The Economic Consequences of Divorce in Germany: What Has Changed since the Turn of the Millennium?


The Economic Consequences of Divorce in Germany: What Has 
Changed since the Turn of the Millennium?

Miriam Bröckel, Hans-Jürgen Andreß

Abstract: Our analysis of data from almost 30 waves of the German Socio-Eco-
nomic Panel Study (SOEP) shows that the economic consequences of divorce are 
still more negative for women than for men despite increased female labour force 
participation and, correspondingly, increased numbers of dual earner households. 
After reviewing recent shifts in the institutional fabric and the social structure of 
the conservative German welfare state with respect to families and marriage, the 
empirical analysis investigates the economic consequences of a sample of 844 men 
and 1,006 women in fi ve dimensions: child custody, support payments, housing, 
employment, and economic well-being. Change is measured by comparing data 
from before and after the turn of the millennium.

Overall, the analyses show that the economic consequences of divorce are still 
more negative for women than for men after the turn of the millennium, although 
female labour participation has increased and the public child care system was 
expanded. Nevertheless, some signs of change become visible indicating that the 
gains and losses of marital disruption are not unilaterally distributed among the 
genders. That men may also depend on incomes of their spouses can be seen for 
the increasing number of dual earner households. Moreover, after the turn of the 
millennium, economic dependence on public transfers increased not only for wom-
en but also for men.

Keywords: Divorce consequences · Household income · Gender inequalities · SOEP

1 Introduction

About a decade ago, we provided evidence on changes in material well-being that 
accompany marital disruption in Germany (Andreß et al. 2003; Andreß/Bröckel 
2007).1 Our analyses were among the fi rst to provide a comprehensive picture of 

Comparative Population Studies
Vol. 40, 3 (2015): 277-312 (Date of release: 17.04.2015)

© Federal Institute for Population Research 2015  URL: www.comparativepopulationstudies.de
DOI: 10.12765/CPoS-2015-04en    URN: urn:nbn:de:bib-cpos-2015-04en0

1 At various places in this article, we repeat conclusions from our former work. For reasons of 
readability, we do not cite them explicitly unless they refer to concrete data.



•    Miriam Bröckel, Hans-Jürgen Andreß278

men’s and women’s occupation, income, housing, living conditions, unemployment 
and welfare dependency before and after separation. We found clear-cut gender in-
equalities among the separating men and women. Women took care of children after 
separation more often than men, but were only partly rewarded for their care work 
by support payments from their former spouse. When there were minor children in 
the household after separation, mothers also moved out of their home more often 
than fathers did. The tight economic situation of their households led many women 
to increase their employment after separation, especially women living together 
with children. Most men, on the other hand, were employed full-time during mar-
riage, and had therefore already exhausted their employment capacities. Contrary 
to public opinion, decreased employment among separated men was somewhat 
rare. As a result of all these changes in household composition, support payments, 
employment and residential mobility, women ended up with much lower average 
disposable household incomes than during marriage, especially when taking the 
number of dependents into account. After housing costs had been deducted, wom-
en’s equivalised household incomes one year after separation equalled two-thirds 
of men’s; women lost one-third of their pre-separation household income, while 
men lost a little more than one-tenth (Andreß/Bröckel 2007: 221-222).

These gender inequalities were not unexpected, and were in line with interna-
tional research on the economic consequences of partnership dissolution (Andreß 
et al. 2006: Table 1). Four risk factors are usually identifi ed for women (Holden/
Smock 1991. 68; Sørensen 1994: 173):

1. After separation, if there are children, economic needs are higher for the resi-
dent parent, which in most cases is the mother. Because she has to care for 
more economically dependent household members, she is often restricted in 
her earning capacities, especially if child care facilities are scarce.

2. Moreover, the risk that the resident parent does not receive the necessary 
child maintenance costs from the non-resident liable parent is high.

3. Income sharing during marriage allows women to participate in men’s market 
incomes, which are higher than women’s market incomes on average. It is 
questionable whether the amount of maintenance payments received from 
their former spouse after separation compensates for the reduction in eco-
nomic support compared to what women received during marriage.

4. Finally, changes in material wealth, pension entitlements, and human capital 
that both partners experience during marriage are only imperfectly meas-
urable and therefore hard to equalise after separation.2 This is especially 
true for the gains and losses in human capital. Given the frequent division of 
labour with a husband employed full-time and a wife working part-time (at 
most) who takes care of the household and the children, it is probable that the 
gains in human capital by continued employment on the side of the husband 
(Kenny 1983) are not set off against the losses of human capital resulting from 
interrupted or delayed working careers on the side of the wife.

2 Even if such changes are measurable, the economic loss may be high when they are sold to pay 
off the partner (e.g. in case of property).



The Economic Consequences of Divorce in Germany    • 279

The magnitude of these risks is naturally moderated by the institutional context 
in which marital disruptions take place. For example, with respect to the fi rst risk, 
if the welfare state provides public child care, this would enable resident parents 
to participate in the labour market despite their family obligations.3 Moreover, the 
number of jobs available for both genders that pay a living wage depends on the 
structure of the labour market. Finally, anti-discrimination legislation may reduce 
potential gender wage gaps. Concerning the second risk, the legal system may 
provide measures and enforcement procedures to ensure that non-resident par-
ents pay the necessary maintenance costs for their children and that the better-off 
spouses make support payments for their former spouses, which would minimise 
the third risk. Looking, fi nally, at the fourth risk, legal procedures may exist that 
equalise material wealth and pension entitlements accrued by both partners during 
marriage. Moreover, employment-friendly policies may support women’s continued 
employment during marriage, thus minimising losses of human capital due to inter-
rupted or delayed working careers.

Ten years ago, we argued that the conservative German welfare state model is 
characterised at least historically by strong support for the male breadwinner family 
(Andreß/Bröckel 2007: 196-202). Given this dependence on the male breadwinner 
model, one would expect marital disruption to be much more consequential for 
German women than for women in other countries. On the other hand, since the 
German welfare state relied so heavily on the male breadwinner model, German 
maintenance law at that time included rather extensive support regulations for the 
economically-dependent spouse and children.4 In addition, a rather comprehensive 
system of public transfers cushioned the most severe economic consequences of 
marital disruption. We therefore had less strong expectations, and hypothesised 
that the economic consequences of marital disruption may be negative for German 
women, but not as severe as might be expected from a male breadwinner model 
without safeguards. Indeed, our empirical analyses showed that the negative ef-
fects of the male breadwinner model on material well-being after marital disruption 
were attenuated, but less so by the legal system or the support payments resulting 
from it. It was rather the generous German system of income support that cush-
ioned the most severe negative changes.

Our analysis was based on longitudinal data from 1984-1999. Policies, legal 
regulations and individual behaviour have changed since that time, among them 
more employment-friendly family policies, but also increased female labour force 
participation rates and more dual earner households. This raises the question of 
whether and how the effects of marital disruption have changed since the turn of 
the millennium. In order to answer this question, we add the most recent waves 
(2000-2012) of the German Socio-Economic Panel (SOEP) study to our former data 

3 Cash transfers or vouchers for buying child care services on the market would have the same 
effect.

4 However, it was unclear how these legal regulations were used by the benefi ciaries and how 
they were enforced by the legal system.



•    Miriam Bröckel, Hans-Jürgen Andreß280

base and compare the period before and after 2000. Section 2 summarises the main 
features of the German welfare state model and how it changed in the last years. It 
also presents some stylised facts about recent socio-demographic changes of eco-
nomic activity and household structures. Section 3 describes our data and meth-
odology. Section 4 presents the results of our data analyses. Section 5 concludes 
with an overall assessment of the economic consequences of marital disruption in 
Germany.

2 Theoretical background and research questions

2.1 The traditional German welfare state model

The German welfare state was traditionally characterised by strong support for the 
male breadwinner family. This view is based on the following institutional features 
(Andreß/Bröckel 2007: 196-198):

• joint taxation of spouses imposing strong incentives to combine a large pri-
mary income, that is the breadwinner’s income, with a comparatively small 
secondary income,

• a health insurance system insuring spouses and minors whose attachment 
to the labour market is marginal along with the household head without any 
extra contribution,

• pensions for surviving dependants being derived from the contributions of 
the head of the household without contributions of their own,

• family policies focusing on the traditional complete family and providing 
mostly cash benefi ts such as child allowances and family tax benefi ts that 
allow the average worker to support a larger family,

• a limited public child care infrastructure intended only to complement care 
work within the family, and

• a generous parental leave system motivating mothers to withdraw from the 
labour market for several years and exclusively care for their children.

These regime characteristics were combined with various legal safeguards for 
the economically-dependent family members in case a marriage breaks down (An-
dreß/Bröckel 2007: 199-200):

• Parents have an obligation to support their children until the completion of 
education or training irrespective of their marital status (payments in kind 
by the resident parent and cash payments by the non-resident parent). The 
amount of cash payments depends on the child’s age, the number of entitled 
persons (other children, previous spouse) and the liable person’s income. 
Courts expect that the non-resident parent does everything possible to be 
able to pay the amount of child support that is set.

• Local youth welfare authorities intervene when a parent cannot or will not 
meet these responsibilities by helping to establish paternity, enforcing le-
gal responsibility and advancing maintenance payments at a minimum level 
(“Regelunterhalt”).



The Economic Consequences of Divorce in Germany    • 281

• Spousal support is restricted to specifi c circumstances pertaining to the for-
mer marriage, among them child care, interrupted educational or working 
careers, and a standard of living below the former standard during marriage 
(“Aufstockungsunterhalt”).

• Marriage-specifi c changes in property can be equalised at the request of one 
spouse. Pension entitlements acquired during marriage must be equalised 
by the court (ex offi cio) in case of a divorce.

However, it was unclear whether and to what degree these regulations were 
being enforced by the legal system (Andreß/Bröckel 2007: 201-202). Moreover, it 
was questionable whether the liable persons would be suffi ciently solvent to pay all 
claims even if this were the case. Liable persons are allowed to retain an amount for 
their basic needs, while the claimants’ basic needs are not considered.

2.2 What has changed in the institutional context?

With respect to families’ material well-being, the most important changes are:
1. an upgrading of the public child care infrastructure, especially in Western 

Germany, and especially for children aged under three,

2. a reform of the parental allowance (from “Erziehungsgeld” to “Elterngeld”),

3. a reorientation of minimum income policies with the so-called “Hartz” labour 
market reforms, and

4. increased expectations on women’s self-supply in reformed German mainte-
nance law.

Since 1996, as part of the law on child and youth welfare (Social Code – Sozial-
gesetzbuch – SGB – VIII), parents in Germany have had the legal right to enrol their 
children in a public day care institution. Originally, this right was only granted to chil-
dren aged three and older, but it has been extended to children aged one and older 
since August 1, 2013. The law also states that the Youth Welfare Offi ces (“Träger der 
öffentlichen Jugendhilfe”) of the districts and municipalities, based on the current 
demand for child care, have to provide the necessary infrastructure. Districts and 
municipalities receive money for this task from “their” federal state.

Although child and youth welfare is a task for the federal states in Germany,5 in 
the late 2000s, the German Federal Government launched a major initiative to sup-
port the expansion of the public child care infrastructure. Various legal regulations 
were adapted in 2008 (all collected in an omnibus act known as the Child Promo-
tion Act “Kinderförderungsgesetz”), and an investment programme of about 12bn € 
was launched, of which the Federal Government paid about a third (BMFSFJ 2010). 
An omnibus act (known as the Day Care Expansion Act “Tagesbetreuungsausbaug-
esetz”) was passed as early as 2005, and this committed local authorities to provide 
an additional 230,000 child care places by 2010, especially in Western Germany. It 

5 It is administered at local level (districts and municipalities), but fi nancial support comes from 
the respective federal state.



•    Miriam Bröckel, Hans-Jürgen Andreß282

was decided in 2007 that there should be enough places to provide 35 percent of all 
children below the age of three with a care place by 2013.

Consequently, the German public child care infrastructure has greatly improved 
since the turn of the millennium in terms of size (number of places) and scope (age 
of children and opening hours) (see Table 1). Nevertheless, the number of places 
available is still considered insuffi cient, especially in large cities in Western Ger-
many. The discrepancy between supply and demand has always been less severe 
in Eastern Germany due to the extensive child care infrastructure that had been 
built up in the former German Democratic Republic (GDR) and then taken over after 
reunifi cation. Instead, a lack of job opportunities for women, due to the economic 
down-turn after reunifi cation and later (comparatively) high unemployment rates, 
was more detrimental to women’s careers in Eastern Germany than employment 
restrictions due to family obligations (see Section 2.3).

With its strong focus on the male breadwinner model, the former (Western Ger-
man) Federal Republic of Germany (FRG) traditionally only needed public child care 
to complement mothers’ care work on a part-time basis, while due to its socialistic 
ideology and labour shortages, the former GDR wanted women to participate in the 
labour market full-time and hence provided full-time child care from the very be-
ginning. Today, with Western Germany catching up on Eastern German standards, 
there is no doubt that there is much more public support for child care than before 
2000. All in all, the majority of political actors is convinced that the State should pro-
vide more opportunities for parents, and especially for mothers to continue employ-
ment despite child care obligations. This new policy orientation is also motivated 
by the understanding that the educational resources of some families (e.g. with 
respect to language education) are rather limited and should be supported in public 
child care facilities. However, the cultural dispute on how to organise child care in 
a modern world is far from being settled. Parts of the conservative political parties 
remain reluctant to provide more public child care, and are also demanding support 
for those who do not use this infrastructure (see the so-called “Betreuungsgeld” in 
footnote 7).

Starting from July 1979, parental care itself was remunerated with a special al-
lowance (labelled as “Erziehungsgeld” from 1986 onwards) which provided a lump 
sum if the parent did not work more than part-time (max. 30 weekly working hours). 
Both the amount and the duration of the allowance have been increased in several 
steps since its introduction,6 and it was not counted against any minimum income 
support payments (e.g. social assistance “Sozialhilfe”). Politically, “Erziehungsgeld” 
was meant as a remuneration of the care work done at home by parents them-
selves. Together with the comparatively long periods of parental leave (up to three 
years), it was considered to have strong negative effects on the female labour sup-

6 Starting from January 1, 2004 (the most recent change), one could either receive 300 € for 24 
months or 450 € for 12 months. On top of these payments, some federal states (Bavaria, Saxony 
and Thuringia; all traditionally governed by Christian Democratic parties) made additional sup-
port payments.



The Economic Consequences of Divorce in Germany    • 283

ply (e.g. Aisenbrey et al. 2009; Drasch 2013), and hence to increase the economic 
risks after marital disruption for the caring parent, which in most cases was the 
mother. Moreover, it was criticised that such a low transfer income would not moti-
vate better-educated women with higher labour incomes to have children. But it is 
especially this group of women that only has few children, if any. The “Elterngeld” 
was therefore introduced in 2007, replacing the former lump sum allowance with an 
allowance that is dependent on the labor income of the applying parent and is paid 
for one year immediately after the child’s birth.7 “Elterngeld” is meant to replace 
labour income, and in doing so to also motivate high-income earners to have chil-
dren. On the other hand, it is counted against minimum income support payments 
(in the beginning only partly, since 2011 completely), which compared to the for-
mer “Erziehungsgeld” is a clear impairment for low-income earners. Nevertheless, 
“Elterngeld” is a success story with a rising labour force participation of mothers 
with children aged between 1-2, namely from 33 percent in 2007 to 40 percent in 
2010 (Wrohlich et al. 2012). 

Germany also saw a major reform of its labour market and minimum income 
policies in 2004. Before that time, there was a separate minimum income payment 
for the long-term employed (unemployment assistance “Arbeitslosenhilfe – ALH”), 
which was merged with the general minimum income programme (social assistance 
“Sozialhilfe”) in 2004 to form a common programme for all employable individuals. 
While the former ALH provided a monthly payment that was indexed to former 
earnings, the new “Arbeitslosengeld II – ALG II” provides monthly payments at the 
level of “Sozialhilfe”. Moreover, both programmes are means-tested, i.e. they take 
partner incomes and economic wealth into account, but the exemptions with regard 
to partner income and wealth are much stricter in the ALG II programme. Even more 
importantly, besides means testing and lower payments, expectations concerning 
recipients’ efforts to fi nd new work, even at a lower level, have increased dramati-
cally. Recipients have to sign target agreements with the job centre which can be 
enforced by lowered or interrupted payments in case of non-compliance. This “car-
rot and stick” policy has increased pressure to take up remunerative employment 
for all individuals who are currently unemployed but are in principle employable. 
Since our former analysis of marital disruption showed that the worst economic 
consequences were cushioned by minimum income payments (Andreß/Bröckel 
2007: 223), this policy change may also be effective for divorcees.

A similar change has taken place in family law. As mentioned in Section 2.1, a 
divorcee can request spousal support from the former spouse. Maintenance claims 
are restricted to specifi c circumstances of the former marriage, the most important 

7 If the other parent also interrupts his/her employment to care for the child, the payment period 
is prolonged by two additional (partner) months. An additional lump sum allowance (“Betreu-
ungsgeld”) was introduced on February 15, 2013 based on an initiative by the conservative 
Christian Social Union (CSU). Starting from August 1, 2013, parents can receive a monthly pay-
ment of 100 € (from January 1, 2014: 150 €) for up to 22 months after “Elterngeld” has expired if 
they do not take their child to a public (or publicly funded) child care institution and rather care 
for it at home or let someone else take care of it.



•    Miriam Bröckel, Hans-Jürgen Andreß284

one being child care (Andreß/Bröckel 2007: 200). Until 2008, the corresponding le-
gal provision (section 1570 of the German Civil Code [Bürgerliches Gesetzbuch –
BGB]) mentioned no time limit for the claim. But since January 1, 2008, the amended 
section states that spousal support can be requested for three years following the 
birth of a child. An extension is only possible beyond that age if it is equitable. 
The decision on the extension must be based both on the needs of the child and 
on available child care facilities. Moreover, the family law case-law shows that the 
family courts are increasingly turning away from the male breadwinner model, and 
in most cases expect a dual earner family in which the wife also works in order to 
maintain her own economic independence (FamRZ 2009: 1391, 2010: 1024, 2011: 
713, 2013: 248, 864). If career interruptions due to child care occur, they are as-
sumed to be rather short-term, or even better to consist of jobs with reduced work-
ing hours. Correspondingly, court decisions on spousal support often only grant a 
fi xed-term claim or a claim that reduces over time in order to accustom claimants 
to their lower standard of living after divorce (FamRZ: 2009: 1391, 2010: 344, 357; 
MDR 2012: 771).

Put differently, as in reformed minimum income programmes, a limited payment 
is guaranteed, but sooner or later the economically dependent spouse has to be 
self-reliant.8 Moreover, the latter expectation is often legitimised with the extended 
child care infrastructure. Hence, with respect to family affairs, we observe a gradual 
transition in the German welfare state after the turn of the millennium away from the 
conservative male breadwinner model and towards the Nordic welfare model that 
aims at providing equal opportunities for men and women; of course, at the cost of 
providing no specialised claims if economic independence is unequally distributed 
between spouses.9

2.3 Changes in the economic behaviour of men and women

Table 1 shows some selected data on economic activity, incomes, and child care 
for 1985-2013 which are mostly based on offi cial statistics. While the employment 
rate of men hardly changed (except for the years with a high unemployment rate, 
around 78 percent of all men are gainfully employed), female labour force partici-
pation increased signifi cantly, from less than 50 percent to almost 70 percent. This 
is also true, although at a lower level, for women with children below the age of 
6. However, this increased economic activity is to a large extent based on part-
time employment. Hence, if one measures female labour force participation in units 
equivalent to full-time positions, the corrected female employment rate actually de-
creases slightly, from 48 percent in 1992 to 46 percent in 2003 (Cornelißen 2005: 

8 It should be noted however that the procedures to determine the amount of child or spousal 
support payments have not been changed, while the reform of minimum income programmes 
also included lower payments for the long-term unemployed.

9 For example, the law provides for spousal maintenance in Sweden, but in fact it is unusual to 
make such claims (Saldeen 1995: 493).



The Economic Consequences of Divorce in Germany    • 285

Indicator 1985 1990 1995 2000 2005 2010 2013

a) Employment rate (aged 15-64) (%)
Men 78.1 78.1 73.7 72.9 71.2 76.0 77.7
Women 48.5 53.6 55.3 58.1 59.6 66.1 68.8
Women (with children aged below 6) 46.7 54.6 60.5

b) Part-time employment (%)
Men 1.4 2.2 3.2 5.0 7.4 9.3 9.9
Women 28.9 33.8 33.9 38.3 44.3 45.9 45.8

c) Unemployment rate (%)
Men 8.6 6.3 9.6 10.5 13.4 7.9 6.9
Women 10.4 8.4 9.0 8.3 10.7 6.5 5.9

d) Earnings
Gross monthly earnings (€) 1675 2018 2114 2230 2387 2594
Net monthly earnings (€) 1174 1341 1411 1513 1610 1727
Real net earnings (€) 1672 1666 1646 1636 1610 1634
Women’s earnings as % of men’s earnings 21.0 21.0 22.0 23.0 22.0

e) Price index for housing costs (2010 = 100) 59.5 75.1 83.0 90.9 100.0 107.5
f) Economic activity, couples with children (%)

Man active/Woman not active 40.0 36.8 33.3 30.4 29.5
Man not active/Woman active 4.1 4.1 5.2 5.6 5.3
Man active/Woman active 50.2 54.0 54.0 54.1 54.9
Man not active/ Woman not active 5.8 5.2 7.5 9.9 10.3

g) Economic activity, couples w/o children (%)
Man active/Woman not active 20.0 18.2 17.5 16.3 16.6
Man not active/Woman active 10.2 10.7 12.1 12.4 11.6
Man active/Woman active 48.6 49.3 53.2 55.2 55.3
Man not active/ Woman not active 21.2 21.8 17.2 16.1 16.6

h) Children in publicly funded child care (%)
Western Germany: aged 0-3 1.6 1.8 2.2 4.2 7.7 17.3 24.2
Western Germany: aged 3-6 62.4 64.6 56.1 73.6 88.6 92.2 93.6
Eastern Germany: aged 0-3 49.8 54.2 41.3 37.0 37.0 48.0 49.8
Eastern Germany: aged 3-6  89.9 84.8 87.1 93.2 95.2 95.6

Tab. 1: Selected statistics on economic activity, incomes, and child care 
(1985-2013)

Note: Some statistics were not available for certain years. If they do not refer exactly to the year in the 
fi rst row, we have chosen the closest value.

Sources: a Number of persons (females, males) aged 15-64 in employment as a share of the total 
population (females, males) of the same age group (Eurostat, http://epp.eurostat.ec.europa.eu/portal/
page/portal/statistics/search_database, retrieved 12 June 2014). b Number of employees reporting be-
ing part-time employed (until 1983: employees who work fewer than the agreed working hours for 
full-time employees) as a share of all employees (GESIS – Social Indicators Monitor (SIMon), Version 
4/2014). c Number of registered unemployed (until 1990: only Western German individuals) as a share 
of the dependent working population (SIMon, Version 4/2014). d Earnings: gross and net nominal 
earnings (Statistisches Bundesamt 2014b), net earnings defl ated with the consumer price index for 
Germany (2010=100); gender pay gap: arithmetic mean of gross hourly wages of all female employ-
ees as a share of the corresponding mean for male employees (Statistisches Bundesamt 2012: 58; 
2014a). e Consumer price index for housing, water, electricity, gas, and other combustibles (Statis-
tisches Bundesamt https://www.destatis.de/DE/ZahlenFakten/GesamtwirtschaftUmwelt/Preise/Ver-
braucherpreisindizes/Tabellen_/VerbraucherpreiseKategorien.html, retrieved 24 July 2014). f Married 
couples and civil partnerships with at least one child aged below 18, data kindly supplied by Matthias 
Keller from the Federal Statistical Offi ce (Keller/Haustein 2012: 58). g Married couples and civil partner-
ships without children, data see (e). h) SIMon (Version 4/2014) andStatistisches Bundesamt (2013).



•    Miriam Bröckel, Hans-Jürgen Andreß286

107). Nevertheless, due to this increased labour force participation, female unem-
ployment has been lower than male unemployment since the mid-1990s.

The increase in female labour force participation was supported by a larger num-
ber of places for children in publicly funded child care facilities. While fewer than 
5 percent of toddlers in Western Germany were taken care of in such institutions up 
to 2000, the care rate for toddlers has now risen to 24.2 percent in 2013. This is not 
suffi cient, but certainly much better than before the turn of the millennium. A similar 
positive development, fortunately on a much higher level, can be observed for pre-
school children in Western Germany. It is almost 100 percent in 2013, thus reaching 
a similar level to that in Eastern Germany. As has already been mentioned, due to its 
socialist heritage public child care was traditionally much more extensive in Eastern 
Germany, both for toddlers and for pre-school children. However, improved public 
child care is not the only explanation for higher female activity rates. Labour market 
reforms which provided increased opportunities for temporary and marginal work, 
structural change entailing a shift from employment in manufacturing to services, 
and changes in parental leave regulations (see Section 2.2), have also contributed 
to the rise in female employment.

This positive development is also consequential for the economic well-being of 
households and families. Although the gender pay gap has not changed signifi cant-
ly since the mid-1990s (women still earn roughly one-fi fth less than men), increased 
female participation rates have signifi cantly decreased the number of households 
in which only the man is economically active. According to the data from Table 1, 
the percentage of households where only the man was active fell by more than 10 
percentage points for couples with children and almost 4 percentage points for cou-
ples without children. On the other hand, households where either both partners 
or only the woman was active increased in number. Hence, the male breadwinner 
model is not dying out, but it is certainly on the downturn. Earnings also increased 
for all employees by about 50 percent from the beginning of the 1990s until the end 
of our observation period. However, as the data on real earnings show, most of this 
increase is eaten up by infl ation, thus severely limiting the purchasing power of the 
working population. While the overall increase in living costs has been modest in 
the observation period, it has been disproportionately high for housing (see the 
price index for housing costs in Table 1). Since marital breakup is often connected 
with residential mobility, this is an additional source of fi nancial strain.

2.4 Research questions and hypotheses

We ask the same fundamental research questions as in our earlier analysis:

Q
1
: How considerable are the economic changes for married men and women 
with respect to separation and divorce?

Q
2
: Who cares for the children, and are the costs of child care covered by support 
payments from the non-resident parent?

Q
3
: Who moves out of the family home and how does this residential mobility 
affect the cost of living for mobile and immobile individuals?



The Economic Consequences of Divorce in Germany    • 287

Q
4
: How does marital disruption change the labour supply of men and women?

Q
5
: To what degree do separated and divorced individuals depend on their own 
gainful employment and private transfers, especially from the former spouse, 
and public transfers?

Q
6
: How does the division of tasks during marriage and the living conditions after 
separation (custody of children, existence of a new partner, gainful employ-
ment) modify economic well-being after the break-up?

The augmented data base however allows us to answer these questions more 
precisely and to add new questions about possible changes after the turn of the 
millennium. Due to the larger sample size, we are now able to control for the multi-
variate relationships between the former marriage model, children, new partners, 
and gainful employment in Q

6
. Given the development of real earnings, we expect 

the economic consequences of marital disruption to be virtually the same as before 
the turn of the millennium (H

1
). With respect to possible gender inequalities, one 

could expect them to be less pronounced after the turn of the millennium due to 
women’s increased economic activity. On the other hand, the discussion in Section 
2.3 showed that this increase was based mostly on part-time or marginal employ-
ment. Hence, we specify a more cautious hypothesis and assume similar gender 
inequalities before and after the turn of the millennium (H

2
). Increased female la-

bour force participation also implies that men are becoming increasingly dependent 
on women’s incomes, and therefore that the economic consequences can also be 
negative for men, at least for some men (H

3
). Since recent amendments of family 

law lead one to expect the economically dependent spouse to be self-reliant earlier 
than before, we anticipate that spousal support payments will play a less signifi cant 
role than in our former analysis (H

4
). Furthermore, having custody over children 

should be less detrimental to economic well-being after the break-up because a 
more extensive public child care infrastructure is available (H

5
). Finally, with respect 

to public transfers, we have no hypothesis as to whether their formerly quite impor-
tant role in cushioning the economic consequences diminishes or decreases. On 
the one hand, we have seen a downsizing of some minimum income payments (e.g. 
ALH), but on the other hand, some new and quite lucrative support payments were 
introduced after the turn of the millennium (e.g., “Elterngeld”).

3 Data and Methodology

3.1 Sample

The following analysis uses data from twenty-nine SOEP panel waves covering 
the period from 1984 to 2012 (Wagner et al. 2007).10 It focuses on the separation 
of married couples, since earlier analyses (Andreß/Güllner 2001) have shown that 

10 The high income sample was excluded. 



•    Miriam Bröckel, Hans-Jürgen Andreß288

separation results in greater economic changes than legal divorce, which follows 
separation, sometimes several years later when the economic situation has already 
stabilised. Separations were identifi ed by combining information from the partner 
variables provided by the panel group at the German Institute for Economic Re-
search (Berlin) and from survey questions about family status and life events (Gül-
lner 2000). The year in which these separations are observed is defi ned by . Since 
we would like to compare the situation before and after these separations, data of 
at least one spouse have to be available for the years t-1 (before) and t+2 (after). 
Therefore, the unit of analysis is the corresponding separation in t, augmented by 
data from the years before and after the event.11

The data we use for the period before the turn of the millennium (1984-99) are 
slightly different from the data in our former analysis (Andreß/Bröckel 2007).12 

Therefore, we compute all statistics anew (rather than copying them from the former 
publication) with in some cases slightly different results. After removing individuals 
who leave the panel in the year of separation, our sample contains data for 393 men 
and 446 women separating in 1984-1999 and 451 men and 560 women separating in 
2000-12. Nearly two-thirds of these separated individuals become legally divorced 
within the observation period: 17 percent in the year of separation, 32 percent one 
year later, and 26 percent two years later. The remaining 25 percent take more than 
two and up to 19 years. It should also be noted that almost all the sample mem-
bers are of working age (only 4 percent are older than 60). The following results are 
based on weighted data, using the cross-sectional SOEP weights that control for the 
sampling design, non-response, and panel attrition (see Andreß/Bröckel 2007: 204 
for a discussion of the effects of panel attrition).13 Because men and women have 
different attrition rates, it is not guaranteed that both spouses are included in our 
analyses. For example, if we measure median household incomes before separa-
tion when both spouses were still living together, the corresponding values for men 
and women can be different because some spouses’ data are missing.

3.2 Income indicators

Changes in income are a central focus of our analysis. The SOEP provides various 
income indicators for different time periods (month, year), at both individual and 

11 For certain analyses the longitudinal comparisons go even further than t-1 and t+1 (see below). 
If individuals experience several separations, only the fi rst is included in the sample.

12 Firstly, in the former publication, all sample members were checked manually and missing part-
ner information was added, thus resulting in a larger sample size for 1984-99. Secondly, and 
as will be explained later, we used a different base year for computing real incomes. Thirdly, 
incomes were expressed in DM, while now they are supplied in Euro even for the years prior 
to the introduction of the Euro in 2002. Fourthly, the data base has been revised with each 
new edition of the SOEP. Fifthly, our former measures of household income included imputed 
rents, while we are now using the housing costs specifi ed by the respondents. Note also that 
the SOEP was extended to East Germany (the former GDR) in 1990. Hence, all data before 1990 
come from West German respondents.

13 More specifi cally, we used the cross-sectional weight attached to the corresponding individual 
in the fi nal year in which the individual contributed to our data.



The Economic Consequences of Divorce in Germany    • 289

household levels. Given our interest in income redistribution by taxes, as well as 
by private and public transfers, we use data on yearly pre- and post-government 
household incomes as provided by the SOEP group in the PEQUIV fi le. Amongst 
other things, the PEQUIV fi le augments the detailed SOEP data on various income 
sources with imputed payroll and income taxes. The use of yearly pre- and post-
government household incomes also has important implications for our longitudinal 
comparisons. Since the event of interest (separation) occurred at some point in time 
between the panel interview in t and the preceding interview in t-1, yearly incomes 
from t-1 are not a valid indicator of the economic situation before the event. In many 
cases they already characterize the time of separation. Therefore, our income com-
parisons are based on data from t-2 and t+1. All selected cases were controlled for 
already living together with their spouse at t-2. Technically, things are even more 
complicated: Because yearly household incomes are based on retrospective ques-
tions, we use data from panel wave t+2 (including retrospective information for t+1) 
and panel wave t-1 (including retrospective information for t-2 ).

These household incomes include private transfers received by the household, 
but do not take into account private transfers paid to other individuals outside the 
household. Given the signifi cance of private transfers for our research problem, 
a few remarks on the underlying survey questions are needed. Private transfers 
received are based on the so-called income calendar, which records the receipt of 
various income components for each month of the year preceding the interview, 
among them (support) payments from individuals outside the household (including 
payments from the advanced payment scheme for child maintenance). Though it is 
possible to distinguish from 2001 onwards between legal support payments made 
by the former spouse (including payments from the advanced child maintenance 
scheme) and other private transfers to the household (and in addition, from 2010 on-
wards, also between payments made by the former spouse and those payments re-
ceived from the advance child maintenance scheme), we use private transfers as the 
sum of all these payments to make data comparable over the whole sample period. 
Thus, all private transfers received are included in PEQUIV pre-government house-
hold incomes, which then are subjected to the PEQUIV tax simulation programme to 
obtain post-government household incomes (after adding social security pensions 
and public transfers). Although the SOEP personal interview also includes a ques-
tion about the amount of (support) payments being made for individuals outside 
the household in the year preceding the interview, this information is not used in 
the PEQUIV. It is however highly important for our analysis, and we have therefore 
deducted these support payments14 from PEQUIV post-government household in-
comes in order to obtain a measure of disposable household incomes for separated 
individuals. Given this procedure, it should be noted that private transfers received 
are taxed and include all kinds of sources, while private transfers that are paid are 

14 The corresponding question distinguishes between different addressees (parents including 
parents-in-law, (former) spouse, children including son-in-law and daughter-in-law, other rela-
tives, other non-related individuals). We have only used (support) payments for children and the 
former spouse.



•    Miriam Bröckel, Hans-Jürgen Andreß290

not taxed, but are restricted to the former family (children, spouse). Until 2000, the 
survey question asks for “donations from individuals outside the household, includ-
ing payments from the advanced maintenance scheme.” This includes gifts and 
other transfers from all private sources outside the household, as well as mandated 
child support and spousal support payments. From 2001 to 2009, private payments 
to the household can be additionally distinguished in terms of legal support pay-
ments from the former spouse (again including payments from the advance child 
maintenance scheme) as against other private payments. Support payments from 
the former spouse and payments from the advance child maintenance scheme are 
separately asked for from 2010 onwards.  As discussed in Andreß and Bröckel (2007: 
206), the amount paid by the liable person and that received by the entitled person 
do not agree, and there is no possibility to correct these discrepancies.

All household incomes were equivalised with respect to household size accord-
ing to the formula:

equivalised household income = household income / sizeθ

These equivalised household incomes were then assigned to each individual 
in the household, and analyses were carried out on the individual level. A value of 
0.5 was used for the elasticity parameter θ, yielding moderate economies of scale 
with increasing household size similar to equivalence scales applied by the OECD 
(Buhmann et al. 1988). Using θ=1, would yield per capita incomes and θ=0 total 
household incomes. In our earlier analysis, θ is varied to check the sensitivity of 
the results against different assumptions concerning economies of scale. It sub-
sequently emerged that gender inequalities persist irrespective of the equivalence 
scale that is used (Andreß/Bröckel 2007: 217).

When analysing incomes over such a long time-span as in our case, it is impor-
tant to keep incomes comparable over time. Thus, we expressed all incomes in 2000 
prices, using the offi cial basket of commodities from the German Federal Statistical 
Offi ce for 2005. The PEQUIV data supply an index which has been published by 
the German Federal Statistical Offi ce and permits incomes to be expressed in 2005 
prices. Our old study used incomes that were expressed in 1995 prices. Preliminary 
analyses with the data showed that using either one of the two indices is not without 
problems for our study. It emerged that expressing incomes in 2005 (1995) prices 
exaggerated (diminished) the level of incomes from the early (late) years of our ob-
servation period, and hence attenuated (aggravated) income changes. We decided 
to compromise by using the newer index but to rescale it to the base year 2000.

3.3 Methods

The following analysis focuses both on the absolute value of various measures of 
economic well-being after separation and their change relative to their amount be-
fore separation. Consider the case of equivalised household income: let Ii,t-2 and 
Ii,t+1 be the equivalised household incomes of individual i two years previously (t-2) 

(1)



The Economic Consequences of Divorce in Germany    • 291

and one year after separation (t+1). As a measure of change, we use the percentage 
change in equivalised household income for each individual:

Below, %Ii is termed a (percentage) change score of income. Because of the 
variance of economic changes (not all women lose and not all men win), we ana-
lyse the median, as well as the fi rst and third quartiles of the distribution of change 
scores. We use violin plots to visualise the distribution of change scores. A violin 
plot is a combination of a box plot and a kernel density plot (Hintze/Nelson 1998). 
Similar change scores were computed for other measures of economic well-being 
(e.g. housing costs).

In principle, quantile regression models can be used to compute confi dence in-
tervals and hypothesis tests for the median, fi rst and third quartiles of the distribu-
tion of change scores. Again, preliminary analyses showed that the basic conclu-
sions remain unchanged if one dichotomises change scores into gains (%Ii≥0) and 
losses (%Ii<0). Therefore, in order to test gender differences and changes after the 
turn of millennium, we use logistic regression to model the probability of a loss of 
income subsequent to separation: Pr(%Ii≥<0). Within these models it is also possi-
ble to control for some important changes in living conditions that take place in par-
allel to the separation of the two partners. The economic situation after separation 
naturally varies depending on whether the person is gainfully employed, has to care 
for children or lives together with a new partner. Moreover, it is important whether 
he or she was economically independent during the marriage. As an indicator of 
economic independence, we measure whether the person lived prior to separation 
in a household with a male breadwinner, a household in which the husband contrib-
uted the main but not the only income (termed “husband main earner household”), 
a dual earner household, or a household with a female breadwinner or main earner.

4 Results

We start with an analysis of gender-specifi c changes in child care, housing and 
employment (Sections 4.1-4.3), before turning to the more specifi c question of avail-
able incomes after marital disruption. Section 4.4 focuses on the effects of private 
and public income redistribution and analyses income changes caused by separa-
tion for men and women and for fi ve different income concepts. Using equivalised 
disposable household incomes, Section 4.5 then controls for relevant living condi-
tions parallel to the separation process.

4.1 Children and support payments

According to the most recent German time budget survey, mothers spend more 
than twice as much time with their children than their fathers do (BMFSFJ/Statis-

%Ii=100·
I
i,t+1

-I
i,t-2

I
i,t-2

 (2)



•    Miriam Bröckel, Hans-Jürgen Andreß292

tisches Bundesamt 2003: 22). This inequality continues when a marriage breaks 
down. As Table 2 shows, 61 percent of all women separating after the turn of the 
millennium live together with minor children (under the age of 18) after separation, 
while only 17 percent of all men do.15 Similar results are found for the period before 
2000 and from representative surveys of divorced men and women (Andreß et al. 
2003). If one compares the two periods, not much has changed, except that couples 
separating after the turn of the millennium have fewer children on average.

We can use the SOEP data to check whether these inequalities are compensated 
for by maintenance payments from the former husband. As already mentioned, the 
SOEP data do not make it possible to identify liable and entitled persons in a strict 
legal sense. But as described in Section 3.2, we can identify those individuals who 
receive and/or make support payments to other persons outside their own private 
household. According to the data shown in Table 2, only 33 percent of all men and 
4 percent of all women give money to their children and/or former spouse after sep-
aration. Even when focusing exclusively on parents, only 38 percent of all fathers 
(3 percent of all mothers) make support payments. On the other hand, as few as 
3 percent of all men and 26 percent of all women receive support payments. The fi g-
ures are slightly higher for individuals living together with children after separation 
(men: 4 percent, women: 35 percent), but 65 percent of all women living together 
with children do not receive any support payments at all.

Has there been any change over time? It is diffi cult to give a defi nite answer. 
With respect to child support, the risk population has changed (fewer children; see 
above). When it comes to support payments received, these payments may origi-
nate from the former spouse, but also from other individuals and public transfers 
via the advance maintenance scheme (see the discussion in Section 3.2). From the 
data in Table 2, it seems as if men make fewer support payments (at least those 
living with children), while women living without children make more support pay-
ments. Given the aforementioned inaccuracies in the data, we hesitate to conclude 
that this is due to a lower economic potential of men and increased dependence on 
women’s incomes.

Accordingly, private transfers comprise only a small part of household income 
after separation, much smaller than public transfers. As Table 2 demonstrates, pri-
vate transfers only make a difference for women and constitute only 9 percent of 
their income package. Women’s (and men’s) economic situation after separation de-
pends much more on public transfers. In the case of women, public transfers consti-
tute more than one-third (39 percent) of their income package on average if they live 
together with children after separation. This proportion is smaller if one looks at all 
separated women (30 percent) or at men’s household incomes (11 percent), but it is 
still a signifi cant fi gure which increases with separation in all cases (from t-2 to t+2), 
except for men without children. We therefore concluded in our previous analysis 

15 After separation, these data may include children from a new partner. Between 2000 and 2012 
(1984-1999 in brackets), 10 percent (13 percent) of all separated men and 10 percent (18 per-
cent) of all separated women live together with minor children and a new partner one year after 
separation.



The Economic Consequences of Divorce in Germany    • 293

T
a

b
. 

2
: 

P
ri

v
a
te

 a
n

d
 p

u
b

lic
 t

ra
n

sf
e

rs
 f

o
r 

se
p

a
ra

te
d

 m
e

n
 a

n
d

 w
o

m
e

n
 2

0
0

0
-2

01
2
 (
19

8
4

-1
9

9
9
 in

 b
ra

ck
e

ts
)

A
ll 

m
e

n
M

e
n

 w
it

h
 

ch
ild

re
n

a
,b

M
e

n
 w

it
h

o
u

t 
ch

ild
re

n
A

ll 
w

o
m

e
n

W
o

m
e

n
 w

it
h

 
ch

ild
re

n
a
,b

W
o

m
e

n
 w

it
h

o
u

t 
ch

ild
re

n

t-
2

t+
1

t-
2

t+
1

t-
2

t+
1

t-
2

t+
1

t-
2

t+
1

t-
2

t+
1

L
iv

e
 t

o
g

e
th

e
r 

w
it

h
 m

in
o

r 
ch

ild
re

n
b
 

in
 t

h
e
 h

o
u

se
h

o
ld

 (
%

)
6

2
 (

6
6

)
17

 (
2

5
)

–
–

–
–

6
7
 (

7
3

)
61

 (
6

7
)

–
–

–
–

D
o

 n
o

t 
liv

e
 t

o
g

e
th

e
r 

w
it

h
 m

in
o

r 
ch

ild
re

n
b
 i
n

 t
h

e
 h

o
u

se
h

o
ld

 (
%

)
3

8
 (

3
4

)
8

3
 (

7
5

)
–

–
–

–
3

3
 (

2
7

)
3

9
 (

3
3

)
–

–
–

–

M
a
d

e
 s

u
p

p
o

rt
 p

a
y

m
e

n
ts

c  
a

t 
t+

1
 (

%
)

–
3

3
 (

41
)

–
3

8
 (

5
3

)
–

2
4
 (

2
3

)
–

4
 (

5
)

–
3
 (

5
)

–
10

 (
5

)

R
e

ce
iv

e
d

 s
u

p
p

o
rt

 p
a
y

m
e

n
ts

d
 a

t 
t+

1
 

(%
)

–
3
 (

2
)

–
4
 (

3
)

–
1
 (

1)
–

2
6
 (

2
4
)

–
3

5
 (

3
0

)
–

9
 (

12
)

P
ro

p
o

rt
io

n
 o

f 
h

o
u

se
h

o
ld

 i
n

co
m

e
 (

%
)

p
ri

v
a

te
 t

ra
n

sf
e

rs
e

1
 (

0
)

1
 (

0
)

0
 (

0
)

2
 (

1)
1
 (

0
)

0
 (

0
)

1
 (

1)
9
 (

10
)

1
 (

1)
12

 (
11

)
1
 (

0
)

4
 (

8
)

p
u

b
lic

 t
ra

n
sf

e
rs

e
14

 (
11

)
11

 (
14

)
17

 (
13

)
2

9
 (

16
)

10
 (

8
)

8
 (

13
)

17
 (

12
)

3
0
 (

19
)

2
0
 (

13
)

3
9
 (

2
6

)
12

 (
8

)
17

 (
6

)

in
d

iv
id

u
a
l l

a
b

o
u

r 
in

co
m

e
f

6
3
 (

6
4
)

7
6
 (

7
6

)
71

 (
7

0
)

7
2
 (

7
6

)
5

4
 (

5
7

)
7

8
 (

7
6

)
2

7
 (

2
8

)
5

5
 (

5
5
)

2
4
 (

2
5
)

51
 (

5
2
)

3
3
 (

3
7

)
6

0
 (

61
)

p
u

b
lic

 t
ra

n
sf

e
rs

 m
o

re
 t

h
a
n

 5
0
 %

 o
f 

h
o

u
se

h
o

ld
 i
n

co
m

e
 (

%
)

5
 (

5
)

10
 (

9
)

6
 (

5
)

12
 (

7
)

3
 (

5
)

9
 (

10
)

6
 (

9
)

2
5
 (

16
)

8
 (

9
)

3
5
 (

2
1)

5
 (

7
)

12
 (

8
)

N
o

te
s:

 a
 M

ad
e
 s

u
p

p
o

rt
 p

ay
m

e
n

ts
: 

in
d

iv
id

u
al

s 
liv

in
g

 w
it

h
 c

h
ild

re
n

 b
e
fo

re
 (

t-
2
) 

se
p

ar
at

io
n

; 
re

ce
iv

e
d

 s
u

p
p

o
rt

 p
ay

m
e
n

ts
/t

ra
n

sf
e
rs

 a
s 

%
 

o
f 

h
h

 i
n

co
m

e
: 

in
d

iv
id

u
al

s 
liv

in
g

 w
it

h
 c

h
ild

re
n

 a
t 

th
e
 t

im
e
 i

n
 q

u
e
st

io
n

 (
t-

2
, 

t+
1
).

 b
 C

h
ild

re
n

 a
g

e
d

 u
n

d
e
r 

1
8
, 

in
cl

u
d

in
g

 c
h

ild
re

n
 o

f 
th

e
 

m
ar

ri
ag

e
 a

n
d

 o
th

e
r 

ch
ild

re
n

 (
e
.g

. c
h

ild
re

n
 o

f a
 n

e
w

 p
ar

tn
e
r)

. c
 S

u
p

p
o

rt
 p

ay
m

e
n

ts
 fo

r 
ch

ild
re

n
 o

r 
fo

rm
e
r 

sp
o

u
se

. d
 P

ay
m

e
n

ts
 fr

o
m

 o
th

e
r 

in
d

iv
id

u
al

s 
o

u
ts

id
e
 t

h
e
 h

o
u

se
h

o
ld

 i
n

cl
u

d
in

g
 a

d
v
an

ce
d

 m
ai

n
te

n
an

ce
 p

ay
m

e
n

ts
. 

e
 M

e
an

 p
e
rc

e
n

ta
g

e
 o

f 
d

is
p

o
sa

b
le

 h
o

u
se

h
o

ld
 i

n
co

m
e
 

(a
ft

e
r 

d
e
d

u
ct

in
g

 t
ax

e
s 

an
d

 s
u

p
p

o
rt

 p
ay

m
e
n

ts
, b

u
t 

in
cl

u
d

in
g

 p
ri

v
at

e
 a

n
d

 p
u

b
lic

 t
ra

n
sf

e
rs

).
 f  

G
ro

ss
 e

ar
n

in
g

s 
as

 m
e
an

 p
e
rc

e
n

ta
g

e
 o

f 
p

re
-

g
o

v
e
rn

m
e
n

t 
in

co
m

e
 (

b
e
fo

re
 t

ax
at

io
n

 a
n

d
 in

cl
u

d
in

g
 p

ri
v
at

e
 t

ra
n

sf
e
rs

).

S
o

u
rc

e
: 

S
O

E
P

 1
9
8
4
-2

0
1
2
, 
w

e
ig

h
te

d
 d

at
a.



•    Miriam Bröckel, Hans-Jürgen Andreß294

that “although a rather comprehensive system of (legal) maintenance regulations 
exists in Germany, it is not the former spouse who bears the main burden of income 
maintenance after separation” (Andreß/Bröckel 2007: 209). And this is obviously 
still the case after the turn of the millennium if one compares the data for the two 
periods in Table 2. It even looks as if this dependence on the public purse has in-
creased over time. For example, before the turn of the millennium, public transfers 
made up 16 percent of men’s and 26 percent of women’s post-separation income 
package if they were living with children in their households. These percentages 
increased to 29 percent and 39 percent after 2000. Also the percentage of individu-
als where more than half of the income package consists of public transfers has 
increased by 5 percentage points (men) and 14 percentage points (women), respec-
tively. But one has to take into account that after the turn of the millennium, income 
packages already consist to a higher degree of public transfers during marriage (see 
the percentage of public transfers at t-2 for all groups).

Although public transfers are of special policy relevance, one should not forget 
that, fi rst and foremost, it is the separated individual who takes care of him- or 
herself through gainful employment (see below). Even in the case of women with 
children, about one-half (51 percent) of their income package comes from their own 
work income (cf. Table 2). Strictly speaking, average percentages of work income in 
Table 2 are not comparable with average percentages of private and public trans-
fers. PEQUIV income data do not include net earnings that we could relate to our 
measure of net disposable household income (after taxation). We therefore used 
gross earnings as a percentage of pre-government household income (before taxa-
tion, but including private transfers). If one assumes that the tax rate is the same for 
earnings as for all other income components, the fi gures in Table 2 roughly indicate 
the (average) signifi cance of earnings within the income package. Nevertheless, the 
number of individuals for whom public transfers are the main source of income 
(i.e. at least half of their disposable household income consists of public transfers) 
increases dramatically in all groups after separation, and especially so for women 
with children. This shows the importance of public transfers, at least for a subgroup 
of separated people.

4.2 Residential mobility

In our sample, 60 percent of all men and 62 percent of all women moved homes 
within two years after separating from their partner. In other words, the majority of 
both genders has to cope with residential mobility after separation. Focusing on in-
dividuals with minor children in the household after separation, the overall mobility 
rate is slightly lower, and noticeable differences exist between men (39 percent) and 
women (56 percent). “On the one hand,” we argued, “this result is in accordance 
with the common assumption in divorce proceedings that children should stay in 
their familiar environment and not change homes. On the other hand, it is surprising 
that this is true especially for men and less so for women” (Andreß/Bröckel 2007: 
210). We suspected that the higher mobility rate of women with children is partly a 



The Economic Consequences of Divorce in Germany    • 295

forced choice. Given their lower average incomes, they have to reduce the fi nancial 
strain caused by separation by moving to smaller and hence less expensive homes.

We will analyse below whether this is still the case after the turn of the millen-
nium. To that end, Table 3 looks beyond straightforward residential mobility and 
describes the concomitant changes in the size, quality and price of the home. The 
data are shown both in absolute (m2, number of rooms, Euro) and relative terms 
(per household member, percentage of household income), and differentiate be-
tween men and women as well as between mobile and immobile persons. Again, 
we focus mainly on the data for 2000-2012, and only comment on the earlier period, 
when marked differences exist. Given the increase in housing costs (see Table 1), 
it is no surprise that respondents separating after the turn of the millennium report 
higher yearly housing costs than those separating in the years 1984-99. Part of this 
increase can also be explained by the fact that the fi rst group lives in slightly larger 
homes than the second.

Just as before the turn of the millennium, mobile persons move to smaller and 
less expensive homes (fewer m2, fewer rooms, lower housing costs). Immobile per-
sons, by defi nition, occupy the same home as before, and therefore have to man-
age to pay the costs of the former marital home from household incomes which are 
smaller in most cases. It may be possible to economise on utilities in some cases, 
or owner-occupied housing may become cheaper when mortgages have been paid 
off. But housing costs will remain the same for immobile persons in many cases.16 

Immobile men spend 29 percent of their household income on housing after sepa-
ration, while mobile men spend 22 percent. On the individual level, housing costs 
increase at least 27 percent for half of immobile men, while the median increase for 
mobile men actually equates to a decrease of 1 percent. Similar differences can be 
observed for immobile (+54 percent) and mobile women (+11 percent). Thus, by 
moving to less expensive homes, the separated individuals are able to reduce the 
fi nancial strain caused by the household split at the expense of lower housing qual-
ity and a loss of neighbourhood contacts.

When interpreting these data with respect to possible gender differences, we 
have to remember the changes in household composition as described in the pre-
vious section. If children mostly stay with their mothers, average household size 
decreases more for men than it does for women, irrespective of their mobility sta-
tus. In other words, women share their homes with more persons on average, even 
when they remain in the family home (see the size of the home and the number of 
rooms per household member in Table 3). The most important difference in our 
context, however, concerns the fi nancial strain caused by the household split. In 
that respect, women always have to pay a larger portion of their household income 

16 A more differentiated analysis shows that the information provided by SOEP respondents on 
housing (size of home, housing costs, etc.) varies over time even for immobile individuals. This 
low reliability over time may explain, for instance, the slight differences in the reported size (m2) 
of their homes or the fact that median housing costs fell in 2000-2012 (but not in 1984-1999). 
Altogether, however, we consider the statistics reported in Table 3 to be plausible and theoreti-
cally meaningful.



•    Miriam Bröckel, Hans-Jürgen Andreß296

Im
m

o
b

ile
M

o
b

ile

t-
1

t
%

 c
h

a
n

g
e

c
t-

1
t

%
 c

h
a
n

g
e

c

M
e

n
 (

m
e

d
ia

n
 v

a
lu

e
s 

u
n

le
ss

 s
ta

te
d

 o
th

e
rw

is
e
)

M
e

a
n

 a
b

so
lu

te
 s

iz
e
 (

m
²)

10
5
 (

9
9

)
10

4
 (

9
8

)
–

10
4
 (

9
2
)

6
4
 (

6
0

)
-4

1
 (

-3
6

)

…
 r

e
la

ti
v
e
 s

iz
e
 (

m
² 

p
e

r 
h

o
u

se
h

o
ld

 m
e

m
b

e
r)

3
7
 (

3
4
)

8
1
 (

7
3

)
+

10
0
 (+

10
0

)
3

4
 (

3
0

)
5

4
 (

51
)

+
5

7
 (+

6
8

)

M
e

a
n

 n
u

m
b

e
r 

o
f 

ro
o

m
sd

4
.3

 (
3

.9
)

4
.2

 (
3

.9
)

–
4

.1
 (

3
.6

)
2

.6
 (

2
.3

)
-4

0
 (

-4
0

)

..
. 
o

f 
ro

o
m

s 
p

e
r 

h
o

u
se

h
o

ld
 m

e
m

b
e

r
1.

5
 (

1.
3

)
3

.3
 (

2
.8

)
+

10
0
 (+

10
0

)
1.

3
 (

1.
2
)

2
.2

 (
2

.0
)

+
6

0
 (+

5
0

)

Y
e

a
rl

y
 c

o
st

se
 (

E
u

ro
) 

4
9

2
6

a  
(4

2
9

4
a )

4
7

8
8

b
 (

4
4

0
5

b
)

-6
 (

-3
)

5
9

0
7a

 (
4

6
0

6
a )

3
7

8
2

b
 (

3
8

5
5

b
)

-3
3
 (

-1
6

)

Y
e

a
rl

y
 c

o
st

se
 a

s 
%

 o
f 

h
o

u
se

h
o

ld
 i
n

co
m

e
2

3
a  

(1
5

a )
2

9
b
 (

2
2

b
)

+
2

7
 (+

3
7

)
19

a  
(1

4
a )

2
2

b
 (

2
1b

)
-1

 (+
5

3
)

W
o

m
e

n
 (

m
e

d
ia

n
 v

a
lu

e
s 

u
n

le
ss

 s
ta

te
d

 o
th

e
rw

is
e
)

M
e

a
n

 a
b

so
lu

te
 s

iz
e
 (

m
²)

10
0
 (

9
3

)
10

0
 (

9
5
)

–
10

2
 (

10
1)

71
 (

6
8

)
-3

1
 (

-3
4

)

…
 r

e
la

ti
v
e
 s

iz
e
 (

m
² 

p
e

r 
h

o
u

se
h

o
ld

 m
e

m
b

e
r)

3
2
 (

2
8

)
4

8
 (

4
3

)
+

3
3
 (+

4
0

)
3

5
 (

3
4
)

4
3
 (

4
0

)
+

18
 (+

15
)

M
e

a
n

 n
u

m
b

e
r 

o
f 

ro
o

m
sd

4
.0

 (
3

.9
)

4
.0

 (
3

.9
)

–
4

.1
 (

4
.0

)
2

.8
 (

2
.7

)
-2

5
 (

-3
3

)

..
. 

o
f 

ro
o

m
s 

p
e

r 
h

o
u

se
h

o
ld

 m
e

m
b

e
r

1.
3
 (

1.
1)

1.
9
 (

1.
7

)
+

3
3
 (+

3
3

)
1.

4
 (

1.
3

)
1.

7
 (

1.
6

)
+

2
5
 (

0
)

Y
e

a
rl

y
 c

o
st

se
 (

E
u

ro
) 

5
,6

4
0

a  
(4

0
3

6
a )

5
,3

0
1b

 (
4

5
7

2
b
)

-1
 (+

1)
5

,9
8

9
a  

(4
8

2
4

a )
4

,5
71

b
 (

41
7

9
b
)

-2
5
 (

-9
)

Y
e

a
rl

y
 c

o
st

se
 a

s 
%

 o
f 

h
o

u
se

h
o

ld
 i
n

co
m

e
2
1a

 (
14

a )
3

7
b
 (

2
2

b
)

+
5

4
 (+

7
5

)
2

4
a  

(1
6

a )
2

6
b
 (

2
1b

)
+

11
 (+

2
3

)

T
a

b
. 

3:
 

R
e

si
d

e
n

ti
a
l m

o
b

ili
ty

 o
f 

se
p

a
ra

te
d

 m
e

n
 a

n
d

 w
o

m
e

n
 2

0
0

0
-2

01
2
 (
19

8
4

-1
9

9
9
 in

 b
ra

ck
e

ts
)

N
o

te
s:

 a
 M

e
as

u
re

d
 a

t 
t-

2
. 

b
 M

e
as

u
re

d
 a

t 
t+

1
. 

c  
M

e
d

ia
n

 o
f 

th
e
 c

h
an

g
e
 s

co
re

s 
o

b
ta

in
e
d

 b
y
 c

o
m

p
ar

in
g

 t
h

e
 b

e
fo

re
 a

n
d

 a
ft

e
r 

m
e
as

u
re

s 
fo

r 
e
ac

h
 i

n
d

iv
id

u
al

. 
d
 E

x
cl

u
d

in
g

 k
it

ch
e
n

, 
b

at
h

ro
o

m
, 

an
d

 r
o

o
m

s 
sm

al
le

r 
th

an
 6

 m
².

 e
 R

e
n

t/
h

o
u

si
n

g
 c

o
st

s,
 i

n
cl

u
d

in
g

 u
ti

lit
y
 c

o
st

s,
 b

u
t 

e
x
cl

u
d

in
g

 h
e
at

in
g

 c
o

st
s.

S
o

u
rc

e
: 

S
O

E
P

 1
9
8
4
-2

0
1
2
, 
w

e
ig

h
te

d
 d

at
a.



The Economic Consequences of Divorce in Germany    • 297

for their home than men, irrespective of whether they are mobile (+4 percentage 
points) or immobile (+8 percentage points). Before the turn of the millennium, there 
were hardly any gender differences with respect to housing costs as a percentage 
of household income. But because housing costs increased faster in the 1990s than 
in the 2000s (see Table 1), the median increase in relative housing costs for individu-
als separating between 1984 and 1999 (+37 percent and +53 percent for immobile 
and mobile men, +75 percent and +23 percent for immobile and mobile women) 
was much larger than after the turn of the millennium, where the median increase 
varied between –1 percent and 54 percent.

The hypothesis that residential mobility is a forced choice, especially for women, 
can be tested by comparing relative housing costs after separation with those dur-
ing marriage. Both mobile men and women experience fewer increases in fi nancial 
strain (men: –1 percent, women: +11 percent) than those men and women who re-
main in the family home (men: +27 percent, women: +54 percent).17 The fi gures 
(not shown in Table 3) are even more remarkable for women living with children: 
Relative housing costs increased for immobile women with children by +50 percent 
(+76), while mobile women with children increased their relative housing costs by 
only +47 percent (+32). These data suggest that residential mobility is partly en-
forced for women. This might also explain the comparatively high mobility rates of 
women with children, who contrary to the legal norms that have been mentioned 
leave their familiar environment more often than men with children.

4.3 Employment

Table 4 shows that, between 2000 and 2012, 91 percent of all men and 67 percent 
of all women (64 percent of all women with children) are either employed full-time, 
part-time or marginally before they separate from their spouses. Compared to the 
overall employment rates in Table 1, these are rather high rates of economic ac-
tivity, especially for women. We are interested in fi nding out how these men and 
women adapt their employment status to the new situation after separation. Table 
4 illustrates the corresponding changes in economic activity one year after sepa-
ration. Some persons have extended their employment in the sense that former 
part-timers are now working full-time, or persons who formerly were not gainfully 
employed are now working at least irregularly or marginally. Since most men work 

17 We observed a similar result for 1984-1999 in our earlier analysis (see Andreß/Bröckel 2007: 
212). But our present replication shows an increase of +53 percent for mobile men, which com-
pared to the earlier value (+25 percent) looks like a data or computation error. Extensive data 
checks showed that it is not a computation error. One explanation of the discrepancy may be 
a large sampling error due to the small sample size in this cell of the table (the change score is 
based on data at t-2 and is computed for a specifi c subgroup of men). Interestingly, mobile men 
have lower incomes after separation than immobile men do, which of course leads to above-
average increases in relative housing costs for mobile men, and may also explain the increase 
of +53 percent. We observe the opposite development for mobile women: Their incomes are 
higher after separation than the incomes of immobile women are, which leads to below-aver-
age increases in relative housing costs.



•    Miriam Bröckel, Hans-Jürgen Andreß298

full-time before separation, further increases of employment are hardly feasible 
(only 2 percent of all men report an increase), so that these kinds of change are 
mostly observed for women (30 percent report an increase), especially for women 
with children (31 percent). Given the predominance of full-time employment before 
separation, the opposite, namely a reduction of employment, is much more prob-
able for men. However, this event is observed less frequently for men (9 percent re-
port a decrease) than for women (13 percent; women without children: 16 percent). 
As we noted earlier, “this contradicts the prejudice often heard in public debates 
that men reduce employment in order to shirk maintenance payments for their 
spouse and children” (Andreß/Bröckel 2007: 213). All in all, women with children 
are the ones who change their activity status most frequently (47 percent report a 
change), mostly by increasing employment (31 percent).

Activity rates during marriage for individuals separating before the turn of the 
millennium are slightly lower (men: 87 percent, women: 63 percent, women with 
children: 56 percent), but still above the overall employment rates. As the data in 

Tab. 4: Employment and earned incomes for separated men and women 
2000-2012 (1984-1999 in brackets)

Men All women Women with childrena

t-2 t+1 t-2 t+1 t-2 t+1

Employment statusb (%)

full-time 88 (86) 84 (81) 28 (36) 38 (41) 16 (22) 24 (25)

part-time 2 (1) 1 (2) 28 (23) 25 (24) 34 (28) 34 (32)

marginal 1 (0) 1 (0) 11 (4) 10 (3) 14 (6) 9 (4)

training 1 (2) 0 (0) 2 (2) 1 (2) 4 (1) 1 (2)

unemployed 6 (11) 12 (16) 31 (24) 26 (25) 32 (28) 33 (28)

other 2 (1) 1 (0) 0 (11) 0 (6) 0 (16) 0 (9)

Change in employmentc (%)

decreased 9 (11) 13 (14) 16 (15)

no change 88 (84) 57 (59) 53 (52)

increased 2 (6) 30 (27) 31 (33)

Gross earningsd 31.029 30.319 12.031 14.681 8.458 13.861

(Median, Euro) (28.976) (30.225) (15.062) (18.557) (11.095) (16.251)

Change in earningse (%)

1st quartile -13 (-7) -6 (-12) -4 (-5)

median 2 (5) 26 (11) 36 (26)

3rd quartile 14 (22) 106 (60) 225 (96)

Notes: a Women living with children after separation. b Individuals aged below 61. c See 
text for further details. d Individuals gainfully employed before and after separation. e Dis-
tribution of change scores obtained by comparing earnings between t-2 and t+1 for each 
individual.

Source: SOEP 1984-2012, weighted data



The Economic Consequences of Divorce in Germany    • 299

Table 4 show, the increase in economic activity after the turn of the millennium is 
due to less unemployment for men and more part-time and marginal employment 
for women. This observation is very much in line with the overall development of 
the labour market (see Section 2.3 and Table 1). But the pattern of changes due to 
separation is not much different from the pattern after the turn of the millennium.

An important question is what these changes in economic activity mean in terms 
of earned incomes. We use yearly earnings before taxes measured in 2000 prices. 
Yearly data are chosen to make this analysis comparable with the following analyses 
of equivalised household incomes. Earnings before taxes are preferred to net earn-
ings because we are interested in the effects of working hours and remuneration, 
and not in the question of how the State privileges or under-privileges changes in 
employment. Furthermore, the income tax bracket for married individuals changes 
one year after separation.

According to these data, the majority of men and women experience positive 
changes; and this is true both before and after the turn of the millennium. Due to 
their lower economic activity before separation, the increases in earnings are es-
pecially high for women (numbers for 1984-1999 in brackets): Half of them experi-
ence increases of +26 percent (+11 percent) and more, while the median increase 
for men amounts to +2 percent (+5 percent). One the other hand, since women 
start from a much lower level, their earnings lag far behind men’s despite the large 
increases. It should also be noted that not all men and women expand their eco-
nomic activity in terms of earnings. One quarter of them has to deal with declin-
ing earnings of at least -13 percent (-7 percent) for men and -6 percent (-12 percent) 
for women. But this should not distract us from the general conclusion that both 
genders predominantly try to expand their economic activity, which results in sig-
nifi cantly increased earnings for women in most cases, albeit on a low level, and in 
minor positive changes for men.

Nevertheless, this positive picture is clouded if one compares the level of earn-
ings before and after the turn of the millennium: Real earnings have hardly increased 
for men, and they are signifi cantly lower for women in 2000-2012 than in the years 
prior to the turn of the millennium. We already know from our discussion in Section 
2.3 that real earnings have tended to decrease rather than to increase in our obser-
vation period (see Table 1), and the stagnation of men’s earnings therefore comes as 
no surprise. We also noticed that increased female labour force participation is to a 
large extent based on part-time and marginal employment, which is refl ected in the 
activity patterns of our female sample members separating between 2000 and 2012 
(see above). Part-time (and also marginal) jobs pay signifi cantly less than full-time 
jobs (St atistisches Bundesamt/Wissenschaftszentrum Berlin für Sozialforschung 
2013: 132), which may go at least some way towards explaining the lower female 
earnings after the turn of the millennium.18 Be this as it may, this result fi ts nicely 

18 Of course, for a fair comparison of earnings before and after the turn of the millennium one 
would have to control for this and other job characteristics as well as for differences in human 
capital.



•    Miriam Bröckel, Hans-Jürgen Andreß300

with the earlier observation that public transfers play a much larger role in women’s 
household incomes in 2000-2012 (see Section 4.1 and Table 2).

4.4 Income and the effects of private and public redistribution

According to the earnings indicator in Table 4, women’s own income – at best – 
amounts to half of men’s own income from employment. It is therefore obvious 
that women, much more than men, need other sources of income to achieve the 
economic status that they had together with their husbands during marriage. On 
the other hand, men’s economic status is not as high as these data suggest because 
part of their income will be transferred to their former wives and children after sepa-
ration in the form of maintenance payments. To see the effects of private and public 
redistribution, we modelled this process in four steps: (1) factor incomes, (2) taxa-
tion, (3) payment and receipt of private transfers and (4) receipt of public transfers 
(including social security pensions). 

For this analysis steps (3) and (4) could be reversed. For this analysis we used 
the following information from the PEQUIV: pre-government household incomes 
(hhpregov), post-government household incomes (hhposgov), private transfers re-
ceived (hhprivat), public transfers (hhpublic). Additionally, we operationalized sup-
port payments being made for spouses and children (hhsupport) as described in 
Section 3.2. Post-government household incomes equal pre-government incomes 
minus taxes plus public transfers and social security pensions. Unfortunately, the 
PEQUIV includes private transfers received in pre-government household incomes, 
which are then subjected to a tax simulation programme. Support payments being 
made for spouses and children are not considered in the PEQUIV. In order to model 
private and public redistribution as the third and fourth steps, we had to deduct 
hhprivat and hhpublic from the corresponding PEQUIV household incomes. More 
specifi cally, we calculated the four redistributive steps as follows: (1) = hhpregov 
– hhprivat, (2) = hhposgov – hhprivat – hhpublic, (3) = hhposgov – hhsupport – hh-
public, (4) = hhposgov – hhsupport. Each of the results was then equivalized ac-
cording to equation (1).  

In a fi nal step (5), we looked at disposable incomes (incomes at step iv) minus 
payments made for the home (i.e. rents, interest and amortisation payments for 
mortgages and utility costs, but excluding heating costs) to account for the fi nancial 
strain resulting from residential mobility. Figure 1 shows, separately for men and 
women and for individuals separating before and after the turn of the millennium, 
the distribution of change scores observed at each step of this income redistribu-
tion process. The middle 50 percent of the distribution (ranging from the fi rst to 
the third quartile) are indicated by a vertical fat blue line, the median change score 
by a white point, and the whole distribution by a kernel density plot. If the median 
change score is below (above) the horizontal red line of no change (%Ii=0), then the 
majority (more than 50 percent) experiences an income loss (gain).

Four observations are noticeable: Firstly, there are clear-cut gender differences 
at each step of the redistribution process. The female median change score is al-
ways lower than its male equivalent, indicating that separation implies less positive 



The Economic Consequences of Divorce in Germany    • 301

Fig. 1: Changes in equivalised household income for men and women 
separating in 1984-1999 and 2000-2012

Notes: About 2 percent of all change scores top-coded at 200 percent. The distribution of 
change scores is shown for the following steps of the income redistribution process: (1) 
pre-government incomes, (2) ... minus taxes, (3) ... incl. private transfers, (4) ... plus public 
transfers, and (5) ... minus housing costs.

Source: SOEP 1984-2012, weighted data.



•    Miriam Bröckel, Hans-Jürgen Andreß302

and more negative income changes for women than for men. Secondly, the distri-
bution of change scores is more skewed for women, with the mass of the change 
scores more in the lower than in the upper end of the distribution, while the men’s 
distribution looks much more symmetrical. This again shows that the balance of 
positive and negative changes tends more in the negative direction for women than 
for men. However, thirdly, it seems as if these gender inequalities decrease with 
each step of the income redistribution process (see the distance between the female 
and male median change scores at each step in Fig. 1). In other words, taxation, 
support payments, and public transfers level out part of the gender inequalities. 
Taxation and support payments predominantly affect men: Their median change 
score drops when taxes are deducted from factor incomes (compare steps 1 and 
2), and when private transfers have been paid and received (compare steps 2 and 
3). Such a decrease is not visible for the women’s median change scores. Note also 
that the female score increases slightly in step (4), when public transfers are added. 
No similar increase is visible for men. In step (5), when housing costs are paid, the 
median change score decreases again for both genders. Fourthly, and fi nally, this 
pattern of income changes does not change dramatically over time. The violin plots 
for 1984-1999 and 2000-2012 look very much the same, except perhaps the male 
change scores in steps (1) and (2). They seem to be more positive in 1984-1999 (the 
median change scores are more clearly above the red reference line). But given the 
possibility of sampling errors, this may be an exaggeration. Therefore, in the next 
section, we will test the gender inequalities observed and possible changes over 
time for their statistical signifi cance.

4.5 Gender inequalities, different living conditions, and changes over 
time

In a series of logistic regression models, we estimated the probability of experienc-
ing a negative income change after separation (i.e. the probability of a negative 
change score). We analyse the probability of such income losses at each step of 
the income redistribution process (Models 1-5). Finally, at the level of disposable 
incomes (incomes at step 4), we control for various aspects of the respondents’ 
living situation before and after separation (Model 6). Our main focus is on gender 
inequalities and changes over time between the two periods before and after the 
turn of the millennium. There has been some concern as to whether logistic regres-
sion coeffi cients can be compared across models; i.e. in our case, across different 
steps of the income redistribution process (Auspurg/Hinz 2011; Breen/Karlson 2013; 
Breen et al. 2013; Karlson et al. 2012; Mood 2009). However, these concerns are not 
relevant for the interpretation of probabilities that are estimated from the regression 
coeffi cients (Long 1997: 50). The estimated probabilities are shown in Figure 2 and 
Table 5.19 

19 The estimated logistic regression coeffi cients, from which these probabilities were computed, 
are shown in Table A1 in the Appendix.



The Economic Consequences of Divorce in Germany    • 303

The probability of an income loss amounts to approximately 75 percent for 
women before the turn of the millennium irrespective of whether one looks at factor 
or at disposable incomes (see Fig. 2). Neither private nor public transfers cushion 
these negative consequences of marital disruption. Moreover, there is no change 
for women after the turn of the millennium: The probability of income loss is still 
about 75 percent.

The situation is slightly better for men: At the level of factor incomes before and 
after taxation, only about 40 percent of all men experience an income loss; and this 
risk is almost the same before and after the turn of the millennium. However, if one 
takes private and public transfers into account, the situation becomes less favour-
able for men. The increasing estimates for men indicate that the probability of an 
income loss increases with each step of the income redistribution process. In other 

Fig. 2: Estimated probability of income loss after separation by gender and 
income concept (in percent)

Notes: Probabilities for men (squares) and women (triangles) estimated from logistic re-
gression models (Models 1-5, see Table A1 in the Appendix). Gender differences are sig-
nifi cant (α=0.05) for each step of the income redistribution process except for disposable 
incomes minus housing costs (step v) in 1984-99.

Source: SOEP 1984-2012, weighted data, estimates based on logistic regression coeffi -
cients (Models 1-5, see Table A1).



•    Miriam Bröckel, Hans-Jürgen Andreß304

words, gender inequalities observed at the level of factor incomes decrease with 
each step. They even disappear (cease being signifi cant) in 1984-1999, once we con-
trol for housing costs (X2=0.58, df=1, p=0.45; all other differences are signifi cant at 
the 5 percent level).20 This overall picture of decreasing gender inequalities appears 
in both observation periods, with two notable differences:

1. Before the turn of the millennium, the probability of an income loss for men 
increases notably, from 41.2 percent to 57.3 percent, if one takes private 
transfers into account. Since a similar increase is not visible after the turn 
of the millennium, it seems as if private transfers became less important for 
adjusting gender inequalities in economic well-being after 2000.

2. Contrary to the situation before the turn of the millennium, gender inequali-
ties do not disappear once one controls for housing costs (X2=5.45, df=1, 
p=0.02).

All in all, however, it would be wrong to conclude that gender inequalities are 
more quickly equalised before the turn of the millennium, given that none of the co-
effi cients that tests for changes between the two observation periods is signifi cant 
(α=0.05).

Tab. 5: Estimated probability of income loss after separation by living 
arrangement (in percent)

Group Before 
separation 

After separation 1984-1999 2000-2012

t-2 t+1

Household 
type

New 
partner

Gainfully 
employed

Minor 
children

Men Women Men Women

1 2 no yes no 70.8 81.3 60.1 80.7b

2 2 yes yes no 44.2a 52.4a 51.3 55.4a

3 2 no no no 65.4 93.8ab 79.6 94.7a

4 2 no yes yes 69.6 70.7 65.5 90.1bc

5 1 no yes yes 59.7 74.3b 43.8 80.2

6 3 no yes no 83.3a 89.3 72.2a 61.8c

7 4 no yes no 48.6 60.5 74.3a 56.2

8 1 no no yes 53.6 91.0b 66.9 94.5b

Notes: Household type: 1 = male single earner, 2 = male main earner, 3 = dual earner, 4 = 
female main or single earner. Signifi cant effects with respect to a the corresponding char-
acteristic, b gender for each period, and c period for each gender (α=0.05). Tests based on 
linear restrictions on the logistic regression coeffi cients.

Source: SOEP 1984-2012, weighted data, estimates based on logistic regression coeffi -
cients (Model 6, see Table A1).

20 These and the following differences are tested by applying linear restrictions to the relevant 
parameters of the logistic regression model.



The Economic Consequences of Divorce in Germany    • 305

A natural extension of this analysis is to ask how these income risks depend on 
men’s and women’s living situations before and after separation. Almost a quarter 
(23.8 percent) of all respondents lived in a traditional male breadwinner household 
during marriage in which the husband was the single earner. However, the most 
common (40.5 percent) household type was a “modernised” male breadwinner 
household in which the husband was the main earner, but not the only one. Dual 
earner households (22.4 percent), or households in which the wife was the main 
or even the single earner (13.3 percent), were less common. Given the increase in 
female labour force participation during the entire observation period (see Section 
2.3), this distribution changed slightly after the turn of the millennium, with “mod-
ernised” male breadwinner households becoming more common (46.2 percent) at 
the expense of the traditional ones (17.9 percent). After separation, the most com-
mon living situation is being employed and living alone without a new partner and 
without any minor children (36.2 percent before and 42.9 percent after the turn 
of the millennium). The second most common living situation is almost identical, 
except for the fact that these men and women live together with minor children 
(21.4 percent before and 22.0 percent after the turn of the millennium).21 Only a mi-
nority lives together with a new partner (27.6 percent before and 21.6 percent after 
the turn of the millennium).

The fi rst row of Table 5 shows the estimated probability of an income loss for the 
most common living situation before and after separation; i.e. men and women who 
have lived in a “modernised” male breadwinner household during marriage, but live 
alone and are unemployed after separation (termed group 1 or simply the reference 
group). The probabilities are derived from a logistic regression model regressing 
disposable incomes (incomes at step 4 neglecting housing costs) on selected as-
pects of the respondent’s living situation before and after separation (see Model 6 
in Table A1 in the Appendix). The probabilities in the fi rst row basically tell the same 
story as Figure 2: An income loss is more probable for women, although the differ-
ences are not particularly pronounced at the level of disposable incomes that are 
shown here. The following rows 2-7 in Table 5 show how this pattern would change 
if one of the characteristics of this reference group were different (group 8 will be 
discussed later). For example, living together with a new partner after separation 
(group 2) has a signifi cant positive effect both for men and women: All loss prob-
abilities are smaller than in the reference group. Unemployment (group 3) increases 
the economic risks (except for men before the turn of the millennium), as does liv-
ing with minor children (group 4) (but only after the turn of the millennium). On 
the other hand, separating from a traditional male breadwinner household entails 
fewer economic risks (group 5). The same is true for individuals separating from 
households in which the wife was the main or the single earner (group 7) (except for 
men after the turn of the millennium), while individuals separating from dual earner 

21 This living situation is naturally more common among women, but differentiating between two 
observation periods, two genders and various living conditions results in highly differentiated 
subgroups of the data, which then provide highly unreliable information due to their small sam-
ple sizes. However, these more differentiated data are available from the authors on request



•    Miriam Bröckel, Hans-Jürgen Andreß306

households experience greater income losses (group 6) (except women after the 
turn of the millennium). But depending on the standard errors of the logistic regres-
sion coeffi cients, these descriptive differences may not be signifi cant. Therefore, 
we formally tested whether there are signifi cant changes if an individual in the refer-
ence group cohabits with a new partner, is gainfully employed, lives together with 
minor children, or experiences separation within another type of marital household. 
These tests are performed for each gender separately for each observation period, 
and a signifi cant effect is indicated with the superscript (a) in the corresponding cell 
of Table 5. 

Each row of Table 5 defi nes another subgroup of our sample. The values of all 
probabilities are however determined by the parameters of Model 6 that have been 
estimated by taking all subgroups into account (and not only the ones shown in 
Table 5). The tests apply linear restrictions on the relevant logistic regression co-
effi cients of the model (cf. footnote 21), and hence test the overall effects of new 
partners, employment, parenting, and type of marital household for each gender 
and observation period. Group 8 is not included in these tests because it is differ-
ent from the reference group with regard to several characteristics (and not just in 
terms of one characteristic as in the case of groups 2-7).

Based on these tests, we can safely say that cohabitation with a new partner 
reduces the income risks of separation (except for men after the turn of the mil-
lennium). Similarly, being gainfully employed after separation makes a signifi cant 
(positive) difference; but only for women, possibly because most men are already 
gainfully employed before separation (see Section 4.3). Living together with minor 
children after separation has no effect, and this holds for both genders. Finally, the 
question of economic independence during marriage is also important for men. If 
they separate from a household where either both partners contributed an income 
or where the wife was the main or single earner, men’s loss probability increases 
signifi cantly (for men separating from female main/single earner households, the 
effect is only signifi cant after the turn of the millennium).

Our main interest however lies in gender inequalities and changes over time. 
All in all, there are signifi cant gender differences to the disadvantage of women in 
both periods.22 However, this overall picture does not necessarily hold true for all 
the subgroups in the sample. If it is true for groups 1-8 in Table 5 (α=0.05), then the 
corresponding cell in Table 5 is marked with the superscript (b). According to these 
group-specifi c tests, women in the reference group faced signifi cantly more eco-
nomic risks before the turn of the millennium if they were unemployed (group 3) or 
if they separated within a traditional male breadwinner household (group 5). After 
the turn of the millennium, this is also true for all women in reference group 1 and 
who live together with minor children (group 4).

22 For 2000-2014 these differences are only signifi cant at the 10 percent level, which is at odds 
with the tests for Figure 2, where they were signifi cant at the 5 percent level. Note however that 
Model 6 is based on fewer observations than Model 4 (see Table A1 in the Appendix). 



The Economic Consequences of Divorce in Germany    • 307

With respect to changes over time, the overall tests show no signifi cant differenc-
es for either men or women. Looking at the probabilities, this comes as no surprise: 
Period 2 probability is larger than its period 1 equivalent in some subgroups, whilst 
in others it is exactly the other way round. Moreover, if group-specifi c differences 
exist, they may not be signifi cant. Nevertheless, some are signifi cant (α=0.05), and 
this is indicated with the superscript (c) in the corresponding cell of Table 5. The 
economic risks faced by women in the reference group who live together with mi-
nor children (group 4) have increased. On the other hand, if they have separated 
from a dual earner household (group 6), their economic risks have decreased.

Finally, Table 5 looks at a subgroup that includes all the high risk factors for wom-
en. Group 8 comprises individuals who separated within traditional male breadwin-
ner households and are unemployed after separation while taking care of minor 
children without having a new partner. As expected, this situation implies signifi -
cantly higher economic risks for women than for men. This living situation fortu-
nately becomes less common for women compared to the 1960s and 1970s. About 
one woman in seven fell into this group during our observation period (1984-1999: 
14.4 percent, 2000-2012: 13.2 percent).

5 Summary and discussion

Our analyses have shown that the economic consequences of divorce are still more 
negative for women than for men after the turn of the millennium. Even now it is 
the woman who takes care of the children in most cases, and the majority of these 
women does not receive any child support payments from their former spouse. 
But we should also mention that men receive virtually no support payments if they 
live together with minor children after separation. One man in three, but only one 
woman in twenty, makes support payments, be it for the children or the former 
spouse. Economically, support payments are of minor importance in the budgets 
of the separated individuals. Moreover, they play a less pronounced equalising role 
in the redistribution process after the turn of the millennium, as we expected in H

4
.

If one counts all private transfers from outside the household, which include the 
aforementioned support payments, these only constitute 1 percent of men’s house-
hold incomes and 9 percent of women’s household incomes. Private transfers only 
make a difference if women live together with minor children after separation. Nev-
ertheless, they constitute only 12 percent of their household incomes. Contrary to 
our expectations in H

5
, living together with minor children becomes a disadvantage 

for women after the turn of the millennium, and this is a clear change as against 
the situation pertaining prior to 2000, when minor children did not increase the risk 
of income loss for either women or men. A more detailed analysis of the change 
scores shows that the change occurred in the upper tail of the distribution, and 
therefore for women with the least negative income changes. Before the turn of 
the millennium, the upper 25 percent of all women experienced a positive income 
change after separation of at least 8 percent, while after the turn of the millennium, 
they also experienced negative income changes of 17 percent at most. Given the 



•    Miriam Bröckel, Hans-Jürgen Andreß308

expansion of the public child care system in the 2000s, this is a surprising result and 
needs further investigation. One possibility is that child care costs offset increases 
in earnings, although child care places are often free of charge for single mothers. 
The earnings increase after separation may also be negligible because women with 
minor children either do not increase their economic activity or choose marginal 
and part-time employment. It could also be the case that separated women with 
minor children make no use of public child care at all, or that it will still take some 
time until this infrastructural change becomes effective.

Public transfers are much more important than private ones, both for men and 
women. Dependence on public transfers is dramatic in some cases: One man in ten 
and one woman in four has to rely on a household income after separation in which 
public transfers make up more than half of the budget. Due to opposing changes in 
minimum income and family policies after the turn of the millennium, we refrained 
from hypothesising about the economic dependence of separated individuals on 
public transfers. But after comparing the data for our two observations periods, we 
have to conclude that economic dependence on public transfers increased for both 
genders after the turn of the millennium.

Both men and women try to ease the fi nancial strain accompanying separation: 
on the one side by economising on consumption and on the other, by increasing 
their own economic activity. Cutting down on expenses is most visible with the 
most important cost factor: the residential home. Almost two-thirds of the spouses 
leave the family home after separation. Although they move to smaller and cor-
respondingly less expensive homes, the increase in housing costs relative to their 
post-separation incomes is still considerable, but still lower than for those individu-
als who stay in the former family home. In a way, residential mobility after separa-
tion is a forced choice. This might also explain the comparatively high mobility rate 
of women with children. Contrary to the legal norm stating that children should stay 
in their familiar environment, they leave the family home more often than men with 
children. Expanding economic activity is a feasible option especially for women 
because only a minority works full-time before separation. This is not the case for 
men, who mostly work full-time before separation. But increased economic activity 
for women often means part-time or marginal work and signifi cantly lower earnings 
than men. Hence, women need other sources of income besides their own labour 
income to maintain at least part of their former standard of living.

But even taking all sources of household income into account, our analyses 
showed that separation implies fewer positive and more negative changes in house-
hold income for women than for men. Some women naturally also experience posi-
tive changes, while some men suffer from income losses, but overall the balance of 
positive and negative changes runs more in the negative direction for women than 
for men. More refi ned tests showed that the changes in household income are sig-
nifi cantly more negative for women than for men. Moreover, these inequalities per-
sist if one controls for the economic situation of the former marital household and 
the living conditions after separation; i.e. whether the person has a new partner, 
lives together with minor children, and is gainfully employed. However, women’s 
disadvantages become more slight with each step of the income redistribution pro-



The Economic Consequences of Divorce in Germany    • 309

cess. They are greatest at the level of factor incomes, and become smaller when 
taxation, private and public transfers are taken into account. Finally, they actually 
become insignifi cant at the level of disposable incomes once we deduct housing 
costs, but only in 1984-1999. Hence, if one takes all the changes into account, in-
cluding residential mobility, the economic consequences of divorce seem to be 
much more balanced between the genders than it appears at fi rst sight.

Our larger data base also allowed us to scrutinize change over time in greater 
detail. On the whole, due to the overall stability of real earnings, we did not antici-
pate any change in the economic consequences of marital disruption (H

1
), and our 

hypothesis was clearly supported by our regression models. We did not expect any 
change with respect to gender inequalities because increased female labour force 
participation is based mostly on part-time and marginal jobs after the turn of the 
millennium. Correspondingly, the women in our sample mostly took up part-time 
and marginal jobs when they increased their economic activity. Therefore, as as-
sumed in H

2
, we found no change of the gender inequalities after the turn of the mil-

lennium. Nevertheless, there are some instances of change: Men separating from 
dual earner households or from households where the woman was the main or 
even the single earner more often experienced negative changes after the turn of 
the millennium, although this risk was already visible for dual earner households in 
1984-1999. In other words, in line with H

3
, the effects of an increasing dependence 

on female incomes can be seen at least for some, though rather small subgroups of 
men. Moreover, economic independence pays for women whose economic risks of 
separation have decreased after the turn of the millennium if they have been living 
in a dual earner marital household. A distressing result is of course that cohabiting 
with minor children became more of a risk for women after the turn of the millenni-
um than for men. More research is needed as to why recent policy changes towards 
more female employment and towards better reconciliation of family and work life 
are having so little effect on the economic consequences of marital disruption.

References

Aisenbrey, Silke; Evertsson, Marie; Grunow, Daniela 2009: Is There a Career Penalty for 
Mothers’ Time Out? A Comparison of Germany, Sweden and the United States. In: 
Social Forces 88,2: 573-606 [doi: 10.1353/sof.0.0252].

Andreß, Hans-Jürgen; Bröckel, Miriam 2007: Marital disruption in Germany: Does the 
conservative welfare state care? Changes in material well-being and the effects of 
private and public transfers. In: Schmollers Jahrbuch 127,2: 193-226.

Andreß, Hans-Jürgen; Güllner, Miriam 2001: Scheidung als Armutsrisiko. In: Barlösius, 
Eva; Ludwig-Mayerhofer, Wolfgang (Eds.): Die Armut der Gesellschaft. Opladen: Les-
ke + Budrich: 169-197 [doi: 10.1007/978-3-322-99629-9_6].

Andreß, Hans-Jürgen et al. 2006: The Economic Consequences of Partnership Dissolu-
tion – A Comparative Analysis of Panel Studies from Belgium, Germany, Great Britain, 
Italy, and Sweden. In: European Sociolgical Review 22,5: 533-560 [doi: 10.1093/esr/
jcl012].



•    Miriam Bröckel, Hans-Jürgen Andreß310

Andreß, Hans-Jürgen et al. 2003: Wenn aus Liebe rote Zahlen werden. Über die 
wirtschaftlichen Folgen von Trennung und Scheidung. Wiesbaden: Westdeutscher 
Verlag.

Auspurg, Katrin; Hinz, Thomas 2011: Gruppenvergleiche bei Regressionen mit binären 
abhängigen Variablen – Probleme und Fehleinschätzungen am Beispiel von Bildungs-
chancen im Kohortenverlauf. In: Zeitschrift für Soziologie 40,1.

BMFSFJ (Bundesministerium für Familie, Senioren, Frauen und Jugend) 2010: Infor-
mation zum Kinderförderungsgesetz (KiföG) vom 13.07.2010 [http://www.bmfsfj.de/
BMFSFJ/gesetze,did=133282.html, 15.05.2014]. 

BMFSFJ (Bundesministerium für Familie, Senioren, Frauen und Jugend); Statistisches 
Bundesamt 2003: Wo bleibt die Zeit? Die Zeitverwendung der Bevölkerung in Deutsch-
land 2001/02. Wiesbaden.

Breen, Richard; Karlson, Kristian Bernt 2013: Counterfactual Causal Analysis and Nonlin-
ear Probability Models. In: Morgan, Stephen L. (Ed.): Handbook of Causal Analysis for 
Social Research. Springer Netherlands: 167-187 [doi: 10.1007/978-94-007-6094-3_10].

Breen, Richard; Karlson, Kristian Bernt; Holm, Anders 2013: Total, Direct, and Indirect 
Effects in Logit and Probit Models. [Article]. In: Sociological Methods & Research 42,2: 
164-191 [doi: 10.1177/0049124113494572].

Buhmann, Brigitte et al. 1988: Equivalence Scales, Well-Being, Inequality, and Poverty: 
Sensitivity Estimates across ten Countires using the Luxembourg Income Study (LIS) 
Database. In: Review of Income and Wealth 34: 115-142 [doi: 10.1111/j.1475-4991.1988.
tb00564.x].

Cornelißen, Waltraud 2005: Gender-Datenreport. Erster Datenreport zur Gleichstellung 
von Frauen und Männern in der Bundesrepublik Deutschland im Auftrag des Bun-
desministeriums für Familie, Senioren, Frauen und Jugend. München: Deutsches Ju-
gendinstitut e.V. in Zusammenarbeit mit dem Statistischen Bundesamt.

Drasch, Katrin 2013: The re-entry of mothers in Germany into employment after fami-
ly-related interruptions. In: IAB-Bibliothek 343 (Dissertationen). Bielefeld: WBV [doi: 
10.3278/300813w].

FamRZ, Zeitschrift für das gesamte Familienrecht mit Betreuungsrecht, Erbrecht, Ver-
fahrensrecht, Öffentlichem Recht, various editions.

Güllner, Miriam 2000: Analyse familiärer Ereignisse (Trennung, Scheidung, Tod) mit 
Daten des Sozio-oekonomischen Panels. Bestimmung der Untersuchungs- und Ver-
gleichsgruppen anhand der Längsschnittdatei ”alleWellen_long.sav”. Bielefeld: Ar-
beitspapier Nr. 2 des Forschungsprojektes “Die wirtschaftlichen Folgen von Trennung 
und Scheidung für Familien” im Auftrag des Bundesministeriums für Familie, Senio-
ren, Frauen und Jugend [http://eswf.uni-koeln.de/forschung/wts/].

Hintze, Jerry L.; Nelson, Ray D. 1998: Violin Plots: A Box Plot-Density Trace Synergism. 
In: The American Statistician 52,2: 181-184 [doi: 10.1080/00031305.1998.10480559].

Holden, Karen C.; Smock, Pamela J. 1991: The Economic Costs of Marital Dissolution: 
Why Do Women Bear a Disproportionate Cost? In: Annual Review of Sociology 17: 51-
78 [doi: 10.1146/annurev.so.17.080191.000411].

Karlson, Kristian Bernt; Holm, Anders; Breen, Richard 2012: Comparing Regression 
Coeffi cients Between Same-sample Nested Models Using Logit and Probit: A New 
Method. In: Sociological Methodology 42,1: 286-313 [doi: 10.1177/0081175012444861].

Keller, Matthias; Haustein, Thomas 2012: Vereinbarkeit von Familie und Beruf. Ergeb-
nisse des Mikrozensus 2010. In: Wirtschaft und Statistik 2012 (Januar): 30-50.



The Economic Consequences of Divorce in Germany    • 311

Miriam Bröckel. Universität Bielefeld, SFB 882 Von Heterogenitäten zu Ungleichheiten. 
Bielefeld, Germany. E-Mail: miriam.broeckel@uni-bielefeld.de
URL: https://ekvv.uni-bielefeld.de/pers_publ/publ/PersonDetail.
jsp?personId=27030353&lang=en

Prof. Dr. Hans-Jürgen Andreß (). Universität zu Köln, Wirtschafts- und Sozialwissen-
schaftliche Fakultät, Institut für Soziologie und Sozialpsychologie (ISS). Köln, Germany.
E-Maill: hja@wiso.uni-koeln.de
URL: http://www.iss-wiso.uni-koeln.de/andress.html?&L=0

Kenny, Lawrence W. 1983: The Accumulation of Human Capital During Marriage by 
Males. In: Economic Inquiry 21: 223-231 [doi: 10.1111/j.1465-7295.1983.tb00627.x].

Long, J. Scott 1997: Regression models for categorical and limited dependent variables. 
Thousand Oaks: Sage.

MDR, Monatsschrift für Deutsches Recht, various editions.

Mood, Carina 2009: Logistic Regression: Why We Cannot Do What We Think We Can 
Do, and What We Can Do About It. European Sociological Review, 26(1), 67-82 [doi: 
10.1093/esr/jcp006]

Saldeen, Åke 1995: Family Law in Sweden. In: Hamilton, Carolyn, Standley, Kate; Hod-
son, D. (Eds.): Family Law in Europe. London/Dublin/Edinbourgh: Butterworth: 472-
510.

Sørensen, Annemette 1994: Women’s Economic Risk and the Economic Position of Sin-
gle Mothers. In: European Sociological Review 10: 173-188.

Statistisches Bundesamt; Wissenschaftszentrum Berlin für Sozialforschung 2013: 
Datenreport 2013. Ein Sozialbericht für die Bundesrepublik Deutschland. Bonn: Bun-
deszentrale für politische Bildung.

Statistisches Bundesamt 2012: Indikatorenbericht 2012 – Nachhaltige Entwicklung in 
Deutschland. Wiesbaden: Statistisches Bundesamt.

Statistisches Bundesamt 2013: Statistiken der Kinder- und Jugendhilfe. Kinder und tätige 
Personen in Tageseinrichtungen und in öffentlich geförderter Kindertagespfl ege am 
01.03.2013. Wiesbaden: Statistisches Bundesamt.

Statistisches Bundesamt 2014a: Gender Pay Gap 2013 bei Vollzeitbeschäftigten be-
sonders hoch. Pressemitteilungen des Statistischen Bundesamtes. Retrieved from 
ht tps://w w w.destatis.de/DE/PresseService/Presse/Pressemit teilungen/2014/03/
PD14_104_621.html, 12.02.2015).

Statistisches Bundesamt 2014b: Volkswirtschaftliche Gesamtrechnungen. Fachserie 18, 
Reihe 1.1. Wiesbaden: Statistisches Bundesamt.

Wagner, Gert G.; Frick, Joachim R.; Schupp, Jürgen 2007: The German Socio-Economic 
Panel Study (SOEP) – Scope, Evolution and Enhancements. In: Schmollers Jahrbuch 
1271: 139-169 [doi: 10.2139/ssrn.1028709].

Wrohlich, Katharina et al. 2012: Elterngeld Monitor: Endbericht; Forschungsprojekt im 
Auftrag des Bundesministeriums für Familie, Senioren, Frauen und Jugend. In: DIW 
Berlin – Politikberatung kompakt 61.

Date of submission: 07.08.2014  Date of acceptance: 12.01.2015



•    Miriam Bröckel, Hans-Jürgen Andreß312

Appendix

We estimated a series of logistic regression models. The most complex Model 6 is 
defi ned as follows:

G and P are dummy variables indicating females (G=1) and the second observa-
tion period (P=1). X is a row vector of covariates. In case of Models 1-5, the vector 
X only includes a “variable” equal to one. In Model 6 it also includes dummies for 
new partner, gainful employment, and cohabitation with minor children. α, β, γ, and 
δ are suitable column vectors of regression coeffi cients. Their estimates are shown 
in columns 3-6 of Table A1.

ln
Pr %Ii<0

1- Pr %Ii<0
=X +G·X +P·X +G·P·X  

Tab. A1: Log-odds of income loss after separation (logistic regression 
coeffi cients)

Model Variables 1984-1999 2000-2012

Men Women Men Women n Pseudo R
2
 

α β γ δ

1 (1) Pre-government incomes -0.4497** 1.5636*** 0.1477 -0.2280 1362 0.090

2 (2) ... minus taxes -0.3545* 1.4807*** 0.1802 -0.1021 1362 0.087

3 (3) ... incl. private transfers 0.2938* 0.7786*** -0.2355 0.2841 1362 0.037

4 (4) ... plus public transfers 0.5341*** 0.5880** -0.2577 0.2466 1362 0.023

5 (5) ... minus housing costs 0.8295*** 0.1879 -0.3028 0.4336 1168 0.009

6 Reference groupa 0.1995 2.7017*** 0.2740 -1.1148 916 0.123

New partner (t-1) -1.1203** -0.2513 0.7622 -0.5998

Gainfully employed (t+1) 0.2498 -1.5024* -1.2029 1.0072

Living with children (t+1) -0.0568 -0.5312 0.2888 1.0830

Husband main earner (t-2) 0.4363 -0.6178 0.4520 0.5450

Dual earner (t-2) 1.1606* -0.6911 0.2758 -0.8752

Wife single/main earner (t-2) -0.5039 -0.7167 2.0449* -1.1856

Notes: a Men separating between 1984 and 1999 from a male breadwinner household, 
without a new partner, not gainfully employed, and living without minor children after 
separation. * p<0.05, ** p<0.01, *** p<0.001.

Source: SOEP 1984-2012, weighted data.



Published by / Herausgegeben von
Prof. Dr. Norbert F. Schneider

Federal Institute for Population Research 
D-65180 Wiesbaden / Germany

Managing Editor /
Verantwortlicher Redakteur
Frank Swiaczny

Assistant Managing Editor /
Stellvertretende Redakteurin
Katrin Schiefer

Copy Editor (German) /
Lektorat (deutsch)
Dr. Evelyn Grünheid

Layout / Satz
Beatriz Feiler-Fuchs

E-mail: cpos@bib.bund.de

Scientifi c Advisory Board / 
Wissenschaftlicher Beirat
Paul Gans (Mannheim) 
Johannes Huinink (Bremen) 
Michaela Kreyenfeld (Rostock)
Marc Luy (Wien) 
Clara H. Mulder (Groningen) 
Notburga Ott (Bochum) 
Peter Preisendörfer (Mainz)
Zsolt Spéder (Budapest)

Comparative Population Studies

www.comparativepopulationstudies.de

ISSN: 1869-8980 (Print) – 1869-8999 (Internet)

Board of Reviewers / Gutachterbeirat
Martin Abraham (Erlangen)
Laura Bernardi (Lausanne)
Hansjörg Bucher (Bonn)
Claudia Diehl (Konstanz)
Andreas Diekmann (Zürich)
Gabriele Doblhammer-Reiter (Rostock)
Jürgen Dorbritz (Wiesbaden)
Anette Eva Fasang (Berlin)
E.-Jürgen Flöthmann (Bielefeld)
Alexia Fürnkranz-Prskawetz (Wien)
Beat Fux (Salzburg)
Joshua Goldstein (Berkeley)
Karsten Hank (Köln)
Sonja Haug (Regensburg)
Hill Kulu (Liverpool)
Aart C. Liefbroer (Den Haag)
Kurt Lüscher (Konstanz)
Emma Lundholm (Umeå)
Nadja Milewski (Rostock)
Dimiter Philipov (Wien)
Roland Rau (Rostock)
Tomáš Sobotka (Wien)
Jeroen Spijker (Barcelona)
Olivier Thévenon (Paris)
Helga de Valk (Brussel)
Heike Trappe (Rostock)
Michael Wagner (Köln)

© Federal Institute for Population Research 2015 – All rights reserved