Académique Documents
Professionnel Documents
Culture Documents
Editors
Social Exclusion
Short and Long Term Causes
and Consequences
Prof. Giuliana Parodi Asst. Prof. Dario Sciulli
Dipartimento di Metodi Quantitativi Dipartimento di Metodi Quantitativi
e Teoria Economica e Teoria Economica
Universit G. dAnnunzio Universit G. dAnnunzio
di Chieti-Pescara di Chieti-Pescara
Viale Pindaro 42 Viale Pindaro 42
65127 Pescara 65127 Pescara
Italy Italy
parodi@unich.it d.sciulli@unich.it
ISSN 1863-916X
ISBN 978-3-7908-2771-2 e-ISBN 978-3-7908-2772-9
DOI 10.1007/978-3-7908-2772-9
Springer Heidelberg Dordrecht London New York
This book contains a collection of papers, most of which have been presented at
the thematic session on Social Exclusion of the XXV Annual Conference of the
Italian Association of Labour Economists (AIEL), held at the University
G. dAnnunzio of Chieti-Pescara in September 2010.
All the papers, before being published, have been submitted to a double blind
peer-review process. This has been possible thanks to the contributions of the
referees: all the papers have benefited from their valuable suggestions.
We are also indebted with various institutions that have provided us the
financial support that has made possible this publication. We would like to thank:
the Italian Association of Labour Economists, the Camera di Commercio, Indu-
stria, Artigianato e Agricoltura di Chieti, the Camera di Commercio, Industria,
Artigianato e Agricoltura di Pescara, the Dipartimento di Metodi Quantitativi
e Teoria Economica of the Universit G. dAnnunzio di Chieti-Pescara, the
Facolt di Scienze Manageriali of the Universit G. dAnnunzio di Chieti-
Pescara and the Universit G. dAnnunzio di Chieti-Pescara.
Referees
Luca Beltrametti, Universit di Genova, Italy
Roger A. Bowles, University of York, UK
Floro Ernesto Caroleo, Universit Parthenope di Napoli, Italy
Sergio Destefanis, Universit di Salerno, Italy
Donata Favaro, Universit di Padova, Italy
Renata Livraghi, Universit di Parma, Italy
Emanuele Millemaci, Universit di Messina, Italy
Catia Nicodemo, Universitat Autonoma de Barcelona, Spain
Carmen Pagliari, Universit G. dAnnunzio di Chieti-Pescara, Italy
Giuliana Parodi, Universit G. dAnnunzio di Chieti-Pescara, Italy
Matteo Picchio, Universiteit van Tilburg, The Netherlands
Cristina Salvioni, Universit G. dAnnunzio di Chieti-Pescara, Italy
Dario Sciulli, Universit G. dAnnunzio di Chieti-Pescara, Italy
v
vi Acknowledgments
vii
viii Contents
ix
x Contributors
The XXV Conference of AIEL (Associazione Italiana degli Economisti del Lavoro)
was held in the Universit dAnnunzio in Chieti-Pescara in September 2010. The
theme chosen for the thematic session was Social Exclusion, in line with the
choice made by the European Union to make 2010 the European year to fight
poverty and social exclusion. The European Commission defines social exclusion
for societies and for individuals: for societies: disintegration and fragmentation of
social relations and hence of social cohesion for individuals a progressive
process of marginalization leading to economic deprivation and various forms of
social and cultural disadvantage. From another perspective, the socially excluded
can include individuals who are unable to participate in the basic economic and social
activities of the society in which they live. In the Conference, the themes under
investigation have ranged from causes to consequences of social exclusion, focusing
on disability, gender gaps, youth, income distribution, discrimination, accidents at
work, and literacy rates (LR). Selected works that were presented at the Conference
have been invited to be part of this book, to which distinguished authors, not present
at the Conference, have also been invited to contribute.
The book is organised in three parts: the first and the second parts deal with
structural, long-term causes of social exclusion (i.e., health, education, disability,
G. Parodi and D. Sciulli (eds.), Social Exclusion, AIEL Series in Labour Economics, 1
DOI: 10.1007/978-3-7908-2772-9_1, Springer-Verlag Berlin Heidelberg 2012
2 G. Parodi and D. Sciulli
crime); the third part deals with short-term causes of social exclusion and, in
particular, the effects of the most recent economic crisis. In most of the papers,
social exclusion is assessed in terms of unemployment, income, and education.
The first part deals with the long-term, structural causes of social exclusion and,
in particular, the problems associated with its possible effects on health (Coppola),
education (Bucciarelli, Pagliari, Muratore, Odoardi, and also Verashchagina), and
work conditions (Nissi and Rapposelli). The second part deals with vulnerable
groups who are particularly at risk of social exclusion [i.e., former convicted
persons (Bowles) and disabled persons (Sloane and Jones, Agovino and Parodi)].
Finally, the third part discusses problems connected with the mechanisms of the
labour market, one of the main direct or indirect causes of social exclusion, with
special emphasis on the effects of the most recent economic crisis on social
exclusion, including unemployment (Tronti and Gatto), poverty (Addabbo, Garca-
Fernndez, Llorca-Rodrguez and Maccagnan), and transitions within the labour
market (Lilla and Staffolani and also Lucarelli and Mussida).
In what follows, we briefly discuss the approaches used to investigate social
exclusion; we also revise the main effects of the most recent economic crisis on the
labour market and policy measures put (or not put) into effect to overcome its
effects. We then present a brief summary of the contributions that appear in the
three parts of the book. Finally, we draw some conclusions.
Social exclusion is a term that has jumped to the forefront of the literature since it
was first used in France in the 1970s. Since then, it has been widely adopted as a topic
of economic policy, both by governments and by researchers; for example, in 1997,
the British government set up a Social Exclusion Unit (SEU), and the London School
of Economics set up the Research Centre for the analysis on Social Exclusion
(CASE); in 2007, the Social Inclusion Unit was set up in Australia; within the
European Union, the definition of social exclusion has been sharpened over time
(Nolan and Whelan 2010).
Social exclusion is interpreted both as a static situation of deprivation and as a
process leading to it. Multidimensional poverty and social exclusion are closely
related concepts, but they do not overlap. Both of these concepts refer to Sens
intuition of capability failures, but in the multidimensional approach to poverty,
failure is viewed in terms of shortfalls from a threshold for each function, and in the
social exclusion approach, failure is interpreted as an inability to participate and,
therefore, identifies functioning failures. These two different approaches refer to
quantitative and qualitative aspects, respectively, of social exclusion (Chakravarty
and Zoli 2009).
Initially, the threshold was defined in terms of income. Social exclusion came
very close to the definition of poverty; subsequently, it was defined in terms of
observed and unobserved heterogeneity. At the individual level, shortfalls in the
1 Introduction and Overview 3
Leschke and Watt (2010), focusing on the role of the 20082009 economic
crisis on incomes, public budget and labour markets, have highlighted the impli-
cations of economic crisis regarding social effects. According to the
European Commission statistics (2010), more than 80 million Europeans,
including 20 million children, live at risk of poverty and social exclusion. The
economic crisis has made things worse by putting an estimated 5 million more
people out of work, potentially exposing many more families to poverty and social
exclusion.
European labour markets were strongly affected by the 20082009 economic
crisis; it caused a severe output decline that was quickly transmitted to the labour
market, producing a reduction in employment and working hours and, in many
countries, a strong increase of unemployment rates and underemployment. These
effects represent a challenge for labour markets and for the social policies of
European governments to mitigate the welfare loss of their citizens. Many
European countries entered the recession with low unemployment rates, partly
because of the effects of structural labour market reforms implemented since the
1980s. In any case, unemployment has increased strongly in the European Union
during the 20082009 economic downturn, reaching 9.6% in March 2010
(EUROSTAT 2010). However, the rise in unemployment has been quite hetero-
geneous across European countries. According to the OECD (2010) analysis, job
losses were particularly high in countries (e.g., Spain, Ireland, USA) where the
housing market contributed the most to the economic downturn. On the contrary,
unemployment has only slightly increased in countries where the economic crisis
was mainly explained by a decline in exports (e.g., Germany, the Netherlands,
Slovak Republic, and Japan). Moreover, job losses have been distributed hetero-
geneously among worker groups and industries. The economic crisis has more
strongly affected construction, mining and manufacturing sectors; young,
low-skilled and temporary workers have been the most affected at the individual
level by the economic downturn. Statistics indicate that youth unemployment is
two to three times greater than the average unemployment rate, and this effect
is especially prevalent among low-qualified workers, for which exit from the
labour market and/or the loss of earnings should be carefully avoided. Similar
considerations apply to immigrants.
Heterogeneous effects at worker level also reflect the structure of the economy
and labour market institutions. For example, Bentolila et al. (2010) have found that
Spain would have avoided 40% of new unemployment if it had adopted French
employment protection legislation. It follows that, while flexibility policies have
strongly contributed to increase employment in Europe until the economic
recession, flexible labour markets during economic crises are associated with
higher unemployment rates, especially for specific worker groups (e.g., youth
unemployment).
The 20082009 recessions has also affected the unemployment dynamics of
many European countries; it has caused an increase in unemployment inflows and
a reduction in unemployment outflows, contributing to an increase in the unem-
ployment pool and in the average duration of unemployment. In this case as well,
6 G. Parodi and D. Sciulli
We now briefly summarise the content of the book, presenting the summary of the
contributions according to the order in which they appear: (Sect. 1.4.1) deals with
general causes of social exclusion (i.e., health, education, and work conditions);
(Sect. 1.4.2) deals with some vulnerable groups who are most likely to be
1 Introduction and Overview 7
associated with social exclusion (i.e., former criminals and disabled people); and
(Sect. 1.4.3) deals directly with the impact that the most recent crisis has had on
social exclusion through the labour market.
Coppolas model predicts that the effect of income growth on health can be
ambiguous, providing an explanation to the observed health poverty trap. This
model provides micro foundations to Solows model of economic growth by
linking growth to health via considering the effect of lifestyle on growth. This
model centres on the concept of the lifestyle returns to scale that expresses the
elasticity of health with respect to income: a bad lifestyle implies that an increase
in income increases the consumption of commodities that are pernicious to health.
At the aggregate level, the Growth Model presented uses the Solow Growth
Model with a constant saving rate, diminishing returns of capital and labour,
labour augmenting technology, and constant returns to scale. The model develops
a health multiplier, on the assumption that health is labour augmenting; therefore,
individual choices about health affect society as a whole via externalities. How-
ever, lifestyle can also be an important channel to transmit economic growth to
health, depending on technological progress. Unlike economic growth, the rate of
health growth may be positive, null or negative, depending on the sign of the
Lifestyle Returns to scale: if it is negative, economic growth negatively affects
health.
Bucciarelli, Pagliari, Muratore, and Odoardi address the important issue of
investigating the connection between social exclusion and LR by analysing data
across 30 very diversified countries, some of which are OECD members, col-
lected for the period between 2007 and 2009. The topic under investigation is
important, as the LR and variables related to social exclusion can be seen as
proxies of human and social capital, which are essential for long-term devel-
opment. Knowledge of the variables that affect education is essential for poli-
cymaking purposes.
The paper has a comprehensive description of the statistical methodology that is
used in the investigation (i.e., multivariate regression analysis, maximum likeli-
hood with VARIMAX rotation, factor transformation matrix, factor analysis and
multivariate regression models with maximum likelihood components, and cluster
analysis). The paper is divided into several sections. Initially, it analyses LR as a
function of three groups of variables: variables strictly related to the educational
situation, i.e., school enrolment at three different levels and children out of school;
public intervention in education, expressed both in terms of general public
expenditures and in terms of GDP; and macro variables, such as real growth rate
and long-term unemployment rate. Because of the heterogeneity of the data, links
8 G. Parodi and D. Sciulli
among the variables are not very evident. But, after applying the various methods
of analysis, it is found that LR can be affected by three components with the nature
of social exclusion variables, i.e., a synthetic indicator of social exclusion, an
indicator regarding educational level, and the GDP real growth rate. Most of the
countries analysed, including almost all of the OECD ones, show similar levels of
macro variables related to education, and instead differ on social exclusion. This
result means that all countries with medium or high levels of income per capita
show good levels of enrolment and attendance at school but that not all of them
can deal effectively with social problems. Finally, a detailed discussion of policy
towards literacy in the USA and Italy is discussed.
Veraschagina addresses the question of the role of education on social mobility
in 12 post-communist countries of Central Europe and the former Soviet Union.
Three hypotheses are tested: during the socialist era, the relation between the
education level of parents and their children weakened; the current levels of
educational mobility in post-communist countries are higher than in their Western
counterparts; and the transition to market economy caused an increase in educa-
tional persistence. Despite heterogeneity among countries, educational persistence
appears to decline sharply until the 1950s; however, evidence from later years does
not confirm this trend. To test the effect of this transition, identification of the
break was necessary. The CUSUM and the Chow test were used for this purpose,
and they showed that possibly not a single break occurred throughout the countries
but that additional breaks in the data referring to educational persistence took
place, on average about 10 years before the transition itself occurred. Moreover,
lower educational persistence in post-communist versus mature countries is con-
firmed by the evidence only for the generation born in the 19501960s. The
question of the sustainability of educational mobility was also investigated: the
increase of private returns to education (PRE), which the data show after the
transition, is expected to be associated with higher educational mobility.
This result is not confirmed by the evidence. A possible explanation is the shift in
the remuneration system, heavily relying on in kind transfers in the pre-transition
period, towards a completely monetary system of remuneration after the transition.
Taking this aspect into account, it is possible that the PRE has actually not been
altered. Unfortunately, the data do not allow an assessment of in kind transfers, so
the explanation provided here is a convincing speculation.
Nissi and Rapposelli apply the nonparametric technique Data Envelopment
Analysis to investigate comparative efficiency in terms of low numbers of acci-
dents at work to EUROSTAT data collected for 15 European countries in 2005, in
the sectors of manufacturing, construction and distribution trades. These sectors
were chosen because Europe registers the greatest number of accidents. Under the
traditional hypothesis of variable returns to scale, and under the specific
assumption of an input oriented approach, the paper develops the idea of unde-
sirable outputs, which only recently has been introduced in the literature and which
has mainly been applied to environmental studies. The results of this analysis show
that countries differ from each other with respect to efficiency in a different way,
according to the sector considered. In the manufacturing sector, four countries are
1 Introduction and Overview 9
efficient (i.e., Germany, Ireland, Luxemburg, and the United Kingdom). They
make a total of four efficient countries in the construction sector as well; in the
distribution trades sector, Belgium and France join the other countries on the
efficient frontier, and Germany returns to being efficient, making a total of seven
efficient countries. The overall conclusion suggests that in the construction and
manufacturing sectors, the efficiency of reducing accidents at work can improve
considerably, while in the distribution trade sector, nearly 50% of the countries
considered are already efficient. Ireland, Luxemburg, and the United Kingdom are
efficient in all sectors considered, and Ireland and the United Kingdom are fre-
quently cited as exemplary efficient Decision Making Units, as they have a low
number of accidents at work.
The fragile groups considered in this part of the book include former convicted and
disabled people. Bowles uses various cross-sectional and longitudinal British data
sets to explore the relationship between crime and social exclusion, concentrating
the investigation on the human capital approach model of offending. According to
this model, crime is the result of a very high discount rate, leading to underin-
vestment in oneself. High discount rates can be explained in terms of a lack of
planning skills and an incapacity to anticipate the future, with special reference to
an inability to assess the dynamic consequences of offending. The formation of
this kind of individual preferences may be explained by variables that are inter-
twined with the concept of social exclusion. This model is well suited to explain
the offending behaviour of young people, who actually represent a large majority
of offenders. The variables that the data show to be correlated with offending are
experience of disadvantage as children, poor attendance at and exclusion from
school, exclusion from mainstream education, leaving school by the minimum
school leaving age, pre-custody drug use, and other crime victimisation rates.
Furthermore, reconviction rates appear to be related to prospects of unemployment
after release, and to problems of accommodation; the probability of reconviction
falls with age. There is also evidence of a comparatively high incidence of mental
health issues among prisoners. Of course, with all these factors, one cannot say
whether they are contributory factors to crime, or consequences of it, or whether
they both have a common cause. The policy implications of this model go well
beyond those of the punitive model of crime, and recommend both prevention
policies aimed at raising aspirations and encouraging investment in human capital,
and resettlement policies. A problem connected with this model is that the
predicted benefits of intervention spill over a wide range of potential savings in
costs, which affect various government departments, and it may be difficult to take
all of them into account if, for the purpose of financing, the benefits have to be
assessed in terms of crime prevention only.
10 G. Parodi and D. Sciulli
Sloane and Jones present a very thorough review of the literature on disability
that is likely to become a standard tool of reference in the field. They pay par-
ticular attention to disabled people and the labour market, commenting on a
variety of datasets, ranging from the European ECHP and the Australian HILDA to
the British BHPS. The review touches on most aspects related to disability:
variability of the percentage of people who declare themselves disabled, the
probability of employment and seriousness of disability, evidence on disability and
low income from work, both with cross-sectional and with longitudinal data, and a
discussion of additional expenditure requirements for disabled people. The defi-
nitional aspect of disability is thoroughly investigated, with special emphasis on
work limited disability, non-work-limited disability, and the non-disabled. Other
issues also thoroughly investigated are problems associated with the age of onset
of disability, with self-reporting disability, with the simultaneous relations
between health and labour supply and vice versa, and with state dependence.
A thorough review of the literature using a UK background is also provided. The
problem of possible discrimination against disabled persons is investigated, with
special attention on disability and productivity, the importance of education and
disability on employment and earnings, work limited and non-work-limited
disability, part-time employment and self-employment, gender, possible job
mismatches, and job satisfaction. A thorough review of policies related to
disability, including income support, employment quota systems, sheltered
employment, anti-discrimination legislation, and employment policies, is also
provided. Individually targeted employment policies are presented as the most
promising policies to help disabled individuals to retain and gain employment for
which they are most suited.
Agovino and Parodi investigate whether Italian civilian invalidity pensions are
used as a form of income support, analysing data at the level of individual
provinces. Their analysis proceeds in two steps: a panel analysis and a Gen-
eralized Method of Moments where the spatial variable is introduced. The first
step of the analysis aims at identifying possible differences in disabling health
conditions in various Italian provinces. For this purpose, information about the
discharges from hospitals of people suffering from possibly invalidating illnesses
is analysed. The results of this analysis show that there is no significant dif-
ference in this variable across provinces. Subsequently, the analysis is developed
on the link between civilian invalidity pensions and indicators of social malaise
such as poverty. The poverty rate appears always significant as a regressor of
civilian invalidity pensions, even when the spatial lag is introduced in order to
take into account similarities between provinces which may go well beyond
administrative boundaries, because of historical and socio-economic factors.
These findings reveal a degree of flexibility in the allocation of civilian invalidity
pensions, which appears to be inconsistent with the strict rules defined by the
law, and suggest discretionary interpretation according to the locally prevailing
socio-economic conditions. The unsuitability of civilian invalidity pensions as a
form of income support is stressed.
1 Introduction and Overview 11
Papers dealing with the effects of the 20082009 economic downturn on labour
markets and its socio-economic consequences include some case-studies from Tronti
and Gatto, Addabbo, Garcia-Fernandez, Llorca-Rodriguez and Maccagnan (Add-
abbo et al. hereafter), Lilla and Staffolani and, finally, Lucarelli and Mussida. Spe-
cifically, Tronti and Gatto investigate the 20082009 recession in the context of the
Italian labour market, focusing especially on the problems that concern the unem-
ployment measurement and its consequences in understanding the phenomena.
Addabbo et al. study the impact of unemployment on the probability of being income
poor and on the difficulties of accessing medical and dental visits treatments in Italy
and Spain to draw lessons that can be extended to the economic crisis. Finally, both
Lilla and Staffolani and Lucarelli and Mussida, focus on unemployment dynamics in
Italy clarifying the role of the 20082009 recession on labour market transitions.
Tronti and Gatto stress the relevance of considering various unemployment
measures to avoid shortcomings in evaluating the socio-economic impact of the
20082009 recession. According to official statistics, according to the standards
defined by the International Labour Office, the unemployment rate has risen to
8.6% in 2010, as a consequence of the strong economic downturn. Nevertheless,
the actual impact of the recession may be greater if the meaning of unemployment
is extended to include the concept of the labour underutilisation rate. Specifically,
Tronti and Gatto show that, considering relevant phenomena such as the dis-
couragement effect, with specific STW schemes providing for working zero hours
(Cassa Integrazione Guadagni), labour hoarding and semi-employment (atypical
workers), the resulting labour underutilisation rate is almost double the standard
unemployment rate. This finding has relevant implications when we consider the
impact of the crisis on the socio-economic conditions of individuals and house-
holds. Moreover, it changes the debate about the adequacy of official definitions of
unemployment in providing a reliable measure of the phenomenon and its
implications (see also Brandolini et al. 2006).
Addabbo et al. analyse the effect of being unemployed on the probability of
being income poor and on the difficulties of accessing medical and dental visits
treatments in Italy and Spain using 2007 IT-SILC/ES-SILC data. The authors also
describe the different unemployment insurance systems and the effect of the
economic crisis on both labour markets. Overall, Addabbo et al. find that being
unemployed increases the probability of being income poor and, partially, the
difficulties of accessing to medical and dental treatments in both countries. In Italy,
the effect is stronger for the previously self-employed now unemployed while in
Spain it is stronger for the unemployed who were never employed before. These
findings are relevant in analysing the possible effects of the 20082009 economic
downturn on households well being, considering the structure of welfare systems
with specific attention to the unemployment insurance systems.
Unemployment dynamics in Italy during the last few years have been studied in
two papers. The evidence from both of the papers is quite consistent, even though
12 G. Parodi and D. Sciulli
1.5 Conclusions
The contributions to the book have emphasised the circular aspects of social
exclusion. Factors that generate social exclusion are themselves an effect of social
exclusion. Groups that are traditionally considered fragile, such as disabled people
1 Introduction and Overview 13
References
Adato M, Carter MR, May J (2004) Sense in sociability? social exclusion and persistent poverty
in South Africa, University of Wisconsin-Madison, Department of Economics and Agricul-
tural Economics, Staff paper, No. 477
Bentolila S, Cahuc P, Dolado JJ, Le Barbanchon T (2010) Two tier labour markets in the great
recession: France vs. Spain. Paper presented at the XXV simposio de la asociacin espanola
de economia, Madrid, December 2010
Brandolini A, Cipollone P, Viviano E (2006) Does the ILO definition capture all unemployment?
J Eur Econ Assoc 4(1):153179
Chakravarty SR, DAmbrosio C (2003) The measurement of social exclusion, DIW, German
Institute for economic research, Berlin, Discussion papers 364
Chakravarty SR, Zoli C (2009) Social exclusion orderings, Working paper series, University of
Verona
European Commission (2010) Combating poverty and social exclusion: a statistical portrait of the
European union 2010. Eurostat statistical books, http://ec.europa.eu/social
EUROSTAT (2010) Official statistics. http://ec.europa.eu/eurostat
ISTAT (2011) www.istat.it
Leschke J, Watt A (2010) How do institutions affect the labour market adjustment to the
economic crisis in different EU countries, ET UI working paper no. 2010-04, European trade
union institute
Nolan B, Whelan CT (2010) Using non monetary deprivation indicators to analyze poverty and
social exclusion: lessons from Europe? J Policy Anal Manage 29(2):305325
OECD (2009) Employment outlook: tackling the jobs crisis. Paris
OECD (2010) Employment outlook: moving beyond the jobs crisis, Paris
Part I
The Structural and Long Term
Causes of Social Exclusion
Chapter 2
Health, Lifestyle and Growth
Gianluigi Coppola
Abstract In this article, I attempt to explain why lifestyle may have a positive
impact on economic growth. First, I consider the ways in which health affects a
consumers utility, and I then define a Health Production Function for which health
is the output and consumer good is the input. In this approach, the Lifestyle Return
to Scale (LRS) parameter is defined. The first result is that an increase in a
consumers personal income may have a positive or a negative effect on health.
That is, health may be a normal or an inferior good, depending on the Lifestyle
Return to Scale value. According to this result, I compute a health multiplier and
then modify the Solow Growth Model in which health is labour-augmenting. The
result is a model in which the Lifestyle Return to Scale positively affects per capita
income and per capita income growth.
2.1 Introduction
At the macro level, stylized facts indicate substantial differences in per capita
income and health status among countries and regions. This may imply that low
per capita incomes negatively affect health and vice versa.
G. Coppola (&)
Dipartimento di Scienze Economiche e Statistiche,
Universita Degli Studi Di Salerno and C.E.L.P.E, Salerno, Italy
e-mail: glcoppola@unisa.it
G. Parodi and D. Sciulli (eds.), Social Exclusion, AIEL Series in Labour Economics, 17
DOI: 10.1007/978-3-7908-2772-9_2, Springer-Verlag Berlin Heidelberg 2012
18 G. Coppola
In the last 20 years, the literature on economic growth has focused primarily on
the role of human capital accumulation, while health has occupied only a marginal
role in economic analyses. Second, if the income and health differences among
countries are significant, those among regions are even stronger and are essential
to economic growth.
The literature on health and health economic growth has not considered the
impacts that lifestyle has on Economic Growth. The aim of this theoretical paper is
to develop the relationship between Health and Growth by accounting for con-
sumer lifestyles.
Contoyannis and Joness hypothesis (2004) introduced a micro-model of con-
sumer choice to define lifestyles more accurately and to explain the effects of
consumers choices on Health. At the macro level, Weil (2005) asked if the forces
driving differences among regions were primarily derived from health or income.
In this context, I provide an answer by computing a health multiplier (Sect. 2.2)
and developing a simple modified Solow growth model in which health is labour-
augmenting (Sect. 2.3). Thus, this model includes the relationship among income,
lifestyle and health status first obtained at the micro level.
The first important result is that an increase in a consumers personal income
may have a positive or a negative effect on health if the consumer has a good or
a bad lifestyle, respectively. At the macro level another result of the model is
that lifestyle may be crucial for growth: a good lifestyle can generate a positive
impact on economic growth, while a bad lifestyle may also negatively affect
growth. The model also explains why health improvement positively affects
income, while increasing income may have a lower effect on health (Weil 2005).
The empirical evidence can be divided into two categories: Long-Run stylized
facts and facts related to governments measures against the unhealthy habits of
the citizens.
The long-run stylized facts show increases in per capita income and life
expectancy. From 1820 to 2001, the per capita World GDP grew from $667 to
$5.709 (Maddison 2003). In the same period, world life expectancy at birth
increased from 28.5 years in 1820 to 65.2 years in 1990.
The differences among regions, in terms of both per capita GDP and life
expectancy at birth, have also increased (Riley 2005). In 1998, the per capita GDP
of the United States was 20 times that of Africa, while in 1820, it was only 3 times
larger.
In the period from 1800 to 1820, the differences among regions, in terms of life
expectancy at birth, were relatively low compared to the period from 1990 to 2001.
In the first period, the difference between the highest and the lowest regional life
expectancies was only 9.2 years. This gap widened to 26.3 years in the period
from 1990 to 2001 (Riley 2005).
2 Health, Lifestyle and Growth 19
1
Financial Times Cash incentives seen as helping nations health, 11 April 2009.
20 G. Coppola
of individuals are of interest to society and, also to firms because they affect labour
productivity.
Ultimately, lifestyle choices generate externalities, a term that indicates pos-
sible conflicts that have not been resolved by the market. Externalities affect labour
productivity and healthcare costs above and beyond the level for which companies
are responsible. This relatively new concept of externalities is explained by Sassi
and Hurst (2008):
Lifestyle choices, as many other forms of consumption, may produce external
effects. There are immediate externalities that derive directly from acts of lifestyle con-
sumption, such as passive smoking, violent and disorderly behaviour associated with
alcohol abuse, or traffic accidents resulting from reckless driving. There are also deferred
externalities, which are generated through the link between lifestyle choices and chronic
diseases. Once chronic diseases emerge, and in some cases even before they emerge (e.g.,
when important risk factors such as hypertension or obesity begin to manifest themselves),
the individuals affected will become less productive, possibly entirely unproductive, they
will make a more intensive use of medical and social services, which may be publicly
funded, they may require care by members of the family and friends. Conversely, a
reduced life expectancy may mean a less prolonged use of publicly funded medical and
social services at the end of life, as well as reduced pension payments, which are not
themselves externalities, but would translate into a less onerous fiscal burden and therefore
less distortion in the way the economy works. All of these phenomena involve external-
ities (negative the former, positive the latter) on society at large, family and friends, which
can be attributed at least to some extent to the lifestyle choices originally made by the
individual. The extent to which externalities can be associated with lifestyle choices
depends, of course, on the strength of the link between lifestyles and disease, i.e., by the
increase in the risk of developing a chronic disease associated with adopting a particular
lifestyle.
These arguments are not new. John Stuart Mill (1859) wrote the book On
Liberty to fight against laws that would limit individual freedom. In the nineteenth
century in Great Britain, social degradation phenomena such as alcoholism were
prevalent. Several social movements asked the government to implement prohi-
bitionist measures to halt these phenomena. Mill contrasted these ideas,
arguing that
what happens inside a persons body or mind is that persons private business, not the
business of society and certainly not the business of the government over himself,
over his own body and mind, the individual is sovereign.
2.3 A Micro-Model
In the following section, I define lifestyle and develop a micro-funded model that
explains the relationship between health and income and the effects of income on
health.
2 Health, Lifestyle and Growth 21
U h; x; z ha xb zd 2:2
Contoyannis and Jones (2004) assume that the utility of a consumer depends on a
set of goods C, Health H, XU ; a vector of observable exogenous influences on
U and lU , a vector of unobservable influences or U, defined by
U uC; H; XU ; lU 2:3
Furthermore, HPF depends on a set of goods C, a vector of observable exog-
enous influences on H, XH ; and a vector of observable exogenous influences on H,
lH : The vector H is defined by
H hC; XH ; lH 2:4
It is assumed that the consumption of a commodity can improve, worsen or
have no effect on the health of a consumer. Consumption can worsen health in the
case of smoking, alcohol and drugs.
For simplicity, let us assume that every commodity can only better or
worsen the health of a consumer. Further, no commodities can positively impact
health in small quantities and negatively impact health in stronger doses.2 It is also
assumed that x improves health, while z worsens health. The commodity x may be
defined as the virtuous, or sustainable, good, and z may be defined as the harmful
good.
Health also depends on the initial level of health h0 ; public health W; time
t and a stochastic component e: The Health Production Function is
2
The ancient Romans said In Medio stat Virtus. In the model that hypothesis doesnt matter
for each single good.
2 Health, Lifestyle and Growth 23
or
3
See for example Berger and Leigh (1989) and Kenkel (1991) for the relationship between
schooling and health. See also Avitabile (2009) for the relationship between health and
information.
24 G. Coppola
d ac cy
z 2:10
b d a q c pz
The weight of health, a; increases the consumption of virtuous goods and
reduces the consumption of harmful goods. Optimally, the health level is
aq b cy q d ac cy c
h 2:11
b d a q c px b d aq c pz
or
q c
aq b d ac pz c
h cyqc 2:12
b d a q c px q
b d a q c
q
aqb
Equation 2.11 is the health demand function, where bda qc and
c
dac
bdaqc are the shares of good x and good z, respectively, weighted for
their relative health elasticities.
The level of health and the price of virtuous good are negatively correlated.
If the price of good x increases (or decreases), it worsens (or improves) the level of
health. Conversely, health improves (or worsens) if the price of z increases
(or decreases).
The health elasticity with respect to income is q c h; the parameter LRS.
Unlike the parameters that can have only one sign, this parameter may be positive
or negative. If q c 0; income growth does not affect the level of health.
If q c\0; income negatively affects health. If q c [ 0; income positively
affects health.
4
This approach may be considered as a generalization of Wagstaffs model (1986). See
Appendix for details.
2 Health, Lifestyle and Growth 25
Thus, income growth does not always positively affect health. The sign and the
degree to which income affects health depend on the parameter h:
A proxy or Index of a consumers Lifestyle (LI) may be given by the weighted
average of the quantity of commodities consumed for the consumers health
elasticity. This variable follows the relationship 1\LI\1 and is given by
aq b d ac
LI q c 2:13
b d a q c b d a q c
Therefore, LRS h is a crucial variable in the model because it indicates the
attitude of a consumer, based on preferences and opportunities, toward leading a
particular lifestyle. Thus, h q c partially and indirectly reflects consumer
preferences because the health production function contains only those com-
modities that consumers prefer or can purchase.5
In the previous section, the effects of income on health were described. Assuming
the existence of a representative agent, Eq. 2.11 can be rewritten as
h
Y
ht 2:14
L
q c c
aqb dac pZ
where m bda qc bdaqc px q
c
A production function with a constant return to scale and for which both
technology and health are labour augmenting is assumed. This may be a Cobb
Douglas Production Function (i.e. Weil 2005; Sala-i-Martin 2005).
Y K a AhL1a 2:15
or
5
Three issues should be highlighted here. First, in this simple consumer model, choices are
made between two commodities. In reality, a commodity may be not consumed for three reasons,
the first two of which were outlined previously: (1) the consumer does not like a commodity;
(2) even if a commodity is liked, the health damage caused by the commodity may be greater than
the commoditys utility, preventing consumption of the commodity; and (3) the relative price of a
commodity may be greater than income, preventing consumption of the commodity. In the first
two cases, the commodity is not consumed as a result of free choice. In the second case, this
choice may be difficult. In the third case, price and income limits restrict access to the
commodity. In this article, we consider only the case in which individuals consume both
commodities.
26 G. Coppola
a
Y K
Ah1a 2:16
L L
From the system given by Eqs. 2.14 and 2.15, the impacts of a health shock
Dm and an income shock DA on health and income can be quantified.
Solving this system yields the effects in terms of elasticity. The results are
reported in Scheme 1 for 0\h\1:
Scheme 1
On health On income
1\h\1
Health shock d log h 1 d log Y 1 a
d log m 1 h1 a d log m 1 h1 a
Income shock d log h h1 a d log Y 1 a
d log A 1 h1 a d log A 1 h1 a
1
In terms of elasticity, the health multiplier is equal to 1h1a for health and
1a
1h1a for income.
For 0\h\1; both multipliers are positive and greater than one. If a lifestyle is
positive, a health shock will more strongly affect health and the growth of labour
productivity.
The effect of a health shock on income depends positively on both LRS and
income elasticity, with respect to labour or labour income share.
h1a
The effect of an income shock on income is equal to 1h 1a ; while the effect
1a
on health is equal to 1h 1a : In this case, for 0\h\1; both effects are greater
than zero.
The question of whether a health shock more strongly affects income or an
income shock more strongly affects health (Weil 2004) can now be answered: both
technological and health shocks produce the same effect on income. This effect is
1a 1
equal to 1h 1a ; which is greater than 1 a because 1h1a [ 1:
A health shock has an impact on health equal to 1h11a ; which is greater than
1a
an income shock 1h1a because 0\h\1: A health shock has an impact on
1a h1a
income equal to 1h1a ; which is greater than
an income shock on health 1h 1a :
The effect that a reduction of the price of commodity x has on health can be
quantified. The multipliers for income and health are
d log h d log h d log t 1
q 2:17
d log px d log t d log px 1 h 1 a
2 Health, Lifestyle and Growth 27
Scheme 2
On health On income
Health shock dh 1 dY 1 a Y
Yh
dm 1 h1 a dm 1 h1 a t
Income shock dh h1 a h dY 1 a Y
dA 1 h1 a A dA 1 h1 a A
For Streeten (1994), one of the reasons for promoting human development is
that a well-nourished, healthy, educated, skilled and alert labour force is the most
important productive asset.
28 G. Coppola
Y t K ta AtLt1a 2:19
where K(t), A(t), L(t) are capital, level of technology, and labour, respectively.
Let us assume all the hypotheses of the Solows Growth Model.
Technological progress and the population growth rate are exogenous and
d ln At d ln Lt
constant: g; n:
dt dt
Assuming that health is a labour-augmenting factor (Weil 2005; Sala-i-Martin
2005), the production function becomes
Y t K ta AthtLt1a 2:20
With
h
Y
h
h ty t h tY h Lh ; 2:21
L
this becomes
1a
Y t K ta AttY th Lt1h 2:22
or
a
1a
Y t K t1h1a AttLt1h
1h1a
2:23
a
1 1
1h1a
1h1a
Y t K t1h1a At1h t1h Lt 2:24
a1h1a
Equation 2.24 shows a constant return to scale because 1h1a 1.
a 1h1a
Substituting a1 1h1a and a2 1h1a 1 a1 yields
1 1
1a1
Y t K ta1 At1h t1h Lt 2:25
The first result is that even if the Solow model with health remains an exog-
enous growth model, the parameter LRS positively affects per capita income
growth and the level of income per capita at the steady state.
For example, if h 0:5; the income growth rate is equal to 2g; while a negative
LRS h\0 results in a per capita income growth rate that is less than that of
technological progress. Thus, a good lifestyle can improve economic growth,
while a bad lifestyle can slow growth.
The second result is that health increases more slowly than per capita income.
Unlike economic growth, the rate of health growth may be positive, null or neg-
ative, depending on the sign of the LRS parameters. For h\0; economic growth
negatively affects health, which worsens.
30 G. Coppola
The second scenario may be the case of a health-poverty trap. Such a trap
was noted in Russia; male life expectancy, which can be considered a health
indicator, plummeted by 7 years from 1989 to 1994 because of high levels of
alcohol consumption (UNDP 2010).
The results of the model can be explained differently. Per capita income growth
depends on the product of the technical progress parameter and LRS, both of
which are exogenous.
Technical progress is considered to be manna from heaven. In fact, the aim
of the endogenous growth theory is to identify those factors and mechanisms that
could be controlled by the government to ensure higher and more durable eco-
nomic growth.
In the model presented in this article, another exogenous parameter, the Life-
style Return to Scale, h, impacts economic growth. In the introduction, we pre-
sented several cases in which governments have attempted to control lifestyles.
This type of governmental behaviour has several possible explanations.
Equation 2.27 can be re-written as
g
h 1 Yt 2:31
d ln Lt
dt
Suppose that the government has established a target for its economic growth
rate, denoted by ^y ; that can be controlled by technological progress g. If ^y [ g;
the economic growth rate fixed by the government is greater than technical pro-
gress, and the government can strive for higher growth by controlling or trying to
modify lifestyles, which are one of the channels that transmit the effects of eco-
nomic growth to health. Conversely, if technological progress is high, lifestyles
may not be important because technological progress can ensure a high level of
economic growth.
Further, health growth is equal to economic growth multiplied the parameter h.
As shown in Fig. 2.1, with a fixed a technological rate g, the effects of economic
growth on health improvement depend on the value of h. Lower values of h, result
in weaker links between economic production and health improvements. This
relationship could become negative for h \ 0. Hence, economic growth is a
2 Health, Lifestyle and Growth 31
2.6 Conclusions
Acknowledgements I would like to thank seminar participants at the University of Salerno and
the University of Pescara. I am grateful for many helpful comments, received during the course of
those presentations. Thanks also to Adalgiso Amendola, Alberto Bennardo, Dimitrios Christelis,
Floro Ernesto Caroleo, Marcello DAmato, Sergio Destefanis, Fernanda Mazzotta, Niall OHi-
gghins, Carmen Pagliari and Giuliana Parodi for their very useful suggestions. The usual dis-
claimer applies.
A.1 Appendix
Starting with Michael Grossmans Model (1972) and Wagstaff (1986) developed a
one-period model of demand for health. The four hypotheses of the model include
the following: (1) an individuals health is determined by the consumption of
health inputs h x xq ; (2) preferences are non-lexicographic: individuals desire
health but not above everything else; (3) individuals also consume other com-
modities that have a positive cost for consumers, so U uh; z with dU dU
dh ; dz [ 0
2 2
and d Udhh;z ; d Udzh;z \0; and (4) consumers have limited economic resources or
budget constraints: px x pz z Y; where px and pz are the prices of commodities
x and z, respectively, and Y is the income.
Assuming a Cobb Douglas Utility function and a Health production function
h x xq ; the Wagstaff Model can be formulated with the following formulas:
U h; z ha zd 2:32
h x x q 2:33
px x pz z Y 2:34
where 0\a\1 and 0\d\1 are the utility elasticities with respect to x and z,
respectively, and 0\q\1 is the elasticity of h with respect to x.
This is a special case of the Consumers model (Sect. 2) with b 0: The
commodity x is not in the Consumers utility function with c 0; thus, z does
not affect health.
The solutions can be obtained from two different methods. The first was pro-
posed by Wagstaff:
In this case the Budget Constraint is not linear. The consumer chooses between
health and z. The second possible solution is
References
Aristotle (2009) Nicomachean Ethics (trans: Ross WD). World Library Classical Books
Avitabile C (2009) Does conditionality matter for adults health? Evidence from a randomized
experiment. CSEF Working Paper 222
Berger MC, Leigh JP (1989) Schooling, self-selection, and health. J Human Res 24(3):433455
(Summer, 1989)
Brillat Savarin JA (2004) The phisiology of taste. Kessinger Publishing, US
Contoyannis P, Jones AM (2004) Socio-economic status, health and lifestyle. J Health Econ
23:965995
Gravelle H, Rees R (1992) Microeconomics, 2nd edn. Longman Group UK Limited, England
Grossman M (1972) On the concept of health capital and the demand for health. J Political Econ
80(2):223255
Heshmati A (2001) On the causality between GDP and health care expenditure in augmented
Solow growth model. SEE/EFI Working paper series in Economics and Finance 423
Kenkel D (1991) Health behaviour, health knowledge and schooling. J Political Econ 99:287305
Kip Viscussi W, Evans WN (1990) Utility function that depend on health status. Am Econ Rev
80(3):353374
Lpez-Casasnovas et al (2005) Health and economic growth: findings and policy implications
In: Lpez-Casasnovas G, Rivera B, Currais L (eds) Cambridge MIT press, Cambridge
Maddison A (2003) The world economy: historical Statistics. Paris Development Centre of the
Organisation for Economic Co operation and Development, Paris
Mankiw NG, Romer D, Weil N (1992) Contribution to the empirics of economic growth.
Quarterly Journal of Economics 107:407437
34 G. Coppola
Marshall A (2006) Principles of Economics Cosmio Inc. original published (1890) by Prometheus
Book
Mas-Colell A, Whinston M, Green J (1995) Microeconomic theory. Oxford University Press,
Oxford
Mill JS (1859) On liberty. Republished by Agora Publication inc (2003)
Morand OF (2005) Economic growth, health, and longevity in the very long term: facts and
mechanisms. In: Lpez-Casasnovas G, Rivera B, Currais L (eds) Health and economic
growth: findings and policy implications. Cambridge MIT press, Cambridge
Riley JC (2005) Estimates of regional and global life expectancy 18002001. Population Dev Rev
31(3):537543
Rivera B, Currais L (1999a) Economic growth and health: direct impact or reverse causation?
Appl Econ Lett 6(11):761764
Rivera B, Currais L (1999b) Income variation and health expenditure: evidence for OECD
countries. Rev Dev Econ 3(3):258267
Sala-i-Martin X (2005) On the healthpoverty trap. In: Lpez-Casasnovas G, Rivera B, Currais L
(eds) Health and economic growth: findings and policy implications. Cambridge MIT press,
Cambridge
Sassi F, Hurst J (2008) The prevention of lifestyle related chronic diseases: an economic
framework. OECD Publishing 32
Solow R (1956) A contribution to the theory of economic growth. Q J Econ 70:6594 (February)
Strauss J (1986) Does better nutrition raise farm productivity? J Political Econ 94(2):297320
Streeten P (1994) Human development: mean and ends. Am Econ Rev 84(2), In: Papers and
Proceedings of the 106th annual meeting of the American Economic Association, May 1994,
pp 232237
UNDP (2010) Human development report. Second printing Palgrave Macmillian, London
Varian H (1992) Microeconomic analysis, 3rd edn. Norton, New York
Wagstaff A (1986) The demand for health: a simplified Grossman model. Bull Econ Res 38:1
Weil DN (2004) Economic growth. Addison-Wesley, Reading
Weil DN (2005) Accounting for the effect of health on economic growth. NBER Working Paper
w1145
Zargosky J (2005) Health and wealth: the late-20th century obesity epidemic in the US. Econ
Human Biol 3(2):296313
Chapter 3
A Comparative Analysis of Literacy
Rate in Contributing to Social
Exclusion Insights
Abstract Our contribution aims to analyze the relationship between the phe-
nomena of social exclusion and literacy levels, and to consider the significant
implication of this relationship on economic growth. The goal, which is reached by
analyzing cross-country data, is firstly to describe the situation of social exclusion
with the use of specific socio-economic variables, and secondly to compare the
levels of education and training for each considered country. These two phe-
nomena are mutually influenced, as a low level of literacy in affecting the
employment status precludes the possibility to enter and operate freely in society,
while poverty and persistent social exclusion of a person or family make difficult
to address appropriate educational and training paths. Therefore, our study has
rejoined two issues which influence almost all decisions adopted by policy makers,
especially in the Western world. The opening issue is the level of education, which
should constitute the human capital of a country, through appropriate investment,
and the second one is the relational condition of social dynamics, which highlight
the so-called social capital. Together these two types of intangible capitals con-
stitute a strong support for the long-term development of a country. Our quanti-
tative analysis is also addressed to detect differences and peculiarities among the
G. Parodi and D. Sciulli (eds.), Social Exclusion, AIEL Series in Labour Economics, 35
DOI: 10.1007/978-3-7908-2772-9_3, Springer-Verlag Berlin Heidelberg 2012
36 E. Bucciarelli et al.
different national realities, with the ultimate purpose to recognize which socio-
economic variables affect more directly the processes of education.
exploit them to coexist in the stable and balanced society they belong to. This
particularly includes the definition offered by the International Adult Literacy
Survey (see Statistics Canada and OECD 2000): The ability to understand and
employ printed information in daily activities, at home, at work and in the com-
munityto achieve ones goals, and to develop ones knowledge and potential.
The definition used by OECD (2010) adds that [] differences in levels of
literacy matter both economically and socially: literacy affects, inter alia, labour
flexibility and quality, employment, training opportunities, income from work and
wider participation in civic society. Furthermore, the IALS Final Report (Sta-
tistics Canada and OECD 2000) shows a number of relationships about literacy:
countries with higher literacy scores have higher labour force participation and
shorter work hours whereas countries with a high proportion of adults with low
prose skills have lower GDP per capita. There is a correlation between a low GDP
per capita of a country and the rising proportion of adults with low prose skills.
Public intervention may partially cope and withstand the mentioned problems
related to the lack of literacy, by trying to obtain a plentiful and robust level of
national human capital in the long run. This is now a fundamental part of many
economic studies concerning growth and development. But the complex phenom-
enon of educating and training individuals cannot be regarded only as a general
educational teaching base of general educational foundations. This means not only
prepared trainers, and adequate public and private investment, but also a rational
project in order to develop in all individuals a sense of common rules through
education (see, for example, Grossman and Kim 1997). These are the foundations
for building a solid social capital which is itself a determinant of socio-economic
development processes (see, among others, Gradstein and Justman 2002, who have
also included the education processes in the relationship between social capital and
economic growth).
People who cannot reach satisfactory levels of literacy either by choice or
compulsion are the weakest in society and go toward the phenomenon of social
exclusion. The problems caused by poor literacy can be observed through two joint
events. The first one concerns the limited prospects of finding safeguarded
employment opportunities, the second one considers the defective contribution
which people offer to the socio-economic system through their labour productivity.
The difficulty in finding a job involves a certain separation from society. This
separation deprives individuals of their ability to fully exploit the possibilities
offered them by contemporary world. The individuals, though not in poverty, will
not have the opportunity to be engaged in specified conducts which are charac-
teristics of each society, as constitutional elements of general needs. Therefore,
this path is degenerative, and leads to conditions which are close to relative
poverty which tends to become a phenomenon that is passed between generations.
In fact, poor parents cannot guarantee the optimal education for their children (see,
among others, Grossmann 2008; Galor and Moav 2004), and socially excluded
parents represent a source of social exclusion for their children. The education
level of a person can be found either through the years of school attended, or
through the qualifications acquired. In particular, in this work, we observe the
38 E. Bucciarelli et al.
influence which some typical variables, related to the processes of social exclu-
sion, can have on the average level of education of a country. It is known that the
economic conditions may affect the ability to achieve high average levels of
education, but also the strength of education and social capital constitute a lever of
development (see, among others, Temple and Johnson 1998). Gradstein and
Justman (2002) analyze the importance of a broad common cultural basis to start
those basic functions of effective interaction between individuals. The spread of
education, and, in general, of literacy may be a fundamental policy goal in trying
to establish virtuous processes from an economic standpoint. Essentially, the main
aim of our empirical study is to demonstrate how certain variables,1 which char-
acterize social exclusion, have obvious influences on the literacy rate. In order to
pursue this aim, we gather data concerning 30 countries and then we briefly
compare the effect of the above mentioned variables on the level of economic
development of the countries considered. In our particular case study, the literacy
rate represents the percentage of people with the ability to read and write without
specifying the level of education.
The framework of the paper is organized as follows. In Sect. 3.2, we start by
showing the methodology used. Then we proceed in Sect. 3.3 with a preliminary
empirical investigation on literacy rate, and on the expectancy variables.
In Sect. 3.4, the results of the multivariate regression model are presented. The
findings of the factor analysis with maximum likelihood method and VARIMAX
rotation are shown in Sect. 3.5. Furthermore, Sect. 3.6 focuses on multivari-
ate regression model for literacy rate with maximum likelihood components.
In Sect. 3.7, we represent and analyze the countries in a hierarchical cluster with
variables related to literacy. In Sect. 3.8 we focus our attention on specific
American literacy policies which are particularly significant for our study. We
present our conclusions in Sect. 3.9.
In this section we analyze literacy rate by using the least-squares method, the
factor analysis with maximum likelihood method and the hierarchical cluster. The
method of least squares is a standard approach to the approximate solution of over
determined systems (Moser 1996; Freund 2003), i.e. sets of equations in which
there are more equations than unknowns. Least squares means that the overall
solution minimizes the sum of the squares of the errors made in solving every
single equation. The most important application is in data fitting: the best fit in the
1
We have collected the cross-country dataset from UNESCO and World Bank. For the cross-
country analysis we provide, see also Levine and Renelt (1991, 1992); de Gregorio and
Lee (2002); Hoover and Perez (2004).
3 A Comparative Analysis of Literacy 39
least-squares sense minimizes the sum of squared residuals, a residual being the
difference between an observed value and the value provided by a model (Bjrck
1996). We consider a linear regression model: hence, the model comprises a linear
combination of the parameters:
X
m
f xi ; b bj /j xi 3:1
j1
of xi ; b
Xij /j xi 3:2
obj
Then we can see in case (3.2) the least square estimate (or estimator, in the
context of a random sample), b is given by:
b X T X1 X T y 3:3
Consequently, we consider the response variable as a linear function of the
regressors:
yt b0 b1 x1 bk xk et 3:4
In our paper, we analyze literacy rate (LR) as a function of school enrolment
consisting in pre-primary (PRE), primary (PRI), secondary (SEC) and tertiary
(TER); GDP real growth rate (GDP); long term unemployment rate (LUR); public
spending on education as percentage of GDP (PSE) and public spending on
education as percentage of government expenditure (PGE); children out of school
(COS). In our analysis, we consider different levels of education: pre-primary,
primary, secondary, and tertiary education. These levels are generally found in
similar ways in many countries, and generally go from kindergarten to the so
called higher education. The analysis regard 30 countries2 from different conti-
nents, and we consider mostly countries of Organizations for Economic Co-
operation and Development (OECD). The model we provide is the following:
2
The countries considered by this research are: Australia, Austria, Belgium, Bulgaria, Canada,
Czech Republic, Denmark, Finland, France, Germany, Greece, Hungary, Iceland, Ireland, Italy,
Japan, Republic of Korea, Latvia, Luxembourg, Mexico, Netherlands, New Zealand, Norway,
Poland, Portugal, Romania, Slovakia, Slovenia, Spain, Sweden, Turkey, United Kingdom, United
States.
40 E. Bucciarelli et al.
used in principal component analysis and factor analysis which maximizes the sum
of the variances of the squared loadings. That is, it seeks a basis that most eco-
nomically represents each individual, so that each country can be well described
by a linear combination of only a few basic functions:
0 !2 1
X k X P
c X k Xp
RVARIMAX arg max@ 4
KRij KRij A
2
3:11
R
j1 i1
p j1 i1
The following analysis refers to the 30 countries some of which are OECD
members, over the period of time 20072009. The countries we chose to include in
our cross-country analysis are those that represent manifold and diversified
3 A Comparative Analysis of Literacy 43
Table 3.1 Average, lowest and highest values for dependent and expectancy variables
Indicator Mean Lowest Highest
Literacy rate 98.25 88.70 (TR) 99.80 (LV)
GDP r.g.r. 0.003 -0.041 (LV) 0.045 (PL)
Pre-primary 0.891 0.159 (TR) 1.233 (ES)
Primary 1.058 0.549 (PT) 1.577 (ES)
Secondary 0.785 0.183 (NL) 1.479 (AU)
Tertiary 0.413 0.01 (NL) 0,968 (U.S.)
l.t.u.r. 1.760 0.045 (MX) 5.988 (SK)
p.e.p. GDP 0.051 0.019 (AT) 0.086 (JP)
p.e.p. GOV 0.132 0.034 (PT) 0.281 (LV)
c.o.s. 0.030 0.0002 (LU) 0.416 (U.S.)
Mean, maximum and minimum values for dependent and expectancy variables. In parentheses is
given the international symbol of the various countries
3
The literacy rate represents the percentage of people with the ability to read and write, and in
this case without specifying the level of education.
4
In our analysis we take into consideration different levels of education, such as pre-primary,
primary, secondary and tertiary education: these levels are generally found in similar ways in
many countries, and represent usually from pre-primary until the so called higher education
(tertiary).
44 E. Bucciarelli et al.
values, while the highest refers to Poland (PL). The average GDP real growth rate
is near to zero, because in recent years there has been stagnation and recession of
the economy. The several education levels get the minimum value by Turkey,
Portugal (PT) and Netherlands (NL) in the four categories considered. Slovakia
(SK) shows the highest value reported for long unemployment rate. Furthermore,
Austria (AT) and Portugal are the countries with lowest usage rates of public
spending in education. Finally the variable related to children out of school shows
its maximum value in United States (US) and this percentage is always condi-
tioned by a high population abundance, and we note simultaneously a high degree
of school drop-outs. In the next section we begin the analysis of the literacy rate
applying multivariate regression method, maximum likelihood method and clus-
tering analysis to highlight the differences in education levels in recent years in the
countries considered.
mathematical model we propose is the (3.4), and in terms of variables is the (3.5).
Before applying the (3.5) we perform other types of investigations in order to
understand the influence of literacy rate compared to single macro groups5 under
analysis: we have first investigated literacy rate as a function only of education
variables, then literacy rate only with variables related to social exclusion.
Observing our results of statistical tests concerning the significance of the param-
eters and considering the best fitting to data, we choose to consider model as a
whole. Indeed, we use also education variables as percentages of total education in
different countries in order to identify the optimal model applied; resulting data are
not satisfactory for the significance of the parameters in any kind of linear combi-
nation we calculate. Thus, the combination of economic growth, education levels
and social exclusion variables leads to a better result than analytical models con-
sidered singly.
We analyze the presence of multicollinearity using the tolerance analysis that is
equal to:tol k R2 . When the values of this analysis are close to zero there is a
combination of other endogenous variables covered by our study. Moreover, we
use the variance inflation factor (VIF) that is the reciprocal of tolerance, in order
to achieve further confirmation of the result found with the tolerance analysis. The
VIF has a threshold value equal to 0.10 and we can check if there is collinearity
among independent variables considered. Tests for investigating endogeneity have
not been reported in this study since they have provided results in the standard and
not particularly interesting for our goals.
In addition, an important test we apply to detect the presence of autocorrelation
between independent variables is the Durbin-Watson statistics which is used in the
erratic component (et) of a regression analysis. The related formula is equal to:
PT 2
t2 et et1
d PT 2 3:14
t1 et
We are also interested in applying the heterogeneity test which concerns the
system of variables we have selected in order to describe the phenomenon of social
exclusion in terms of illiteracy as our data set is quite multifaceted (many vari-
ables, different from each other and variability between countries). For identifying
any heterogeneity we use the ANOVA test (analysis of variance), which locates
the dispersion of observed values compared with an expected value, so that we can
detect the possible homogeneity between the variables; the resulting variances
should be quite similar to each other otherwise it will confirm the hypothesis of
heterogeneity of the system. Finally, all the explanatory variables should be
uncorrelated with the residual terms (uncorrelated residual values), that is the
residue of an observation must be independent from the residual value resulted
from another observation. The results are shown in Tables 3.2 and 3.3.
5
For macro groups we mean groupings of the endogenous variables based on their similarities.