Beruflich Dokumente
Kultur Dokumente
Masters thesis
for the Masters degree programme
Quantitative Finance
in the Faculty of Business, Economics and Social Sciences
at the Christian-Albrechts-Universitat zu Kiel
submitted by
Acknowledgment
I am extremely grateful to Almighty God, for His grace and guidance
from the beginning to the completion of this thesis. I would like to express my very great appreciation to my supervisor Prof. Dr. Uwe Jensen
for his valuable and constructive suggestions during the planning and
development of this research work. His willingness to give his time so
generously has been very much appreciated. I would like to also thank
my friend Victoria Sam Abaidoo for her insightful inputs and useful suggestions.
I am also ever grateful to my parents and siblings, for their
prayers, encouragement, support and love in challenging moments.
Abstract
This paper explores how the share of immigrants in Germany affect the
life satisfaction of German residents. In particular, immigrants are also
categorized into five different immigrant subgroups- European Economic
Area (EEA), Turkey, Other Europeans, Asia and the rest of the world.
The number of immigrants from these regions are also independently explored to know their effect on the satisfaction levels of German residents.
Three models are used for the analysis, namely the Ordered probit model,
Ordaniry Least Squares (OLS) regression and the Fixed effect model.
The results indicate a positive significant effect of the total immigration
share on satisfaction when the Ordered probit and OLS modelS are used,
but a negative non-significant effect when the Fixed effect model is used.
The Ordered probit model is therefore used as our benchmark model
with which the various immigrant subgroup effects are also analysed.
Immigrants from EEA, Turkey and the Other European countries had a
significant positive effect on the satisfaction of German residents, while
those from Asia had a negative non-significant effect and those from the
rest of the world also had a positive but non-significant effect on the
satisfaction of German residents. The results therefore indicate that in
general German residents are more happy when more immigrants come
to Germany, especially immigrants from Europe; therefore there should
not be much concern in Germany with regards to extending the EEA to
include other European countries.
ii
Contents
Acknowledgment . . . . . . . . . . . . . . . . . . . . . . . . .
Abstract . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
ii
List of Acronyms . . . . . . . . . . . . . . . . . . . . . . . . .
1 Introduction
3 Literature Review
10
19
4.1
Data Sources . . . . . . . . . . . . . . . . . . . . . . . .
19
4.2
. . . . . . . .
21
4.3
Econometric specifications . . . . . . . . . . . . . . . . .
26
4.3.1
26
4.3.2
31
4.3.3
The Model . . . . . . . . . . . . . . . . . . . . . .
32
iii
5 ESTIMATION RESULTS
35
5.1
Main Results . . . . . . . . . . . . . . . . . . . . . . . .
35
5.2
51
5.3
52
6 Conclusion
55
Appendix
Bibliography
Affirmation
VIII
. . . . . . . . . . . . . . . . . . . . . . . . . . XVI
iv
List of Acronyms
OECD
Totimmshare
Totimmsh hat
immshareEEt
immEurOth t
YrsofEdu
Age2
Age squared
Empdum
NotinLabFor
NWinEduTra
MatLeave
NWUnem
EasGer
logNumPrsHH
WrkHrs
WrkHrs2
logHhInc
logNumDocVis
logUnempExp
ThreeChild
loggdp
unemprate
vi
Chapter 1
Introduction
The impact of immigration on the welfare of host countries has been a
topic of grave concern for both policy makers and economists over the
years now; with a large number of people migrating to foreign countries
for one reason or the other. The United Nations (UN) and the Organisation for Economic Co-operation and Development (OECD) in a joint
report stated that the number of international immigrants increased from
154.2 million in 1990 to 231.5 million people in 2013, which is about a 50
% increase (UNDESA, 2013). This has led to quite a number of studies
and academic literature on immigration, addressing a broad variety of
topics.
Generally, researchers and policy makers would want an answer
to the question of whether immigrants have a positive or negative impact on the welfare of the residents in the host country. In answering
this question, economists and researchers have over the years traditionally employed the use of objective measures of welfare such as wages
and employment ( Card (1990), Card (1997), Dustmann et al. (2005),
Butcher and Card (1991), Borjas (1994), Borjas (2003), Ottaviano and
Chapter 2
Evolution of Immigration
Policy in Germany
In this chapter, I will succinctly review the historical evolution of immigration policy in Germany and address the essential findings from the
studies which are closely associated with this policy evolution.
According to an OECD ranking in 2012, Germany was the second largest recipient of migrants in the world, after the United States,
and the number one in Europe. Migration to and from Germany has
a long history. The reasons for such migration have primarily been the
same over the years: to seek greener pastures; flight from ethnic, political, or religious persecution; forced expulsion. The history of German
immigration policies can be mainly dated back to the post World War II
era right through to the Immigration Act in 2005.
The first group of immigrants in Germany are those called the
expellees or the ethnic Germans. These were people who had a German
background but had been living outside of Germany before the World
War II. These group of people settled in Eastern and some parts of Cen4
tral Europe, mainly present day Poland and Czech Republic. They were
expelled as a result of the Nazi German invasion on the Eastern bloc.
According to German Ministry of Interior (2014), between mid 1944 and
the end of 1949, roughly 7.7 million German expellees had been admitted into the Federal Republic of Germany. Between that time and by
the end of 1981, 1.8 million more ethnic Germans and expellees were
admitted and brought this figure to 9.5 million. Between the years 1982
and 2013, another 3.5 million ethnic German re-settlers and their families
from Eastern Europe and the former Soviet Union came into the Federal Republic. This has resulted in a total of 12 million ethnic Germans
admitted into Germany over these years.
The second influx of foreign population comprised of foreign
workers, also known as guest workers, who were admitted into the country from 1955 to 1973. During the 1950s, the West Germanys economic
miracle led to an increasing demand for both semi-skilled and unskilled
labour. The labour supply from the local Germans was not sufficient in
meeting this demand, so the government signed recruitment agreements
with Italy (1955), Spain and Greece (1960), Turkey (1961), Morocco
(1963), Portugal (1964), Tunisia (1965) and Yugoslavia (1968) (Ministry
of Interiors publication on Migration and Integration (2014)) . Also after 1961, East Germanys decision to build the Berlin Wall and close its
borders to the West, cut off the supply of workers from East Germany,
thereby contributing to increased shortage of labour. The number of
Germans in the labour force fell by 2.3 million from 1960 to 1972, which
led to an increment in the recruitment of foreign workers. In 1960, only
1.3% of those in employment were foreigners and by 1973 this number
had risen to 11.9%. Reportedly, most of these foreign workers were employed in the states of North Rhine-Westphalia, Baden-Wuerttemberg ,
Bavaria and Hesse. (German Ministry of Interior, 2014)
the law of soil; meaning children who were born in Germany with at
least one parent being a German citizen, under some circumstances also
became a citizen. The second was the green card regulation, which was
intended to attract more highly skilled foreign professionals to Germany.
The latter was primarily focused on IT-specialists (Constant and Tien
(2011)). However, the program could not be as successful as expected
failing to attract highly skilled migrants, as evidenced by the sharp drop
in visa applications. This caused the government to officially abandon
this scheme in 2005 (Constant et al., 2010b). This is believed to be as
a result of the fact that the German government spent a lot of time
debating whether high-skilled immigrants should be admitted, and how
to ensure that they leave after expiration of their contracts.
After about four years of political controversies and deliberations, as well as a couple of rejections of the immigration bill, the
Federal Government finally passed an Immigration Act which took effect from January 1, 2005. Through this Act, Germanys policy-makers
built the foundation for immigration policy and the social integration
of migrants, thereby finally recognising that Germany is an immigration
country. These reforms were brought about mainly in response to the
problem of high unemployment; mainly due to mismatches of demand
and supply as well as labour shortages for high skilled immigrants (Constant and Tien (2011)). Chaloff and Lemaitre (2009), made an insightful
assertion about the motives behind immigration policy reforms made in
recent years by countries like Germany. They explained that the policy
has been to regulate immigration, while still leaving an allowance for employers to employ high-skilled workers. It is worth noting that despite the
Immigration Act upholding the ban on the recruitment of foreign labour,
especially for unskilled and low-skilled workers, Section 18 Subsection 1
of the Residence Act required that the recruitment of foreign workers
Chapter 3
Literature Review
In this thesis, I explore the effect of the main immigrant subgroups share
on the subjective well-being of German residents. As indicated earlier,
Germany is a typical immigration country and therefore policy makers
are concerned about the impact of international migration on the welfare
of residents, for that matter natives of the country. There has been a
surge in the number of studies examining the use of the broader subjective well-being measures rather than the traditional objective welfare
measurements. It is therefore essential to apply this area of research to
explore the direct impact of the different immigrants subgroups on the
subjective well-being of the resident population.
In this chapter, I will briefly present the findings of existing literature in Germany and the world as whole, that examines the impact of
immigrants. I will also review contemporary economic studies in terms of
happiness and life satisfaction. By so doing, I will compare and contrast
the two types of well-being measures with supporting literature . I then
argue why it is important to measure the impact of immigrants on a host
country, subjectively too, but not just objectively.
10
11
Grubel and Grady (2011) find that recent immigration to Canada has
created a fiscal burden to the countrys economics. Now turning to Germany, Sinn and Werding (2001) concluded that as at 1997, immigration
had a net fiscal burden on Germanys public finances. But they noted
that in the long-term, immigrants who stayed over 25 years produced
a net surplus. Bonin et al. (2000) and Bonin (2001) also argued that
immigration however produces a slight net benefit for the public sector
over the entire lifespans of the immigrants owing to their young average
arrival age and the manner through which the German pension system
is tied to ones earned income.
Recent studies also examine the relationship between immigration and the attitudes of natives. For example, Card (2005) analyse
European Social Survey data and find that even though attitudes towards immigrants are partly forged by economic factors, other aspects
such as culture, and natives social status are essential in affecting perception about immigration. Also, Boeri (2010) argues that the business
cycle affect natives attitudes towards immigrants. Other studies also
examine the determinants of attitudes toward immigrants (Facchini and
Mayda (2009), Mayda (2006), Rustenbach (2010), Senik et al. (2009),
Bauer et al. (2000)).
While welfare and other traditional economic measurements are
essential in exploring the impact of immigration, our understanding can
be deepened using the relatively new method of subjective well-being.
There is a burgeoning consent among governments and international institutions on two points: first, that GDP is a very limited and imperfect measure, and second, that measures of subjective well-being have a
paramount role to play in defining welfare(ODonnell (2013)). ODonnell
(2013) explains how countries are using well-being data to improve policy making, and concludes that this approach leads to better policies and
13
a better policy process, since at the end of the day, happiness is what
matters most to most citizens. He give examples of countries which have
started to measure their progress with regards to the happiness of their
citizens. He indicates that Bhutan is the best known example where the
government has adopted the objective of maximizing its Gross National
Happiness (GNH) Index; but others, like the UK, US, Canada and New
Zealand are also now systematically collecting data on happiness and
life satisfaction. For example, British Prime Minister David Cameron
has set up a system demanding the Office for National Statistics (ONS)
to measure well-being frequently (ODonnell (2013)). ODonnell (2013)
also noted that the OECD is leading the way in establishing fair standards so that cross-country comparisons can be made; and also some of
these measurements use survey evidence to measure how happy people
feel at the moment, while others ask about overall satisfaction with life.
De Neve et al. (2013) also argues that it is essential to balance economic
measures of societal development with subjective well-being measures,to
ensure that economic growth leads to broad development across all life
domains, not just greater economic capacity.
In recent years, the amount of literature that evaluate subjective well-being has increased extensively. Quite a number of studies in
this area aim to investigate the determinants of subjective well-being
(e.g., Dolan et al. (2008), Clark et al. (2008); Deaton (2010); DeVoe and
Pfeffer (2009); Blanchflower and Oswald (2011)). Most of these studies
tend to find out the factors that make citizens of a country happy, and
they do this mostly through correlations and just recently some going
further with test of hypothesis. Most researchers use a happy equation
to measure happiness. Normally, economists tend to use a cardinal version of happiness or life satisfaction as a dependent variable in their
econometric analysis. With respect to the independent variables, many
14
that of immigrants. For example, Bartram (2011) using data from the
World Values Survey, concludes that the relationship between income and
happiness is much weaker for USA natives than for immigrants; however,
he noted that even for immigrants that relationship is still relatively
weak. He also notes that migration is most likely a journey to better
ones economic welfare, but an increase in income does not necessarily
yield better happiness. Gokdemir and Dumludag (2012) also examine the
role of several socio-economic and non-economic factors explaining the
differences of happiness levels of Turkish and Moroccan Immigrants in
the Netherlands; being the two largest non-EU immigrant communities
in the Netherlands. They find that Moroccans, despite having lower
income levels and higher unemployment rates than Turkish immigrants,
their happiness level is higher than the Turkish immigrants. They explain
that this is because there is insignificant effect of absolute income for
Turkish immigrants, however, the effect of relative income, which largely
explains the lower life satisfaction, matters for Turkish immigrants.
Normally it is presumed that people migrate in search of better income, from rural areas towards urban areas, or from developing
countries to well-to-do countries, and therefore it is most probable that
the less happy ones choose to migrate. Therefore, another aspect of the
literature tends to address the change in happiness for immigrants after they have migrated. For example, Nowok et al. (2011) investigate
the question of Does migration make you happy? by exploring if persons who migrate within the UK become happier than they were before
and whether the effect is permanent or temporary. They find that the
internal migrants within the UK on average experience a significant decline in subjective well-being (SWB), in the period just before the time
of migration; however there is a boost associated with migration which
tends to bring people back to their initial level of happiness. Knight and
16
Gunatilaka (2012) also explore the change in life satisfaction for internal migrants who move from rural to urban China, since they constitute
18% of the total population; and they find that generally this internal
migration leads to a drop in happiness.
To focus on the literature for this paper, there is a relatively new
strand of economic study which examines the effect of migration on natives subjective well-being. Akay et al. (2012) is the first research paper
that attempts to address this situation by combining information from
the German Socio-Economic Panel and regional data from INKAR at the
RaumOrdnungsRegionen level; between 1997 to 2007. They investigate
the impact of immigration on the subjective well-being of natives and
immigrants and conclude that immigration positively affects the SWB
of native Germans; after conducting robustness checks and addressing
endogeneity issues. Another example is Betz and Simpson (2013), who
use the European Social Survey to examine the impact of aggregate immigration inflows on the subjective well-being of native-born European
populations in a panel of 26 countries in the time period between 2002
and 2010. Their main conclusion is that, and I quote: recent immigrant
flows have a non-linear, yet overall positive impact on the well-being of
natives. Specifically, we find that immigrant flows from two years prior
have larger positive effects on natives well-being than immigrant inflows
from one year prior. Our findings are very small in magnitude and in
practical application; only large immigrant flows would affect native wellbeing significantly (Betz and Simpson (2013)). In Canada, Ding (2013)
also examines the effect of immigration on both natives and immigrants
subjective well-being but finds that there is generally a negative impact.
Nonetheless, there is no such similar study that examines the direct effect of migration from different immigrant sub-groups on the subjective
well-being of residents. To the best of my knowledge, this thesis is the
17
18
Chapter 4
Data and Methods
4.1
Data Sources
The sources for the empirical analysis used in this paper are mainly
from two distinct sources. I combine a dataset extracted from the German Socio-Economic Panel (GSOEP) and rich regional data at the state
level(Bundeslander) from official statistics of Germany. The GSOEP
has been widely used in the SWB literature (Winkelmann and Winkelmann (1998), Ferrer-i Carbonell and Frijters (2004), Van Praag et al.
(2003)). The GSOEP is a wide-ranging representative longitudinal study
of about 11,000 private households consisting of about 30000 residents;
this includes both natives and immigrants. This annual panel survey
was first executed in the Federal Republic of Germany in 1984, collecting data on German and immigrant households who are reinterviewed
every year. The sample has been enlarged and refreshed over the years,
particularly with the inclusion of about 2,000 East German households
in 1990 and a sample of Eastern European immigrants who migrated to
Germany after the collapse of the Soviet Block. The GSOEP surveys
19
are conducted by the German Institute for Economic Research, and contains data with respect to topics like household composition, occupation,
employment, earnings, health and life satisfaction. I extract a rich set
of socio-economic variables at the individual level; particularly, I retrieve
information to formulate the SWB variable. The SWB variable is derived
from the question How satisfied are you at present with your life as a
whole?, which grants responses on an ordinal scale from 0 to 10, where 0
stands for completely dissatisfied and 10 for completely satisfied. Since
my focus is to examine the impact of the various immigrant subgroups
on the SWB of the entire population, I do not separate the immigrant
sample from the natives, but rather treat them as one- residents of Germany. The definition of immigration employed is based on citizenship,
with respect to the law of blood and the law of soil in Germany.
The second data source is the Federal Office of Statistics (Statistisches Bundesamt), from which I extract statistics for the 16 federal
states of Germany. Since the GSOEP contains information on the federal states of the sampled individuals, it is possible to match the microdata with the regional statistics. The advantages of using the federal
states level data are manifold. First, federal states are well-defined regions which are somehow distinct in their culture, economic policies and
labour market characteristics. This precise geographical level allows for
the adequate capturing of the heterogeneity of German local labour markets. Moreover, my main interest of statistics from immigrant subgroups
is readily available only at the federal states level, even though it would
have been more efficient to get these statistics at the ROR level; but then
also at the ROR level, the immigration subgroups would constitute small
percentages . The key variable of interest, the immigrant share in the
federal states, is defined as the ratio between the number of immigrants
(or immigrant subgroup) and the total resident population (for the sake
20
4.2
As stated earlier, the SWB variable is derived from the question How
satisfied are you at present with your life as a whole?. Originally the
responses are on an ordinal scale from 0 to 10, where 0 stands for completely dissatisfied and 10 for completely satisfied. Table 1 shows the
summary of responses for this question and as we can see, only 8.36% of
residents who rate their life satisfaction use a number less than 5 and
just 3.58% pick a rating of 10. It is evident that too few observations
in many cells, so I decided to recode the responses into fewer groups by
recoding 0-5 to 0, 6-7 to 1, 8 to 2 and 9-10 to 3. This is done to
reduce the variation in the responses in order to get more efficient estimates. The summary of responses for the recoded variable, SWB, is
presented in Table 2. But I must say that using the original variable in
my analysis did not yield significantly different results from the that of
21
Table 2
SWB
Freq. Percent Cum.
0
24,144
19.24 19.24
1
42,341
33.74 52.98
39,114
31.17 84.15
2
3
19,895
15.85 100.00
Total 125,494 100.00
liar immigration policies; for example immigrants from the EEA can just
come to Germany and secure jobs without requiring a visa as well as
easier integration. I singled out Turkey because it is by far the most
populous immigrants country, even representing a greater proportion
than America and Africa combined. The subgroup Others comprises
of mainly immigrants from Africa, America and other undeclared nationalities. And from Table 4, it is evident that the largest number of
immigrants come from EEA constituting about 3.58% of the entire German population; with the lowest number coming from Others at a meagre 0.72% of the population. It can also be seen that the East German
states- Brandenburg, Mecklenburg-Vorpommern, Saxony, Saxony-Anhalt
and Thuringia, have the lowest share of immigrants in comparison with
the West German states.
Table 5 shows all the variables I use in my analysis which were
partly extracted from the GSOEP responses on individual characteristics
such as age, gender, marital status, employment status, household size,
health status (number of doctor visits), Education, number of children,
amount of work hours and household income. I also extracted the region and federal states of individuals, to address regional heterogeneity,
in addition to labour market characteristics with respect to the various
federal states; here I consider only GDP and unemployment rates. But
it must be noted that I decomposed most of these variables and used
them as dummies, leaving out the ones with the highest ratios out. The
abbreviated variables are expressed fully in the list of Acronyms section.
I extracted the individual characteristics variables from the GSOEP
version 29 (long) dataset; specifically, from the pl, pequiv and pgen
datasets. However, I did some re-coding, notably by excluding all individuals who did not have data on life satisfaction, for any given time
period, and also by adding a one to all metric variables before finding
23
Table 3:
24
25
4.3
Econometric specifications
In this section I will give an insight into the econometrics methods I will
use and choose one as my benchmark model. I start by briefly explaining
panel or longitudinal data estimation and then I highlight on the reason
for the choice of models I use. I then present the econometric model and
explain its parameters.
4.3.1
longitudinal data allows researchers to address a number of essential economic questions that cannot be addressed using cross-sectional or timeseries datasets in empirical studies. For instance, panel data helps in
clarifying the often heard assertion that the real reason one finds (or
does not find) certain effects is the presence of omitted (mismeasured
or unobserved) variables that are correlated with the independent variables (Hsiao (2014), chapter 1). Panel data therefore allows to control
for omitted (unobserved or mismeasured) variables. One practical example is that, the least-squares regression coefficients of yit on xit are well
known to be biased; but under panel data estimation, when we take the
first difference of individual observations over time, Least squares regression now provides unbiased and consistent estimates of (Hsiao (2014),
chapter 1).
After talking about the importance of panel data estimation,
I now look at some dynamics with some panel data estimation models.
There is a part of linear panel data models where the error in each time
period is assumed to be uncorrelated with the independent variables in
the same time period. This assumption is too strong for most panel data
applications. As stated earlier, a primary motivation for using panel data
is to solve the omitted variables problem.
With regards to this thesis I study population models that explicitly contain a time-constant, unobserved effect. This is in line with
modern econometrics in the sense that unobserved effects are treated as
random variables, drawn from the population along with the observed
explained and independent variables, as opposed to parameters to be
estimated. In this regard, the paramount issue is whether the unobserved effect is uncorrelated with the independent variables. (Wooldridge
(2002), chapter 10)
28
E(y | x1 , x2 , ..., xk , ) = 0 + x +
29
t = 1, 2, ..., T
If i serves as an index for individuals, then i could be called
an individual effect or individual heterogeneity; it follows analogously for
families, firms, cities, states and other cross-sectional units.
In quite a number of methodological literature, and also in applications, there is often the debate about whether i will be treated as a
random effect or a fixed effect. Traditionally, in longitudinal data models,
i is called a random effect when it is treated as a random variable and
a fixed effect when it is seen as a parameter to be estimated across each
cross section observation i. Considering a large number of random draws
from the cross section, it usually makes sense to treat the unobserved
effects, i , as random draws from the population, along with yit and
xit . This approach is absolutely convenient from an omitted variables
or neglected heterogeneity perspective. However, in modern econometric
argot, random effect is used when there is no correlation between the
observed independent variables and the unobserved effect; and fixed effect does not necessarily mean that i is being treated as non-random
but rather, arbitrary correlation between the unobserved effect i and
the observed independent variables xit is allowed. So, if i is called an
individual fixed effect or a federal state fixed effect, then, for practical
reasons, this terminology means that i is allowed to be correlated with
xit . These lead us to two different estimation methods random effects
estimation and fixed effects estimation. (Wooldridge (2002), chapter 10).
In this thesis, I stick to the modern econometric definitions.
In fixed effects analysis, one can consistently estimate partial
effects in the presence of time-constant omitted variables that can be arbitrarily correlated with the explanatory variables xit . In this sense, the
fixed effects estimation is more robust than random effects estimation.
30
i = +
m
X
l wil + i
l=1
to recover j via l .
Where wi s are the time-invariant exogenous variables including the decomposed dummies (e.g. gender, federal state, etc.). But I must say that
another shortfall with this arises when the time period T is small. This
makes i inconsistent, which implies inconsistent l and j
4.3.2
Now I switch the focus to the ordered probit model which is a suitable
candidate for my analysis due to ordinal nature of my dependent variable.
If y is an ordered response, then the values we assign to each outcome
are no more arbitrary. For example, in this thesis, y is the SWB variable
on a scale from zero to three, with y = 3 representing the highest life
31
satisfaction and y = 0 the lowest rating. The fact that three is a higher
life satisfaction rating than 2 carries useful information, even though the
life satisfaction ratings itself only has ordinal meaning. For example, we
cannot say that the difference between three and one is somehow twice
as important as the difference between one and zero.(Wooldridge (2002),
chapter 15)
Assuming y is an ordered response which takes on the values
0, 1, 2, ..., J for some known integer J. The ordered probit model for y
(conditional on independent variables x) can be derived from a latent
variable model. I further assume that a latent variable y is determined
by:
y = x + ,
| x Normal(0,1)
4.3.3
The Model
(4.1)
it = i + r + it
(4.2)
where,
SW Bit - captures the latent well-being of an individual i at time t.
IMrt - immigrant share in federal state of residence r at time t.
Xit - comprises individual socio-demographic and economic characteristics such as age, marital status and income.
Zrt - includes time-varying labour market characteristics, such as unemployment rate and GDP per capita in each federal state of residence r at
a given time t.
The error term it and its components are represented in equation (4.2): captures individual unobservable heterogeneity as well as
federal-state specific time-invariant attributes; and represents periodspecific effects captured in the regression by time dummies; whereas is
an error term that follows a standard normal distribution due to identification in the ordered probit specifications.
Although the ordered nature of the dependent variable favour
an ordered probit model, for simpler and better interpretation of results, I will run some linear regressions (OLS and Fixed Effects model)
as well. The merits of using a linear specification are that it makes the
interpretation of the parameter estimates easier, and enables controlling
for individual unobservable characteristics in a simpler way (Akay et al.
(2012)). Moreover, Ferrer-i Carbonell and Frijters (2004) demonstrate
that using the cardinal model or the ordinal model in SWB analysis,
33
do not yield any significant difference in the estimation results. Therefore, the ordered probit will be my main model but I will also present
the linear regression models only for comparison purposes. There have
been some studies that have sought to use an ordered probit model or
a linear regression model (Boes and Winkelmann (2010); Ding (2013);
Akay et al. (2012)). Akay et al. (2012Akay et al. (2012) use both a
random-effect ordered probit model and a fixed-effect model to examine
the impact of migration on the life satisfaction of German natives and
immigrants. They find that random effects ordered probit and fixed effects have similar results. In this study, I will use the ordered probit,
OLS and fixed-effect model for the analysis of only the total immigrants
share; for comparison purposes. But with the rest of the results, that is,
from the immigrant subgroups, I will use only the ordered probit model.
There are quite a number of shortcomings in the analysis using
the ordered probit model. First of all, the role of unobserved individual
characteristics, such as personality traits, as well as unobserved federal
state characteristics, is highly essential when analysing subjective wellbeing (Boyce (2010)). If these factors are in anyway correlated with the
explanatory variables, then a specification controlling for time-invariant
individual characteristics is preferred; but that is not the case with the
ordered probit model.(Akay et al. (2012)). Furthermore, there could
also be some multicollinearity between the immigrants share and the
federal state labour characteristics. I therefore use instrumental variable
approach in getting an estimate for the immigrants share before running
the main analysis.
Futhermore, in view of accounting for serial correlation in the
error term, I cluster the standard errors at the individual level.
34
Chapter 5
ESTIMATION RESULTS
5.1
Main Results
35
Table 6:
the various immigrant subgroups like I did for the total in Table 6. I then
present the results of the instrumented immigrant subgroups population
share on the SWB of residents, using the ordered probit model in Tables
11 to 15. I will interpret the results from table to table, and then I
will make comparisons within these interpretations with respect to the
various models.
One thing to note is that I decomposed some variables into
dummy variables, thereby leaving out the reference groups in the analysis.
Reference groups for having children is No children; for marital status is
Married; for employment status is Employed; for gender is Female; for
region is West Germany; for West German federal states is North-Rhine
Westfalia; and for East German federal states is Saxony. For consistency,
the largest groups in the undecomposed variables were chosen as the
reference groups. Another is that I control for federal state characteristics
in all the models by including federal state indicators.
Now, before discussing the main results of interest, it is essential
to discuss how the estimates of my model compare with those from ex-
36
Table 7:
37
isting literature. With respect to the OP model, the full estimates of the
socio-economic characteristics listed in Table 5, are reported in Table 7.
As one can see from a quick preview of this table, my results are consistent with previous literature regarding the subject of SWB in Germany
(Ferrer-i Carbonell (2005), Winkelmann and Winkelmann (1998), Akay
et al. (2012)). For example, the pattern of SWB over a persons life cycle
exhibits the classic U-shaped behaviour, suggesting that well-being is
high when one is young, and then decreases to a lowest level around the
age of 40, and then increases again (though it is not so clear in Figure
1) (see Frey and Stutzer (2002), Dolan et al. (2008)). Being married
has a positive effect on ones life satisfaction. Moreover, an increase
in the educational attainment will lead to a higher level of well-being.
Bad health is negatively correlated with life satisfaction as evident in
the sign of Number of Doctor Visits. Higher income also leads to a
higher life satisfaction as seen in Figure 2. As unsurprisingly entrenched
in the SWB literature, being unemployed is negatively associated with
life satisfaction (Wilson and Walker (1993), Clark and Oswald (1994),
Frey and Stutzer (2002) and Dolan et al. (2008)). However I find an
interesting result in the amount of worked hours; in the estimates there
appears to be a concave relationship between a SWB and worked hours.
This suggests that those with lower working hours are less satisfied with
their lives; and this satisfaction increases and is maximum around the
average hours worked, and then finally decreases to a lower satisfaction
again as the worked hours exceeds this average. The first part does not
make much sense and the reason could be that the lower working hours
are associated with those not working- either not in the labour force or
unemployed. Another reason could be that since this group of people
work less, they therefore earn less, hence less satisfied.
Regarding my key variable of interest, it is evident from Table 7
38
Figure 1:
Figure 2:
39
that a higher immigrant share in the federal state leads to a positive and
significant increase in the SWB of German residents; thus immigration
has a positive impact on the residents SWB. This is in line with Akay
et al. (2012). These results were not significantly different when I used
the original 0-10 satisfaction scale (Table A6)
In Table 8, the coefficients are estimated using the OLS model
and the interpretation is direct and easier. The results are not too different from that of the OP model. Generally the signs of the coefficients
are the same, with the exception of the coefficient for Bremen which
is positive however non-significant for the OP model, but negative and
significant for the OLS model. Also it can be seen that the coefficients of
the OLS model are generally smaller in magnitude than that of the OP
model. For example with regards to my main variable of interest, while
the OP model gave a coefficient of 0.0332 to the total immigrant share,
the OLS model gave a coefficient of 0.0276. With regards to the total
immigrant share, the results presented in the Table 8 imply that there is
0.0276 life satisfaction increase in the 4-scale life satisfaction of German
residents associated with a 1% increase in the immigrant share in each
federal state. Also, in terms of standardized coefficients, it can be seen
that: an increase of one standard deviation in the immigrant share in
each federal state is estimated to increase residents life satisfaction by
0.0945 standard deviation units (Table A7 in Appendix). This seems to
be a rather large effect, considering that the standardized coefficient for
an individual being unemployed is -0.0539 and for household income is
0.0532.
Tables 9 and 10 gives the estimates using the fixed effects
model. Table 9 contains individual and federal state specific characteristics which cannot be decomposed, whereas Table 10 contains th auxiliary
OLS of the residuals (res) from Table 9 on the decomposed variables
40
Table 8:
41
Table 9:
42
Table 10:
43
45
Table 11:
46
Table 12: Impact of Immigrant Share from Turkey on SWB (OP model)
47
Table 13:
48
Table 14:
49
Table 15:
50
Asia and the coefficient of the immigrant share here is different from
that of the total and those from Europe. There is a negative however
insignificant effect of immigrants from Asia on the SWB of German residents. This insignificant effect may be as a result of the relatively smaller
amount of Asian immigrants in the various federal states.
In Table 15, I estimate the impact of immigrants from Other
parts of the world on the SWB of German residents. Generally these
estimates are not far from that of the total immigrant share; with the
coefficient of immigrant share from Other parts of the world being also
positive, however insignificant. Nonetheless in both Tables 14 and 15,
there are changes of sign of some of the coefficients of the federal state
indicators, in comparison with that of the total population. Specifically,
the coefficients of Hamburg and Hessen which were insignificantly negative under the total immigrant share, become significantly positive; and
that of Mecklenburg-Vorpommern changes from insignificantly positive
to significantly negative. This could possibly mean that immigrants from
Asia and Other parts of the world have a higher positive impact on
Hamburg and Hessen than those from EEA, Other European countries
and the total immigrant share. However, these same subgroups in Tables 14 and 15 could possibly have a higher negative impact on the SWB
of residents in Mecklenburg-Vorpommern, as compared with the other
subgroups.
5.2
skilled; using the OP model. I use the Erikson and Goldthorpe Class
Categorie in the GSOEP data and recode those in high service, selfemployed with employees and Skilled manual jobs as High skilled; low
service, routine service-sales, routine non-manual and self-employed with
no employees as Middle skilled jobs. And then finally semi-skilled and
unskilled manual, farm labour and self-employed farm as Low skilled
jobs. From the results of the estimation, the immigrant share has a positive impact across the three skilled groups. And the coefficients are not
far from each other, with just a slight increase in that of the immigrant
share as we move from High skilled to Low skilled. This suggest that
none of the labour force participants should feel threatened by the influx
of immigrants.
5.3
ity in Stata, after running an OLS regression, and the null hypothesis
of homoskedasticity was rejected. And so some of the methods I use in
minimizing this effect of heteroskedasticity is by taking the natural logs
and in some cases the squares of the metric variables. Also in my OP
analysis I employ robust standard errors and also cluster these standard
error at the individual level.
54
Chapter 6
Conclusion
The main objective of this study is to analyse whether the geographic
concentration of various immigrant subgroups in Germany has an impact on the residents subjective well-being. I merge panel data from
Germany with official statistics on immigrants and local labour markets
from the period 2005 to 2012, and then I estimate three regression models in which I correlate residents subjective well-being variable with the
immigrant share in the federal state, as well as a wealth of other control
variables. The main findings of this paper indicate that, in general, an
increasing number of the total immigrant share in each federal state lead
to an increase in the subjective well-being of residents in Germany. In
other words, German residents life satisfaction increases as the number
of immigrants increases. Also I find out that immigrants from the EEA,
Turkey and the rest of the European countries have a significant positive impact on the life satisfaction of Germans, with those from Other
Europeans, which includes EU candidate countries, having the highest
effect. In addition, those from Asia have a negative, however insignificant
effect and the last group which I term as the Other parts of the world
also had a positive but insignificant effect on the SWB of residents. In
55
56
consideration as well.
57
Appendix
Table A1:
Table A2:
Table A3:
OLS Instrumenting of Immigrant Share from Other European Countries
Table A4:
Table A5:
II
Table A6:
III
Table A7: Impact of Total Immigrant Share on SWB (OLS model with standardized coefficients)
IV
Table A8: Impact of Total Immigrant Share on High Skilled (OP model )
Table A9:
VI
Table A10:
VII
Bibliography
Akay, A., Constant, A. F., and Gulietti, C. (2012). The impact of immigration on the well-being of natives, iza dp no. 6630, bonn: Iza.
Online: http://ftp. iza. org/dp6630. pdf (10.09. 2013).
Basten, C. C. and Siegenthaler, M. (2013). Do immigrants take or create
residents jobs? Quasi-experimental evidence from Switzerland KOF
Working Papers.
Barrett, A., Kahanec, M., Zimmermann, K. F., Barrett, A., and Matre,
B. (2013). Immigrant welfare receipt across europe. International
Journal of Manpower, 34(1):823.
Bartel, A P (1989). Where do the new US immigrants live? Journal of
Labor Economics, pages 371391.
Bartram, D. (2011). Economic migration and happiness: Comparing immigrants and natives happiness gains from income. Social Indicators
Research, 103(1):5776.
Bauer, T. K., Zimmermann, K. F., and Lofstrom, M. (2000). Immigration policy, assimilation of immigrants and natives sentiments towards
immigrants: evidence from 12 oecd-countries.
Betz, W. and Simpson, N. B. (2013). The effects of international migration on the well-being of native populations in europe. IZA Journal of
Migration, 2(1):121.
VIII
Analysis, 57(1):121.
Borjas, G. J. (1994). The economics of immigration. Journal of economic
literature, pages 16671717.
Borjas, George J (1999). The economic analysis of immigration Handbook
of labor economics, 3:16971760.
Borjas, G. J. (2003). The labor demand curve is downward sloping:
Reexamining the impact of immigration on the labor market. Technical
report, National Bureau of Economic Research.
Boyce, C. J. (2010). Understanding fixed effects in human well-being.
Journal of Economic Psychology, 31(1):116.
Br
ucker, H., Epstein, G. S., McCormick, B., Saint-Paul, G., Venturini,
A., and Zimmermann, K. F. (2002). Managing migration in the european welfare state. Immigration Policy and the Welfare System, Oxford
University Press, Oxford, pages 1168.
IX
Germany.
115(507):F324F341.
Dustmann, C., Frattini, T., Halls, C., et al. (2010). Assessing the fiscal
costs and benefits of a8 migration to the uk. Fiscal Studies, 31(1):141.
Dustmann, C., Frattini, T., and Preston, I. P. (2013). The effect of
immigration along the distribution of wages. The Review of Economic
Studies, 80(1):145173.
XI
Dustmann, C., Glitz, A., and Frattini, T. (2008). The labour market
impact of immigration. Oxford Review of Economic Policy, 24(3):477
494.
Facchini, G. and Mayda, A. M. (2009). Does the welfare state affect
individual attitudes toward immigrants? evidence across countries.
The review of economics and statistics, 91(2):295314.
Ferrer-i Carbonell, A. (2005). Income and well-being: an empirical analysis of the comparison income effect. Journal of Public Economics,
89(5):9971019.
Ferrer-i Carbonell, A. and Frijters, P. (2004). How important is methodology for the estimates of the determinants of happiness?*. The Economic Journal, 114(497):641659.
Frey, B. S. and Stutzer, A. (2002). What can economists learn from
happiness research? Journal of Economic literature, pages 402435.
Gokdemir, O. and Dumludag, D. (2012). Life satisfaction among turkish
and moroccan immigrants in the netherlands: The role of absolute and
relative income. Social indicators research, 106(3):407417.
Grubel, H. and Grady, P. (2011). Immigration and the canadian welfare
state 2011. H. Grubel & P. Grady, Studies In Immigration and Refugee
Policy, Fraser Institute.
Hatton, T. J. and Tani, M. (2005). Immigration and Inter-Regional Mobility in the UK, 19822000*. The Economic Journal, 115(507):F342
F358.
Hsiao, C. (2014). Analysis of panel data. Cambridge university press, 54.
XII
Hunt, J. (2012). The impact of immigration on the educational attainment of natives. Technical report, National Bureau of Economic Research.
Islam, A. (2007). Immigration unemployment relationship: The evidence
from canada*. Australian Economic Papers, 46(1):5266.
Kahneman, D. and Sugden, R. (2005). Experienced utility as a standard of policy evaluation. Environmental and resource economics,
32(1):161181.
Knight, J. and Gunatilaka, R. (2012). Income, aspirations and the hedonic treadmill in a poor society. Journal of Economic Behavior &
Organization, 82(1):6781.
Manacorda, M., Manning, A., and Wadsworth, J. (2012). The impact
of immigration on the structure of wages: Theory and evidence from
britain. Journal of the European Economic Association, 10(1):120151.
Mayda, A. M. (2006). Who is against immigration? a cross-country
investigation of individual attitudes toward immigrants. The Review
of Economics and Statistics, 88(3):510530.
Ministry of Interior, (2014). Residence law and policy on migration and
integration in germany. German Ministry of Interiors Publication on
Migration and Integration, pages 1046.
Myrskyla, Mikko and Margolis, Rachel (2014). Happiness: Before and
after the kids. Demography, 51(5):18431866.
Nowok, B., Van Ham, M., Findlay, A., and Gayle, V. (2011). Does
migration make you happy? a longitudinal study of internal migration
and subjective well-being (discussion paper). IZA Discussion Paper
6140.
XIII
XIV
XV
Affirmation
I hereby declare that I have composed my Masters thesis Immigration
Effects on Satisfaction independently using only those resources mentioned, and that I have as such identified all passages which I have taken
from publications verbatim or in substance. Neither this thesis, nor any
extract of it, has been previously submitted to an examining authority,
in this or a similar form.
I have ensured that the written version of this thesis is identical
to the version saved on the enclosed storage medium.
Kiel, November 3, 2015
XVI
.....................................