
Central drawbacks of this method
Data intense
Information on AEPs for all levels of education = required
Query-based AEPs for individuals and small groups = often erratic
Smoothing of such AEPs = (frequently) required
ldquoTypicalrdquo (private) ROE based on algebraic calculation 10
Quantitatively returns on physical capital
Algebraic method
14
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 14lt
2-tier approach
1 Construct AEPs for specific groups of the population
2 Compute IRR on the base of these AEPs
A rare example of the use of the algebraic methods =
the works by Giora Hanoch (Jo Human Resources 1967)
Her seminal paper is entitled
ldquoAn Economic Analysis of Earnings and Schoolingrdquo
Algebraic method
15
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 15lt
Earnings by age and years of schooling (Hanoch 1967)
Whites North Whites South
Algebraic method
16
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 16lt
w
Yt = annual income for t years of education
Ct = total cost of investment in t-th year of education
rt = return
Assumption
Mincer-Regressions
A simple way to develop a testable Mincerian function
17
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 17lt
Back-tracing to period 0 (1st non-compulsory year of schooling) gives
Applying natural logs (ln) and the standard approximation
we get
Mincer-Regressions
18
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 18lt
For an identical return over all years of education this implies
Considering the impact of experience on income we finally end up w
the legendary Mincerian specification
Mincer-Regressions
19
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 19lt
Interpretation of (semi-log- or log-lin-specification)
Increasing S by 1 yr increases ln(Y) by
NB
therefore if ln(Y) increases by this = an increase of Y by
Mincer-Regressions
20
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 20lt
A slightly more sophisticated version of a Mincerian income function
Mincer-Regressions
w S1 S2 and S3 denoting dummy (01) variables for which
S1 = 1 for a level of education of at least first grade
S2 = 1 for a level of education of at least a high school grade
S3 = 1 for a level of education of at least a college degree
21
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 21lt
auf
Example
An individual who completed high school but not college is characterized by
S1 = 1 S2 = 1 and S3 = 0
Let Nx = yrs of education to attain level of education x
the private ROE on
primary education then is given by r1 = bN1
secondary education then is given by r2 = cN2
college education then is r3 = dN3
Mincer-Regressions
22
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 22lt
Advantages Mincerian vs algebraic method
less data-intense
Completely based on individual data
(instead of groupedaggregated data)
Disadvantages
Primarily suited to compute private ROEs
Completely based on individual data
(instead of groupedaggregated data)
Mincer-Regressions
23
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 23lt
We should keep 3 caveats in mindhellip
1 Education = (partially) a consumption good
- fun-factor bandwagon (business admin fad) political motivation
- eg Angrist and Krueger (JASA 1992)
run for universities during Vietnam War
9 increase in students enrolled for 1965-69
lsquoVietnam Era Draft Lotteryrsquo
1970-73 Prob(study) 5 higher for small norsquos
Mincer-Regressions
24
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 24lt
and alsohellip
2 incomplete labor markets wages may not reflect productivity
3 lsquoSignallingrsquo and lsquoScreeningrsquo Mincerian approach
Mincer-Regressions
Returns to Investment in Education in global perspective
Psacharopoulos and Patrinos (EE 2004)
25
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 25lt
ROEs w increasing level of education
ROEs w increasing level of development
private ROE gt social ROE
MincerndashRegressions
3 major styilized facts (international perspective)
26
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 26lt
MincerndashRegressions
27
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 27lt
Overall women benefit more from education (in terms of returns) but
MincerndashRegressions
28
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 28lt
Major problems of global comparative studies
Sample selection and stratification
1 Developing countries bias towards cities major companies etc
2 Civil servants soldiers etc sample etc
(wages market determined)
Methodology
questionable really same estimation strategy for all countries
MincerndashRegressions
29
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 29lt
Some further critical aspects
Control variables = problematic in MincerndashRegressions (eg occupation)
Functional form linear quadratic in parts linear log
Sheepskin effects (lsquomilestonesrsquo after 12 and 16 yrs)
Selection models
(labor market activity = endogenous decision)
MincerndashRegressions
30
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 30lt
Some sample findings (German Labor Force Survey 2004)
Experience-Income profiles (university male)
MincerndashRegressions
31
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 31lt
Some sample findings (German Labor Force Survey 2004)
Experience-Income profiles (university female)
MincerndashRegressions
32
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 32lt
Some sample findings (German Labor Force Survey 2004)
High school graduates wo further vocational training (control group)
MincerndashRegressions
33
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 33lt
Some sample findings (German Labor Force Survey 2004)
High school graduates wo further vocational training (control group)
MincerndashRegressions
34
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 34lt
Some sample findings (German Labor Force Survey 2004)
High school graduates wo further vocational training (control group)
MincerndashRegressions
35
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 35lt
Some sample findings (German Labor Force Survey 2004)
High school graduates wo further vocational training (control group)
MincerndashRegressions
9
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 9lt
Objective Calculation of ROI for educational investment of level x
Period of education starts now and takes S years
(working life ends in T years)
x = 1 eg primary education (completed typ compulsory school)
x = 2 eg secondary education (completed secondary school)
x = 3 eg tertiary education (university degree)
K-12
Algebraic method
10
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 10lt
Definition
The IRR is that value r that fulfills the following equation
where
Yx = income for an x-level of education
C = costs of education
t = time index (year)
Algebraic method
11
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 11lt
Special case
cost of education = foregone earnings (exclusively)
Algebraic method
12
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 12lt
private ROEs exclusively consider individual costs of education
social ROEs imply costs for the society as a whole
(financing education through taxes)
typically
social ROE lt private ROE
Note We can sharply discriminate private from social ROEs
Algebraic method
13
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 13lt
Central drawbacks of this method
Data intense
Information on AEPs for all levels of education = required
Query-based AEPs for individuals and small groups = often erratic
Smoothing of such AEPs = (frequently) required
ldquoTypicalrdquo (private) ROE based on algebraic calculation 10
Quantitatively returns on physical capital
Algebraic method
14
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 14lt
2-tier approach
1 Construct AEPs for specific groups of the population
2 Compute IRR on the base of these AEPs
A rare example of the use of the algebraic methods =
the works by Giora Hanoch (Jo Human Resources 1967)
Her seminal paper is entitled
ldquoAn Economic Analysis of Earnings and Schoolingrdquo
Algebraic method
15
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 15lt
Earnings by age and years of schooling (Hanoch 1967)
Whites North Whites South
Algebraic method
16
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 16lt
w
Yt = annual income for t years of education
Ct = total cost of investment in t-th year of education
rt = return
Assumption
Mincer-Regressions
A simple way to develop a testable Mincerian function
17
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 17lt
Back-tracing to period 0 (1st non-compulsory year of schooling) gives
Applying natural logs (ln) and the standard approximation
we get
Mincer-Regressions
18
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 18lt
For an identical return over all years of education this implies
Considering the impact of experience on income we finally end up w
the legendary Mincerian specification
Mincer-Regressions
19
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 19lt
Interpretation of (semi-log- or log-lin-specification)
Increasing S by 1 yr increases ln(Y) by
NB
therefore if ln(Y) increases by this = an increase of Y by
Mincer-Regressions
20
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 20lt
A slightly more sophisticated version of a Mincerian income function
Mincer-Regressions
w S1 S2 and S3 denoting dummy (01) variables for which
S1 = 1 for a level of education of at least first grade
S2 = 1 for a level of education of at least a high school grade
S3 = 1 for a level of education of at least a college degree
21
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 21lt
auf
Example
An individual who completed high school but not college is characterized by
S1 = 1 S2 = 1 and S3 = 0
Let Nx = yrs of education to attain level of education x
the private ROE on
primary education then is given by r1 = bN1
secondary education then is given by r2 = cN2
college education then is r3 = dN3
Mincer-Regressions
22
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 22lt
Advantages Mincerian vs algebraic method
less data-intense
Completely based on individual data
(instead of groupedaggregated data)
Disadvantages
Primarily suited to compute private ROEs
Completely based on individual data
(instead of groupedaggregated data)
Mincer-Regressions
23
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 23lt
We should keep 3 caveats in mindhellip
1 Education = (partially) a consumption good
- fun-factor bandwagon (business admin fad) political motivation
- eg Angrist and Krueger (JASA 1992)
run for universities during Vietnam War
9 increase in students enrolled for 1965-69
lsquoVietnam Era Draft Lotteryrsquo
1970-73 Prob(study) 5 higher for small norsquos
Mincer-Regressions
24
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 24lt
and alsohellip
2 incomplete labor markets wages may not reflect productivity
3 lsquoSignallingrsquo and lsquoScreeningrsquo Mincerian approach
Mincer-Regressions
Returns to Investment in Education in global perspective
Psacharopoulos and Patrinos (EE 2004)
25
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 25lt
ROEs w increasing level of education
ROEs w increasing level of development
private ROE gt social ROE
MincerndashRegressions
3 major styilized facts (international perspective)
26
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 26lt
MincerndashRegressions
27
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 27lt
Overall women benefit more from education (in terms of returns) but
MincerndashRegressions
28
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 28lt
Major problems of global comparative studies
Sample selection and stratification
1 Developing countries bias towards cities major companies etc
2 Civil servants soldiers etc sample etc
(wages market determined)
Methodology
questionable really same estimation strategy for all countries
MincerndashRegressions
29
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 29lt
Some further critical aspects
Control variables = problematic in MincerndashRegressions (eg occupation)
Functional form linear quadratic in parts linear log
Sheepskin effects (lsquomilestonesrsquo after 12 and 16 yrs)
Selection models
(labor market activity = endogenous decision)
MincerndashRegressions
30
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 30lt
Some sample findings (German Labor Force Survey 2004)
Experience-Income profiles (university male)
MincerndashRegressions
31
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 31lt
Some sample findings (German Labor Force Survey 2004)
Experience-Income profiles (university female)
MincerndashRegressions
32
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 32lt
Some sample findings (German Labor Force Survey 2004)
High school graduates wo further vocational training (control group)
MincerndashRegressions
33
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 33lt
Some sample findings (German Labor Force Survey 2004)
High school graduates wo further vocational training (control group)
MincerndashRegressions
34
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 34lt
Some sample findings (German Labor Force Survey 2004)
High school graduates wo further vocational training (control group)
MincerndashRegressions
35
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 35lt
Some sample findings (German Labor Force Survey 2004)
High school graduates wo further vocational training (control group)
MincerndashRegressions
10
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 10lt
Definition
The IRR is that value r that fulfills the following equation
where
Yx = income for an x-level of education
C = costs of education
t = time index (year)
Algebraic method
11
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 11lt
Special case
cost of education = foregone earnings (exclusively)
Algebraic method
12
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 12lt
private ROEs exclusively consider individual costs of education
social ROEs imply costs for the society as a whole
(financing education through taxes)
typically
social ROE lt private ROE
Note We can sharply discriminate private from social ROEs
Algebraic method
13
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 13lt
Central drawbacks of this method
Data intense
Information on AEPs for all levels of education = required
Query-based AEPs for individuals and small groups = often erratic
Smoothing of such AEPs = (frequently) required
ldquoTypicalrdquo (private) ROE based on algebraic calculation 10
Quantitatively returns on physical capital
Algebraic method
14
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 14lt
2-tier approach
1 Construct AEPs for specific groups of the population
2 Compute IRR on the base of these AEPs
A rare example of the use of the algebraic methods =
the works by Giora Hanoch (Jo Human Resources 1967)
Her seminal paper is entitled
ldquoAn Economic Analysis of Earnings and Schoolingrdquo
Algebraic method
15
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 15lt
Earnings by age and years of schooling (Hanoch 1967)
Whites North Whites South
Algebraic method
16
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 16lt
w
Yt = annual income for t years of education
Ct = total cost of investment in t-th year of education
rt = return
Assumption
Mincer-Regressions
A simple way to develop a testable Mincerian function
17
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 17lt
Back-tracing to period 0 (1st non-compulsory year of schooling) gives
Applying natural logs (ln) and the standard approximation
we get
Mincer-Regressions
18
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 18lt
For an identical return over all years of education this implies
Considering the impact of experience on income we finally end up w
the legendary Mincerian specification
Mincer-Regressions
19
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 19lt
Interpretation of (semi-log- or log-lin-specification)
Increasing S by 1 yr increases ln(Y) by
NB
therefore if ln(Y) increases by this = an increase of Y by
Mincer-Regressions
20
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 20lt
A slightly more sophisticated version of a Mincerian income function
Mincer-Regressions
w S1 S2 and S3 denoting dummy (01) variables for which
S1 = 1 for a level of education of at least first grade
S2 = 1 for a level of education of at least a high school grade
S3 = 1 for a level of education of at least a college degree
21
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 21lt
auf
Example
An individual who completed high school but not college is characterized by
S1 = 1 S2 = 1 and S3 = 0
Let Nx = yrs of education to attain level of education x
the private ROE on
primary education then is given by r1 = bN1
secondary education then is given by r2 = cN2
college education then is r3 = dN3
Mincer-Regressions
22
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 22lt
Advantages Mincerian vs algebraic method
less data-intense
Completely based on individual data
(instead of groupedaggregated data)
Disadvantages
Primarily suited to compute private ROEs
Completely based on individual data
(instead of groupedaggregated data)
Mincer-Regressions
23
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 23lt
We should keep 3 caveats in mindhellip
1 Education = (partially) a consumption good
- fun-factor bandwagon (business admin fad) political motivation
- eg Angrist and Krueger (JASA 1992)
run for universities during Vietnam War
9 increase in students enrolled for 1965-69
lsquoVietnam Era Draft Lotteryrsquo
1970-73 Prob(study) 5 higher for small norsquos
Mincer-Regressions
24
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 24lt
and alsohellip
2 incomplete labor markets wages may not reflect productivity
3 lsquoSignallingrsquo and lsquoScreeningrsquo Mincerian approach
Mincer-Regressions
Returns to Investment in Education in global perspective
Psacharopoulos and Patrinos (EE 2004)
25
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 25lt
ROEs w increasing level of education
ROEs w increasing level of development
private ROE gt social ROE
MincerndashRegressions
3 major styilized facts (international perspective)
26
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 26lt
MincerndashRegressions
27
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 27lt
Overall women benefit more from education (in terms of returns) but
MincerndashRegressions
28
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 28lt
Major problems of global comparative studies
Sample selection and stratification
1 Developing countries bias towards cities major companies etc
2 Civil servants soldiers etc sample etc
(wages market determined)
Methodology
questionable really same estimation strategy for all countries
MincerndashRegressions
29
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 29lt
Some further critical aspects
Control variables = problematic in MincerndashRegressions (eg occupation)
Functional form linear quadratic in parts linear log
Sheepskin effects (lsquomilestonesrsquo after 12 and 16 yrs)
Selection models
(labor market activity = endogenous decision)
MincerndashRegressions
30
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 30lt
Some sample findings (German Labor Force Survey 2004)
Experience-Income profiles (university male)
MincerndashRegressions
31
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 31lt
Some sample findings (German Labor Force Survey 2004)
Experience-Income profiles (university female)
MincerndashRegressions
32
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 32lt
Some sample findings (German Labor Force Survey 2004)
High school graduates wo further vocational training (control group)
MincerndashRegressions
33
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 33lt
Some sample findings (German Labor Force Survey 2004)
High school graduates wo further vocational training (control group)
MincerndashRegressions
34
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 34lt
Some sample findings (German Labor Force Survey 2004)
High school graduates wo further vocational training (control group)
MincerndashRegressions
35
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 35lt
Some sample findings (German Labor Force Survey 2004)
High school graduates wo further vocational training (control group)
MincerndashRegressions
11
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 11lt
Special case
cost of education = foregone earnings (exclusively)
Algebraic method
12
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 12lt
private ROEs exclusively consider individual costs of education
social ROEs imply costs for the society as a whole
(financing education through taxes)
typically
social ROE lt private ROE
Note We can sharply discriminate private from social ROEs
Algebraic method
13
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 13lt
Central drawbacks of this method
Data intense
Information on AEPs for all levels of education = required
Query-based AEPs for individuals and small groups = often erratic
Smoothing of such AEPs = (frequently) required
ldquoTypicalrdquo (private) ROE based on algebraic calculation 10
Quantitatively returns on physical capital
Algebraic method
14
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 14lt
2-tier approach
1 Construct AEPs for specific groups of the population
2 Compute IRR on the base of these AEPs
A rare example of the use of the algebraic methods =
the works by Giora Hanoch (Jo Human Resources 1967)
Her seminal paper is entitled
ldquoAn Economic Analysis of Earnings and Schoolingrdquo
Algebraic method
15
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 15lt
Earnings by age and years of schooling (Hanoch 1967)
Whites North Whites South
Algebraic method
16
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 16lt
w
Yt = annual income for t years of education
Ct = total cost of investment in t-th year of education
rt = return
Assumption
Mincer-Regressions
A simple way to develop a testable Mincerian function
17
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 17lt
Back-tracing to period 0 (1st non-compulsory year of schooling) gives
Applying natural logs (ln) and the standard approximation
we get
Mincer-Regressions
18
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 18lt
For an identical return over all years of education this implies
Considering the impact of experience on income we finally end up w
the legendary Mincerian specification
Mincer-Regressions
19
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 19lt
Interpretation of (semi-log- or log-lin-specification)
Increasing S by 1 yr increases ln(Y) by
NB
therefore if ln(Y) increases by this = an increase of Y by
Mincer-Regressions
20
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 20lt
A slightly more sophisticated version of a Mincerian income function
Mincer-Regressions
w S1 S2 and S3 denoting dummy (01) variables for which
S1 = 1 for a level of education of at least first grade
S2 = 1 for a level of education of at least a high school grade
S3 = 1 for a level of education of at least a college degree
21
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 21lt
auf
Example
An individual who completed high school but not college is characterized by
S1 = 1 S2 = 1 and S3 = 0
Let Nx = yrs of education to attain level of education x
the private ROE on
primary education then is given by r1 = bN1
secondary education then is given by r2 = cN2
college education then is r3 = dN3
Mincer-Regressions
22
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 22lt
Advantages Mincerian vs algebraic method
less data-intense
Completely based on individual data
(instead of groupedaggregated data)
Disadvantages
Primarily suited to compute private ROEs
Completely based on individual data
(instead of groupedaggregated data)
Mincer-Regressions
23
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 23lt
We should keep 3 caveats in mindhellip
1 Education = (partially) a consumption good
- fun-factor bandwagon (business admin fad) political motivation
- eg Angrist and Krueger (JASA 1992)
run for universities during Vietnam War
9 increase in students enrolled for 1965-69
lsquoVietnam Era Draft Lotteryrsquo
1970-73 Prob(study) 5 higher for small norsquos
Mincer-Regressions
24
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 24lt
and alsohellip
2 incomplete labor markets wages may not reflect productivity
3 lsquoSignallingrsquo and lsquoScreeningrsquo Mincerian approach
Mincer-Regressions
Returns to Investment in Education in global perspective
Psacharopoulos and Patrinos (EE 2004)
25
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 25lt
ROEs w increasing level of education
ROEs w increasing level of development
private ROE gt social ROE
MincerndashRegressions
3 major styilized facts (international perspective)
26
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 26lt
MincerndashRegressions
27
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 27lt
Overall women benefit more from education (in terms of returns) but
MincerndashRegressions
28
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 28lt
Major problems of global comparative studies
Sample selection and stratification
1 Developing countries bias towards cities major companies etc
2 Civil servants soldiers etc sample etc
(wages market determined)
Methodology
questionable really same estimation strategy for all countries
MincerndashRegressions
29
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 29lt
Some further critical aspects
Control variables = problematic in MincerndashRegressions (eg occupation)
Functional form linear quadratic in parts linear log
Sheepskin effects (lsquomilestonesrsquo after 12 and 16 yrs)
Selection models
(labor market activity = endogenous decision)
MincerndashRegressions
30
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 30lt
Some sample findings (German Labor Force Survey 2004)
Experience-Income profiles (university male)
MincerndashRegressions
31
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 31lt
Some sample findings (German Labor Force Survey 2004)
Experience-Income profiles (university female)
MincerndashRegressions
32
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 32lt
Some sample findings (German Labor Force Survey 2004)
High school graduates wo further vocational training (control group)
MincerndashRegressions
33
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 33lt
Some sample findings (German Labor Force Survey 2004)
High school graduates wo further vocational training (control group)
MincerndashRegressions
34
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 34lt
Some sample findings (German Labor Force Survey 2004)
High school graduates wo further vocational training (control group)
MincerndashRegressions
35
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 35lt
Some sample findings (German Labor Force Survey 2004)
High school graduates wo further vocational training (control group)
MincerndashRegressions
12
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 12lt
private ROEs exclusively consider individual costs of education
social ROEs imply costs for the society as a whole
(financing education through taxes)
typically
social ROE lt private ROE
Note We can sharply discriminate private from social ROEs
Algebraic method
13
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 13lt
Central drawbacks of this method
Data intense
Information on AEPs for all levels of education = required
Query-based AEPs for individuals and small groups = often erratic
Smoothing of such AEPs = (frequently) required
ldquoTypicalrdquo (private) ROE based on algebraic calculation 10
Quantitatively returns on physical capital
Algebraic method
14
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 14lt
2-tier approach
1 Construct AEPs for specific groups of the population
2 Compute IRR on the base of these AEPs
A rare example of the use of the algebraic methods =
the works by Giora Hanoch (Jo Human Resources 1967)
Her seminal paper is entitled
ldquoAn Economic Analysis of Earnings and Schoolingrdquo
Algebraic method
15
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 15lt
Earnings by age and years of schooling (Hanoch 1967)
Whites North Whites South
Algebraic method
16
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 16lt
w
Yt = annual income for t years of education
Ct = total cost of investment in t-th year of education
rt = return
Assumption
Mincer-Regressions
A simple way to develop a testable Mincerian function
17
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 17lt
Back-tracing to period 0 (1st non-compulsory year of schooling) gives
Applying natural logs (ln) and the standard approximation
we get
Mincer-Regressions
18
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 18lt
For an identical return over all years of education this implies
Considering the impact of experience on income we finally end up w
the legendary Mincerian specification
Mincer-Regressions
19
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 19lt
Interpretation of (semi-log- or log-lin-specification)
Increasing S by 1 yr increases ln(Y) by
NB
therefore if ln(Y) increases by this = an increase of Y by
Mincer-Regressions
20
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 20lt
A slightly more sophisticated version of a Mincerian income function
Mincer-Regressions
w S1 S2 and S3 denoting dummy (01) variables for which
S1 = 1 for a level of education of at least first grade
S2 = 1 for a level of education of at least a high school grade
S3 = 1 for a level of education of at least a college degree
21
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 21lt
auf
Example
An individual who completed high school but not college is characterized by
S1 = 1 S2 = 1 and S3 = 0
Let Nx = yrs of education to attain level of education x
the private ROE on
primary education then is given by r1 = bN1
secondary education then is given by r2 = cN2
college education then is r3 = dN3
Mincer-Regressions
22
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 22lt
Advantages Mincerian vs algebraic method
less data-intense
Completely based on individual data
(instead of groupedaggregated data)
Disadvantages
Primarily suited to compute private ROEs
Completely based on individual data
(instead of groupedaggregated data)
Mincer-Regressions
23
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 23lt
We should keep 3 caveats in mindhellip
1 Education = (partially) a consumption good
- fun-factor bandwagon (business admin fad) political motivation
- eg Angrist and Krueger (JASA 1992)
run for universities during Vietnam War
9 increase in students enrolled for 1965-69
lsquoVietnam Era Draft Lotteryrsquo
1970-73 Prob(study) 5 higher for small norsquos
Mincer-Regressions
24
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 24lt
and alsohellip
2 incomplete labor markets wages may not reflect productivity
3 lsquoSignallingrsquo and lsquoScreeningrsquo Mincerian approach
Mincer-Regressions
Returns to Investment in Education in global perspective
Psacharopoulos and Patrinos (EE 2004)
25
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 25lt
ROEs w increasing level of education
ROEs w increasing level of development
private ROE gt social ROE
MincerndashRegressions
3 major styilized facts (international perspective)
26
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 26lt
MincerndashRegressions
27
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 27lt
Overall women benefit more from education (in terms of returns) but
MincerndashRegressions
28
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 28lt
Major problems of global comparative studies
Sample selection and stratification
1 Developing countries bias towards cities major companies etc
2 Civil servants soldiers etc sample etc
(wages market determined)
Methodology
questionable really same estimation strategy for all countries
MincerndashRegressions
29
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 29lt
Some further critical aspects
Control variables = problematic in MincerndashRegressions (eg occupation)
Functional form linear quadratic in parts linear log
Sheepskin effects (lsquomilestonesrsquo after 12 and 16 yrs)
Selection models
(labor market activity = endogenous decision)
MincerndashRegressions
30
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 30lt
Some sample findings (German Labor Force Survey 2004)
Experience-Income profiles (university male)
MincerndashRegressions
31
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 31lt
Some sample findings (German Labor Force Survey 2004)
Experience-Income profiles (university female)
MincerndashRegressions
32
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 32lt
Some sample findings (German Labor Force Survey 2004)
High school graduates wo further vocational training (control group)
MincerndashRegressions
33
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 33lt
Some sample findings (German Labor Force Survey 2004)
High school graduates wo further vocational training (control group)
MincerndashRegressions
34
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 34lt
Some sample findings (German Labor Force Survey 2004)
High school graduates wo further vocational training (control group)
MincerndashRegressions
35
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 35lt
Some sample findings (German Labor Force Survey 2004)
High school graduates wo further vocational training (control group)
MincerndashRegressions
13
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 13lt
Central drawbacks of this method
Data intense
Information on AEPs for all levels of education = required
Query-based AEPs for individuals and small groups = often erratic
Smoothing of such AEPs = (frequently) required
ldquoTypicalrdquo (private) ROE based on algebraic calculation 10
Quantitatively returns on physical capital
Algebraic method
14
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 14lt
2-tier approach
1 Construct AEPs for specific groups of the population
2 Compute IRR on the base of these AEPs
A rare example of the use of the algebraic methods =
the works by Giora Hanoch (Jo Human Resources 1967)
Her seminal paper is entitled
ldquoAn Economic Analysis of Earnings and Schoolingrdquo
Algebraic method
15
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 15lt
Earnings by age and years of schooling (Hanoch 1967)
Whites North Whites South
Algebraic method
16
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 16lt
w
Yt = annual income for t years of education
Ct = total cost of investment in t-th year of education
rt = return
Assumption
Mincer-Regressions
A simple way to develop a testable Mincerian function
17
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 17lt
Back-tracing to period 0 (1st non-compulsory year of schooling) gives
Applying natural logs (ln) and the standard approximation
we get
Mincer-Regressions
18
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 18lt
For an identical return over all years of education this implies
Considering the impact of experience on income we finally end up w
the legendary Mincerian specification
Mincer-Regressions
19
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 19lt
Interpretation of (semi-log- or log-lin-specification)
Increasing S by 1 yr increases ln(Y) by
NB
therefore if ln(Y) increases by this = an increase of Y by
Mincer-Regressions
20
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 20lt
A slightly more sophisticated version of a Mincerian income function
Mincer-Regressions
w S1 S2 and S3 denoting dummy (01) variables for which
S1 = 1 for a level of education of at least first grade
S2 = 1 for a level of education of at least a high school grade
S3 = 1 for a level of education of at least a college degree
21
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 21lt
auf
Example
An individual who completed high school but not college is characterized by
S1 = 1 S2 = 1 and S3 = 0
Let Nx = yrs of education to attain level of education x
the private ROE on
primary education then is given by r1 = bN1
secondary education then is given by r2 = cN2
college education then is r3 = dN3
Mincer-Regressions
22
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 22lt
Advantages Mincerian vs algebraic method
less data-intense
Completely based on individual data
(instead of groupedaggregated data)
Disadvantages
Primarily suited to compute private ROEs
Completely based on individual data
(instead of groupedaggregated data)
Mincer-Regressions
23
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 23lt
We should keep 3 caveats in mindhellip
1 Education = (partially) a consumption good
- fun-factor bandwagon (business admin fad) political motivation
- eg Angrist and Krueger (JASA 1992)
run for universities during Vietnam War
9 increase in students enrolled for 1965-69
lsquoVietnam Era Draft Lotteryrsquo
1970-73 Prob(study) 5 higher for small norsquos
Mincer-Regressions
24
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 24lt
and alsohellip
2 incomplete labor markets wages may not reflect productivity
3 lsquoSignallingrsquo and lsquoScreeningrsquo Mincerian approach
Mincer-Regressions
Returns to Investment in Education in global perspective
Psacharopoulos and Patrinos (EE 2004)
25
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 25lt
ROEs w increasing level of education
ROEs w increasing level of development
private ROE gt social ROE
MincerndashRegressions
3 major styilized facts (international perspective)
26
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 26lt
MincerndashRegressions
27
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 27lt
Overall women benefit more from education (in terms of returns) but
MincerndashRegressions
28
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 28lt
Major problems of global comparative studies
Sample selection and stratification
1 Developing countries bias towards cities major companies etc
2 Civil servants soldiers etc sample etc
(wages market determined)
Methodology
questionable really same estimation strategy for all countries
MincerndashRegressions
29
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 29lt
Some further critical aspects
Control variables = problematic in MincerndashRegressions (eg occupation)
Functional form linear quadratic in parts linear log
Sheepskin effects (lsquomilestonesrsquo after 12 and 16 yrs)
Selection models
(labor market activity = endogenous decision)
MincerndashRegressions
30
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 30lt
Some sample findings (German Labor Force Survey 2004)
Experience-Income profiles (university male)
MincerndashRegressions
31
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 31lt
Some sample findings (German Labor Force Survey 2004)
Experience-Income profiles (university female)
MincerndashRegressions
32
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 32lt
Some sample findings (German Labor Force Survey 2004)
High school graduates wo further vocational training (control group)
MincerndashRegressions
33
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 33lt
Some sample findings (German Labor Force Survey 2004)
High school graduates wo further vocational training (control group)
MincerndashRegressions
34
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 34lt
Some sample findings (German Labor Force Survey 2004)
High school graduates wo further vocational training (control group)
MincerndashRegressions
35
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 35lt
Some sample findings (German Labor Force Survey 2004)
High school graduates wo further vocational training (control group)
MincerndashRegressions
14
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 14lt
2-tier approach
1 Construct AEPs for specific groups of the population
2 Compute IRR on the base of these AEPs
A rare example of the use of the algebraic methods =
the works by Giora Hanoch (Jo Human Resources 1967)
Her seminal paper is entitled
ldquoAn Economic Analysis of Earnings and Schoolingrdquo
Algebraic method
15
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 15lt
Earnings by age and years of schooling (Hanoch 1967)
Whites North Whites South
Algebraic method
16
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 16lt
w
Yt = annual income for t years of education
Ct = total cost of investment in t-th year of education
rt = return
Assumption
Mincer-Regressions
A simple way to develop a testable Mincerian function
17
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 17lt
Back-tracing to period 0 (1st non-compulsory year of schooling) gives
Applying natural logs (ln) and the standard approximation
we get
Mincer-Regressions
18
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 18lt
For an identical return over all years of education this implies
Considering the impact of experience on income we finally end up w
the legendary Mincerian specification
Mincer-Regressions
19
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 19lt
Interpretation of (semi-log- or log-lin-specification)
Increasing S by 1 yr increases ln(Y) by
NB
therefore if ln(Y) increases by this = an increase of Y by
Mincer-Regressions
20
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 20lt
A slightly more sophisticated version of a Mincerian income function
Mincer-Regressions
w S1 S2 and S3 denoting dummy (01) variables for which
S1 = 1 for a level of education of at least first grade
S2 = 1 for a level of education of at least a high school grade
S3 = 1 for a level of education of at least a college degree
21
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 21lt
auf
Example
An individual who completed high school but not college is characterized by
S1 = 1 S2 = 1 and S3 = 0
Let Nx = yrs of education to attain level of education x
the private ROE on
primary education then is given by r1 = bN1
secondary education then is given by r2 = cN2
college education then is r3 = dN3
Mincer-Regressions
22
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 22lt
Advantages Mincerian vs algebraic method
less data-intense
Completely based on individual data
(instead of groupedaggregated data)
Disadvantages
Primarily suited to compute private ROEs
Completely based on individual data
(instead of groupedaggregated data)
Mincer-Regressions
23
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 23lt
We should keep 3 caveats in mindhellip
1 Education = (partially) a consumption good
- fun-factor bandwagon (business admin fad) political motivation
- eg Angrist and Krueger (JASA 1992)
run for universities during Vietnam War
9 increase in students enrolled for 1965-69
lsquoVietnam Era Draft Lotteryrsquo
1970-73 Prob(study) 5 higher for small norsquos
Mincer-Regressions
24
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 24lt
and alsohellip
2 incomplete labor markets wages may not reflect productivity
3 lsquoSignallingrsquo and lsquoScreeningrsquo Mincerian approach
Mincer-Regressions
Returns to Investment in Education in global perspective
Psacharopoulos and Patrinos (EE 2004)
25
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 25lt
ROEs w increasing level of education
ROEs w increasing level of development
private ROE gt social ROE
MincerndashRegressions
3 major styilized facts (international perspective)
26
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 26lt
MincerndashRegressions
27
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 27lt
Overall women benefit more from education (in terms of returns) but
MincerndashRegressions
28
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 28lt
Major problems of global comparative studies
Sample selection and stratification
1 Developing countries bias towards cities major companies etc
2 Civil servants soldiers etc sample etc
(wages market determined)
Methodology
questionable really same estimation strategy for all countries
MincerndashRegressions
29
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 29lt
Some further critical aspects
Control variables = problematic in MincerndashRegressions (eg occupation)
Functional form linear quadratic in parts linear log
Sheepskin effects (lsquomilestonesrsquo after 12 and 16 yrs)
Selection models
(labor market activity = endogenous decision)
MincerndashRegressions
30
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 30lt
Some sample findings (German Labor Force Survey 2004)
Experience-Income profiles (university male)
MincerndashRegressions
31
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 31lt
Some sample findings (German Labor Force Survey 2004)
Experience-Income profiles (university female)
MincerndashRegressions
32
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 32lt
Some sample findings (German Labor Force Survey 2004)
High school graduates wo further vocational training (control group)
MincerndashRegressions
33
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 33lt
Some sample findings (German Labor Force Survey 2004)
High school graduates wo further vocational training (control group)
MincerndashRegressions
34
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 34lt
Some sample findings (German Labor Force Survey 2004)
High school graduates wo further vocational training (control group)
MincerndashRegressions
35
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 35lt
Some sample findings (German Labor Force Survey 2004)
High school graduates wo further vocational training (control group)
MincerndashRegressions
15
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 15lt
Earnings by age and years of schooling (Hanoch 1967)
Whites North Whites South
Algebraic method
16
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 16lt
w
Yt = annual income for t years of education
Ct = total cost of investment in t-th year of education
rt = return
Assumption
Mincer-Regressions
A simple way to develop a testable Mincerian function
17
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 17lt
Back-tracing to period 0 (1st non-compulsory year of schooling) gives
Applying natural logs (ln) and the standard approximation
we get
Mincer-Regressions
18
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 18lt
For an identical return over all years of education this implies
Considering the impact of experience on income we finally end up w
the legendary Mincerian specification
Mincer-Regressions
19
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 19lt
Interpretation of (semi-log- or log-lin-specification)
Increasing S by 1 yr increases ln(Y) by
NB
therefore if ln(Y) increases by this = an increase of Y by
Mincer-Regressions
20
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 20lt
A slightly more sophisticated version of a Mincerian income function
Mincer-Regressions
w S1 S2 and S3 denoting dummy (01) variables for which
S1 = 1 for a level of education of at least first grade
S2 = 1 for a level of education of at least a high school grade
S3 = 1 for a level of education of at least a college degree
21
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 21lt
auf
Example
An individual who completed high school but not college is characterized by
S1 = 1 S2 = 1 and S3 = 0
Let Nx = yrs of education to attain level of education x
the private ROE on
primary education then is given by r1 = bN1
secondary education then is given by r2 = cN2
college education then is r3 = dN3
Mincer-Regressions
22
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 22lt
Advantages Mincerian vs algebraic method
less data-intense
Completely based on individual data
(instead of groupedaggregated data)
Disadvantages
Primarily suited to compute private ROEs
Completely based on individual data
(instead of groupedaggregated data)
Mincer-Regressions
23
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 23lt
We should keep 3 caveats in mindhellip
1 Education = (partially) a consumption good
- fun-factor bandwagon (business admin fad) political motivation
- eg Angrist and Krueger (JASA 1992)
run for universities during Vietnam War
9 increase in students enrolled for 1965-69
lsquoVietnam Era Draft Lotteryrsquo
1970-73 Prob(study) 5 higher for small norsquos
Mincer-Regressions
24
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 24lt
and alsohellip
2 incomplete labor markets wages may not reflect productivity
3 lsquoSignallingrsquo and lsquoScreeningrsquo Mincerian approach
Mincer-Regressions
Returns to Investment in Education in global perspective
Psacharopoulos and Patrinos (EE 2004)
25
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 25lt
ROEs w increasing level of education
ROEs w increasing level of development
private ROE gt social ROE
MincerndashRegressions
3 major styilized facts (international perspective)
26
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 26lt
MincerndashRegressions
27
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 27lt
Overall women benefit more from education (in terms of returns) but
MincerndashRegressions
28
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 28lt
Major problems of global comparative studies
Sample selection and stratification
1 Developing countries bias towards cities major companies etc
2 Civil servants soldiers etc sample etc
(wages market determined)
Methodology
questionable really same estimation strategy for all countries
MincerndashRegressions
29
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 29lt
Some further critical aspects
Control variables = problematic in MincerndashRegressions (eg occupation)
Functional form linear quadratic in parts linear log
Sheepskin effects (lsquomilestonesrsquo after 12 and 16 yrs)
Selection models
(labor market activity = endogenous decision)
MincerndashRegressions
30
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 30lt
Some sample findings (German Labor Force Survey 2004)
Experience-Income profiles (university male)
MincerndashRegressions
31
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 31lt
Some sample findings (German Labor Force Survey 2004)
Experience-Income profiles (university female)
MincerndashRegressions
32
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 32lt
Some sample findings (German Labor Force Survey 2004)
High school graduates wo further vocational training (control group)
MincerndashRegressions
33
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 33lt
Some sample findings (German Labor Force Survey 2004)
High school graduates wo further vocational training (control group)
MincerndashRegressions
34
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 34lt
Some sample findings (German Labor Force Survey 2004)
High school graduates wo further vocational training (control group)
MincerndashRegressions
35
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 35lt
Some sample findings (German Labor Force Survey 2004)
High school graduates wo further vocational training (control group)
MincerndashRegressions
16
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 16lt
w
Yt = annual income for t years of education
Ct = total cost of investment in t-th year of education
rt = return
Assumption
Mincer-Regressions
A simple way to develop a testable Mincerian function
17
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 17lt
Back-tracing to period 0 (1st non-compulsory year of schooling) gives
Applying natural logs (ln) and the standard approximation
we get
Mincer-Regressions
18
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 18lt
For an identical return over all years of education this implies
Considering the impact of experience on income we finally end up w
the legendary Mincerian specification
Mincer-Regressions
19
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 19lt
Interpretation of (semi-log- or log-lin-specification)
Increasing S by 1 yr increases ln(Y) by
NB
therefore if ln(Y) increases by this = an increase of Y by
Mincer-Regressions
20
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 20lt
A slightly more sophisticated version of a Mincerian income function
Mincer-Regressions
w S1 S2 and S3 denoting dummy (01) variables for which
S1 = 1 for a level of education of at least first grade
S2 = 1 for a level of education of at least a high school grade
S3 = 1 for a level of education of at least a college degree
21
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 21lt
auf
Example
An individual who completed high school but not college is characterized by
S1 = 1 S2 = 1 and S3 = 0
Let Nx = yrs of education to attain level of education x
the private ROE on
primary education then is given by r1 = bN1
secondary education then is given by r2 = cN2
college education then is r3 = dN3
Mincer-Regressions
22
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 22lt
Advantages Mincerian vs algebraic method
less data-intense
Completely based on individual data
(instead of groupedaggregated data)
Disadvantages
Primarily suited to compute private ROEs
Completely based on individual data
(instead of groupedaggregated data)
Mincer-Regressions
23
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 23lt
We should keep 3 caveats in mindhellip
1 Education = (partially) a consumption good
- fun-factor bandwagon (business admin fad) political motivation
- eg Angrist and Krueger (JASA 1992)
run for universities during Vietnam War
9 increase in students enrolled for 1965-69
lsquoVietnam Era Draft Lotteryrsquo
1970-73 Prob(study) 5 higher for small norsquos
Mincer-Regressions
24
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 24lt
and alsohellip
2 incomplete labor markets wages may not reflect productivity
3 lsquoSignallingrsquo and lsquoScreeningrsquo Mincerian approach
Mincer-Regressions
Returns to Investment in Education in global perspective
Psacharopoulos and Patrinos (EE 2004)
25
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 25lt
ROEs w increasing level of education
ROEs w increasing level of development
private ROE gt social ROE
MincerndashRegressions
3 major styilized facts (international perspective)
26
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 26lt
MincerndashRegressions
27
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 27lt
Overall women benefit more from education (in terms of returns) but
MincerndashRegressions
28
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 28lt
Major problems of global comparative studies
Sample selection and stratification
1 Developing countries bias towards cities major companies etc
2 Civil servants soldiers etc sample etc
(wages market determined)
Methodology
questionable really same estimation strategy for all countries
MincerndashRegressions
29
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 29lt
Some further critical aspects
Control variables = problematic in MincerndashRegressions (eg occupation)
Functional form linear quadratic in parts linear log
Sheepskin effects (lsquomilestonesrsquo after 12 and 16 yrs)
Selection models
(labor market activity = endogenous decision)
MincerndashRegressions
30
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 30lt
Some sample findings (German Labor Force Survey 2004)
Experience-Income profiles (university male)
MincerndashRegressions
31
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 31lt
Some sample findings (German Labor Force Survey 2004)
Experience-Income profiles (university female)
MincerndashRegressions
32
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 32lt
Some sample findings (German Labor Force Survey 2004)
High school graduates wo further vocational training (control group)
MincerndashRegressions
33
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 33lt
Some sample findings (German Labor Force Survey 2004)
High school graduates wo further vocational training (control group)
MincerndashRegressions
34
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 34lt
Some sample findings (German Labor Force Survey 2004)
High school graduates wo further vocational training (control group)
MincerndashRegressions
35
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 35lt
Some sample findings (German Labor Force Survey 2004)
High school graduates wo further vocational training (control group)
MincerndashRegressions
17
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 17lt
Back-tracing to period 0 (1st non-compulsory year of schooling) gives
Applying natural logs (ln) and the standard approximation
we get
Mincer-Regressions
18
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 18lt
For an identical return over all years of education this implies
Considering the impact of experience on income we finally end up w
the legendary Mincerian specification
Mincer-Regressions
19
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 19lt
Interpretation of (semi-log- or log-lin-specification)
Increasing S by 1 yr increases ln(Y) by
NB
therefore if ln(Y) increases by this = an increase of Y by
Mincer-Regressions
20
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 20lt
A slightly more sophisticated version of a Mincerian income function
Mincer-Regressions
w S1 S2 and S3 denoting dummy (01) variables for which
S1 = 1 for a level of education of at least first grade
S2 = 1 for a level of education of at least a high school grade
S3 = 1 for a level of education of at least a college degree
21
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 21lt
auf
Example
An individual who completed high school but not college is characterized by
S1 = 1 S2 = 1 and S3 = 0
Let Nx = yrs of education to attain level of education x
the private ROE on
primary education then is given by r1 = bN1
secondary education then is given by r2 = cN2
college education then is r3 = dN3
Mincer-Regressions
22
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 22lt
Advantages Mincerian vs algebraic method
less data-intense
Completely based on individual data
(instead of groupedaggregated data)
Disadvantages
Primarily suited to compute private ROEs
Completely based on individual data
(instead of groupedaggregated data)
Mincer-Regressions
23
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 23lt
We should keep 3 caveats in mindhellip
1 Education = (partially) a consumption good
- fun-factor bandwagon (business admin fad) political motivation
- eg Angrist and Krueger (JASA 1992)
run for universities during Vietnam War
9 increase in students enrolled for 1965-69
lsquoVietnam Era Draft Lotteryrsquo
1970-73 Prob(study) 5 higher for small norsquos
Mincer-Regressions
24
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 24lt
and alsohellip
2 incomplete labor markets wages may not reflect productivity
3 lsquoSignallingrsquo and lsquoScreeningrsquo Mincerian approach
Mincer-Regressions
Returns to Investment in Education in global perspective
Psacharopoulos and Patrinos (EE 2004)
25
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 25lt
ROEs w increasing level of education
ROEs w increasing level of development
private ROE gt social ROE
MincerndashRegressions
3 major styilized facts (international perspective)
26
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 26lt
MincerndashRegressions
27
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 27lt
Overall women benefit more from education (in terms of returns) but
MincerndashRegressions
28
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 28lt
Major problems of global comparative studies
Sample selection and stratification
1 Developing countries bias towards cities major companies etc
2 Civil servants soldiers etc sample etc
(wages market determined)
Methodology
questionable really same estimation strategy for all countries
MincerndashRegressions
29
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 29lt
Some further critical aspects
Control variables = problematic in MincerndashRegressions (eg occupation)
Functional form linear quadratic in parts linear log
Sheepskin effects (lsquomilestonesrsquo after 12 and 16 yrs)
Selection models
(labor market activity = endogenous decision)
MincerndashRegressions
30
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 30lt
Some sample findings (German Labor Force Survey 2004)
Experience-Income profiles (university male)
MincerndashRegressions
31
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 31lt
Some sample findings (German Labor Force Survey 2004)
Experience-Income profiles (university female)
MincerndashRegressions
32
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 32lt
Some sample findings (German Labor Force Survey 2004)
High school graduates wo further vocational training (control group)
MincerndashRegressions
33
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 33lt
Some sample findings (German Labor Force Survey 2004)
High school graduates wo further vocational training (control group)
MincerndashRegressions
34
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 34lt
Some sample findings (German Labor Force Survey 2004)
High school graduates wo further vocational training (control group)
MincerndashRegressions
35
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 35lt
Some sample findings (German Labor Force Survey 2004)
High school graduates wo further vocational training (control group)
MincerndashRegressions
18
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 18lt
For an identical return over all years of education this implies
Considering the impact of experience on income we finally end up w
the legendary Mincerian specification
Mincer-Regressions
19
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 19lt
Interpretation of (semi-log- or log-lin-specification)
Increasing S by 1 yr increases ln(Y) by
NB
therefore if ln(Y) increases by this = an increase of Y by
Mincer-Regressions
20
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 20lt
A slightly more sophisticated version of a Mincerian income function
Mincer-Regressions
w S1 S2 and S3 denoting dummy (01) variables for which
S1 = 1 for a level of education of at least first grade
S2 = 1 for a level of education of at least a high school grade
S3 = 1 for a level of education of at least a college degree
21
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 21lt
auf
Example
An individual who completed high school but not college is characterized by
S1 = 1 S2 = 1 and S3 = 0
Let Nx = yrs of education to attain level of education x
the private ROE on
primary education then is given by r1 = bN1
secondary education then is given by r2 = cN2
college education then is r3 = dN3
Mincer-Regressions
22
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 22lt
Advantages Mincerian vs algebraic method
less data-intense
Completely based on individual data
(instead of groupedaggregated data)
Disadvantages
Primarily suited to compute private ROEs
Completely based on individual data
(instead of groupedaggregated data)
Mincer-Regressions
23
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 23lt
We should keep 3 caveats in mindhellip
1 Education = (partially) a consumption good
- fun-factor bandwagon (business admin fad) political motivation
- eg Angrist and Krueger (JASA 1992)
run for universities during Vietnam War
9 increase in students enrolled for 1965-69
lsquoVietnam Era Draft Lotteryrsquo
1970-73 Prob(study) 5 higher for small norsquos
Mincer-Regressions
24
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 24lt
and alsohellip
2 incomplete labor markets wages may not reflect productivity
3 lsquoSignallingrsquo and lsquoScreeningrsquo Mincerian approach
Mincer-Regressions
Returns to Investment in Education in global perspective
Psacharopoulos and Patrinos (EE 2004)
25
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 25lt
ROEs w increasing level of education
ROEs w increasing level of development
private ROE gt social ROE
MincerndashRegressions
3 major styilized facts (international perspective)
26
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 26lt
MincerndashRegressions
27
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 27lt
Overall women benefit more from education (in terms of returns) but
MincerndashRegressions
28
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 28lt
Major problems of global comparative studies
Sample selection and stratification
1 Developing countries bias towards cities major companies etc
2 Civil servants soldiers etc sample etc
(wages market determined)
Methodology
questionable really same estimation strategy for all countries
MincerndashRegressions
29
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 29lt
Some further critical aspects
Control variables = problematic in MincerndashRegressions (eg occupation)
Functional form linear quadratic in parts linear log
Sheepskin effects (lsquomilestonesrsquo after 12 and 16 yrs)
Selection models
(labor market activity = endogenous decision)
MincerndashRegressions
30
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 30lt
Some sample findings (German Labor Force Survey 2004)
Experience-Income profiles (university male)
MincerndashRegressions
31
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 31lt
Some sample findings (German Labor Force Survey 2004)
Experience-Income profiles (university female)
MincerndashRegressions
32
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 32lt
Some sample findings (German Labor Force Survey 2004)
High school graduates wo further vocational training (control group)
MincerndashRegressions
33
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 33lt
Some sample findings (German Labor Force Survey 2004)
High school graduates wo further vocational training (control group)
MincerndashRegressions
34
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 34lt
Some sample findings (German Labor Force Survey 2004)
High school graduates wo further vocational training (control group)
MincerndashRegressions
35
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 35lt
Some sample findings (German Labor Force Survey 2004)
High school graduates wo further vocational training (control group)
MincerndashRegressions
19
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 19lt
Interpretation of (semi-log- or log-lin-specification)
Increasing S by 1 yr increases ln(Y) by
NB
therefore if ln(Y) increases by this = an increase of Y by
Mincer-Regressions
20
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 20lt
A slightly more sophisticated version of a Mincerian income function
Mincer-Regressions
w S1 S2 and S3 denoting dummy (01) variables for which
S1 = 1 for a level of education of at least first grade
S2 = 1 for a level of education of at least a high school grade
S3 = 1 for a level of education of at least a college degree
21
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 21lt
auf
Example
An individual who completed high school but not college is characterized by
S1 = 1 S2 = 1 and S3 = 0
Let Nx = yrs of education to attain level of education x
the private ROE on
primary education then is given by r1 = bN1
secondary education then is given by r2 = cN2
college education then is r3 = dN3
Mincer-Regressions
22
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 22lt
Advantages Mincerian vs algebraic method
less data-intense
Completely based on individual data
(instead of groupedaggregated data)
Disadvantages
Primarily suited to compute private ROEs
Completely based on individual data
(instead of groupedaggregated data)
Mincer-Regressions
23
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 23lt
We should keep 3 caveats in mindhellip
1 Education = (partially) a consumption good
- fun-factor bandwagon (business admin fad) political motivation
- eg Angrist and Krueger (JASA 1992)
run for universities during Vietnam War
9 increase in students enrolled for 1965-69
lsquoVietnam Era Draft Lotteryrsquo
1970-73 Prob(study) 5 higher for small norsquos
Mincer-Regressions
24
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 24lt
and alsohellip
2 incomplete labor markets wages may not reflect productivity
3 lsquoSignallingrsquo and lsquoScreeningrsquo Mincerian approach
Mincer-Regressions
Returns to Investment in Education in global perspective
Psacharopoulos and Patrinos (EE 2004)
25
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 25lt
ROEs w increasing level of education
ROEs w increasing level of development
private ROE gt social ROE
MincerndashRegressions
3 major styilized facts (international perspective)
26
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 26lt
MincerndashRegressions
27
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 27lt
Overall women benefit more from education (in terms of returns) but
MincerndashRegressions
28
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 28lt
Major problems of global comparative studies
Sample selection and stratification
1 Developing countries bias towards cities major companies etc
2 Civil servants soldiers etc sample etc
(wages market determined)
Methodology
questionable really same estimation strategy for all countries
MincerndashRegressions
29
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 29lt
Some further critical aspects
Control variables = problematic in MincerndashRegressions (eg occupation)
Functional form linear quadratic in parts linear log
Sheepskin effects (lsquomilestonesrsquo after 12 and 16 yrs)
Selection models
(labor market activity = endogenous decision)
MincerndashRegressions
30
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 30lt
Some sample findings (German Labor Force Survey 2004)
Experience-Income profiles (university male)
MincerndashRegressions
31
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 31lt
Some sample findings (German Labor Force Survey 2004)
Experience-Income profiles (university female)
MincerndashRegressions
32
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 32lt
Some sample findings (German Labor Force Survey 2004)
High school graduates wo further vocational training (control group)
MincerndashRegressions
33
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 33lt
Some sample findings (German Labor Force Survey 2004)
High school graduates wo further vocational training (control group)
MincerndashRegressions
34
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 34lt
Some sample findings (German Labor Force Survey 2004)
High school graduates wo further vocational training (control group)
MincerndashRegressions
35
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 35lt
Some sample findings (German Labor Force Survey 2004)
High school graduates wo further vocational training (control group)
MincerndashRegressions
20
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 20lt
A slightly more sophisticated version of a Mincerian income function
Mincer-Regressions
w S1 S2 and S3 denoting dummy (01) variables for which
S1 = 1 for a level of education of at least first grade
S2 = 1 for a level of education of at least a high school grade
S3 = 1 for a level of education of at least a college degree
21
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 21lt
auf
Example
An individual who completed high school but not college is characterized by
S1 = 1 S2 = 1 and S3 = 0
Let Nx = yrs of education to attain level of education x
the private ROE on
primary education then is given by r1 = bN1
secondary education then is given by r2 = cN2
college education then is r3 = dN3
Mincer-Regressions
22
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 22lt
Advantages Mincerian vs algebraic method
less data-intense
Completely based on individual data
(instead of groupedaggregated data)
Disadvantages
Primarily suited to compute private ROEs
Completely based on individual data
(instead of groupedaggregated data)
Mincer-Regressions
23
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 23lt
We should keep 3 caveats in mindhellip
1 Education = (partially) a consumption good
- fun-factor bandwagon (business admin fad) political motivation
- eg Angrist and Krueger (JASA 1992)
run for universities during Vietnam War
9 increase in students enrolled for 1965-69
lsquoVietnam Era Draft Lotteryrsquo
1970-73 Prob(study) 5 higher for small norsquos
Mincer-Regressions
24
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 24lt
and alsohellip
2 incomplete labor markets wages may not reflect productivity
3 lsquoSignallingrsquo and lsquoScreeningrsquo Mincerian approach
Mincer-Regressions
Returns to Investment in Education in global perspective
Psacharopoulos and Patrinos (EE 2004)
25
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 25lt
ROEs w increasing level of education
ROEs w increasing level of development
private ROE gt social ROE
MincerndashRegressions
3 major styilized facts (international perspective)
26
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 26lt
MincerndashRegressions
27
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 27lt
Overall women benefit more from education (in terms of returns) but
MincerndashRegressions
28
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 28lt
Major problems of global comparative studies
Sample selection and stratification
1 Developing countries bias towards cities major companies etc
2 Civil servants soldiers etc sample etc
(wages market determined)
Methodology
questionable really same estimation strategy for all countries
MincerndashRegressions
29
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 29lt
Some further critical aspects
Control variables = problematic in MincerndashRegressions (eg occupation)
Functional form linear quadratic in parts linear log
Sheepskin effects (lsquomilestonesrsquo after 12 and 16 yrs)
Selection models
(labor market activity = endogenous decision)
MincerndashRegressions
30
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 30lt
Some sample findings (German Labor Force Survey 2004)
Experience-Income profiles (university male)
MincerndashRegressions
31
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 31lt
Some sample findings (German Labor Force Survey 2004)
Experience-Income profiles (university female)
MincerndashRegressions
32
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 32lt
Some sample findings (German Labor Force Survey 2004)
High school graduates wo further vocational training (control group)
MincerndashRegressions
33
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 33lt
Some sample findings (German Labor Force Survey 2004)
High school graduates wo further vocational training (control group)
MincerndashRegressions
34
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 34lt
Some sample findings (German Labor Force Survey 2004)
High school graduates wo further vocational training (control group)
MincerndashRegressions
35
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 35lt
Some sample findings (German Labor Force Survey 2004)
High school graduates wo further vocational training (control group)
MincerndashRegressions
21
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 21lt
auf
Example
An individual who completed high school but not college is characterized by
S1 = 1 S2 = 1 and S3 = 0
Let Nx = yrs of education to attain level of education x
the private ROE on
primary education then is given by r1 = bN1
secondary education then is given by r2 = cN2
college education then is r3 = dN3
Mincer-Regressions
22
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 22lt
Advantages Mincerian vs algebraic method
less data-intense
Completely based on individual data
(instead of groupedaggregated data)
Disadvantages
Primarily suited to compute private ROEs
Completely based on individual data
(instead of groupedaggregated data)
Mincer-Regressions
23
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 23lt
We should keep 3 caveats in mindhellip
1 Education = (partially) a consumption good
- fun-factor bandwagon (business admin fad) political motivation
- eg Angrist and Krueger (JASA 1992)
run for universities during Vietnam War
9 increase in students enrolled for 1965-69
lsquoVietnam Era Draft Lotteryrsquo
1970-73 Prob(study) 5 higher for small norsquos
Mincer-Regressions
24
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 24lt
and alsohellip
2 incomplete labor markets wages may not reflect productivity
3 lsquoSignallingrsquo and lsquoScreeningrsquo Mincerian approach
Mincer-Regressions
Returns to Investment in Education in global perspective
Psacharopoulos and Patrinos (EE 2004)
25
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 25lt
ROEs w increasing level of education
ROEs w increasing level of development
private ROE gt social ROE
MincerndashRegressions
3 major styilized facts (international perspective)
26
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 26lt
MincerndashRegressions
27
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 27lt
Overall women benefit more from education (in terms of returns) but
MincerndashRegressions
28
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 28lt
Major problems of global comparative studies
Sample selection and stratification
1 Developing countries bias towards cities major companies etc
2 Civil servants soldiers etc sample etc
(wages market determined)
Methodology
questionable really same estimation strategy for all countries
MincerndashRegressions
29
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 29lt
Some further critical aspects
Control variables = problematic in MincerndashRegressions (eg occupation)
Functional form linear quadratic in parts linear log
Sheepskin effects (lsquomilestonesrsquo after 12 and 16 yrs)
Selection models
(labor market activity = endogenous decision)
MincerndashRegressions
30
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 30lt
Some sample findings (German Labor Force Survey 2004)
Experience-Income profiles (university male)
MincerndashRegressions
31
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 31lt
Some sample findings (German Labor Force Survey 2004)
Experience-Income profiles (university female)
MincerndashRegressions
32
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 32lt
Some sample findings (German Labor Force Survey 2004)
High school graduates wo further vocational training (control group)
MincerndashRegressions
33
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 33lt
Some sample findings (German Labor Force Survey 2004)
High school graduates wo further vocational training (control group)
MincerndashRegressions
34
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 34lt
Some sample findings (German Labor Force Survey 2004)
High school graduates wo further vocational training (control group)
MincerndashRegressions
35
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 35lt
Some sample findings (German Labor Force Survey 2004)
High school graduates wo further vocational training (control group)
MincerndashRegressions
22
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 22lt
Advantages Mincerian vs algebraic method
less data-intense
Completely based on individual data
(instead of groupedaggregated data)
Disadvantages
Primarily suited to compute private ROEs
Completely based on individual data
(instead of groupedaggregated data)
Mincer-Regressions
23
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 23lt
We should keep 3 caveats in mindhellip
1 Education = (partially) a consumption good
- fun-factor bandwagon (business admin fad) political motivation
- eg Angrist and Krueger (JASA 1992)
run for universities during Vietnam War
9 increase in students enrolled for 1965-69
lsquoVietnam Era Draft Lotteryrsquo
1970-73 Prob(study) 5 higher for small norsquos
Mincer-Regressions
24
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 24lt
and alsohellip
2 incomplete labor markets wages may not reflect productivity
3 lsquoSignallingrsquo and lsquoScreeningrsquo Mincerian approach
Mincer-Regressions
Returns to Investment in Education in global perspective
Psacharopoulos and Patrinos (EE 2004)
25
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 25lt
ROEs w increasing level of education
ROEs w increasing level of development
private ROE gt social ROE
MincerndashRegressions
3 major styilized facts (international perspective)
26
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 26lt
MincerndashRegressions
27
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 27lt
Overall women benefit more from education (in terms of returns) but
MincerndashRegressions
28
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 28lt
Major problems of global comparative studies
Sample selection and stratification
1 Developing countries bias towards cities major companies etc
2 Civil servants soldiers etc sample etc
(wages market determined)
Methodology
questionable really same estimation strategy for all countries
MincerndashRegressions
29
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 29lt
Some further critical aspects
Control variables = problematic in MincerndashRegressions (eg occupation)
Functional form linear quadratic in parts linear log
Sheepskin effects (lsquomilestonesrsquo after 12 and 16 yrs)
Selection models
(labor market activity = endogenous decision)
MincerndashRegressions
30
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 30lt
Some sample findings (German Labor Force Survey 2004)
Experience-Income profiles (university male)
MincerndashRegressions
31
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 31lt
Some sample findings (German Labor Force Survey 2004)
Experience-Income profiles (university female)
MincerndashRegressions
32
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 32lt
Some sample findings (German Labor Force Survey 2004)
High school graduates wo further vocational training (control group)
MincerndashRegressions
33
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 33lt
Some sample findings (German Labor Force Survey 2004)
High school graduates wo further vocational training (control group)
MincerndashRegressions
34
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 34lt
Some sample findings (German Labor Force Survey 2004)
High school graduates wo further vocational training (control group)
MincerndashRegressions
35
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 35lt
Some sample findings (German Labor Force Survey 2004)
High school graduates wo further vocational training (control group)
MincerndashRegressions
23
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 23lt
We should keep 3 caveats in mindhellip
1 Education = (partially) a consumption good
- fun-factor bandwagon (business admin fad) political motivation
- eg Angrist and Krueger (JASA 1992)
run for universities during Vietnam War
9 increase in students enrolled for 1965-69
lsquoVietnam Era Draft Lotteryrsquo
1970-73 Prob(study) 5 higher for small norsquos
Mincer-Regressions
24
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 24lt
and alsohellip
2 incomplete labor markets wages may not reflect productivity
3 lsquoSignallingrsquo and lsquoScreeningrsquo Mincerian approach
Mincer-Regressions
Returns to Investment in Education in global perspective
Psacharopoulos and Patrinos (EE 2004)
25
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 25lt
ROEs w increasing level of education
ROEs w increasing level of development
private ROE gt social ROE
MincerndashRegressions
3 major styilized facts (international perspective)
26
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 26lt
MincerndashRegressions
27
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 27lt
Overall women benefit more from education (in terms of returns) but
MincerndashRegressions
28
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 28lt
Major problems of global comparative studies
Sample selection and stratification
1 Developing countries bias towards cities major companies etc
2 Civil servants soldiers etc sample etc
(wages market determined)
Methodology
questionable really same estimation strategy for all countries
MincerndashRegressions
29
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 29lt
Some further critical aspects
Control variables = problematic in MincerndashRegressions (eg occupation)
Functional form linear quadratic in parts linear log
Sheepskin effects (lsquomilestonesrsquo after 12 and 16 yrs)
Selection models
(labor market activity = endogenous decision)
MincerndashRegressions
30
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 30lt
Some sample findings (German Labor Force Survey 2004)
Experience-Income profiles (university male)
MincerndashRegressions
31
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 31lt
Some sample findings (German Labor Force Survey 2004)
Experience-Income profiles (university female)
MincerndashRegressions
32
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 32lt
Some sample findings (German Labor Force Survey 2004)
High school graduates wo further vocational training (control group)
MincerndashRegressions
33
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 33lt
Some sample findings (German Labor Force Survey 2004)
High school graduates wo further vocational training (control group)
MincerndashRegressions
34
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 34lt
Some sample findings (German Labor Force Survey 2004)
High school graduates wo further vocational training (control group)
MincerndashRegressions
35
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 35lt
Some sample findings (German Labor Force Survey 2004)
High school graduates wo further vocational training (control group)
MincerndashRegressions
24
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 24lt
and alsohellip
2 incomplete labor markets wages may not reflect productivity
3 lsquoSignallingrsquo and lsquoScreeningrsquo Mincerian approach
Mincer-Regressions
Returns to Investment in Education in global perspective
Psacharopoulos and Patrinos (EE 2004)
25
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 25lt
ROEs w increasing level of education
ROEs w increasing level of development
private ROE gt social ROE
MincerndashRegressions
3 major styilized facts (international perspective)
26
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 26lt
MincerndashRegressions
27
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 27lt
Overall women benefit more from education (in terms of returns) but
MincerndashRegressions
28
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 28lt
Major problems of global comparative studies
Sample selection and stratification
1 Developing countries bias towards cities major companies etc
2 Civil servants soldiers etc sample etc
(wages market determined)
Methodology
questionable really same estimation strategy for all countries
MincerndashRegressions
29
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 29lt
Some further critical aspects
Control variables = problematic in MincerndashRegressions (eg occupation)
Functional form linear quadratic in parts linear log
Sheepskin effects (lsquomilestonesrsquo after 12 and 16 yrs)
Selection models
(labor market activity = endogenous decision)
MincerndashRegressions
30
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 30lt
Some sample findings (German Labor Force Survey 2004)
Experience-Income profiles (university male)
MincerndashRegressions
31
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 31lt
Some sample findings (German Labor Force Survey 2004)
Experience-Income profiles (university female)
MincerndashRegressions
32
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 32lt
Some sample findings (German Labor Force Survey 2004)
High school graduates wo further vocational training (control group)
MincerndashRegressions
33
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 33lt
Some sample findings (German Labor Force Survey 2004)
High school graduates wo further vocational training (control group)
MincerndashRegressions
34
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 34lt
Some sample findings (German Labor Force Survey 2004)
High school graduates wo further vocational training (control group)
MincerndashRegressions
35
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 35lt
Some sample findings (German Labor Force Survey 2004)
High school graduates wo further vocational training (control group)
MincerndashRegressions
25
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 25lt
ROEs w increasing level of education
ROEs w increasing level of development
private ROE gt social ROE
MincerndashRegressions
3 major styilized facts (international perspective)
26
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 26lt
MincerndashRegressions
27
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 27lt
Overall women benefit more from education (in terms of returns) but
MincerndashRegressions
28
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 28lt
Major problems of global comparative studies
Sample selection and stratification
1 Developing countries bias towards cities major companies etc
2 Civil servants soldiers etc sample etc
(wages market determined)
Methodology
questionable really same estimation strategy for all countries
MincerndashRegressions
29
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 29lt
Some further critical aspects
Control variables = problematic in MincerndashRegressions (eg occupation)
Functional form linear quadratic in parts linear log
Sheepskin effects (lsquomilestonesrsquo after 12 and 16 yrs)
Selection models
(labor market activity = endogenous decision)
MincerndashRegressions
30
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 30lt
Some sample findings (German Labor Force Survey 2004)
Experience-Income profiles (university male)
MincerndashRegressions
31
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 31lt
Some sample findings (German Labor Force Survey 2004)
Experience-Income profiles (university female)
MincerndashRegressions
32
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 32lt
Some sample findings (German Labor Force Survey 2004)
High school graduates wo further vocational training (control group)
MincerndashRegressions
33
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 33lt
Some sample findings (German Labor Force Survey 2004)
High school graduates wo further vocational training (control group)
MincerndashRegressions
34
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 34lt
Some sample findings (German Labor Force Survey 2004)
High school graduates wo further vocational training (control group)
MincerndashRegressions
35
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 35lt
Some sample findings (German Labor Force Survey 2004)
High school graduates wo further vocational training (control group)
MincerndashRegressions
26
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 26lt
MincerndashRegressions
27
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 27lt
Overall women benefit more from education (in terms of returns) but
MincerndashRegressions
28
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 28lt
Major problems of global comparative studies
Sample selection and stratification
1 Developing countries bias towards cities major companies etc
2 Civil servants soldiers etc sample etc
(wages market determined)
Methodology
questionable really same estimation strategy for all countries
MincerndashRegressions
29
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 29lt
Some further critical aspects
Control variables = problematic in MincerndashRegressions (eg occupation)
Functional form linear quadratic in parts linear log
Sheepskin effects (lsquomilestonesrsquo after 12 and 16 yrs)
Selection models
(labor market activity = endogenous decision)
MincerndashRegressions
30
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 30lt
Some sample findings (German Labor Force Survey 2004)
Experience-Income profiles (university male)
MincerndashRegressions
31
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 31lt
Some sample findings (German Labor Force Survey 2004)
Experience-Income profiles (university female)
MincerndashRegressions
32
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 32lt
Some sample findings (German Labor Force Survey 2004)
High school graduates wo further vocational training (control group)
MincerndashRegressions
33
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 33lt
Some sample findings (German Labor Force Survey 2004)
High school graduates wo further vocational training (control group)
MincerndashRegressions
34
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 34lt
Some sample findings (German Labor Force Survey 2004)
High school graduates wo further vocational training (control group)
MincerndashRegressions
35
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 35lt
Some sample findings (German Labor Force Survey 2004)
High school graduates wo further vocational training (control group)
MincerndashRegressions
27
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 27lt
Overall women benefit more from education (in terms of returns) but
MincerndashRegressions
28
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 28lt
Major problems of global comparative studies
Sample selection and stratification
1 Developing countries bias towards cities major companies etc
2 Civil servants soldiers etc sample etc
(wages market determined)
Methodology
questionable really same estimation strategy for all countries
MincerndashRegressions
29
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 29lt
Some further critical aspects
Control variables = problematic in MincerndashRegressions (eg occupation)
Functional form linear quadratic in parts linear log
Sheepskin effects (lsquomilestonesrsquo after 12 and 16 yrs)
Selection models
(labor market activity = endogenous decision)
MincerndashRegressions
30
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 30lt
Some sample findings (German Labor Force Survey 2004)
Experience-Income profiles (university male)
MincerndashRegressions
31
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 31lt
Some sample findings (German Labor Force Survey 2004)
Experience-Income profiles (university female)
MincerndashRegressions
32
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 32lt
Some sample findings (German Labor Force Survey 2004)
High school graduates wo further vocational training (control group)
MincerndashRegressions
33
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 33lt
Some sample findings (German Labor Force Survey 2004)
High school graduates wo further vocational training (control group)
MincerndashRegressions
34
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 34lt
Some sample findings (German Labor Force Survey 2004)
High school graduates wo further vocational training (control group)
MincerndashRegressions
35
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 35lt
Some sample findings (German Labor Force Survey 2004)
High school graduates wo further vocational training (control group)
MincerndashRegressions
28
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 28lt
Major problems of global comparative studies
Sample selection and stratification
1 Developing countries bias towards cities major companies etc
2 Civil servants soldiers etc sample etc
(wages market determined)
Methodology
questionable really same estimation strategy for all countries
MincerndashRegressions
29
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 29lt
Some further critical aspects
Control variables = problematic in MincerndashRegressions (eg occupation)
Functional form linear quadratic in parts linear log
Sheepskin effects (lsquomilestonesrsquo after 12 and 16 yrs)
Selection models
(labor market activity = endogenous decision)
MincerndashRegressions
30
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 30lt
Some sample findings (German Labor Force Survey 2004)
Experience-Income profiles (university male)
MincerndashRegressions
31
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 31lt
Some sample findings (German Labor Force Survey 2004)
Experience-Income profiles (university female)
MincerndashRegressions
32
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 32lt
Some sample findings (German Labor Force Survey 2004)
High school graduates wo further vocational training (control group)
MincerndashRegressions
33
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 33lt
Some sample findings (German Labor Force Survey 2004)
High school graduates wo further vocational training (control group)
MincerndashRegressions
34
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 34lt
Some sample findings (German Labor Force Survey 2004)
High school graduates wo further vocational training (control group)
MincerndashRegressions
35
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 35lt
Some sample findings (German Labor Force Survey 2004)
High school graduates wo further vocational training (control group)
MincerndashRegressions
29
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 29lt
Some further critical aspects
Control variables = problematic in MincerndashRegressions (eg occupation)
Functional form linear quadratic in parts linear log
Sheepskin effects (lsquomilestonesrsquo after 12 and 16 yrs)
Selection models
(labor market activity = endogenous decision)
MincerndashRegressions
30
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 30lt
Some sample findings (German Labor Force Survey 2004)
Experience-Income profiles (university male)
MincerndashRegressions
31
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 31lt
Some sample findings (German Labor Force Survey 2004)
Experience-Income profiles (university female)
MincerndashRegressions
32
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 32lt
Some sample findings (German Labor Force Survey 2004)
High school graduates wo further vocational training (control group)
MincerndashRegressions
33
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 33lt
Some sample findings (German Labor Force Survey 2004)
High school graduates wo further vocational training (control group)
MincerndashRegressions
34
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 34lt
Some sample findings (German Labor Force Survey 2004)
High school graduates wo further vocational training (control group)
MincerndashRegressions
35
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 35lt
Some sample findings (German Labor Force Survey 2004)
High school graduates wo further vocational training (control group)
MincerndashRegressions
30
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 30lt
Some sample findings (German Labor Force Survey 2004)
Experience-Income profiles (university male)
MincerndashRegressions
31
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 31lt
Some sample findings (German Labor Force Survey 2004)
Experience-Income profiles (university female)
MincerndashRegressions
32
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 32lt
Some sample findings (German Labor Force Survey 2004)
High school graduates wo further vocational training (control group)
MincerndashRegressions
33
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 33lt
Some sample findings (German Labor Force Survey 2004)
High school graduates wo further vocational training (control group)
MincerndashRegressions
34
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 34lt
Some sample findings (German Labor Force Survey 2004)
High school graduates wo further vocational training (control group)
MincerndashRegressions
35
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 35lt
Some sample findings (German Labor Force Survey 2004)
High school graduates wo further vocational training (control group)
MincerndashRegressions
31
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 31lt
Some sample findings (German Labor Force Survey 2004)
Experience-Income profiles (university female)
MincerndashRegressions
32
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 32lt
Some sample findings (German Labor Force Survey 2004)
High school graduates wo further vocational training (control group)
MincerndashRegressions
33
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 33lt
Some sample findings (German Labor Force Survey 2004)
High school graduates wo further vocational training (control group)
MincerndashRegressions
34
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 34lt
Some sample findings (German Labor Force Survey 2004)
High school graduates wo further vocational training (control group)
MincerndashRegressions
35
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 35lt
Some sample findings (German Labor Force Survey 2004)
High school graduates wo further vocational training (control group)
MincerndashRegressions
32
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 32lt
Some sample findings (German Labor Force Survey 2004)
High school graduates wo further vocational training (control group)
MincerndashRegressions
33
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 33lt
Some sample findings (German Labor Force Survey 2004)
High school graduates wo further vocational training (control group)
MincerndashRegressions
34
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 34lt
Some sample findings (German Labor Force Survey 2004)
High school graduates wo further vocational training (control group)
MincerndashRegressions
35
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 35lt
Some sample findings (German Labor Force Survey 2004)
High school graduates wo further vocational training (control group)
MincerndashRegressions
33
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 33lt
Some sample findings (German Labor Force Survey 2004)
High school graduates wo further vocational training (control group)
MincerndashRegressions
34
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 34lt
Some sample findings (German Labor Force Survey 2004)
High school graduates wo further vocational training (control group)
MincerndashRegressions
35
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 35lt
Some sample findings (German Labor Force Survey 2004)
High school graduates wo further vocational training (control group)
MincerndashRegressions
34
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 34lt
Some sample findings (German Labor Force Survey 2004)
High school graduates wo further vocational training (control group)
MincerndashRegressions
35
QEE Prof Bernd Suumlssmuth IEW Winter 20102011 35lt
Some sample findings (German Labor Force Survey 2004)
High school graduates wo further vocational training (control group)
MincerndashRegressions