DEMOGRAPHIC RESEARCH
VOLUME 31, ARTICLE 44, PAGES 13111336
PUBLISHED 3 DECEMBER 2014
http://www.demographic-research.org/Volumes/Vol31/44/
DOI: 10.4054/DemRes.2014.31.44
Research Article
Ageing dynamics of a human-capital-specific
population:
A demographic perspective
Dimiter Philipov
Anne Goujon
Paola Di Giulio
© 2014 Dimiter Philipov, Anne Goujon & Paola Di Giulio.
This open-access work is published under the terms of the Creative Commons
Attribution NonCommercial License 2.0 Germany, which permits use,
reproduction & distribution in any medium for non-commercial purposes,
provided the original author(s) and source are given credit.
See http:// creativecommons.org/licenses/by-nc/2.0/de/
Table of Contents
1
Introduction
1312
2
Data and methods
1314
2.1
Multistate education-specific population projections
1315
2.2
Human-capital-specific age composition of the working-age
population
1319
2.3
Human-capital-specific age pyramid
1322
3
Ageing dynamics: The human-capital-specific old-age dependency
ratio
1323
4
An alternative measurement: Education-specific OADR
1327
5
Summary and discussion
1328
6
Acknowledgements
1331
References
1332
Appendix: The Italian context
1335
Demographic Research: Volume 31, Article 44
Research Article
http://www.demographic-research.org 1311
Ageing dynamics of a human-capital-specific population:
A demographic perspective
Dimiter Philipov
1
Anne Goujon
2
Paola Di Giulio
3
Abstract
BACKGROUND
Research on how rising human capital affects the consequences of population ageing
rarely considers the fact that the human capital of the elderly population is composed in
a specific way that is shaped by their earlier schooling and work experience. For an
elderly population of a fixed size and age-sex composition, this entails that the higher
its human capital, the greater the total amount of public pensions to be paid.
OBJECTIVE
The main purpose of this paper is to analyse the link between human capital and retiree
benefits and its effect on population ageing from a demographic viewpoint.
METHODS
We construct an old age dependency ratio (OADR), in which each person, whether in
the numerator or the denominator, is assigned the number of units corresponding to
his/her level of human capital. Based on data for Italy, we study the dynamics of this
human-capital-specific OADR with the help of multistate population projections to
2107.
RESULTS
Our results show that under specific conditions a constant or moderately growing
human capital may aggravate the consequences of population ageing rather than
alleviate them.
1
Wittgenstein Centre for Demography and Global Human Capital (IIASA, VID/ÖAW, WU), Austria.
E-Mail: dimiter.philipov@oeaw.ac.at.
2
Wittgenstein Centre for Demography and Global Human Capital (IIASA, VID/ÖAW, WU), Austria.
3
Wittgenstein Centre for Demography and Global Human Capital (IIASA, VID/ÖAW, WU), Austria.
Philipov, Goujon & Di Giulio: Ageing dynamics of a human-capital-specific population
1312 http://www.demographic-research.org
CONCLUSIONS
With those findings, the authors would like to stimulate the debate on the search for
demographic and/or socio-economic solutions to the challenges posed by population
ageing.
1. Introduction
Many authors consider an increase in human capital as a major factor for alleviating the
negative effects of population ageing (Chawla, Betcherman, and Banerji 2007; Lee and
Mason 2010, 2011; Fougère and Mérette 1999; Lutz, Sanderson, and Scherbov 2004).
In the macro-economic framework, the classical works on human capital by Theodore
Schultz (e.g., 1961) and Gary Becker (e.g., 1964) are followed by the new growth
theory, whose proponents, among them Lucas (1988) and Romer (1990), view human
capital as an input that increases the labour force‟s productivity (Skirbekk 2008). They
believe that this will stimulate long-term economic growth and thus also facilitate the
allocation of the resources required for the retired population (Kemnitz and Wigger
2000). From a demographic viewpoint, the increase in human capital will offset the
shrinking working-age population, as those most educated tend to work longer and
retire at later ages (Crespo Cuaresma, Lutz, and Sanderson 2009; Lutz, Crespo
Cuaresma, and Sanderson 2008; Lutz, Goujon, and Wils 2008; Lutz and KC 2011;
Striessnig and Lutz 2014). Advocates of the neo-classical view (Mankiw, Romer, and
Weil 1992) are, however, convinced that human capital will affect the productivity
growth rate in the short run, while its return will decrease in the long run. In the same
vein, Börsch-Supan (2003) argues that a rise in human capital cannot fully compensate
population decline.
Research on how rising human capital affects the consequences of population
ageing does not explicitly consider the impact of the elderly population‟s human capital
that is shaped by earlier schooling, training and work experience. Retirees‟ human
capital, utilised during their working lives, is decisive for the level of their income. In
this paper we consider public pensions of a pay-as-you-go scheme as a proxy for the
income of the older population
4
. Hence, for an elderly population of a fixed size and
age-sex composition, the total amount of public pensions to be paid increases with the
level of its human capital. In a cohort perspective, a rise in the human capital of the
4
This assumption is made to simplify the discussion. Pay-as-you-go is the preponderant pension scheme in
Italy and in many other European countries. The validity of our inferences for other pension schemes requires
analyses that are beyond the scope of this paper.
Demographic Research: Volume 31, Article 44
http://www.demographic-research.org 1313
working-age population will generate economic growth, but upon retirement it will also
increase the economic burden by a corresponding rise in pensions.
The main purpose of this paper is to analyse the link between human capital and
retiree benefits and its effect on population ageing from a demographic viewpoint. Our
research method is based on the construction of a population disaggregated by age (as
of age 20) and sex where all individuals are weighted according to the level of their
respective human capital. We defined the weights for the working-age population by
using earning functions and the weights for the elderly population by using human-
capital-specific public pension levels. Striessnig and Lutz (2014) apply education-
weighted dependency ratios with the purpose to determine the optimal fertility needed
to achieve the lowest total dependency ratio. In their calculation they assume the
retirement benefits to be the same across all education categories; productivity of
human capital is assumed to differ with the ratio of 1 : 1.25 : 1.5 between primary,
secondary and tertiary education (Striessnig and Lutz 2014: Appendix table A1).
Instead, we infer these ratios by applying earning functions for the working population
and empirical data for the retirees.
We examined the dynamics of ageing by using a human-capital-specific old age
dependency ratio (OADR) relying on multi-state population projections by levels of
education. We applied the method to the population of Italy where changes among
cohorts in terms of educational attainment have been radical. In Italy, the share of adults
with tertiary education is lower than in many other EU countries. It is expected to rise
considerably in line with the target of “at least 40% of 30–34–year-olds completing
third-level education” stated in the European Union‟s Europe 2020 programme for the
present decade (European Commission 2010a). The Appendix provides more
information on the Italian context.
Our analysis of how human capital affects ageing is not based on forecasts, but on
population projections. We also want to emphasise that these projections permit us to
examine the effect of one specific force – namely human capital – on ageing under the
ceteris paribus assumption, i.e., net of the effect of any other factors such as economic
growth, level of un/employment, economic in/activity, ability to work and health, part-
time employment, wage and pension differentials by length of working life, pension
schemes, savings and spending, consumption patterns, labour-force participation of
elderly people, age at retirement and entry into the labour force. Indeed, the future
impact of the human-capital-specific OADR on population ageing can be strongly
linked with most of these factors. However, before adding these additional factors in the
analysis, we need to know how human capital affects ageing and whether it is strong
enough to be considered in rigorous analyses of the future of population ageing. Our
analysis is purely demographic, and extends inferences based on the use of the
conventional OADR.
Philipov, Goujon & Di Giulio: Ageing dynamics of a human-capital-specific population
1314 http://www.demographic-research.org
2. Data and methods
We used the European Union Statistics on Income and Living Conditions (EU-SILC)
for Italy, 2007, made available in the 2008 wave (IT-SILC XUDB 2010 - version
December 2011. For more information on EU-SILC see: http://epp.eurostat.ec.europa.
eu/portal/page/portal/microdata/eu_silc [consulted on 10 Dec. 2013].). Launched in
2003, EU-SILC was the first longitudinal micro-level survey that provided
comprehensive data on income and a broad range of other social and economic
information across all 28 Member States of the enlarged EU and a number of other
countries. We used the information on respondents aged 20–64 who work full-time
(16,765, representing 70% of total working population) or who are aged 65 and above
and receive a public pension (10,464 persons). The selection of full-time respondents is
based on their self-defined current economic status; income is the self-declared
individual gross income obtained by gainful work. We computed the gross public
pension received by a respondent as the sum of old-age benefits, survivor‟s benefits and
disability benefits, for all those who receive a public pension. All private pensions were
excluded. Education is indicated as the highest education level attained, using the
following categories of the International Standard Classification of Education: ISCED
0–2 (junior-secondary and below), ISCED 3–4 (completed upper-secondary and post-
secondary non tertiary), ISCED 5–6 (tertiary education). In this paper, we refer to these
three levels also as low, middle and high education.
Table 1 shows the populations aged 20–64 and 65+ and their incomes/pensions
disaggregated by three education levels. The population with low levels of education is
the largest in both age groups. However, educational improvements are visible across
the different cohorts. The old-age population has significantly lower levels of education
than the younger cohorts: In 2007, only 15% of the 65+ population had an upper
secondary or higher education, as compared to 54% in the 20–64 and 75% in the 20–24
age group in the same year (not shown in the Table). Until the end of the 1970s,
educational enrolment in Italy increased most at the lower secondary level. It was only
at the beginning of the 1990s that the transition to upper-secondary education became
the norm for students.
Table 1 also shows that income and pensions
5
significantly differ by education
level: The median annual pension received by retirees with a high education is more
than twice that received by their low-educated peers. This differentiation validates our
research topic. The pension median levels observed in 2007 are the result of the
different pension reforms that were actuated in Italy starting gradually from the 1990s
5
The median income and pension are computed for all those who receive, respectively, a positive income and
a positive pension. People aged less than 65 who receive no income or have a negative income, and people
aged more than 64 with no public pension are therefore excluded from the calculation.
Demographic Research: Volume 31, Article 44
http://www.demographic-research.org 1315
(see appendix for more detail). The median income levels observed in table 1 reflect the
distribution by age and education of the working population. As much as 43% of the
total income produced by highly educated people was actually earned by the population
aged 50 and over, with people aged 20–39 (about 43% of the total working population
with high education level) having a median income of actually 22500 euros (data not
shown). For further details on income level see section 2.2.
Table 1: Population 20+ and median value of income/pension by three levels
of education, Italy, 2007
Numbers (in
thousands)
Percentages
Income per
year, median
(Euro)
Pension per
year, median
(Euro)
Level of
education
Age group
20–64
Age group
65 and
higher
Age group
20–64
Age group
65 and
higher
Age group
20–64
Age group 65
and higher
Low
(ISCED 0–2)
16,714
10,033
46
85
18,892
10,749
Middle
(ISCED 3–4)
13,140
712
41
11
22,259
19,495
High
(ISCED 5–6)
4,718
282
13
4
27,769
27,445
Total
34,572
11,027
100
100
21,383
11,741
Source: ISTAT (2008) and ISTAT (2007); for income and pensions: EU-SILC 2008
2.1 Multistate education-specific population projections
The multi-state population projections used in this research are based on the projections
developed by Goujon (2009) in the framework of the MicMac project,
6
which
implement the calculation of transition probabilities between four levels of education
(ISCED 0–1, ISCED 2, ISCED 3–4 and ISCED 5–6). For the purposes of this paper, the
first two education states were aggregated into one (ISCED 0–2) in the result section.
The initial year is 2007 and the base-year population was taken from the 2007 Labour
Force Survey for Italy (ISTAT 2007); projections were made until 2057 for the first
phase and prolonged to 2107 for the second phase, according to the main demographic
assumptions presented in Table 2.
6
The MicMac project was funded under the 6th Framework Programme of the European Union. Details about
the projection can be found in Deliverable 3 of the project. See: www.nidi.nl/Content/NIDI/output/
micmac/micmac-d3.pdf [last viewed on January 16, 2014].
Philipov, Goujon & Di Giulio: Ageing dynamics of a human-capital-specific population
1316 http://www.demographic-research.org
Table 2: Fertility, mortality and migration assumptions, Italy, 2007, 2057,
and 2107
Variable
Total and by
education
2007
2057
2107
Total Fertility Rate
Total
1.4
1.6
1.6
ISCED 0-1
1.7
2
2
ISCED 2
1.4
1.6
1.6
ISCED 3-4
1.4
1.6
1.6
ISCED 5-6
1.4
1.6
1.6
Male life expectancy at
age 0
Total
78.7
86.1
86.6*
ISCED 0-1
76.1
81.7
81.7
ISCED 2
77.0
82.9
82.9
ISCED 3-4
82.2
88.1
88.1
ISCED 5-6
82.5
88.4
88.4
Female life expectancy at
age 0
Total
84.0
90.8
91*
ISCED 0-1
82.6
88.2
88.2
ISCED 2
83.7
89.6
89.6
ISCED 3-4
85.8
91.7
91.7
ISCED 5-6
86.0
91.9
91.9
Net migration male
(in thousands)
Total
196
97
84
ISCED 0-1
97
50
50
ISCED 2
38
18
13
ISCED 3-4
42
20
14
ISCED 5-6
18
9
6
Net migration female
(in thousands)
Total
237
94
79
ISCED 0-1
104
42
42
ISCED 2
45
18
13
ISCED 3-4
58
22
16
ISCED 5-6
29
11
8
* Although the mortality rates by education are constant after 2050, the overall life expectancy changes between 2050 and 2107
because the weight of the different populations in the different education categories changes. Table 2 shows the life
expectancies in 2107 according to the trend scenario.
Demographic Research: Volume 31, Article 44
http://www.demographic-research.org 1317
In our analysis, changes in fertility corresponded to the medium variant of the
ISTAT projections (ISTAT 2008), further disaggregated until 2050 by educational
attainment according to estimates by KC et al. (2010). After 2050, fertility rates and
education differentials remained constant. Mortality was disaggregated by education
using estimates for 2008, which show that male life expectancy at age 30 is 48 years for
men with low education, 52.9 years for men with middle education and 53.1 years for
men with high education. The corresponding life expectancies for women are 54, 56.6
and 56.7 years, respectively (European Commission 2010b: 39, Table I.3.7), i.e., a
significant difference was observed between the first and the second levels. The
differential in life expectancy at birth by education shown in Table 2 was obtained from
United Nations standard life tables, applied to the ISTAT 2008 overall life expectancy
projection assumptions until 2050 and kept at this level thereafter. Figures for the net
number of migrants were based on the latest projections by ISTAT to 2065 (ISTAT
2011, central scenario) and the rate of decrease estimated by ISTAT for net-migration
between 2064 and 2065 was kept constant across the rest of the projection period (to
2107), i.e., fewer migrants as the population declines
7
. Migration was also
disaggregated by education according to Docquier, Lowell, and Marfouk (2009) for
population ages 25+. The distribution of migrants for ages up to 24 followed that of the
Italian population.
Although the mortality rates by education are constant after 2050, the overall life
expectancy changes between 2050 and 2107 because the weight of the different
populations in the different education categories changes. Table 2 shows the life
expectancies in 2107 according to the trend scenario.
Our projections were based on two scenarios for the educational transition rates: a
constant and a trend scenario. The constant version assumed that the rates observed
during the last period of observation (2004–2007) remain unchanged. The trend
scenario was set by using transitions across three periods (1995–1999, 2000–2003 and
2004–2007) and extrapolating them until the 2050s. The following targets for reaching
maximum transition rates were set along the projection period:
Maximum value for transitions from ISCED 0–1 to ISCED 2 is 1.0 (i.e., all
pupils have at least junior secondary education); achieved in 2032 for men
and in 2027 for women.
7
Although it might be discussed whether a country with a declining population will turn or not to
immigration to replace the missing productive population, migration tends to be very volatile, and hardly
predictable, hence most – if not all – projection exercises (by national statistical institutions and international
organizations, like the United Nations Population Division) tend to lower progressively net migration figures
compared to the base year through the projection period and have them converging to 0. Moreover, many
European countries faced with the economic crisis tend to lean presently towards more restrictions on
immigration.
Philipov, Goujon & Di Giulio: Ageing dynamics of a human-capital-specific population
1318 http://www.demographic-research.org
Maximum value for transitions from ISCED 2 to ISCED 3–4 is 0.85 (as
targeted by the EU in its Lisbon strategy); achieved in 2037 for both sexes.
Maximum value for transitions from ISCED 3–4 to ISCED 5–-6 is 0.45 (i.e.,
reach the target set in the Europe 2020 strategy document, levels comparable
to those currently reached in the United States; achieved in 2040 for both
sexes.
In order to validate our results, we compared the population by age, sex and
education obtained from the projections in 2012 in the trend and constant scenarios to
that of the Italian Labour Force Survey (LFS) values for the same year. The results were
very close in terms of total population by large age groups and sex, and also in terms of
levels of education, i.e., the overall difference was less than 5%. It is interesting to see
that the 2012 „real‟ distribution according to the 2012 LFS was closer to that of the
constant scenario than to that of the trend scenario.
Table 3 shows the projection results in the first phase of 50 years. We compared
the results of the projections to that of ISTAT (2011) available until 2065 and the
difference is negligible in all years: Below 1% difference in total population in 2065
between the constant/trend scenario and the ISTAT central scenario. The difference
between ISTAT and our projections is slightly larger in the middle of the ISTAT
projection period (2020 to 2045) due most likely to different age schedules of migration
but never exceeds 2.5%. The age composition (proportion 0–14, 15–64, 65+, and 85+)
is also not affected by the introduction of education.
Comparisons with Table 1 show that even in the constant scenario the proportion
of the aged population with lower than secondary education (ISCED 0–2) will decline
considerably as a result of the recent increases in enrolment rates among younger
cohorts. The trend scenario yields a considerably higher education level in the working-
age population than the constant scenario.
Table 3: Distribution of the population 20+ by three levels of education, under
constant and trend scenario assumptions, Italy 2057, percentages
Constant scenario
Trend scenario
Level of education
Age group
20–64
Age group 65
and higher
Age group
20–64
Age group 65
and higher
Low (ISCED 0–2)
28.5
32.9
17.4
32.7
Middle (ISCED 3–4)
52.9
47.6
21.1
46.0
High (ISCED 5–6)
18.6
19.5
61.4
21.3
Total
100.0
100.0
100.0
100.0
Source: Authors’ calculations
Demographic Research: Volume 31, Article 44
http://www.demographic-research.org 1319
2.2 Human-capital-specific age composition of the working-age population
Let us recapitulate what we have done so far. In the first step, we projected the age, sex
and education composition of the population. In this framework, each individual
contributes exactly one unit to the overall distribution, irrespective of his/her age, sex or
level of education. In the next step, we differentiate the units according to the achieved
level of human capital. To this end, we apply earning functions to the working-age
population (ages 20–64). This can be done in at least two different ways: The first
approach is education-specific, considers the number of years spent in school and
differentiates persons according to the length of their schooling. As the latter hardly
ever changes after age 30, it has the disadvantage that the human capital accumulated
per person will not change in the remaining lifespan. The approach is appropriate when
education is the sole item of interest in the estimation of human capital.
The second approach, advocated by Mincer (1974), seemed more appropriate for
our purposes. It considers work experience in addition to the length of schooling.
Hence, human capital can grow throughout the working life of all individuals. We
applied a simple form of Mincer‟s earning function to the three educational levels
separately for men and women. We prefer to use the term „income function‟ here as the
function is estimated based on income from labour before taxes. This function assumes
that labour income is positively correlated with schooling and work experience (the
longer the schooling/work experience, the higher the income). We linked each level of
education with an average number of years spent in school, and work experience with
age: the higher the age, the longer the work experience. However, we wanted to take
into account the fact that human capital based on work experience accumulated at the
end of working life might be outdated and mark a relative decline. Therefore, the
income function levels off towards old working ages or takes the form of an inverse U-
shape (as a result of approximating long work experience with age-squared in the
functional form of the equation). This yields the following regression equation, where Y
denotes labour income and ε the error term:
8
Labour income is defined as income before taxation (gross income) received by a
person who is employed full-time.
8
Earning functions are usually expressed in a log-linear functional form. Other functional forms are also
considered, such as the log-log and the linear; see reviews by Lemieux (2003) and Polachek (2007). The
debate on the functional form refers to interpretations of the coefficients. It is not relevant here as we apply
the coefficients without reference to their interpretation.
Philipov, Goujon & Di Giulio: Ageing dynamics of a human-capital-specific population
1320 http://www.demographic-research.org
Less than a dozen persons with extremely high incomes were considered outliers
and excluded (limits: more than € 200,000 for persons with lower than tertiary
education, and more than € 300,000 for persons with tertiary education). Figure 1
shows the income functions for the two sexes. The figures for the higher education
graphs were estimated over the age span 23–63 because of the small number of cases
outside this age interval, and the estimates were extrapolated to ages 20 and 65. An
example: The values of the income functions at age 50 for men are € 25,000, € 35,000
and €58,000 for levels below junior secondary, upper secondary and tertiary education,
respectively. For women, the corresponding figures are € 19,000, € 27,000 and €
40,000.
The mean values by age yielded by each of the income functions matched very
well with the means directly estimated from the sample. Comparisons were made for
10-year age groups: 25–34, 35–44, 45–54 and 55–64, separately for males and females
and for each education level. Only in 3 out of 24 cases was the mean of the income
function outside the 95% confidence interval of the sample mean, namely for males
aged 25–34 with secondary and tertiary education and for females aged 35–44 with
tertiary education. In most of the other cases, the differences between the two means
were smaller than 1%.
Figure 1: Gross income functions of women (left) and men (right) by age and
education level, Italy, 2007
0
10
20
30
40
50
60
70
20 30 40 50 60
Euros in thousands
age
Low
Middle
High
0
10
20
30
40
50
60
70
20 30 40 50 60
Euros in thousands
age
women
men
Demographic Research: Volume 31, Article 44
http://www.demographic-research.org 1321
Figure 1 shows well-known characteristics of the distribution of income by age,
gender and level of education: higher education leads to higher income, men earn more
than women, levels of income flatten (but do not decrease) before retirement. We only
mention these findings, as a detailed discussion would be beyond the scope of this
article. It is important to note that while the income functions are differentiated by
education in Figure 1, they also reflect accumulated human capital, because they
depend on the length of working life. If the latter had not been considered, the curves
would be straight lines running parallel to the horizontal axis and would not reflect real
income by age accurately.
We applied the income functions to construct a human-capital and age-specific
composition of the population aged 20–64. Consider as an example the values of the
income functions at age 50 as outlined above. For men, the ratios are 1:1.42 for middle
education and 1:2.34 for high education, relative to the income of a male with low
education. Hence, if a man aged 50 with low education contributes 1 unit to the human-
capital-specific age composition of the working-age population, a male aged 50 with
middle education contributes 1.42 and a male aged 50 with high education contributes
2.34 units. Similar ratios can be computed for other ages, but they are incomparable
across ages, because the income of a less-educated male differs by age. Similarly, the
income of a female aged 50 is neither comparable to that of a male aged 50 nor to the
income of individuals at other ages. In order to achieve comparability across ages and
sexes, we pivoted all values to the lowest income, which is that of a female aged 20
having an education equal to or lower than junior secondary level. If she contributes 1
unit to the age composition, a man aged 50 contributes 2.2, 2.87 or 4.72 units,
respectively, depending on his level of education.
The human-capital composition of the population aged 65 and above is constructed
differently. We assumed that this population was retired, so we examine its retired
human capital. To this end, we made use of the median of education-specific gross
public pensions over the whole age span 65 and above. The person-units assigned to a
person aged 65 or older with a specific sex and education were once more related to the
income of a female aged 20 having an education equal to or lower than secondary level.
The median gross public pensions for both males and females and by each level of
education were taken from the EU-SILC data (Table 1).
Our method of constructing a human-capital and age-sex composition of the
working-age population combines individual-level estimates of income differentiated
by human capital, age, and sex, with the observed working-age population distributed
by the same three components. One may argue that estimates of the income functions
derived for full-time employment are assigned to persons employed part-time or not
employed at all. Our rationale is that, when assigned to the whole working-age
population, these estimates present the full productive capacity, independent of
Philipov, Goujon & Di Giulio: Ageing dynamics of a human-capital-specific population
1322 http://www.demographic-research.org
employment and health/disability status. In case the distribution by employment status
was introduced, we would have to assume that it would remain unchanged during the
next several decades; the effect of similar assumptions on our results is discussed in the
last section. Similar note is valid for the population aged above 65, where median
human-capital and sex-specific pensions computed from survey data are assigned to the
observed population differentiated by education and sex.
2.3 Human-capital-specific age pyramid
The human-capital-specific age pyramid shows an age and sex-specific population
composition, in which each person‟s contribution of one person-unit is weighted with
his/her level of human capital. The total number of weighted person-units depends on
the selected pivot value and on survey information on gross income and public pensions
and does not correspond to the actual population size, which, however, is insignificant
for the purposes of the present study. Hence, the age composition can be standardised to
a total population of say 10,000. Figure 2 shows the age compositions of the studied
population broken down into three educational levels (left) and standardised for 10,000
person-units, and the human-capital-specific population (right) standardised for 10,000
human-capital-weighted (or human-capital-specific) person-units. Ages are considered
from 20 upwards because our method of constructing human-capital-specific person-
units is inapplicable for the population aged 0–19.
Figure 2: Age composition of the studied population by sex and education (left
pyramid), and of the sex and human-capital-specific population
(right pyramid), Italy 2007, both weighted to 10,000 units
150 100 50 0 50 100 150
20
25
30
35
40
45
50
55
60
65
70
75
80
85
90
95
100
age
150 100 50 0 50 100 150
20
25
30
35
40
45
50
55
60
65
70
75
80
85
90
95
100
age
Low
Middle
High
men
women
men
women
Demographic Research: Volume 31, Article 44
http://www.demographic-research.org 1323
Pyramids like the one in the left part of Figure 2 have been extensively studied by
Lutz, Goujon, and Wils (2008), Lutz and KC (2011) and, specifically for Italy, by
Goujon (2009). Results of population forecasts by levels of education are presented in
this form. Pyramids like the one in the right part of Figure 2 have not been discussed in
the literature. This is a pyramid of person-units specified by age, sex and human capital,
in which the latter is weighted by the size of gross labour income and public pensions.
Since income and pensions are of primary importance in the construction of a person-
unit, the form of the pyramid is typically influenced by income dynamics: it is thickest
around ages 40–50 when labour income is high, and it is very thin from age 65
onwards, when the pension size constitutes the basis for person-unit construction.
3. Ageing dynamics: The human-capital-specific old-age dependency
ratio
Demographers traditionally study ageing dynamics with the help of such indicators as
median age, the share of the population aged 65+ and the old-age dependency ratio
(OADR). Figure 2 shows that all of them can be estimated for a population
disaggregated by the level of human capital. The analysis in this paper is based on the
OADR in conjunction with the dynamics of changes in the population aged 65 and
above.
In the conventional OADR, defined as the ratio of the population aged 65 and
higher to the population aged 20–64 (the choice of 20 and 65 as cutting ages is
insignificant in this paper), each person contributes exactly one unit to the denominator
or numerator, irrespective of his/her level of human capital. Instead, we use the
population composition disaggregated by human capital as explained above and
presented in the right part of Figure 2. The OADR extended in this way thus includes
the effect of accumulated human capital. It is a new rate, which we call HC-OADR. For
the pyramid in the left part of Figure 2, the conventional old age dependency ratio is
33%, while the HC-OADR is 16% for the population shown in the right pyramid of
Figure 2. The latter ratio is considerably lower, because public pensions are lower than
gross labour income and hence a pensioner‟s contribution to a human-capital-specific
person-unit is smaller.
We used projections by education till 2057 and constant assumptions thereafter up
to 2107 to estimate the HC-OADR for each of the projection years and applied both the
constant and the trend scenarios. The ratios of pensions and labour income by age and
gender were kept constant at the level of the initial year, i.e., 2007. Figure 3 shows the
HC-OADR over the 50 years till 2057 in absolute terms (left) and relative to 1 in the
initial year (right).
Philipov, Goujon & Di Giulio: Ageing dynamics of a human-capital-specific population
1324 http://www.demographic-research.org
Our analysis yielded the following results: First of all, the HC-OADR is two times
lower than the conventional ratio (16% versus 33%). When measured in person-units
based on income and pensions, population ageing seems to be less severe than when
measured with the conventional ratio. However, this does not lower the economic
burden: it is simply a more precise way of computation than the conventional
demographic method. Second, both the conventional and the HC-OADR will grow in
the forthcoming decades (see left graph in Figure 3). Education, whether it is kept
constant at the initial year of projection or extrapolated to grow till 2057, does not turn
around the ageing process, although the latter is attenuated after 2045 according to the
trends of the HC-OADR(t) and the conventional OADR. Third, each of the HC-OADR
grows faster than the conventional OADR (see graphs in the right part of Figure 3).
Fourth, when education increases as depicted in the trend scenario, the HC-OADR
grows at a slower pace after 2030 than the HC-OADR depicted in the constant
education scenario. We also tested the robustness of the projection results by removing
the education differentials for fertility, mortality, and migration (not shown here) and it
did not affect significantly the relative measure of the HC-OADR compared to the
conventional OADR.
Figure 3: Trends in the conventional OADR and the HC-OADR [constant (c)
and trend scenario (t)], 2007-2057, absolute values (left) and relative
to 1 in the initial year (right)
The last two results outlined above are peculiar. According to contemporary
literature, an increase in human capital is expected to decelerate the negative
consequences of population ageing. However, at least for Italy and for the next 50
years, our results do not support this view.
0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
2007 2017 2027 2037 2047 2057
Absolute values
OADR
HC-OADR(c)
HC-OADR(t)
0
0.5
1
1.5
2
2.5
3
2007 2017 2027 2037 2047 2057
Relative values
OADR
HC-OADR(c)
HC-OADR(t)
Demographic Research: Volume 31, Article 44
http://www.demographic-research.org 1325
Before explaining these findings, let us take a look at the long-term projection up
to the point when population stability has been reached. After 2057, the two scenarios
were continued for an additional 50 years, based on the assumption that the educational
transition rates do not change. When keeping all components of population change
constant, projections over a longer period showed that the values observed for 2107 are
stabilised to the extent required for the needs of the discussion in this paper. Figure 4
shows the ratios over a period of 100 years.
Figure 4: Conventional and HC-OADR [constant (c) and trend scenario (t)],
2007–2107, absolute values (left) and relative to 1 in the initial year
(right)
When population stability has been reached, the absolute values of the three
indicators (left part) are 0.58 for the OADR and 0.37–0.39 for the HC-OADR. The
conventional OADR remains higher than the HC-OADR. The relative dynamics of
change are better depicted in the right part where the three indicators are weighted so
that their value in 2007 equals 1. The corresponding values at the point of stability are
1.8 for the OADR, 2.3 for the HC-OADR(c) and 2.5 for the HC-OADR(t): i.e., while
the conventional OADR is 1.8 times higher as compared to the observed value for 2007,
the HC-OADR increases by 2.4 to 2.5 times.
The OADR-based results show that population ageing is faster when quantitative
differences between levels of human capital are appropriately measured. What is the
explanation for this unusual finding? We refer to the initial and projected composition
of the population by education given in Tables 1 and 2. While the less-educated
dominated among the old-age population in the initial year, their share declined
drastically over a period of 50 years, i.e., the education level of the elderly population
increased considerably.
0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
2007 2027 2047 2067 2087 2107
Absolute values
OADR
HC-OADR(c)
HC-OADR(t)
0
0.5
1
1.5
2
2.5
3
2007 2027 2047 2067 2087 2107
Relative values
OADR
HC-OADR(c)
HC-OADR(t)
Philipov, Goujon & Di Giulio: Ageing dynamics of a human-capital-specific population
1326 http://www.demographic-research.org
Figure 5 shows how the populations aged 65 and above and the working-age
populations change up to 2057 both in the conventional projection and in the constant
HC-scenario (where person-units are weighted with their human capital), all relative to
1 in 2007, i.e., the initial year. The working-age populations do not change much during
the 50-year period; their relative values fluctuate around 1. The elderly populations,
however, increase considerably and the HC-weighted populations grow much faster.
Hence, while the denominators of both the OARD and the HC-OADR do not change
much, their numerators grow significantly, but those of the HC-OADR grow faster.
The faster growth of the HC-weighted aged population is due to the larger number
of highly educated persons whose pensions are higher. Thus an earlier increase in
human capital entails a higher demand for public pensions (provided the other
conditions remain unchanged). From a demographic viewpoint, this link is sound and
obvious. In real life, it is only one among numerous economic and social forces that
should be taken into account to better control the overall effect that raising human
capital has on the economic and social consequences of ageing. This has not yet been
explicitly considered in contemporary research. Our analysis shows that the link can be
significant and should not be neglected.
Figure 5: Relative projected trends in numerators (age group 65+) and
denominators (age group 20–64) of old-age ratios, conventional
(conv.) and constant HC-scenarios (c)
0
0.5
1
1.5
2
2.5
3
2007 2017 2027 2037 2047 2057
Relative values
c65+
c20-64
conv.65+
conv.20-64
Demographic Research: Volume 31, Article 44
http://www.demographic-research.org 1327
4. An alternative measurement: Education-specific OADR
Human capital comprises at least three important components: education, work
experience and health. The impact of the first two was taken into account when
constructing the HC-OADR. As data on work experience and health are scarce, human
capital is frequently equated with education. It is therefore interesting to examine how
education affects the ageing dynamics of populations while disregarding the impact of
other human-capital components.
Education can be measured in different ways, one being the number of years spent
studying. When three levels of education are considered, they correspond to study
lengths of around 8 years for students with a junior-secondary or less education, 12
years for those with an upper-secondary education and 16 years for the tertiary
educated. This ratio of 1:1.5:2 can be assigned to persons of any age. A disadvantage of
this approach is that an elderly person and a working-age person with the same
education contribute the same amount of education-weighted person-units to their
corresponding age group, while in the approach outlined above, elderly people
contribute a lower number of person-units because pensions are lower than labour
income. Hence, trends can differ.
The construction of an education-weighted population size for one specific age
does not affect other ages because no pivot value is applied. This is a distinctive
difference to the method outlined above. Yet the age composition will not be identical
to the conventional one because the person-units contributed by individuals with a high
education are doubled, and those of individuals with a middle education are increased
by one half. Moreover, the proportion of individuals with a specific level of education
differs across age groups over time. These differences are reflected in the ED-OADR,
which is lower than the conventional OADR (Figure 6, left part) because the proportion
of higher-educated people is lower among the elderly than among the younger cohorts
and the weighted person-units increase more for the young than for the elderly
population. When transition rates and components of change remain constant in the
long run, the educational and age composition of the population stabilise and the
OADR and the ED-OADR from the two scenarios equalise, which is not the case when
using the HC-OADR.
Figure 6 (right part) describes the pace of changes noted for the three types of
OADR. Relative to the conventional OADR, the dynamics of the ED-OADR differ
from those of the HC-OADR (Figure 4, right part). It is interesting to note that an
increase in education keeps the pace of increase in the ED-OADR(t) similar to that of
the conventional ratio for some six decades, while the ED-OADR(c) in the scenario
based on a constant level of education shows a steeper ageing curve. The population
ages faster when human capital is measured in this way instead of using the
Philipov, Goujon & Di Giulio: Ageing dynamics of a human-capital-specific population
1328 http://www.demographic-research.org
conventional OADR, but slower than indicated by the ratio that also takes into account
the length of work experience.
Figure 6: Conventional and ED-OADR [constant (c) and trend scenario (t)],
2007–2107, absolute values (left) and relative to 1 in the initial year
(right)
5. Summary and discussion
Governments in developed countries make every effort to relieve the unsustainable
pressure of growing population ageing. Raising the education level is frequently
advocated as a solution because of its established positive impact on the productivity of
the labour-force and economic growth. The aim of the article is not to cast doubt on the
necessity of education and the many positive externalities associated with enlarged
quality education at the individual and macro level, that range from better health to
increased economic wealth to mention just a few. However, researchers rarely, if at all,
tackle the fact that a better-educated labour force will require higher pensions once it
has retired. This obvious consequence shows that raising the level of education is not a
cure-all for easing the burden of population ageing. By integrating human capital into
the calculation of the conventional population age composition and dependency ratio
we show that an increase in education accelerates population ageing.
9
9
An anonymous reviewer put it neatly: “Can human capital solve the ageing crisis? Of course not, for the
same reason that encouraging migration is no solution: migrants themselves become old. Today‟s highly
qualified high earners become tomorrow‟s expensive pensioners.”
0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
2007 2027 2047 2067 2087 2107
Absolute values
OADR
ED-OADR(c)
ED-OADR(t)
0
0.5
1
1.5
2
2.5
3
2007 2027 2047 2067 2087 2107
Relative values
OADR
ED-OADR(c)
ED-OADR(t)
Demographic Research: Volume 31, Article 44
http://www.demographic-research.org 1329
We show these findings for Italy by elaborating the fundamental concept of a
person-unit. In conventional calculations, each person contributes one person-unit to the
age composition of a population. Instead, we weight person-units with human-capital-
based weights. The human-capital-based age composition of the population can be used
to compute the share of the population aged 65 and above as well as the human-capital-
specific OADR. We found that the share of the weighted population aged 65 and above
increases faster than that in the non-weighted equivalent. As a consequence, the HC-
OADR increases faster than the conventional OADR.
Constructing weighted population age compositions is a new approach. Let us first
consider the OADR. As a measure of population ageing it is subject to a range of
assumptions, among them fixed cutting ages, and in particular the upper one. This
assumption can be crucial when life expectancy beyond age 65 increases. It was relaxed
by Sanderson and Scherbov (2005) who introduced the concept of prospective age
based on a reverse measurement of ageing, i.e., not from the start of life but relative to
its expected length. Their method of measuring ageing showed a much slower trend of
ageing as compared to a measurement based on the OADR.
In a later article, Sanderson and Scherbov (2010) measured ageing based on the
ability to work (ratio of persons with disabilities to persons without disabilities). They
defined this measure with one cutting age at 30, formulating the ratio for persons aged
30 and above. This indicator of ageing relaxes other implicit assumptions in the
conventional OADR, namely that all persons in the economically active age group
(approximately 16 to 65) contribute to the economy independently of their health status,
and that all persons aged 65 and above are considered to be consumers but not
producers of wealth. In a related approach, the so-called Rostock index of ageing
considers employed versus unemployed persons in the numerator and denominator of
the OADR (Vaupel and Loichinger 2006). While relaxing assumptions of the
conventional OADR, their extensions add new assumptions: for example, that the share
of persons with abilities/disabilities, or employed/unemployed will remain constant
during the projection periods. Nevertheless, they supply useful information and enrich
the understanding of population ageing.
The HC-OADR does not incorporate these extensions of the conventional OADR
while it relaxes another assumption, namely that each person contributes exactly one
person-unit to either the denominator or the numerator (or in terms of earnings, each
person receives one and the same income, independently of age, sex, education, work
experience, and independently of whether it is labour income or pension). Instead, we
differentiate individuals according to the level of their human capital; moreover, we
distinguish between working and retired human capital.
Like the above extensions, our HC-OADR also depends on assumptions about
such unchanging shares as those related to employment and ability. Its sensitivity to
Philipov, Goujon & Di Giulio: Ageing dynamics of a human-capital-specific population
1330 http://www.demographic-research.org
these assumptions does not differ significantly from that in the extensions described
above. Consider, for example, the share of unemployed persons. Our assumption is that
all people are fully employed. If we relax it and use the share of employment instead,
but keep the latter unchanged during the period of projection, we get the same trend in
the HC-OADR as the one depicted in the right parts of Figures 3 and 4. The absolute
values of the HC-OADR shown in the left parts of these figures would be different and
the corresponding trend would run parallel to the one depicted in the figures
(presumably it would be even higher, as excluding the unemployed from the
denominator would decrease it). Thus our conclusions remain fully valid, as they do not
depend on the absolute values of the HC-OADR. Similarly, fixed age at retirement does
not have any effect, whether it is set to the level of 63 or 67 or any other level that
remains unchanged during the projection period. Relatedly a stepwise increase in
retirement age that is enacted in a range of countries (including the Fornero reform in
Italy, see appendix) does not eliminate the influence of the specific increase in human
capital as discussed in this paper. We also assume constant returns to education.
We applied ratios of human-capital-specific gross income and pensions as they
were observed around 2007 and kept them constant during the period of projection. Our
findings do not rely on absolute values because their units of measurement depend on
assumptions. Hence, we used relative values to restrict the effect of these assumptions.
The issue if the results are applicable to other countries remains open, and will be
explored in future research. The Italian case, although atypical in some aspects is not an
outlier in the European context. Similar relationship between the income and the
pensions of high, medium and low educated people are observed in other countries of
Europe. Preliminary analysis based on EU SILC data show that differences in education
affect more markedly pensions, and less income in the same manner as in Italy in Spain,
the Netherlands, Ireland, France and Finland. In some other countries, the pattern is
even more pronounced, like in Portugal, Cyprus or Greece. For example the pensions of
the high educated are about five-time higher than the pensions of the low educated in
Portugal.
We once more emphasise that our findings are not based on forecasts. They rest on
projections that are based on specific assumptions, as is the case with the extensions of
the OADR outlined above. The use of projections permits us to analyse a selected force
of population change while keeping a range of other relevant issues constant.
Besides the „no changes‟ assumption, we did not consider additional important
aspects such as health and social care for the elderly population. Although they are
unrelated to education we can assume that persons with higher human capital are
accustomed to a higher quality of life and request more quality in care. Hence, an
increase in retired human capital may increase the demand for better quality care.
Private pensions were not considered either. It can be argued that they also are
Demographic Research: Volume 31, Article 44
http://www.demographic-research.org 1331
proportional to human capital and the assumption about keeping them constant remains
valid in the same way as the assumptions discussed above. Numerous other issues
related to the increase in the human capital of the elderly population, such as changes in
patterns of consumption, private savings and investment, travel, politics and voting etc.,
remain out of the scope of this paper although they might refer to consequences of
population ageing.
Extending our findings towards relaxing some of the assumptions will stimulate
the debate on the search for demographic and/or socio-economic solutions to the
challenges posed by population ageing. Finally we share a word of caution. In the last
section we discussed many assumptions that relate to the method used, mostly
following on comments raised by colleagues and the reviewers. Most of these
assumptions refer to turning a projection into a forecast, which is not the purpose of this
paper. The long list of assumptions may leave the impression that our indicators are less
parsimonious than preceding indicators. Such an impression is wrong because other
indicators are also subject to many assumptions although they are not explicitly stated.
6. Acknowledgments
Part of this work was accomplished with EU funding for the FP6 MicMac project. We
would like to thank Marco Marsili from ISTAT for sending us most of the demographic
data for Italy. The responsibility for all conclusions drawn from the IT-SILC data lies
entirely with the authors. We are grateful to the following scientists for their most
useful comments on an earlier draft of this paper: Bilal Barakat, Graziella Caselli,
Dalkhat Ediev, Michael Kuhn, Wolfgang Lutz, Klaus Prettner, Warren Sanderson,
Erich Striessnig, and Peter Vanhuysse. Our thanks also go to the participants of several
meetings and to the two anonymous reviewers for their constructive comments. Thanks
are due to Sylvia Trnka who edited the manuscript.
Philipov, Goujon & Di Giulio: Ageing dynamics of a human-capital-specific population
1332 http://www.demographic-research.org
References
Becker, G. S. (1964). Human capital. New York: Columbia University Press.
Börsch-Supan, A. (2003). Labor market effects of population aging. Labour 17: 5–44.
doi:10.1111/1467-9914.17.specialissue.2.
Chawla, M., Betcherman, G., and Banerji, A. (2007). From red to gray: the “third
transition” of aging populations in Eastern Europe and the former Soviet Union.
Washington DC: The World Bank. doi:10.1596/978-0-8213-7129-9.
Crespo Cuaresma, J., Lutz, W., and Sanderson, W.C. (2009). The dynamics of age
structured human capital and economic growth. Paper presented at the New
directions in the Economic analysis of Education. Milton Friedman Institute,
University of Chicago, November 21.
Docquier, F., Lowell, B.L., and Marfouk, A. (2009). A gendered assessment of highly
skilled emigration. Population and Development Review 35(2): 297–322.
doi:10.1111/j.1728-4457.2009.00277.x.
European Commission (2010a). Europe 2020: A strategy for smart, sustainable and
inclusive growth. Brussels: European Commission.
European Commission (2010b). Demography Report 2010, Commission staff working
document. Brussels: European Commission.
Fougère, M. and Mérette, M. (1999). Population ageing and economic growth in seven
OECD countries. Economic Modelling 16(3): 411-427. doi:10.1016/S0264-
9993(99)00008-5.
Goujon, A. (2009). Report on changes in the educational composition of the population
and the definition of education transition scenarios: The example of Italy and the
Netherlands. Extended version of Deliverable D3 in Work Package 1 (Multistate
Methods) of EU (6th Framework) funded project MicMac: Bridging the micro-
macro gap in population forecasting.
ISTAT (2007). Resident population by sex, age and level of education - Italy -
Continuous Labour Force Survey – 2007. Rome: ISTAT.
ISTAT (2008). Previsioni demografiche 1° gennaio 2007-1° gennaio 2051. Rome:
ISTAT.
ISTAT (2011). Il futuro demografico del Paese Previsioni demografiche al 2065:
peggiora la dinamica naturale, sostenute ma in calo le migrazioni con l’estero,
in aumento stranieri e anziani. Rome: ISTAT.
Demographic Research: Volume 31, Article 44
http://www.demographic-research.org 1333
KC, S., Barakat, B., Goujon, A., Skirbekk, V., Sanderson, W.C., and Lutz, W. (2010).
Projection of populations by level of educational attainment, age, and sex for
120 countries for 2005-2050. Demographic Research 22(15): 383–472.
doi.10.4054/DemRes.2010.22.15.
Kemnitz, A. and Wigger. B.U. (2000). Growth and social security: the role of human
capital, European Journal of Political Economy 16(4): 673–683. doi:10.1016/
S0176-2680(00)00020-3.
Lee R. and Mason, A. (2011). Population Aging and the Generational Economy: A
Global Perspective. Northampton: Edward Elgar Publishing.
Lee, R. and Mason, A. (2010). Fertility, human capital, and economic growth over the
demographic transition. European Journal of Population 26(2): 159–182.
doi:10.1007/s10680-009-9186-x.
Lemieux, T. (2003). The “Mincer Equation” thirty years after Schooling, Experience,
and Earnings. Berkeley: Center for Labor Economics, University of California
(Working Paper 62).
Lucas, R.E. (1988). On the Mechanics of Economic Development. Journal of Monetary
Economics 22(1): 3-42. doi:10.1016/0304-3932(88)90168-7.
Lutz W., Crespo Cuaresma, J., and Sanderson W.C. (2008). The demography of
educational attainment and economic growth. Science 319(5866):1047–1048.
doi:10.1126/science.1151753.
Lutz, W., Goujon, A., and Wils, A. (2008). The population dynamics of human capital
accumulation. In Prskawetz A., Bloom, D., and Lutz, W. (eds.). Population
aging, human capital accumulation, and productivity growth. New York:
Population Council.
Lutz, W. and KC, S. (2011). Global human capital: integrating education and
population. Science 333(6042): 587–592. doi:10.1126/science.1206964.
Lutz, W., Sanderson, W.C. and Scherbov, S. (2004). The end of world population
growth in the 21st century: New challenges for human capital formation and
sustainable development. London: Earthscan.
Mankiw, N.G., Romer, D., and Weil, D.N. (1992). A contribution to the empirics of
economic growth. The Quarterly Journal of Economics 107(2): 407–437.
doi:10.2307/2118477.
Mincer, J. (1974). Schooling, experience, and earnings. New York: NBER Press.
Philipov, Goujon & Di Giulio: Ageing dynamics of a human-capital-specific population
1334 http://www.demographic-research.org
Polachek, S. (2007). Earnings over the lifecycle: the mincer earnings function and its
applications. Bonn: Institute for the Study of Labor (IZA Discussion Paper
3181).
Romer, P.M. (1990). Endogenous technological change. Journal of Political Economy
98(5): 71–102.
Sanderson, W.C. and Scherbov, S. (2005). A new perspective on population ageing.
Vienna: Vienna Institute of Demography of the Austrian Academy of Sciences
(European Demographic Research Papers 3).
Sanderson, W.C. and Scherbov, S. (2010). Remeasuring aging. Science 329(5997):
1287–1288. doi:10.1126/science.1193647.
Schultz, T. (1961). Investment in human capital. The American Economic Review
51(1): 1– 17.
Skirbekk, V. (2008). Age and productivity potential: A new approach based on ability
levels and industry-wide task demand. In Prskawetz A., Bloom, D., and Lutz, W.
(eds.). Population aging, human capital accumulation, and productivity growth.
New York: Population Council.
Striessnig, E. and Lutz, W. (2014). How does education change the relationship
between fertility and age-dependency under environmental constraints? A long-
term simulation exercise. Demographic Research 30(16): 465–492. doi:10.4054/
DemRes.2014.30.16.
Vaupel, J.W. and Loichinger, E. (2006). Redistributing work in aging Europe. Science
312 (5782): 1911–1913. doi:10.1126/science.1127487.
Demographic Research: Volume 31, Article 44
http://www.demographic-research.org 1335
Appendix: The Italian context
We briefly outline the Italian context as our results are shaped by past reforms in the
Italian education and pension systems.
Until 1859, compulsory education only comprised two years of elementary school,
from 1877 onwards three and from 1907 four years. It took until 1962-63 before the
norms proposed in the famous 1926 Gentile reform were implemented in the framework
of a new reform and 8 years of schooling (lower secondary education) became
mandatory in Italy. The reform covered all people born in or after 1949, i.e., those who
turn 65 in 2014 or later. Our projection thus captures the massive increase in education
of the cohorts born after 1949, which was, in fact, almost invisible among the elderly
population before 2007. Differently to most other European countries, upper secondary
but not tertiary education steadily increased for the generations born after 1960.
Completion of tertiary education only prevailed among the generations born during and
after the 1970s. The very dynamic and recent evolution of the education system in Italy
makes this country particularly interesting for the topic of our research. We expect the
impact of education on ageing dynamics to be quite different in countries with other
education histories. More specifically, the evolution of the proposed ageing indicators
in the analysed time span will be less dynamic in countries, in which high education
was already widespread in the past decades.
Besides the change in the education system, the recent history of pension reforms
in Italy is one other specificity of the Italian context which needs pointing out. Until
1992, the Italian pension system was based on an earnings-related scheme, in which the
pension level depended on the income the retiree had earned during the last years before
retirement and the pension usually amounted to at least 80% of the last pay. Thereafter,
and especially from 1995 onwards, the Dini reform introduced a gradual transition from
this earnings-related scheme to a contribution-based-system (planned to be completed
after 2030), in which the pension depends on the amount the retiree contributed during
his/her working life. As a result, the pension amounts to 50–60% of employees‟ last
salary and even less for self-employed retirees. The fact that most of the people aged
65+ in 2007 (the base year for our projections) retired under the earnings-related system
might explain why the median pension is comparatively high as compared to the
income in the highest education group. First of all, the highly educated are only a very
small fraction of the total population aged 65+. They may therefore have had jobs,
which were much better paid in the past than nowadays. Secondly, the best-educated in
the age group 20–64 include both very young people, whose earning capacity has not
yet been exhausted, and highly-educated people, who are forced to accept poorly paid
jobs due to difficulties on the labour market.
Philipov, Goujon & Di Giulio: Ageing dynamics of a human-capital-specific population
1336 http://www.demographic-research.org
The new pension reform (known as the Fornero reform) became effective as of 1
January 2012. It incorporated all pensions into the contribution system, raised the
pension age for both women and men to at least age 66 by 2018 and higher thereafter,
indexed the pension age to changes in life expectancy and introduced strong
disincentives for all those who want to retire before the given age limit even if they
have already paid their contributions for the required minimum number of years (42
years + 1 month for men and one year less for women in 2012, more thereafter).
