Cumulative inequality along the life course Long

Steffen Hillmert
Cumulative inequality along the life course
Long-term trends on the German labour market
ESOC Working Paper 1/2010
Steffen Hillmert
Department of Sociology
University of Tübingen
Wilhelmstr. 36
72074 Tübingen
Germany
[email protected]
Abstract:
Cumulative inequalities in life courses mean that relatively small individual (dis-)advantages at earlier stages
generate larger (dis-)advantages at later stages. This paper discusses conceptual links between individual
mobility and developments of inequality along the life course, with a special focus on the question to what extent
employment careers are characterised by forms of cumulative advantage and disadvantage. Moreover, it is asked
how intra-generational developments are related to inter-generational social mobility, and the paper also
discusses possible associations between inequality patterns and specific institutional configurations. In its
empirical part, the paper presents descriptive life-course analyses based upon data from a broad range of West
German birth cohorts and covering a period of approximately 60 years. On the basis of two longitudinal datasets
(survey and register data), it is investigated to what extent the employment careers of both men and women in
Germany have been characterized by cumulative developments of advantage and disadvantage and how these
developments have changed across cohorts.
Steffen Hillmert
Cumulative inequality along the life course: Long-term trends on the German labour
market
1. Introduction
Various theories (from both the supply- and the demand side) and empirical evidence support
the assumption that such cumulative advantages and disadvantages play a major role in
labour-market careers and characterize particular mobility regimes in country-specific labour
markets. Apart from such possible system specificities, however, successive cohorts in most
Western societies have also been confronted with very different labour-market conditions
since World War II. These were characterized by economic growth, educational expansion
and occupational upgrading as well as increasing labour-market participation of women.
During the last few decades, cohort members have experienced growing uncertainty reflected
in delayed labour-market entry and increasing employment risks. The question is how these
changes have modified the extent and the patterns of cumulative (dis-)advantage in
employment careers. In particular, have these uncertainties blurred the boundaries of
established stratification and counteracted the traditional forms of cumulative inequality? The
case of Germany is particularly interesting in this regard, as the changes in labour-market
conditions have been accompanied by long-term institutional stability in the form of
standardized (selective) systems of education and occupation, collective wage bargaining
combined with compressed wages, and coordinated employment relations.
On the basis of two longitudinal datasets (survey and register data), this descriptive paper
investigates to what extent the employment careers of both men and women in Germany have
been characterized by cumulative developments of advantage and disadvantage; how these
developments have changed across cohorts and how this has been related to the historical
development of employment insecurity and other changes in labour-market conditions. The
basis is a multi-dimensional concept of measuring cumulative advantage in labour-market
careers. For a longer sequence of cohorts, trajectories of absolute and relative inequalities are
analyzed in terms of income and occupational social status. The historical period covered by
the various analyses ranges approximately from 1940 to 2005. Cumulative inequality is
2
measured in the form of inequality trends within a cohort; a further reference is the individuallevel stability of positions.
The paper starts with a discussion of conceptual links between intra-generational mobility and
both the development of inequality along the life course and the transmission of social status
across generations (section 2). Connections with questions of cumulative (dis-)advantage are
of special interest. Section 3 discusses theoretical links with specific institutions and historical
circumstances. The central hypothesis is that specific institutional configurations and labour
market developments affect mobility patterns as well as the accumulation of advantage and
disadvantage in individual careers. In its empirical part, the paper presents evidence from
cohort-specific analyses based on German life-course data which is introduced in section 4.
These analyses (section 5) look at age-related developments of occupational status and earned
income within a cohort and at individual-level stability of positions and upward and
downward mobility. Section 6 draws some conclusions for further research.
2. Conceptual considerations
2.1 Inter-generational and intra-generational social mobility
Intra-generational mobility can be regarded to be a function of opportunity structures and
individual efforts of career advancement (Sørensen 1975). In principle, upward and
downward mobility may reflect social advantage and disadvantage, but in order to prevent
wrong conclusions, the results of mobility analyses need to be interpreted in close association
with analyses of the absolute status levels on which mobility processes start. Another
possibility to take both absolute levels of inequality and questions of stability into account is
to use cumulative measures of social advantage, e.g. life-time income. Inequalities at any
given point in time may then be offset by mobility. When combining analyses of intragenerational mobility with those of inter-generational mobility, status levels are compared not
only within a career but also between an individual’s career and his or her social origin – i.e.,
in most cases, the parental status level. Depending on when status is measured, one may find
downward, upward or no inter-generational mobility while observing always the same life
course. This means that the timing of measurement may be decisive for the degree of intergenerational stability and mobility; in a strict sense, the issue of timing applies to the
measurement of both children’s and parents’ status.
3
There are also theoretical links between inter-generational and intra-generational mobility.
Following the normative idea of securing family status it is likely that children who have not
(yet) attained their parents’ status will have a particular intention of moving upward in order
to reach (at least) this status level. This idea is in line with theoretical arguments emphasizing
the relevance of perceived gains and losses rather than final assets (Kahneman and Tversky
1979). Such movements may be called „counter-mobility“, i.e. the “work-life movement
which has the effect of returning an individual back to his class of origin, following some
initial shift away on his entry into employment, and which thus serves to promote
intergenerational stability” (Goldthorpe et al. 1987: 54). As especially persons from higherclass background are rather unlikely to enter immediately in a position which is equal to their
origin level, inter-generational stability depends to some extent on the existence of such
processes. An alternative explanation of career mobility highlights the long-term benefits of
(transient) phases of overqualified employment (Sicherman and Galor 1990).
2.2 Cumulative (dis-)advantage
There are very different terminologies and conceptual frameworks in which intra-generational
change and career developments are studied. However, on the measurement level, most of the
relevant questions can be framed within one of three basic perspectives on the dynamics of
intra-cohort inequality, each of them with specific analytical potential: Mobility analyses look
at the individual-level dynamics or stability of positions. In distributional analyses, status
distributions are compared at various points in time, whereby aggregate measures of
dispersion and inequality are of particular interest. Yet another aspect is represented by
analyses of co-variation looking at associations between individual status positions at various
points in time – or put differently, the predictability of later stages by earlier stages.
A specific concept which can be associated with both aspects of mobility and absolute status
levels is the idea of cumulations of advantage and disadvantage along the life course. The
Cumulative Advantage concept (also called the “Matthew Effect”) describes the fact that
small status differences at the beginning of a career tend to result in even greater differences
later on. One reason is that initial small advantages can foster early success, providing an
actor disproportionally with new resources and by these means generating growing
advantages at later points in time. The basic concept derived from an analysis of careers in
science (Merton 1968). Cumulative Advantage (CA) – or Cumulative Disadvantage – has
proven to be a simple, but fruitful background concept describing time-dependent processes in
4
different research fields like crime, poverty and human development (e.g., Sampson and Laub
1997; Halleröd and Bask 2007; Dannefer 2003). However, to be applicable in empirical
analyses, the concept needs to be specified as it is rather heterogeneous (DiPrete and Eirich
2006; O’Rand 2009). In the following descriptive, macro-level analyses, three definitions and
measurement concepts of cumulative (dis-)advantage along the life course are distinguished:
(1) Social closure: This (weak) definition of cumulative inequality is directly related to
mobility. It implies that status changes become increasingly difficult over time as mobility
decreases. In other words, the status order which represents advantage and disadvantage
becomes increasingly consolidated.
(2) Collective polarisation: This concept means that intra-cohort differentiation of advantage
and disadvantage – as measured by the (cross-sectional) status distribution at any given point
in time – increases over time.
(3) Selective accumulation: This perspective analyses to what extent intra-cohort
differentiation represented by the distribution of individually accumulated status increases
over time. Such differences can be compared in absolute terms, so that developments can be
described as proportionate or disproportionate with regard to steady accumulation (cf. DiPrete
and Eirich 2006). Career interruptions are part of cumulative measures. For results based on
this definition of cumulative inequality, individual-level path dependencies, i.e. the fact that
later mobility depends on previous experiences, may therefore be decisive. This applies to
both path-dependent mobility and path-dependent interruptions or gaps in the relevant process
which may result from the experience of unemployment and non-activity. Such path
dependencies determine to which degree labour-market insecurities are reflected in (interindividual) selective accumulation.
When comparing these definitions of CA with intra-generational mobility, it becomes obvious
that it can itself be regarded to be one dimension of CA and that it directly influences the
other two. Changes in the (cross-sectional) status distribution necessarily entail some degree
of (absolute) individual-level mobility, and mobility as well as possible interruptions
determine to what extent individual-level status accumulates over time. Again, these concepts
can be linked to questions of inter-generational mobility. There is no necessary relationship
between (intra-generational) cumulative advantage and inter-generational social reproduction,
5
but social origin tends to have a major effect already on the entry level at the beginning of
individual careers; hence, when there is cumulative advantage resulting from selective
positioning at labour market entry, inter-generational stability will be perpetuated or even
enforced.
3. Institutional and historical aspects
3.1 Macro-level determinants of cumulative inequality and mobility
Both mobility and CA are formal (measurement) concepts which abstract from underlying
causal processes, but different theories support the assumption that cumulative advantages or
disadvantages play a prominent role in labour market careers and characterise a specific
mobility regime in the labour market. This applies to both supply- and demand-side
perspectives. From a perspective of individual resources, it can be argued that the economic,
social and cultural capital associated with higher positions provides the basis for further
promotion by more information on vacancies, larger networks, greater familiarity with new
situations and financial security protecting against the risks of job changes. In particular,
human capital theory (Becker 1964) states that timing of investment in employees is
important: the investment in younger persons (in situations of recruitment or promotion) is
more promising given higher rates of return and larger possibilities for development. For
example, Elman and O´Rand (2004) describe that income curves are flatter independent of
work experience, if degrees are achieved later in life time; the rate of return for education is
lower then – or in other words, room for mobility is limited. Cunha et al. (2005) further
investigate questions on the optimal timing of educational investments. Referring to
Signalling Theory (Spence 1979) and Rosenbaum’s (1979) theory of tournament mobility it
can be expected that when recruiting and promotion decisions are made under uncertainty,
previous success is taken as an indicator for ability and future success, especially when there
is a small degree of regulation. Given such mechanisms, it seems obvious that not all forms of
labour markets are equally susceptible for cumulative advantage. Major macro-level
differences can be attributed to institutional differences and economic conditions.
Type of the labour market: In qualification-based labour markets (Maurice et al. 1982)
cumulative inequality of type (2) can hardly develop because careers proceed in specific
occupational fields with limited career ladders. Labour market coordination is especially
structured by occupation-specific qualification at the time of job entry and in situations of
promotion. Moreover, collective agreements in coordinated market economies (Hall and
6
Soskice 2001) often lead to relatively compressed wages. Under such conditions, reduced
status mobility will rather foster cumulative inequality of type (3). Employees achieving a
high position at the beginning of their career will remain in high positions over the life-course
and hence accumulate advantage. In contrast to this, in organisational labour markets workers
experience more upward and downward mobility which reflect increasingly less differences in
educational qualifications (Büchtemann et al. 1994). On the other hand, individual success
regardless of formal qualifications may be rewarded more, and this might rather increase
inequalities between persons with different entry positions.
Educational systems and their interplay with labour market institutions: The concept of
cumulative inequality describes that small initial differences tend to grow along the life
course. The main question for analysing intergenerational social mobility is how these initial
differences are generated. Educational level is a crucial factor for careers in all labour
markets, but the associations between education and position in the employment system vary
by educational systems and transition regimes (Hillmert 2001; 2002). Highly stratified
educational systems with a close connection between education and labour market (through
qualificational coordination) strongly transport educational differences to the first labour
market positions. Low stratified systems with weak connection of educational and
employment system create smaller differences among labour market entrants. It can be
assumed that these differences are more easily compensated by other factors. Moreover,
educational systems vary in the degree of social selectivity. The more stratified an educational
system is and the earlier selection processes occur, the more likely it is that social background
affects educational outcome and that there is a close connection between education and first
job (Allmendinger 1989; Müller and Shavit 1998), i.e. the more the social background effects
of education will be perpetuated in the labour market system. Given the relative stability of
occupational careers, it can be expected that the indirect (education-related) effect of social
background is rather constant in qualification-based labour markets. But also in organisational
labour markets background-related resources will create initial and later differences.
Welfare regimes and gender-specific division of labour: Interruptions are crucial determinants
of further careers. Frequency and type of such interruptions are highly connected with
welfare-state regimes (Esping-Andersen 1990), especially when looking at gender differences.
For example, duration and consequences of parental leave are dependent on the provision of
childcare and which welfare institutions exist. The more the re-entry after parental leave is
7
supported by welfare institutions and the less gender-specific labour division is prevalent the
less differences and cumulative developments regarding mobility should be found between
genders.
Economic and demographic conditions: Both have their strongest impact in labour market
segments which are open to market forces. This holds especially for access to apprenticeships
in firms as ports of entry into skilled work and for jobs on external labour markets. In times of
increasing labour market problems both initial differences and social background effects are
likely to increase as a result of intensified competition. The same may result from large cohort
sizes (Easterlin 1980). Times of turbulence in labour markets will also increase the likelihood
of career interruptions due to unemployment, again contributing to the accumulation of
advantage and disadvantage on the individual level.
To sum up, institutional factors favouring cumulative disadvantage in various forms are:
stratified educational and training systems which transfer social inequality to the labour
market and lead to unequal starting positions; labour markets which (disproportionally)
reward individual success (CA type 2); but also labour markets which support status
continuity in employment careers (CA type 3). In labour markets where coordination is
predominantly structured by qualifications, a disproportionate development of (cumulative)
inequalities is less likely, but inequalities may result from a permanent exclusion of labour
market outsiders. Very competitive labour markets may promote absolute cumulative
advantages. The clearer hierarchies in education are and the stronger labour market positions
reflect educational differences, the more indirect social background effects turn into
inequalities in employment. Additional factors affecting accumulation of advantage and
disadvantage are gender differences and economic and demographic conditions leading to
interruptions in individual careers.
3.2 The German case
The following analyses concentrate on one particular case, the situation in Germany, which is
characterised by specific institutions and historical developments in life-course patterns (see
also Mayer and Hillmert 2003; Grunow and Mayer 2007). The German labour market is
characterised by a strong qualification-based segment associated with a differentiated system
of education and training. Occupational boundaries are strong, and occupational mobility
tends to be much lower than job mobility (Hillmert 2001). There is also evidence that the
8
degree of social selectivity in the educational system is relatively high (Shavit and Blossfeld
1993; OECD 2004). The (West) German labour market has been characterised by a high
degree of division of labour between genders (“male breadwinner model”) and a relatively
high degree of gender segregation in employment; for most men, this has meant a high level
of continuity in their employment careers (Kurz et al. 2006), at least after the period of
integration into the labour market. A system of collective wage bargaining has also
contributed to standardisation in employment and comparatively small wage dispersion. With
regard to historical conditions, most notable are the turbulences caused by two world wars,
separate developments in the two German states, and finally their (re-)unification. After an
economically difficult period with high levels of unemployment immediately after World War
II, there was considerable economic growth accompanied by an increasing demand for
workers and occupational upgrading. The labour market in the late 1950s and 1960s was close
to full employment, and this situation affected the labour market entry predominantly of the
cohorts born between 1930 and 1950. Since the mid-1970s, workers have experienced
growing uncertainty reflected in delayed labour market entry and increasing employment risks
(Kurz et al. 2006). Unemployment has again become a growing problem (cf. Figure 1).
Figure 1: Unemployment rates in Germany, 1948 – 2003
14
12
10
8
6
4
2
2002
2000
1998
1996
1994
1992
1990
1988
1986
1984
1982
1980
1978
1976
1974
1972
1970
1968
1966
1964
1962
1960
1958
1956
1954
1952
1950
1948
0
Until 1990: West Germany; excluding Berlin (until 1949) and Saarland (until 1958)
Source: German Labour Office
However, job stability has not declined dramatically (Erlinghagen and Knuth 2004). While
wage inequality was relatively stable in the post-war decades, one can see an increasing
polarization of the income distribution over the recent years, which has been subject of
9
controversy in the recent political debate (Giesecke and Verwiebe 2009). The supply side of
the labour market was characterised by a marked educational expansion (cf. Hadjar and
Becker 2009), affecting in particular the birth cohorts after 1950. Women benefitted in
particular from the expansion of secondary and tertiary education. Not least due to better
qualifications, women have increasingly participated in the labour market. This development
was accompanied by various waves of immigration, in particular by labour migrants from
southern Europe. Together with an increasing proportion of flexible forms of work it buffered
the core of the (male) labour market against excessive turbulence (Blossfeld et al. 2006).
While these developments may have resulted in changes in the relative position of various
groups on the labour market, they have also been accompanied by a remarkable degree of
long-term institutional stability, most notably represented by the (selective) systems of
education and occupation. Regarding the combined effects of institutional stability and
secular trends on the labour market we have the following expectations:
* Basic patterns of intra-general mobility and Cumulative (dis-)advantage: In a
qualificational labour market like in Germany, upper and lower positions are allocated early
in a career, but labour market entry takes place relatively late. Formal qualifications are the
major predictors of occupational positions. After a period of settling in, occupational mobility
is rather low and the following careers are relatively stable. This means that there are only
moderate changes in overall inequality at any given point in time. Stable career patterns lead
to continuous (but not disproportionate) accumulation of advantage and disadvantage on the
individual level, carrying the risk of social exclusion for the disadvantaged.
* Historical changes in mobility and inequality: We expect only moderate changes in the
patterns of life course developments across the cohorts entering the labour market after World
War II. Cohorts entering the labour market during or immediately after the War will
experience a higher level of mobility at the beginning of the careers. After a period of
consolidation, there is also long-term trend towards higher levels of mobility due to increasing
difficulties of finding (adequate) employment after completing education. Moreover, due to
an increasing number of career interruptions, heterogeneity of career patterns has increased.
For women, this may be counteracted by their increasing attachment to the labour force. In
addition to that, particular cohorts may be influenced by specific conditions. For example,
10
large birth cohorts (like the 1964 cohort) may show relatively high levels of cumulative
inequality due to increased competition.
* Gender differences: Women’s careers tend to be less stable than men’s, in spite of (or rather
along with) their increasing participation in the labour market. As a result of educational
expansion and shorter interruptions in women’s careers, we expect increasing similarities
between the inequality patterns of men and women.
* Links to inter-generational mobility: Social origin is important for positioning in the labour
market from the beginning of careers. As a result of both the selective educational system and
the close links between qualifications and labour market positions, its effect mainly works
through education. This indirect effect of social background is relatively stable across the
career. Thus, inter-generational stability is perpetuated, carrying the permanent risk of social
exclusion of marginalised groups. Moreover, the specific individual situation remains decisive
for mobility chances. In particular, when people have not yet attained an “adequate” status
level, i.e. a status that is at least as high as their parents’, such a situation of “status mismatch”
tends to lead to processes of (counter-)mobility.
4. Data and operationalisation
We use two datasets with complementary strengths and weaknesses. Our empirical analyses
base, first, upon the German Life History Study (GLHS), a series of retrospective cohort
studies conducted in various years (1981/83; 1985/86; 1989; 1998/99); for more information
in these studies see also table 1 (including approximate references to important contemporary
events). Using the GLHS as a dataset has specific advantages: It consists of a broad range of
cohorts, so most of the 20th century is covered by the data. The cohort data have been
harmonised and they are of high quality, not least with regard to the collected information on
occupations (Brückner 1993; Brückner and Mayer 1995; Mayer and Brückner 1989; Hillmert
et al. 2004).
11
Table 1: The (West) German Life History Study: birth cohorts and corresponding
historical periods
Birth cohort
Data
collection in...
n (Men)
n (Women)
1919-21
1929-31
1939-41
1949-51
1954-56
1959-61
1964
1971
1985-86
1981-83
1981-83
1981-83
1989
1989
1998-99
1998-99
559
347
375
353
517
482
745
697
837
340
344
361
471
466
700
604
Transition to labour
market during
historical period...
1930/40s
1940/50s
1950/60s
1960/70s
1970s
1970/80s
1980s
1980/90s
Important historical
events/trends
World War II
Immediate Post-war period
Two German states
‘Economic miracle’
Educational expansion
Oil crisis
Mass unemployment
German unification and
beyond
However, our analyses do not use the full dataset of the GLHS. As employment in the
German Democratic Republic was characterised by the conditions of a very different political
and economic system, and consistent with our theoretical considerations, our analyses are
limited to the life courses from West German data. For reasons of consistency among all
cohorts, we restrict our analyses to persons of German nationality. Due to the retrospective
design and different times of interviewing, the cohorts in the GLHS have very different
observation windows. As processes of cumulative advantage and disadvantage may become
relevant only over a longer period in the life course, we restrict some analyses to cohorts with
broader observation windows. The main variable of interest using the GLHS data is social
status in the form of occupational status. It is represented by occupational prestige measured
according to Treiman (SIOPS) (Treiman 1977), a measure which is available for all cohorts.
As measures of dispersion, we use variation coefficients in order to detect possible
disproportionate accumulation (or “strict” CA in the terminology of DiPrete and Eirich 2006).
The main lines of differentiation are gender, cohort, age and social origin. Social origin is
measured as father’s occupational status at respondent’s age 15 (or according to a slightly
different definition). For some of the analyses, social origin was grouped into a dichotomous
variable (above or below the median); this allows making use of the full sample size.
The second data source is a sample from the pension register (VKST 2010) drawn in 2006. It
contains detailed and reliable, process-based (monthly) information on income and is a large
sample (see table 2). However, additional information on individual characteristics (including
social origin) is very limited, and the sample is restricted to workers covered by the public
pension system. For comparisons with the first dataset, birth cohorts were clustered and
analyses are restricted to West Germany. Note that the birth cohorts before 1940 only appear
in the GLHS data.
12
Table 2: Data from the German pension register (VKST 2006 sample)
Birth cohort
N (Men)
N (Women)
1940-44
1945-49
1950-54
1955-59
1960-64
1965-69
1970-74
2231
2400
2581
2796
3190
3323
3278
3240
2872
2836
2960
3315
3489
4034
Transition to labour
market during
historical period...
1950/60s
1960s
1960
1970s
1950/60s
1960/70s
1970s
Analyses with VKST data concentrate on earned income. Earnings are measured indirectly by
pension entitlements (credit points). As the credit points are calculated relative to the average
earned income, this leads to a standardisation of income for every calendar year.
5. Empirical results
5.1 First occupations
Our descriptive analyses first look at long-term historical trends in the patterns of labour
market entry and early careers. Figure 2 compares the distributions of occupational status for
specific cohorts, presenting the range (i.e., the mean +/- 1 standard deviation) for both the
respondents’ first occupation and their occupation at age 30.
Figure 2: Occupational status (mean +/- 1 std) at the beginning of careers, by birth
cohort: Men (left) – women (middle) – by social origin (right)
70
60
50
40
30
20
10
0
Left and middle: Light-coloured bars: First occupation, dark-coloured bars: Occupation at age 30.
Right: First occupation; Light-coloured bars: Low origin, dark-coloured bars: high origin.
In the youngest cohort, respondents had not reached the age of 30 when they were interviewed.
West German Life History Study
13
Over the cohorts, there has been a long-term trend of occupational upgrading, and it is
plausible that such structural changes predominantly affect labour market entrants rather than
mobility processes at later career stages (see also DiPrete et al. 1997). With the 1964 birth
cohort (which was also the largest birth cohort in West Germany) this trend seems to have
come to an end. Also, there appears to have been a growing dispersion of occupational status
in the younger cohorts. Finally, the mean distance in occupational status between the first
occupation and the occupation at age 30 has obviously increased across cohorts, meaning that
there has been an extended period of ‘settling in’ at the beginning of labour market careers.
5.2 Individual-level mobility along the life course
Using intra-generational mobility as an indicator of social closure, the following analyses look
at the likelihood of status changes along the life course. Differences in the status level
between earlier and later occupations as they are indicated by figure 1 necessarily imply some
mobility processes, but this defines rather a minimum level of mobility. An inter-cohort
comparison of mobility is presented in table A1 (in the appendix), models 1, 3, 5 and 6. On
the basis of simple Cox regression models, these longitudinal analyses compare the relative
rates of upward and downward occupational mobility across cohorts, whereby upward and
downward mobility is defined by a change of at least 10 percent. There has been a long-term
non-monotonic trend regarding the level of mobility after labour market entry: The cohorts
born around 1930, 1940 and 1950 had the lowest level of mobility, which increased
significantly afterwards. A notable exception in the long-term development is the cohort of
born around 1920. Their labour market entry was overshadowed by a period of particular
turbulence, the period of the Second World War and the years immediately afterwards, so
they had a relatively high level of career interruptions, postponement, transitory jobs and
mobility as a result. The figures 3 and 4 indicate clearly decreasing upward and downward
mobility with age.
14
Figure 3: Upward mobility: Relative mobility rates (odds ratios) by age and birth
cohort: Men (left) – women (right)
2
1919‐21
1,5
1929‐31
1939‐41
1
1949‐51
1954‐56
1959‐61
0,5
1964
1971
0
25
30
35
40
45
50
55
25
30
35
40
45
50
55
Results from a Cox-regression. See appendix for the parameter estimates (table A2, models 1 and 3).
Upward mobility: >10% increase in occupational status; Odds ratios: 1:= cohort 1929-31, age 25
West German Life History Study
Figure 4: Downward mobility: Relative mobility rates (odds ratios) by age and birth
cohort: Men (left) – women (right)
2
1919‐21
1,5
1929‐31
1939‐41
1
1949‐51
1954‐56
1959‐61
0,5
1964
1971
0
25
30
35
40
45
50
55
25
30
35
40
45
50
55
Results from a Cox-regression. See appendix for the parameter estimates (table A2, models 5 and 6).
Downward mobility: >10% decrease in occupational status; Odds ratios: 1:= cohort 1929-31, age 25
West German Life History Study
This means that there is less and less change in the status order along the life course, which in
that way becomes increasingly consolidated. Occupational mobility steadily declines along
the life course: While there are less and less changes with age, no specific point in time
(“point of occupational maturity”) can be identified for such a consolidation. Additional
analyses indicate that apart from the ages below 30 (where probably initial status mismatch is
15
resolved), workers from higher social origin tend to have lower mobility rates than those from
lower classes.
How is this intra-generational mobility related with inter-generational mobility? A simple
distinction of entry positions by social origin (columns in the right panel of figure 1) indicates
that the labour market careers of people from lower and higher social origin differ from the
very beginning. These differences mainly result from social selectivity in the preceding stages
of education and training. Moreover, our theoretical considerations predict that present
individual status in its relation to parental status is an important determinant of mobility. In
particular, a negative inter-generational status difference (i.e., the situation that an individual’s
present status is lower than his or her parents’ status) is a major driving force of intragenerational mobility. There is considerable potential for such processes: In all cohorts, there
has been a considerable proportion of young men (between 40 and 50 percent) who had a
lower occupational status than their fathers when they entered their first job – in spite of the
trend towards occupational upgrading. Table A2 in the appendix (models 2 and 4) presents the
relative effect that such an inter-generational loss of status has on mobility after labour market
entry. Regarding the magnitude of these effects, it also demonstrates that compared to the
overall long-term trends, such a situation is a very important determinant of individual
(counter-)mobility. Any interactions with cohort and age are not significant. What do the
processes of individual-level mobility mean for inter-generational mobility? As a result of
upward mobility processes, the proportion of workers who experience an inter-generational
loss of occupational status moderately, but constantly declines along the life course.
5.3 Developments of inequality
Our next analyses use indicators of possible collective polarisation. When looking at agerelated developments of the aggregated status distributions in these cohorts (charts not
presented), one can see a moderate increase in mean status along the life course but only a
very moderate increase in dispersion. In fact, variation coefficients are remarkably stable for
both men and women (see figure 5). Structural change in the labour market obviously
becomes salient across cohorts of labour market entrants rather than within cohorts, especially
when looking at the employment careers of men.
16
Figure 5: Variation coefficient of status at age X, by birth cohort: Men (left) – women
(right)
1
1919‐21
0,8
1929‐31
1939‐41
0,6
1949‐51
0,4
1954‐56
1959‐61
0,2
1964
1971
0
25
30
35
40
45
50
55
25
30
35
40
45
50
55
West German Life History Study
Additional analyses combine the data of the cohorts born around 1920 and around 1930, i.e.
the two oldest cohorts with the largest observation windows, to look at the long-term
importance of social origin along the life course. While this combination abstracts from the
differences between these two cohorts, it results in both a large observation window and a
reasonable sample size. When then distinguishing between workers of lower and higher origin
(figure A1 in the appendix), one finds that there is very little change with age in the status
distributions of these two groups. Significant, however, is the relatively clear separation
between the two distributions which persists throughout the life course.
Looking at the dispersion in the aggregate distribution of earned income (figure 6) one can see
a moderate increase with age and an increasing level of inequality across cohorts. Trends
among women are more significant. Note that our indicator of earned income is influenced
not only by the wage dispersion, but also by the heterogeneity of working hours.
17
Figure 6: Variation coefficient of earned income at age X, by birth cohort: Men (left) –
women (right)
1
0,8
1940‐44
1945‐49
0,6
1950‐54
1955‐59
0,4
1960‐64
1965‐69
0,2
1970‐74
0
25
30
35
40
45
50
55
25
30
35
40
45
50
55
Data: VKST 2006
Our next analyses look at status accumulation along the life course. “Accumulated status” is
probably an unusual measure, but it can be a proxy for cumulative processes of advantage and
disadvantage, especially when the retrospective data design makes income information
unreliable. The dimension of the scale is “status times years”. Given the close association of
the status scale with income (as well as other measures of social advantage), there is also a
high correlation with lifetime income. In contrast to the aggregate status distributions, the
accumulated measures are influenced by career interruptions due to unemployment and nonemployment. When calculating the overall distribution of accumulated status until a particular
age (charts not presented), one can see a constant increase in mean cumulated status, so that a
constant variation coefficient represents increasing – but not disproportionally increasing –
absolute differences. Figure 7 presents the relative dispersion of these distributions. The
patterns are rather different for men and women. Variation coefficients for men decrease and
finally consolidate with age, while they constantly increase for women. Heterogeneity of
labour-force attachment is probably the primary reason why the dispersion among women is
rather higher than among men. Note that accumulation starts already before age 25, so
workers who enter the labour market at an early age – who typically have lower qualifications
– have a “starting advantage”, before the higher qualified workers take over between the age
35 or 40. Variation in the age of entry into labour market explains why the level of
heterogeneity starts from a high level and declines in the first few years. After this point,
however, we can see (or expect) a moderate increase of dispersion in all cohorts.
18
Figure 7: Variation coefficient of cumulated status at age X, by birth cohort: Men (left)
– women (right)
1
1919‐21
0,8
1929‐31
1939‐41
0,6
1949‐51
0,4
1954‐56
1959‐61
0,2
1964
1971
0
25
30
35
40
45
50
55
25
30
35
40
45
50
55
West German Life History Study
In combination with the remarkable stability in the aggregate status distributions in figure 5
this result raises the question of how strong individual path dependencies are. Further
mobility analyses (figures not presented) indicate that there are no significant path
dependencies with regard to status mobility. For upward mobility, we find rather the opposite:
previous experiences of (upward) mobility tend to decrease the likelihood of further upward
moves. To some extent, the same applies to interruptions (of any kind) in men’s careers.
However, we do find strong positive path-dependencies regarding the career interruptions of
women. The growing variation across cohorts is probably also due to an increasing
heterogeneity employment risks and interruptions in careers.
Figure A2 in the appendix comparing again two origin groups gives an account of the lifecourse development and its relation to inter-generational reproduction. For a better
interpretation, the chart presents not the absolute values, but the deviations from the overall
mean. The cumulated status distributions of men and women of different social origin become
increasingly separated with when the cohort members become older, especially among
women. This is due to the fact that women’s employment is characterised not only by status
heterogeneity among occupations but particularly in employment stability and labour-force
attachment. Workers from lower social origin – who typically enter the labour market earlier
– have again some “starting advantage” when accumulation status before being overtaken by
persons from higher class origin after age 30.
19
Finally, our analyses turn to income accumulation along the life course. We look at lifetime
earned income up to a certain age and the dispersion of these age-specific distributions (figure
8). Lifetime income is susceptible to processes of unequal accumulation, since it is affected
not only by the dispersion of wages and working hours but also by differences in career
interruptions. Heterogeneity of labour-force attachment is the primary reason why the
dispersion of cumulated income among women is higher than among men. Note that
accumulation starts already before age 25, so workers who enter the labour market at an early
age – who typically have lower qualifications – have a “starting advantage”, before the higher
qualified workers take over between the age 35 or 40. Age differences in labour market entry
also explain why the level of heterogeneity starts from a high level and declines in the first
few years. After this point, however, we see a moderate increase of dispersion in all cohorts.
Figure 8: Variation coefficient of cumulated earned income at age X, by birth cohort:
Men (left) – women (right)
1
1940‐44
0,8
1945‐49
0,6
1950‐54
1955‐59
0,4
1960‐64
0,2
1965‐69
1970‐74
0
25
30
35
40
45
50
55
25
30
35
40
45
50
55
Data: VKST 2006
6. Summary and conclusions
On the basis of these analyses on West Germany, our hypotheses can at least partially be
confirmed. The basic patterns of mobility and cumulative advantage follow the theoretical
expectations. There is little change in the status distributions at any given point in time (i.e.,
no collective polarisation) but a growing dispersion in earned income. There is evidence for at
least constantly cumulating advantage and disadvantage in mid-career employment. In the
course of a career it becomes increasingly unlikely to change the individual occupational
status; in other words, the status order becomes consolidated. What is probably most striking
is the clear stratification of attained occupational status by social origin, which persists or
20
even increases throughout the careers. Such a pattern is typical for an institutional system like
the German system characterised by a differentiated educational system and a qualification
based labour market; a system which makes it likely that the impact of social origin is
transferred to the labour market through education and that it has a long-lasting effect on
employment careers. As pension entitlements in Germany are closely related to employment
careers, it is likely that these inequalities are transferred even beyond retirement.
Historical trends are non-monotonic. The cohorts born around 1930 and 1940 experienced a
minimum of mobility and heterogeneity; the careers of the 1920 birth cohort reflect the
turbulence caused by World War II; from the 1950 birth cohort onwards, there are indications
of growing mobility and (cumulative) inequality. This is due to increasing risks of
employment and interruptions rather than a polarisation among occupations and labour market
positions. Members of any single cohort have obviously been affected by specific conditions,
which have also changed over their lifetime. Further research should therefore look at detailed
series of cohorts in combination with long-term analyses along the life course.
Moreover, analyses of occupational status could be complemented with analyses of other
indicators of (dis-)advantage like assets and housing to allow generalisations regarding
cumulative advantage and disadvantage. Another way of extending these analyses is to
investigate mobility processes in greater detail, e.g. to better distinguish between direct and
indirect job moves, i.e. occupational mobility mediated by phases of unemployment and nonemployment (Grunow and Mayer 2007): As in our cumulative measures, such a perspective
reveals the importance of employment continuity for occupational careers, especially when
(dis-)continuities in careers are path-dependent. Beyond specific empirical results this paper
underlines the relevance of inherent connections among the various conceptual perspectives
of life-course research, in particular analyses of aggregate inequality, mobility and covariation between careers stages.
21
References
Allmendinger, Jutta (1989): ‘Educational Systems and Labour Market Outcomes’, European
Sociological Review 5: 231-250.
Becker, Gary Stanley (1964): Human Capital, New York: Columbia Univ. Press.
Blossfeld, Hans-Peter, Mills, Melinda and Bernradi, Fabrizio (eds.) (2006): Globalization,
Uncertainty and Men’s Careers: An International Comparison, Cheltenham: Edward
Elgar.
Breen, Richard (ed.) (2004): Social Mobility in Europe, Oxford: Oxford Univ. Press.
Brückner, Erika (1993): Lebensverläufe und gesellschaftlicher Wandel: Konzeption, Design
und Methodik der Erhebung von Lebensverläufen der Geburtsjahrgänge 1919 – 1921,
Berlin: Max-Planck-Institut für Bildungsforschung.
Brückner, Hannah and Mayer, Karl Ulrich (1995): Lebensverläufe und gesellschaftlicher
Wandel: Konzeption, Design und Methodik der Erhebung von Lebensverläufen der
Geburtsjahrgänge 1954-1956, 1959-1961, Berlin: Max-Planck-Institut für
Bildungsforschung.
Büchtemann, Christoph F., Schupp, Jürgen and Soloff, Dana J. (1994): ‘From School to
Work: Patterns in Germany and the United States’, in Schwarze, Johannes, Buttler,
Friedrich and Wagner, Gert G. (eds.): Labour Market Dynamics in Present Day
Germany, Frankfurt a. M.: Campus, pp. 112-141.
Cunha, Flavio, Heckman, James J., Lochner, Lance and Masterov, Dimitriy V. (2005):
Interpreting the Evidence on Life Cycle Skill Formation. NBER Working Paper No.
11331, Cambridge.
Dannefer, Dale (2003): ‘Cumulative Advantage/Disadvantage and the Life Course: CrossFertilizing Age and Social Science Theory’, Journal of Gerontology: Social Sciences
6: 327-337.
DiPrete, Thomas A., De Graaf, Paul, Luijkx, Ruud, Tåhlin, Michael and Blossfeld, HansPeter (1997): ‘Collectivist versus Individualist Mobility Regimes? Structural Change
and Job Mobility in Four Countries’, American Journal of Sociology 103: 318-358.
DiPrete, Thomas A. and Eirich, Gregory M. (2006): ‘Cumulative Advantage as a Mechanism
for Inequality: A Review of Theoretical and Empirical Developments’, Annual Review
of Sociology 32: 271-297.
Easterlin, Richard A. (1980): Birth and Fortune: The Impact of Numbers on Personal
Welfare, New York: Basic Books.
Elman, Cheryl and O’Rand, Angela (2004): ‘The Race is to the Swift: Socioeconomic
Origins, Adult Earnings and Wage Attainment’, American Journal of Sociology 110:
123-160.
Erikson, Robert and Goldthorpe, John H. (1992): The Constant Flux: A Study of Class
Mobility in Industrial Societies, Oxford: Clarendon Press.
Erlinghagen, Marcel and Knuth, Matthias (2004): ‘In Search of Turbulence. Labour Market
Mobility and Job Stability in Germany’, European Societies 6: 49-70.
Featherman, David F. and Hauser, Robert (1978): Opportunity and Change. New York:
Academic Press.
Giesecke, Johannes and Verwiebe, Roland (2009): ‘The Changing Wage Distribution in
Germany between 1985 and 2006’, Schmollers Jahrbuch. Zeitschrift für Wirtschaftsund Sozialwissenschaften 129: 191-201.
Goldthorpe, John H., Llewellyn, Catriona and Payne, Clive (1987): Social Mobility and Class
Structure in Modern Britain, Oxford: Oxford Univ. Press.
Grunow, Daniela and Mayer, Karl Ulrich (2007): How Stable are Working Lives?
Occupational Stability and Mobility in West Germany 1940s – 2005, CIQLE Working
Paper 2007-03. Yale University, New Haven.
22
Hadjar, Andreas and Becker, Rolf (eds.) (2009): Expected and Unexpected Consequences of
the Educational Expansion in Europe and USA: Theoretical approaches and empirical
findings in comparative perspective. Bern: Haupt.
Hall, Peter A. and Soskice, David (2001): Varieties of Capitalism: The Institutional
Foundations of Comparative Advantage, Oxford: Oxford Univ. Press.
Halleröd, Björn and Bask, Miia (2008): ‘Accumulation of Welfare Problems in a Longitudinal
Perspective’, Social Indicators Research 88: 311-327.
Hillmert, Steffen (2001): Ausbildungssysteme und Arbeitsmarkt. Lebensverläufe in
Großbritannien und Deutschland im Kohortenvergleich, Wiesbaden: Westdeutscher
Verlag.
Hillmert, Steffen (2002): ‘Labour Market Integration and Institutions: An Anglo-German
Comparison’, Work, Employment & Society 19: 675-701.
Hillmert, Steffen, Künster, Ralf, Spengemann, Petra and Mayer, Karl Ulrich (2004): Projekt
‘Ausbildungs- und Berufsverläufe der Geburtskohorten 1964 und 1971 in
Westdeutschland’. Dokumentationshandbuch. Berlin: Max-Planck-Institut für
Bildungsforschung.
Kahneman, Daniel and Tversky, Amos (1979): ‘Prospect Theory: An Analysis of Decision
under Risk’, Econometrica 47, 263-291.
Kurz, Karin, Hillmert, Steffen and Grunow, Daniela (2006): ‘Increasing Instability in
Employment Careers? Men’s Job Mobility and Unemployment in West Germany’, in:
Blossfeld, Hans-Peter, Mills, Melinda and Bernardi, Fabrizio (eds.), Globalization,
Uncertainty and Men’s Careers: An International Comparison, Cheltenham: Edward
Elgar, pp. 75-113.
Maurice, Marc, Sellier, François and Silvestre, Jean-Jacques (1982): Politique d´éducation et
d´organisation industrielle en France et en Allemagne: Essai d´analyse societal, Paris:
Presses Universitaires de France.
Mayer, Karl Ulrich and Brückner, Erika (1989): Lebensverläufe und Wohlfahrtsentwicklung.
Konzeption, Design und Methodik der Erhebung von Lebensverläufen der
Geburtsjahrgänge 1929-1931, 1939-1941, 1949-1951, Berlin: Max-Planck-Institut für
Bildungsforschung.
Mayer, Karl Ulrich and Hillmert, Steffen (2003): ‘New Ways of Life or Old Rigidities?
Recent Changes in Social Structures and Life Courses in Germany and their Political
Impacts’, West European Politics 26: 79-100.
Merton, Robert K. (1968): ‘The Matthew Effect in Science: the Reward and Communication
System in Science’, Science 199: 55-63.
Müller, Walter and Shavit, Yossi (1998): From School to Work. A Comparative Study of
Educational Qualifications and Occupational Destinations, Oxford: Oxford Univ.
Press.
OECD (2004): Learning for Tomorrow’s World: First Results from PISA 2003. Paris: OECD.
O’Rand, Angela (2009): ‘Cumulative Processes in the Life Course’, in: Elder, Glen H. Jr. And
Giele, Janet Z. (eds.), The Craft of Life Course Research. New York: Guildford Press,
pp. 121-140.
Rosenbaum, James E. (1979): ‘Tournament Mobility: Career Patterns in a Corporation’.
Administrative Science Quarterly 24: 220-241.
Sampson, Robert J. And Laub, John H. (1997): ‘A Life-Course Theory of Cumulative
Disadvantage and the Stability of Delinquency’, in Thornberry, Terence P. (ed.):
Developmental Theories of Crime and Delinquency. New Brunswick: Transaction, pp.
133-161.
Shavit, Yossi and Blossfeld, Hans-Peter (eds.) (1993): Persistent Inequality. Changing
Educational Attainment in Thirteen Countries, Boulder: Westview Press.
23
Sicherman, Nachum and Galor, Oded (1990): ‘A Theory of Career Mobility’, Journal of
Political Economy 98: 169-192.
Sørensen, Aage B. (1975): ‘The Structure of Intragenerational Mobility’, American
Sociological Review 40: 456-471
Spence, Michael (1973): ‘Job Market Signalling’, Quarterly Journal of Economics 87: 355374.
Treiman, Donald (1977): Occupational Prestige in Comparative Perspective. New York:
Academic Press.
VKST
(2010):
Versicherungskontenstichprobe
2006.
http://forschung.deutscherentenversicherung.de/ ForschPortalWeb (15.03.2010)
24
Appendix
Table A1: Risk of upward and downward mobility (Cox-Regression), all cohorts
Modell
Upward
mobility
1
Men
Coeff.
2
SE
Coeff.
SE
.08
Coeff.
SE
.08
.11
.12
.11
.11
.10
.12
.57**
0
.13
.20
.42**
.43**
.57**
.63**
.08
.09
.10
.09
.10
.08
.10
.42**
0
-.11
-.30*
-.22*
.00
-.06
-.08
-.05**
.00
.67**
0
-.12
-.05
.25**
.25*
.33**
.29**
.08
.09
.10
.09
.10
.08
.10
.72**
0
.06
.29**
.60**
.53**
.59**
.51**
Age
-.08**
.00
-.05**
.00
Occupational status
-.07**
Status below father
.56**
13190
2266
478.6
13190
2266
1714.6
4
SE
Cohort 1919-21
Cohort 1929-31 (Ref.)
Cohort 1939-41
Cohort 1949-51
Cohort 1954-56
Cohort 1959-61
Cohort 1964
Cohort 1971
N= (Episodes)
Events
-2*diffLL
3
Women
Coeff.
Data: West German Life History Study
SE
6
Women
Coeff.
SE
.11
.12
.10
.11
.08
.12
.00
.10
.11
.12
.11
.11
.10
.12
.99**
0
-.02
.06
.46**
.55**
.42**
.48**
.12
.14
.11
.12
.11
.13
.67**
0
-.00
-.16
-.03
.15
.23*
-.01
-.07**
.00
-.08**
.00
-.05**
.00
-.07**
.00
.05
.40**
.06
12001
1871
346.7
12001
1871
1502.9
Upward mobility: >10% increase in occupational status; Downward mobility: >10% decrease in occupational status
* p < .05 ** p < .01
Downward
mobility
5
Men
Coeff.
13190
1469
400.2
12001
1441
303.5
Figure A1: Range of occupational status (as mean deviations from the overall mean +/0.5 std) at age x, by social origin: Men (left) – women (right)
15
10
5
Low origin 0
25
30
35
40
45
50
25
30
35
40
45
50
High origin
‐5
‐10
‐15
West German Life History Study, cohorts born around 1920 and 1930
Figure A2: Dispersion in accumulated status (as mean deviations from overall mean +/0.5 std) at age X, by social origin: Men (left) – women (right)
400
300
200
100
Low origin
0
‐100
25
30
35
40
45
50
25
30
35
‐200
‐300
‐400
West German Life History Study, cohorts born around 1920 and 1930
40
45
50
High origin