Available via license: CC BY
Content may be subject to copyright.
TYPE Original Research
PUBLISHED 16 December 2022
DOI 10.3389/fhumd.2022.1037778
OPEN ACCESS
EDITED BY
Agnieszka Kanas,
Erasmus University
Rotterdam, Netherlands
REVIEWED BY
Philipp Jaschke,
Institute for Employment Research
(IAB), Germany
Verena Seibel,
Utrecht University, Netherlands
*CORRESPONDENCE
Jana Kuhlemann
jana.kuhlemann@mzes.uni-mannheim.de
SPECIALTY SECTION
This article was submitted to
Refugees and Conflict,
a section of the journal
Frontiers in Human Dynamics
RECEIVED 06 September 2022
ACCEPTED 22 November 2022
PUBLISHED 16 December 2022
CITATION
Kuhlemann J (2022) Refugees’ time
investments—Dierences in the time
use of refugees, other immigrants, and
natives in Germany.
Front. Hum. Dyn. 4:1037778.
doi: 10.3389/fhumd.2022.1037778
COPYRIGHT
©2022 Kuhlemann. This is an
open-access article distributed under
the terms of the Creative Commons
Attribution License (CC BY). The use,
distribution or reproduction in other
forums is permitted, provided the
original author(s) and the copyright
owner(s) are credited and that the
original publication in this journal is
cited, in accordance with accepted
academic practice. No use, distribution
or reproduction is permitted which
does not comply with these terms.
Refugees’ time
investments—Dierences in the
time use of refugees, other
immigrants, and natives in
Germany
Jana Kuhlemann1,2*
1Mannheim Centre for European Social Research (MZES), University of Mannheim, Mannheim,
Germany, 2Graduate School of Economic and Social Sciences, Center for Doctoral Studies in Social
and Behavioral Sciences, University of Mannheim, Mannheim, Germany
Introduction: Since the 2015/16 refugee influx to Germany and other
European countries, these host societies have been challenged with the
integration of culturally distant refugees. These recent arrivals can strategically
invest their time in activities promoting their integration, thereby rendering
time use as a channel of integration. Refugees are a vulnerable group that
diers from other immigrants with respect to their migration motivation,
experience, and conditions in the receiving countries. Accordingly, refugees
might also dier from other immigrants with respect to their time use. This
might play a role in explaining dierences in refugees’ and other immigrants’
integration outcomes.
Methods: Using a cluster analysis approach, this contribution (1) descriptively
examines whether and to what extent refugees’ time use diers from that
of other immigrants and the host-country population in Germany and (2)
examines the role of refugees’ legal status for their time use. The study
examines time allocation to dierent activities of refugees, other first-
generation immigrants, and native Germans, using data collected from 2016
to 2019 of the German Socio-Economic Panel, including the IAB-BAMF-SOEP
Survey of Refugees and the IAB-SOEP Migration Sample.
Results and discussion: Results from (1) the cluster analysis approach
show dierent clusters of time use patterns for the three population groups
of refugees, other immigrants, and natives. For native Germans and other
immigrants, the dominant time use cluster is characterized by full-time
investment in employment activities. For refugees, the dominant time use
pattern is characterized by low overall invested hours to the measured activities
(low activity cluster). In contrast to the other two groups, a cluster of refugees
predominantly allocating their time to employment activities is not found.
Pooled analyses (2) of the role of refugees’ legal status show some evidence
that those who have a form of protection status, in comparison to those who
have asylum seeker status, have a lower probability to display childcare- and
household-related activities than to report low activity. However, fixed eects
analyses show that refugees receiving a positive decision on their asylum
application do not change with respect to their time use patterns.
KEYWORDS
refugees, time use, cluster analysis, migration, integration
Frontiers in Human Dynamics 01 frontiersin.org
Kuhlemann 10.3389/fhumd.2022.1037778
Introduction
In 2015 and 2016, Germany and other OECD countries
experienced a large influx of refugees seeking protection
from war, persecution, and violence in their home countries
(OECD/EU, 2018;Jaschke and Kosyakova, 2021). Refugees can
be distinguished from other immigrant groups by the forced
character of their migration (Echterhoff et al., 2020;Kogan and
Kalter, 2020). Coming from destabilized countries (Kosyakova
and Brücker, 2020), refugees do not have a perspective of
returning to their home countries in the short-term (Brücker
et al., 2020). The reasons for leaving their home countries
make the group of refugees, who arrive in countries like
Germany, a very selective one. Previous research has shown
that refugees from the 2015/16 influx to Germany and other
European countries are selective in so far that they have
higher human capital and education than their counterparts
who stayed in the home countries (Guichard, 2020;Aksoy
and Poutvaara, 2021). Before and during their migration,
many refugees experience trauma (Mansouri and Cauchi, 2007;
Ambrosetti et al., 2021). This and the uncertainty and waiting
in the asylum process of the host country lead to refugees’ high
post-migration stress (Ambrosetti et al., 2021). These factors of
refugees’ selectivity, their reasons for migration, their migration
experience, and the restrictions they face in the asylum process
in the host countries, distinguish refugees from other immigrant
groups. More importantly, these factors restrict refugees in
their opportunity and ability to integrate into the host society
(Ambrosetti et al., 2021).
The integration of refugees and immigrants in general is a
central goal of receiving societies and benefits these societies
as well as the immigrants (Ager and Strang, 2008;Cheung
and Phillimore, 2014). Integration is reflected in immigrants’
participation in the labor market or in the educational system, in
their housing and their health conditions in comparison to the
native population of the host country (Ager and Strang, 2008).
In short, integration describes the opportunity for immigrants
to participate in life in the host country (Seidle and Joppke,
2012;Echterhoff et al., 2020). Immigrants can make active
investments in their integration by strategically allocating time
to activities that foster such integration in their everyday lives.
This means that the allocation of time to activities on a regular
everyday basis can be an indicator of immigrants’ integration
efforts. In addition, the resulting time use patterns themselves
can be an indicator of integration, in so far that immigrants who
display similar time use patterns as the members of the receiving
society can be considered highly integrated into the host society.
Integration via similarity in time use patterns mainly shows
integration in terms of engagement in activities and habits, but I
argue that the prioritization of investing time in some activities
over others also reflects values and beliefs. For instance, the
decision of immigrants to invest time in going to school in the
host country reflects that they value education. Furthermore,
the amount of time invested in this activity is an indicator
of how much they value the activity. Investing 8 h a day in
educational activities thereby reflects a higher value being placed
on education than investing 3 h a day. Hence, the time use of
immigrants, in terms of what activities time is invested in and
how much of it, is an important indicator of integration that
needs to be examined.
The time use of immigrants has rarely been studied
as a whole and has rather been investigated in isolated
activities. Most prominently, time investment in human capital
accumulation, labor market activities, or housework has been
examined (e.g., Cortes, 2004;Vargas and Chavez, 2010;Ribar,
2012). Previous research on the time use of refugees is mostly
qualitative or looked at isolated activities as well (e.g., Cortes,
2004;Brekke, 2010). The goal of the present paper is to
examine time use more holistically by describing it with a cluster
analysis approach. This approach uses the time individuals
allocated to different activities and generates clusters or groups
of similar time use. These groups represent a categorization of
time use patterns for the population in focus (refugees), which
can be descriptively compared to the categorizations of time
use patterns found for other populations (other immigrants
and natives).
Given the differences between refugees and other
immigrants as stated above, refugees’ and other immigrants’
time use is likely to differ. However, refugees’ agency in time
allocation, depending on the institutional restrictions in the
host society, might further increase this difference. The present
paper focuses on refugees in Germany, which is one of the
European countries with the highest share of received asylum
seekers from the 2015/16 influx (OECD, 2017), and therefore
provides an ideal setting to study this heterogeneous group of
refugees in the context of a European host society. Refugees
face several institutional and legal barriers when first arriving in
Germany. The amount of time since arriving in Germany and
their legal status determine refugees’ access to the labor market,
their opportunity to move between municipalities as well as
German federal states, and their financial situation. A positive
decision on their asylum application lifts many of these barriers,
which increases refugees’ agency in time allocation and thereby
increases their integration prospects into the host society.
Hence, this paper (1) describes refugees’ time use patterns and
contrasts them with the time use patterns of other immigrants
and the native German population. Additionally, it examines
(2) the role of refugees’ legal status for their time use by looking
at how a refugee’s time use changes after receiving a positive
decision on their asylum application.
Frontiers in Human Dynamics 02 frontiersin.org
Kuhlemann 10.3389/fhumd.2022.1037778
Theoretical considerations and
previous research on time use
Time use
Time is a limited resource, and the choice of how to
allocate it has implications for many areas of individuals’ lives
(Zerubavel, 1985). Allocating time to one activity means that
one has less time available to allocate to other activities on each
day, given that the day has a fix amount of 24 h. How one
decides to spend this time is a product of external requirements
(if one has a job, he/she is required to work a set amount of
time), personal preferences (if one prefers to do sports rather
than to take a language course, he/she invests time in sports),
and other potential factors. In any case, allocating time to an
activity serves a certain purpose, such as making money, trying
to keep being healthy, or developing interpersonal relationships.
This reasoning is in line with Becker’s (1965) economic theory
of the allocation of time, stating that individuals maximize their
utility by allocating time to different activities. Depending on
the individual and the context the individual lives in, utility
takes different forms, such as income, productivity, or wellbeing
(“psychic income”; Becker, 1965, p. 498). However, not only
the composition of utility, in other words individuals’ goals,
but also the way of maximizing this utility differs between
individuals in terms of how they allocate time to different
activities to achieve these goals (Becker, 1965). The combination
of goals that individuals want to reach, in other words their
utility, and their strategy of reaching them therefore determines
their overall time use patterns. While the goals that are
leading time investment should differ based on individuals’
characteristics and preferences, they are also likely to differ
between the populations of refugees, other immigrants, and
natives due to structural and systematic differences between
them. Consequently, this would suggest differences in the time
use patterns of the three populations.
The time use of refugees, other
immigrants, and natives
In order to understand how refugees’ time use might differ
from other immigrants’ and natives’ time use, it is crucial to
look at the systematic and structural differences between these
groups of the population. Refugee migration is characterized by
being forced due to war, violence, or persecution in the home
country (Echterhoff et al., 2020;Kogan and Kalter, 2020). Their
motivation to leave their home countries is not mainly driven by
pull factors of the host countries, but rather by push factors of
the country of origin (Borjas, 1987;Chiswick, 1999;Aksoy and
Poutvaara, 2021). The process of migration of refugees itself is
often dangerous and long (Echterhoff et al., 2020), where they
have to rely on smugglers, suffer physical deprivation and harm,
or experience other life-threatening situations and trauma on
their way to the host countries (Mansouri and Cauchi, 2007;
Gillespie et al., 2018;Alencar et al., 2019;Ambrosetti et al., 2021).
Refugees come to the host country without a visa and have to
apply for asylum before receiving legal status, whereas other
immigrants apply for visas before migrating. This means that
other immigrants know the legal conditions and preliminary
duration of their stay in advance. In contrast, refugees cannot
know the outcome of their asylum process and length of their
stay when they arrive in the host country. After migrating,
refugees might also be more prone to experiencing post-
migration stress compared to other immigrants (Li et al., 2016).
Post-migration stress is determined by factors like refugees’
legal status, their housing situation, labor market access, and
host-country language skills (Ambrosetti et al., 2021). Hence,
in contrast to most other immigrants, refugees’ migration
motivation and migration process involve a lot of traumatic
experiences that negatively affect their mental health (Mansouri
and Cauchi, 2007;Lindert et al., 2009;Allsopp et al., 2014;
Kogan and Kalter, 2020;Ambrosetti et al., 2021) and physical
health (Allsopp et al., 2014;Jaschke and Kosyakova, 2021). These
health issues are further amplified by the circumstances that
refugees face in the host countries. Health issues affect refugees’
everyday lives (Brücker et al., 2019) and make them more prone
to having difficulties adapting to the host country compared to
other immigrants. In addition, they might affect refugees’ ability
to use their time effectively, resulting in differences in the time
use patterns of refugees and other immigrants (Expectation 1).
Due to the forced character of refugees’ migration, they
have little time to prepare for migrating to a host country in
comparison to other immigrants. This means that refugees have
less time to take language courses, make additional money for
financial security, or invest in educational attainment that is
transferable to the host country. Hence, refugees’ human capital
is less likely to be transferable to the host country compared
to other immigrants’ human capital (Brell et al., 2020;Kogan
and Kalter, 2020), which poses incentives to invest time in host-
country-specific human capital accumulation after migration to
the host country. This can be explained with a rational choice
approach to immigrants’ human capital investment (Duleep and
Regets, 1999), which expects investment in education rather
than employment based on the transferability of skills and
credentials acquired in the country of origin. An individual will
invest in host-country education if opportunity costs are lower
than the expected returns of the educational degree. This is
the case if origin country-acquired human capital is not fully
transferable to the host-country labor market. Furthermore,
foreign education certificates, even if recognized in the host
country, cannot fully close the gap to host country educational
certificates in terms of employment probability (Damelang
et al., 2020). This means that immigrants whose skills and
educational credentials are less transferable to the host country
Frontiers in Human Dynamics 03 frontiersin.org
Kuhlemann 10.3389/fhumd.2022.1037778
are more inclined to invest in education compared to those
whose credentials are fully transferable (Duleep and Regets,
1999). In addition, refugees expect to stay in the host countries
longer since they have no short-term perspective of returning to
their home countries (Kogan and Kalter, 2020), which further
amplifies the expected returns of host-country specific human
capital (Cortes, 2004). In combination with the young age of
recently arrived refugees (in Germany, more than 60% of arrived
refugees are between 20 and 35 years old; Brücker et al., 2020),
these factors make the investment of time in education and
training in the host country more likely for refugees compared
to other immigrants (Expectation 2).
In contrast to all groups of first-generation immigrants,
including refugees, individuals of the native population of
the host country are brought up in the host country. This
means they usually lack experience of migration, which
includes issues surrounding transferability of human capital,
visa applications and granting procedures, and host-country
language acquisition. Hence, natives usually do not have to deal
with any forms of post-migration stress and other migration-
related issues, making them more equipped to allocate their
time with their own agency and oriented toward a different
form of utility compared to immigrants and refugees. Hence, I
expect that refugees’ time use differs from the time use of natives
(Expectation 3).
Previous research examining the overall time use patterns of
refugees is rare. A qualitative study by Brekke (2010) examining
refugees in Sweden used information on their time allocation
during a regular workday and found that refugees experience
long periods each day without being engaged in a specific
activity. These periods include staring out of windows or looking
at themselves in the mirror (Brekke, 2010). Similarly, Dupont
et al. (2005), who examined refugees in the Netherlands, found
that they invest time in drug use in order to deal with their
trauma and the long waiting periods in the asylum process.
Quantitative research on refugees’ and other immigrants’
time use focused on isolated activities. For instance, research
on the labor market behavior of immigrants in various host
countries has shown that during the first 5 years after migration,
immigrant men allocate more time to paid labor and immigrant
women more time to household work compared to natives
(Vargas and Chavez, 2010;Ribar, 2012). Over time, their patterns
became more similar to the native population (Vargas and
Chavez, 2010;Ribar, 2012). Concerning refugees in particular,
research showed that they are less likely to work compared to
other immigrants (Cortes, 2004;Brell et al., 2020) and natives
(Bratsberg et al., 2017;Bevelander and Luik, 2020) immediately
after their arrival in the host country. But Cortes (2004) found
that refugees make larger gains in working hours and income
compared to economic immigrants in the 10 years since their
arrival. Eventually, refugees surpass economic immigrants in
terms of income and working hours. In the same 10-year
interval, refugees also have higher rates of human capital
accumulation in comparison to economic immigrants (Cortes,
2004).
Institutional background and legal
barriers for refugees in Germany
As mentioned above, the initial circumstances that refugees
face when arriving a new host country, mainly related to
the asylum regulations, are an important factor distinguishing
refugees from other immigrants. Upon arriving in the host
country, refugees find themselves in a situation in which they
have rather low control over their allocation of time due to legal
restrictions. In the case of Germany, asylum seekers are directed
to a reception facility, where they register and apply for asylum.
They are assigned to an accommodation which they cannot
leave without official permission (Residenzpflicht) and where
they have to live during their first 6–18 months in Germany
or until they receive an answer about their asylum application
(BAMF, 2021). Asylum seekers can receive a positive decision on
their asylum application in form of asylum, refugee protection,
or subsidiary protection, if their application is not rejected due
to their lacking entitlement to asylum (Kosyakova and Brenzel,
2020;BAMF, 2021). Each of these forms of protection come with
a temporary residence permit and several rights. Among them
are the opportunity to participate in integration and language
courses as well as the permission to work in Germany (BAMF,
2021). This means that refugees gain more autonomy in the
allocation of their time once they receive a positive decision
on their asylum application. Refugees whose asylum application
is rejected can receive a temporary suspension of deportation
(Duldung). Similar to refugees who are still in the asylum
process, those with a temporary suspension of deportation can
enter the German labor market only after having received a
permission to work from the authorities (BAMF, 2021).
Previous research on the effects of refugees’ legal status
on time use-related aspects has found that refugees who have
a temporary residence permit are less likely to be employed
compared to refugees who hold the citizenship of the host
country, irrespective of their time spent in the host country
(Bakker et al., 2014). Similarly, a lower likelihood to be employed
in comparison to refugees with a citizenship status has been
found for refugees with a permanent residence permit (Bakker
et al., 2014). For Germany, Kosyakova and Brenzel (2020) found
that refugees who have received a decision on their asylum
application are faster in taking up employment and enrolling in
language courses compared to those who have not yet received
a decision. These findings suggest that receiving a decision on
the asylum application fosters changes in time use in terms of
investing time in educational and labor market activities.
When looking at refugees’ legal status, it is also important to
consider the country of origin. Asylum seekers from countries
with good prospects to remain in the host country, meaning
Frontiers in Human Dynamics 04 frontiersin.org
Kuhlemann 10.3389/fhumd.2022.1037778
countries with more than 50% of asylum applications being
accepted (BAMF, 2022b), can access language and integration
courses even before receiving a positive decision about their
asylum application (BAMF, 2017;Kosyakova and Brenzel,
2020). In addition, they only have a residence obligation
(Residenzpflicht) within the first 3 months since their arrival
in Germany (BAMF, 2021), which can be longer for refugees
from save origin countries. The obligation to keep living in the
federal state where they applied for asylum (Wohnsitzauflage)
remains for all refugees, even after receiving legal protection
status (BAMF, 2022a). Hence, for refugees with good prospects
to remain, the decision on the asylum application should have
less impact on their time use, since they have more autonomy
even before the decision compared to refugees from other
origin countries. Between the end of 2015 and the end of 2019,
countries of origin with good prospects to remain were Eritrea,
Iraq, Iran, Somalia, and Syria (Kosyakova and Brenzel, 2020).
Consequently, I expect to find that refugees’ time use
changes when they receive legal status (Expectation 4) and
that this change in time use is less likely for refugees from
countries with good prospects to remain in Germany compared
to refugees from origin countries with lower prospects to remain
in Germany (Expectation 5).
Data and analytic strategy
The focus of this paper lies on describing the time use
of refugees in contrast to other immigrants and the native
population in Germany, as well as assessing the role of refugees’
legal status in their allocation of time. For this analysis, I rely
on data from the German Socio-Economic Panel (Goebel et al.,
2019;SOEP v36, 2021). The SOEP is a household panel starting
in 1984 that randomly samples households in Germany and
surveys all members of these households annually. The data
also include a sample of refugees in form of the IAB-BAMF-
SOEP Survey of Refugees (Kühne et al., 2019;IAB-BAMF-SOEP,
2021), which includes data from 2016 to 2019 of refugees who
arrived in Germany between 2013 and 2016. The survey is
conducted jointly by the Institute for Employment Research
(IAB), the research data center of the German Federal Office
for Migration and Refugees (BAMF), and the German Socio-
Economic Panel (SOEP) at the German Institute for Economic
Research (DIW). For information on other immigrants, I rely
on the IAB-SOEP Migration Sample (Brücker et al., 2014;IAB-
SOEP, 2021), which is a cooperation project between IAB and
SOEP. Observations were pooled over waves from 2016 to 2019
and respondents between the age of 18 and 40 were considered,
representing the young working age population in Germany
with a more similar age structure to the refugee group (75%
of refugee respondents were included in this age group). First-
generation immigrants whose first interview was more than 6
years after immigration were excluded to make sure that refugees
and other immigrants were observed in similar stages of living
in Germany.
The data measured time use in a stylized way: every
respondent indicated the number of hours that they allocated
to a set of activities on a typical weekday. I used information
for all activities that are available in the data for natives,
immigrants, and refugees. These are employment, running
errands, childcare, care and support for other persons,
education/training, repairs and gardening, physical leisure
activities and sports, leisure activities, and housework. I excluded
all observations in which respondents did not receive one of
the time use (sub-) questions, refused to answer, or gave an
improbable answer for at least one of the activities. Hence, I
coded time use in nine activities as the number of hours allocated
to this activity, ranging from 0 to 24 allocated hours.
To construct the variable of interest, time use, I used k-
means clustering to identify clusters of similar time allocation,
whilst taking into account all nine activities as well as the
number of hours allocated to the activities. The time use
variables were standardized across all respondents used for the
cluster analysis to ensure the equal influence of all activities
for cluster assignment (Hastie et al., 2009, chap. 14.3.3). The
k-means algorithm thereby randomly chose a number k of
starting cluster centers, determined the closest cluster center
for each observation, and assigned each observation to the
cluster with its closest center (Hastie et al., 2009, chap. 14.3.6).
To determine the closeness, Euclidean distance was used. In
the next step, the initial cluster center in each cluster was
replaced by the new cluster mean. This process was repeated
until “within each cluster the average dissimilarity of the
observations from the cluster mean, as defined by the points
in that cluster, is minimized” (Hastie et al., 2009, p. 509). The
resulting clusters depended on the starting values of the k
centers (Makles, 2012); hence I estimated 20 rounds of clustering
with different seeds for selection of the k starting centers
with ks 1–15. Population groups were analyzed separately in
samples of 16,022 native German, 12,316 refugee, and 2,013
immigrant observations.
The resulting clusters of time use were (1) analyzed
descriptively and compared among the population groups.
Further, (2) I assessed the role of refugees’ asylum application
status in a multinomial conditional fixed effects regression
analysis. Here, I made use of the longitudinal structure of the
data, looking at refugees only. The event in focus was receiving a
positive decision on the asylum application. All waves after this
positive decision were coded 1, while all waves before were coded
0. In addition, I estimated models with observations pooled over
waves with multinomial logistic regressions, where legal status
was coded as seven-category variable (1 =asylum seeker status,
2=no permit, 3 =refugee or asylum status, 4 =subsidiary
status or other humanitarian permit, 5 =Duldung (temporary
suspension of deportation), 6 =permanent residence permit, 7
=other residence permit).
Frontiers in Human Dynamics 05 frontiersin.org
Kuhlemann 10.3389/fhumd.2022.1037778
Analyses (2) took into account institutional factors
(the time since arriving in Germany, country of origin,
housing characteristics) and individual characteristics (age,
marital status, number of persons and children in household,
educational attainment at immigration, physical health, life
satisfaction). The time since arriving in Germany was measured
in years. Country of origin was grouped into nine categories [1
=Western countries (including Western Europe and Northern
America as well as New Zealand and Australia), 2 =Eastern
European countries, 3 =Middle Eastern countries (without
Syria and Iraq), 4 =Asia (without Afghanistan; combined
Central, East, South, and South East Asia mainly due to low
numbers of observations for these regions), 5 =Latin America,
6=Africa, 7 =Syria, 8 =Afghanistan, 9 =Iraq (these
countries have sufficiently high numbers to be examined in
separate categories)]. Housing characteristics were included in
a dummy variable, which was 0 for living in a shared housing
environment, such as refugee shelters or reception facilities,
and 1 for living in private housing. Age was measured in years
(centered), and marital status was measured dichotomously (0
not married, 1 married). I also included the number of persons
and children living in the respondent’s household. For education
at immigration, I used an ordinal variable of education attained
abroad with four categories (1 =compulsory school, no degree,
2=compulsory school degree, 3 =secondary school degree,
4=other school degree). Physical health was measured as
subjective health reported on a 5-point scale from 1 “bad” to 5
“very good,” and life satisfaction as proxy for mental health was
measured on a 11-point scale from 0 “completely dissatisfied”
to 10 “completely satisfied.” The models also controlled for
survey year. The analytical sample for analysis (2) included
9,780 refugee observations for the pooled models. The fixed
effects analyses of refugees’ legal status included between 1,358
and 4,435 observations, depending on the changes occurring in
the dependent variable between waves.
Table 1 shows some of the descriptive statistics of the
analytical sample of refugees in contrast to the samples of natives
and other immigrants. The full descriptive tables can be found in
the Supplementary Tables 1–3. The samples cut off observations
below 18 and above 40 years, which was implemented in order
to make the native and other immigrant samples more similar to
the refugees in terms of their characteristics. This was successful
in terms of age: the samples were rather balanced, all having a
mean age of respondents around 30. Other imbalances between
the samples remained. The refugee sample included more than
60% male observations, while the native and other immigrant
samples were slightly female dominated. Refugees also tended
to have more children in the household (almost two children in
comparison to only one or less in the other samples). Among
the refugee observations, the share of those only having primary
education was more than 25 percentage points higher compared
to the other samples. Refugees’ health status on the other
hand was very high (almost half of the refugee observations
reported very good health). Concerning labor force status, the
refugee sample included almost 22% of refugee observations in
employment compared to 70% for other immigrants and 75%
for natives. This is also reflected in the 19% of the refugee
sample who reported at least 1 h spent in employment during
a regular day. Of this share, the average hours reported to be
spent in employment was 6.5 h. Further descriptive information
on the allocation of hours to the different activities can be found
in the Supplementary Tables 4–6. Refugees mostly came from
origin countries in the Middle East (more than half were from
Syria), while most other immigrants reported their origin to
be in Western countries. Concerning their legal status, most
observations in the refugee sample had asylum or refugee status
(55.09%), with 19% of observations reporting asylum seeker
status and around 14% reporting subsidiary protection status
or another form of humanitarian protection. Over the four
survey points from 2016 to 2019, the share of observations
with asylum seeker residence permit declined, while the share
of observations with asylum or refugee status and especially of
those with subsidiary or other humanitarian status increased
drastically. In 2019, <10% of the refugee sample had an asylum
seeker status and over 80% had some form of protection status.
Results
The optimal number of clusters
In order to decide on the optimal number of clusters and
to assess the stability of this number across different starting
points of k, I estimated the proportional reduction of error
coefficient (PRE) for each round of clustering. This measure
estimates the reduction in the within cluster sum of squares
of each round proportional to the previous round (Makles,
2012). Figure 1 shows the optimal number of clusters for the
different rounds of clustering by population group, suggested
by the PRE. The size of the circles represents the frequency
with which the numbers occurred as optimum. Figure 1A on
the left shows the one best solution for each clustering round
and displays a concentration around the lower numbers of
optimal clusters. Given that the PRE of the one best solution
per round of clustering is often very similar to the second best
and further solutions, I rely on the results from Figure 1B on the
right. This graph shows the four best solutions for the number
of clusters for each round. Here, we can see more spread in
the suggested optimal number of clusters, but still the highest
concentration among the numbers between two and five. Hence,
the cluster analysis was conducted for ks of two, three, four,
and five clusters by population group, using the same seed per
k for each population group. Additionally, in order to look at
the stability of the resulting clusters across different seeds, the
cluster analysis was conducted twice for each number of clusters
and population group.
Frontiers in Human Dynamics 06 frontiersin.org
Kuhlemann 10.3389/fhumd.2022.1037778
TABLE 1 Descriptive statistics.
Population group Refugees Natives Other immigrants
Variable NMean/Percentage NMean/Percentage NMean/Percentage
Gender 9,780 15,314 1,841
Female 3,769 38.54% 7,960 51.98% 1,042 56.60%
Male 6,011 61.46% 7,354 48.02% 799 43.40%
Number of children in HH 9,780 1.67 15,314 0.71 1,841 1.11
Number of persons in HH 9,780 3.68 15,314 2.98 1,841 3.10
Education 9,780 15,314 1,841
Primary education 3,257 33.30% 1,058 6.91% 68 3.69%
Lower secondary education 2,454 25.09% 2,121 13.85% 204 11.08%
Upper secondary education 1,924 19.67% 6,298 41.13% 439 23.85%
Post-secondary non-tertiary education 263 2.69% 1,706 11.14% 348 18.90%
Tertiary undergraduate education 1,804 18.45% 2,401 15.68% 485 26.34%
Tertiary post-graduate education 78 0.80% 1,730 11.30% 297 16.13%
Physical health 9,780 15,314 1,841
Bad 166 1.70% 170 1.11% 21 1.14%
Poor 541 5.53% 1,144 7.47% 133 7.22%
Satisfactory 1,178 12.04% 3,406 22.24% 288 15.64%
Good 3,302 33.76% 7,693 50.24% 928 50.41%
Very good 4,593 46.96% 2,901 18.94% 471 25.58%
Life satisfaction 9,780 7.28 15,314 7.58 1,841 7.86
Legal status 9,780 – – – –
Asylum seeker status 1,874 19.16%
No permit 34 0.35%
Refugee +asylum status 5,511 56.35%
Subsidiary protection +other humanitarian status 1,407 14.39%
Duldung 499 5.10%
Permanent residence permit 88 0.90%
Other residence permit 367 3.75%
Country of origin 9,780 – – 1,841
Western countries – – 1,286 69.85%
Eastern Europe 460 4.70% 222 12.06%
Middle East (without Syria and Iraq) 460 4.70% 76 4.13%
Asia (without Afghanistan) 265 2.71% 114 6.19%
Latin America – – 47 2.55%
Africa 1,090 11.15% 90 4.89%
Syria 5,406 55.28% 4 0.22%
Afghanistan 926 9.47% 0 0%
Iraq 1,173 11.99% 2 0.11%
Age 9,780 28.96 15,314 29.33 1,841 32.41
Labor force status: Employed 2,122 21.70% 11,532 75.30% 1,292 70.18%
Includes observations with no missing on any of the displayed variables per population group.
Time use patterns
I descriptively examined the resulting clusters in terms of
the average investment of hours in each activity by cluster.
Based on these descriptive statistics, the clusters were labeled
according to their main activity, i.e., the activity to which the
most hours were allocated. If a cluster was characterized by two
or more activities with similar amounts of invested time (<2 h
difference), this is reflected in the cluster name by the prefix
mixed. I also distinguished between full-time, when the main
Frontiers in Human Dynamics 07 frontiersin.org
Kuhlemann 10.3389/fhumd.2022.1037778
FIGURE 1
Optimal number of clusters by population group. Circle size represents frequency, with larger circles representing higher frequency. (A) Best
solution. (B) Four best solutions.
activity was alloc ated at least 7 h, and part-time, when the main
activity was allocated between 4 and 7 h. Clusters were labeled
with low activity if the overall hours reported by the respondents
in the cluster was on average <10 per day. Tables 2–4show
these descriptive statistics and labels for both seeds of the five-
cluster solutions for refugees, natives, and immigrants, in order
to exemplify how the labeling was done.
In order to compare the time use patterns of refugees, other
immigrants, and natives, the labeling was done for all performed
cluster analyses according to the rules above. Figures 2–4show
the resulting cluster labels by k and the used seed, sorted by
cluster size (Figure 2 for refugees, Figure 3 for natives, Figure 4
for other immigrants). For refugees, across different numbers
of k clusters, I consistently found the largest cluster to be
characterized by low activity, meaning <10 h overall reported by
the respondents in this cluster. Secondly, I consistently found
a cluster characterized by time investment in childcare and
household activities. Other time use clusters were only found
for larger ks and were either characterized by investment in
education or also in childcare and care activities.
When comparing the patterns of time use clusters, one can
see differences between the patterns of natives and immigrants
in comparison to the refugees. Natives and other immigrants
were consistently grouped into a large cluster of respondents
investing time in employment. Only after this largest cluster
did natives and other immigrants display similarities with the
time use of refugees. For the group of other immigrants, the
second largest cluster was characterized by time investment
in childcare and household, similarly as for refugees. I also
found some mixed clusters for other immigrants when looking
at larger ks, meaning that in these clusters, there was no
clear main activity. Such clusters are also found for refugees.
For natives, on the other hand, the cluster analyses found
no consistent second largest cluster. Instead, the clusters were
overall very similar to each other in the sense that there
was always a rather large time investment in employment,
see e.g., the clusters full-time employment and other activities
or mixed part-time leisure and part-time employment. Like
refugees, however, natives displayed a cluster characterized
by investment in education, which was only found once
in all cluster solutions for other immigrants. A childcare
cluster for natives was only displayed for k larger than two,
but it was consistently among the smallest clusters. Overall,
when looking at the clusters that the k-means clustering
algorithm found for different numbers of k, the clusters were
rather consistent across different ks and across different seeds
within the population groups. However, the clusters were
very different across the population groups, meaning that
refugees, other immigrants, and natives seemed to invest their
time differently.
Frontiers in Human Dynamics 08 frontiersin.org
Kuhlemann 10.3389/fhumd.2022.1037778
TABLE 2 Mean hours invested in activities by cluster (refugees).
Refugees Solution seed 1
Cluster
1 2 3 4 5
Mean of hours allocated to
Employment 2.30 1.22 0.10 0.31 0.46
Errands 1.01 1.66 1.57 1.51 0.99
Childcare 1.62 1.63 8.02 7.07 0.95
Care 0.05 0.09 0.08 5.63 0.04
Education 0.10 0.44 0.27 0.48 6.10
Repairs 0.13 0.46 0.08 0.24 0.13
Leisure Activities 0.88 1.50 0.68 0.72 1.10
Sports 0.37 1.81 0.26 0.45 0.82
Housework 1.24 1.68 4.33 3.60 1.27
N4,938 2,096 2,987 298 1,997
Mean number of reported
hours
7.70 10.48 15.40 20.00 11.85
Mean number of reported
activities
3.50 4.69 3.80 4.92 4.49
Cluster name Mixed, low activity Mixed FT Childcare and
Household
Mixed FT
Childcare and PT
Care
PT Education
Solution seed 2
1 2 3 4 5
Mean of hours allocated to
Employment 0.11 1.48 1.56 2.24 0.32
Errands 1.61 1.14 1.03 1.40 1.50
Childcare 8.05 0.98 1.74 2.28 7.02
Care 0.08 0.03 0.05 0.14 5.59
Education 0.29 1.86 1.23 1.08 0.52
Repairs 0.06 0.02 0.00 1.25 0.22
Leisure Activities 0.68 1.25 0.87 1.06 0.71
Sports 0.31 1.45 0.01 0.79 0.44
Housework 4.35 1.38 1.33 1.33 3.63
N2,956 4,109 3,470 1,478 303
Mean number of reported
hours
15.53 9.59 7.82 11.57 19.93
Mean number of reported
activities
3.81 4.32 3.05 5.32 4.90
Cluster name FT Childcare and
Household
Mixed, low activity
(education and
leisure)
Mixed, low activity
(childcare)
Mixed Mixed FT
Childcare and PT
Care
Since these results still show some similarities across
the population groups, I examined the distribution of the
respondents across the clusters in an additional step. Figure 5
shows these distributions in percent, with clusters again being
sorted by size. The percentages of respondents in similar clusters
were similar across populations groups. Natives’ and other
immigrants’ largest cluster of full-time employment was similar
in relative size for the same k. Refugees’ and other immigrants’
Frontiers in Human Dynamics 09 frontiersin.org
Kuhlemann 10.3389/fhumd.2022.1037778
TABLE 3 Mean hours invested in activities by cluster (natives).
Natives Solution seed 1
Cluster
12345
Mean of hours allocated to
Employment 7.15 8.59 4.81 1.96 0.86
Errands 1.14 0.79 0.88 1.35 0.65
Childcare 2.19 0.97 0.38 11.66 0.17
Care 0.07 0.03 0.04 0.25 0.02
Education 0.34 0.29 0.70 0.12 7.37
Repairs 2.53 0.25 0.20 0.35 0.14
Leisure Activities 1.58 1.24 5.22 1.11 1.98
Sports 0.74 0.70 0.75 0.44 0.99
Housework 1.49 1.01 1.13 2.79 0.91
N660 9,370 1,683 1,917 2,392
Mean number of reported hours 17.23 13.88 14.09 20.03 13.08
Mean number of reported activities 5.36 4.58 4.21 4.81 4.37
Cluster name FT Employment
and other
activities
FT Employment Mixed PT Leisure
and PT
Employment
FT Childcare FT Education
Solution seed 2
12345
Mean of hours allocated to
Employment 4.59 0.82 1.70 7.99 8.59
Errands 0.87 0.65 1.36 0.96 0.76
Childcare 0.39 0.17 12.37 1.56 0.88
Care 0.04 0.02 0.26 0.05 0.03
Education 0.73 7.40 0.11 0.30 0.30
Repairs 0.15 0.13 0.34 1.29 0.00
Leisure activities 5.42 1.99 1.12 1.35 1.26
Sports 0.73 0.98 0.43 0.75 0.69
Housework 1.11 0.91 2.86 1.25 0.98
N1,519 2,369 1,740 3,297 7,097
Mean number of reported hours 14.02 13.08 20.54 15.50 13.49
Mean number of reported activities 4.10 4.36 4.73 5.62 4.21
Cluster name Mixed PT Leisure
and PT
Employment
FT Education FT Childcare FT Employment
and other
activities
FT Employment
full-time childcare and household and full-time childcare clusters
only had up to a 4-percentage point difference in size for the
same k. For refugees’ and natives’ part-time education cluster,
this difference was even smaller with one percentage point. One
difference between the distributions was very striking: natives’
childcare-related clusters (full-time childcare) were only about
half the size of the childcare-related clusters of refugees and
other immigrants (full-time childcare and household, full-time
childcare). Overall, it is important to note that the analyses
showed some similarities between the time use of refugees, other
immigrants, and natives, but the three groups never displayed
the same clusters for the same k. Hence, these results are in
line with Expectations 1 and 3, that refugees’ time use differs
from the time use of other immigrants and natives. In addition, I
rather consistently find a cluster of time investment in education
for larger ks for refugees, but only once for other immigrants.
Frontiers in Human Dynamics 10 frontiersin.org
Kuhlemann 10.3389/fhumd.2022.1037778
TABLE 4 Mean hours invested in activities by cluster (other immigrants).
Other immigrants Solution seed 1
Cluster
12345
Mean of hours allocated
to
Employment 1.03 8.83 0.81 8.56 5.12
Errands 0.96 0.66 1.18 0.79 8.32
Childcare 2.19 1.56 10.61 1.02 0.79
Care 0.10 0.02 0.05 0.03 1.53
Education 1.99 0.17 0.15 0.35 0.97
Repairs 0.37 0.18 0.10 0.23 1.62
Leisure Activities 1.91 0.93 0.89 1.13 0.06
Sports 0.49 0.00 0.27 1.19 0.82
Housework 1.82 1.02 3.42 1.04 1.00
N383 660 408 528 34
Mean of number of
reported hours
10.85 13.38 17.49 14.34 20.24
Mean of number of
reported activities
4.58 3.73 4.14 5.06 5.62
Cluster name Mixed FT
Employment
FT Childcare
and
Household
FT
Employment
and other
activities
FT Errands
and
Employment
Solution seed 2
12345
Mean of hours allocated
to
Employment 1.00 0.97 5.12 8.72 6.73
Errands 0.87 1.16 8.32 0.57 0.98
Childcare 1.57 8.99 0.79 1.49 1.03
Care 0.12 0.05 1.53 0.04 0.03
Education 6.18 0.12 0.97 0.20 0.33
Repairs 0.49 0.12 1.62 0.13 0.38
Leisure activities 1.26 1.05 0.06 0.67 1.94
Sports 0.72 0.25 0.82 0.20 1.00
Housework 1.48 3.21 1.00 0.95 1.23
N118 538 34 769 554
Mean of number of
reported hours
13.70 15.91 20.24 12.97 13.63
Mean of number of
reported activities
4.91 4.22 5.62 3.74 5.15
Cluster name PT Education FT Childcare
and
Household
FT Errands
and
Employment
FT
Employment
PT
Employment
Frontiers in Human Dynamics 11 frontiersin.org
Kuhlemann 10.3389/fhumd.2022.1037778
FIGURE 2
Time use clusters by size (refugees).
Furthermore, this education cluster of other immigrants is much
smaller than the ones found for refugees. This is in line with
Expectation 2, stating that refugees are more likely to invest time
in education in the host country than other immigrants.
Heterogeneity in time use
patterns—Descriptive analyses
As for the heterogeneity in time use patterns within
population groups, several interesting findings emerged. An
examination of how immigrants’ migration motivation was
associated with cluster membership showed that economic
immigrants were significantly more likely to sort into the
employment clusters compared to family immigrants. This
was consistently around 6–11 percentage points for all k
cluster solutions. However, family immigrants were not always
significantly more likely to sort into the childcare-oriented
clusters compared to economic immigrants. For natives, a
gender analysis showed that women were significantly more
likely to sort into the full-time employment cluster than men,
if the clustering algorithm also detected a cluster of full-time
employment and other activities, which was much more likely
to be sorted into by men (more than 15 percentage points
more than women). For both analyses, results can be found in
Supplementary Tables 7, 8.
FIGURE 3
Time use clusters by size (natives).
FIGURE 4
Time use clusters by size (other immigrants).
Refugees, in comparison to the other two samples had
on average about 0.5 to one child more in the household.
Furthermore, 64% of the refugee sample reported at least
Frontiers in Human Dynamics 12 frontiersin.org
Kuhlemann 10.3389/fhumd.2022.1037778
FIGURE 5
Distribution of observations across the clusters by
population group.
one child in the household (43% for natives), which might
have driven the large childcare-oriented clusters. Indeed,
looking at the percentage of refugee observations without
children in the household by cluster, showed that in the
clusters involving childcare, shares of observations without
children were small (between 0.31 and 4.06%). Accordingly,
in the refugee clusters that do not involve childcare, the
shares of observations without children were higher. For
instance, in the part-time education cluster the share of
observations without children in the household was around
24% (see Supplementary Table 9). However, the consistently
largest cluster of low activity, which comprised between 40
and 70% of the refugee sample overall, included between 70
and 97% of observations without children in the household
for most ks. This suggests that separate analyses for refugees
without children in the household would likely yield very
small childcare and care clusters, but still very large low
activity clusters.
Refugees’ cluster membership by country of origin showed
that refugees from African, Asian, and Eastern European
countries were represented with the highest shares in the low
activity clusters amongst all refugee groups. The childcare-
oriented clusters included high shares of refugees from Eastern
European countries. Lastly, the education-oriented cluster:
Refugees from the Middle Eastern countries had the highest
share in this cluster (21% of this group belonged to the
education-oriented cluster). The shares of Syrian, Afghan, Iraqi,
and African refugees in this cluster, however, were also rather
high with around 16% (see Supplementary Table 10).
Heterogeneity in refugees’ time use
patterns—Refugees’ legal status
The role of refugees’ legal status for their time use patterns
was examined in two ways. First, I used pooled multinomial
logistic regression models for each cluster solution. Interestingly,
for all numbers of clusters, results indicated that respondents
with refugee or asylum status, or subsidiary protection or
other humanitarian status, were rather consistently significantly
more likely to sort into the mixed and low activity clusters
compared to the full-time childcare and household cluster. Effect
sizes were around 2–4 percentage points. Having a permanent
residence permit showed a significant effect only in some
models: respondents with a permanent residence permit were
significantly more likely to sort into the education- and care-
related clusters compared to the low activity clusters. Here,
effect sizes were even larger with around 10 percentage points.
However, in most models these effects were not significant.
As example, Table 5 shows the effects for the three cluster
solutions. The solutions for the other clusters can be found in
the Supplementary Tables 12–15.
Second, I used the longitudinal structure of the data with
multinomial logistic conditional fixed effects models, which
estimated the effect of receiving a positive decision on the
asylum application on sorting into the different clusters. These
models found only very small and insignificant changes in the
probability to sort into different clusters when the respondents
received their positive decision (see Table 6). Hence, refugees
do not seem to drastically change their time allocation to the
measured activities once they receive their legal asylum status,
at least in the short term (not in line with Expectation 4).
Additional sensitivity analyses showed similar results when
only looking at refugees from countries of origin with good
prospects to remain in Germany (between 2015 and 2019,
countries of origin with good prospects to remain were Eritrea,
Iraq, Iran, Somalia, and Syria). On the other hand, for fixed-
effects analyses on the group of refugees without good prospects
to remain, case numbers were too low. For these refugees,
receiving legal asylum should result in the most changes
within their lives compared to refugees with good prospects
to remain, who have some privileges before receiving a legal
status such as access to integration courses. In order to address
potential differences between refugees with and without good
prospects to remain in Germany, additional analyses on the
Frontiers in Human Dynamics 13 frontiersin.org
Kuhlemann 10.3389/fhumd.2022.1037778
TABLE 5 Pooled analyses: the role of refugees’ legal status.
Three clusters Solution seed 1 Solution seed 2
AME in p.p. Mixed FT
Childcare
and PT
Care
Mixed, low
activity
FT
Childcare
and HH
Mixed, low
activity
FT
Childcare
and HH
PT
Education
Reference: asylum
seeker status
Refugee +asylum
status
0.35 2.28 –2.64 2.42 –2.66 0.24
Subsidiary
protection +other
humanitarian status
0.50 3.14 –3.63 3.30 –2.34 –0.96
Duldung 0.66 1.90 –2.56 3.47 –2.44 –1.03
Permanent
residence permit
8.53 4.60 –13.10 –3.78 –6.33 10.10
Other residence
permit
0.34 0.21 –0.55 0.32 –0.46 0.15
This table shows average marginal effects (AME) in percentage points. Effects are marked bold if the coefficients in the models the AME are based on are significant on at least 5%-level
with respect to the reference cluster (Mixed, low activity). Models control for: gender, educational attainment at immigration, years since migrating to Germany, country of origin, survey
year, marital status, number of children and persons living in respondent’s household, physical health, life satisfaction, housing situation, age (centered).
pooled refugee sample were conducted. These analyses looked
at the effect of being from an origin country with good
prospects to remain on sorting into the different time use
clusters. In Table 7, the results of this analysis show overall
rather small and insignificant effects. This shows that refugees
from origin countries with and without good prospects to
remain in Germany largely do not differ in terms of their
time use, irrespective of their legal status (not in line with
Expectation 5).
Discussion
The time use of refugees has rarely been quantitatively
studied by looking at more than one activity. The present
study looks at refugees’ time use in a more holistic way
and explores the differences in time use of refugees, other
immigrants, and natives in Germany. The use of time is an active
investment in integration for refugees and other immigrants
and hence, a difference in time use between these two groups
might contribute to understanding differences in integration
outcomes. Indeed, the present study found differences in
the time use patterns of refugees, other immigrants, and
natives. Members of the majority German population and
other immigrants seemed to be more likely to invest time in
employment, whereas refugees formed large clusters of either
childcare or low activity. However, there were clusters of
refugees’ time use that were also found for the other two groups.
Immigrants also displayed large clusters of time investment in
childcare-oriented activities, whereas natives displayed clusters
of time investment in education, just like refugees. Overall, even
though there were some similar time use clusters, the exact same
time use clusters for same k and seed for all three population
groups were never found. This is in line with Expectations 1
and 3 stating that the time use of refugees differs from the time
use of other immigrants and natives in Germany. In addition,
in line with Expectation 2, refugees’ time use patterns showed a
group that invests time in education, which was not found for
other immigrants.
This difference of refugees’ time use from the time use of
other immigrants and natives was mostly driven by refugees’
low time investment in employment. Even though almost
20% of the refugee sample reported a time investment of at
least 1 h in employment activities, this did not lead to the
formation of an employment cluster. One reason for this is
that the other activities were much more important for the
cluster formation among the refugee sample since they were
more consistently invested in by all respondents in the sample,
meaning that the share of respondents who invested time in
these activities was higher. Employment was among the three
activities with the lowest share of observations investing time in
this activity for refugees. In contrast for natives, employment
was the activity with the third highest share of observations
who invest time in this activity, with only 6 percentage points
Frontiers in Human Dynamics 14 frontiersin.org
Kuhlemann 10.3389/fhumd.2022.1037778
TABLE 6 Fixed eects analyses: the role of refugees’ legal status.
AME in p.p. 2 clusters 3 clusters 4 clusters 5 clusters
Solution Seed 1 Seed 2 Seed 1 Seed 2 Seed 1 Seed 2 Seed 1 Seed 2
Positive decision on
asylum case, reference:
no decision, yet
Mixed, low activity 1.16 1.12 0.35 –0.87 –1.03 –1.03 –3.40 –1.64
Mixed, low activity
(education and leisure)
–0.99
Mixed 0.66 –0.38
FT Childcare and HH –1.16 –1.12 –1.39 –1.44 –0.98 –0.98 1.92 1.72
Mixed FT Childcare and
PT Care
1.04 1.92 1.29
Mixed PT Childcare and
PT Care
1.32 1.32
PT Education 2.31 0.69 0.69 –1.11
N1,360 1,358 1,530 3,032 3,152 3,152 4,435 4,365
Table shows average marginal effects (AME) in percentage points. Effects are marked bold if the coefficients in the models the AME are based on are significant on at least 5%-level with
respect to the reference cluster (Mixed, low activity). Models control for: marital status, number of children and persons living in respondent’s household, physical health, life satisfaction,
housing situation, age (centered).
TABLE 7 Pooled analyses: the role of good prospects to remain.
AME in p.p. 2 clusters 3 clusters 4 clusters 5 clusters
Solution Seed 1 Seed 2 Seed 1 Seed 2 Seed 1 Seed 2 Seed 1 Seed 2
Origin country with
good prospects to
remain, reference: safe
origin country
Mixed, low activity 1.09 1.15 1.06 –1.10 –1.08 –1.08 –1.53 –0.24
Mixed, low activity
(education and leisure)
–1.34
Mixed 0.25 2.43
FT Childcare and HH –1.09 –1.15 –1.20 –0.91 –1.47 –1.47 –1.04 –1.18
Mixed FT Childcare and
PT Care
0.13 0.31 0.34
Mixed PT Childcare and
PT Care
0.36 0.36
PT Education 2.01 2.18 2.18 2.00
Table shows average marginal effects (AME) in percentage points. Effects are marked bold if the coefficients in the models the AME are based on are significant on at least 5%-level with
respect to the reference cluster (Mixed, low activity). Models control for: gender, educational attainment at immigration, years since migrating to Germany, country of origin, survey year,
marital status, number of children and persons living in respondent’s household, physical health, life satisfaction, housing situation, age (centered).
difference to the activity with the highest share. This is similar
for the immigrant sample (detailed information can be found in
Supplementary Tables 4–6).
The second factor that drove clustering in the refugee sample
is the high share of those who reported only few hours in
the surveyed activities overall. In the refugee sample, 43% of
observations reported <10 h of activities, while this share was
much lower for natives (7%) and for other immigrants (10%).
Since the analysis is limited to the information on time spent
in the measured activities, it neglects potential other activities.
This is especially problematic if refugees were to allocate more of
their time to such unmeasured activities (rather than being really
inactive) compared to the other population groups, because
it makes comparing the groups more difficult. For instance,
Frontiers in Human Dynamics 15 frontiersin.org
Kuhlemann 10.3389/fhumd.2022.1037778
previous qualitative research suggests that some refugees spent
time volunteering and helping other refugees (Kallio et al., 2021;
Lubit, 2022). Activities such as this are not captured in the data.
Another activity not captured in the data but likely important
for refugees is job search effort. A short descriptive analysis
of the shares of refugee observations out of labor force and
registered as unemployed by cluster, showed that indeed the low
activity and the mixed clusters consistently show higher shares
of observations that are registered as unemployed (proxying
looking for a job) than of observations out of labor force
(proxying not employed but also not looking for work). For the
other clusters, this difference in shares was either the opposite or
there was no difference in shares (see Supplementary Table 11).
This suggests that at least some of the observations sorted into
the low activity clusters because actively looking for work was
not captured in the data. Similarly, the measurement of the
education-related activities might result in differences between
refugees and the other population groups. Since more than 70%
of the refugee observations in the sample have been enrolled
in some form of integration or language course, more could
have reported to invest time in education than the 23% who
did report it. This might be due to the refugee respondents
not counting their participation in the language classes as
educational activity, which would additionally increase the
sorting of refugee observations into the low activity clusters.
However, even if both the additional measuring of other
activities and the use of a different method to look at time
use patterns would lead to refugees’ time use clusters looking
different than in the present study, the clusters of refugees’ time
use would still likely not be similar to the time use clusters
of natives and other immigrants. Instead, they would likely
form additional clusters of job searching activities, volunteering
activities, or similar activities, which in turn would not be found
in the same size for natives and other immigrants. Hence, the
difference in time use patterns would prevail.
This difference in time use patterns between refugees
and other immigrants should have important implications
for their integration outcomes. The deliberate and strategic
investment of time in certain activities that foster integration
will affect integration outcomes and, in addition, the time use
patterns of an immigrant or refugee becoming more similar
to the patterns of the native population is an indicator of
integration itself. For instance, if we deem time investment
in employment to foster integration, then refugees would lag
behind other immigrants at least in the first years since their
arrival in the host country. Similarly, given that refugees’ time
use clusters are more different from the natives’ compared
to the other immigrants’ clusters, the similarity of time use
patterns as indicator of integration would also suggest that
refugees lag behind other immigrants in terms of integration.
As discussed, multiple differences between refugees and other
immigrants, such as migration motivation and experience,
make such differences in time use patterns likely. Especially
the conditions in the host country that come with many
legal barriers for refugees should play a role for time
use patterns.
Indeed, my analyses showed that having asylum or refugee
status as well as subsidiary protection or other humanitarian
status was associated with a lower probability to sort into the
childcare- and household-oriented cluster in comparison to the
low activity cluster. On the other hand, the fixed effects analyses
looking at the effect of receiving a positive decision on the
refugees’ asylum cases for their time use found insignificant and
small effects (no support for Expectation 4). Since receiving
legal asylum lifts many of the institutional restrictions which
refugees face in the host society, their time allocation after
reception should be much more free than before, meaning
that they potentially have more similar agency in their time
allocation than natives and other immigrants. However, the
results suggest that it might be difficult for refugees to change
their time use and their situation immediately after receiving
their legal status. Kosyakova and Brenzel (2020) reported that
refugees in 2017 in Germany had waited on average 6 months
for a positive decision on their asylum application. This means
6 months of living without a legal status and the restrictions
that come with this situation. In addition, these refugees took up
their first employment in Germany on average 20 months after
applying for asylum (Kosyakova and Brenzel, 2020), suggesting
that refugees need time to adjust to the host country, find a job,
and learn the German language, even after they receive their
legal status. Hence, the period of 4 years that I examined in this
paper, of which the first observed year had to be without legal
status in the fixed effects models, might be too short to observe
meaningful changes in the time use of refugees. In addition, over
50% of the refugee sample were from Syria meaning they had
good prospects to remain in Germany, which comes with fewer
restrictions even before gaining legal status. With Expectation 5,
I argued that refugees from such countries with good prospects
to remain should be less likely to show a change in time use
when receiving legal status since there were fewer restrictions
lifted with it. However, my analyses showed that refugees from
countries with and without good prospects to remain largely
showed no difference in time use, and hence, Expectation 5 was
not supported.
The study’s drawbacks mostly concern the data. They
measure time use in a stylized way, yet it would be preferable
to have time use diaries in which respondents chronologically
report the hours spent doing certain activities throughout the
course of a day. Such data can be used to study time use
sequences, which would be useful to get a holistic view of
the time allocation per day. Hence, the use of unordered time
information is only an approximation of what a usual day in
a refugee’s life looks like. In addition, the lack of measurement
of other activities, which are potentially more relevant for the
group of refugees, prevents the analysis from more accurately
and holistically describing the time use of all three population
groups. Further research is therefore needed to more deeply
investigate the differences in the time use of refugees, other
Frontiers in Human Dynamics 16 frontiersin.org
Kuhlemann 10.3389/fhumd.2022.1037778
immigrants, and natives, as well as its predictors such as legal
status or individual characteristics. Furthermore, it is necessary
to explicitly study the implications of differences in time use
for integration outcomes and time use patterns as indicator
of integration. This would further the understanding of the
factors holding back refugee integration and the development
of measures to eliminate such factors in order to facilitate the
integration of refugees.
Data availability statement
Publicly available datasets were analyzed in this study. This
data can be found here: SOEP Research Data Center: https://
www.diw.de/en/diw_01.c.601584.en/data_access.html.
Ethics statement
Ethical review and approval was not required for the
study on human participants in accordance with the
local legislation and institutional requirements. Written
informed consent for participation was not required for this
study in accordance with the national legislation and the
institutional requirements.
Author contributions
JK developed the concept and design of the study, performed
all data preparation and statistical analyses, wrote all drafts
of the manuscript, and revised and approved the submitted
manuscript version.
Funding
This work was supported by the University of Mannheim’s
Graduate School of Economic and Social Sciences, and the
European Research Council (ERC) under the European Union’s
Horizon 2020 research and innovation programme (Grant
agreement No. 864683). The publication of this article was
funded by the University of Mannheim.
Conflict of interest
The author declares that the research was conducted in the
absence of any commercial or financial relationships that could
be construed as a potential conflict of interest.
Publisher’s note
All claims expressed in this article are solely those of the
authors and do not necessarily represent those of their affiliated
organizations, or those of the publisher, the editors and the
reviewers. Any product that may be evaluated in this article, or
claim that may be made by its manufacturer, is not guaranteed
or endorsed by the publisher.
Supplementary material
The Supplementary Material for this article can be found
online at: https://www.frontiersin.org/articles/10.3389/fhumd.
2022.1037778/full#supplementary-material
References
Ager, A., and Strang, A. (2008). Understanding integration: a conceptual
framework. J. Refug. Stud. 21, 166–191. doi: 10.1093/jrs/fen016
Aksoy, C. G., and Poutvaara, P. (2021). Refugees’ and irregular
migrants’ self-selection into Europe. J. Dev. Econ. 152, 102681.
doi: 10.1016/j.jdeveco.2021.102681
Alencar, A., Kondova, K., and Ribbens, W. (2019). The smartphone as a lifeline:
an exploration of refugees’ use of mobile communication technologies during their
flight. Media Cult. Soc. 41, 828–844. doi: 10.1177/0163443718813486
Allsopp, J., Sigona, N., and Phillimore, J. (2014). Poverty Among Refugees and
Asylum Seekers in the UK. Working Paper 1/2014. IRIS Working Paper Series.
Birmingham: Intsitute of Research into Superdiversity, University of Birmingham.
Ambrosetti, E., Dietrich, H., Kosyakova, Y., and Patzina, A. (2021). The impact
of pre- and postarrival mechanisms on self-rated health and life satisfaction among
refugees in Germany. Front. Sociol. 6, 1–17. doi: 10.3389/fsoc.2021.693518
Bakker, L., Dagevos, J., and Engbersen, G. (2014). The importance of resources
and security in the socio-economic integration of refugees. A study on the impact
of length of stay in asylum accommodation and residence status on socio-economic
integration for the four largest refugee groups in the Netherlands. J. Int. Migr.
Integr. 15, 431–448. doi: 10.1007/s12134-013-0296-2
BAMF (2017). Das Bundesamt in Zahlen 2016. Asyl, Migration Und Integration
[The Federal Office in Numbers 2016. Asylum, Migration, and Integration].
Report. Nuremberg: Bundesamt für Migration und Flüchtlinge [Federal Office for
Migration and Refugees].
BAMF (2021). The Stages of the German Asylum Procedure.An Overview
of the Individual Procedural Steps and the Legal Basis. Nuremberg:
Bundesamt für Migration und Flüchtlinge [Federal Office for Migration
and Refugees].
BAMF (2022a). Steht Die Wohnsitzauflage Einer Arbeitsaufnahme Entgegen?
[Does the Residence Obligation ContradictTaking up Employment?]. Bundesamt Für
Migration Und Flüchtlinge [Federal Office for Migration and Refugees]. Available
online at: https://www.bamf.de/SharedDocs/FAQ/DE/ZugangArbeitFluechtlinge/
011-wohnsitzauflage- arbeitsaufnahme.html?nn=282388 (accessed October
25, 2022).
BAMF (2022b). Was heißt gute Bleibeperspektive? [What does good prospects
to remain mean?]. Bundesamt für Migration und Flüchtlinge [Federal Office
for Migration and Refugees]. (2022). https://www.bamf.de/SharedDocs/FAQ/DE/
IntegrationskurseAsylbewerber/001-bleibeperspektive.html?nn=282388 (accessed
April 14, 2022).
Becker, G. S. (1965). A theory of the allocation of time. Econ. J. 75, 493–517.
doi: 10.2307/2228949
Bevelander, P., and Luik, M. A. (2020). Refugee employment integration
heterogeneity in sweden: evidence from a cohort analysis. Front. Sociol. 5, 44.
doi: 10.3389/fsoc.2020.00044
Frontiers in Human Dynamics 17 frontiersin.org
Kuhlemann 10.3389/fhumd.2022.1037778
Borjas, G. J. (1987). Self-selection and the earnings of immigrants. Am. Econ. Rev.
77 (4): 531–553. doi: 10.3386/w2248
Bratsberg, B., Raaum, O., and Roed, K. (2017). Immigrant Labor Market
Integration across Admission Classes. Working Paper 10513. IZA Discussion
Papers. Bonn: IZA—Institute of Labor Economics. doi: 10.2139/ssrn.
2911451
Brekke, J.-P. (2010). Life on hold: the impact of time on young asylum seekers
waiting for a decision. Discourse. J. Childhood Adolesc. Res. 5, 159–167. Available
online at: https://nbn-resolving.org/urn:nbn:de:0168- ssoar-354603
Brell, C., Dustmann, C., and Preston, I. (2020). The labor market integration
of refugee migrants in high-income countries. J. Econ. Perspect. 34, 94–121.
doi: 10.1257/jep.34.1.94
Brücker, H., Croisier, J., Kosyakova, Y., Kröger, H., Pietrantuono, G., Rother, N.,
et al. (2019). Second Wave of the IAB-BAMF-SOEP Survey: Language Skills and
Employment Rate of Refugees Improving with Time. IAB-Kurzbericht. Nuremberg:
Institute for Employment Research (IAB).
Brücker, H., Kosyakova, Y., and Vallizadeh, E. (2020). Has there been a
“refugee crisis”? New insights on the recent refugee arrivals in Germany and
their integration prospects. Soziale Welt 71, 24–53. doi: 10.5771/0038-6073-2020-
1-2-24
Brücker, H., Kroh, M., Bartsch, S., Goebel, J., Kühne, S., Liebau, E., et al. (2014).
The New IAB-SOEP Migration Sample: An Introduction into the Methodology and
the Contents. 216. SOEP Survey Papers, Series C. Berlin/Nuremberg: German
Institute for Economic Research DIW.
Cheung, S. Y., and Phillimore, J. (2014). Refugees, social capital,
and labour market integration in the UK. Sociology 48, 518–536.
doi: 10.1177/0038038513491467
Chiswick, B. R. (1999). Are immigrants favorably self-selected? Am. Econ. Rev.
89, 181–85. doi: 10.1257/aer.89.2.181
Cortes, K. E. (2004). Are refugees different from economic immigrants?
Some empirical evidence on the heterogeneity of immigrant groups in the
United States. Rev. Econ. Stat. 86, 465–480. doi: 10.1162/0034653043230
31058
Damelang, A., Ebensperger, S., and Stumpf, F. (2020). Foreign credential
recognition and immigrants’ chances of being hired for skilled jobs—evidence
from a survey experiment among employers. Social Forces 99, 648–671.
doi: 10.1093/sf/soz154
Duleep, H. O., and Regets, M. C. (1999). Immigrants and human-capital
investment. Am. Econ. Rev. 89, 186–191. doi: 10.1257/aer.89.2.186
Dupont, H. J., Kaplan, C. D., Verbraeck, H. T., Braam, R. V., and
Van de Wijngaart, G. F. (2005). Killing time: drug and alcohol problems
among asylum seekers in the Netherlands. Int. J. Drug Policy 16, 27–36.
doi: 10.1016/j.drugpo.2004.06.002
Echterhoff, G., Hellmann, J. H., Back, M. D., Kärtner, J., Morina, N.,
and Hertel, G. (2020). Psychological antecedents of refugee integration
(PARI). Perspect. Psychol. Sci. 15, 856–879. doi: 10.1177/17456916198
98838
Gillespie, M., Osseiran, S., and Cheesman, M. (2018). Syrian refugees and the
digital passage to europe: smartphone infrastructures and affordances. Social Med.
Soc. 4, 1–12. doi: 10.1177/2056305118764440
Goebel, J., Grabka, M. M., Liebig, S., Kroh, M., Richter, D., Schröder, C., et al.
(2019). The German socio-economic panel (SOEP). J. Econ. Stat. 239, 345–360.
doi: 10.1515/jbnst-2018-0022
Guichard, L. (2020). Self-selection of asylum seekers: evidence from Germany.
Demography 57, 1089–1116. doi: 10.1007/s13524-020-00873-9
Hastie, T., Tibshirani, R., and Friedman, J. (2009). The Elements of Statistical
Learning: Data Mining, Inference, and Prediction. 2nd Edition. Springer Series in
Statistics. New York: New York: Springer.
IAB-BAMF-SOEP (2021). IAB-BAMF-SOEP Survey of Refugees 2019. Data
2016–2019. IAB-BAMF-SOEP.
IAB-SOEP (2021). IAB-SOEP Migration Samples (M1, M2), Data of the Years
2013–2019. IAB-SOEP.
Jaschke, P., and Kosyakova, Y. (2021). Does facilitated and early access
to the healthcare system improve refugees’ health outcomes? Evidence
from a natural experiment in Germany. Int. Migr. Rev. 2021, 1–31.
doi: 10.1177/0197918320980413
Kallio, K. P., Meier, I., and Häkli, J. (2021). Radical hope in asylum seeking:
political agency beyond linear temporality. J. Ethn. Migr. Stud. 47, 4006–4022.
doi: 10.1080/1369183X.2020.1764344
Kogan, I., and Kalter, F. (2020). An empirical-analytical approach to
the study of recent refugee migrants in Germany. Soziale Welt 71, 3–23.
doi: 10.5771/0038-6073-2020-1-2-3
Kosyakova, Y., and Brenzel, H. (2020). The role of length of asylum procedure
and legal status in the labour market integration of refugees in Germany. Soziale
Welt 71, 123–159. doi: 10.5771/0038-6073-2020-1-2-123
Kosyakova, Y., and Brücker, H. (2020). Seeking asylum in Germany: do human
and social capital determine the outcome of asylum procedures? Eur. Sociol. Rev.
36, 663–683. doi: 10.1093/esr/jcaa013
Kühne, S., Jacobsen, J., and Kroh, M. (2019). Sampling in Times of High
Immigration: The Survey Process of the IAB-BAMF-SOEP Survey of Refugees.
Available online at: https://surveyinsights.org/?p=11416 (accessed November
19, 2022).
Li, S. S., Liddell, B. J., and Nickerson, A. (2016). The relationship between post-
migration stress and psychological disorders in refugees and asylum seekers. Curr.
Psychiatry Rep. 18, 1–9. doi: 10.1007/s11920-016-0723-0
Lindert, J., von Ehrenstein, O. S., Priebe, S., Mielck, A., and Brähler, E. (2009).
Depression and anxiety in labor migrants and refugees: a systematic review and
meta-analysis. Soc. Sci. Med. 69, 246–257. doi: 10.1016/j.socscimed.2009.04.032
Lubit, A. J. (2022). A mother’s experience of asylum: conflicting temporalities,
belongng, and evolving care relations. J. Intercult. Stud. 43, 464–479.
doi: 10.1080/07256868.2022.2086226
Makles, A. (2012). Stata tip 110: how to get the optimal k-means cluster solution.
Stata J. 12, 347–351. doi: 10.1177/1536867X1201200213
Mansouri, F., and Cauchi, S. (2007). A psychological perspective on Australia’s
asylum policies. Int. Migr. 45, 123–150. doi: 10.1111/j.1468-2435.2007.00398.x
OECD (2017). Finding Their Way: The Labour Market Integration of Refugees in
Germany. OECD, International Migration Division, Directorate for Employment,
Labour and Social Affairs.
OECD/EU (2018). Settling in 2018: Indicators of Immigrant Integration. Paris,
Brussels: OECD Publishing and European Union. doi: 10.1787/9789264307216-en
Ribar, D. C. (2012). Immigrants’ Time Use: A Survey of Methods and
Evidence. Working Paper 6931. IZA Discussion Papers. Bonn: IZA—Institute of
Labor Economics.
Seidle, F. L., and Joppke, C. (2012). “Introduction,” in Immigrant Integration in
Federal Countries, ed. C. Joppke and F. Leslie Seidle. Montreal, Canada: Montreal,
Canada: McGill-Queen’s University Press, 3–22.
SOEP v36 (2021). Socio-Economic Panel (SOEP), Version 36, Data from Years
1984-2019 (SOEP Core V36,EU-Edition). SOEP v36.
Vargas, A. J., and Chavez, M. (2010). Assimilation Effects Beyond the Labor
Market: Time Allocations of Mexican Immigrants to the US. Lubbock: Texas
Tech University.
Zerubavel, E. (1985). Hidden Rhythms: Schedules and Calendars in Social Life.
First California paperback ed. Berkeley: Berkeley/Los Angeles/London: University
of California Press.
Frontiers in Human Dynamics 18 frontiersin.org
