Você está na página 1de 25

Unfunded Pension Systems: Ageing and Variance

Qualitative Aspects of Migration


Silke Uebelmesser
Article information:
To cite this document: Silke Uebelmesser. "Qualitative Aspects of Migration" In
Unfunded Pension Systems: Ageing and Variance. Published online: 10 Mar 2015;
Downloaded by North South University At 21:59 21 November 2015 (PT)

183-206.
Permanent link to this document:
http://dx.doi.org/10.1108/S0573-8555(2004)0000264009
Downloaded on: 21 November 2015, At: 21:59 (PT)
References: this document contains references to 0 other documents.
To copy this document: permissions@emeraldinsight.com
The fulltext of this document has been downloaded 26 times since NaN*

Access to this document was granted through an Emerald subscription provided by


emerald-srm:600887 []
For Authors
If you would like to write for this, or any other Emerald publication, then please
use our Emerald for Authors service information about how to choose which
publication to write for and submission guidelines are available for all. Please visit
www.emeraldinsight.com/authors for more information.
About Emerald www.emeraldinsight.com
Emerald is a global publisher linking research and practice to the benefit of society.
The company manages a portfolio of more than 290 journals and over 2,350 books
and book series volumes, as well as providing an extensive range of online products
and additional customer resources and services.
Emerald is both COUNTER 4 and TRANSFER compliant. The organization is a
partner of the Committee on Publication Ethics (COPE) and also works with Portico
and the LOCKSS initiative for digital archive preservation.

*Related content and download information correct


at time of download.
CHAPTER 8

Qualitative Aspects of Migration


Downloaded by North South University At 21:59 21 November 2015 (PT)

Who is most likely to emigrate

Comes over one an absolute necessity to move. And what is more,


to move in some particular direction. A double necessity then: to
get on the move, and to know whither. David H. Lawrence
(1885 1930)

Abstract

Migration is low for most European countries. But migration puts nevertheless
pressure on unfunded pension systems if those who pay contributions to an
unfunded pension system leave the country. This will lead to competition for
contributors and set a race to the bottom in motion. Growing integration of the
countries of the European Union will intensify this development.
In order to shed some light on the aspect of emigration from an industrialised
country by natives, we analyse who is most likely to react to these incentives. As
statistical information is thin with respect to emigration, we present estimations
about intended emigration from Germany. We nd that young individuals with an
above-average school degree are most likely to think about emigrating. But these
individuals are exactly those who are needed as contributors to the unfunded
pension systems. Reactions of national pension systems must therefore be
expected.

As we have seen in Chapter 3, population growth is primarily dominated by


fertility rates and life expectancies. Migration does not play a very important role
for most European countries and will be very probably insufcient to provide a
solution to the ageing problem. It is, however, possible that despite the low
general level of migration, those few who migrate are enough to put pressure on
184 Unfunded Pension Systems: Ageing and Migration
the unfunded pension systems. If the wrong individuals, i.e. young individuals
with high incomes who pay contributions to the pension system, start migrating in
order to escape the rising pension burden, the nancial stability of the system
deteriorates. Even worse, the danger of an unbalanced pension budget might
induce the national pension systems to react and to engage in a competition over
contributors which might result in a race to the bottom.
It is not sufcient to observe that labour mobility is still limitedeven
though the majority of Germans, for example, is not yet mobile internationally,1
which hints at a high degree of general inertia or insufcient incentives. As Sinn
(1990, 1998) states, all that is needed for governments to feel the political
Downloaded by North South University At 21:59 21 November 2015 (PT)

pressure of factor mobility is marginal mobility.2 As soon as some people react


to different situations at home and abroad by migrating, the unfunded pension
systems are affected because they rely on a well-balanced ratio of contributors
to retirees.
In this chapter, we, therefore, analyse in detail who is most likely to react to
these incentives. We present estimations about the quality of migration from
Germany as an example of an industrialised country. By knowing which
characteristics make it more likely for an individual to think about emigrating and
by considering the projected frequency of these characteristicsnotably with
respect to the demographic developmentwe can formulate hypotheses about the
possible evolution of emigration in the coming decades.
Various studies have looked into the social and economic integration of
immigrants in countries like the US, Canada, Australia and Israel beneting from
an exhaustive collection of data. These analyses focus on who immigrates (for
example, Borjas, 1987, 1994) and on how immigrants coming from different
countries of origin and arriving at different points in time adapt to the new
environment (for example, Chiswick, 1978, and Borjas, 1994, for a survey).3
Although emigration and immigration are only two different sides of the same
coin, the migration literature is mostly about immigration. Emigration, on the
contrary, has not been much examined with the exception of emigration from
industrialised countries in the form of return migration (see DaVanzo (1983) and
Dustmann (1996) for a survey) and emigration from developing countries linked

1
See Chapter 3.
2
According to Sinn (1998), the concept of marginal mobility is similar to the marginal concept
predominant on private goods markets where it is also marginal demand that drives the competitive
behaviour of rms and not intramarginal demand.
3
Examples of empirical analyses can be found in Beggs and Chapman (1988, 1990, 1991), Chiswick
and Miller (1985), Dustmann (1993), Greenwood and McDowell (1991), Mayer and Riphahn (2000),
and Schmidt (1997).
Qualitative Aspects of Migration 185
to the brain drain problem (for example, Hamada, 1996). So far, there are not
many studies which analyse emigrationin particular by nativesfrom
industrialised countries.
One reason for this observation might be that in general, information about
emigrationin contrast to immigrationis hard to nd. The US Census Bureau
has recently developed some techniques to estimate the number of emigrants,
which underlines the difculty of obtaining reliable emigration data.4 In Germany,
emigrants are legally obliged to give notice when leaving the country. These data
collected by the German Federal Statistical Ofce thus give an initial idea about
the quantity and quality of emigration from Germany. However, the number of
Downloaded by North South University At 21:59 21 November 2015 (PT)

emigrants is very probably underreported due to registration problems. This must


be kept in mind when evaluating the plausibility of the data we use for our
estimations.
In both cases, statistical information is thin and comprises at most an
approximation of the volume without much additional information about the
characteristics of emigrants, let alone information about the destination
country. We, therefore, want to analyse emigration from Germany in detail
with a special interest in the qualitativenot quantitativeaspects and
estimate the effects of different characteristics on the probability of thinking
about emigrating. We consider detailed information at the individual,
household, and regional level which helps us to determine the characteristics
of those who intend to emigrate relative to those who want to stay in Germany.
These ndings allow us to evaluate who is most likely to emigrate in the next
few decades.
In section 8.1, we describe the data. The intention to migrate is estimated in
sections 8.2 and 8.3 concludes.

8.1. DESCRIPTION OF THE DATA

The data for this analysis stem from the German Socio-Economic Panel (GSOEP)
which consists of about 6000 households.5 We use the waves 10 from 1993, 13
from 1996, 14 from 1997, and 15 from 1998 because these waves are the only ones
where individuals are asked about their intentions to emigrate. Wave 13 in
addition comprises information about the reasons for moving. We exclude
foreigners as they are self-selected in the sense that they have already displayed

4
See Bashir and Robinson (1994) for the foreign-born population and Fernandez (1995) for the US
born population.
5
See the Appendix A5 for some additional information about the GSOEP.
186 Unfunded Pension Systems: Ageing and Migration
Table 8.1: Description of the dependent variable and sample characteristics.

Move Womena Menb

Cases Percent Cases Percent

Never 1842 18 1300 14


Probably not 3643 36 3419 36
Yes, it depends 3399 34 3735 39
Yes, very much 1111 11 1158 12
Sum 9995 100 9612 100

Mean and standard error are calculated for the values 1 never, and 4 yes, very much.
Downloaded by North South University At 21:59 21 November 2015 (PT)

Percentages may not add to 100% due to rounding. Source: Waves of 1993, 1996, 1997 and 1998
(GSOEP).
a
Mean 2.378; Std. error 0.909.
b
Mean 2.494; Std. error 0.873.

their mobility in at least one instance.6 As the gender can be expected to inuence
the propensity to move in different waysfor example, through stronger family
ties for women (Naskoteen and Zimmer, 1980) and through different educational
and professional careersthe sample will be subdivided into a female and a male
sub-sample. After excluding individuals with missing values for relevant
variablesmostly concerning the propensity to migrate and the schooling and
work history, the male sample population reduces to 9612 observations and the
female sample population to 9995 observations. Potential correlations in the error
terms due to the fact that the sample includes repeated answers from given
individuals in subsequent years are taken into account. Table A8.1 describes
sample characteristics for the variables used in the empirical analysis.

8.1.1. Intention to migrate

We use the stated propensity to move to another country (move) as the


dependent variable. Table 8.1 presents sample characteristics for this variable
with the response categories never, probably not, yes, it depends, and yes,
very much. Men and women show a very high willingness to emigrate. We see
that 51% of the men and 45% of the women are thinking about emigrating
strongly (yes, very much) or at least to some extent (yes, it depends). We will
comment below on the order of magnitude of these responses given the quite low
numbers of individuals who actually emigrate.

6
For an analysis of the migration propensity based on a sample of Germans and foreigners see
Uebelmesser (2004).
Qualitative Aspects of Migration 187
Table 8.2: Reasons of those with a positive propensity to migrate.

Reasons to migrate Women Men

Cases Percent Cases Percent

Better job 269 24 522 42


Retirement 176 16 190 15
Training/education 83 8 115 9
Family 140 13 34 3
Other reason 367 33 314 25
No answer 70 6 67 5
Sum 1105 100 1242 100
Downloaded by North South University At 21:59 21 November 2015 (PT)

move yes, it depends or move yes, very much. Percentages may not add to 100% due to
rounding. Source: Waves of 1993 (GSOEP).

Table 8.2 gives an overview of the reasons for migration stated by those who
show a positive propensity to migrate.7 Better professional opportunities play an
important role for 42% of the men and for 24% of the women. This hints at
relevant differences in wages and/or employment probabilities in the destination
country relative to Germany. However, reasons which are not directly linked to
economic differences prevail. Sixteen percent of the women and 15% of the men
want to spend their retirement period abroad. The motivation to migrate is thus not
related to wage or employment differentials, although the general economic
situation in the destination country is important to judge the purchasing power of
the pension benets abroad.8
Better institutions for training and education are the reason given by 8% of the
women and 9% of the men with a positive propensity to migrate. Again, the
economic situation only plays a role in an indirect way when assuming that these
individuals hope to give themselves better job opportunities abroador at
homeafter having completed their studies abroad.9 For women, family reasons
are of importance in 13% of the cases compared to 3% of the cases for men. Here,
the economic situation indirectly inuences the decision to migrate when thinking
that those individuals may follow their partner or other family members who have

7
A more extensive discussion of the economic and non-economic reasons of the potential migrants can
be found in Uebelmesser (2004).
8
Note, however, that migration after retirement does not affect an unfunded pension system.
9
We do not have information about the intention of individuals to return after having completed their
education. See Dustman (1995, 1997) for an analysis of the long-run effects of return migrants on the
welfare state.
188 Unfunded Pension Systems: Ageing and Migration
migrated earlierpossibly for economic reasons. Thus, in general, the economic
situation in the destination country compared to Germany plays some role for the
propensity to migrateeither directly or indirectly.
Tables 8.3 and 8.4 show the breakdown of the responses for women and men
according to a number of personal, household and regional characteristics.
Various patterns are immediately apparent.

Table 8.3: Intention to emigrate (women).

Variable Yes, very much Yes, it depends Probably not Never Total sample
Downloaded by North South University At 21:59 21 November 2015 (PT)

Individual data
Age ,20 95 (24) 152 (38) 108 (27) 41 (10) 396
Age 2029 372 (15) 940 (37) 942 (37) 297 (12) 2551
Age 3039 305 (10) 1040 (36) 1149 (40) 414 (14) 2908
Age 4049 175 (9) 728 (36) 792 (40) 301 (15) 1996
Age 5059 127 (10) 384 (31) 417 (34) 298 (24) 1226
Age 60 37 (4) 155 (17) 235 (26) 491 (53) 918
Married 508 (9) 2985 (33) 2333 (39) 1134 (19) 5960
Foreign partner 10 (11) 53 (59) 22 (24) 5 (6) 90
German partner 498 (8) 1932 (33) 2311 (39) 1129 (19) 5870
Not married 603 (15) 1414 (35) 1310 (32) 708 (18) 4035
Children 0 6 327 (10) 1129 (35) 1216 (38) 513 (16) 3185
Children 7 16 499 (10) 1770 (36) 1952 (39) 731 (15) 4952
No children under 16 697 (12) 1940 (34) 1901 (33) 1146 (20) 5684
West Germany 908 (13) 2558 (38) 2273 (34) 1030 (15) 6769
East Germany 203 (6) 841 (26) 1370 (42) 812 (25) 3226
Education
Elementary 304 (9) 898 (28) 1063 (33) 960 (30) 3225
Secondary 460 (10) 1534 (34) 1813 (40) 711 (16) 4518
Higher secondary 347 (15) 967 (45) 767 (34) 171 (8) 2252
University degree 154 (14) 490 (44) 383 (35) 81 (7) 1108
Occupational training 695 (10) 2301 (33) 2716 (38) 1350 (19) 7062
Occupation
Worker 104 (11) 271 (29) 353 (38) 203 (22) 931
Self-employed 44 (13) 145 (43) 106 (32) 41 (12) 336
Trainee 77 (17) 186 (41) 141 (31) 48 (11) 452
Employee 457 (11) 1472 (37) 1593 (40) 459 (12) 3981
Civil servant 35 (11) 140 (45) 113 (37) 22 (7) 313
Unemployed 101 (11) 257 (27) 353 (38) 226 (24) 937
Income
Yes 1053 (12) 3167 (36) 3352 (38) 1276 (14) 8848
Low net income 275 (13) 646 (31) 776 (38) 357 (17) 2054
Middle net income 463 (10) 1524 (34) 1731 (39) 741 (17) 4459
High net income 315 (13) 997 (43) 846 (36) 178 (8) 2335
No (retired) 58 (5) 232 (20) 291 (25) 566 (49) 1147

Values in parenthesis are percentages. Source: Waves of 1993, 1996, 1997 and 1998 (GSOEP).
Qualitative Aspects of Migration 189
Table 8.4: Intention to emigrate (men).

Variable Yes, very much Yes, it depends Probably not Never Total sample

Individual data
Age ,20 70 (17) 132 (33) 154 (38) 49 (12) 405
Age 2029 395 (16) 1074 (44) 782 (32) 184 (8) 2435
Age 3039 348 (12) 1166 (41) 1027 (36) 316 (11) 2857
Age 4049 170 (9) 731 (39) 750 (40) 231 (12) 1882
Age 5059 131 (10) 465 (35) 477 (36) 240 (18) 1313
Age 60 44 (6) 167 (23) 229 (32) 280 (39) 720
Married 528 (9) 2024 (36) 2110 (38) 899 (16) 5561
Foreign partner 16 (26) 33 (53) 11 (18) 2 (3) 62
Downloaded by North South University At 21:59 21 November 2015 (PT)

German partner 512 (9) 1991 (36) 2099 (38) 897 (16) 5499
Not married 630 (16) 1711 (42) 1309 (32) 401 (10) 4051
Children 0 6 327 (10) 1129 (35) 1216 (38) 513 (16) 3185
Children 7 16 489 (12) 1567 (39) 1511 (37) 499 (12) 4952
No children under 16 748 (13) 2325 (40) 1994 (34) 819 (14) 5886
West Germany 927 (14) 2715 (42) 2127 (33) 737 (11) 6506
East Germany 231 (7) 1020 (33) 1292 (42) 563 (18) 3106
Education
Elementary 381 (11) 1082 (33) 1146 (35) 712 (21) 3321
Secondary 371 (11) 1248 (37) 1312 (39) 400 (12) 3331
Higher secondary 406 (14) 1405 (47) 961 (32) 188 (6) 2960
University degree 221 (12) 859 (47) 601 (33) 132 (7) 1813
Occupational training 758 (11) 2507 (36) 2589 (38) 1046 (15) 6900
Occupation
Worker 269 (10) 886 (34) 1021 (39) 458 (17) 2634
Self-employed 99 (16) 259 (43) 196 (33) 49 (8) 603
Trainee 75 (15) 195 (38) 188 (37) 55 (11) 513
Employee 344 (12) 1331 (45) 1082 (36) 213 (7) 2970
Civil servant 85 (12) 287 (42) 241 (35) 71 (10) 684
Unemployed 85 (11) 253 (33) 272 (36) 152 (20) 762
Income
Yes 1078 (12) 3503 (40) 3135 (36) 990 (11) 8706
Low net income 225 (10) 782 (35) 885 (40) 322 (15) 2214
Middle net income 563 (13) 1672 (40) 1484 (35) 469 (11) 4188
High net income 290 (13) 1049 (46) 766 (32) 199 (9) 2304
No (retired) 80 (9) 232 (26) 284 (31) 310 (34) 906

Values in parenthesis are percentages. Source: Waves of 1993, 1996, 1997 and 1998 (GSOEP).

Men in general are more likely to consider emigrating than women and younger
individuals think more often about leaving Germany than older individualswith
the exception of men under 20.
A higher school degree makes thinking about emigrating more likely for men
and women. Occupational training and a university degree, however, do not
190 Unfunded Pension Systems: Ageing and Migration
further increase the probability compared to a secondary degree or a higher
secondary degree, respectively. Concerning occupation as an important factor, we
nd for men and for women that employed people in general are more likely to
consider emigrating than the unemployed and retired. For the specic forms of
occupations considered explicitly, probabilities are very high for most of the
occupationsincluding self-employed and civil servants. In addition, the
probability to think about emigrating increases with income levels.10
Single men and women are more likely to consider emigrating than those who
are married to a German partner, while those with a foreign partner show a higher
propensity to emigrate. Children, however, do not play an important role in the
Downloaded by North South University At 21:59 21 November 2015 (PT)

decision. Concerning differences between East and West Germany, individuals


living in West Germany are more likely to emigrate than individuals living in the
Eastern part of Germany.

8.1.2. Discussion of the intention variable

Given the few actual emigrants in the GSOEP, and given the fact that in general
information about emigrationin contrast to immigrationis hard to nd, the
variable on the intention to move allows an approximation of the underlying data.
As Manski (1990, p. 935) states, intentions data do potentially convey
information about behaviour.
To get a feeling for the reliability of the data of the GSOEP, we compare them
with similar data from the study on Performance of the European Union Labour
Market by the European Commission (1995). In this study, individuals are asked
whether they would be willing to work in an EC member state different from the
one of which they are a national (Table 8.5). Thirty-four percent of the men and
21% of the women gave a positive answer.
This allows us to compare the results of the two data sets at least indirectly. If
we only consider data from the 10th wave (1993) of the GSOEP with information
about the reasons to move, we nd that 22% of the men and 11% of the women
name better job opportunities abroad as a possible reason to think about
emigrating.11 In both data sets, men are more willing to migrate for professional
reasons than women. It is not surprising that the numbers from the GSOEP are
smaller than those in the study of the European Commission (1995), given the fact

10
For those individuals without any (information about) net wage income, we have simulated net wage
income using the Heckman procedure (1979).
11
Ratio of those who name better job opportunities as the reason to move (Table 8.2) to all
individuals (2489 women and 2420 men) of wave 1993.
Qualitative Aspects of Migration 191
Table 8.5: Willingness to work in a foreign EC State.

Variable Yes (%) No (%)

Total 28 72
Of which
Men 34 66
Women 21 79

Source: European Commission (1995).


Downloaded by North South University At 21:59 21 November 2015 (PT)

that in the GSOEP individuals have to choose the most likely reason among
several reasons so that other reasons might crowd out work-related reasons.12
Taking this into account, one can state that the order of magnitude is comparable.
It is, however, still necessary to give an explanation as to how one can reconcile
the different orders of magnitude of the statistical information of the German
Federal Statistical Ofce (see Chapter 3) and the responses in the GSOEP and the
survey of the European Commission. For this, we break down the responses into
age groups. From the GSOEP data (Tables 8.3 and 8.4), we see that the propensity
to emigrate (move yes, very much or move yes, it depends)
decreases with age for men and women, with few exceptions. For women, it is
62% for those under 20 and decreases for the 20 to 59-year-old from 52 to 41%.
Women aged 60 and over are less likely to consider emigrating (21%). The pattern
is very similar for men although the level of those who consider emigrating
is higher compared to women except for men under 20 (50%). For the 20 to
59-year-old, the level decreases from 60 to 45% and drops to 29% for those aged
60 and over.13
The observations from the German Federal Statistical Ofce (Statistisches
Bundesamt, 2000) are quantitatively different but not qualitatively. The absolute
number of emigrants is much lower than what one would expect from the answers
to the willingness-to-migrate question in both surveys. In 1993, only 86,619
Germans emigrated and in 1996, 1997 and 1998, the volumes were only slightly
higher with 118,430, 109,903 and 116,403 emigrants, respectively. If we
breakdown the data by age groups (Table 8.6), we nd, however, a similar
prole for intended and real emigration. The data from the German Federal
Statistical Ofce show a relative increase in emigration up to the age of 25 30

12
See Table 8.2 for the other reasons.
13
The study of the European Commission (1995) conrms this general trend. The willingness to work
abroad is highest for those below 31 years with 39% and decreases to 27% for the 3149-year-old and
to 15% for the 50-year-old and older individuals.
192
Downloaded by North South University At 21:59 21 November 2015 (PT)

Unfunded Pension Systems: Ageing and Migration


Table 8.6: Emigration from Germany.

Age groups 1993 1996 1997 1998

Absolute Per 1000 of Absolute Per 1000 of Absolute Per 1000 of the Absolute Per 1000 of
number the age group number the age group number age group number the age group

, 18 20,260 1.3 25,312 1.6 22,426 1.4 22,443 1.4


1825 9900 1.4 12,545 2.0 11,493 1.8 12,435 2.0
2530 12,984 1.8 15,521 2.4 12,124 2.3 14,845 2.7
3150 30,835 1.3 46,359 1.8 44,472 1.7 48,600 1.9
51 12,640 0.5 18,693 0.7 17,388 0.6 18,080 0.6
Total 86,619 1.1 118,430 1.4 109,903 1.3 116,403 1.4

Source: Statistisches Bundesamt (2000).


Qualitative Aspects of Migration 193
and a decline thereafter with a sharp drop for those aged 51 years and older. This is
exactly reected by the answers of men whose propensity to emigrate also peaks
for the age group 20 29 and falls markedly for those 60 years and older, while it is
closely mirrored by the answers of women, whose willingness declines over all
age groups.
For the following analysis, we thus follow Burda et al. (1998) in assuming that
intentions are a monotonic function of the underlying driving variables which
motivate migration. We will, therefore, concentrate on identifying those
characteristics which make it more likely for an individual to think about
emigrating. We will interpret the results of the estimation accordingly, namely
Downloaded by North South University At 21:59 21 November 2015 (PT)

that individuals with these characteristics will in fact be over-proportionally


represented among the future emigrants.

8.2. ESTIMATION OF THE INTENTION TO MIGRATE

According to the standard human capital model,14 the mobility decision of an


individual is guided by the comparison of the present value of lifetime earnings
labour income and pension benetsin the home country and in the foreign
country net of migration costs for migration at a certain age. As with all other
decisions, the individual chooses the alternative that maximises the utility: in this
case the utility of lifetime earnings. Thus, within this framework, migration occurs
when the utility with migration exceeds the utility without migration. The human
capital model thus suggests comparing the economic situation in the source
and in the destination country by considering the monetary and non-monetary
migration costs.
This modelling has, however, two shortcomings in our context. First, it neglects
any reasons which are not earnings-related, but which play an important role when
thinking about migration as illustrated in Table 8.2. Second, it requires that the
destination country with its specic characteristics is known. In general, however,
information about the volume of emigration in general and about characteristics of
the destination country in particular is mostly lacking.15
Our approach alleviates both problems. We assume that the emigration decision
is a function of individual characteristics, characteristics of the household, and
characteristics of the (home) region. We thus include economic but also
14
See Sjaastad (1962) for an early version of this model.
15
In Germany, emigrants are legally obliged to give notice when leaving the country. However, the
number of emigrants is probably underreported due to registration problems and information about the
destination country is very limited.
194 Unfunded Pension Systems: Ageing and Migration
non-economic factors which can be important for the (potential) migration
decision and aim at identifying their effects. In addition, we abstract from
variables concerning the destination countries for the analysis which can be
justied as we are only interested in the attitude towards migration and not in the
probability of migrating to a specic country. It is reasonable to assume that there
is at least one country for individuals with a positive propensity to migrate for
which the utility exceeds the utility without migration.
Following this approach, we now focus on a systematic analysis of the effect
of each independent variable on the dependent variable. We implicitly assume
that the decision to emigrate can be approximated by these variables.16 As the
Downloaded by North South University At 21:59 21 November 2015 (PT)

dependent variable, we use the reported propensity to move to another country


which can be viewed as an ordered response with four categories: never,
probably not, yes, it depends and yes, very much. The statistical model for
categorical data is an ordered model, for example, an ordered Probit model.17
The interpretation of the estimated coefcients in an ordered Probit model is not
straightforward as the value and the sign can differ from those of the marginal
effects. Therefore, the interpretation focuses on the statistical signicance of
the coefcients and on the marginal effects derived from simulations. The estima-
tion results of the ordered Probit model for the male and female sub-sample are
given in Tables B8.1 and B8.2, with column 1 presenting results of the basic
conguration and columns 2 and 3 adding measures of the professional situation
and the environment, respectively.
Simulations are carried out for the whole sample such that rst the baseline
probabilities for each outcome are predicted on the basis of the estimated
coefcients (Tables B8.1 and B8.2). Then single variables are xed at two different
values and the probabilities are predicted for each of the two values. The
continuous variables age and unemployment rate are rst set at the average
value and then at the average value plus one standard deviation. Dummy variables,
for example, residence in West Germany, are set at 0 and at 1, respectively. The
difference in the predicted probability for each variable set at two different values
is then divided by the baseline probabilities. Tables 8.7 and 8.8 display changes of
the propensity to migrate measured in percent of the baseline probabilities, thus
making the relative impact of a variable comparable across different outcomes and
across sub-samples.
The results of the simulation conrm mostly what one would expect. If we
have a human capital theory a la Sjaastad (1962) in mind, we would expect

16
Burda et al. (1998) follow a similar approach in their analysis of the intention to migrate from East to
West Germany.
17
See Appendix C8 for a short description of the probit model.
Qualitative Aspects of Migration 195
Table 8.7: Simulation results (women).

Variable Yes, very much Yes, it depends Probably not Never

Baseline probability 0.122 0.340 0.359 0.178


Individual data
Age: 38 vs. 52* 23.180 20.486 0.057 1.705
Residence: West German vs. East 0.779 0.302 20.140 20.755
German***
Married to German partner vs. 20.116 20.042 0.019 0.082
not married**
Married to foreign partner vs. 0.507 0.119 20.086 20.214
not married***
Downloaded by North South University At 21:59 21 November 2015 (PT)

Children 0 6 vs. none*** 20.412 20.172 0.082 0.429


Children 7 16 vs. none 20.064 20.024 0.013 0.056
Education
Secondary vs. elementary*** 0.339 0.140 20.065 20.343
Higher secondary vs. elementary*** 0.771 0.263 20.150 20.612
University degree vs. none** 0.044 0.010 20.020 20.052
Occupational training vs. none 20.039 20.014 0.008 0.033
Occupation
Unemployment rate: 12% vs. 16% 20.061 20.023 20.013 0.054
Worker vs. not 20.026 20.010 0.005 0.025
Self-employed vs. not** 0.386 0.125 20.079 20.279
Trainee vs. not 0.072 0.027 20.015 20.063
Employee vs. not** 0.143 0.052 20.029 20.120
Civil servant vs. not 20.119 20.050 0.024 0.121
Unemployed vs. not 0.012 0.004 20.002 20.010
Income
Middle net income vs. low 20.059 20.020 0.011 0.042
net income
High net income vs. low 0.006 0.002 20.001 20.004
net income
None (retired) vs. low net income*** 20.596 20.266 0.103 0.652

Reference categories are elementary degree, East German residence, not married and low net income.
***, ** and * denote statistical signicance at the 1, 5 and 10% level, respectively (see Table B8.1
most extensive estimation).

the propensity to move to decrease with age as the shorter payoff period of the
human capital investment decreases the net gains of migrationat least if
migration is considered for economic reasons.18 The coefcients for the age
variables are signicant at the 10% level in the female sub-sample but not in

18
But the human capital theory cannot explain migration after retirement. For those individuals who
think about emigrating in order to spend their years as retirees abroad, the willingness to migrate is
supposed to increase with age.
196 Unfunded Pension Systems: Ageing and Migration
Table 8.8: Simulation results (men).

Variable Yes, very much Yes, it depends Probably not Never

Baseline probability 0.130 0.388 0.353 0.129


Individual Data
Age: 38 vs. 52 20.433 20.153 0.116 0.519
Residence: West German vs. East 0.670 0.227 20.195 20.759
German***
Married to German partner vs. 20.113 20.037 0.025 0.088
not married**
Married to foreign partner vs. 0.966 0.140 20.183 20.328
not married***
Downloaded by North South University At 21:59 21 November 2015 (PT)

Children 0 6 vs. none*** 20.226 20.078 0.070 0.247


Children 7 16 vs. none 20.074 20.024 0.023 0.074
Education
Secondary vs. elementary*** 0.255 0.088 20.078 20.285
Higher secondary 0.467 0.144 20.141 20.455
vs. elementary***
University degree vs. none 0.016 0.003 20.008 20.020
Occupational training vs. none 20.054 20.017 0.016 0.051
Occupation
Unemployment rate: 12 vs. 16% 0.046 0.014 20.014 20.043
Worker vs. not*** 20.336 20.108 0.103 0.347
Self-employed vs. not 0.249 0.056 20.071 20.172
Trainee vs. not*** 20.334 20.107 0.103 0.344
Employee vs. not 20.050 20.013 0.015 0.041
Civil servant vs. not* 20.239 20.072 0.073 0.228
Unemployed vs. not** 20.244 20.087 0.076 0.278
Income
Middle net income vs. low 0.072 0.024 20.022 20.073
net income
High net income vs. low 0.183 0.056 20.056 20.172
net income
None (retired) vs. low net 20.053 20.019 0.016 0.059
income

Reference categories are elementary degree, East German residence, not married and low net income.
***, ** and * denote statistical signicance at the 1, 5 and 10% level, respectively (see Table
B8.1most extensive estimation).

the male sub-sample. As to the marginal effect, an increase in the age of the
individual increases the probability of staying in Germany and decreases the
probability of migrating. Compared to a 38-year-old, the alternative never
becomes 171% more likely for a 52-year-old woman and 52% more likely for
a 52-year-old man whereas the probability for the alternative yes, very much
Qualitative Aspects of Migration 197
decreases by 318% for women and 43% for men.19 The rather low signicance
of the effects hint at other potential reasons for thinking about migrating which
are not captured by the human capital theory, e.g. joining friends and family
members or emigrating in order to spend the years as retirees abroad.
As we focus on emigration from an industrialised country, the propensity
to migrate should increase with the years of education and training (Borjas,
1996). First of all, the geographic region which makes up the relevant labour
market is larger for highly educated individuals than for less educated ones.
Second, highly educated individuals might be more efcient at learning about
employment opportunities in alternative labour markets which reduces
Downloaded by North South University At 21:59 21 November 2015 (PT)

migration costs. Last but not the least, higher education implies better
knowledge of foreign languages which is an essential prerequisite for economic
and social integration.20 The extent to which human capital is transferable from
the home country to the destination country depends in many cases on general
communication skills. The simulation shows that the signicant school and
university variables have effects in the expected direction. Holding a higher
secondary degree increases the probability for the yes, very much alternative
by 77% (47%) in the female (male) sub-sample and decreases the probability
for the never alternative by 61% (45%) compared to an elementary degree.
Holding a university degree has a small signicant effect on the propensity to
migrate for women, but does not have a signicant effect for men.
Occupational training, however, is without any signicant inuence in both
sub-samples.
The occupational situation plays a signicant role for a self-employed and
for an employee in the female sub-sample and for a worker, a trainee, a
civil servant, an unemployed and a self-employedalbeit at the 11%
signicance levelin the male sub-sample. Being self-employed positively
inuences the probability to migrate for women and men. It seems, therefore, that
the entrepreneurial spirit of the self-employed outweighs the counter-arguments
brought forward by Naskoteen and Zimmer (1980).21 For civil servants and
unemployed individuals in the male sub-sample, the effect is signicant and
negative, which implies that both groups are less likely to migrate in order not to

19
This huge effect is partly caused by the fact that a change in age by 14 years (one standard deviation)
is considered.
20
See, for example, for the relevance of language skills for social integration Chiswick and Miller
(1995) and for economic integration Dustmann (1994).
21
Naskoteen and Zimmer (1980) argue that self-employment should lead to a smaller propensity to
move as the self-employed are less susceptible to promotion opportunities.
198 Unfunded Pension Systems: Ageing and Migration
lose the safe job at the government or the claims of the unemployment insurance,
respectively. Apparently, unemployed individuals are afraid to forego their
insurance claims, although it might be worthwhile to consider migrating if the
probability of nding a suitable job abroad, i.e. the expected income abroad, is
sufciently high.
The level of the wage income does not signicantly inuence the propensity to
emigrate, the only exception being no income (retired) in the female sub-sample,
which has a negative effect.
As to the private environment, the partner variable for those who are married
to a German should have a negative impact on the propensity to move abroad as
Downloaded by North South University At 21:59 21 November 2015 (PT)

it is both partners together or the family as a whole who must gain by


migrating.22 Moving with the partner or the familyespecially when there are
childreninduces higher migration costs as all members of the family incur
monetary and non-monetary costs when trying to adapt to a foreign
environment. In contrast, those with a foreign partner should show a higher
propensity to emigrate, which implies that for those couples the migration costs
are lower. We nd that being married signicantly inuences the propensity to
migrate in both sub-samples. With a German partner, women (men) are 12%
(11%) less likely to think about migration very much compared to being not
married. A foreign partner, however, increases the probability to consider
migrating very much by 51% (97%) for women (men). This shows that
moving with a partner makes it more difcult and, therefore, less likely than
moving aloneat least when the partner is German; but this conclusion is
reversed when the partner is foreign.
Children in the household have a signicant effect in the expected direction in
both sub-samplesthough this effect is almost twice as large for small children of
0 6 years in the female sub-sample. This shows that the mobility of women is
more affected by family ties.
What is quite surprising at rst sight is the signicance of living in the
western part of Germany in both sub-samples and the impact this variable has
on the propensity to migrate. If an individual lives in West Germany, the
probability for the alternative yes, very much will increase by 78% for
women and by 67% for men, whereas the probability for the alternative never
decreases by 76% for women and for men. One explanation for this
phenomenon could be that more mobile individuals from East Germany have
already migrated either to the West or to a foreign country orto put it
differentlythat there is a selection bias regarding the mobility of individuals

22
See Mincer (1978) for an analysis of migration decisions of families.
Qualitative Aspects of Migration 199
who still live in East Germany. The state unemployment rate, however, has no
signicant effect.

8.3. CONCLUSION

In order to shed some light on a so far rather neglected aspect of migration, namely
emigration from an industrialised country by natives, we determine the
characteristics of an individual and his or her environment which positively or
negatively inuence the propensity to migrate. It is important to know more about
Downloaded by North South University At 21:59 21 November 2015 (PT)

the qualitative aspects of emigration, especially with respect to the growing


integration of the countries of the European Union and the consequences of
migration for unfunded pension systems.
We have seen in Chapter 3 that net migration is not an important phenomenon
for countries of the European Union and is not expected to increase over the next
50 years. There might be more intra-EU migration at the expense of immigration
from non-EU countries but without a signicant increase in total migration
volumes. It is, however, possible that the low level of migration is sufcient to
put pressure on the unfunded pension systems in the EU countries if those
individuals who are mostly needed to carry the pension burden start emigrating.
We have, therefore, focused the analysis on providing some indications about the
qualitative aspects of emigration from Germany, i.e. who is most likely to
emigrate.
What the analysis has shown is that older men and women are less likely to
migrate. In the light of an older society, this effect would reduce the total mobility
of Germans in the future. However, another result of the analysis is that children in
the household reduce the propensity to migrate. Fewer children in the future
would thus lead to a more mobile society. Besides, a rise in the educational levels
for more recent cohortssomething which we have not explicitly dealt with in our
estimationswould also increase mobility. Which effect dominates is a priori
difcult to say. What can, however, be derived from the analysis concerns the
characteristics of the probable emigrants: young and with an above-average
school level. But these individuals are exactly those who are needed as
contributors for the unfunded pension systems.
There is the argument that if the young and well educated individuals in other
European countries show comparable migration behaviour we will merely
observe an exchange of mobile individuals with similar characteristics. This
argument, however, is only true if there is no competition for migrants between
countries. As the young and well-educated are particularly mobile and are
200 Unfunded Pension Systems: Ageing and Migration
particularly needed in all countries as contributors, there will be much competition
over precisely these individuals.
Thus, even though the effect of ageing on total mobility is ambiguous, it is quite
clear that emigration of the young and well educated will exert pressure on
the German pension system and in a similar way on pension systems in other EU
countries if it is not possibleas legally prohibited for migration within the
European Unionto discriminate between immigrants and the local population.
As Sinn (1998, p. 117) has stated, marginal mobility and non-discrimination are
sufcient to produce strong competitive pressure on the institutions concerned.
There is always the illusion of being able to attract more young and well-educated
Downloaded by North South University At 21:59 21 November 2015 (PT)

individuals by scaling down the pension system and reducing thus in a twofold
way the burden for contributors, i.e. through a reduced system and more
contributors, without hurting the pensioners too much, thanks to the increased
number of contributors. And again, immigration and emigration are just different
sides of the same coin as the attempt to attract young and highly skilled
immigrants is identical to the ght to hold back the potential emigrantsboth
leading to a likely race to the bottom of inter- and intragenerationally
redistributive activities.23
In Chapter 9, we will focus on the possible competition for mobile individuals
among European countries which might be intensied by the growing integration
within the European Union. The effect of migration on the nancial sustainability
of unfunded pension systems depends on the distribution of competencies at the
national and European level. We will, therefore, analyse in detail intra-European
migration with respect to efciency and distribution against the background of
European regulations.

APPENDIX A8. DESCRIPTIVE STATISTICS

See Table A8.1.

APPENDIX B8. PROBIT ESTIMATION

See Tables B8.1 and B8.2.

23
Immigrants from non-European countries may react even more strongly to differences in the
economic conditions in individual European countries. Their (differential) mobility is very high as they
need to choose only the destination country given that they have already decided to move (Sinn, 1998).
Qualitative Aspects of Migration 201
Table A8.1: Descriptive statistics.

Variable Dummy Women Men

Mean SD Mean SD

Individual data
Age 38.842 14.099 38.323 13.408
Residence: West Germany 0.677 0.468 0.677 0.468
Married 0.596 0.491 0.579 0.494
Married to German partner 0.587 0.492 0.572 0.495
Married to foreign partner 0.009 0.094 0.006 0.080
Children 0 6 0.170 0.375 0.155 0.362
Downloaded by North South University At 21:59 21 November 2015 (PT)

Children 7 16 0.267 0.442 0.238 0.426


Educationa
Elementary degreeb 0.323 0.468 0.346 0.476
Secondary degree 0.452 0.498 0.347 0.476
Higher secondary degree 0.225 0.418 0.308 0.462
University degree 0.111 0.314 0.189 0.391
Occupational training 0.707 0.455 0.718 0.450
Occupation
Worker 0.093 0.291 0.274 0.446
Self-employed 0.034 0.180 0.063 0.242
Trainee 0.045 0.208 0.053 0.225
Employee 0.398 0.490 0.309 0.462
Civil servant 0.031 0.174 0.071 0.257
Unemployed 0.094 0.291 0.079 0.270
Incomec
Low net incomeb 0.206 0.404 0.230 0.421
Middle net income 0.446 0.497 0.436 0.496
High net income 0.234 0.423 0.240 0.427
None (retired) 0.115 0.319 0.094 0.292
Unemployment rate (State level) 12.461 4.223 12.475 4.235
No. of observations 9995 9612

Source: Waves of 1993, 1996, 1997 and 1998 of the GSOEPexcept for Statistisches Bundesamt
(1994, 1999a) for the unemployment rate.
a
Variables indicating the highest degree obtained by the individualscorresponding in the German
system to Hauptschule, Realschule and Fachoberschule/Gymnasium, respectively.
b
Omitted in the estimation to avoid multicollinearity.
c
Low net income referring to the rst quartile of the wage distribution and high net income to the fourth
quartile.
202
Downloaded by North South University At 21:59 21 November 2015 (PT)

Unfunded Pension Systems: Ageing and Migration


Table B8.1: Parameter estimates (women).

Variable Coefcient Std. error Coefcient Std. error Coefcient Std. error

Age 20.080*** 0.026 20.096** 0.027 20.051* 0.027


Age2 0.001 0.001 0.002** 0.001 0.001* 0.001
Age3 20.000*** 0.000 20.000** 0.000 0.000** 0.000
Education
Secondary degree 0.207*** 0.040 0.199** 0.040 0.197*** 0.041
Higher secondary degree 0.427*** 0.052 0.418** 0.053 0.387*** 0.061
University degree 0.141** 0.062 0.126* 0.063 0.128** 0.064
Occupational training 20.009 0.041 20.028 0.042 20.021 0.042
Occupation
Unemployment rate (state level) 20.051*** 0.004 20.008 0.008 20.009 0.008
Unemployed 0.106* 0.048 0.006 0.049
Trainee 0.152* 0.069 0.040 0.070
Worker 0.106* 0.053 20.015 0.055
Self-employed 0.319** 0.077 0.194** 0.080
Civil servant 0.075 0.079 20.073 0.083
Employee 0.215** 0.036 0.078** 0.038
Individual data
West German residency 0.443** 0.074 0.430*** 0.076
Married to foreign partner 0.321*** 0.122
Married to German partner 20.098** 0.038
Children 0 6 20.256*** 0.041
Downloaded by North South University At 21:59 21 November 2015 (PT)

Children 7 16 20.036 0.033


Income
Retired 20.362*** 0.085
Middle net income 20.028 0.048
High net income 0.003 0.071

Qualitative Aspects of Migration


l1 : threshold for probit 22.788 0.222 22.077 0.255 21.046 0.283
l2 : threshold for probit 21.673 0.221 20.952 0.255 20.563 0.283
l3 : threshold for probit 20.514 0.221 0.216 0.255 0.612 0.283

Pseudo R2 0.051 0.057 0.062

Log-likelihood 212,235.948 212,164.948 212,101.354

***, ** and * denote statistical signicance at the 1, 5 and 10% level, respectively. Reference categories are elementary degree, East German
residence, not married and low net income.

203
204
Downloaded by North South University At 21:59 21 November 2015 (PT)

Unfunded Pension Systems: Ageing and Migration


Table B8.2: Parameter estimates (men).

Variable Coefcient Std. error Coefcient Std. error Coefcient Std. error

Age 20.037 0.025 20.053* 0.026 20.019 0.027


Age2 0.001 0.001 0.001 0.001 0.000 0.001
Age3 0.000* 0.000 20.000* 0.000 20.000 0.000
Education
Secondary degree 0.168*** 0.040 0.159** 0.041 0.153** 0.041
Higher secondary degree 0.354*** 0.048 0.296** 0.050 0.261** 0.052
University degree 0.054 0.052 0.048 0.054 0.035 0.057
Occupational training 20.050 0.044 20.041 0.045 20.031 0.045
Occupation
Unemployment rate (state level) 20.039*** 0.004 0.007 0.008 0.006 0.008
Unemployed 20.111 0.063 20.153* 0.064
Trainee 20.210** 0.071 20.196** 0.075
Worker 20.169** 0.054 20.197** 0.055
Self-employed 0.146* 0.074 0.120 0.075
Civil servant 20.093 0.072 20.135 0.077
Employee 0.003 0.053 20.026 0.055
Individual data
West German residency 0.435** 0.074 0.393** 0.082
Married to foreign partner 0.559** 0.150
Married to German partner 20.099* 0.041
Children 0 6 20.139** 0.041
Downloaded by North South University At 21:59 21 November 2015 (PT)

Children 7 16 20.044 0.033


Income
Retired 20.032 0.098
Middle net income 0.042 0.054
High net income 0.103 0.081

Qualitative Aspects of Migration


l1 : threshold for probit 22.175 0.236 21.755 0.275 21.380 0.280
l2 : threshold for probit 21.023 0.236 20.594 0.275 20.214 0.280
l3 : threshold for probit 0.218 0.236 0.659 0.275 1.045 0.280

Pseudo R2 0.037 0.042 0.045

Log-likelihood 211,678.270 211,607.723 211,572.045

***, ** and * denote statistical signicance at the 1, 5 and 10% level, respectively. Reference categories are elementary degree, East German
residence, not married and low net income.

205
206 Unfunded Pension Systems: Ageing and Migration

APPENDIX C8. PROBIT MODEL

Let ypij be a continuous, latent variable which represents the attitude towards
emigration of individual i in the region j: Assume that ypij is a linear function of Xij
and Zj ; parameters b and g; and a stochastic term uij such that
ypij X 0ij b Z 0j g uij C8:1
The unobserved variable ypij is represented by a variable yij which is related to the
four categories in the following way
8
> 1; iff ypij # l1
Downloaded by North South University At 21:59 21 November 2015 (PT)

>
>
>
>
< 2; iff l1 , ypij # l2
yij C8:2
>
> 3; iff l2 , ypij # l3
>
>
>
:
4; iff ypij . l3 ;
where ll ; l 1; 2; 3; are unobservable thresholds to be estimated. With the
standard normal distribution F; the probabilities for an individual to be part of the
four categories are given by
Pry 1 Fl1 2 X 0 b 2 Z 0 g;

Pry 2 Fl2 2 X 0 b 2 Z 0 g 2 Fl1 2 X 0 b 2 Z 0 g;


C8:3
Pry 3 Fl3 2 X 0 b 2 Z 0 g 2 Fl2 2 X 0 b 2 Z 0 g;

Pry 4 1 2 Fl3 2 X 0 b 2 Z 0 g:
For all probabilities to be positive, we need
l1 , l2 , l3 : C8:4

Você também pode gostar