1 INTERGENERATIONAL ECONOMIC MOBILITY IN GERMANY: LEVELS AND TRENDS Iryna Kyzyma* Luxembourg Institute of Socio-Economic Research Olaf Groh-Samberg University of Bremen July 12, 2018 Abstract This paper provides new evidence on intergenerational economic mobility in Germany by analyzing the degree of intergenerational persistence in ranks – positions, which parents and children occupy in their respective income distributions. Using data from the German Socio- Economic Panel, we find that the association of children’s ranks with ranks of their parents is about 0.242 for individual labor earnings and 0.214 for household pre-tax income. The evidence points that mobility of earnings across generations is higher for daughters than for sons whereas the opposite applies to the mobility of household pretax income. We also find that intergenerational rank mobility of earnings decreased twice for children born in 1973-1977 as compared to children born in 1968-1972. JEL codes: D31, J31, J62. Keywords: intergenerational economic mobility, absolute rank mobility, relative rank mobility, income inequality, changes over time ________ * Corresponding author, Luxembourg Institute of Socio-economic Research, 11 Porte des Sciences, L-4366 Esch- sur-Alzette, G.D. Luxembourg; e-mail: [email protected].
41
Embed
INTERGENERATIONAL ECONOMIC MOBILITY IN ......2. Rank-based approach to measuring intergenerational economic mobility 2.1. The intergenerational association of ranks In this paper,
This document is posted to help you gain knowledge. Please leave a comment to let me know what you think about it! Share it to your friends and learn new things together.
Transcript
1
INTERGENERATIONAL ECONOMIC MOBILITY IN GERMANY:
LEVELS AND TRENDS
Iryna Kyzyma*
Luxembourg Institute of Socio-Economic Research
Olaf Groh-Samberg
University of Bremen
July 12, 2018
Abstract
This paper provides new evidence on intergenerational economic mobility in Germany by
analyzing the degree of intergenerational persistence in ranks – positions, which parents and
children occupy in their respective income distributions. Using data from the German Socio-
Economic Panel, we find that the association of children’s ranks with ranks of their parents is
about 0.242 for individual labor earnings and 0.214 for household pre-tax income. The evidence
points that mobility of earnings across generations is higher for daughters than for sons whereas
the opposite applies to the mobility of household pretax income. We also find that
intergenerational rank mobility of earnings decreased twice for children born in 1973-1977 as
Parents’ log household income 11.00 0.40 11.05 0.44 10.96 0.37
Child’s log household income 10.90 0.53 10.96 0.50 10.84 0.55
Note: Athours’ calculations based on GSOEP data, weighted estimates.
26
Table 2. Regression estimates of intergenerational economic mobility, all cohorts
Income measure
All children Only sons Only daughters
Absolute
mobility
Relative
mobility
Absolute
mobility
Relative
mobility
Absolute
mobility
Relative
mobility
Panel A: Intergenerational rank mobility
Individual labor earnings 0.379***
(0.024)
0.242***
(0.045)
0.310***
(0.031)
0.380***
(0.059)
0.423***
(0.040)
0.154**
(0.068)
447 246 201
Household pre-tax income 0.393***
(0.024)
0.214***
(0.040)
0.410***
(0.034)
0.180**
(0.058)
0.380***
(0.033)
0.241***
(0.056)
565 281 284
Panel B: Intergenerational income mobility
Individual labor earnings 6.34***
(0.734)
0.368***
(0.070)
7.78***
(0.743)
0.265***
(0.071)
5.36***
(1.368)
0.420***
(0.130)
447 246 201
Household pre-tax income 7.52***
(0.520)
0.274***
(0.050)
8.47***
(0.650)
0.185**
(0.063)
6.66***
(0.818)
0.353***
(0.080)
565 281 284
Note: Authors’ calculations based on the GSOEP data, all cohorts pulled together. For each specification of the sample, the first line provides the estimated coefficients from
the OLS regression model, where children’s economic outcomes are regressed on respective economic outcomes of their parents, the second line lists robust standard errors of these
estimates, and the third line indicates the sample size. * means significant at 0.05 level, ** means significant at 0.01 level, and *** means significant at 0.001 level.
27
Table 3. Transition matrices for individual gross labor earnings
Note: The quantiles in Panels B and C are defined separately for each gender.
29
Figure 3. Estimates of intergenerational mobility for individual earnings, by rolling
cohorts
Figure 4. Estimates of intergenerational mobility for household pretax income, by
rolling cohorts
0.1
.2.3
.4.5
Para
mete
r estim
ate
1968-7
2
1969-7
3
1970-7
4
1971-7
5
1972-7
6
1973-7
7
Year born
Relative Absolute
Panel A: Rank mobility
56
78
Absolu
te e
arn
ing
s
0.1
.2.3
.4.5
Earn
ings e
lasticity
1968-7
2
1969-7
3
1970-7
4
1971-7
5
1972-7
6
1973-7
7
Year born
Relative Absolute
Panel B: Earnings mobility0
.1.2
.3.4
.5P
ara
mete
r est
imate
1968-7
2
1969-7
3
1970-7
4
1971-7
5
1972-7
6
1973-7
7
Year born
Relative mobility Absolute mobility
Panel A: Rank mobility
66.5
77.5
Absolu
te in
com
e
0.1
.2.3
.4.5
Inco
me e
last
icity
1968-7
2
1969-7
3
1970-7
4
1971-7
5
1972-7
6
1973-7
7
Year born
Relative Absolute
Panel B: Income mobility
30
Table 5. Changes in the estimates of intergenerational economic mobility over time, entire sample
Individual labor earnings Household pre-tax income
Estimates Model 1 Model 2 Model 3 Model 1 Model 2 Model 3
Panel A: Intergenerational rank mobility
Constant 0.379***
(0.024)
0.477***
(0.051)
0.427***
(0.033)
0.393***
(0.023)
0.438***
(0.047)
0.402***
(0.031)
Father’s rank 0.242***
(0.045)
0.045
(0.096)
0.145*
(0.063)
0.214***
(0.040)
0.123
(0.081)
0.195***
(0.052)
Year born -0.019*
(0.008)
-0.009
(0.008)
Cohort -0.114*
(0.048)
-0.023
(0.047)
Father’s rank*year 0.038*
(0.015)
0.019
(0.014)
Father’s rank*cohort 0.228*
(0.089)
0.045
(0.083)
Panel B: Intergenerational income mobility
Constant 6.34***
(0.0734)
8.82***
(1.748)
7.85***
(1.235)
7.52***
(0.520)
8.45***
(0.936)
7.55***
(0.712)
Father’s log earnings 0.368***
(0.070)
0.147
(0.166)
0.230*
(0.117)
0.274***
(0.050)
0.189*
(0.091)
0.272***
(0.069)
Year born -0.413
(0.263)
-0.188
(0.162)
Cohort -2.50
(1.540)
-0.084
(1.037)
Father’s log earnings*year 0.036
(0.025)
0.017
(0.016)
Father’s log earnings*cohort 0.225
(0.146)
0.003
(0.101) Note: Model 1 provides baseline estimates from Table 2. Model 2 tests for the presence of a linear trend in the estimates of intergenerational mobility by year of child’s birth. Model
3 tests for the significance of the change in the estimates of intergenerational mobility between two cohorts of children – those born in 1968-1972 and those born in 1973-1977.
Standard errors in the parentheses. * means significant at 0.05 level, ** means significant at 0.01 level, and *** means significant at 0.001 level.
31
Appendix A
Table 1. Summary of the studies on intergenerational income mobility in
Germany1
Study The period used
for income
measurement
Age when child’s income
is measured
Age when father’s
income is measured
Elasticity
estimate
Father-son pairs
Couch and
Dunn (1997)
1984 - 1989 Annual earnings,
multiyear average (up to
six years) when sons were
18 years old and more
(the period between 1984
- 1989)
Annual earnings,
multiyear average (up to
six years) for the period
1984 - 1989
0.112
Lillard (2001) 1984 - 1998 Annual earnings,
multiyear average (up to
six years) when sons were
18 years old and more
(the period between 1984
- 1998)
Annual earnings,
multiyear average (up to
six years) when fathers
were up to 65 years old
(the period 1984 – 1998)
0.109
Vogel (2006) 1984 - 2005 Annual earnings,
multiyear average (at
least over five years)
when sons were 25 years
and older
Annual earnings,
multiyear average (at
least over five years)
when fathers were up to
60 years old
0.235
Eisenhauer and
Pfeiffer (2008)
1984 - 2006 Monthly earnings when
sons were between 30 and
50 years old
Monthly earnings,
multiyear average (at
least over five years)
when fathers were
between 30 and 50 years
old
0.282
0.205 (without
multiyear
average of
fathers’
earnings)
Schnitzlein
(2009)
1984 – 2004 Annual earning,s
multiyear average over
the period between 2000
and 2004, when sons
were 30-40 years old
Annual earnings,
multiyear average (at
least over five years
between 1984-2004)
when fathers were 30-55
years old
0.263
Schnitzlein
(2016)
1984 - 2011 Annual earnings,
multiyear average over
the period between 1997
and 2011, when sons
were 35-42 years old
Annual earnings,
multiyear average (at
least over five years
between 1984-1993)
when fathers were 30-55
years old
0.318
Father-daughter pairs
Schnitzlein
(2009)
1984 – 2004 Annual earnings,
multiyear average over
the period between 2000
and 2004, when daughters
were 30-40 years old
Annual earnings,
multiyear average (at
least over five years)
when fathers were 30-55
years old
0.361
Note: All estimates listed in the table are based on data from the German Socio-economic panel.
1 In this paper, we consider only intergenerational mobility of income-related outcomes. For the evidence on
intergenerational mobility of educational and occupational outcomes see, among others, Ermisch et al. (2006),
Heineck and Riphahn (2009), Riphahn and Schiederdecker (2012), and Braun and Stuhler (forthcoming).
32
Appendix B
Table 1: The description of the GSOEP sub-samples
Name of the
sample
Year of
collection
Description Size
Sample A
“Residents in
the FRG”
1984 Includes people living in private households in
the Federal Republic of Germany (FRG), where
the head of the household does not belong to
one of the main groups of foreigners (Turkish,
Greek, Yugoslavian, Spanish or Italian)
4528
Sample B
“Foreigners in
the FRG”
1984 Includes people living in private households in
the FRG, where the head of the household is of
Turkish, Greek, Yugoslavian, Spanish or Italian
origin
1393
Sample C
“German
residents in the
GDR”
1990 Includes people living in private households
where the head of the household is a citizen of
the German Democratic Republic (GDR)
2179
Sample D
“Immigrants”
1994/1995 Includes households in West Germany, in which
at least one household member has moved from
abroad after 1984.
531
Sample E
“Refreshment”
1998 Includes people living in private households in
Germany without any restrictions to their origin
1060
Sample F
“Refreshment”
2000 Includes people living in private households in
Germany without any restrictions to their origin
but with a slightly higher selection probability
for households with a non-German than with a
German head
6043
Sample G
“High income”
2002 Includes private households with a monthly
income of at least 3835 Euros
1224
Sample H
“Refreshment”
2006 Includes people living in private households in
Germany without any restrictions to their origin
1506
Sample I
“Incentive
sample”
2009 Includes people living in private households in
Germany without any restrictions to their origin
1531
Sample J
“Refreshment
sample”
2011 Includes people living in private households in
Germany without any restrictions to their origin
3136
Sample K
“Refreshment
sample”
2012 Includes people living in private households in
Germany without any restrictions to their origin
1526
Migration
sample
2013 Includes immigrants using register information
of the German Federal Employment Agency
2700
Source: Composed by the authors using Haisken-DeNew and Frick (2005) and on-line SOEP Desktop
Figure D.1. Estimates of intergenerational mobility for individual earnings by rolling
cohorts, only sons
Figure D.2. Estimates of intergenerational mobility for individual earnings by rolling
cohorts, only daughters
0.1
.2.3
.4.5
Par
amet
er e
stim
ate
1968
-72
1969
-73
1970
-74
1971
-75
1972
-76
1973
-77
Year born
Relative Absolute
Panel A: Rank mobility
6.5
77.
58
8.5
Abs
olut
e ea
rnin
gs
0.1
.2.3
.4.5
Ear
ning
s el
astic
ity
1968
-72
1969
-73
1970
-74
1971
-75
1972
-76
1973
-77
Year born
Relative Absolute
Panel B: Earnings mobility0
.1.2
.3.4
.5P
aram
eter
est
imat
e
1968
-72
1969
-73
1970
-74
1971
-75
1972
-76
1973
-77
Year born
Relative Absolute
Panel A: Rank mobility
44.
55
5.5
66.
5A
bsol
ute
earn
ings
0.1
.2.3
.4.5
Ear
ning
s el
astic
ity
1968
-72
1969
-73
1970
-74
1971
-75
1972
-76
1973
-77
Year born
Relative Absolute
Panel B: Earnings mobility
35
Appendix E
Figure E.1. Estimates of intergenerational mobility of household pretax income by
rolling cohorts, only sons
Figure E.2. Estimates of the intergenerational mobility of household pretax income by
rolling cohorts, only daughters
0.1
.2.3
.4.5
Para
mete
r est
ima
te
1968-7
2
1969-7
3
1970-7
4
1971-7
5
1972-7
6
1973-7
7
Year born
Relative Absolute
Panel A: Rank mobility
7.5
88.5
9A
bso
lute
inco
me
0.1
.2.3
.4.5
Inco
me e
last
icity
1968-7
2
1969-7
3
1970-7
4
1971-7
5
1972-7
6
1973-7
7
Year born
Relative Absolute
Panel B: Income mobility
0.1
.2.3
.4.5
Para
mete
r est
imate
1968-7
2
1969-7
3
1970-7
4
1971-7
5
1972-7
6
1973-7
7
Year born
Relative Absolute
Panel A: Rank mobility
4.5
55.5
66.5
7A
bso
lute
inco
me
0.1
.2.3
.4.5
Inco
me e
last
icity
1968-7
2
1969-7
3
1970-7
4
1971-7
5
1972-7
6
1973-7
7
Year born
Relative Absolute
Panel B: Income mobility
36
Appendix F
Table F.1. Changes in intergenerational mobility of individual labor earnings, by gender sub-groups
Estimate Only sons Only daughters
Model 1 Model 2 Model 3 Model 1 Model 2 Model 3
Constant 0.310***
(0.031)
0.361***
(0.064)
0.338***
(0.041)
0.423***
(0.040)
0.546***
(0.088)
0.471***
(0.059)
Father’s rank 0.380***
(0.059)
0.277*
(0.119)
0.324***
(0.079)
0.154**
(0.068)
-0.092
(0.149)
0.058
(0.100)
Year born -0.011
(0.011)
-0.023
(0.013)
Cohort -0.071
(0.063)
-0.104
(0.078)
Father’s rank*year 0.021
(0.021)
0.045*
(0.022)
Father’s rank*cohort 0.143
(0.119)
0.208
(0.136)
Constant 7.78***
(0.743)
7.97***
(1.580)
8.17***
(1.225)
5.36***
(1.368)
9.24**
(3.142)
6.63**
(2.081)
Father’s log earnings 0.265***
(0.071)
0.255
(0.152)
0.230*
(0.118)
0.420***
(0.130)
0.057
(0.297)
0.305
(0.196)
Year born -0.036
(0.240)
-0.634
(0.520)
Cohort -0.690
(0.521)
-1.996
(2.958)
Father’s log earnings *year 0.001
(0.023)
0.060
(0.049)
Father’s log earnings *cohort 0.059
(0.146)
0.182
(0.282)
Number of obervations 246 246 246 201 201 201
Source: SOEP data, authors’ calculations.
Note: Model 1 provides baseline estimates from Table 4.1. Model 2 tests for the presence of a linear upward trend in the estimates of intergenerational mobility by year of
child’s birth. Model 3 tests for the presence of a linear upward trend in the estimates of intergenerational mobility between two cohorts of children – those born in 1968-1972 and
those born in 1973-1977. * means significant at 0.05 level, ** means significant at 0.01 level, and *** means significant at 0.001 level.
37
Table F.2. Changes in intergenerational mobility of pretax household income, by gender sub-groups
Estimate Only sons Only daughters
Model 1 Model 2 Model 3 Model 1 Model 2 Model 3
Constant 0.410***
(0.034)
0.428***
(0.067)
0.405***
(0.044)
0.380***
(0.033)
0.439***
(0.069)
0.391***
(0.045)
Parental rank 0.180***
(0.058)
0.143
(0.111)
0.189*
(0.073)
0.241***
(0.056)
0.123
(0.117)
0.218**
(0.076)
Year born -0.004
(0.012)
-0.011
(0.011)
Cohort 0.013
(0.068)
-0.027
(0.065)
Parental rank*year 0.008
(0.021)
0.023
(0.020)
Parental rank*cohort -0.026
(0.121)
0.053
(0.113)
Constant 8.47***
(0.650)
8.29***
(1.159)
8.11***
(0.943)
6.66**
(0.818)
8.91***
(1.534)
7.08***
(1.087)
Parental log earnings 0.185**
(0.063)
0.207
(0.112)
0.221*
(0.091)
0.353***
(0.080)
0.141
(0.150)
0.314**
(0.106)
Year born 0.027
(0.174)
-0.417
(0.274)
Cohort 0.728
(1.276)
-0.859
(1.646)
Parental log income *year -0.004
(0.017)
0.039
(0.026)
Parental log income *cohort -0.074
(0.123)
0.079
(0.161)
Number of obervations
Source: SOEP data, authors’ calculations.
Note: Model 1 provides baseline estimates from Table 4.1. Model 2 tests for the presence of a linear upward trend in the estimates of intergenerational mobility by year of
child’s birth. Model 3 tests for the presence of a linear upward trend in the estimates of intergenerational mobility between two cohorts of children – those born in 1968-1972 and
those born in 1973-1977. * means significant at 0.05 level, ** means significant at 0.01 level, and *** means significant at 0.001 level.
38
Appendix G
Table G.1. Sensitivity analysis for the treatment of zero values in earnings, all cohorts
Sample specification
All children
Absolute
mobility
Relative
mobility
Excluding observations with earnings below 1200
Euros (Main sample)
0.379***
(0.024)
0.242***
(0.045)
447
Excluding observations with zero earnings 0.385***
(0.024)
0.230***
(0.045)
455
Including observations with zero values 0.416***
(0.023)
0.167***
(0.043)
536
Excluding observations with earnings below 1200
Euros (Main sample)
6.34***
(0.734)
0.368***
(0.070)
447
Excluding observations with zero earnings 6.45***
(0.777)
0.354***
(0.074)
455
Recording zero earnings to 1 Euro 8.39***
(0.931)
0.093
(0.089)
536
Note: Authors’ calculations based on the GSOEP data, all cohorts pulled together. For each specification
of the sample, the first line provides the estimated coefficients from the OLS regression model, where children’s
economic outcomes are regressed on respective economic outcomes of their parents, the second line lists standard
errors of these estimates, and the third line indicates the sample size. * means significant at 0.05 level, ** means
significant at 0.01 level, and *** means significant at 0.001 level.
39
Table G.2. Sensitivity analysis for the life-cycle bias
Age of children used for earnings
measurement
All children
Absolute
mobility
Relative
mobility
Number of
observations
Panel A: Intergenerational rank mobility
30-32 0.396***
(0.023)
0.207***
(0.042)
557
34-36 (the main sample) 0.379***
(0.024)
0.242***
(0.045)
447
35-37 0.371***
(0.026)
0.258***
(0.048)
402
36-38 0.378***
(0.028)
0.244***
(0.051)
351
37-39 0.385***
(0.031)
0.230***
(0.057)
297
Panel B: Intergenerational earnings mobility
30-32 7.51***
(0.642)
0.252***
(0.061)
557
34-36 (the main sample) 6.34***
(0.734)
0.368***
(0.070)
447
35-37 6.27***
(0.772)
0.380***
(0.073)
402
36-38 6.16***
(0.869)
0.387***
(0.082)
351
37-39 5.51
(1.064)
0.448***
(0.101)
297
Note: Authors’ calculations based on the GSOEP data, all cohorts pulled together. For each specification of the
sample, the first line provides the estimated coefficients from the OLS regression model, where children’s
economic outcomes are regressed on respective economic outcomes of their parents and the second line lists
standard errors of these estimates. * means significant at 0.05 level, ** means significant at 0.01 level, and ***
means significant at 0.001 level. The results from the baseline model are in bold.
40
Table G.3. Sensitivity analysis for the attenuation bias
Number of observations used for
averaging of earnings
All children
Absolute
mobility
Relative
mobility
Number of
observations
Panel A: Intergenerational rank mobility
3 observations for children and 5 for
fathers
0.379***
(0.024)
0.242***
(0.045)
447
1 observation for children and 5 for
fathers
0.374***
(0.025)
0.251***
(0.047)
432
2 observations for children and 5 for
fathers
0.382***
(0.024)
0.234***
(0.044)
464
3 observations for children and 1 for
fathers
0.386***
(0.026)
0.227***
(0.047)
424
3 observations for children and 3 for
fathers
0.390***
(0.024)
0.218***
(0.044)
477
Panel B: Intergenerational earnings mobility
3 observations for children and 5 for
fathers
6.34***
(0.734)
0.368***
(0.070)
447
1 observation for children and 5 for
fathers
5.81***
(0.835)
0.415***
(0.079)
432
2 observations for children and 5 for
fathers
5.76***
(0.857)
0.414***
(0.070)
464
3 observations for children and 1 for
fathers
7.83***
(0.754)
0.231***
(0.072)
424
3 observations for children and 3 for
fathers
7.28***
(0.679)
0.280***
(0.065)
477
Note: Authors’ calculations based on the GSOEP data, all cohorts pulled together. For each specification of the
sample, the first line provides the estimated coefficients from the OLS regression model, where children’s
economic outcomes are regressed on respective economic outcomes of their parents, the second line lists standard
errors of these estimates, and the third line indicates the sample size. * means significant at 0.05 level, ** means
significant at 0.01 level, and *** means significant at 0.001 level. The results from the baseline model are in bold.
41
Endnotes
i Because in the SOEP a reference year for income measurement is always the previous year, we have moved all
income variables by one year to eliminate the time mismatch between the age and reported incomes. Hence, income
reported in 2014 was linked to the age reported in 2013. ii These findings, however, do not imply that absolute outcomes of girls are better than absolute outcomes of boys
because the ranks are defined within the gender-specific distributions of income. iii In his study on intergenerational mobility in Germany Schnitzlein (2009) find that the intergenerational elasticity
of earnings constitutes 0.26 for sons and 0.36 for daughters. Higher estimates of intergenerational earnings
persistence for daughters than for sons were also found in Bratberg et al. (2005).