Third Demographic Transition and Demographic Dividend: An Application Based On Panel Data Analysis

Bulletin of Geography. Socio–economic Series / No.
42 (2018): 59–82
BULLETIN OF GEOGRAPHY. SOCIO–ECONOMIC SERIES DE

journal homepages:
G
http://www.bulletinofgeography.umk.pl/
http://wydawnictwoumk.pl/czasopisma/index.php/BGSS/index
http://www.degruyter.com/view/j/bog
ISSN 1732–4254 quarterly
Third demographic transition and demographic dividend:

An application based on panel data analysis
Alejandro López González1, CDFMR, María Jesús González-González2, CFMR
University of León, Departament of Geography and Geology, Campus de Vegazana, s/n 24071 León, Spain; 1phone: +34987293509,
e-mail: [email protected] (corresponding author); 2phone: +34987291051, e-mail: [email protected]
How to cite:
López-González, A. and González-González, Mª J., 2018: Third demographic transition and demographic dividend: An applica-
tion based on panel data analysis. Bulletin of Geography. Socio-economic Series, 42(42), 59-82. DOI: http://doi.org/10.2478/bog-
2018-0031.
Abstract. The third demographic transition, barely mentioned by some authors Article details:
and implicit for others, refers to changes in the demographic structures of the Received: 26 October 2017
most developed countries promoted by the most recent migratory flows, with Revised: 23 March 2018
repercussions in aspects such as age structure or the composition of the labour Accepted: 08 August 2018
market. The concept of the third demographic transition revolves around the in-
creasing presence of foreigners, many of whom take up jobs that nationals reject,
as well as other more skilled posts. Using the panel data methodology, we try to
explain the third demographic dividend whose impact can be seen in the labour Key words:
market. The results enable us to conclude that the foreign worker differential puts ageing,
downward pressure on salaries, which affects other groups. If workers are avail- demographic dividend,
able and policies are constructive, this leads to positive results and social wealth. migration
© 2018 Nicolaus Copernicus University. All rights reserved.
Contents:
1. Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 60
2. Theoretical framework . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 60
2.1 A third transition and a third demographic dividend? The effects of migration and tensions
in the labour market and the welfare state . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 62
3. Stylized facts: inmigration in European countries . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 64
4. Empirical approach . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 68
4.1. Data . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 68
4.2. Methodology . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 70
4.3. Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 72
5. Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 78
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 80
© 2018 Nicolaus Copernicus University. All rights reserved. © 2018 De Gruyter Open (on-line).
60 Alejandro L. González, María J. González-González / Bulletin of Geography. Socio-economic Series / 42 (2018): 59–82
1. Introduction literature portrays them as being parallel phenom-

ena. The first demographic transition generates a
demographic dividend with its own characteristics
The theory of the demographic transition is the (the bonus of economic growth due to a constant
framework sustaining the basis of the debate on increase in the working age population group). The
what is known as the demographic dividend associ- second demographic transition is fostered by qual-
ated with the growth of the potential workforce and, itative improvement in the working age group gen-
more specifically, the reduction in the rate of de- erated by the increasing proportional importance of
pendence. The first demographic dividend would be older workers.
characteristic of countries starting to consider the Our aim is to analyse the correlation between
decline in birth rates within a context of low mor- transitions and demographic dividends with the
tality, associated with an increase in a mostly very support of scientific production. We attempt to deal
young workforce; the second dividend appears in with the relationship between what some authors
countries at the end of the demographic transition, (Coleman, 2006) refer to as the third demographic
with aged populations and a declining workforce transition and a potential third demographic divi-
that is beginning to generate work for older people dend. We understand that the third dividend should
(Lesthaeghe, Karel, 2002). In recent decades, the de- correspond to the driving idea behind this new
velopment of demographic methods and techniques transitional cycle: the main engine is immigration,
and the availability of more statistical information and therefore the triggering force is not population
has made it possible to analyse more accurately the growth but the sustained arrival of immigrants from
impact of changes in structure by age or sex on the developing countries who fill gaps in the labour
labour market, welfare, education or health, that is, market resulting from accumulated years of low
on the economy. It is in this scenario that the key birth rates in the local population (Szreter, 2015).
concept of demographic dividend appears. This is The study comprises three parts. The first is a
the favourable relationship between the potential theoretical approach to the concepts of demograph-
workforce (adults) and the passive or dependent ic transition and demographic dividend; we con-
population (the elderly and young) (Erikson, 1981; sider these terms to be intrinsically related and in
Carstensen, Fried, 2011; Fried, 2015). continuous evolution, since patterns in structural
Thus, the third demographic transition, with re- changes in the population imposed by each tran-
percussions in aspects such as age structure or the sitional model can be applied to different models
composition of the labour market, contemplates of the demographic dividend. The second part is
complementarity and substitution in the labour an empirical analysis of Western European coun-
market (Dribe et al., 2017). The third demograph- tries, considering the third demographic dividend
ic transition revolves around the increasing pres- on the basis of the evolution of distribution be-
ence of immigrants, and its most worrying aspect tween foreign and domestic populations in cohorts
is, for different authors, the immigrants’ impact on of the working-age population during 2006–2014.
the destination labour market (Marozzi, 2016). The third part is an analysis of panel data regres-
In this paper, we used Eurostat statistics for large sion based on two scenarios, and a study of the re-
age groups to estimate the dependence rate. The sulting model.
main aim is to measure the impact of immigrant
workers on the salaries of the population as a whole.
These salaries constitute one of the indicators of the
2. Theoretical framework
dividend attributed to immigrants as we believe that
salaries are one of the underlying causes of current
migratory movements. We also consider economic The concept of the demographic dividend derives
growth, labour, level of education and ageing (No- from the theory of the demographic transition, one
testein, 1953; Pew Research Center, 2015). of the most important theoretical contributions of
Thus, demographic transition and demographic demography, traditionally inclined towards quanti-
dividend are two closely linked concepts. Scientific tative methods (Kirk, 1996).
Alejandro L. González, María J. González-González / Bulletin of Geography. Socio-economic Series / 42 (2018): 59–82 61
The first outlines of the demographic transi- the workforce as a result of periods when fertility
tion can be found in the identification of regular- exceeded mortality (transitional period). This first
ities and the corresponding proposals for country demographic dividend coincides with the conven-
taxonomies, regarding the evolution of population tional theory of the first demographic transition and
size in different countries (Thompson, 1929; Lan- is currently valid, though limited to developing and,
dry, 1934), though the term itself was still not used to a lesser extent, emerging countries, i.e. countries
at the time. F. Notestein (1945) established the ba- still immersed in the demographic transition or at
sis of the transnational theory. the end of it, where higher levels of fertility over
Between 1950 and 1990, theoretical development mortality ensure that the flow of new job appli-
focused on what might be called the first demo- cants continues to rise. Scientific literature focuses
graphic transition in which the transition towards on questions associated with fertility (Bloom et al.,
levels of greater socioeconomic development were 2009; Moultrie et al, 2012) and on differential mor-
accompanied by changes in demographic behaviour tality by age cohorts (Canudas-Romo, Guillot, 2015;
(Coale, 1973, 1984). Different rhythms, followed by Masquelier et al., 2014; Timaeus, 2004). Issues that
the descending evolution of mortality and fertility, concern this current are related to economic devel-
at different rates, resulted in stages of large accu- opment (Bloom, 2003; Bloom. et al., 2013), main-
mulated volumes of demographic excess. As it en- ly in Sub-Saharan Africa and South Asia, where the
tered the labour market, this demographic surplus majority of countries that have not concluded the
brought about a demographic dividend, or at least transitional stage are located. Such issues also in-
what became known as the first dividend, due to its clude cases in which the first dividend is coming to
analogy with the first demographic transition. an end, for example, China. These studies mainly
In the field of migration and mobility a model deal with the precise conditions required for con-
analogous to the demographic transition was pre- verting the spurt of potential economic growth as-
sented. The mobility transition hypothesis posed sociated with the dividend into real growth given
by W. Zelinsky (1971) describes this process in the the institutional weaknesses existing in many devel-
populations of developed countries in four stages oping countries (Bloom et al., 2007 in Sub-Saharan
(the pre-modern traditional society, the early tran- Africa; James, 2008 and Chandrasekhar et al., 2006
sitional society, the late transitional society and the for India; Wongboosin et al., 2005 for Thailand).
advanced society) which synthesize regularities ob- They also deal with the implications that the tran-
served in the spatial mobility of the population fol- sient nature of this dividend has on the sustainabil-
lowing a process parallel to the standard transitional ity of the growth model in some countries, such as
theory. This theory detected regularities for interna- China, facing drastic changes in age structure (Fang
tional migration in agrarian colonisation process- , Wen, 2012; Wang, 2007), and which is linked to
es in border areas, population movement from the the emergence of a second dividend (Cai, 2010).
country to the city and inter- and intraurban mi- In its first stages, the transition theory soon met
gration. Zelinsky ended his proposal with the pro- with new processes that undermined its explanato-
jection of an ulterior phase (a future superadvanced ry power with regard to contemporary populations.
society) in which he established forecasts that pre- The heterogenous nature of the demographic paths
dicted events such as waves of unqualified labour of different populations and complexity of demo-
force migration from developing countries or in- graphic phenomena limit the explanatory value of
creasing control of population mobility. this theoretical framework (Kreager, 2009), opening
The demographic dividend refers to extra eco- up a new approach that refers to an ulterior second
nomic growth strictly attributable to the population; demographic transition in industrialised countries
it cannot therefore be considered the same as pro- since 1960, this time supported by both changes in
ductivity gains associated with technological (inno- basic demographic magnitudes, e.g. fertility, and by
vations) or institutional (changes in modes of social the appearance of new family models (Van der Kaa,
organisation) changes. This added value associated 2002, 2004).
with demography was, at least originally, conceived These changes imply a new perspective of the
as an increasing number of young workers joining demographic dividend, not so much concerned
with the contributions of working age groups as es, will be culturally distant. This mutation would
with older age cohorts. New contributions have have severe consequences on a social and cultural
deepened and expanded the concept of the demo- level; a true shock that is incipiently starting to be
graphic dividend (Mason , Lee, 2006; Mason, 2007). felt especially, but not only, in European countries.
They acknowledge two different stages in the demo- These consequences, whilst interacting with the mi-
graphic transition (first and second transitions) as gratory tradition in each country, affect the dispari-
a result of different levels of socio-economic devel- ty of the migration policies of different states (Bauer
opment in different countries. Logically, this implies et al., 2000). Leaving aside the social repercussions
a split in the concept of the dividend: one demo- of immigration, clearly seen in tensions occurring
graphic dividend corresponding to a higher rate of in some important aspects of the welfare state (Hu-
growth in the working age population (net produc- ber, Oberdabering, 2015), we shall focus on the la-
er) in comparison with the dependent population bour market.
(net consumer); a second dividend associated with The growing presence of immigrants will mul-
the increasingly ageing working population, which tiply friction with the national population in rela-
results in increased production as the strengthened tion to sensitive issues, such as the labour market
group of older workers strive to build assets to have or some components of welfare. This, together with
a more financially comfortable retirement (Lee,Ma- identity and culture shock, will cause tension lead-
son, 2006). As these studies refer to demograph- ing to situations of maximum stress and trigger ep-
ic transitions that have almost ended, developed isodes of maximum violence, as occurred with the
countries and emerging powers experiencing rap- open doors policy in Germany after the 2015 Syri-
id demographic change, such as East Asia or Latin an refugee crisis (Jäckle , König, 2016).
America, are chosen as empirical cases. Relevant is- Scientific evidence seems to sustain the theory of
sues no longer focus on economic development, but a dual national population: when immigrant work-
on developed countries with populations in the ad- ers are less qualified than the national labour force,
vanced stages of the demographic transition. They the latter is in a more favourable position, whereas
concern matters such as intergenerational transfers the opposite occurs when the situation is reversed
(Mason, Lee, 2007, for Taiwan; Phang, Mason, 2007, (Mayda, 2006). Assuming this to be true, a brief ob-
for the Republic of Korea), sustainability of pension servation would be appropriate.
systems (Turra, Queiroz, 2007, for Brazil) and, al- Gang and Rivera-Batiz (1994) estimated com-
though formally addressing the first demograph- plementary and substitution relationships between
ic dividend, they also analyse the repercussions of the resident and immigrant labour force in both
possible demographic ageing (Mejia Guevara et al., the USA and the European Union at the time. The
2010, for Mexico). method applied considered subjective aspects of
the work force (low-skilled employment, educa-
tion, work experience). The impacts of immigra-
2.1. A third transition and a third demo- tion depend on the magnitude of migratory flows
graphic dividend? The effects of migra- and on the relative levels of the mentioned attrib-
tion and tensions in the labour market utes, and also on the degree of complementarity and
and the welfare state substitutability amongst immigrants and nationals.
The conclusions reached by the authors at the time
David Coleman (2006) proposed a thought-provok- pointed to the complexity of the impact of immi-
ing theory: Western Europe and the United States grants on residents’ salaries which were very une-
are undergoing a decisive change in their demo- qual; though small in absolute terms. Thus, when
graphic composition associated with growing im- immigrant labour complements national labour
migration. This transformation is fundamentally (they diverge as far as training or experience are
ethnic and cultural, and in the not too distant future concerned), salaries increase. However, if they are
the native population will be progressively displaced substitutive (the mentioned attributes coincide or
by groups from other countries which, in many cas- are similar), salaries decrease.
With regard to the response of the labour mar- In any case, the impact is weak when analysed at
ket to economic cycles, Dustmann et al. (2010) an- skills level.
alysed the cases of Germany and the UK. They Another particular case is Israel. Cohen-Gold-
concluded that different types of behaviour exist- ner and Paserman (2011) studied the impact of
ed in both cases, as the diminution of employment massive immigration into the country from Russia
amongst immigrants occurs more rapidly during during the 1990s, based on the assumption that im-
times of recession and hiring occurs equally as rap- migrants are good substitutes of the local workforce.
idly during times of recovery. However, the case of The effect may not be noticeable to begin with but
the UK is asymmetrical because in times of eco- it increases in time as they acquire the skills and
nomic growth there are discrepancies in the cre- competence of the local population. The authors
ation of jobs between both groups. More jobs are concluded that the effect of this migratory episode
created for foreigners than nationals. Also, with re- was minimal in terms of employment. However, in
gard to the German labour market, in this case re- terms of salary, a negative correlation was detect-
stricted to the federate states composing former ed as the high level of education of the newcomers
Western Germany, DÁmuri et al. (2010) studied had no effect on the salaries they received in their
the impact of immigration throughout the 1990s, first employment. The imbalance was corrected over
and pointed out that during this period it was not time and the salaries of both groups became simi-
the native workers who were negatively affected by lar as immigrants began to occupy posts that were
the arrival of foreigners, but immigrants that had on a par with their level of education.
settled in the previous decades, due to a large ex- Regarding the difference in salary between im-
tent to the high level of substitution between the migrants and nationals, Islam and Parasnis (2016)
two groups within a context of marked rigidity in studied the case of Australia, distinguishing be-
the labour market. However, Glizt (2012), based on tween the salaries of white- and blue-collar work-
the peculiar treatment by the authorities of the large ers. The authors observed how in the case of highly
influx of immigrants with German ancestors from qualified white-collar posts, the salaries of immi-
western Europe (they were channelled towards cer- grants were even higher than those of nationals; the
tain regions according to family and work criteria), complete opposite occurred in the case of blue-col-
detected that during the 1990s and the start of this lar workers.
century the presence of immigrants had a signifi- The impact of immigration is not only felt in
cant effect on increased unemployment in the local terms of employment but also in different levels of
population, though not on salaries. employment and salaries. Borjas (2006) studied the
One specific case is Spain (Carrasco et al., 2008). impacts of immigration on the labour market in
This country, until recently one of emigration, has the United States and changed the perspective on
become the destination for large volumes of immi- this issue, pointing to the adjustment mechanisms
grants since the second half of the 1990s. In this that the native population adopts to avoid the ad-
particular case, after analysing both legal and illegal verse effects of immigration in this respect. One of
immigration, the authors found no evidence of neg- these adjustments is migratory movements. There
ative impacts on employment rates or on the sala- is evidence that the main areas receiving foreign
ries of nationals. immigrants are characterised by a reduction in the
Competition between immigrants and nationals number of immigrants from other parts of the Unit-
can also be seen in the impact on employment lev- ed States and by natives emigrating to other parts
els amongst the latter, attributable to immigration. of the country.
Jean and Jiménez (2011) investigated this in OECD The idea of an eventual demographic dividend,
countries and found no permanent effect of immi- similar to that seen in the first and second demo-
gration on unemployment amongst nationals. They graphic transitions, is not suggested by any of the
discovered that changes in the percentage of immi- above authors; neither do they refer to the concept
grants in the workforce may have a temporary ef- of the third demographic transition mentioned by
fect on unemployment in the national population. Coleman. However, they do address the repercus-
sions of immigration on labour market variables
(salaries, employment) or on social and political as Austria, Belgium and, in particular, Luxemburg,
tension occurring between the local population and where it has surpassed one tenth of the total popu-
immigrants. lation (reaching almost 45% in the third dividend).
The foreign population in Estonia and Spain has
also gained considerable importance. However, con-
3. Stylised facts: immigration in European trary to the previously-mentioned countries, in Es-
tonia there is a constant withdrawal of immigrants
countries
and in Spain the situation reached a peak at some
time during the study period and there is now a
There are 20 European Union countries in the Or- certain ebb in the immigration tide. In other terri-
ganisation for Economic Co-operation and Devel- tories, such as Denmark, France, Italy, Sweden and
opment (OECD) and the impact of the presence of the UK, levels have reached an intermediate mag-
a foreign population on the population as a whole nitude of between 5% and 10%, with a positive dy-
and on labour differs in each one. Figure 1 repre- namic throughout the period; however, in Germany
sents the overall percentage of foreigners in the total and Greece, with similar figures, there has been a
number of population for 2006 and 2014. The graph significant slowdown with regard to the foreign
is divided by a diagonal line starting from zero. In population. At the other end of the scale, Hungary,
the countries to the left, the percentage of foreign- Slovakia and Poland have hardly any foreigners in-
ers is higher in 2014 than in 2006 and the opposite side their borders.
occurs in those to the right. The vast majority of Figure 2 shows analogous information, though
countries are situated above or along the line, with it is restricted to the working population segment.
very few below it; Estonia is possibly the most sig- The graph reading is identical to that of Figure 1.
nificant case due to the progressive naturalization of As in the case of the population as a whole, the pe-
its Russian minority, but slight decreases have also riod in question also witnessed an increase in the
been observed for Germany and Greece. Thus, the number of foreigners in the labour force subgroup.
foreign population is undeniably dynamic, reaching The exceptions are Estonia, for the previously-men-
estimable percentages of growth in countries such tioned reasons, and Spain and Portugal, due to the
Fig. 1. Participation rate of foreigners in the total population in 2006 (FP2006) and 2014 (FP2014)
Source: Eurostat. Own compilation
Fig. 2. Participation rate of foreigners in the total labour force in 2006 (FW2006) and 2014 (FW2014)
Fig. 3. Relationship between the number of foreigners in the total population (FP2006) and the labour force (FW2006)
Fig. 4. Relationship between the number of foreigners in the total population (FP2014) and the labour force (FW2014)
severe recession affecting both countries. At the end ent authors have decided to consider persons born
of the study period, the weight of foreign workers abroad as a criterion for measuring the real impact
is overwhelming in Luxembourg, where it exceeds of immigration (Coleman, 2006). In this study, for
50%; significant in Austria, Estonia, Italy and Spain, reasons derived from the structure of the sources,
where it exceeds one tenth; or in Belgium, Germa- we have opted for the nationality criterion as Eu-
ny and the United Kingdom, slightly below the rostat provides data for 2006 and 2014, just before
above-mentioned level. the wave of migration in 2015. Figure 5 relates per-
Economic theory highlights motivation to work centages of the total population of foreign citizens
as the main cause of immigration, with an expect- with persons born in other countries. The points
ed higher number of foreigners in the labour force above the diagonal line on the map indicate coun-
than in the population as a whole. Figures 3 and tries where the percentage of foreigners exceeds the
4 link both percentages for 2006 and 2014 respec- percentage of the foreign-born population, an unu-
tively. Points above the diagonal line imply that the sual and anecdotal situation, brought about by spe-
fraction of foreigners is higher in the labour market cific situations such as small nations and strong
than in the total population. With the exception of employment areas close to their borders (Luxem-
Luxembourg, the percentages are close to each oth- bourg fits this profile quite well; the situation in the
er, with some convergence throughout the studied Czech Republic is more difficult to explain).
period. This effect would be influenced by financial In the balance between both percentages, four
difficulties at the time, via a decrease in the number factors favour a greater difference between the for-
of foreigners in the formal labour market, or by the eign-born population and foreign citizens: the first,
activation of family reunification processes. which affects some countries, is the result of the
The fact that ‘immigrant’ is not synonymous eventful history of Europe and the forced move-
with ‘foreigner’ means that in many countries pro- ment of people following the latest global confla-
cesses for granting citizenship tend to underestimate gration, which still has implications in countries
the importance of immigration associated with the such as Germany (Douglas, 2012), and has gener-
condition of being foreign. For this reason, differ- ally left its mark on practically all of Europe (Rei-
Fig. 5. Foreigners (FP 2014) versus those born in other countries (FB 2014)
nisch ,White, 2011); repatriations following the ing as they are based on derisory percentages; in
emancipation of former colonies, as in the case the case of Italy, the fact that it is the first coun-
of France and Algeria (Jordi, 2002) or the United try on the central Mediterranean routes of irregular
Kingdom and the Commonwealth (Hansen, 2002); immigration, is significant. In Finland, there was a
the wave of migrants from south to south east Eu- modest proportion of foreigners in 2006, which has
rope towards countries in the north and west of the increased due to the growing attraction of Nordic
continent, encouraged by the strong demand for la- countries. Many of these immigrants are refugees
bour during the economic boom in 1950 and 1970 (as in the case of Sweden, where rapid growth in the
(Messina, 2007); the disparate generosity of nation- number of younger foreigners can be seen). Final-
al legislations in granting citizenship. Logically, the ly, amongst countries that traditionally receive im-
incidence of any of these factors would have a low- migrants, only the United Kingdom remains as one
er impact in the present day, especially movements of the destinations with the highest level of growth
following the Second World War, but could be de- due to the labour market and attractive cultural en-
cisive in countries where immigration is currently vironment for immigrants. At the other end of the
scarce, such as Eastern European countries. On the scale are the countries where a decrease in the im-
contrary, those that have only recently acquired the migrant population is highlighted, particularly Es-
role of immigration countries (Spain, Italy) are sit- tonia and Denmark, though with nuances (Bench,
uated on the equiproportionality line, the difference Mouritsen, 2013): in the first case, specific tensions
between the birth and citizenship criteria increasing associated with its wide Russian minority could
only by applying the mechanisms of naturalisation. have led to part of this group being naturalised; in
Figure 6 shows the variation in the percentage the second case, we believe that a restrictive pol-
of foreign workers aged 30–39, 40–49 and 50–59. icy with regard to the entry of foreigners may be
Finland, Italy, the Czech Republic and the United behind these data. With regard to the other cases,
Kingdom can be seen as the leading countries in only Spain appears in the 30 to 39-year-old group,
the growth of this group in all age ranges. In the where the severity of the crisis has put a stop to
case of the Czech Republic, the data are mislead- the flow of mostly young immigrants; no new cas-
es have been mentioned for the 40 to 49-year-old

group. Finally, Austria and Germany, which have
been receiving immigrants for decades, appear in
the 50 to 59-year-old group. Nationalization may be
the cause of the variation observed.
From the perspective of spatial contrasts, this
approximation enables our essentially economistic
theses to be modulated. The cases of Italy, Estonia
and even Finland or Sweden underline the impor-
tance of other motivations: the pursuit of a better
future means that these countries, either because of
their geographical location on immigration routes
or their positive attitude to asylum for humanitari-
an reasons (for the case of Sweden, see vid. G. Le-
maitre -2007-), receive large numbers of immigrants
who care little about differences in salary in one la-
bour market or another.
4. Empirical approach
We selected a number of indicators that quantify

the impact of the workforce on one of the main
elements defining the labour market: salaries. We
propose a methodology based on the analysis of re-
gression with panel data, using a combination of the
fraction of foreign workers organized by age groups,
and the measurement of disparities between for-
eigners and nationals for different attributes. Final-
ly, we propose two approaches: a strictly linear one
and another in which the reciprocal transformation
of determined variables is used to give more rele-
vance to countries where fewer differences exist be-
tween local and foreign population attributes.
4.1. Data
Twenty countries belong both to the European Un-

ion and the OECD. However, there are gaps in the
disaggregated information available in the Eurostat
data base for 2006 and 2014 in certain countries,
specifically Greece, Hungary, Poland, Portugal, Slo-
vakia and Slovenia. In this context, countries with
the lowest incomes and levels of immigration are
Fig. 6. Variation in the representativeness of foreign work- not included in the analysis. These gaps have also
ers led to the study of impacts on groups of workers
being restricted. National specificities regarding ac-
cess to and withdrawal from the labour market are weighted average of the percentage of foreigners in
reflected in disparate initial and final intervals. Co- each quarter.
horts of workers aged between 30 and 59 were tak- The final block, generically labelled by the pre-
en into account to avoid problems with statistical fix GAP, consists of a succession of variables used
consistency. to quantify discrepancies for determined attributes
Thirteen variables were selected (Table 1), four among foreigners and nationals in each country.
of which measure the impact of migration on so- These variables acquire the proxis value to intro-
ciety as a whole, and nine are indicators of migra- duce the effect of the differential between the two
tion itself. groups on the dependent variables that do not dis-
All of the variables collected for the impact of criminate them (the information on annual salaries
the foreign population are dependent and corre- supplied by Eurostat refers to total amounts). The
spond to the annual average pay of full-time work calculation procedure is homogenous.
by age group (SAL3039, SAL4049, SAL5059). This
information, originally expressed in euros, was (1)
transformed into common logarithms, mitigating
the distorting effect of extreme variables on the re- where X is a percentage relative to any variable.
sults obtained. We can only access this information It denotes an observation in country i and at time
every four years. t, ext represents the foreign population, and nac the
Immigration in the world of labour was quanti- national population of the country in question. The
fied by the percentage of foreign workers in com- resulting values may be positive or negative, imply-
parison with the total work force. Together with ing different situations with regard to the position
salaries, it is the only information that has been of foreigners in the host society as a whole.
particularized by age groups (EXT3039, EXT4049, These variables refer to three different fields, lev-
EXT5059). It is calculated based on four-month- el of education, poverty risk amongst workers and
ly observations, annual data being obtained as the unemployment. The estimation procedure consists
Table 1. Dependent and independent variables

Variable Type Description
SAL3039 Dependent Mean annual salaries of full-time workers aged between 30 and 39. Expressed in
logarithms
logarithms
logarithms
EXT3039 Independent Percentage of foreign workers aged between 30 and 39
GAPEDU02 Independent Difference between the percentage of foreign and national workers with basic
studies (levels 0 to 2)
GAPEDU34 Independent Difference between percentages of foreign and national workers with intermedi-
ate studies (levels 3 and 4)
GAPEDU58 Independent Difference between percentages of foreign and national workers with higher stud-
ies (levels 5 to 8)
GAPRPW Independent Difference between percentages of foreign and national workers at risk of poverty
GAPUNEMP1539 Independent Difference between percentages of unemployed foreign and national workers
aged between 15 and 39
GAPUNEMP Independent Difference between percentages of foreign and national workers
Source: Own compilation
of calculating the difference corresponding to per- there is a broad spread amongst the data for the dif-
centages that represent a specific attribute between ferent countries, and also amongst practically all of
foreigners and nationals in each country. A posi- the variables, as indicated by the values for standard
tive value implies higher values for the mentioned deviation. Immigrants have a better primary level of
attribute amongst foreigners, and a negative one the education than nationals, noticeably lower at inter-
opposite. Statistics provided by Eurostat enable the mediate level and similar with regard to higher ed-
level of education to be disaggregated in three stra- ucation; in this series of indicators, some countries
ta: primary (GAPEDU02), secondary (GAPEDU34) can be highlighted for different reasons: in primary
and higher studies (GAPEDU58). If foreigners have studies the positive differential is maximum in Ger-
a positive differential in primary studies, we can in- many, France, Finland (2014) and Sweden (2014),
fer that they compete for unskilled jobs. If this oc- whereas Estonia and Spain show negative values;
curs in the case of secondary studies it would imply the differentials for higher studies are positive for
greater competition for “blue collar” posts, where- the Czech Republic, Denmark, Luxemburg, Sweden
as in the case of overqualification (a higher percent- and the United Kingdom (the latter in 2014 only),
age of foreigners with higher education) they will which are now destinations for highly qualified im-
presumably compete more for “white collar” posts. migrants. They are also poorer than nationals, the
Generally speaking, the integration of immi- risk of poverty being very similar only in Holland
grants is difficult. This is particularly true in first (2006, 2014) and the Czech Republic (2014). Fi-
generations, above all in the case of those who have nally, unemployment rates are higher, particularly
not acquired nationality. A good indicator of the de- amongst young people; unemployment is higher in
gree of integration is the number of workers at risk both population groups only in the Czech Republic.
of poverty, salaries somehow reflecting the difficul-
ties faced by immigrants when trying to improve
their position in the labour market. 4.2 Methodology
Unemployment is the last element in the set of
indicators depicting differences between the two Scientific literature dealing with the impact of im-
groups. The idea is to measure the incidence of un- migration on the labour market, as seen in section
employment in the immigrant population, for both 2, focuses on its influence on employment and sal-
the working population as a whole (GAPUNEMP) aries. Also, generally speaking, works that use re-
and younger age groups (GAPUNEMP1539). gressions on existing data employing fixed effects
The statistical descriptions are shown in table 2. predominate (Carrasco et al., 2008; Glitz, 2012;
The percentage of foreigners decreases with age and Borjas, 2006; Cohen-Goldner, Paserman, 2011;
Table 2. Statistical descriptions

Standard
N Minimum Maximum Mean
deviation
SAL3039 42 9.165 10.929 10.410 0.460
SAL4049 42 9.016 11.088 10.503 0.521
SAL5059 42 8.878 11.156 10.510 0.561
EXT3039 42 1.34 57.88 12.44 13.10
EXT4049 42 0.84 52.47 9.68 11.19
EXT5059 42 0.46 42.46 8.24 10.12
GAPEDU02 42 -8.30 28.30 7.58 10.09
GAPEDU34 42 -24.80 14.50 -7.43 11.02
GAPEDU58 42 -12.20 19.70 -0.13 8.08
GAPRPW 42 -0.50 26.90 9.41 5.86
GAPUNEMP1539 42 -4.70 17.60 5.66 4.49
GAPUNEMP 42 -2.80 15.30 6.41 3.95
Dustman et al., 2010) in comparison with other al- cal model is more complex to interpret because
ternatives such as panel-type data regressions with although the coefficients are linear, the result
random effects (Islam, Parasnis, 2016), analyses of of the transformation is not. The field of var-
substitution elasticities (Gang et al., 1994; DÁ- iation has a basic characteristic, it asymptoti-
muri et al., 2010), and analyses of delays in time
cally tends towards zero, so that the effects of
series (Jean, Jimenez, 2011). In this paper, pooled
high values tend to disappear whereas low val-
OLS fixed effects have also been used, as well as
heteroskedasticity-robust HAC errors. The gener- ues have a very large impact; logically, the mag-
al model employed has the following specification: nitude of the impact on the Dependent variable
depends on the coefficient of regression which
(2) is constant. The reason justifying this is the na-
ture of the information provided by the >GAP
where y is the different target variable for variables: they measure discrepancies between
each regression (salaries per age group), is the foreign and local populations, considering
the constant term, X is a different independ- that their real impact on salaries becomes less
ent variable according to the dependent varia- noticeable as it increases, until both variables
ble used (percentage of foreign workers per age become independent because, generally, in ex-
group), and Z is a pool of j variables that vary treme cases they would refer to small marginal
in time and with each transversal unit, but do groups (for example, if we record the existence
not change for the different regressions (rela- of extreme overqualification, i.e. a very high
tive variables at training, risk of poverty and positive value of foreigners with higher educa-
unemployment level). Finally, is the error term tion, this very likely reflects the existence of a
so that. very small minority pool of immigrants, usual-
This model was particularised based on two ly employed in universities or research centres;
approaches to the data for differences between conversely, when there is a very strong posi-
foreign and national workers: the first consid- tive differential for foreign workers with basic
ers that these variables are linear, whereas in studies, this could be related to minority groups
the second they are considered non-linear by with a very low impact on the labour market
their transformation into a reciprocal model. due to precariousness). In other words, we tend
The former is not difficult to read, the effect to magnify the impacts of immigration on sal-
of the regressor is proportional to the field of ary when the status of the immigrant is similar
variation of the variable, depending on the fi- to that of the national worker, and minimize it
nal sign of the product of the coefficient signs in the opposite case.
and the variation of the variable. The recipro- Considering both specifications the general
model was particularised as follows:
A. Non-transformed GAP variables:
(3)
(4)
(5)
where
(6)
B. Transformed GAP variables:
(7)
(8)
(9)
where
(10)
rec indicating that the variable was transformed thus the analysis focuses on differentials between
by applying the reciprocal model, and are coeffi- foreigners and nationals in the country in question
cients of regression, whereas u, v and w refer to er- (the variables assigned GAP).
rors, so that . The regressions using non-transformed GAP var-
The specifications will undergo transformations iables show different results depending on whether
because as the coefficients are estimated, restricted salaries of workers aged between 30 and 39 years or
equations will be designed, eliminating statistical- those of older workers are taken as the regressand.
ly insignificant variables (18 models were estimat- In the first case, the education differential does not
ed, ten if the original values were maintained and seem to be important in the variation in salaries,
eight for the transformed results). which explains why a second regression was made
to estimate the coefficients of the unemployment
differentials (in the restricted specification GAPUN-
4.3. Results EMP1539 is significant at 0.05, whereas GAPUN-
EMP is significant at 0.10; both are significant at
Tables 3 and 4 show the results of the estimation 0.01 in the restricted model). The explanatory ca-
of the models, the first using original GAP varia- pacity of specifications (1) and (2) is moderate in
bles, the second transforming them using the re- the first case but decreases substantially in the sec-
ciprocal model. Each table consists of three panels, ond (see R-squared intra), the lower quality of the
each corresponding to a different dependent varia- restricted model translates into a risk of correlation
ble. The columns inside each panel correspond to a between the independent variables and errors, albeit
different specification, the column with estimations the quality of information is slightly better. Adher-
on the far left are unrestricted models, whilst those ing to the explanatory variables, higher unemploy-
on the left are models successively restricted to the ment amongst young foreigners implies a decrease
statistically significant parameters. The rows show in salaries; on the other hand, a higher rate of un-
the results of the estimations. The upper one shows employment in the group of foreigners as a whole
the constant of the regression, the seven rows im- implies slight increases in salaries. However, accord-
mediately below the estimation of the coefficients ing to economic theory, the larger pool of the un-
of regression correspond to the independent varia- employed increases the number of job opportunities
bles; whereas the four lower rows include parame- and therefore salaries tend to decrease, although
ters for goodness of fit (coefficient of determination some caution is required when interpreting the var-
–R squared LSVF, R squared intra-), suitability of iable in model (1) due to its little significance.
the specification (AIC, Akaike information criteri- Salaries in older age groups, 40 to 49 and 50 to
on) and correlation between independent variables 59, give substantially different results. Focusing on
and the error term (Durbin-Watson criterion). the salaries of middle-aged workers, three models—
Fourteen different specifications, seven models (3), (4) and (5)—were estimated. The first is unlim-
for each of the two procedures used, were estimat- ited, with a noticeably higher explanatory capacity
ed. Non-statistical significance of the percentage of than its restricted versions, greater informative qual-
foreign workers in comparison with the whole la- ity, and there are no consistent signs of correlation
bour force is the element common to all of them, with errors. The significant variables in the first ver-
Table 3. Results of the regressions with non-transformed GAP variables

Dependent variable: SAL_30_39 (1) (2)
10.08*** 10.292***
Constant
(0.188522) (0.027001)
0.0953141
EXT3039
(0.113916)
0.253868
GAPEDU02
(0.113916)
0.259801
GAPEDU34
(0.239494)
0.259531
GAPEDU58
(0.291695)
0.00954191
GAPRPW
(0.00688716)
-0.336836** -0.0538157***
GAPUNEMP1539
(0.0160626) (0.0117341)
0.0414868* 0.0658394***
GAPUNEMP
(0.0209098) (0.0118298)
R2 LSDV 0.972605 0.967359
R2 Intra 0.372266 0.252059
AIC -56.12573 -58.76705
Durbin-Watson 1.363284 1.266606
Dependent variable: SAL_40_49 (3) (4) (5)
10.2141*** 10.449*** 10.4475***
Constant
(0.188594) (0.0657088) (0.0534241)
0.0790542
EXT4049
(0.127983)
0.449033* 0.610904** 0.610632**
GAPEDU02
(0.252562) (0.218446) (0.219531)
0.45346* 0.60492** 0.604668**
GAPEDU34
(0.248197) (0.216979) (0.217987)
0.455302* 0.606348** 0.606142**
GAPEDU58
(0.249933) (0.219864) (0.220854)
0.0110262
GAPRPW
(0.00773521)
-0.0362213* -0.000302315
GAPUNEMP1539
(0.0195158) (0.0058792)
Table 3. Continuation
0.0417558
GAPUNEMP
(0.0253846)
R2 LSDV 0.971970 0.961494 0.961493
R2 Intra 0.356473 0.115957 0.115926
AIC -44.70351 -37.36680 -39.36530
Durbin-Watson 1.310305 1.3424413 1.343636
Dependent variable: SAL_50_59 (6) (7)
10.268*** 10.4624***
Constant
(0.141557) (0.0525686)
0.075835
EXT5059
(0.107202)
0.434271** 0.538798**
GAPEDU02
(0.169056) (0.231929)
0.439654** 0.53367**
GAPEDU34
(0.163946) (0.230132)
0.443782** 0.536568**
GAPEDU58
(0.164546) (0.232335)
0.0102547
GAPRPW
(0.00752463)
-0.0371733
GAPUNEMP1539
(0.0264311)
0.0418272
GAPUNEMP
(0.0264311)
R2 LSDV 0.978857 0.971470
R2 Intra 0.338368 0.107219
AIC -50.283311 -45.690856
Durbin-Watson 1.356342 1.389672
(1), (3) and (6), unrestricted models; (2), specifications restricted to GAPUNEMP1539 and GAPUNEMP; (4), only GAP variables relative to the
level of education and GAPUNEMP1539 are considered; (5) and (7), only the three GAP variables relative to studies are considered to be regres-
sors. Note: ***p < 0.01, ** p < 0.05, * p < 0.10
sion are the three relative to the education level dif- is on the limit. The final model for this regression,
ferential and GAPUNEMP1539, all of which are at (5), implies that only the three education differen-
0.10. Having considered the four significant varia- tials, which are still significant at 0.05, are consid-
bles, we specified a new model (4), in which only ered; however, the quality of the results remains at
the differentials in education (all at 0.05) detract very discreet levels because despite the goodness of
influence from the statistics perspective; the exclu- fit being close to that of model (4), the value for
sion of non-significant variables in the unrestricted the Durbin-Watson coefficient indicates correlations
model implies a significant decrease in explana- with the error term. Adhering to the reading for the
tory power (R2 intra of 0.116), and, although the statistically significant coefficients, overqualification
Durbin-Watson coefficient does not imply the exist- of foreigners in comparison with nationals implies
ence of correlation between regressors and errors, it higher salaries at any level; similarly, a higher rate of
unemployment amongst young foreigners has an in- the incidence of disparities in basic and average lev-
direct effect on the salaries of middle-aged workers, els of education are not significant. However, over-
which decrease. However, this effect is quite weak, or underqualification in terms of higher education
as exemplified by the trajectory of this coefficient in are important as they determine salaries. Another
successive estimations. regularity comes in the form of discrepancies in un-
Two models, (6) and (7), were calculated for the employment amongst young people, which, except
incomes of workers aged 50 to 59. In this case, the in the case of salaries of younger workers, do not
only variables that have an effective impact on sal- have a very significant impact on the regressands.
aries are those related to discrepancies between the Generally, the explanatory power of the regressions
level of studies reached by foreigners and nationals using this procedure is substantially lower, at least
(they are significant at 0.05 in both the unrestricted in the case of the unrestricted specifications. This
and restricted variant). The exclusion of four regres- does not occur with the models restricted to statis-
sors in the restricted version implies a sharp drop tically significant variables which produce slightly
in the goodness of fit of the model (R-squared intra lower results from the point of R2 intra.
is 0.338 in the unrestricted specification and 0.107 Effects on the salaries of workers aged between
in the restricted one), not compensated for by bet- 30 and 39 are shown in models (8) and (9). As pre-
ter quality information or by greater significance of viously mentioned, the regression has a low R2 in-
the coefficients, though in both cases there is no tra, specifically, 0.233 in the complete specification
clear evidence of correlation with errors. As in the and 0.120 in the restricted one. Both models have
case of middle-aged workers, a higher level of edu- similar informative quality, though the former has
cation amongst foreigners in comparison with na- a slightly more favourable Durbin-Watson value in
tionals implies higher salaries, even though values the restricted specification (the risk of positive cor-
for the coefficients are slightly lower. relation between independent variables and errors
Different levels of education play a major role in is lower). The impact of variations in the regres-
the salaries of older groups of workers. The more sors on salaries is only significant in GAPEDU58
highly-qualified the foreign worker is at any level and GAPUMEMP1539 (both at 0.05 significance),
of education, the more salaries increase. This effect FAPRPW and GAPUNEMP (at 0.10); this signif-
can be seen from two perspectives: a standard ex- icance increases in the restricted model (with the
planation would indicate that a higher number of exception of youth unemployment which is signifi-
better-educated foreigners implies a greater supply cant at 0.10; 0.01 in the remaining regressors). The
of workers and therefore a negative impact on sal- impact of variations in these independent variables
aries should be expected, which is not the case in is negative, except for youth unemployment, the
the light of these results. The alternative explanation impact of divergences on the risk of poverty being
carries the greatest weight. The higher the level of particularly noticeable. On the other hand, and dis-
education amongst foreigners, the greater their con- cordantly with economic theory, a higher level of
tribution to productivity, the higher the company’s youth unemployment increases the salaries of un-
income and therefore the higher salaries are. Ac- skilled workers aged between 30 and 39.
cording to these data, the second explanation would The effect of the independent variables on the
appear to be more feasible. salaries of workers aged between 40 and 49 is de-
Table 4 shows the results of regressions obtained scribed in three models, with specifications that are
from the reciprocal transformation of variables rel- the result of subsequent restrictions based on the
ative to differences between attributes of foreign- statistical significance of the regressors. Model (10),
ers and nationals in each country. This approach which is unrestricted, determines that only the high-
implies overweighting observations in which both er education differential (significant at 0.01), and, to
groups are close to each other (the section of values a lesser extent, the risk of poverty and unemploy-
between -1 and 1 tend to infinity as differences be- ment rates have an impact (the three coefficients
tween them are annulled, whilst they asymptotical- corresponding to these indicators are significant at
ly tend to zero between 1 and -1). A comprehensive 0.10). In model (11), taking into account only the
approach enables us to reach an initial conclusion: significant exogenous variables in specification (10),
Table 4. Results of the regression using reciprocal transformations of GAP variables
Variable Dependent: SAL3039 (8) (9)
10.252*** 10.4191***
Constant
(0.23731) (0.00430422)
0.094617
EXT3039
(0.101056)
0.0171654
GAPEDU02rec
(0.030359)
0.116277
GAPEDU34rec
(0.0687299)
-0.00745803** -0.00963019***
GAPEDU58rec
(0.00292487) (0.00271379)
-0.163624* -0.06966634***
GAPRPWrec
(0.0888689) (0.0242508)
0.0259396** 0.00890515*
GAPUNEMP1539rec
(0.0111406) (0.00482128)
-0.097482* -0.0495105***
GAPUNEMPrec
(0.0509853) (0.0132106)
R2 LSDV 0.966539 0.961616
R2 Intra 0.233262 0.120460
AIC -47.72455 -47.95989
Durbin-Watson 1.495400 1.336127
Variable Dependent: SAL4049 (10) (11) (12)
10.3658*** 10.5206*** 10.5159***
Constant
(0.274561) (0.00389974) (0.00463913)
0.100832
EXT4049
(0.132307)
0.0268465
GAPEDU02rec
(0.0281198)
0.126842
GAPEDU34rec
(0.0796851)
-0.0101865*** -0.012456*** -0.0115176***
GAPEDU58rec
(0.00186916) (0.00134848) (0.00239513)
-0.201209* -0.084647*** -0.0980548***
GAPRPWrec
(0.1005508) (0.0217038) (0.0249552)
0.0238189* -0.000141514
GAPUNEMP1539rec
(0.0133055) (0.00220955)
Table 4. Continuation
-0.108362* -0.0533368*** -0.0630282***

GAPUNEMPrec
(0.0593293) (0.0111554) (0.013333)
R2 LSDV 0.967541 0.972376 0.962779
R2 Intra 0.254783 0.135554 0.145462
AIC -38.54163 -45.05314 -40.79247
Durbin-Watson 1.385944 1.271084 1.248976
Variable Dependent: SAL5059 (13) (14)
10.4782*** 10.5206***
Constant
(0.222403) (0.00396965)
0.0474597
EXT5059
(0.127142)
0.0217663
GAPEDU02rec
(0.0277805)
0.109568
GAPEDU34rec
(0.0810588)
-0.0127196*** -0.01246446***
GAPEDU58rec
(0.00217412) (0.00130956)
-0.178687* -0.0846281***
GAPRPWrec
(0.0929647) (0.0214739)
0.0162651
GAPUNEMP1539rec
(0.014948)
-0.0975466* -0.0533379***
GAPUNEMPrec
(0.0528241) (0.0111633)
R2 LSDV 0.974684 0.972376
R2 Intra 0.207795 0.135553
AIC -42.71842 -47.05308
Durbin-Watson 1.412180 1.271137
(8), (10) and (13), unrestricted models; (9) and (11), models restricted to GAPEDU58, GAPRPW, GAPUNEMP3539 and GAPUNEMP; (12) and
(14), only consider variables GAPEDU58, GAPRPW and GAPUNEMP. Note: ***p < 0.01, ** p < 0.05, * p < 0.10
they all have statistically significant coefficients at the restricted versions, which have a lower Akai-
0.01, with the exception of differences in employ- ke value (the minimum corresponds to specification
ment amongst young people. Model (12) confirms, (11)). The risk of correlation between regressors
in essence, the conclusions obtained by specifica- and errors is always present: although we cannot
tion (11). The explanatory power of the restricted affirm that it exists in the unrestricted regression,
version is low (R2 intra of 0.255), but the decline the Durbin-Watson values indicate that it hampers
in this coefficient of determination is lower than its the restricted versions. Generally, the coefficients of
non-transformed equivalent in the exogenous var- regression operate by reducing salaries; the effect
iables, since for models (11) and (12) it is quanti- of overqualification on higher studies in foreigners
fied at 0.136 and 0.145, respectively. Analogically, tends to reduce salaries, which is consistent with the
the informative quality of these models is better in economic theory because the supply of workers in-
creases; the same occurs in the case of the greater cial results and increased social wealth (Lee, Ma-
risk of poverty amongst foreign workers and a high- son, 2006).
er rate of unemployment among them, these results As we approach a world in which life expectancy
being expected: the first indicator refers to groups at birth is 76 globally, we have achieved something
of population with a lower wage-bargaining pow- unprecedented in history: we have added a new 30-
er, the second group, again, has the effect of an in- year stage to human existence.
crease in job offers. Evidence suggests that the second demographic
The final regression corresponds to the sala- dividend does not capture all the potential of this
ries of workers aged between 50 and 59. The results demographic change. Sustainable benefits could be
of its specifications have lower explanatory values possible if we design and create significant roles for
(the unrestricted version has an R2 intra of 0.208, older adults that greatly benefit society. We could
and the restricted one, 0.136) contrasting results be- build a new stage of human development, a “third
ing observed for the quality of information and risk demographic dividend”; a dividend in which the
of correlation of errors: model (13), which is un- roles and responsibilities of older adults in the last
restricted, has a higher Akaike value (it offers less third of life bring new sustained social capital to
information) than the restricted one (14); on the satisfy social needs and create better social welfare
other hand, the Durbin-Watson index points to the beyond the second demographic dividend (Fried,
presence of correlation with errors in the restricted 2016).
version, whereas this cannot be affirmed in the un- Many of our politicians believe that this addi-
restricted one. Focusing on the coefficients, a fairly tional stage of life is an obstacle in every sense of
similar situation to estimations made for the salaries the word, with the elderly devouring dispropor-
of middle-aged workers was found. The same ex- tionate parts of our public and financial resourc-
ogenous variables are statistically significant in the es at the expense of younger generations. Political
unrestricted version, with the exception of unem- leaders worldwide are concerned about the “rate
ployment amongst young people. In model (13), the of old-age dependency” and the balance between
GAPEDU58rec variable is significant at 0.01, and working-age taxpayers and those who have retired
GAPRPWrec and GAPUNEMPrec at 0.1. Including and are “dependent” on society and use public sec-
these three variables only in the restricted model, tor resources. The main engine of this transition-
they all become significant at 0.01. As in the regres- al cycle will be immigration. Migratory movements
sion proposed for the salaries of workers aged be- may come to represent the principal force govern-
tween 40 and 49, once again, negative relationships ing the evolution of the economy in this century.
are found: an inversely proportional impact is ob- We should therefore be aware of international mi-
served when disparity between foreigners and na- gratory movements, structural changes and new de-
tionals is higher. mographic trends.
The set of regressions that we have made un-
der the assumption of a linear relationship shows a
5. Conclusions clear disparity between the response of salaries of
younger workers and those of middle-aged and old-
er workers. In the former case, the factors most in-
During the last 100 years, investment in social wel- fluencing variations in salaries are related to higher
fare has transformed our life expectancy. Now, al- or lower levels of unemployment amongst foreign-
most all societies have passed the “first demographic ers in comparison with nationals from the country
dividend” of the transition from high mortality in in question. As can be seen, this impact is nega-
an agrarian society to one with a lower infant mor- tive in the case of unemployment which has a di-
tality rate. Most societies are now in different stag- rect effect on this group, and on the contrary, it is
es of the second demographic dividend, promoting positive with regard to differences in unemploy-
the accumulation of capital and more individu- ment affecting all workers, regardless of their age.
al savings. If employment is available and policies The first result is almost to be expected, a higher
are constructive, this will lead to positive finan-
level of unemployment amongst foreign youths im- of unemployment amongst foreigners, means lower
plying that this group forces salaries down. salaries. The explanations in this paragraph are valid
We have seen that the impact on the third divi- for the remuneration of other age groups.
dend is caused by immigration which would fill the We end our conclusions by indicating that this
gaps in the labour market; therefore, the trigger- paper has focused on salaries, which is an exoge-
ing force will be the continuous arrival of people nous variable. This strategy implies that we have put
from other countries. This is due to the low natural other equally relevant indicators to one side from
growth rate in the countries in this study. the perspective of demand in the labour market.
Regressions under the assumption of inverse re- These would include the rate of job creation, , or
lationships indicate a specific characteristic of these other reasons such as aspiring to improve the liv-
transformations, which imply important nuances ing conditions of many immigrants from develop-
when reading the coefficients: if they are positive ing countries, or the phenomenon of refugees from
or negative, the reading has the usual interpreta- countries torn by war or persecution on ideological
tion (positive values imply growth in the regres- or religious grounds. We are aware of the implica-
sand when the change in direction of the regressor tions of a limited number of observations in a sta-
is positive). This does not occur with the intensi- tistical analysis of the panel, which was due to the
ty of growth of the independent variable; that is, if time sequencing of some relevant variables in the
the difference between the levels of basic education Eurostat data base, thus limiting our analysis to a
amongst foreigners in comparison with nationals of short panel (three years and 14 countries). There-
the country in question is positive and the coeffi- fore, there is a risk that the coefficients may not be
cient is negative, the impact on salary is negative, robust. On the one hand, the sources used, and on
but this repercussion is maximum when the levels the other, the deliberate choice of economic varia-
of education in both groups are similar and lower bles rather than other social or even cultural ones,
when the difference is more obvious, until it dis- limit the scope of this paper which should neces-
appears (it should be remembered that reciprocal sarily be enhanced by other future papers that may
models asymptotically tend towards zero). overcome shortcomings implicit in this study.
If the difference in the level of higher studies
between foreigners and nationals increases, point-
ing to either overqualification or underqualification, References
this displacement would indicate a reduction in sal-
aries in the former case and an increase in salaries
in the latter. The regressions that used exogenous Bauer, T. K.; Lofstrom, M. and Zimmermann, K. F.,
non-transformed variables and found the impact of 2000: Immigration policy, assimilation of immi-
overqualification on the level of education in mid- grants and natives´ sentiments towards immigrants:
dle-aged and older workers relevant, gave positive evidence from 12 OECD-countries. In: IZA Discus-
results. A possible explanation for this could be sion Paper, 187. Available at http://ftp.iza.org/dp187.
the impact of the work factor on productivity; but pdf (DoA: 20/03/2017).
it was also indicated that overqualification due to Bench, E. C. and Mouritsen, P., 2013: Restricting the
this concept producing negative coefficients asso- right to family migration in Denmark: When human
ciated to possible excess demand for work would rights collide with a welfare state under pressure. In:
be equally plausible. In other words, to the impact Jurado, E. and Brochmann, G. (Eds.), Europe´s immi-
that the extra labour supply brought about by for- gration challenge. Reconciling work, welfare and mobil-
eign workers has on salaries. Similar explanations ity. (Chapter 9). London: IB Tauris.
can be found for the other two explanatory varia- Bloom, D. E.; Canning, D. and Sevilla, J., 2003: The
bles: the differential of foreign workers outside the demographic dividend. A new perspective on the
system, poor workers, with lower wage-bargaining economic consequences of population change. San-
power, promote lower salaries; a greater discrepancy ta Monica: Rand.
between the supply and demand of workers in the Bloom, D. E., Canning, D., Fink, G. and Finlay, J. E.,
labour market, directly influenced by a higher rate 2007: Realizing the demographic dividend: is Afri-
ca any different?. In: Program on the Global Demog- idence from Israel. In: European Economic Review, 55:
raphy of Aging Working Paper Series, 23. Available at 1027-1045.
http://www.hsph.harvard.edu/program-on-the-glob- Coleman, D., 2006: Immigration and ethnic change in
al-demography-of-aging/WorkingPapers/2007/pgda_ low-fertility countries: a Third Demographic Transi-
wp23_2007.pdf (DoA: 22/03/2017). tion. In: Population and Development Review, 32(3):
Bloom, D. E., Canning, D., Fink, G. and Finlay, J. E., 401-446.
2009: Fertility, female labor force participation, and DÁnuri, F.; Ottaviano, G. I. P. and Peri, G., 2010: The
the demographic dividend. In Journal and Economic labor market impact of immigration in Western Ger-
Growth, 14: 79-101. many in the 1990s. In: European Economic Review,
Bloom, D. E.; Humair, S.; Rosemberg, C.; Sevilla, J. 54: 550-570.
and Trusell, J., 2013: A demographic dividend for Douglas, R. M., 2012: Orderly and humane. The expul-
sub-saharan Africa: source, magnitude and realiza- sion of the germans after the Second World War. New
tion. In: Program on the global demography of aging Haven: Yale University Press.
working paper, 110. Available at http://www.hsph.har- Dribe, M. et al., 2017: Socio-economic status and fer-
vard.edu/program-on-the-global-demography-of-ag- tility decline: Insights from historical transitions in
ing/WorkingPapers/2013/PGDA_WP_110.pdf (DoA: Europe and North America. In: Population Studies,
20/03/2017). 71(1): 3-21. DoA: dx.doi.org/10.1080/00324728.2016
Borjas, G. J., 2006: Native internal migration and the .1253857
labor market impact of migration. In: The Journal of Dustmann, C., Glitz, A. and Vogel, T., 2010: Employ-
Human Resources, XLI(2): 221-258. ment, wages, and the economic cycle: differences be-
Carstensen, L. L. and Fried, L. P ., 2011: The mean- tween immigrants and natives. In: European Economic
ing of old age. In: Beard, J.R., Biggs, S. Bloom, D.E., Review, 54: 1-17.
Fried, L. P., Hogan, P., Kalache, A. and. Olshansky, Durbin, J., Watson, G. S., 1950: Testing for Serial Cor-
S. J. (Eds.), Global population ageing: Peril or promise relation in Least Squares Regression, I. In: Biometrika,
(Chapter 1). Geneva: World Economic Forum. 37(3–4): 409–428. DoA:10.1093/biomet/37.3-4.409.
Cai, F., 2010: Demographic transition, demographic div- Durbin, J., Watson, G. S., 1951:Testing for Serial Corre-
idend, and Lervis turning point in China. In: China lation in Least Squares Regression, II. In: Biometrika,
Economic Journal, 3(2): 107-119. 38(1–2): 159–179. DoA:10.1093/biomet/38.1-2.159
Canudas-Romo, V. and Gillot, M., 2015: Truncated Erikson E., and Erikson, J., 1981: On generativity and
cross-sectional average length of life: a measure of identity: From a conversation with Erik and Joan
comparing the mortality history of cohorts. In: Pop- Erikson. In: Harvard Educational Review, 51(2): 249–
ulation Studies, 69(2): 147-159. 269.
Carrasco, R.; Jimeno, J. F. and Ortega, A. C., 2008: Fried L. P . 2015: A prescription for the next 50 years of
The effect of immigration on the labor market per- Medicare. In: Generations, Vol. 39 (2), pp. 180– 188.
formance of native-born workers: some evidence of Fang, C. and Wen, Z., 2012: When demographic divi-
Spain. In: Journal of Population Economics, 21: 627- dend disappears. Growth sustainability in China. In:
648. Aoki, M. and Wu, J. (eds), The Chinese economy: a
Chandrasekhar, C. P.; Gosh, J. and Roychowdury, A., new transition. New York: Palmgrave-MacMillan: 75-
2006: The “demographic dividend” and young India´s 90.
economic future. In: Economic and Political Weekly, Fried L. P., 2016: Investing in health to create a third de-
41(49): 5055-5064. mographic dividend. In: The Gerontologist, 56 (Suppl.
Coale, A. J., 1973: The Demographic Transition con- 2): S167– S177. DoA: 10.1093/geront/gnw035
sidered. Proceedings of the International Population Gang, I. N. and Rivera-Batiz, F., 1994: Labour market
Conference, Liege. In: Liege, International Union for effects of immigration in the United States and Eu-
the Scientific Study of Population: 53-57. rope. Substitution vs. complementarity. In: Journal of
Coale, A. J., 1984: The Demographic Transition. In: The Population Economics, 7: 157-175.
Pakistan Development Review, XXIII(4): 531-552. Glitz, A., 2012: The labour impact of immigration: a
Cohen-Goldner, S. and Paserman, M. D., 2011: The dy- quasi-experiment exploiting immigration location
namic impact on native´s labor market outcomes: ev-
rules in Germany. In: Journal of Labor Economics, France, Belgium and Switzerland. In: European Jour-
30(1): 175-213. nal of Population/Revue Européenne de Démographie,
Huber, P. and Oberdabernig, D., 2015: The impact of 18(4): 325–360. DoA: 10.1023/A:102112580007
welfare benefits on natives´ and immigrants´ attitudes Marozzi, M., 2016: Construction, robustness assessment
towards immigration. In: WWW for Europe Work- and application of an index of perceived level of so-
ing Paper, No. 82. Available at http://www.foreurope. cio-economic threat from immigrants: A study of 47
eu/fileadmin/documents/pdf/Workingpapers/WWW- European countries and regions. In: Social Indicators
forEurope_WPS_no082_MS17_Backgroundpaper.pdf Research, 128(1): 413-437. Available at link.springer.
(DoA: 21/03/2017) com/article/10.1007/s11205-015-1037-z.
Hansen, R., 2002: Citizenship and immigration in post- Mason, A. and Lee, R., 2006: Reform and support sys-
war Britain. The institutional origin of a multicultur- tems for the elderly in developing countries: captur-
al nation. Oxford: Oxford University Press. ing the second dividend”. In: Genus, LXII(2): 11-35.
Islam, A. and Parasnis, J., 2016: Native-migrant wage Mason, A., 2007: Demographic transition and demo-
differential across occupations: evidence from Aus- graphic dividends in developed and developing coun-
tralia. In: International Migration, 54(3): 89-109. tries. In: United Nations Expert Group Meeting on
Jäckle, S. and König, P. D., 2016: The dark side of the Social and Economic Implications of Changing Pop-
German “welcome culture”: investigating the causes ulation Age Structure. Mexico City, 31 August-2 Sep-
behind attacks on refugees in 2015. In: West Euro- tember 2005. New York: United Nations: 81-101.
pean Politics. DoA: dx.doi.org/10.1080/01402382.201 Mason, A. and Lee, R., 2007: Transfer, capital, and con-
6.1215614. sumption over the Demographic Transition. In: Clark,
James, K. S., 2008: Glorifying Malthus current debate on R., Ogawa, N. and Mason, A. (Eds.), Population Ag-
“demographic dividend” in India. In: Economic and ing, Intergenerational Transfers and the Macroecon-
political Weekly, 43(25): 63-69. omy. Cheltenham: Edward Elgar: 128-162.
Jean, S. and Jiménez, M., 2011: The unemployment im- Masquelier, B., Reiners, G. and Pison, G., 2014: Di-
pact of immigration in OECD countries. In: European vergences in trends in child and adult mortality in
Economic Review, 55: 241-256. sub-saharan Africa: survey evidence on the survival
Jordi, J. J., 2002: Les pieds noir: constructions identi- of children and sibling. In: Population Studies, 68(2):
taires et réinvention des origines. Hommes et Migra- 161-177.
tions, 1236: 14-25. Mayda, A. M., 2006: Who is against immigration? A
Kirk, D., 1996: Demographic Transition Theory. In: cross-country investigation of individual attitudes to-
Population Studies, 50(3): 361-387. DoA: dx.doi. wards immigrants. In: Review of Economics and Sta-
org/10.1080/0032472031000149536 tistics, 88(3): 510-530. DoA: 10.1162/rest.88.3.510.
Kreager, Ph., 2009: Two concepts of population: the long Mejia Guevara, I., Vélez Hernández Varela, F. and
view. Paper prepared for the XXVI IUSSP Interna- García López, J. E., 2010: El primer dividendo de-
tional Population Conference. Available at http:// mográfico y los sistemas de protección social en Méx-
iussp2009.princeton.edu/papers/90656. ico. In: Notas de Población, 7(90): 133-162.
Landry, A., 1934: La Révolution Démographique: Etudes Messina, A. M., 2007: The logics and politics of post-
et essais sur les problemes de la population. Paris: Li- WWII migration to western Europe. Cambridge:
brairie du Recueil Sirey. Cambridge University Press.
Lee, R., and Mason, A., 2006: What is the Demograph- Moultrie, T. A.; Sayi, S. T. and Timaeus, M., 2012: Birth
ic Dividend?. In: Finance and Development, 43(3). intervals, postponement and fertility decline in Afri-
Available at http://www.imf.org/external/pubs/ft/fan- ca: a new type of transition?. In: Population Studies,
dd/2006/09/basics.htm (DoA: 24/03/2017). 66(3): 241-258.
Lemaitre, G., 2007: The integration of immigrants into Notestein, F. W., 1945: Population: the long view. In:
the labour market: The case of Sweden. In: OECD So- Shulz, T. W. (Ed.), Food for de World. Chicago: Uni-
cial, Employment and Migration Working Papers, 48. versity of Chicago Press: 36-57.
Lesthaeghe, R. and Karel N., 2002: From the first to the Notestein, F. W., 1953: Economics problems of popula-
second demographic transition: An interpretation of tion change. Proceedings of the Eight Conference of
the spatial continuity of demographic innovation in
Agricultural Economists. London: Oxford Universi- security problems in Brazil. In: United Nations Expert
ty Press: 140-152. Group Meeting on Social and Economic Implications
Pew Research Center, 2015: Chapter 1: Main factors of Changing Population Age Structure. Mexico City,
driving population growth. The Pew Charitable Trust. 31 August-2 September 2005. New York: United Na-
Available at www.pewforum.org/2015/04/02/main- tions: 103-118.
factors-driving-population-growth/. Van der Kaa, D. J., 2002: The idea of a second tran-
Phang, H. S. and Mason, A., 2007: Demographic div- sition in industrialized countries. Paper presented at
idend and labour force transformation in Asia: the the Sixth Welfare Policy Seminar of the National In-
case of Republic of Korea. In: United Nations Expert stitute of Population and Social Security, Tokyo, Ja-
Group Meeting on Social and Economic Implications pan. Available at www.ipss.go.jp/web-ad/webjournal.
of Changing Population Age Structure. Mexico City, files/population/2003_4/kaa.pdf (DoA: 21/03/2017).
31 August-2 September 2005. New York: United Na- Van der Kaa, D. J., 2004: Is the second demographic
tions: 119-139. transition a useful research concept. Questions and
Reinisch, J. and White, E., 2011: The disentanglement answers. In: Vienna Yearbook of Population Research,
of population. Migration, expulsion and displacement 2: 4-10.
in post-war Europe. New York Palgrave Macmillan: Wang, F., 2007: Demographic dividend and prospect
1944-9. for economic development in China. In: United Na-
Szreter, S., 2015: Fertility, social class, gender, and the tions Expert Group Meeting on Social and Economic
professional model: Statistical explanation and histor- Implications of Changing Population Age Structure.
ical significance. In: Economic History Review, 68(2): Mexico City, 31 August-2 September 2005. New York,
707–722. DoA: 10.1111/ehr.12102. United Nations: 81-141-154.
Thompson, W. S., 1929: Population. In: American Jour- Wongboosin, K., Guest, Ph. and Prachuabmoh, V.,
nal of Sociology, 34(6): 959-975. 2005: Demographic change and the demographic div-
Timaeus, I. M. and Jasseh, M., 2004: Adult mortality in idend in Thailand. In: Asian Population Studies, 1(2):
sub-saharan Africa: evidence from demographic and 245-256.
health surveys. In: Demography, 41(4): 757-772. Zelinsky, W., 1971: The hypothesis of mobility transition.
Turra, C. M. and Queiroz, B. L., 2007: Before it´s too In: Geographical Review, 61(2): 219-249.
late: demographic transition, labour supply, and social
The proofreading of articles, positively reviewed and approved for publishing in the ‘Bulletin of Geography. Socio-economic Series’, was financed from
the funds of the Ministry of Science and Higher Education earmarked for activities popularizing science, in line with Agreement No 509/P-DUN/2016.
© 2018 Nicolaus Copernicus University. All rights reserved.

Third Demographic Transition and Demographic Dividend: An Application Based On Panel Data Analysis

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Third Demographic Transition and Demographic Dividend: An Application Based On Panel Data Analysis

Uploaded by

Copyright:

Available Formats

Bulletin of Geography. Socio–economic Series / No.

42 (2018): 59–82

BULLETIN OF GEOGRAPHY. SOCIO–ECONOMIC SERIES DE

Third demographic transition and demographic dividend:

Alejandro López González1, CDFMR, María Jesús González-González2, CFMR

© 2018 Nicolaus Copernicus University. All rights reserved.

1. Introduction literature portrays them as being parallel phenom-

es have been mentioned for the 40 to 49-year-old

We selected a number of indicators that quantify

Twenty countries belong both to the European Un-

Table 1. Dependent and independent variables

Table 2. Statistical descriptions

A. Non-transformed GAP variables:

B. Transformed GAP variables:

Table 3. Results of the regressions with non-transformed GAP variables

Table 4. Results of the regression using reciprocal transformations of GAP variables

Variable Dependent: SAL3039 (8) (9)

-0.108362* -0.0533368* -0.0630282*

© 2018 Nicolaus Copernicus University. All rights reserved.

You might also like