€¦ · web viewhow does population ageing affect savings? empirical evidence for selected...
TRANSCRIPT
HOW DOES POPULATION AGEING AFFECT SAVINGS?
EMPIRICAL EVIDENCE FOR SELECTED EUROPEAN
COUNTRIES
Abstract
Ageing is going to be greater and greater in the forthcoming decades. At the same time,
old-aged dependency and longevity occur due to rising life expectancy at birth. Our
study aims to examine the effect of population ageing on private savings. A panel data
of a sample of selected European countries collected from the World Bank (WB, 2016) - World Development Indicators (WDI) database is used, in order
to analyze possible heterogeneity across and multiple subsamples. The span period is
1990-2013. Besides, economic growth and crisis provide empirical support. Our
findings show that longevity and dependency rates have both significant impacts on
savings. However, the results are influenced by the way in which the data is managed.
Therefore, this calls into question the practice of applying for a reform in the
government’s old-age support programs. Besides, the paper makes a good contribution
to knowledge: firstly, it is innovative since it puts together demographic and economic
variables among a selected group of developed countries; secondly, it uses a very up-to-
dated database; and thirdly, it fills a gap in the literature.
Keywords Population Ageing, Longevity, Dependency, Savings
JEL Classification J11, E21
1
1. Introduction
According to data from World Population Prospects (United Nations, 2015), the number
of older persons—those aged 60 years or over—has increased substantially in recent
years, and between 2015 and 2030, it is projected to grow by 56%, from 901 million to
1.4 billion. It can be seen that the number of people aged 80 years or over, the “oldest-
old” persons, are growing even faster than the number of older persons overall. Again,
continued improvements in living standards, health styles, and medical care also helped
to lower mortality rate from chronic diseases; these circumstances will result in higher
life expectancy around the world (Guest, 2006; Raftery et al., 2014; Kolasa and
Rubaszek, 2016) and, through channels like fertility and mortality, in opposite effects
on capital accumulation (Lau, 2014).
Nevertheless, there is a close relationship of life expectancy with social and
economic conditions, physical environment and individual lifestyle. For example, if we
focus on socioeconomic variables, the relationship between income (understood as a
measure of socioeconomic status) and health is probably very complicated, depending
on the context and the aggregation level. Even when the positive correlation is clear,
causal interpretations may include income influencing health, health influencing income
and/or “third variables” affecting both indicators in the same direction and at the same
time. However, there are special cases. For example, some southern countries of the
European Union that are relatively poor have a life expectancy indicator greater than
that of the rich countries of northern Europe.
2
Among geographic areas with more of a tradition of this type of applied studies,
we can find Great Britain and its “Black Report” (Black, 1980), which was updated in
the following years. For example, the “Achenson Report” is a continuation of previous
studies, from the perspective of the wide differences in the United Kingdom between
those at the low and those at the high end of the social scale (Rivera et al., 2016). These
differences were registered in stages of prosperity and, at the same time, periods when
there were reductions in the mortality rate across the geographic area considered at an
aggregated level. Hence, in the “Achenson Report” a conceptual structure is presented
that defines the determinants of health in a complex system: income, education,
housing, employment, smoking, alcohol intake and lifestyles.
Based on this view, the “Achenson Report” denotes 39 recommendations, or
priority policy directions, for not increasing health inequalities. This advice is based on
greater income of the poorest to enhance their lifestyle and nutrition and the basic
facilities to give them better health.
Among the most recent papers, there is also a key issue in solving the paradox
that income appears to be related to health within economies but not between them. The
best proxy relies on the fact that in developed countries, which have developed a certain
standard of living, increases in per capita Gross Domestic Product (GDP) make little
difference to the levels of health. The reason for it is the called epidemiological
transition (Mc Keown, 2009). It shows changing trends of population age distributions,
mortality, fertility rates, life expectancy, and death causes. Nevertheless, within
countries, differences in living standards help to create a social order in the population.
3
Under these assumptions, the epidemiological transition lead to the absolute
deprivation loses its relevance and has to be replaced by relative deprivation
(Wilkinson, 1996). This could be a good explanation to understand why, after a given
time, income and health are inversely related in developing countries but lose this
relationship in developed economies.
However, the real nature of the relationship between health and income is still
not clearly defined, because of methodological problems. The academic literature
pointed out a wide range of questions about this relationship and shows the sensitivity
in the different works to the methodology employed. It is important to remember that
the findings depend on a great extent on the kind of indicator used to measure health,
the level of data aggregation and the causal impacts among the variables that we
consider.
The central argument of understanding health indicators is to measure the health
status of a population. This is not an easy issue because there is no comparable health
index among geographic areas. The indicators commonly used, which are available for a
large number of these areas, are mortality rates (infant and adult) and life expectancy.
Nevertheless, these indicators are not sensitive to improvements in quality of life, a
basic issue in developed countries where high levels of health have already been
registered. In spite of these disadvantages, mortality is an indicator widely employed in
academic studies linking income and health, as the information are more readily
available when making comparisons between countries.
4
Given these influences, health surveys collect a wide range of indicators that get
a broader view of health, since they ask about individuals’ opinions of their own health
status, health behaviour and their health care utilization. If individual information is
available, it is possible to make comparisons between several socioeconomic groups.
The main problem stems comes from that these health surveys do not usually collect a
longitudinal follow-up, because usually they are cross-sectional studies. Another
important problem that is attributed to them is that the questions are usually restricted to
a short time period. Besides, these surveys are unrepresentative for certain high-risk
groups or marginalized parts of the society. However, these individual-level data are
advised by several authors when the objective is to study the most advanced and
compelling hypotheses about the relationships between income and health.
Moreover, data aggregation employed in several studies examining the health
status of the population in different countries and its relationship to the income level,
can also cause problems from an empirical point of view. At this regards, a first
problem is the availability of comparable data for long time periods. In this case, the
observations are often measures at national or regional level, in contrast to individual
panel data for which there are a large number of observations of cross-sectional
measurements. Hence, it is true that several problems differ depending on the
observation unit adopted. In other words, the individual or an aggregate geographical
area is the decision to be made.
Another issue of relevance in the causality of the variables that are considered in
the analysis is the relationship between income and health because it is another
methodological aspect that is particularly important. Although several papers show a
5
positive relationship between health and income, few of them are worried about the
causality of this association. Thus, this causality is difficult to test (Babones, 2008). The
author points out that although there is a “strong, consistent and statistically significant
correlation between national income inequality and population health”, it is true that
there is also evidence showing that this correlation is causal.
Moreover, several studies have focus on the interesting relationship between
childhood health and later life earnings. For instance, Behrman and Rosenzweig (2004)
employ a United States sample of female twins and find a (positive) association
between birth weight and hourly wages in mid adulthood (at ages of 39–58 years). In
addition, Almond (2006) using the 1918 Influenza Pandemic as a measure of a health
shock around birth, demonstrates that it reduced annual wage income of United States
men in mid adulthood (ages 40, 50 or 60 year). After that, Black et al. (2007) show
(with information based on administrative data for a sample of Norwegian twins) a
positive association between birth weight and earnings in early adulthood (at ages of
25–35 years). Also, Chen and Zhou (2007), using the 1959-1961 famine in a country
study like China, find only limited evidence of a negative effect on survivors´ earnings
in rural areas in adulthood (at ages of 24–37 years).
Besides, Smith (2009) uses a subsample of United States siblings from the Panel
Study of Income Dynamics (PSID) aged 25–47 to estimate the relationships of both
childhood self-reported health and parental income during childhood. The author show
that about 50% of their overall impact is already present at age 25, while the remaining
50% is the consequence of faster individual income growth after age 25. Nevertheless,
Nelson (2010), using the 1918 Influenza Pandemic, does not find a significant effect on
hourly wages of relatively old males (above age 65) in a country like Brazil.
6
More recent studies have shown the effects of social mobility in the lifecycle
applied to health economics. For example, in Flores et al. (2015), the analysis is
conducted using the Survey of Health, Aging and Retirement in Europe (SHARE),
which includes retrospective information on early life circumstances and full work
histories for over 20,000 individuals in 13 European countries. The authors find a
smaller, positive long-term association between childhood health and lifetime earnings
operating mainly through annual earnings. Besides, most of these life cycle profiles
differ between European country-groups. And for women the authors find a buffering
effect, i.e. that a higher parental SES declines the negative impact of poor health during
childhood on accumulated earnings over the lifecycle.
Papageorge and Thom (2016) using the Health and Retirement Study (HRS) for
the United States, demonstrate that the genetic gradient in wages has grown in more
recent birth cohorts, that is explained by the interactions between technological change
and labour market ability. Thus, individuals who grew up in economically
disadvantaged households are less likely to go to college. Their findings provide support
for the idea that childhood poverty limits the educational attainment of high-ability
individuals suggests the existence of unrealized human potential.
Although previous empirical literature presents different ways of understanding
the evidence, most analyses report that average health is worse in more unequal
societies. However, this relationship is not perfect, since there are several determinants
that can influence it. In addition, countries with lower per capita income levels have
lower mortality rates. This issue could be explained by arguing that while it is true that
some factors (food and housing) are positively associated with a level of income above
7
a certain minimum, there may be others (alcohol, tobacco or drug consumption) that
have the opposite impact.
Nevertheless, population health would also help to disentangle differences in
income levels between individuals and countries. The relevance of investment in health
has been re-called by the theories of human capital. Better health diminishes
productivity losses caused by disease in the workforce, reducing disability, weakness
and the number of days off work. Also, they increase assistance to schools and the
learning capacity of school children.
Hence, the published health economics literature on socioeconomic status and
health is characterized by several papers that show the complexity of those
relationships. Improving this information is basic if we are able to capture the value of
socioeconomic measures, and to understand the most relevant determinants of health
status.
As populations grow more and more aged, it is more relevant than ever that
governments design social policies and public services targeted to older persons in order
to achieve the goals laid out in the 2030 Agenda for Sustainable Development. It is
conceivable that the response of savings to rising longevity has attracted a great deal of
attention among researchers.
Thus, this work is related to the strand of the literature that investigates the joint
effects of higher age dependency and rising longevity on savings. This paper uses a
panel dataset of European selected countries over the period 1990-2013. The main
8
contributions of this empirical study are summarized as follows. Firstly, the main
control variables are consistent and significant. Secondly, longevity has a significant
positive effect on savings, whereas old-aged dependency rate has a negative one.
Thirdly, from an economic policy perspective the aforementioned results would
condition retirement age and/or pensions decisions and policies.
The rest of the paper is organized as follows. The next section presents the
literature review of empirical literature and Section 3 describes the basic framework of
the research. Section 4 presents our main empirical results. The Section 5 discusses our
findings and Section 6 concludes.
2. Review of Empirical Literature
Firstly, with respect to the empirical evidence on health-related growth, the focus of
studies has moved from the exploration of direct impacts to the indirect ones. Thus,
Bloom and Canning (2005) demonstrated that a 1% increase in adult survival rates
enhances labour productivity by about 2.8%. Also, Aghion et al. (2011) found that only
the reduction in mortality rates below forty leads to more productivity growth in
Organization for Economic Cooperation and Development (OECD) countries.
Acemoglu and Johnson (2007) analysed the effect of life expectancy on economic
performance with a model based on a predicted mortality instrument. These authors
found that there is no evidence that the large increase in life expectancy raised per
capita income indicators.
9
In the same line, Bloom et al. (2013) revisited Acemoglu and Johnson (2007)
and showed that their main finding is mostly determined by a priori exclusion of initial
life expectancy. Therefore, Cervellati and Sunde (2011) studied that life expectancy
may have direct impacts on economic growth. These effects seem to be non-monotonic
and it depends on the amount of demographic development. French (2012) positively
tested for some OECD economies that better health indicators improves income while
the latter in turn also affects health status.
Cooray (2013) tested that health capital does not have a significant impact on
economic growth, unless through their relationships with health expenditure and
education. So, Kumar and Chen (2013) demonstrated that health and education
enhanced the growth rate of total factor productivity. The authors focused on the
relevance of including health capital on the policies design, which helps to rise
technology diffusion.
Thus, rising health care expenditures in a framework of an ageing population
had concerned about the sustainability of health care systems due to additional pressure
by introducing drugs and high-cost techniques conditioned by the income possibilities
of each country. Besides, governments` polities to cover the future health care
expenditure of an aging population will likely depend on other factors such as health
supply or innovations in health care delivery that improve cost- effectiveness and trade-
offs among health coverage and taxation.
Overall, much of the empirical literature examining the relationship between
ageing (De Serres et al., 2003; Verbic and Spruk, 2014) and savings (Edwards, 1996)
10
vary both in terms of data set and econometric techniques. Thus, a main issue, often
emphasized in the previous studies, is that in terms of aiding the ageing policy debate,
the Life-Cycle Hypothesis (Modigliani, 1986) assumes that individuals save during their
economically active years to finance their consumption when retirement comes. It is a
well-known result from these findings that the average propensity to consume is greater
in both young and aging individuals, since they are borrowing against future income (in
the case of young people) or employing savings (in the case of retired individuals).
Empirical validation of these findings was demonstrated in the seminal paper of
Leff (1969). Moreover, two main important criticisms of Leff’s results focused on
information quality (Goldberger, 1973) and the pooling of developing and developed
economies in the same data sample (Gupta, 1971; Ram, 1982).
In other words, these latter results are relevant, due to Leff’s findings seems to
be modeled by the developed countries in the data base. More papers to support Leff’s
conclusions could be as follows like Edwards (1996), Masson et al. (1998), Loayza et
al. (2000) and De Serres and Pelgrin (2003).
Nevertheless, it is still usual practice to pool developed and developing countries
in a same group. Therefore, in spite of recent empirical validation of Leff’s results, it is
clearly showed that the earlier criticisms have not been sufficiently applied.
Thus, greater dependency ratio will increase the relative number of non-savers
reducing private saving rates and national savings (Leff 1969; Ram 1982; Masson et al.
1998; Loayza et al. 2000; Kelley and Schmidt, 2005). For example, Li et al. (2007)
11
using panel data methods, found that the impact of rising elderly dependency ratio is
negative on aggregate savings (Goldberger, 1973; Gupta, 1971). Besides, Wong and
Tang (2013) using panel data for 22 OECD countries achieved the same result but
focusing on the fact that it depends on the manner in which the data is handled and/or
the sample is selected.
Nevertheless, longevity-driven ageing has an opposite effect on aggregate saving
rates; nevertheless a significant exception to this phenomenon is analyzed in Feldstein
(1974). The author considers that longevity increasing but retirement ages staying
constant is hold if private savings are declining (Kaier and Müller 2015).
Greater longevity can explain that the elderly people go on saving because of
higher uncertainty about future health care expenditure (De Nardi et al. 2009). In other
words, the retirement’s age may change as life expectancy rises and the positive effect
of longevity on private savings is supported by several studies (Bloom et al., 2003; Lee
et al, 2003a and b; Kinugasa and Mason, 2007; Wong and Tang, 2013). Besides, the
retirement age may not stay constant as life expectancy rises (Blondal and Scarpetta,
1998).
Ando et al. (1995), for instance, show that the elderly in Japan have a high
probability of maintaining employment, and tend not to talk out as much as theory
predicts. Ehrlich and Lui (1991) and Sheshinski (2009) add theoretical support for the
positive effect of longevity on private savings.
12
This implies that old-age dependency exerts a negative impact on savings and it
is sensitive to the way in which the data is managed. Since higher age dependency and
rising longevity have opposite effects on private savings, the overall impact of ageing
will differ across countries and across social security systems.
3. Data and Methodology
In this Section, we explain the data set used in our empirical analysis,
considering the selected set of countries, the sample period, the
variables and the sources of information from which we obtained the
relevant information. Besides, a descriptive analysis is presented in
the following paragraphs.
Basic data for this study is gathered from the World Development
Indicators (World Bank, 2016). The dataset contains a wide range of
variables, such as the ratio of savings to Gross Domestic Product
(GDP), demographic and labor statistics, income or school enrollment.
Thus, the primary World Bank collections of development indicators
are compiled from officially-recognized international sources. It
presents the most current and accurate global development data
base available, and includes national, regional and global estimates.
13
Our empirical results are based on a complete and balanced
panel data set for a selected sample of twelve European Union
countries over the time period 1990-2013 (where all the relevant
variables are available): Austria, Denmark, Finland, France, Germany,
Ireland, Italy, Netherlands, Portugal, Spain, Sweden, and United
Kingdom. We do not include more European countries since there are
not enough observations for all the variables considered in the
studied period. Hence, we have restricted our analysis to the period
that allows us to compare results across countries.
Table 1 summarizes some descriptive statistics for the variables
included in the data set. Gross saving - GDP ratio is the dependent
variable, the conditioning or explicative variables/factors include aged
dependency ratio, life expectancy at birth, fertility rate, annual growth
rate of per capita GDP , labor force participation rate and primary school
enrollment. It can be seen that our sample countries grew, on
average, around 1.39% per capita per annum, with aged dependency
ratio of 23.84% and labor force around 59.1%. Among the
demographic variables, life expectancy yields about 78.51 years
(mean) while the fertility rate is 1.39.
(Insert Table 1)
14
Following Li et al. (2007) and Wong and Tang (2013), the empirical panel specification
for savings (complete model) is formalized as follows:
gns¿=∝0+∝1depold ¿+∝2¿¿+∝3 fertility i ,t−1+∝4 growth¿−1+∝5 labor¿+∝6 primary¿+u¿
(1)
where subscript i refer to countries and subscript t to years. According to
economic criteria and previous evidence, we anticipate that coefficients are as follows:
∝1<0, ∝2>0, ∝3<0, and ∝4>0. Thus, we expect positive signs for longevity and growth
effects, and negative signs for old-age dependency and fertility. Furthermore, other
control variables as labor and primary enrollment rates are included in the analysis in
order to avoid spurious regression.
4. Empirical results
Before presenting the main estimates of the panel data models, we perform some
preliminary tests. Since this work is based on a time-series cross-country panel data we
should analyse the variables involved in in order to ensure that the regressions are not
spurious. Additionally, Figures 1, 2 and 3 in the Appendix show time series plots of the
main variables for the twelve EU countries studied. As expected in these Figures, gross
savings- GDP ratios exhibit long term upward or downward trends that are not uniform
across countries. Hence, the observation that gross savings - GDP ratios are much
volatile series than the demographic ones as aged dependency ratios (% of working-age
population) clearly show that its short-run dynamics are more likely to be influenced by
business cycles than by the slowly evolution of demographic variables. Upward trends
15
are common in old-age dependencies and life expectancies, this being explained by the
population ageing process in Europe.
As a common feature of first generation of panel unit root tests is that they suffer
from a loss of power when individual specific trends are included, then a second
generation CIPS test (Table 2), which assumes cross-section dependence (Pesaran,
2007) is performed. Taking into account lag orders p = 0, 1, 2 and 3, the corresponding
tests show that in most of the cases the variables are I(1). Thus, an implication of these
findings is that we need to model the deterministic trend factor of the time series, as
well as we have to use the first difference of growth in the corresponding regressions.
(Insert Table 2)
Moreover, we present the main findings from the estimation of the panel data
models. Estimates for the full sample of the selected EU countries can be found in Table
3. Results are consistent and significant along the specifications. Interestingly, we report
a saving regression on the main explanatory variables (old-age dependency and life
expectancy), and from these specifications, we add the rest of the above-mentioned
determinants in order to test the robustness of our estimates.
We found a negative effect from old-age dependency on savings which is robust
and significant in all the specifications. Regarding life expectancy, the reverse positive
effect is obtained. Similar results are obtained for the growth variable. However, the
additional control variables are not statistically significant.
16
(Insert Table 3)
Besides, we briefly discuss the robustness of the results
presented above. It is important to indicate that we check the
sensitivity of the estimates in the sample of EU countries used in our
empirical analysis. In sum, we deal with heterogeneity across
countries.
Doing so, we focus on specification (v), the most complete, and
consider three subsamples of the twelve EU selected countries from
macro-areas (based on geographical location). These subsamples are
called Mediterranean countries, Nether-Nordic countries and Anglo-
Saxon countries. Mediterranean countries include France, Italy,
Portugal and Spain; Nether-Nordic countries include Denmark,
Finland, Netherlands and Sweden; and Anglo-Saxon countries include
Austria, Germany, Ireland and United Kingdom. Table 4 shows the
results for the different subsamples and for the full sample.
Findings for Mediterranean and Nether-Nordic countries are
similar to those of the full sample, and these results agree with
previous empirical evidence (Wong and Tang, 2013). However, for the
case of Anglo-Saxon countries, our findings show a turning effect for
old aged and life expectancy.
17
(Insert Table 4)
In addition, we must highlight the significant economic growth
impact on savings that is clearly displayed (Tables 3 and 4). The
savings rise as the level of income per capita growths, which in turn is
esteemed by sample Nether-Nordic countries that get higher
coefficients. These results are in line with previous empirical
evidence.
Altogether, and from a policy economic perspective, the
aforementioned results would provide more informed understanding
for policy makers about condition retirement age and/or pensions
decisions. For example, the generosity of pension systems in
European countries is expected to discourage private savings; and
there are vast differences in social security systems across them. In
this sense, population ageing should be analyzed as a multi-faceted
phenomenon because countries with higher income often have lower
fertility rates and higher life expectancy. The implication is that some
mechanism of reverse causality from savings as a source of future
income to dependency and life expectancy can explain the evidence
that we have found.
5. Discussion
18
It is widely believed that longevity rates are likely to have large impacts on life-cycle
behaviors, i.e., healthier people who live longer have stronger incentives to invest in
developing skill generate more investment or save more. However, rising old-aged
dependency tends to be jointly considered with greater longevity. As both factors are
expected to have opposite impacts on savings, the omission of one of them could not
conduct to a good estimate of the impact of the other on savings. In this paper, we first
construct a sample of EU countries where it is found that longevity, dependency rate
and growth variables have significant effects on savings.
As expected, our model results support findings on savings given in the previous
literature. Nevertheless, a distinctive feature of this study is that the significance of
demographic changes is based on having non-linear effects on saving rates because they
are bounded. The theoretical assumptions are supported by the econometric implication
of our information. In particular, our empirical findings highlight that the trend
properties of the data require econometric treatment and the need to use much more
expansive information than in other previous empirical works where the robustness of
the results is tested. Results for the full sample are compared within three subsamples
(Mediterranean, Nordic, Anglo-Saxon countries) of the twelve EU selected countries.
Thus, healthcare needs will continue putting more pressure on public budgets
over the next decades. Besides, European Union is probably the most influenced by this
problem, with both population aging and health models characterized by basic health
coverage and on cost-sharing systems.
Due to these changes, income covers children from the negative impacts of
adverse childhood health and it is based on the called buffering hypothesis (Currie and
19
Stabile 2003). Moreover, it could be relevant to explore whether there are differences in
the life cycle profiles between European countries, which is what one could expect
given the cross-country differences in levels of socioeconomic conditions over the
period of analysis that we consider.
It is also interesting pointing out some basic limitations of the paper. The first
one concerns endogeneity issues. It is well recognised that economies with higher
income usually have lower fertility rates and higher life expectancy (years). This could
be understood as a chance of some effect of reverse causality from savings to old-age
dependency and life expectancy (years). Thus, it is not quite easy how to enable
instruments for old-aged dependency and life expectancy (years) even if one drops the
basic assumption of exogeneity of these variables.
6. Conclusions and policy implications
Population ageing and savings determinants are likely to remain an interesting area of
debate. In this paper, the effect of population ageing on private savings using a panel
data of a sample of selected European countries collected from the World Bank to
analyze possible heterogeneity across and multiple subsamples for the period 1990-2013
is investigated.
Summarizing, the overall message of our findings is based on the idea that the
results resemble those of the previous literature (De Nardi et al., 2009; Wong and Tang,
2013), in that old-aged dependency has a significant negative impact on savings.
Eventually, savings are also conditioned by institutional factors or policies, like
retirement age and/or pensions.
20
As well, economic growth always plays a major role in these fields. This calls
into question the practice of applying for a reform in the government old-age support
programs.
Nonetheless, we speculate that if income inequality continues increasing, public
policies should better understand how income could help to get a more efficient
allocation of resources. In other words, there is a clear effect regarding the political and
economic processes that generate income which influence population outcomes (ageing
and, consequently, savings). In addition, individual resources may also have indirect
impact on public resources or social welfare (health care, labor market, schooling…).
Hence, several solutions would be needed depending on the country: rich or poor
A question arises about policies that would manage the potential problems
related to ageing. Let us notice that policymakers should reduce negative effects of the
rising population ageing and prevent savings from further cuts. An extension of the
approach should include the analysis if this legal factors affecting savings. It can be
argued that there is a need for a more coordinated pension system within the Euro area
as well as a greater allowance for savings is assumed among these countries. The results
suggest that in designing policies to facilitate savings catch-up process one needs to
broaden the concept of human capital to include ageing.
21
References
Acemoglu, D., & Johnson, S. (2007). Disease and development: the effect of life
expectancy on economic growth. Journal of Political Economy, 115, 925-985
Aghion, P., Howitt, P. & Murtin, F. (2011). The relationship between health and
growth: when Lucas meets Nelson-Phelps. Review of Economics and
Institutions, 1, 1-24
Almond, D. (2006). Is the 1918 Influenza Pandemic over? Long-term effects of in utero
influenza exposure in the post-1940 U.S. population. Journal of Political
Economy, 114, 672–712.
Ando, A., Moro, A., Cordoba, J.P. & Garland, G. (1995). Dynamics of demographic
development and its impact on personal saving: case of Japan. Ricerche
Economiche, 49(3), 179-205.
22
Babones, S.J. (2008). Income inequality and population health: Correlation and
causality. Social Science & Medicine, 66(7), 1614-1626.
Behrman, J.R. & Rosenzweig, M.R. (2004). Returns to birthweight. Review of
Economics and Statistics, 86, 586–601
Black, D. (1980). The Black Report. Penguin Books, London.
Black, S.E., Devereux, P.J. & Salvanes, K.G. (2007). From the cradle to the labor
market? The effect of birth weight on adult outcomes. Quarterly Journal of
Economics, 122, 409–439
Blondal, S. & Scarpetta, S. (1998). The retirement decision. OECD Economic Outlook,
63, 179-192.
Bloom, D.E., Canning, D. & Graham, B. (2003). Longevity and life-cycle savings.
Scandinavian Journal of Economics, 105(3), 319-338.
Bloom, D.E. and Canning, D. (2005). Health and economic growth: reconciling the
micro and macro evidence. CDDRL Working Paper, 42.
Bloom, D.E., Canning, D. & Fink, G. (2014). Disease and development revisited.
Journal of Political Economy, 122(6), 1355-1366.
Carroll, C.D. & Weil, D.N. (1994). Saving and growth: a reinterpretation. Carnegie-
Rochester Conference Series on Public Policy, 40(1), 132-192.
Cervellati, M. & Sunde, U. (2011). Life expectancy and economic growth: the role of
the demographic transition. Journal of Economic Growth, 16, 99-133
Chen, Y. & Zhou Li, A. (2007). The long-term health and economic consequences of
the 1959–1961 famine in China. Journal of Health Economics, 26, 659–681
Cooray, A.V. (2013). Does health capital have differential effects on economic growth?.
Applied Economics Letters, 20, 244-249
23
Currie, J. & Stabile, M. (2003). Socioeconomic status and health: Why is the
relationship stronger for older children?. American Economic Review, 93, 1813–
1823
De Nardi, M., French, E. & Jones, J.B. (2009). Life expectancy and old-age savings.
American Economic Review, 99(2), 110-115.
De Serres, A. & Pelgrin, F. (2003). The decline in private saving rates in the 1990s in
OECD countries: how much can be explained by non-wealth determinants?.
OECD Economic Studies, 36, 117-153.
Edwards, S. (1996). Why are Latin America’s savings rates so low? An international
comparative analysis. Journal of Development Economics, 51( 1), 5-44.
Ehrlich, I. & Lui, F.T. (1991). Intergenerational Trade, longevity and economic growth.
Journal of Political Economy, 99(5), 1029-1059.
Feldstein, M. (1974). Social security, induced retirement, and aggregate capital
accumulation. Journal of Political Economics, 82(5), 905-926.
Flores, M.., Garcia-Gomez, P. & Kalwij, A. (2015). Early life circumstances and life
cycle labor market outcomes. Tinbergen Institute DP 2015-04/V.
French, D. (2012). Causation between health and income: a need to panic. Empirical
Economics, 42, 583-601
Goldberger, A. (1973). Dependency rates and savings rates: further comment. American
Economic Review, 63(1), 232-233.
Guest, R.S. (2006). Population ageing, social mobility and capital saving. Journal of
Policy Modelling, 28(1), 89-102.
Gupta, K.L. (1971). Dependency rates and savings rates: comment. American Economic
Review, 61(3), 469.
24
Haque, N., Pesaran M.H. & Sharma, S. (1999). Neglected heterogeneity and dynamics
in cross-country savings regressions. IMF Working Paper, 99(128).
Kaier, K. & Müller, C. (2015). New figures on unfunded public pension entitlements
across Europe: concept, results and applications. Empirica, 42(4), 865-895.
Kolasa, A. & Rubaszek, M. (2016). The effect of ageing on the European economies in
a life-cycle model. Economic Modelling, 52A, 50-57.
Kumar, A. & Chen, W. (2013). Health, education and the dynamics of cross-
country productivity differences. Applied Economics Letters, 20, 1160-1164
Lau, S.P. (2014). Fertility and mortality changes in an overlapping-generations model
with realistic demography. Economic Modelling, 38, 512-521.
Lee, R.D., Mason, A.W. & Miller, T. (2003a). Saving, wealth and the transition from
transfers to individual responsibility: the cases of Taiwan and the United States.
Scandinavian Journal of Economics, 105(3), 339-358.
Lee, R.D., Zhang, J. & Zhang, J. (2003b). Rising longevity, education, savings, and
growth. Journal of Development Economics, 70(1), 83-101.
Leff, N.H. (1969). Dependency rates and savings rates. American Economic Review,
59(5), 886-896.
Loayza, N.V., Schmidt-Hebbel, K. & Serven, L. (2000). What drives private savings
across the world. The Review of Economics and Statistics, 82(2), 165-181.
McKeown, R.E. (2009). The epidemiologic transition: changing patterns of mortality
and population dynamics. Americal Journal Lifestyle Medicine, 3(1 Suppl), 19S-
26S.
Masson, P..R., Bayoumi, T. & Samiei, H. (1998). “International evidence on the
determinants of private saving”. World Bank Economic Review, 12(3): 483-501.
25
Modigliani, F. (1986). Life cycle, individual thrift, and the wealth of nations. American
Economic Review, 76(3), 297-313.
Nelson, R.E. (2010). Testing the Fetal Origins Hypothesis in a developing country:
Evidence from the 1918 Influenza Pandemic. Health Economics, 19, 1181–1192.
Papageorge, N.W. & Thom, K. (2016). Genes, Education, and Labor Market Outcomes:
Evidence from the Health and Retirement Study. IZA Discussion Paper No.
10200, September 2016.
Raftery, A.E., Alkema, L. & Gerland, P. (2014). Bayesian Population Projections for
the United Nations. Statistical Science, 29(1), 58-68.
Ram, R. (1982). Dependency rates and aggregate savings: a new international cross-
section study. American Economic Review, 72(3), 537-544.
Rivera, B., Casal, B., Lago, S., Cantarero, D., Pascual, M., Reyes, F. & Blazquez, C.
(2016). A systematic review of the impact of inequalities on Non-Communicable
Diseases. Working Paper FRESHER project H2020
Sheshinski, E. (2009). Longevity and aggregate savings. Centre of the Study of the
Rationality Discussion Paper No. 519.
Smith, James P. 2009. The impact of childhood health on adult labor market outcomes”.
Review of Economics and Statistics, 91, 478–489
United Nations (2015). World population ageing report. United Nations, New York.
Verbic, M. & Spruk, R. (2014). Aging Population and Public Pensions: Theory and
Macroeconometric Evidence. Panoeconomicus, 61(3), 289-316.
Wilkinson, R. G. (1996). Unhealthy societies: the afflictions of inequality. Routlegde,
London.
Wong, B. & Tang, K.K. (2013). Do ageing economies save less? Evidence from OECD
data. International Journal of Socio Economics, 40(6), 591-605.
26
World Bank (2016). World Development Indicators. The World Bank.
Table 1
Variables and summary statistics
Variable Mean Standarddeviation
Minimum Maximum
gns (gross savings-GDP ratio, dependent variable) 22.973 4.547 10.295 34.161
depold (aged dependency ratio (people older than 64 to working-age population, those ages 15-64))
23.846 3.604 15.249 33.446
le (Life expectancy (years)) 78.512 2.007 73.966 83.078fertility (fertility rate: total births per woman) 1.621 0.251 1.160 2.130
growth (annual growth rate of per capita GDP) 1.391 2.576 -9.109 9.593
labor (labor force participation rate) 59.104 4.887 47.100 68.300
primary (primary school enrollment)104.52
9 5.536 95.712 123.212
27
Source: Authors’ own elaboration.
Table 2
Second generation CIPS test: Pesaran (2007)
VariableINTERCEPT ONLY
number of lags0 1 2 3
gns 1.917 2.270 2.522 3.760depold 7.711 -3.775 *** -0.906 -0.534le -0.572 0.346 -1.199 0.101fertility 0.217 -1.436 * -0.439 -0.506growth -4.848 *** -2.974 *** -1.473 ** 0.447labor -1.964 ** -2.002 ** -2.938 *** -0.914primary 1.135 0.245 0.490 2.130
VariableINTERCEPT and TREND
number of lags0 1 2 3
gns 1.606 2.278 2.813 3.240depold 10.053 -1.574 * 1.547 2.696
28
le 1.309 3.138 -0.572 1.631fertility -1.027 -2.544 *** -1.267 -0.553growth -3.178 *** -2.614 *** 0.262 2.224labor 0.162 -0.160 0.092 1.462primary 2.740 1.285 0.835 3.771
Notes: Null hypothesis CIPS: series are I(1). ***, **, and * denote statistical significance at 1%, 5%, and 10%.
Table 3
Saving regressions (fixed effects)
Variable / Specification (i) (ii) (iii) (iv) (v)
depold-0.245 *** -
0.314*** -0.284 *** -0.278 ** -0.299 **
(3.26) (2.59) (2.47) (2.40) (2.42)
le0.110 0.382 *** 0.395 *** 0.460 ***
(0.72) (2.57) (2.62) (2.55)
fertility-1-1.349 -1.178(0.75) (0.63)
growth-10.583 *** 0.570 *** 0.570 ***
(9.41) (8.82) (8.78)
labor-0.077(0.75)
29
primary0.011(0.21)
R-squared 0.043 0.051 0.157 0.143 0.100Source: Authors’ own elaboration.
Notes: Notes: z-statistics in parentheses. ***, **, and * denote statistical significance at 1%, 5%, and 10%
respectively.
Table 4
Sensitivity to subsamples of countries: Saving regressions (fixed effects)
Variable / Specification
Full set of sample countries
Sample Mediterranean countries
Sample Nether-Nordic
countries
Sample Anglo-Saxon
countries
depold-0.299 ** -0.093 -1.500 *** 0.601 ***
(2.42) -0.320 (6.28) (4.59)
le0.460 -0.616 2.397 *** -0.450 **
(2.55) *** -1.290 (8.82) (2.08)
fertility-1-1.178 3.562 -2.692 -16.229 ***
(0.63) -0.930 (1.04) (4.61)
growth-10.570 *** 0.274 ** 0.426 *** 0.394 ***
(8.78) -2.710 (5.09) (5.07)
30
labor-0.077 -0.070 0.151 0.379 ***
(0.75) -0.420 (1.03) (2.64)primary 0.011 0.240 -0.050 0.125
(0.21) -1.920 (0.72) (1.24)R-squared 0.100 0.001 0.513 0.007
Source: Authors’ own elaboration.
Notes: Notes: z-statistics in parentheses. ***, **, and * denote statistical significance at 1%, 5% and 10%
respectively.
Appendix
Figure 1
Gross Savings GDP ratio
31
1020
3040
1020
3040
1020
3040
1990 2000 2010 2020 1990 2000 2010 2020 1990 2000 2010 2020 1990 2000 2010 2020
1 2 3 4
5 6 7 8
9 10 11 12
GN
S
yearGraphs by country
Source: World Bank data indicators.
Notes: Austria (1), Germany (2), Denmark (3), Spain (4), Finland (5), France (6), United Kingdom (7),
Ireland (8), Italy (9), Netherlands (10), Portugal (11) and Sweden (12).
Figure 2
Aged dependency ratio
32
1520
2530
3515
2025
3035
1520
2530
35
1990 2000 2010 2020 1990 2000 2010 2020 1990 2000 2010 2020 1990 2000 2010 2020
1 2 3 4
5 6 7 8
9 10 11 12
DEPO
LD
yearGraphs by country
Source: World Bank data indicators.
Notes: Austria (1), Germany (2), Denmark (3), Spain (4), Finland (5), France (6), United Kingdom (7),
Ireland (8), Italy (9), Netherlands (10), Portugal (11) and Sweden (12).
Figure 3
Life expectancy
33
7580
8575
8085
7580
85
1990 2000 2010 2020 1990 2000 2010 2020 1990 2000 2010 2020 1990 2000 2010 2020
1 2 3 4
5 6 7 8
9 10 11 12
LE
yearGraphs by country
Source: World Bank data indicators.
Notes: Austria (1), Germany (2), Denmark (3), Spain (4), Finland (5), France (6), United Kingdom (7),
Ireland (8), Italy (9), Netherlands (10), Portugal (11) and Sweden (12).
34