provision of human capital evidence review · 2019-09-02 · background, health and job mobility as...
TRANSCRIPT
1
Provision of Human Capital Evidence Review
A Report for the Office for National Statistics
by
Lea Samek1, Christian Darko2, William King4, Sunny Valentino Sidhu4, Meera Parmar4,
Francesca Foliano1, Jamie Moore1, Mary O’Mahony1,3, Chris Payne4, Ana Rincon-Aznar1 and
Gueorguie Vassilev4
1 National Institute of Economic and Social Research
2 University of Birmingham
3 King’s College London
4Office for National Statistics
September 2019
Acknowledgements: We gratefully acknowledge comments from Andrew Aitken (NIESR) as well as
detailed suggestions from Gueorguie Vassilev (ONS), Sunny Sidhu (ONS) and Meera Parmar (ONS).
2
Table of Contents
1. Introduction ................................................................................................................................... 5
2. Factors affecting human capital accumulation and their measurement .................................. 8
2.1. Family background and household structure................................................................... 10
2.1.1. Parental education and educational attainment ........................................................... 10
2.1.2. Parental education and cognitive and non-cognitive skills .......................................... 15
2.1.3. Family structure and offspring education ..................................................................... 17
2.1.4. Wider family and environmental factors ....................................................................... 22
2.2. The role of health in human capital accumulation .......................................................... 24
2.2.1. Ageing ........................................................................................................................... 25
2.2.2. Health and education .................................................................................................... 27
2.2.3. Health and labour supply .............................................................................................. 29
2.2.4. Provision of and access to health care ......................................................................... 30
2.2.5. Earning effects on health .............................................................................................. 32
2.2.6. Lifestyle, access and health conditions ......................................................................... 34
2.3. Job mobility, skills, and training ....................................................................................... 40
2.3.1. Firm, industry and occupation-specific human capital ................................................ 41
2.3.2. Work characteristics ..................................................................................................... 41
2.3.3. Estimating the costs of mobility .................................................................................... 43
2.3.4. Job mobility in the UK .................................................................................................. 47
2.4. Within-country migration .................................................................................................. 49
2.5. Crime .................................................................................................................................... 53
2.5.1. Economics of education and crime ............................................................................... 54
2.5.2. Family and crime .......................................................................................................... 58
2.5.3. Educational/vocational programs and recidivism ........................................................ 58
2.6. Learning ............................................................................................................................... 59
2.6.1. Education environment ..................................................................................................... 59
2.6.2. Adult learning ................................................................................................................... 63
3. Determinants of earnings and their use in valuing human capital ......................................... 65
3.1. Returns to education ........................................................................................................... 66
3.1.1. Educational qualification, degree subject, and institutional quality ............................ 67
3.1.2. Higher education and earnings ..................................................................................... 68
3.1.3. Institutional characteristics .......................................................................................... 69
3
3.1.4. Identifying causal effects ............................................................................................... 71
3.2. Skills and personality traits ................................................................................................ 72
3.2.1. Cognitive skills .............................................................................................................. 72
3.2.2. Personality traits ........................................................................................................... 74
3.3. Family background and earnings ...................................................................................... 77
3.4. Health as an earnings determinant .................................................................................... 79
3.5. Job mobility ......................................................................................................................... 82
3.6. On-the-job training ............................................................................................................. 84
4. Conclusion ................................................................................................................................... 87
5. References .................................................................................................................................... 92
4
Abbreviations
AI Artificial Intelligence
BCS British Cohort Study
BHPS British Household Panel Survey
DiD Difference-in-Difference
GLS Generalised Least Squares
HESA Higher Education Statistics Agency
ILO International Labour Organization
IGE Intergenerational Elasticity
IQ Intelligence Quotient
IV Instrumental Variable
LFS Labour Force Survey
NCDS National Child Development Study
NHS National Health Service
NLSY National Longitudinal Survey of Youth
NPD National Pupil Database
OECD Organization for Economic Co-operation and Development
OLS Ordinary Least Squares
ONS Office for National Statistics
PIAAC OECD Survey of Adult Skills
PISA Programme for International Student Assessment
STEM Science, Technology, Engineering and Mathematics
TTWA Travel To Work Area
UNICEF United Nations International Children's Emergency Fund
WHO World Health Organisation
5
1. Introduction
The measurement of human capital stock is high on the policy agenda given its importance as a driver
of economic growth. Although there has been increased interest in the measurement of human capital,
particularly in recent times, the concept of human capital goes back almost 250 years. Adam Smith
(1776) described human capital as “the acquired and useful abilities of all the inhabitants or members
of the society” which, once acquired, are “a capital fixed and realised, as it were, in his person”. Starting
in the 1960s, economists treated these abilities as assets and used the notion of human capital to explain
output differences which they could otherwise not explain by traditional inputs. As a consequence, the
definition of human capital evolved over time and later was broadened to encompass the knowledge,
skills, capacities and abilities embodied in an individual that contribute to economic production and
social well-being (Schultz, 1961; OECD, 1998; World Bank, 2006). Despite this evolved definition,
previous research tended to proxy human capital with education, measured mainly by schooling
(Schultz, 1961; Mincer, 1974; Lucas, 1988; Barro, 1991; and Mankiw et al., 1992). The fundamental
insight of human capital is that education can be considered as an investment into the knowledge and
skills of individuals (Checchi, 2006). Existing evidence so far indicates that higher levels of acquired
schooling improve individual earnings (Psacharopoulos and Patrinos, 2018) and also productivity
(Wössmann, 2016).
At the national level, human capital is strongly correlated with economic growth (Hanushek and
Wössmann, 2012). Recently, the World Bank developed a human capital index that attempts to identify
the trajectory of individuals and how they impact on the productivity of the next generation of workers.
Specifically, the index captures the amount of human capital a child born today could expect to attain
by age 18, taking into account current risks of poor health and poor education.1 It uses under-age-5
mortality rates taken from the United Nations Child Mortality Estimates, adult survival rates, age-
specific enrolment rates, prevalence of stunting taken from the United Nations International Children's
Emergency Fund (UNICEF), the World Health Organisation (WHO) as well as the World Bank Joint
Malnutrition Estimates, and harmonised test scores from international student achievement testing
programs. Another recently developed human capital index measure is by the World Economic Forum
which covers 130 countries.2 The index is based on 21 indicators including literacy, numeracy,
educational attainment and enrolment, (un)employment rates, school quality, among many others,
obtained mostly from publicly available data originally collected by the International Labour
1 For more detailed information on the data and methodology, please refer to the following report published in
2018 http://documents.worldbank.org/curated/en/300071537907028892/pdf/WPS8593.pdf 2 For more detailed information on the data and methodology, please refer to Appendix B of the following report
published in 2017 (see https://weforum.ent.box.com/s/dari4dktg4jt2g9xo2o5pksjpatvawdb).
6
Organization (ILO), the United Nations Educational, Scientific and Cultural Organization and
qualitative survey data from the World Economic Forum’s Executive Opinion Survey.3
In the UK, the Office for National Statistics (ONS) current estimates of human capital stock are based
on the Jorgenson and Fraumeni (1989, 1992) income-based approach.4 In this approach, potential
lifetime earnings of each individual in the active population are calculated, differentiated by gender,
age and education and taking account of the probabilities that those in younger age groups remain in
education. These earnings are then summed across people in each age, gender and education
qualification group to yield an overall figure for human capital stock. Although this is in line with the
United Nations Economic Commission for Europe guidelines on Measuring Human Capital and allows
for education, employment and demography to interact and jointly explain economic changes, this
approach is heavily reliant on education as a human capital driver and leaves no room for other
determinants. Therefore, this literature review has been carried out on recent important developments
that are necessary to help the ONS to identify other possible factors which can be relevant in its own
attempts at developing a more comprehensive measure of human capital.
In this review, focus is put on the UK context and studies published within the last 10 years although
we also cover relevant and timely evidence for other countries as well as older highly influential papers
that help set the context and contribution of the review. We find that while some factors offer insightful
views and findings on the underlying determinants of human capital, the lack of evidence in the UK
and occasional contradicting results plague some of the conclusions drawn. The review is based mostly
on evidence from studies with a quantitative approach which have been published in high-ranked
academic journals that are regarded as leaders or among the leaders in their respective discipline or
area. As a guide for selecting the journals, we rely on the Association of Business Schools rankings.5
In exceptional cases, we review articles and publications that are not listed in this ranking but are
innovative, of good methodological quality or are widely cited in their respective field. We also resort
to published works by reputable organisations such as the Organisation for Economic Co-operation and
Development (OECD), the World Bank, the Institute for Fiscal Studies, among others.
3 Very recently, Angrist et al. (2019) with the help of the World Bank, constructed a database of globally
comparable learning outcomes for 164 countries and territories covering 98 per cent of the global population from
2000 to 2017. They aim for this dataset to be updated annually to allow observing human capital accumulation
over time. For more detailed information on the data employed, please refer to their Appendix A. 4 There are essentially three methods employed to measure human capital stock, cost-based, attainment-based or
income-based (see Jones and Chiripanhura (2010) for reviews). Other international examples of measuring human
capital accumulation include: Australia (see Wei, 2004, 2008), Canada (see Gu and Wong, 2010), China (see Li
et al., 2013), New Zealand (see Le et al., 2006), Norway (see Liu and Greaker, 2009), Sweden (see Ahlroth et al.,
1997) and the US (see Christian, 2010). 5 https://charteredabs.org/academic-journal-guide-2018/
7
The review is structured as follows: first, it discusses the individual and combined role of factors that
are considered to be relevant in the process of human capital accumulation. These include family
background and household structure, ageing and health, job mobility, skills and training, within-country
migration and crime. Second, it reviews determinants of earnings and their use in valuing human capital.
This section focuses on studies on returns to education, cognitive skills and personality traits, family
background, health and job mobility as well as on-the-job training. A large majority of the studies
covered in this part of the review are considered of high methodological quality both in terms of data
and econometric techniques. We end this literature review with a concluding chapter on the emerging
findings.
8
2. Factors affecting human capital accumulation and their measurement
This section aims to highlight some of the factors associated with human capital development. First, the
review starts with a look at the role of family background, namely, parental education, and family
structure. These two are known to play an influential part on early childhood development and in
forming the foundation for future economic success. Secondly, the interest in health (and also ageing)
as a component of human capital is considered. Health is viewed as an investment good, which has the
ability to improve individual well-being and, at the same time, increases time that people can allocate
to work in order to accumulate wealth. Thirdly, an efficiently functioning labour market is continuously
reallocating labour in response to changing demands. Therefore, we review literature on job mobility
and the need to develop new skills and enhance existing skills through training which all become
increasingly relevant in today’s modern economy. Fourthly, we review the role of regional migration
in generating knowledge and shaping the redistribution of human capital. Finally, human capital
accumulation can be affected by criminal activities and, therefore, we review literature on the
relationship between criminal participation and education, family background as well as intervention
programs.
Key findings show that indeed family background plays a fundamental role in shaping future economic
prospects of individuals (Dearden et al., 1997; Delaney et al., 2011; Corak, 2013). Studies that examine
causal mechanisms, mostly using education policy reforms as an identification strategy for the case of
the UK, provide further understanding of the linkage between parental education and children’s human
capital development (Chevalier et al., 2013). Results from these studies suggest positive effects on
children’s education, although the magnitude of maternal and paternal schooling differs among studies.
There is however limited understanding of causal mechanisms for the UK, owing largely to small
sample sizes and limited (administrative) data availability on twins, siblings and adopted children. In
contrast, many studies for Nordic and other Scandinavian countries use samples of twins, siblings and
adopted parents as identification strategies (Plug, 2004; Björklund et al., 2006; Sacerdote, 2007;
Holmlund et al., 2011; Pronzato, 2012). In terms of children’s cognitive and non-cognitive skills
development, results indicate a positive association with parental cognitive skills at younger ages
(Blanden et al., 2007; Brown et al., 2011; de Coulon et al., 2011). These findings suggest that children
pick up most of the important attributes and skills needed to perform well in life at earlier stages of their
lives. Family structure, such as size of family (Black et al., 2010; Angrist et al., 2010), birth order (Booth
and Kee, 2009; Pavan, 2016; Lehmaann et al., 2018), parental presence (Kalil et al., 2016; Gould et al.,
2019), gender of siblings (Cools and Patacchini, 2017; Rao and Chatterjee, 2017; Brenøe, 2018) and
spillover from siblings (Breining, 2014; Breining et al., 2015; Black et al., 2017; Nicoletti and Rabe,
2017; Qureshi, 2018) also seem to have impacts on children’s education. Later-born children appear to
9
have lower educational outcomes than their older siblings and this effect tends to be larger in bigger
families. This may be mitigated to some extent by spillovers from older siblings performing well in
school to younger siblings. Since most studies on family structure and children’s development relate to
Scandinavian countries, more evidence is needed for the UK (with the exception of Booth and Kee,
2009 as well as Nicoletti and Rabe, 2017).
A growing body of literature suggests that education has benefits beyond increases in labour market
productivity which extend to health. The literature review has highlighted that the majority of UK based
studies exploit the increase in the minimum school-leaving age in 1947 and/or 1973 as an exogenous
change in the amount of schooling received to examine the causal effect of education on health
conditions (e.g. biomarkers), subjective general health and health behaviour, such as smoking. Albeit
the common strategy, scholars draw different conclusions and find mixed evidence for the causal effect
of education on health (Oreopoulos, 2006; Blanchflower and Oswald, 2008; Silles, 2009; Powdthavee,
2010; Devereux and Hart, 2010; Clark and Royer, 2013; Jürges et al., 2013). Although evidence for the
opposite direction of the relationship suggests a health selection process which generates socioeconomic
inequalities in child- and young adulthood, the large majority of work is carried out in the US and
mainly relies on standard logit regression models. When looking at the context of labour supply, the
majority of UK based evidence comes from the British Household Panel Survey (BHPS) and the
Understanding Society Survey and the indirect role of informal care provision associated with third
parties’ health rather than from the role of individual health (Michaud et al., 2010; Carmichael and
Ercolani, 2016; Carr et al., 2016). The same data sources are used in studies on the impact of earnings
on health, which mainly use the minimum wage legislation implemented in 1999 as an instrument
(Kronenber et al., 2017; Lenhart, 2017; Reeves et al., 2017).
Parts of the literature find evidence suggesting the loss of human capital as a result of movement
between jobs and/or occupations; this is evident not only for the individual who may face a significant
change in the sets of tasks to perform, but also for firms, which need to bear some of the costs of
(re)training the worker (Cortes and Gallipoli, 2017; Artuç et al., 2010; Dix-Carneiro, 2014). In contrast,
some literature highlights that human capital may be more portable across occupations than previously
thought, as long as job mobility is accompanied to a certain extent by skills transfer (Autor et al., 2003),
albeit empirically this has been shown to be more relevant for highly-skilled workers (Poletaev and
Robinson, 2008; Carrillo-Tudela et al., 2016). For the UK, the evidence also shows under-utilisation of
skills may result from moving from a full-time to a part-time job (Connolly and Gregory, 2008).
On the role of migration, highly educated individuals are found to exhibit highest levels of spatial
mobility (Faggian et al., 2015), which contribute to the process of knowledge transfer. The human
capital externalities associated with migration behaviour of skilled individuals are found to be a
10
significant determinant of innovation in British local areas (Gagliardi, 2015) but this relationship is bi-
directional as the innovativeness of a region appears as one of the major factors encouraging graduates
to seek employment in that region.
Lastly, the review on crime and human capital highlights the role of early educational intervention in
reducing criminal activity. Although different features of crime incidence affect human capital, the most
widely studied area involves education which reveals that human capital increases the opportunity costs
of crime from foregone work and expected costs with incarceration (Lochner, 2004; Lochner and
Moretti, 2004). It is important to note that this area of research is growing but it is also mainly carried
out using US and Scandinavian administrative data. The few UK based exceptions use the increase in
the minimum schooling age in 1997 as an identification strategy (Sabates and Feinstein, 2008; Machin
et al., 2011; 2012). The review has further highlighted the lack of empirical evidence for the opposite
direction which is mainly due to the difficulty of identifying exogenous variations in criminal
involvement.
2.1. Family background and household structure
Among the many factors shown to contribute to human capital development and adulthood labour
market outcomes, family background has been found to be the most important characteristic (Heckman
et al., 2006a; D’Addio, 2007). The range of family characteristics that have been shown to influence
human capital development are parental education, family structure (Ermisch et al., 2012), cultural
background and ethnicity, among others. This section provides a review of the literature in relation to
the role of family background, namely parental education and family structure. Parental education is a
direct input into the education production (see Checchi, 2006) and can affect the choice of other inputs
(Chevalier et al., 2013).
2.1.1. Parental education and educational attainment
Numerous studies have found a strong association between parental background, often father’s
earnings, and children’s earning (mostly sons). This correlation ranges between 0.2 and 0.5, although
some studies find a slightly higher correlation for the UK (see Corak, 2013). The importance of parental
education cannot be ruled out when examining the process of human capital development. Estimates of
the education intergenerational elasticity (IGE), which shows by how much a child’s earnings change
with changes in his/her parents’ earnings, range from 0.14 to 0.45 in the US (Mulligan, 1999) and from
0.25 to 0.40 in the UK (Dearden et al., 1997).6 Explanations for the role of parental education on
6 See Narayan et al. (2018) and OECD (2018) for a thorough discussion on intergenerational mobility.
11
children’s education have often been associated with the fact that more educated parents are able to
produce a better home learning environment and are able to contribute to the teaching of children
(Duncan et al., 2005). Other arguments support the view that financial constraints significantly impact
on children’s educational attainment (Checchi, 2006; Krueger, 2018). Studies, such as Carneiro et al.
(2004), Cameron and Heckman (1998) and Chevalier and Lanot (2002), further suggest that parental
education has a much more positive effect on educational attainment of offspring, compared to other
factors such as current parental income (see Chevalier et al., 2013).
Evidence of the positive relationship between parental education and children’s education is well
documented in the literature, both for the UK and elsewhere. For instance, for the UK Ermisch et al.
(2001) use data from the BHPS to show that parental educational attainment is a powerful predictor of
children’s educational attainment (see also Blanden and Gregg, 2004). In a recent study on higher
educational attainment, Delaney et al. (2011) use data designed and collected on behalf of the Irish
University Association as part of the Irish Universities Study to provide decomposed estimates of the
magnitude of the intergenerational relationship between parental education and grade performance at
university (and also returns from university education). They find that, although the effect of parental
education is small, on average 10 years of additional combined parental education is associated with
one percentage point increase in grade outcome in university, with the effect being higher for males
than for females. They also find different parental-gender effects. For instance, they find maternal
education to be more important in predicting educational attainment of males in university.
Even more recently, Anders et al. (2017) use data from Next Steps, previously known as Longitudinal
Study of Young People in England, and the National Pupil Database (NPD) to analyse the importance
of subject choice in the probability of attending university or a highly competitive university in England.
The authors also consider the association between socioeconomic status and young people's subject
choices, and the extent to which this acts as a transmission mechanism between socio-economic status
and inequality in attendance at university. Their result suggests that high achieving students are directed
towards particular subjects. Undoubtedly, such a finding has important policy implication for the UK
where recent government initiatives have aimed to facilitate equal opportunity and access to children
and families from diverse backgrounds.
Understanding causal mechanisms of parental education and children’s education
A strand of the literature focuses on the causal mechanisms underlying the link between parental
education and children’s education. Understanding the mechanisms by which parental education affects
children outcomes, although challenging empirically, could be more relevant for policy. Causal effects
12
of parental education on children’s education have often been examined using different identification
strategies such as instrumental variable (IV) techniques and, more recently, using different samples of
twin parents, siblings or adoptees. This section provides a review of the literature on some of the causal
mechanisms using these identification strategies with a focus on the UK where possible.
In his 2004 paper, Chevalier investigated whether the effect of parental education on children’s
educational attainment in the UK is causal. In particular, the author focused on the intergenerational
educational choice, that is, children staying in education beyond the compulsory school-leaving age.
The author used data on individuals aged between 16 and 18 from the British Family Resources Survey,
a national dataset that contains data on both children and parents. To determine the causal link between
parental education and children’s education, the author uses School Leaving Age, which was introduced
in the 1970s, as an instrument for parental education. The use of policy reform as an IV for parental
education has been well documented for other countries - Black et al. (2005) for Norway; Oreopoulos
(2006) for the US; and Lundborg et al. (2014) for Sweden. By assuming exogeneity of parental
education, similar to Black et al. (2005), the author finds parental education to be significant for
children’s schooling. Specifically, each additional year of parental education increases the probability
of children staying on in school by around 4 percentage points. The effects of mother’s and father’s
schooling on children education were however not statistically different from each other. In a relaxed
specification of the exogeneity of parental education, the authors instrument for parental education
using school-leaving age and find similar effects of maternal education on children’s education, while
for fathers the effect was found to be insignificant and negative. In a further analysis, when the sample
is restricted only to natural parents, the author finds paternal and maternal education to have a similar
effect on children’s propensity of staying on in school. This finding is a possible indication for own
birth parents, rather than step-mothers or –fathers, to be more altruistic towards their own children (Case
et al., 2000). Interestingly, the author also finds possible evidence of the role model effect; that is,
maternal education is important for daughters’ schooling while paternal education is important for sons.
In a similar and more recent study, Chevalier et al. (2013) used data from the UK Labour Force Survey
(LFS) to investigate the relationship between early school-leaving (at age 16) and parental education
(and also income). Ordinary Least Squares (OLS) estimates indicated stronger effects of maternal
education compared to paternal education, with effects on sons being stronger than daughters. In IV
estimates that used paternal union status, and its interaction with occupation as instruments, the result
improved the significance of paternal education on the offspring’s likelihood of staying in school. The
IV estimates showed no significant effects of maternal education. Dickson et al. (2016) perform a
similar exercise using the Avon Longitudinal Study of Parents and Children to examine the causal
effects at different stages of the child’s life. They find parental education to be important at earlier age
13
(4 years) and throughout the child’s schooling years up until the age of 16. Other studies including
Sabates and Duckworth (2010), who only consider maternal schooling, and Silles (2010) have used the
school leaving age reform as an instrument to examine the causal effects on children’s cognitive and
non-cognitive outcomes. The former paper found that an additional year of maternal schooling is
significantly associated with improvements in mathematics attainment for their children, but fail to find
convincing evidence for differences in reading and behavioural outcomes. The latter paper fails to find
sufficient evidence to suggest that parents’ school had a positive impact on children’s cognitive and
non-cognitive development.
Other recent studies that use an IV technique to address endogeneity of maternal schooling include
Carneiro et al. (2013). The authors instrument maternal education with variation in schooling costs
during maternal adolescence. This study focused on a host of outcomes including cognitive
achievement, measured as test score, grade repetition and behavioural problems, using a matched data
of females from the National Longitudinal Survey of Youth (NLSY) 1979 and their children. The
novelty of this study as claimed by the authors is related to the systematic treatment of a large range of
inputs and outputs in the child’s development process at different ages. The authors further compared
the relative roles of maternal education and cognitive ability and demonstrated how maternal education
varied with different characteristics of the child, i.e. gender and race, and also with the cognitive ability
of the mother. Their results in relation to educational outcomes showed that maternal education
increases children’s performance in math and reading at the ages 7 to 8, with a reduced effect at the
ages 12 to 14. Maternal education was also found to reduce grade repetition. In terms of investment in
other inputs, the authors found that more educated mothers are more likely to invest in their children
through provision of books, special lessons, and availability of computers.
In relation to the use of other identification strategies, Pronzato (2012) uses a sample of twin parents
from the Norwegian register data to examine the intergenerational transmission of education. The use
of twins to address issues of unobserved heterogeneities is interesting in a number of ways: firstly, twins
share the same family background; experience similar lifetime events and; most importantly, share the
same genes. The twin-estimator study indicated positive and significant effects of parental education
on children’s education, although the effects of maternal education were found to be smaller (0.096 for
mother’s compared with 0.158 for fathers). The OLS estimator also showed significant effects of
mother’s (0.242) and father’s schooling (0.214) on children’s schooling. To distinguish between
monozygotic and dizygotic twins, the authors perform an exercise that allows them to select siblings
born very close together (9 to 13 months of age difference) and compare these with estimates from the
twin sample. The authors performed this exercise partly because their data did not allow them to
distinguish directly between monozygotic and dizygotic twins. The results from this exercise indicated
14
similar directional effects with mother’s schooling increasing children’s schooling by 0.139, and 0.123
for fathers. The authors did not find significant differences between the effects of mother’s and father’s
schooling using the twin estimator and the sibling estimator (with the 9 to 13 months of age difference).
Focusing on different parts of the distribution of parents’ education, the author found the effects of
maternal education strongest at the lowest part of the distribution (albeit insignificant), compared to
fathers’ where the effects were significant and stronger at the top part. This result is similar to Ermisch
and Pronzato (2011) who found the effects of maternal education to be larger among poorer educated
parents, whereas the effects of paternal education is larger among better educated parents. Their results,
which are based on Norwegian data7, indicated that an additional year of either mother’s or father’s
education increased their children’s education by as little as one-tenth of a year. Holmlund et al. (2011)
illustrate the findings across methodologies using cohorts of parents and children from Swedish register
data. Their dataset includes data on dizygotic twins with parents born between 1935 and 1943, and
children born before 1983 (at 23 or older) at the time of the study. They find no effect of the maternal
education on child outcomes when controlling for assortative mating, i.e. similarities of education levels
of mothers and fathers achieved by including the spouse’s education, but they do find a positive effect
otherwise.
A recent working paper by Papageorge and Thom (2018) further exploits advances in genetics to
explore the relationship between a genetic index, educational attainment and labour market outcomes,
using US data from the Health and Retirement Study. They use a polygenic score (a weighted sum of
individual genetic markers8) to predict educational attainment and they interact this genetic information
with childhood environments to assess their role in determining educational outcomes. They
demonstrate a strong relationship between the polygenic score and educational attainment and show
that this association differs by childhood socio-economic status. Their results also suggest the
possibility that early investments in human capital may substitute for genetic endowments in preventing
very low levels of educational attainment. However, these same investments could complement genetic
endowments in generating higher levels of educational attainment such as college completion.
Again, in Pronzato (2012), the author examines intergenerational transmission of education using a
sample of siblings with different distances in age as the identification strategy. The aim of this
experimentation was to determine whether the findings would be similar to the ones from the twin-
estimator. Similar to the twins, siblings share the same family characteristics but differ in terms of
genes. The author finds the results to be in the same direction to the results from the twin sample,
7 See Amin et al. (2015) for a more recent study on Sweden. 8 The markers most heavily weighted in this index involve early brain development and processes related to neural
communication.
15
although the effects from the sibling estimates underestimated the effects of paternal education in all
estimations. The effects of mothers’ education were larger, averaging around 0.128. While using
siblings instead of twins for studies on causal effects of parental education is more advantageous, due
to larger sample sizes, and can provide more precise estimates, evidence from such studies are limited
for the UK. An exception is Ermisch and Francesoni (2000) who analyse the BHPS and find parents’
educational attainment to be strongly associated with that of children with maternal education having a
stronger positive association with the child’s educational attainment than the educational attainment of
the father.
The use of adopted children as identification strategy has also gained considerable attention in the
literature (Plug, 2004; Björklund et al., 2006; Sacerdote, 2007; Holmlund et al., 2011). Plug (2004), for
instance, found significant effects of mothers’ and fathers’ education on schooling of own birth children.
However, when they examine the effect of parental education on a sample of adopted children, they do
not find any significant effects of maternal education on children’s schooling (they found a significant
effect for fathers). Their findings showed a significant reduction in the effects of parental schooling
(especially mothers’) between own birth children and adopted children. Björklund et al. (2006) showed
that pre and post-birth factors influence intergenerational earnings and education transmissions where
the pre-birth factors play a greater role for mothers’ education, and less so for fathers’ income. Sacerdote
(2007) uncovered evidence to show that the shared family environment explains 14 per cent of the
variation in educational attainment, 35 per cent of the variation in college selectivity, and 33 per cent
of the variation in the drinking behaviour. Additionally, there are significant, positive effects on the
adopted children’s education, income, and health when the assigned parents are more educated and have
smaller families. There is evidence to suggest that the correlation of parental education as well as family
size and adoptee outcomes is much greater than the correlation between parental income, or
neighbourhood characteristics and adoptee outcomes. This approach is appealing because there is no
genetic transmission of ability between parents of adopted children and adopted children themselves.
A major downside of this method, however, is that the process of allocating children to adopted homes
is not random and sample sizes may be smaller. In addition, parents that tend to adopt are often more
educated (hence wealthier). To the best of our knowledge, there is no evidence for the UK that uses
data on adoption to examine causal effects of parental education on children’s educational attainment.
See Holmlund et al. (2011) for a comparison of different methodological approaches.
2.1.2. Parental education and cognitive and non-cognitive skills
The role of parental education extends beyond just years of schooling. Similar to many developed
countries, the UK government is pursuing policies to raise literacy and numeracy levels of children,
16
especially in the aftermath of the introduction of the PISA (Programme for International Student
Assessment) test for OECD countries. For instance, a large proportion of the UK adult population has
poor literacy and numeracy skills (see Leitch, 2006). This has implications on individuals during
adulthood since they are more likely to be unemployed and, even if employed, they are more likely to
be in low paid jobs. It also impacts on firms, which have a higher demand for skilled workers. Available
evidence for the Czech Republic (Fischer and Lipovska, 2013) shows that parental educational
attainment has a significant impact on the initial educational level of children as well as on their lifelong
learning participation.
Many of the studies on intergenerational transmission of cognitive skills have been analysed in the
contexts of Scandinavia (Black et al., 2009; Björklund et al., 2010; Grönqvist et al., 2017), the US (Agee
and Crocker, 2002) and Germany (Anger and Heineck, 2010; Anger, 2011). Exceptions for the UK are
work carried out by Brown et al. (2011) and de Coulon et al. (2011). There is also evidence for OECD
countries of indirect effects of parental education on cognitive ability. Better-educated parents have
higher awareness of quality of life that indirectly raises the cognitive ability of their children – parental
education significantly increases children's schooling and reduces their fertility rates (Burhan et al.,
2017). In a related study by Lipovska and Fischer (2016) for the Czech Republic, the authors try to
identify who talented students are (since it could signify ability), which background they come from,
and how family background influences them. The authors find that most talented students come from
highly educated families with tradition in their field of interest and long positive attitudes towards
knowledge acquisition. They also show that there is a close relationship between talented children’s
reading ability and the accumulation of human capital. For example, if all six ancestors, comprising of
both parents and the maternal and paternal grandparents of the child gained a university degree, 82 per
cent of talented students could read before entering elementary school.
Despite the limited evidence for the UK, existing findings show a positive intergenerational relationship
between parents’ childhood cognitive skills and offspring skills (Blanden et al., 2007; Brown et al.,
2011). Using data from the British National Child Development Study (NCDS), Brown et al. (2011)
show that a parent’s test scores at age seven have a positive influence on the performance of their
children when they were as young as five years old. Using the same dataset, de Coulon et al. (2011)
draws on the 1970 cohort and uses test scores of parents at age 34, which they compare with test scores
of their children at the ages three to six. Their results show strong evidence that parents with better
numeracy and literacy in adulthood have children who perform better in early cognitive and non-
cognitive tests.
17
Evidence on the intergenerational transmission of non-cognitive skills is limited. In a recent study,
Anger and Schnitzlein (2017) estimate sibling correlations in cognitive (Intelligence Quotient (IQ) tests)
and non-cognitive skills (locus of control, reciprocity, and the Big Five personality traits) based on data
from the German Socio-Economic Panel Study. Although the study does not clearly distinguish between
genetic and environmental effects, the authors find sibling correlations of personality traits range from
0.22 to 0.46, while sibling correlations in cognitive skills were found to be higher than 0.50. Their
results show that around half of the inequality in cognitive abilities can be explained by shared family
background.9 A similar exercise by Grönqvist et al. (2017) reveals that, despite the positive relationship
shown by previous studies, not correcting for measurement error in cognitive and non-cognitive abilities
substantially underestimates the effect of intergenerational ability correlations. They use two sets of IVs
to correct for paternal abilities; ability evaluations of father at age 14 and ability evaluations of the son’s
uncle. For maternal abilities they use evaluations of their brothers. Specifically, they find mother–son
correlations in cognitive abilities are stronger than father–son correlations, whereas no such difference
is apparent for non-cognitive abilities.
Other aspects of family background have been considered in the literature. For instance, Coelli (2011)
investigates the link between parents’ job loss and the enrolment of children in universities or post-
secondary education institutions. Using data for Canadian households from the Survey of Labour and
Income Dynamics for the period 1993 to 2007, the author found a negative effect of job loss from
layoffs or business failure on children enrolment in universities or community colleges. The authors
attempted to address possible sample selection bias but there was not enough evidence to suggest that
estimated results suffered from selectivity bias in the first place. Chowdry et al. (2013) also analysed
the impact of both family income and education on the participation rates of pupils in UK higher
education. They used linked data from the NPD, the National Information System for Vocational
Qualifications, and the Higher Education Statistics Agency (HESA) for two cohorts of individuals
between 2004 and 2005 and between 2006 and 2007 aged from 11 to 20 years old. They show that,
although family income matters in explaining lower participation rates in higher education for pupils
from poorer families, the most important factor seems to be pupils’ achievements in secondary
education. These findings were true for both private and public-school pupils and suggested the need
for policy to improve the access to higher education for pupils from a lower socio-economic background
(as also suggested by Coelli, 2011).
2.1.3. Family structure and offspring education
9 See also Anger and Heineck (2010).
18
An extensive body of research has identified family characteristics as key factors associated with later
adulthood achievements and education choices (McLanahan and Sandefur, 1994; McLanahan et al.,
2013). Findings from these studies are informative for policymakers and suggest where interventions
could address the issue of early inequality and intergenerational transmission of disadvantage. This
literature emphasises that there may be differences in children’s outcomes as a result of growing up in
different families. Family structure affects children’s outcome by determining parental patterns of
investment. Empirical studies have mostly focussed on five home environment characteristics: family
size, birth order, parental presence, gender of siblings and spillover from siblings. By using linked
administrative data, researchers have tried to identify the causal impact of childhood conditions on
achievement with precision. As these studies require rich longitudinal information on all family
members, they have mostly been based on Scandinavian, US and Israeli administrative records, with
the exception of one study based on English education records. Below we describe the most recent
literature grouped by family characteristics.
Family size
Early work based on observational studies usually finds large negative associations between family size
and the attainment of the offspring. These results may be driven by the fact that family size is usually
correlated with unobservable characteristics of the family that also affect children’s achievements.
Empirical research that tries to identify the causal link between quantity of children and their outcomes
is limited. Black et al. (2010) use rich administrative data linked to military records for recent cohorts
of Norwegian men and study the effect of family size on IQ by instrumenting the family size with two
different exogenous variables: 1) the event of a twin second or third birth in the family; 2) the sex
composition of the first two births. They find negative effects of third or higher parity births on IQ (-15
per cent) of first and second born individuals when they use the twin instrument. When they use the
gender composition as an instrument, they find no effect of family size on IQ. It is however likely that
the difference in these IV estimates results is driven by idiosyncratic characteristics of twin births; twins
are usually born with lower birth weight and no spacing between them, requiring therefore more
investment in terms of money and time from their parents and therefore a reallocation of resources from
other siblings. Further data analysis by the same authors shows that estimates based on twins as
exogenous variation in family size should not be generalised to other types of variation of family size
and pertains to families with twins.
Angrist et al. (2010) combine the two instruments used by Black et al. (2010), which are the event of a
twin second or third birth in the family and the sex composition of the first two births and use Israeli
19
census data linked to the population registry. This study finds no effect of family size on schooling of
older children. The authors argue that their results are driven by the choice of empirical strategy, which
is based on the joint use of the instruments and the high fertility population of interest, with demographic
and social characteristics closer to developing countries.
Birth order
There is an extensive literature on the relationship between birth order and educational attainment that
finds consistent negative associations: later-born children have lower educational outcomes than their
older siblings and this effect tends to be larger in bigger families. This literature review will consider
three recent interesting studies that investigate the channels through which birth order is likely to affect
education outcomes. Booth and Kee (2009) use retrospective family background data from wave 13 of
the BHPS and create a birth order index that accounts for family size; they find that later-born children
receive a smaller share of their parents’ resources. Similarly, Pavan (2016) uses a US survey and family
fixed effects to estimate the birth-order effect. The results show a sizeable positive firstborn effect on
many cognitive tests. An investigation of the possible mechanisms explaining these results documents
considerable gaps in parental investment across birth order. Lehmann et al. (2018) examine the possible
causes of the achievement gap between later-born children and their older siblings by exploring US
survey data. In particular, they find that mothers take more risks during pregnancy and are less likely
to breastfeed and to provide cognitive stimulation for later-born children.
In a more recent non-UK study, Breining et al. (2017) use data from the Danish Birth Register and data
from Florida Departments of Education and Health to examine possible mechanisms of differences in
birth order for Denmark and Florida. They find surprisingly that second-born children are no worse off
in terms of the quality of school attended and are also more likely to attend pre-school. The authors
however find that second-born boys, compared to older siblings, achieve lower reading and math test
scores in Denmark and in reading scores in Florida.
Parents’ absence
The absence of one of the parents is likely to have a lasting effect on children’s achievement through a
substantial reduction of investments in their human capital. The absence of a parent because of his or
her death is more likely to be an exogenous event compared to the absence caused by divorce or
abandonment. However, the empirical evidence is scant as the precise identification of causal
relationships requires large administrative datasets. Kalil et al. (2016) use administrative data from
Norway to analyse how fathers’ presence affects the intergenerational transmission of educational
20
attainment. They exploit within-family variation in father exposure that occurs across siblings in the
event of paternal death, and find that longer paternal exposure amplifies the father-child association in
education and attenuates the mother-child association. These effects are stronger for boys than for girls.
Gould et al. (2019) use a big sample of Israeli children born between 1974 and 1991 and estimate the
effect of parental death and divorce. Specifically, they test the effects of parental schooling on three
separate sets of children: children that did not lose either parent by age 18; children that lost a mother
or a father before the age of 18; and children that lost a mother or a father, but after the age of 18. Using
variation in the age of a child when the parent died across these groups, the authors test the idea that a
child’s human capital should depend on whether that parent was alive and able to interact with the child
before the end of secondary education. Using divorce as a source of parental influence is analogous
where it is tested whether mothers’ education (the custodial parent in most cases) becomes more
important relative to the father if the divorce occurred before the age of 18. Their results show that
parental education impacts significantly on children’s human capital accumulation and that the size of
this effect depends on the length of time parents and children spend together. For instance, when a
parent dies, his or her education becomes less important for the child’s outcomes while the other
parent’s education plays a larger role. This is in line with much earlier findings which suggest that
children growing up in single-parent headed households, rather than with both parents present, achieve
lower education levels (McLanahan and Sandefur, 1994) and that children raised with both biological
parents present have better educational outcomes than children from blended families, i.e. are
stepchildren or have half-siblings (Ginter and Pollack, 2004).
Children who are, or have been, in care are among the lowest performing groups in educational
outcomes internationally (Flynn et al., 2013). By linking data from the NPD and the Children Looked
After Database (CLAD), Sebba et al. (2015) assess how educational attainment differed between
Children in Need (CIN), Children Looked After (CLA) and Children not in care or need (Comparison
Group). CLA were split into separate categories: long term early entry (in care 12 months or more
continuously at the end of KS4 who were also in care at the end of KS2), long term late entry (in care
12 months or more continuously at the end of KS4 who were not in care at the end of KS2), and short
term (under 12 months in care at the end of KS4). The authors find that the comparison group
significantly outperformed all other groups in terms of mean GCSE points scores. Out of CIN and CLA
groups, CLA short term significantly underperformed, with the authors citing a possible reason for this
being that, those entering care in adolescence are associated with more challenging difficulties. CLA
short term – are therefore less likely to achieve greater educational attainment. The author’s cite
absences, exclusions and school moves as explanations for the large amount of the disadvantage that
CIN and CLA suffer. DFE (2019) showed supporting evidence of CIN under achieving academically,
followed by CLA, followed by children not in need nor looked after. Results are further supported by
21
analysis carried out by Sebba et al. (2015), that CIN and CLA are much more likely to have at least one
fixed period exclusion.
Sibling gender
Gender composition of siblings is a key aspect of the childhood family environment. In recent years,
economics and education researchers have studied its role in the determination of intra-family gender
norms that affect education choices, particularly of girls. Brenøe (2018) uses Danish administrative data
to estimate the effect of having a younger sister on first-born education choices. The identification is
possible because of the random assignment of gender of the second-born child. This study finds that
having a second-born brother relative to a sister increases first-born women’s gender conformity:
women with a brother are less likely to study STEM (Science, Technology, Engineering and
Mathematics) subjects. Further analysis shows that parents of mixed-sex children invest their time more
gender-specifically to their first-born daughter compared with parents of same-sex children. These
findings are supported by Cools and Patacchini (2017) and Rao and Chatterjee (2017); their research
based on US survey data provides evidence that women with brothers hold more traditional gender
attitudes than those without brothers.
Sibling spillover
Black et al. (2017) use administrative data matched to education and medical records of very recent
cohorts from Florida and Denmark to estimate the effect of siblings spillovers in the presence of a family
shock. Specifically, the authors focus on families with three children and use a difference-in-difference
(DiD) research design that controls for birth order to estimate the effect of having a disabled third
sibling. With both sets of data, the study finds consistent evidence indicating that the second child in a
family is more adversely affected relative to the first born when the child is disabled, and these results
survive several robustness checks. Qureshi (2018) uses US administrative data to investigate the
spillover effect of an older sibling’s achievement on younger siblings. As siblings share the same home
environment, it is likely that some unobservable characteristics of the family affect the human capital
of all siblings. To estimate the causal effect of older siblings’ human capital on younger siblings’
outcomes, Qureshi instruments the achievement of older siblings with an indicator of the quality of their
teachers. This study finds that an improvement in older siblings’ test scores leads to a statistically
significant increase in younger siblings’ test scores in maths and reading. This result is important as it
shows the role of schools in reinforcing social inequalities as well as its role in social remediation.
Similar results are found by Breining et al. (2015) as they investigate the spillover effects of early-life
medical treatments on the siblings of treated children. By using administrative data from Denmark and
22
a regression discontinuity design that exploits the changes in medical treatments across the very low
birth weight cut-off, the study finds substantial positive effects on academic achievement of siblings of
treated children. An analysis of the possible mechanisms suggests that improved interactions within the
family may be an important pathway behind the observed spillover effects. Breining (2014) uses high
quality register data from Denmark to study the spillover effects on firstborns from having a younger
sibling suffering from an attention deficit hyperactivity disorder. Estimating a model with cousin fixed
effects the author finds that the educational outcomes of healthy firstborn children are significantly
reduced by the presence of a disordered sibling.
Finally, Nicoletti and Rabe (2017) estimate the direct effect of sibling spillover in school achievement
by using the NPD which includes administrative education records for pupils in England. They exploit
the variation in school test scores across three subjects observed at the two ages as well as variation in
the composition of school peers between siblings. These results provide empirical evidence of positive
spillover effects from the older sibling to the younger, but not vice versa, and reinforces the conclusion
of the paper by Qureshi (2018) that school can have an important role in social remediation through
intra-family spillovers.
2.1.4. Wider family and environmental factors
Case et al. (2005) investigate the lasting effects of mother’s prenatal health, childhood health and
economic circumstances on adult health, employment and socioeconomic status. Using data from the
NCDS the authors follow a birth cohort from birth into middle age. OLS Regression is used to predict
total O-level passes by age 16, controlling for parental income, education and social class. Among other
findings, the authors show that mother’s smoking during pregnancy is associated with significantly
fewer O-levels passes, with children whose mothers had reported smoking heavily during pregnancy
passing 0.4 fewer O-level exams. Korhonen et al. (2012) use longitudinal data to assess how maternal
prenatal, postnatal and/or current maternal depressive symptoms is associated with measures of the
child’s wellbeing (social competence). Evidence suggest that concurrent depressive symptoms are a
risk for both sexes of children, whereas prenatal and postnatal depressive symptoms are only a risk for
boys.
Desforges and Abouchaar (2003) carry out a literature review assessing how parental involvement,
parental support and family education affect pupil achievement. Numerous studies show how differing
parental involvement variables linked to the child’s outcomes. Parental involvement/support and family
education variables included: reading, library visits, playing with letters and numbers, painting and
drawing, teaching (through play) the letters of the alphabet, playing with numbers and shapes, teaching
23
nursery rhymes, singing, having an educated mother, a helpful father, parental involvement in support
of schooling and degree of discipline exerted by the parents. These factors are primarily assessed against
educational attainment, as well as other more specific mechanisms such as cognitive development,
sociability, confidence and resilience. The common theme across much of the literature was that
parental involvement does have a significant positive effect on children’s behaviour and achievement,
whilst controlling for factors such as social class or family size. There was also some evidence that the
parental involvement effect diminished slightly over time as children reached school leavers age, though
the relationship is relatively strong. The authors suggested there is perhaps less of a relationship with
achievement, but a greater relationship with staying on rates and pupils’ educational aspirations. Lessof
et al. (2018) use the Longitudinal Study of Young People in England (LSYPE) to assess how different
aspects of the home environment, pupils’ actions and personal characteristics affect GCSE scores.
Bivariate analysis and multilevel modelling approaches were implemented. Key findings show that
children were more likely to attain higher GCSE scores if they: are living with both their biological
parents, have a mother who is highly educated, have parents who always discussed their school reports,
are living in a mortgaged/owned property, are provided with internet-connected home desktops/laptops,
are less psychologically stressed, do not truant, usually do all their homework and are female. The
authors also went on to explain the gender differences in GCSE scores, citing that: boys have greater
likelihood of special educational needs and disabilities which affect schooling, as well as parents being
more likely to expect their daughters to attend university and less likely to want their daughters to look
for options outside education after year 11 (e.g. apprenticeships).
Sammons et al. (2015) investigate the effects of pre-school and early home learning on A-level
outcomes. Logistic regression and multilevel model techniques are employed on the NPD and Effective
Pre-school, Primary and Secondary Education dataset. Conclusions from the analysis were that,
attending pre-school improved students are more likely to partake in A-levels; although, there is an
insignificant effect on the students grades in A-levels. However, the quality of the home learning
environment that pupils experience before they attended school does have a significant effect on both
staying on for A-levels as well as the grades achieved at this stage. Home learning environment is based
on the frequency of specific activities involving parent and child, including: teaching the child the
alphabet, playing with letters and numbers, library visits, reading to the child, teaching the child songs
or nursery rhymes. The authors suggest that pre-schooling largely operates through increasing student’s
GCSE grades, whereas early home learning has a more lasting effect through to overall A-level
attainment. Currie and Almond (2011) survey recent work into human capital development under the
age of 5, stating “Child and family characteristics measured at school entry do as much to explain future
outcomes as factors that labour economists have more traditionally focused on, such as years of
education”. Studies surveyed are based from data from the NLSY and the NCDS. NLSY studies
24
included that of Cunha and Heckman (2008), who assess the extent to which parental investments (e.g.,
number of child’s books, whether the child has a musical instrument, whether the child receives special
lessons, how often the child goes to museums, how often the child goes to the theatre, etc) increases the
child’s future earnings and likelihood of graduation from high school, and how much of the increase is
as a result of cognitive or non-cognitive skills. Results showed increasing parental investment by 10%
at ages 6-7 increases children’s earnings by 24.9% (with both cognitive and non-cognitive skills equally
as important) and increases likelihood of graduating high school by 64.4% (mainly through cognitive
skills). Additionally, Dearden (1998) uses data from the NCDS to show that child’s reading and maths
scores (age 7), type of school, family composition, teachers assessment of interest shown by parent in
their child’s education, type of school attended, family financial status, region, fathers socio-economic
status and parental education are all significant predictors of the number of years the child stays in full
time education.
Covay and Carbonaro (2010) analyse after-school activities in more detail. Using a US early childhood
longitudinal study, OLS regression is used to predict cognitive and non-cognitive skills. Different
models were created including extracurricular activities (EAs) only; SES only; SES and EAs; EAs, SES
and the interaction between EAs and SES. The authors found that participation in different types of
extracurricular activities (sports, dance, music, arts and performing arts) explain a moderate portion of
the SES advantage in non-cognitive and cognitive skills. Results suggest that pupils who participate in
sports benefit more than those who took part in other extra curriculum activities. Schuller and
2.2. The role of health in human capital accumulation
Although the vast majority of the early literature on human capital focuses mainly on returns to
education (Mincer, 1958; Psacharapoulos and Patrinos, 2018), health has been increasingly recognised
as a crucial component of human capital (see Becker, 1962; Mushkin, 1962; Becker, 2007). This was
especially true since Grossman (1972a, 1972b and subsequently 2000, 2006) developed a model for the
demand for health capital in the early 1970s. The model assumes that an initial health stock is inherited
by individuals, which depreciates with age and can be increased through health care consumption,
nutrition and exercise. These types of investment improve well-being and increase the available time in
which individuals can work and accumulate wealth. Though scholars have acknowledged that health
plays a significant role in generating human capital, existing measures of its stocks do not adequately
account for health status.10 The existing measures take account of mortality by incorporating survival
10 Recommendations for new measures have been developed by the Atkinson Review (Atkinson, 2005), the
Commission on the Measurement of Economic Performance and Social Progress (Stiglitz et al., 2009) and a
consortium of 15 OECD countries, Israel, Russia, Romania, Eurostat, the ILO and ONS.
25
rates into the model but ignore any other aspect of health, focusing mainly on the contribution of
education.
With an ageing population, health has a profound but ambiguous impact on labour market activity and
its outcomes and, therefore, on human capital accumulation. Longevity may not necessarily be
associated with good health, as survival into older age is increasingly accompanied by morbidity.
Therefore, the willingness to supply labour may decline, which is reflected in earlier retirement or
increasing absenteeism from work. Against this, poor health is associated with medical care needs,
which often require financial commitments at an accelerating rate for the elderly. Therefore, labour
supply may actually increase in order to compensate for these financial constraints. If poor health does
not lead to absence from work or early retirement, it can still reduce on-the-job productivity and work
quality, which is referred to as presenteeism in the occupational health literature (Burton et al., 1999).
Consequently, morbidity impacts on the quality, as well as the quantity of labour supplied which affect
human capital stock by reducing the productive capacity of the workforce. In addition, there are likely
to be spillover effects on the work routine of healthy co-workers (Goetzel et al., 2003), which further
increases the costs of ill-health to the economy. Work by O’Mahony and Samek (2019) has quantified
the combined effects of both absenteeism and presenteeism for the UK economy, and has found that
human capital would increase by 11 per cent if workers in poor health moved to good health.11 This is
found to be driven by a number of factors, including a larger effect for males than females reflecting
higher retirement rates, lower relative wages due to poor health, and the fact that males account for a
greater share of the aggregate human capital stock due to their higher earnings. Likewise, while the
human capital stock reduction due to ill-health for those with low qualifications is relatively lower than
for those with university degrees, most people in poor health also have lower qualifications.
This section reviews studies on the relationship between health and human capital. In particular, it
discusses ageing and the relationship between health and education, also known as the education-health
gradient; how health impacts on labour supply; health care access and its provision. Finally, the link
between earnings and health is considered by reviewing studies which have examined the causal impact
of earning effects on health. The reverse relationship, i.e. presenteeism, is further discussed in the
literature review on earnings determinants.
2.2.1. Ageing
11 They evaluated the extent to which poor health reduces the potential human capital by removing people from
the active population and reducing the productive human capital stock by diminishing the effectiveness of human
capital, through absenteeism and presenteeism.
26
Almost every country is projected to experience an increase in the proportion of the population aged 60
and above between 2005 and 2050 (United Nations, 2009). This will undoubtedly affect economic
growth through changes in labour supply and human capital accumulation. This is related to the fact
that the elderly tend to work and save less and often require financial support to afford costly health
care. The shift in the age distribution affects aggregate human capital stocks because longevity is often
accompanied by worsening health, and relatively large cohorts with low or moderate levels of formal
education are replaced by relatively small cohorts with high levels of formal education. While this
suggests a reduction in overall labour supply, migration, which is discussed in detail in chapter 2.4, can
mitigate some of these adverse effects of ageing by shaping the redistribution of human capital within
countries. Additionally, Bloom et al. (2010) predict that OECD countries will experience only a modest
growth decline as a result of these demographic changes because of increases in the statutory retirement
age and behavioural changes. Although there is no empirical evidence yet on the implications of the
abolishment of the UK Default Retirement Age in 2011, similar policy reforms in other European
countries find expected or actual retirement ages to increase as a result (see Mastrogiacomo, 2004;
Bottazzi et al., 2006 on Italy12; Coppola and Wilke, 2014 on Germany; Manoli and Weber, 2016 on
Austria). Coppola and Wilke (2014), for instance, apply a DiD procedure in studying the German case
and observe an increase in the expected retirement age by approximately two years. Their results are in
line with figures obtained from studies on Italy. Manoli and Weber (2016) use a regression kink design
and find individuals delay their retirement in Austria by five months with every one-year increase in
the expected retirement age. If working lives are prolonged, returns to investments in education and the
updating of skills to increase productivity increase and take place more frequently. While it can be
argued that different conclusions can be drawn when focussing on individuals in poor health, the elderly
have become healthier over time (either because of shorter periods and/or later periods of morbidity, or
because of longer lives) which is often referred to as the compression of morbidity.
One strand of the literature argues that the compression of morbidity is a result of intergenerational
spillovers from well-educated children to their parents. This arises as the offspring has better knowledge
about health, more resources to invest in their elderly parents’ health, they can improve their parents’
morale and hence their mental health, and arte able to provide informal care, medication adherence or
act as their agents in the health and long-term care system (Zimmer et al., 2007; Torssander, 2013;
Friedman and Mare, 2014). Empirical evidence for this causal relationship is found by Lundborg and
Majlesi (2018) who use a Swedish compulsory schooling reform, which was implemented in the 1950s
and 1960s at different points in time across municipalities. They find that children’s years of schooling
(imputed from the highest educational attainment according to Holmlund et al.’s (2011) approach)
12 Brugiavini (1997) exploits the same strategy and uses Italian data but does not find any significant changes in
retirement behaviour.
27
drives their parents’ longevity.
2.2.2. Health and education
Understanding the sources of health inequalities requires understanding the role of an individual’s
education, behaviour and preferences. While it is widely accepted that there is a gradient linking
educational attainment and health, the direction of causation remains unclear. Does better education
result in better health or do healthier people become more educated? Or are there confounding factors
which improve health and education at the same time? A growing body of literature suggests that
education has benefits beyond increases in labour market productivity which extend to health and other
outcomes. There are different ways in which education possibly improves health, which are discussed
by Cutler and Lleras-Muney (2008) in detail and are summarised as follows: more educated people can
be more likely to work in a relatively safe environment (as opposed to lower skilled occupations); they
can be more likely to exercise and engage in healthy behaviour as they can to be more aware of the
health consequences of their behaviour; they can have better health insurance and experience better
access and higher quality of health care provision; and, lastly, they can be more future-oriented and
hence invest more in long-term health.
The majority of work, which examines the causal effect of education on health using biomarkers from
the Health Survey for England, exploits the increase in the minimum school-leaving age in 1947 and/or
1973 in the UK, which acts as an exogenous change in the amount of schooling received. However,
evidence is mixed. Neither Clark and Royer (2013) nor Jürges et al. (2013) find any evidence that more
schooling reduces the risk of adult hypertension or blood fibrinogen and C-reactive protein levels,
respectively and Blanchflower and Oswald (2008) find an adverse effect of schooling on blood pressure,
albeit of very small magnitude. Powdthavee (2010) finds that one extra year of schooling reduces
hypertension by slightly more than 10 per cent for both men and women. Using the New Earnings
Survey Panel Data, Devereux and Hart (2010) also detect an impact on hypertension but only for men
and not for women. Besides these studies, there is little evidence on the impact of education on
objectively measured health in the context of the UK.
Other work that also exploits the increase in the UK school-leaving age as an instrument for years of
schooling but produces findings derived from self-assessed health outcomes, also comes to mixed
conclusions. For instance, Silles (2009) uses the General Household Survey for England, Scotland and
Wales in her IV approach and finds that an extra year of secondary school increases the probability of
being in good health by 4.5 percentage points. A very similar approach was employed by Oreopoulos
(2006). Although this paper’s main focus was on the education-earnings nexus, other health outcomes
28
were taken into account and by using the General Household Survey in an IV approach, he only finds
small and statistically insignificant effects of education on self-assessed health. These results are
consistent with findings by Clark and Royer (2013), who build on Oreopoulos’s (2006) work by
combining the General Household Survey with the Health Survey of England in a regression
discontinuity design.
A very different approach is adopted by Conti et al. (2010), who use the British Cohort Study (BCS) to
follow individuals born in 1970 through the age of 30. They estimate a multi-factor model of schooling,
earnings and health and find that education has an important causal effect on smoking behaviour (it
explains 60 to 70 per cent of the differences in smoking behaviour), self-reported health (35 to 55 per
cent) and men’s obesity (one third). They observe the largest effects for the highly cognitive skilled
population and that non-cognitive skills are almost as important as early health endowments in
explaining these three health outcomes in adult life.
Less commonly examined in the literature, especially in the UK context, but equally relevant is the
health gradient in education, i.e. healthier adolescents selecting into higher education. Poor health can
delay cognitive development; it can lead to social isolation and disengagement from school; and it can
impact on perceived prospects and life expectancy reducing long-term investments in education.
Equally, factors such as family background, psychological traits and cognitive as well as non-cognitive
skills can all be confounding factors explaining the education-health gradient. The large majority of
work on this relationship is based on the US and uses the 1997 cohort NLSY to examine the mechanisms
linking health to the educational attainment of adolescents (Haas and Fosse, 2008; Lynch and von
Hippel, 2016; Benson et al., 2018; Carroll et al., 2018). By estimating standard logit and sibling fixed-
effects models, Haas and Fosse (2008) find that, after controlling for sociodemographic family
characteristics, adolescents in poor self-assessed health are less likely to complete high school by their
20th birthday and to transition to post-secondary education, which is partly driven by cognitive and
academic achievements as well as psychosocial factors. These findings are consistent with the results
reported by Lynch and von Hippel (2016) who use self-assessed health at age 17 in an ordinal logistic
regression model to predict highest degree obtained by age 31 and argue that the positive relationship
is a result of early advantages in academic performance, college plans, and family background. Benson
et al. (2018) and Carroll et al. (2018) also use an ordinal logistic regression but use obesity and a number
of different health impairments, respectively. Benson et al. (2018) find that overweight adolescents are
less likely to obtain high school diplomas and bachelor's degrees while Carroll et al. (2018) find that
individuals, who have mental or multiple health impairments prior to high school completion or college-
going, are less likely to initially enroll in 4-year postsecondary institutions than healthy adolescents. A
slightly different approach is applied in early work by Palloni (2006), who uses the 1958 cohort of the
29
NCDS in a Monte Carlo simulation model and finds that childhood health is important but not the only
determinant in accession to adult social class positions.
This empirical evidence highlights the role of health selection processes in generating socioeconomic
inequalities in early adolescence to young adulthood. Moreover, together with the fact that causal effects
of education on health are often smaller than the observed cross-sectional relationship between
education and health amongst adults, these results suggest that covariates are inadequate in cross-
sectional analyses as they cannot fully account for the selection of healthier adolescents into higher
education and thus the causal effect of education on health is likely to be overestimated.
2.2.3. Health and labour supply
Latest figures published by the ONS (2018) show that the average absence due to sickness has almost
halved since records began in 1993 with employees now taking around 4 days per annum, compared to
approximately 7 days 25 years ago. Sickness absence has been falling since 1999 and the proportion of
working hours lost due to sickness absence has decreased from 2.4 to 1.9 per cent since the financial
crisis. While this is a very promising development and suggests health improvements, the observed
reduction in absenteeism might take place at the expense of rising presenteeism, where individuals work
in poor health. Reducing workplace absence due to poor health has always been of interest but its
importance is reinforced by Dame Carol Black’s review in 2008 on the health of the working age
population, which estimates that the total economic cost of ill health is over £100 billion in the UK
alone. A report published by the Sainsbury Centre for Mental Health (2017) highlights that mental
health problems alone costed employers almost £35 billion in 2016, of which almost £11 billion came
from sickness absence. Presenteeism contributed nearly twice that and around £3 billion was a result of
staff turnover. Interestingly, the cost of sickness absence has increased by 26 per cent from £8.4 billion
in 2006, mainly as a result of an increase in the total number of employees in the UK (increasing the
costs at the aggregate level) and the number of workers with mental health problems.13
There have been many governmental initiatives in promoting health and increasing awareness of the
role employers can play in preventing poor health and facilitating job retention. However, the literature
on the link between health and hours worked is scarce (an exception is a study carried out by Cai et al.
13 They calculate the average cost of a working day lost based on national accounts data for average gross
compensation per employee (i.e. wage plus national insurance and pension contributions), adjusted downwards to
take account of the evidence given in absence surveys that lower-paid workers tend to take more time off work
than those on higher earnings.
30
(2014) in Australia14), especially in the context of the UK. Existing UK-based research has instead
focused on the effects of health on labour force entry and exit (e.g. García-Gómez et al., 2010) and in
particular on the effect of third party’s health on labour supply decisions. The latter literature strand has
been growing in recent years due to the rising need for informal care (Carers UK, 2014), which is likely
to have an impact on labour market outcomes of healthier individuals. We further elaborate on this
literature in the following section.
2.2.4. Provision of and access to health care
Although life expectancy has increased in England over the past few decades, a social gradient across
the country exists, with people living in the most deprived areas having on average the shortest lives
(Public Health England, 2017). For instance, life expectancy is highest for both men and women in the
South East with 80.5 and 84.1 years, respectively. It is the lowest in the North East with and expectancy
of 77.9 and 81.6 years for men and women, respectively.
Lenhart (2019) uses the BHPS and finds that negative health shocks increase health care usage and the
likelihood that individuals pay for this care out of their own pocket. This is particularly likely
immediately after the onset of the health shock, although its effect remains large and significant even
in later periods. Since annual GP visits and hospital night stays, which are two of the health care usages
he captures, imply absenteeism from work, they may also partially explain associated earning losses.
Even though health care is universally provided by the National Health Service (NHS), it is likely that
individuals want to forego long waiting times and receive immediate treatment through private health
care. This is underpinned by the fact that for the first time since the implementation in 2015 the NHS
did not meet the goal of treating at least 92 per cent of patients within 18 weeks (Murray, 2016).
These types of health care usage are likely to reduce labour and to a larger extent household income, if
other household members also take time off work to help with GP appointments and hospital visits or,
in extreme cases, engage additionally in informal care (see previous section). Therefore, health care
increases longevity, labour participation and productivity but it can also divert labour away from
production because the available time to provide care is constrained, increasing pressure on combining
caregiving and work duties. The additional burden of caregiving is associated with an opportunity cost
since time needs to be taken away from work or leisure to carry it out, which is often called the
14 They use the Household, Income and Labour Dynamics in Australia and find that lower health status results in
fewer working hours. They use a dynamic random effects Tobit model to account for zero working hours and
follow Heckman by predicting individual health status to account for measurement error and endogeneity.
31
substitution effect. Carmichael and Ercolani (2016) estimate that unpaid caregiving will draw in more
than a third of the UK adult population at some point during their lives.
The literature on employment outcomes and health effects of informal caregiving has been growing in
recent years, especially as the ageing population increases in the UK. Older individuals are increasingly
expected to care for relatives while governments extend their working lives at the same time by
increasing retirement ages. By following individuals aged 50 to 75 from the Understanding Society
Survey and using discrete-time survival models, Carr et al. (2016) find that co-residential caregiving,
i.e. when the caretaker and the care recipient live within the same household, increases the likelihood
of exiting work. When distinguishing between care recipients, they find that women are more likely to
exit work when caring for their spouse while men do not adjust their working hours if their spouse is in
poor health. Female part-time and full-time workers who cared for more than 10 hours per week were
between 2.64 and 4.46 times more likely to leave work, compared to female non-carers.
Bauer and Sousa-Poza (2015) find that, albeit the small proportion of the labour force affected,
caregiving tends to lower the quality of the caregiver’s psychological health, which also has a negative
impact on physical health outcomes. Michaud et al. (2010) use the BHPS and rely on caregiving
dynamics to account for unobserved heterogeneity. They find a negative effect of co-residential
caregiving on future employment and a negative effect of employment on future co-residential and
extra-residential caregiving, i.e. when the caretaker and the care recipient do not live in the same
household. 15 In an earlier mixed methods study, Carmichael et al. (2008) find that the length and the
intensity of the caring episode, together with potential financial considerations and employers' attitude,
are amongst aspects of the decision-making process that constrain carer’s labour supply.
However, empirical evidence of caregiving histories has often drawn on time lags and leads of
caregiving, employment or family circumstances to explain the relationship at a certain point in time
(see Stern, 1995; Heitmueller, 2007; Michaud et al., 2010; Carmichael et al., 2010). Only in recent work
by Carmichael and Ercolani (2016) the caregiving trajectories over life-courses are explored. Using the
harmonised BHPS-Understanding Society Survey in an optimal matching and cluster analysis, they find
that individuals differ in terms of life-stage, gender and attitudes towards family and gender roles prior
to their caregiving duties. By applying a DiD approach, the authors show that those following the most
caregiving-intensive pathways not only end up poorer but also experience a relative decline in
subjective health and wellbeing.
15 Carmichael et al. (2010) use the BHPS to examine the reverse relationship and find that employment
participation and earnings both impacts negatively on willingness to supply informal care.
32
2.2.5. Earning effects on health
There is now a very substantial literature that has established a strong association between higher socio-
economic status and better health of individuals, which has been observed in numerous countries and
across a wide range of different health outcomes. These results are also extended to broader family
relationships as studies show a link between parental income and their children’s health (Adda et al.,
2009; Currie et al., 2007; Propper et al., 2007). The WHO highlights that children in the bottom
percentiles of the income distribution are twice as likely of experiencing harmful illnesses or premature
death compared to the ones at the top of the distribution (UNICEF, 2018).
However, this raises the question of whether poor health is the result of low income as it reduces the
ability to work, or whether it is rather the cause. If individuals maximise their utility, which is
considered to be a function of health and the consumption of other goods, subject to budget and time
constraints, a sudden income increase relaxes the budget constraint resulting in better health (assuming
health is a normal good). However, health is likely to depend on other components of the utility
function, which can be positively correlated with utility but negatively correlated with health. Such
examples include risky activities or smoking, which have an adverse effect on health, and as such higher
income would have an ambiguous effect on health overall. Moreover, genetic endowment or school
quality are likely to impact on both health and earnings, which explains their correlation but does not
suggest any causal relationship. Therefore, scholars have increasingly employed natural experiments to
address the issue of reverse causality (health as a determinant of earnings is discussed at a later stage in
this literature review) by examining health effects of the Great Recession of 200816, or exploiting
governmental assistance programs of lower income families17, lottery winnings18 or the introduction of
a minimum wage. The majority of the literature has focused on other countries than the UK, and only
very recently have these approaches been applied to the context of the UK using exclusively the BHPS.
Recently published work by Reeves et al. (2017), Kronenberg et al. (2017) and Lenhart (2017) use the
minimum wage legislation, implemented in 1999, which increased hourly wages to at least £3.60, to
16 Iceland: Ásgeirsdóttir et al., 2014; Sweden: Cesarini et al., 2016; US: McInerney et al., 2013 17 US: Hoynes et al., 2011; Evans and Garthwaite, 2014; Hoynes et al., 2015; Canada: Milligan and Stabile, 2011 18 See Apouey and Clark (2015) for a UK based study which uses monetary lottery winnings as exogenous shocks
to individuals’ wealth to estimate its effect on general self-assessed (whether the respondent is in excellent health
or not), mental and physical health as well as health behaviours, such as smoking and social drinking. They find
that the sudden increase in income has a significant effect on mental but not on overall health, which they observe
is driven by the association between lottery winnings and smoking and social drinking. Since general health
encompasses mental health as well as these behaviours, it does not improve following the positive income shock.
For earlier work, see Gardner and Oswald (2007).
33
explain earnings effects on health in the UK.19 They all use different time periods from the BHPS and
employ a DiD approach to compare health effects of higher wages on recipients of the minimum wage
with otherwise similar but unaffected individuals. Although Reeves et al. (2017) and Kronenberg et al.
(2017) both examine the legislation’s effect on mental health, using the 12-item General Health
Questionnaire as a continuous variable, they come to different conclusions. While Reeves et al. find
that higher wages improve mental health with an effect size (0.37 of a standard deviation) comparable
to the impact of antidepressants on depressive symptoms (0.39 of a standard deviation), Kronenberg et
al. do not identify any significant effect. However, Kronenberg et al. use a longer pre-legislation period
(waves 7 to 9 compared to waves 8 and 9 in Reeves et al.) and both studies differ with respect to the
definitions of treatment and control groups. Both scholars use a wage based-comparison, as one set of
groups where they examine eligible recipients versus ineligible non-recipients. In an alternative set of
treatment and control groups, Reeves et al. base their analysis on the initial treatment assignment rather
than the actual treatment received.
When Reeves et al. (2017) examine blood pressure, smoking or hearing ability (which was used as a
control) as potential health outcomes, they do not detect any significant earnings effects. This contrasts
findings by Lenhart (2017), who used 10 waves of the BHPS to identify significant health improvements
after the legislation was implemented when using self-assessed health, health conditions and health care
usage as key health outcomes. Given that the UK health care service is mainly paid for by taxes, a
reduction in the usage of any kind of health services, including doctor visits, is considered a health
improvement by the authors. The reduction in health conditions is driven by a decrease in those that
can potentially be treated by over-the-counter medicine (e.g. body pain, skin problems and allergies or
hearing and eye sight difficulties) and not by changes in relation to chronic conditions, such as epilepsy
or asthma. These findings indicate that self-treated medication is a normal good and, if the budget
constraint is relaxed, consumption behaviour may change, explaining the observed health
improvements.
While all three papers identify a reduction in low-wage workers’ financial strains on health, Lenhart
finds that there are a combination of channels explaining the relationship between wages and health,
including health-related behaviour (e.g. reduced smoking) and leisure expenditure (e.g. increased
holidays and sports club memberships). Like many natural experiments in this subject area, these papers
are limited by small sample sizes, attrition and the potential for an anticipation effect, selection bias
between the treatment and the control group, and the spillover effect to wages from individuals just
19 Previous work on the implementation of the national minimum wage in the UK has shown no significant
employment effects (Connolly and Gregory, 2002; Dickens and Manning, 2004; Stewart, 2004) and no effects on
hours worked (Connolly and Gregory, 2002).
34
above the minimum wage threshold. Furthermore, they only examine short-run effects and they are
likely to capture some counteracting effects to mental health improvement based on the stigma often
associated with minimum wages.
2.2.6. Lifestyle, access and health conditions
There are many aspects of individuals’ lifestyles that are identified as having an impact on human
capital outcomes. This section will review the effects of sleep, nutrition, exercise and stress as factors
which enable the accumulation (or, indeed, lead to a loss due to poor health, for example) of human
capital. The mechanism in which some of these aspects of an individual’s lifestyle affect human capital,
however, is often through effects on self-reported or objective health that then enables human capital to
be accumulated or lost.
One widely identified trend in the literature is that, across the lifetime of an individual, there are age-
related changes in sleeping patterns (Scullin and Bliwise, 2015). While these findings are often disputed
(Ohayon et al, 2004), the effect of sleep quality cannot be overlooked despite natural changes, or the
lack thereof, as an individual gets older. By looking at indicators of sleep quality, assessed using the
Pittsburgh Sleep Quality Index (PSQI), Gadie et al (2017) find that better self-reported sleep is
associated with better health outcomes – especially so for mental health and moderately so for cognitive
and physical health. Furthermore, Eide and Showalter (2012) investigate the effect of sleep on student
achievement and find that there is an optimal number of hours of sleep that boost test scores which
varies by age. Too much sleep, they find, leads to a reduction in test scores and hence a less desirable
outcome in terms of improving an individual’s human capital. Alhola and Polo-Kantola (2007) find
that, across the sleep-deprivation literature, there is a consistent negative effect of both acute total and
chronic partial sleep deprivation on both attention and working memory. Within this review are also
documented effects of a lack of sleep on long term memory, visuomotor performance and reasoning
ability. Drummond et al (2000) find that, with 35 hours of sleep deprivation, free recall of verbal
learning in normal young volunteers is impaired compared to the rested state. Furthermore, there are
wider labour market related outcomes affected by the level of sleep an individual gets.
Continuing with sleeping patterns, Hafner et al (2017) find that, across five different OECD countries
(Canada, Germany, Japan, the UK and the US), workers who sleep less than six hours per day report
on average 2.4 percentage point higher productivity loss due to the implication of absenteeism or
presenteeism – this is compared to workers sleeping between seven to nine hours per day. They also
estimate that an individual who gets less than six hours sleep loses around 6 working days due to
absenteeism/presenteeism per year more than an individual sleeping seven to nine hours, while a person
sleeping six to seven hours loses on average 3.7 working days more per year. Hafner et al (2017) use
the data generated during the ‘Britain’s Healthiest Workplace’ competition for the 2015 and 2016
35
combined waves. The sleep variables is derived from the question in the survey on how many hours of
sleep the individual gets on average in a 24 hour period which is then grouped into the following four
categories: < 6 hours, 6 to 7 hours, 7 to 9 hours, ≥ 9 hours. In order to quantify the economic costs of
different sleep amounts, the authors use absenteeism and presenteeism as a proxy for productivity by
using the Work Productivity and Activity Impairment Questionnaire (General Health) which is
incorporated into the BHW survey. Two empirical methods were used to measure the effect of sleep on
the ‘% work impairment (WPAI-GH scale) absenteeism/presenteeism’: OLS and a fractional response
estimator – of the two, the parameter estimates are preferred over the fractional response estimator.
The effect of nutrition is another aspect of individuals lifestyles that can affect their outcomes. Many
studies tend to focus on the investigation of malnutrition and its subsequent negative effects on health
and other life outcomes. However, Cade et al (2004) use the UK Women’s Cohort Study to compare
various types of diet in the UK. They use a food-frequency questionnaire to look at the difference
between meat eaters, fish-eaters and vegetarians and find that oily fish-eaters have the highest total
energy intake with vegetarians having the lowest which also resulted in higher nutrient intakes for oily
fish eaters than the other groups. Vegetarians also have the lowest absolute protein, fat and saturated
fat intakes. Diet is important to look at in this way and aids the literature which looks at the different
effects of diet on other health outcomes – as diet is found to significantly affect the likelihood of
developing certain diseases (see Ascherio et al, 1996; Key et al, 1996 for example). Appleby et al (2016)
look at the long-term health outcomes of vegetarians and vegans, and use a type of systematic review
of the literature to investigate the likelihood of having obesity, diabetes, cardiovascular disease, cancer
and other such diseases. Meredith and Dwyer (1991) look specifically at adolescent health and the
important role both nutrition and exercise must have for young adults to improve their health. This is a
review of the nutritional needs of adolescents as they age and the risk different demographics face to
developing eating problems such as anaemia, obesity and eating disorders while also accounting for the
effects of puberty on a change in nutritional and exercise requirements.
Exercise is another of the biggest positive influences over an individual’s health. By exercising the
recommended amount, the risk of developing long-term conditions such as heart disease, type 2
diabetes, stroke and some cancers (NHS, 2018) is reduced significantly. The National Health Service
website states that, across the lifetime, different lengths of daily exercise are required to maintain good
health. However, the effect of exercise has focused mainly on the benefit to an individual’s health, while
wider labour market outcomes, among others, are not well established. Lechner and Downward (2013)
undertook a matching analysis to examine the effects of sports participation on labour market outcomes
in England. Three surveys were used: Active People Survey (England-based survey of sports
participation by people aged 16 and over), Annual Population Survey (socio-economic information
aggregated to local authority level to be matched to Active People Survey), and the Active Places
36
Survey (reports on the actual supply of sports facilities at local authority level). Using these data, the
authors apply non-parametric econometric matching methods to find the relationship between
participation in sports and people’s labour market outcomes. They also used probit analysis to
investigate how participation varies with exogenous factors not influenced by current sports
participation. The study finds that sports participation has positive associations with individual income
of working age males and females and that individuals are more likely to be in work aged 26 to 45.
Further analysis suggest that these differences varies by age between gender (it is found that, compared
to males, females who switch from racquet and outdoor sports to team sports experience higher
employability when they are older). Barron et al (2000) look at the links between athletic participation
in high school and the acquisition of human capital. The US study finds that men are no more likely to
be employed if they participated in high school athletics. However, the wage for males who participated
in sports in high school is 12% higher in the National Longitudinal Study of the High School Class of
1971 (NLS-72) and 32% higher in the NLSY. Lechner and Sari (2015) use Canadian panel data which
follows a population of age 20 to 44 in 1994 through to 2008. It is based on the three classifications of
activity (inactive, moderate, active) which were defined in 1996 and look at the earnings and
employment consequences of moves between these groupings. Their empirical analysis suggests that
increasing activity from moderate to active positively affects earnings (increases in wages of 10-20%
after 8-12 years) however the effect of moving from inactivity to moderate sports activity has no
detectable effect. The effects on employment, however, suggest that moving from inactive to moderate
and also active participation down to moderate increases employment, while the effect on hours worked
is negligible. Lechner and Sari argue that one of the reasons for the surprising finding that moving from
active down to moderate participation may be “a further increase of sports and exercise activities beyond
a moderate level increases the value of leisure time” therefore this individual is more likely to substitute
work for leisure which increases their sports participation. The authors do recognise that this is a
speculative justification for this finding and cite that further research is required to back up this finding.
Exercise is also known to have effects on cognitive ability.
Stress is another factor which contributes to poorer health and increase the risk of developing severe
long-term issues. Stress in the workplace (organizational stress) also has implications on an individual’s
performance at work as it is seen to hinder day-to-day tasks. Sullivan and Bhagat (1992) describe four
main hypotheses surrounding the effect of stress on performance; the inverted U-shape, a positive linear
relationship, the negative linear relationship and then there being no relationship between stress and
performance. The first of these argues that a moderate amount of stress is necessary to facilitate
performance at work as it motivates an individual to get more work done. Below this level of stress,
there is little motivation and too much stress diverts attention away from work and towards coping with
stress. Their findings support a negative relationship between the amount of stress an individual feels
and their job performance. However, they also note that there is some inconsistency across studies
37
where the measures of stress were different (for example, the difference of functional vs dysfunctional
stress, plus looking at short-term effects on performance against long-term effects). Some studies focus
on occupation-specific stress – such as nurses, child welfare workers, physicians and call centre
employees – and discuss the effects of organizational stress on their work specifically. Travis et al
(2015) look at role conflict and role ambiguity as work stressors for social welfare workers, where role
conflict refers to the conflicting demands employees deal with and role ambiguity is where the
expectations of others are unclear (Kahn et al, 1964). The study found that work stressors including
work-family conflict and role conflict were positive and significant predictors of emotional exhaustion
which is consistent with literature surrounding job-burnout (see Allen et al, 2000; Maslach, 2003 and
Peeters et al, 2005). Work-family conflict is based on the work of Kahn et al (1964) and relates
specifically to the study of organizational stress as a stressor whereby it is “a form of interrole conflict
in which the role pressures from the work and family domains are mutually incompatible in some
respect” which Greenhaus and Beutell (1985) categorise into: Time-based conflict, Strain-based
conflict, and Behaviour-based conflict. Emotional exhaustion and job-burnout have negative effects on
an individual’s job performance. (see article “The relationship of emotional exhaustion to work
attitudes, job performance and organizational citizenship behaviours” –need to properly reference)
As mentioned previously in this review, Lenhart (2019) uses the BHPS to analyse the relationship
between the labour market and health outcomes. Through this, the author looks at changes in the
frequency of healthcare usage which affects the labour market outcomes of individuals through taking
time away from work and work-related activities. Part of the focus of the study was to determine
potential mechanisms through which health shocks negatively affect labour market outcomes. The
author finds that increasing usage of health care can contribute to the “persistent negative effects of
health shocks on labour market outcomes. One factor limiting the ease of access to healthcare services
is the disparity between rural and urban areas in most countries – including more developed countries
such as Canada, Australia and the US. Sibley and Wiener (2011) state that the conclusions as to whether
rural populations have limited access to healthcare compared to urban areas have been “contradictory
and inconclusive”. They state that the inconsistency in results is dependent upon which measure of
access to healthcare is used and whether geographic characteristics – which go beyond whether the
individual is in a rural or urban area – are controlled for. Their econometric method is the calculation
of unadjusted odds ratios to compare access measures across the rural-urban line.
‘Distance decay’ is one central theory which investigates the relationship between the distance to
healthcare services and the likelihood of use. A report by Imison et al (2014) for the King’s fund stated
that the distance decay effect is where “distance from hospital services reduces patients’ utilisation of
them (services are taken less often or later)”, however they state that the exact relationship between
distance from health services and the outcome on health is yet to be conclusively corroborated across
38
academic papers. A systematic review by Kelly et al (2016), however, find 77% of the 108 studies they
reviewed found a distance decay relationship for healthcare where those living in rural areas had
reduced healthcare usage as a result of being further away from services.
Many papers cite the Behavioural Model of Health Services Use (see Andersen, 1995; Babitsch et al,
2012; Litaker et al, 2005) which was modelled to incorporate both individual and contextual factors
which determine individual’s use of healthcare. The model identifies ‘predisposing’ (including
demographic information about an individual such as their age, sex, ethnicity etc.), ‘enabling’ (such as
financial situation, education level and employment status) and ‘need’ factors (e.g. self-reported health,
diagnosed medical conditions), all of which have sub-themes and include a multitude of different
indicators – hence studies that use this model have observed different outcomes depending on the
research question. Babitsch et al (2012) state that, alongside income factors which enable healthcare
usage, transportation, travel times to and waiting times for healthcare services all affect healthcare use.
There is also some evidence for the effects of specific health conditions and their effect on labour market
outcomes. Seuring et al (2015) conduct a cross-country systematic review of the economic costs of type
2 diabetes and find that, across all but 2 studies, the employment probabilities of men with type 2
diabetes were more negatively affected than those of women (however the extent of interpretation of
this relationship is dependent on the statistical methods used in the studies to account for reverse
causality or other unobservable factors). Holmes et al (2003) use the TARDIS study which is UK-based
and captures the full costs of care for a sample of people with type 2 diabetes – collecting data both on
patients and carers. There is a question on the questionnaire which asks “Are you not working, or
working only part time, because of your diabetes?” which was used in the empirical analysis for those
aged <65 years where the patient or the carer had responded that they were not working or only working
part time because of their diabetes (or due to caring for someone with diabetes). They find, looking at
the mean earnings loss, that patients lose £869 of their earnings whereas carers lose £1,300. For context,
using the Annual Survey of Hours and Earnings data for 1998, the total loss of earnings for patients was
5.1% of average earnings, and for carers, it is 7.6%.20 There are many existing cost-of-illness reports
which look at the burden of specific health conditions such as back pain, inflammatory bowel disease
and bipolar disorder (see Bassi et al, 2004; Maniadakis and Gray, 2000; Gupta and Guest, 2002).
Maniadakis and Gray (2000) extrapolated data from the Census and Labour Force Survey for Great
Britain to the UK using the population ratio published by the ONS. The authors estimate that in 1998,
£9090 million (1.9% of total compensation received by employees)21 were “lost due to incapacity to
20 Calculations were made using the Annual Survey of Hours and Earnings (ASHE) data for 1998, from the Office for National Statistics. Gross weekly earnings for all employees were multiplied by 52 to obtain the annual earnings estimate. 21 Calculations were made using the Office for National Statistics compensation of employee’s series from the GDP Income tables.
39
work attributed to back pain” - £6538 million (1.4% of total compensation received by employees) 20of
which was due to males and £2552 million (0.5% of total compensation received by employees) 20 for
females. The number of days out of work attributed to back pain is the main empirical focus of the study
– which is a focus on the absenteeism as a result of a specific health condition. Begley and Beghi (2002)
review existing literature of cost-of-illness reports on the economic cost of epilepsy and cite the indirect
costs of epilepsy include the cost of unemployment (attributed to the presence of epilepsy in the
individual). This is estimated as the lost earnings and/or the imputed value of lost household work
associated with morbidity and mortality referred to as the human capital approach. The human capital
approach is also reviewed by Luppa et al (2007) who define this to measure “the potential loss in
production for a society as the consequence of an illness, namely in terms of lost earnings”. This is
another literature review but of the cost-of-illness of depression – which looks to examine both studies
finding direct cost effects and indirect cost effects and finds that the indirect morbidity costs per case
were between $3656.34 and $5360.99 per year were reported.
Jones et al (2016) study the outcomes of acute health shocks on labour market outcomes and find some
evidence of causal impacts of the incidence of acute health shocks affect labour supply decisions. They
define acute health shocks as the onset of a cancer or stroke or myocardial infarction (heart attack) and
that labour market participation reduces significantly after a health shock where the average labour
market exit risk doubles afterwards. However, for those who remain active following a deterioration in
health, there is no significant effect on hours and earnings detected. The study uses Understanding
Society data from the UK Household Longitudinal Study (UKHLS) to identify the short run labour
supply response to a health shock which occurred between t-1 and t. The sample size is reduced to 428
looking at those who experienced an acute health shock in the UKHLS which is small, but they state
this is consistent with the literature. A matching algorithm is implemented using coarsened exact
matching (CEM) and propensity score matching following a method set out by Ho et al (2007) to
estimate the average treatment effect on the treated (ATT). To estimate the ATT, parametric models
(probit or OLS depending on the distribution of the outcome) are used with the weights obtained from
the CEM which contrasts purely nonparametric comparison of weighted means and allows the authors
to condition further on observable and time-invariant unobservable confounders, proxied by lagged
outcomes.
Much research has been carried out to show how air and noise pollution effect a person’s ability to
develop human capital. Suglia et al. (2007) use longitudinal data on children in Boston, Massachusetts,
USA, to assess whether higher levels of black carbon (which is common near busy roads) predict
decreased cognitive function (assessed through verbal and nonverbal intelligence and memory). To
estimate black carbon levels, an existing spatiotemporal regression model is used to predict traffic
exposure throughout a day. Cognition is assessed through two assessments: Kaufman Brief Intelligence
40
Test (K-BIT) and the Wide Range Assessment of Memory and Learning (WRAML). The effects of
predicted black carbon on cognition is assessed via the implementation of a linear regression, adjusting
for child’s age at cognitive assessment, gender, race/ethnicity, exposure to in-utero/postnatal second-
hand smoke, birth weight, blood lead level and maternal education. Results show black carbon to be a
significant predictor of cognitive function. Clark et al. (2005) investigate the association between noise
pollution and measures of cognition. The 2001–2003 Road Traffic and Aircraft Noise Exposure and
Children’s Cognition and Health (RANCH) project is analysed using multilevel modelling. It is found
that aircraft noise exposure at home and at school are highly correlated with impaired reading
comprehension, however road traffic noise exposure is not. Clark et al. (2012) however hypothesize
that air pollution explained the association between noise exposure and child’s cognition. Using the
same RANCH data, the authors find that adjustment for air pollution has little influence on the
association previously found between aircraft noise exposure at school and children’s cognition.
2.3. Job mobility, skills, and training
In this section we review a number of recent papers that aim to shed light on the implications that the
increase in job mobility may have on the re-training needs of workers (and different types of workers),
and the extent to which these costs will be borne by the firms. This issue is strongly related to the
literature that debates on whether human capital is more firm or industry specific, with some recent
contributions suggesting that it may be increasingly linked to specific occupations or tasks; which may
mean an increase in the degree of transferability of skills across firms and even industries.
Understanding the effects of a rapid introduction of Artificial Intelligence and robotisation on the labour
market, and on the economy as a whole, is emerging as an issue of key interest. Undoubtedly, the
resulting changes in production processes could be profound and likely to have important consequences
for the organisation of work. One strand of the literature investigates the degree as to which robots are
likely to be substituting for labour (Acemoglu and Restrepo, 2017; Arntz et al., 2017), while others
explore the rise of new job roles and occupations and attempt to understand what types of skills are
likely to be more complementary to new forms of capital. The displacement of labour in automated
tasks could be counteracted by an increase in the productivity in non-automated tasks which may
enhance the demand for new tasks and counterbalance the negative effects of automation on
employment. In this respect, the adjustment will depend, largely, on the appropriateness of the match
between the skills of the workforce and those required by new technologies (Acemoglu and Restrepo,
2018).
We highlight issues surrounding job mobility and occupational change in particular for the UK, but we
find that the evidence is scant compared to other countries that can make use of richer surveys. All of
41
the UK evidence we have reviewed mainly uses the BHPS or the LFS. In addition, it is worth noting
that there is to date little evidence specific for the UK on the relationship between automation and
robotisation and the displacement of workers while Frey and Osborne (2017) find that in the US 47 per
cent of workers are at risk due to automation.
2.3.1. Firm, industry and occupation-specific human capital
Initially the literature emphasised the importance of firm-specific capital acquired by workers; Becker’s
(1964) initial discussion focused on the dichotomy between firm-specific and general capital. This
distinction was crucial because while workers could have strong incentives to invest optimally in
general human capital, there could be potential incentive problems in financing firm-specific human
capital. Subsequent research suggested that an important component of human capital could be industry-
specific and can be lost only when the worker switches to a different industry (Neal, 1995; Parent,
2000). A more recent strand of literature concludes that a major component of human capital is
occupation-specific (Kambourov and Manovskii, 2009; Sullivan, 2010). In particular, Kambourov and
Manovskii (2009) argue that the evidence supporting the prevalence of industry-specific capital is
misleading as they find that tenure in an industry has a small effect on wages once the effect of
occupational experience is accounted for. Lazear (2009) proposes a theoretical skill-weight model
where most capital is general human capital and uses are specific to the firm. The assumption is that
there are a variety of skills used on each job, and each of these skills is general in the sense that it is
used at other firms as well. The difference, however, is that firms vary in their weighting of the different
set of skills. Therefore, this theoretical model predicts that most forms of skills will not be truly “firm
specific”, and there should be less loss of human capital following an involuntary job loss, especially
when industry and occupation effects are taken into account.
The process of changing occupation (not just jobs) however should not necessarily be associated with
loss of human capital (which can be the case mostly when the job change is involuntary) as some
occupational change can also be the result of natural career progression, for example as workers achieve
promotions that result in wage gains. Moreover, mobility across firms and occupations can be an
important adjustment mechanism in a dynamic labour market; for instance, Autor et al. (2003)
introduces a tasked-based approach to model job skill demand and point out that there is a decline in
the demand for routine intensive occupations, to which workers can adjust through occupational
mobility.
2.3.2. Work characteristics
42
There has been a growing area of research around structured management practices and firm
performance – including productivity (Bloom et al. 2013, ONS 2017, Bender et al. 2016, Broszeit et al.
2016). Using the survey of management practices in the US, conducted in over 30,000 plants, Bloom
et al (2013) found that structured management practices for performance monitoring, targets and
incentives are linked to better performance of firms, with firms adopting these practices displaying
greater productivity, profitability, innovation, and growth. Furthermore, firms with a higher share of
managers and employees with a degree, tend to have a higher management score. In addition, Bloom et
al (2017) found casual impacts of education on management practices. They find large significant
effects on management practices of being near a land-grant college with a range of controls for other
local variations in population density, income and other county-level and firm-level controls.
Within the work environment, it is important that workers are engaged - defined as a positive,
fulfilling, work-related state of mind that is characterized by vigor, dedication, and absorption
(Schaufeli et al., 2002). Using structured qualitative interviews with a group of Dutch employees from
different occupations, Schaufeli et al. (2002) found that engaged employees have high energy and self-
efficacy, while Schaufeli et al (2006) found that engaged employees often experience positive emotion,
which may explain why they are more productive. In addition, an alignment of values can also have an
impact on worker engagement.
One approach taken by employers to increase the development of their workers is the
implementation of a mentoring system. Through this, mentees are able gain and develop new
knowledge, skills and abilities, as well as recognition within the organisation from more senior workers
(Tong & Kram, 2013). However, the benefits of mentoring are not limited to mentees, with mentors
having increased opportunity to demonstrate leadership as they guide and advise their mentees (Dawson
et al, 2015).
A study by Orpen (1995) looked at 97 British employees in their first job. During the 6th month of their
employment, each participant had assessed the extent of the mentoring they received, measured by a
15-item scale. Data on the number of promotions and salary growth Four years later were used as
measures of career success. In addition, there was a significant relationship between career coaching
mentoring and positive outcomes. Those who experienced more career-related support from their
mentors had a higher organizational reward. Further work done by Orpen (!997) looked at the effects
of a formal mentoring programme on the work motivation, organizational commitment and job
performance of mentees. The performance of each mentee was given by ratings from their managers,
with results showing that formal mentoring can improve employee attitudes without necessarily raising
their performance, at least in the short term.
The move towards flexible working arrangements has been identified as a way for workers to
balance out work responsibilities and other commitments (Evans 2001; Glass and Estes, 1997: Dex and
43
Smith, 2002). Using survey data of 398 health professionals who had children aged 16 years or younger
at home, supportive practices at work, such as flexible working, decreased work/family conflict,
decreased depression, fewer somatic complaints, and lower blood cholesterol (Thomas & Ganster
1995). In addition, supportive supervisor behaviour of non-work demands also showed consistent
positive effects on job satisfaction and health outcomes for those responding to the survey. Using the
Workplace Employee Relations Survey (WERS), Dex and Smith (2001) examined the effects on firm’s
performance associated with firms giving their employees an entitlement to any one of 10 family-
friendly or flexible working arrangements in 1998. Using an OLS, logistic regression or ordered Probit
to estimate models of labour productivity, firm performance, among others, they find that the provision
of family-friendly policies relating to child care and working from home were associated with greater
employee commitment, captured by employees’ feelings of loyalty or shared values with the
organisation. Further to this, it has been noted in the literature that once becoming a mother, women
might choose to offset higher earnings for more desired working conditions (Felfe, 2012). Using data
from the 1979 National Longitudinal Survey of Youth, and an endogenous switching wage equation
model to account for the self-selection, Amuedo-Dorantes and Kimmel (2008) assessed the role of
employment-based health insurance offers in explaining the motherhood wage gap. The authors find a
negative compensating wage differential arising for the relative job preference of mothers from an
important component of non-wage compensation, namely, health insurance coverage.
Kidd et al. (2003) examines the outcomes on employee capability that resulted from
the career discussions of 104 employers. The results show that employees benefit from discussions
about their careers, with multiple outcomes being experienced, from short-term outcomes on improved
self-insight and feeling more confident and reassured, to longer-term outcomes on job moves, or
increase opportunities to develop new skills. It is important to recognise the role that line managers can
play in creating an environment that empowers self-development.
2.3.3. Estimating the costs of mobility
Cortes and Gallipoli (2017) estimate, using US data, that the transition costs across occupations can be
substantial, which complements other evidence regarding transition costs across industries (Artuç et al.,
2010; Dix-Carneiro, 2014). They draw from a model of occupational choice with a gravity equation
linking worker flows to occupation characteristics, and to transition costs. In evaluating the cost of
switching occupations Cortes and Gallipoli stress the importance of ‘task distance’, that is, the degree
of dissimilarity in the mix of requirements needed to perform a job. The costs of occupational mobility
should be increasing in task distance, and a considerable share of human capital would be lost when a
worker experiences a large change in the set of tasks she/he performed. Cortes and Gallipoli’s model
44
also allows for occupation-specific entry costs, which are independent of task content and may vary
over time; these entry costs capture institutional barriers faced by potential entrants to an occupation,
such as qualification credentials, professional training and union membership requirements. They
estimate the gravity equation using a range of proxies for occupational mobility costs and draw from
data on monthly worker flows across two-digit occupations from the Current Population Survey for the
period 1994 to 2012. This study exploits the fact that individuals' occupations are observed over
consecutive months, as data on job-to-job transitions are available at a monthly frequency. They
characterise occupations according to a number of variables regarding a number of dimensions of
complexity of work, general education development, specific vocational preparation requirements,
aptitudes, temperaments and physical demands, among others.
Cortes and Gallipoli find that task distance is a significant component of the cost of switching
occupations, suggesting an important role for task-specific human capital. In particular, an increase of
one standard deviation in task distance is estimated to increase the cost of switching occupations by
nearly 20 per cent. They consider fixed cost of switching across occupations that belong to different
major task groups (non-routine cognitive, routine cognitive, routine manual or non-routine manual), as
there may be costs associated with these switches that exceed what is captured by the distance measure.
They find that if the switch is to a different major task group, the increases are larger, that is, depending
on the type of transition. For example, the additional costs are estimated to be between 14 percentage
points (for transitions into ‘routine cognitive’ occupations22) and 58 percentage points (for transitions
into ‘routine manual’ jobs23). The results are robust when the analysis is restricted to younger workers
for whom occupational mobility rates tend to be higher. Despite the considerable role of the ‘task
content’ variable, they find that the largest part of occupation mobility costs can be attributable to task-
independent entry costs into the occupations.
Gathman and Schonberg (2010) also study the transferability of skills accumulated in the labour market,
building on the concept of ‘task-specific human capital’; this is in line with previous literature that
consider that occupations are characterised by a vector of skill or task characteristics (Autor et al., 2003).
The underlying notion is that if human capital is task-specific, it should be partly transferable across
occupation pairs, in which a similar mix of tasks is performed. To measure the transferability of skills
22 Routine cognitive occupations include: Sales supervisors and sales reps, finance and business; retail and other
salespersons, office supervisors and computer operators, secretaries, stenographers, and typists, information and
records processing, except financial; financial records processing occupations, office machine operators and mail
distributing; other administrative support occupations, including clerical. 23 Routine manual occupations include: mechanics and repairers; construction trades; other precision production
occupations; machine operators and tenders, not precision; fabricators, assemblers and hand working occupations;
production inspectors and graders; transportation and material moving; helpers, construction and production
occupations; freight, stock and material handlers.
45
empirically, they combine a high- quality panel on complete job histories and wages, with information
on tasks performed in those occupations. They use data on job histories and wages from a 2 per cent
sample of all social security records in Germany, providing a complete picture of job mobility and
wages for more than 100,000 workers during the period 1975 to 2001. These data have several distinct
advantages over data used in the previous literature on occupational mobility. First of all, and compared
to household data, the use of administrative data allows for a more precise record of the dates of job
changes; second, there is more guarantee that occupational titles are consistent across firms as they form
the basis for wage bargaining between unions and employers; and third, there should be less
measurement error in earnings and occupational titles compared to typical surveys, as misreporting is
subject to severe penalties. The authors then use data on detailed information on tasks performed in
occupations (from the repeated cross-section German Qualification and Career Survey), to identify how
similar occupations are in their skill requirements. In this survey, individuals are asked whether they
perform any of 19 different tasks in their job, and they categorise these 19 tasks into three aggregate
groups: ‘analytical tasks’, ‘manual tasks’, and ‘interactive tasks’. This study introduces a continuous
measure of how occupations are related to each other in terms of their skill requirements, and concludes,
after exploring wage changes, that specific human capital should not be fully lost after an individual
leaves an occupation. They conclude that human capital is more portable across occupations,
particularly for the high-skilled, than previously considered. Poletaev and Robinson (2008) find a
similar result for the US, building also on the idea that jobs can be characterised by their skill vector
rather than their narrow industry or occupation codes. Drawing from evidence from the US Displaced
Workers Survey (for the period 1984 to 2000), they find that displaced workers, which in a large
majority were reallocated across standard occupations and industry codes (after in this case plant
closures), tend to experience wage losses to some degree. They observe that those workers with the
largest losses were also those that experienced more significant switch in their skill portfolio. If human
capital is specific and can be narrowed down to a small number of basic skills, then a substantial
deviation in the skill portfolio mix between pre- and post-displacement jobs would result in a loss of
specific capital, and the patterns of wage losses are consistent with this idea. Conversely, many switches
of industry or occupation codes are not significant skill portfolio switches, and in these cases the wage
losses are relatively lower.
A number of OECD papers focus on how education and training policies can facilitate transitions across
occupations, while maintaining workers in quality jobs that can maximise the use of their skill set
(Bechichi et al., 2018; Andrieu et al., 2018). Bechichi et al. (2018) provide evidence on the quantity
and type of training needed to facilitate certain job-to-job transitions and investigate what influences
skills distances across occupations have. They use PIAAC data on 31 countries and their results show
that training needs are not equal across all occupations and level of skills. Skills distances in terms of
46
cognitive skills are higher among groups of low-skilled occupations or from mid-skilled to high-skilled
occupations than among higher skilled occupations. The analysis of occupational moves within high-
skilled occupations, for instance moving from professionals to managers or within the group of
professionals, conversely shows that skill distances are larger when it comes to task-based skills and
relatively smaller for cognitive skills.
Building on these studies, Andrieu et al. (2018) propose an experimental methodology to estimate
empirically the monetary cost of training needed to move workers across occupations for a sample of
OECD countries. They estimate re-training costs of both direct and indirect nature. Indeed, the direct
cost of re-training can be proxied with two main elements: a) a country’s yearly expenditure in
education per pupil and b) the average re-training time between the occupation of origin and all the
‘acceptable’ occupations of destination. Occupations are regarded ‘acceptable’ when they are relatively
close in terms of skills requirements to those of the occupation of origin, and any wage cut is of
maximum 10 per cent. The direct cost of undertaking the training is however only one of the
components of the total costs, and this methodology also considers the workers’ opportunity costs in
terms of the foregone wages throughout the duration of the training spell. The indirect costs of training
are defined here as the wages and bonuses a worker sacrifices while in training, which corresponds to
the opportunity cost of retraining from a worker’s point of view. This study estimates that the average
direct (re)training costs per worker is estimated to be around $1,600 for small training needs (up to 6
months); it is double than that ($3,200) in a moderate training scenario (up to 1 year), and rises sharply
in a more intensive training scenario (up to 3 years) to $12,600. The average per-person opportunity
cost (i.e. the indirect cost) is about $4,300 in the case of the first scenario, $9,900 in the moderate one,
and rises sharply to $37,000 in the case of the intensive training scenario. They estimate the opportunity
cost of training represents approximately 75 per cent of the per-person total cost. The total cost (that is,
the sum of the direct and indirect costs) is generally larger for highly skilled than for low-skilled
occupations, given the larger opportunity costs.
This study also finds that a workers’ average age is positively correlated with the cost of moving the
worker to a different occupation, which highlights the challenges of occupational mobility for older
workers. They also find, similarly, occupations with higher proportions of young (older) workers
display significantly lower (higher) total retraining cost per worker on average. These results are derived
using country-occupation data, and conclude that an increase of one year in the average age of workers
(in a given occupation and cluster of countries24) would be associated with an increase of approximately
11 per cent in the average cost of re-training for each worker. Similarly, an increase in the proportion
of young workers by one percentage point would be associated with a decrease in average cost of
24 Cluster countries are based on relative similarities of type of tasks performed on jobs.
47
retraining each worker by approximately 3.7 per cent. The observation of an association between age
and per-worker cost of retraining and acquiring the cognitive skills need to move across occupations, is
however likely to be driven by differences in the age composition of occupations, and not by a direct
relationship between the average worker’s age and the cost of re-training in a given occupation. Once
one accounts for the fact that occupations with lower total cost of re-training have a higher proportion
of younger workers, the remaining association between age and cost becomes often statistically
insignificant. This is however not a main result of the paper and deserves further exploration.
2.3.4. Job mobility in the UK
Longhi and Brynin (2010) investigate the implications of changes in occupation using UK (BHPS)25
and German (German Socio-Economic Panel) panel data. This study finds that year on year, over 10
per cent of people change their occupation in Britain (they consider movements across two-digit
occupation codes), while this figure is 5 per cent in the case of Germany. The UK ranks relatively high
in terms of occupational mobility compared to other countries in Europe too, and recent evidence shows
that the level of occupational mobility in Britain is one of the highest only behind Estonia and Sweden
(Bachmann et al., 2017). These figures are still lower than those available for the US, with evidence of
an annual mobility of 15 per cent (Kambourov and Manovskii, 2009).
Longhi and Brynin (2010) analyse the factors associated with job and occupational moves using Cox
proportional hazard models, where they take into account the effect of time variant as well as time-
invariant covariates. They distinguish between two types of changes: job moves within and moves
between occupations. Longhi and Brynin find that those who change occupation tend to be relatively
poorly matched to their job compared to those who change job but remain in the same occupation. The
results suggest that factors specific to the individual, as indicated by (proxy) measures of ability and
motivation, represent a major impetus to change occupation, and that occupational changes might be
rational choices that relate to an initial poor occupational decision and to general aspirations about the
job and work-life balance. They conclude that despite differences in occupational turnover between
Britain and Germany, the factors driving change as well as the effects on the workers are not too distant
across the two countries. In Britain, however, occupational change is less likely at the top of the
occupational ladder, and compared to Germany, there is less evidence of an orderly career progression
channelled through change of occupation.
In terms of job satisfaction, German data is richer and allows us to understand the beneficial effects of
changing occupation, based on a number of subjective evaluations by workers. In the case of Britain, it
25 Nationally representative household survey providing rich information on individual demographic,
socioeconomic and work-related characteristics.
48
is only possible to know if job satisfaction has improved or worsened after an occupational change. The
results overall show both positive and negative effects of change, but overall slightly more marked
improvement in circumstances after an occupation change, relative to a job change within the same
occupation. In terms of skills, the results show that those who change occupation experience greater
decline in the use of their skills than do other job changers. The ‘use of skills’ is evaluated using a
measure of job match, which aims to capture whether the worker is overqualified. For Germany, this
analysis is more straightforward; it is derived from a question on “What training is required for your
job?”, which is then compared to actual qualifications. No equivalent exists in the BHPS, so for the UK,
they use the ‘average’ method, which implies computing the education typically needed to perform a
certain job, derived using the LFS. These are merged back into the BHPS per occupation and a dummy
indicating ‘overqualification’ is computed, by a simple comparison of qualifications ‘held’ and
‘needed’.
Panos et al. (2014) investigate the inter-related dynamics of dual job-holding, human capital,
occupational choice and mobility, using a panel sample of UK employees from the BHPS during the
period 1992 to 2004. Multiple job-holding is estimated to be an important determinant of job mobility
decisions, and individuals doing different occupations in their secondary employment are more likely
to change jobs in the following year. From a policy point of view there is interest in understanding
whether multiple job-holding can lead to the more efficient acquisition of skills, and to promote future
potential entrepreneurial initiatives. This UK study shows that individuals may be using multiple job-
holding as a vehicle for obtaining new skills and expertise and as a stepping stone to new careers, also
involving self-employment opportunities, but it depends on the level of education and/or income. The
study finds that individuals who enjoy a relative sense of financial security are found to be more likely
to explore different occupational paths in their secondary employment to satisfy their intrinsic
preferences. Instead, those individuals facing increased financial constraints are found to be more likely
to do the same occupation in both their primary and secondary job, thus exploiting the higher earnings
opportunities associated with their accumulated occupational experience.
Connolly and Gregory (2008) provide further evidence for the UK of the nature of occupational change
studying transitions of women between full-time and part-time work. Unsurprisingly, they show higher
occupational mobility for part-time workers than for full-time employees in the UK. Using information
from the New Earnings Survey Panel Dataset and the BHPS for the period 1991 to 2001, they show that
around one-quarter of women moving from full- to part-time work move to an occupation that requires
lower level of qualifications. Downgrading affects approximately 29 per cent of women from
professional and corporate management jobs, and up to 40 per cent in jobs for those with intermediate
skills. While some professional women switching to part-time work do move into a new career requiring
49
a higher level of skills, at least as many take up jobs in a range of low-skill occupations, underutilising
between three and five years of higher education and professional training.
Carrillo-Tudela et al. (2016) look at both occupational and sectoral mobility of workers in the UK labour
market during the period 1993 to 2012, drawing from the UK LFS. Given the relatively long timeframe
of this study, the authors are able to explore the incidence of career changes over time, in particular in
relation to the Great Recession. They estimate that workers who change employers have around a 50
per cent chance of switching to another occupation or industry. Career changes are more likely for (a)
those workers actively searching for a job, (b) those that made voluntary transitions (i.e. those who
‘resigned׳ from jobs, or gave up for ‘family or personal reasons’, as opposed to those that were made
‘redundant’ or ‘dismissed’) and (c) those workers that work part-time or as temps. Across occupations,
career changes that involve an upgrade in the skill level are more likely through voluntary employer-
to-employer transitions, while instead, career changes that involve a step down in skill level are more
likely after spells of non-employment.
Across industries, they find that low-skilled workers have a higher probability of a career change than
medium and high skilled workers. Across occupations, it is only the unemployed high and medium
skilled workers that have shown a higher propensity to change career. These authors find that broad-
based shortfall in economic activity has significantly prevented workers from pursuing alternate careers
thus sacrificing substantial wage gains. In terms of the effects of human capital on the probability of a
career change, they find that age decreases the probability of a career change, pointing to the importance
of on-the-job human capital accumulation.
Using comparative panel data of 14 European economies for the period 1994 to 2001 (European
Community Household Panel), Muffels and Luijkx (2008) find, in line with most of the literature, that
with age, occupational mobility slows down. However, they find a positive sign for the age-squared
variable in their regression, which indicates that beyond some age threshold both upward and downward
mobility increase again. This upward mobility at higher ages is more likely to occur just before
retirement age (at 50 years), whereas downward mobility is likely to occur somewhat earlier (at 48
years) either due to demotion on the job, due to moving to a lower level job with another employer, or
due to moving into self-employment. Muffels and Luijkx (2008) also conclude that education strongly
increases the chances for upward mobility and lowers the likelihood of downward mobility into lower-
level jobs. These results are consistent with human capital predictions.
2.4. Within-country migration
50
Regional education, mainly through education of workers, of entrepreneurs, and the realisation of
human capital externalities, emerges as a critical determinant of regional development (Gennaioli et al.,
2013). At the same time the literature stresses, that patterns of internal migration can significantly shape
the redistribution of human capital within countries and thus affect regional economies. In this section
we focus on the how within-country migration patterns can be closely related to the process of
knowledge generation and development of regions. Wherever possible we focus on evidence for the
UK and also review some relevant and current international studies. The analysis of the determinants
and consequences in relation to international migration flows is beyond the scope of this study.
Focusing on the UK, a number of papers focus on the relationship between migration flows and
innovation performance of a region. Faggian and McCann (2009) employ a simultaneous equation
model to relate human capital inflows to a region’s innovativeness as well as to other regional factors
(including the number of job vacancies, the level of regional unemployment, the quality of life, the
regional real wage, the geographic periphery of the region, the density of universities, and the regional
R&D expenditure, etc.). The main sources used in this study include labour market information on
student leavers for the year 2000 (from HESA) and measures of innovation based on number of patent
applications. They show that the innovativeness of a region is one of the major factors encouraging
university graduates to seek employment in that region, while at the same time, inflows of highly mobile
university graduates promote regional innovation, at least within England and Wales.
Using British data, Gagliardi (2015) also provide evidence of a causal link between an increase in the
average stock of human capital, which arises from the inflow of skilled migration inflows, and the
innovative performance of local areas. An OLS regression is estimated in differences where the
dependent variable is the innovation performance of a region measured as the share of innovative firms
in a travel to work area (TTWA) (that is, the difference between time t and time t-1). The main variable
of interest is the influx of migrants, proxied by the share of skilled population in TTWA; but the
regression includes other controls such as the change in the average ratio of skilled over unskilled labour
and other indicators of TTWA controls. The specification in differences allows for removing any time-
invariant unobserved heterogeneity across regions. The paper focuses on the role of human capital
externalities as a crucial determinant of local productivity and innovative performance. In this context,
skilled migration becomes a crucial channel of knowledge diffusion broadening the geographical scope
of human capital externalities. The estimation procedure adopted is constructed around a place-based
knowledge production function defined for TTWA level. The empirical analysis draws from the
Community Innovation Survey and the LFS. The main challenge in performing this estimation strategy
is related to the endogeneity of the regressors of interest. Firstly, migrant inflows and innovative
performance may be correlated due to common fixed influences, as immigrant population may be
51
concentrated in certain areas as a consequence of historic settlement patterns. This may explain a
positive correlation between skilled migration and innovative performance of a region, even when there
is no a genuine casual effect. Endogeneity in this case is minimised by estimating the relationship using
differences, that is, by relating changes in immigrant concentration between two points in time to
changes in the innovative performance of the areas of destination. Second, this type of estimation is
likely to be affected by reverse causality bias. Skilled migration inflows can be considered a
fundamental determinant of innovation since they act as channels of knowledge transfer, and reduce the
geographically localised nature of human capital externalities. To address this source of bias, this paper
adopts an IV approach, drawing from variables that are likely to be correlated with inflows, but not
otherwise associated with the dependent variable through unobserved local characteristics. For
example, the variation in the share of skilled migrants is instrumented by a shift-share instrument
constructed using LFS data on country of birth (this identification strategy combines local economic
compositions with shifts on the aggregate level to predict variation in a variable of interest). The results
of this paper show human capital externalities related to the migration behaviour of skilled individuals
are a significant determinant of innovation in British local areas; but there is not much evidence for the
existence of additional human capital externalities in cities related to the effects of urban agglomeration.
A recent strand of literature draws from micro-level administrative data to show that patterns of internal
migration play a decisive role is the spatial distribution of human capital. The study by Kooiman et al.
(2018) follows the birth cohort 1979 from age 16 until age 35, comparing spatial trajectories between
university graduates and their lower educated peers. This paper uses data from the System of Social
statistical Datasets of Statistics which covers the complete registered population of the Netherlands;
individuals are traced longitudinally and spatially from 1995 until 2014 and this data is enriched with
demographic and socio-economic information. They observe that the highest educated individuals are
more than others attracted to metropolitan core areas. They also find that employees in the largest cities
make more wage progression than their peers in smaller cities and villages because they increase the
number of hours worked and they are employed in industries characterised by above-average wage
progression. A question of importance highlighted in the literature relates to the reason why lower
educated workers do not migrate to the same degree as university graduates do.
The results in Kooiman et al. are consistent with those in Moretti (2012), who shows for the US that
lower educated workers tend to stay put in regions with relatively scant opportunities. This relative
immobility of the lower educated is argued to be attributed to a number of reasons. For example, they
have a lower ability to process and analyse information with regards to the availability of alternative
opportunities elsewhere and they have a stronger reliance on family and friends (Faggian et al., 2015).
A possibility is that for the lower educated these spatial bounded benefits are predominantly financial,
52
as they may enjoy better wages for the same job, while for university graduates metropolitan areas offer
more opportunities for career progression.
Comunian et al. (2016) discusses the contribution of different ‘types’ of human to local development.
They investigate whether creative human capital, that is, ‘the human capital specifically developed via
education and advanced training in creative and artistic subjects’, play the same role as more scientific-
oriented human capital (e.g. STEM) in fostering local development, as the recent literature on human
capital and regional economic development has become increasingly concerned with the role of the
‘creative occupations’. They explore the transition period from university education to employment of
creative graduates, with the aim of advancing the understanding of the relationship between creativity
and mobility of human capital. This study provides the first empirical analysis of the role played by
creative graduates’ subject background in influencing their migration choices in the UK. It expands
from other prior studies in the field by looking more specifically at the migration behaviour of sub-
groups of creative graduates. The main sources for this study are the Students in Higher Education and
the Destinations of Leaves from Higher Education surveys, both collected by HESA. The first one
contains data on all students enrolled in UK Higher Education Institutions, while the second one is a
survey undertaken every year that generally covers British domiciled students to collect information
about graduates’ employment activities six months after graduation. Since the main interest is
migration, the focus here on British-domiciled students (both part-time and full-time) for which full
location information (post code information for pre-university, university, and job location) is available.
The results confirm the role of London as a hub for talent, but also as magnet attracting creative students
from all over the UK and retaining almost three quarters of them. They also find, consistent with
previous literature, that creative graduates have lower salaries but they find that migration mitigates
some of their difficulties allowing them to find a better occupation more fitted to their skills. The fact
that ‘return migration’ is the most common choice of creative graduates suggest that networks and peer-
to-peer support play a key role after graduation. For music graduates in particular they find university
networks play a crucial role in helping them to eventually secure successful careers.
Huttunen et al. (2018) analyse the geographic mobility of workers after a permanent job loss, and
investigate factors that influence workers’ migration decision. They provide evidence on the importance
of non-economic factors in explaining migration decisions and income losses following job
displacement. They draw from rich Norwegian register data that includes information about the
workers’ characteristics, location and employment histories, as well as information about spouses,
children, parents and siblings. A key feature of the data is that it tracks individuals even after leaving
the labour force. Moreover, it allows the analysis of earnings and employment patterns several years
before the job loss, thus assessing the selection into mobility. Having detailed information on spouses,
53
the age of children and the location of parents and siblings help assessing the effect of family networks
on mobility decisions. Also, this study provides evidence of how earning losses after job displacements
are related to migration. This research assesses the post-displacement labour market experience of
movers and stayers, comparing displaced movers and stayers with a control group of non-displaced
workers. A key aim is to understand whether earnings differential between displaced movers and stayers
is a causal effect of moving and the selection issue in the decision to move. This is an important
consideration as while some literature finds that migrants tend to be favourably self-selected, it is also
possible that the most productive workers are best rewarded in their local labour market, and hence
have less need to move.
The results show that family ties are very important for workers’ mobility decisions. Workers who move
to a new region after a job loss suffer larger income losses than stayers, although the difference between
the movers and the stayers in total family income is smaller and tends to disappear over time. They find
that the difference in individual earnings between displaced movers and stayers is driven entirely by
workers who move to rural regions or to regions where they have close family networks.
2.5. Crime
The labour market outcome of an individual is strongly related to their human capital which, as
discussed earlier, is linked to family background, health and many other factors. In this section of the
literature review, the role of criminal activity is discussed. In particular, papers that examine the causal
effect of educational attainment on crime (and vice versa) are studied. Policymakers put a lot of
emphasis on enforcement and punishment to tackle crime, but a lot of empirical evidence highlights the
role of early intervention with regards to education in reducing criminal activity. Although different
features of crime incidence are worth studying, the most widely examined area (by economists) involves
education. This focus is not surprising given that a survey of newly sentenced prisoners in the UK
showed that in 2005/2006 47 per cent had no educational qualification, compared to 15 per cent for the
general population (Hopkins, 2012). Therefore, most of the reviewed studies in this section examine
the economics of education and crime, while the remaining section discusses the role of family life, and
how educational or vocational intervention programs in prison may affect human capital.
According to a review carried out by Draca and Machin (2015) the prominence of economics of crime
as a research field on its own has risen drastically over the last two decades. However, this review shows
that the majority of papers published, especially the most widely cited ones use US data, and more
recently Scandinavian administrative data (for example, Bennet, 2018). Only very little work has been
54
carried out using recent UK data, which is partly the result of limited availability for identification
strategies and partly because of the lack of administrative data.
2.5.1. Economics of education and crime
The existing socio-economic literature has identified some mechanisms through which education
impacts on crime: income, preferences, social interaction and time availability. Education is considered
a main driver in human capital accumulation which improves work opportunities and in turn
discourages criminal activities. Widely cited work by Lochner based on the US argues that human
capital increases the opportunity costs of crime from foregone work and expected costs with
incarceration (Lochner, 2004; Lochner and Moretti, 2004). In other words, investments in human
capital, through education for example, should reduce crime if human capital increases marginal returns
from working more than returns from participating in crime. This is in line with empirical evidence
which finds a negative relationship between wages and criminal activities (Machin and Meghir, 2004)
and a positive relationship between education and wages (see section on earnings determinants for a
detailed review). While this is plausible for most types of street crime, there is the argument that white
collar crime and education are positively correlated because illegal activities, such as fraud or
embezzlement, are likely to require skills obtained in school (Lochner, 2004).
Besides the wage-related mechanism, Hjalmarsson and Lochner (2012) bring forward the argument that
education can reduce crime through changes in risk-related preferences and social interaction. This is
based on the assumption that education can encourage risk-aversion and more educated people tend to
interact with like-minded people who are discouraged from engaging in crime. However, the latter
argument can be questionable since this is likely to depend on the quality of school provided. Although
there is no empirical evidence for this relationship, research on school choice and desegregation in the
US offer some support for the role of school quality in explaining the education-crime nexus (Guryan,
2004; Cullen et al., 2006; Weiner et al., 2009; Deming, 2011).
Another literature strand builds on the incapacitation effect and argues that education discourages crime
not only later in life, but also during school enrolment, since students are less likely to commit any
crime while being in a classroom (Tauchen et al., 1994; Anderson, 2014; Bell et al., 2016, 2018).
Furthermore, skills obtained as a result of school attendance may enhance labour market prospects and
discourage a life of crime by making juvenile arrests more costly. Early empirical evidence on the
incapacitation effect in the US is presented by Jacob and Lefgren (2003) who use exogenous teacher
training days, and Luallen (2006), who uses unexpected school closures as a result of teacher strikes,
as instruments for student absence from school. Both find incapacitation effects of education on
55
criminal activity, with higher crime rates during days off. Similar findings are observed by Anderson
(2014) who exploits variation across states in the minimum high school dropout ages to estimate the
effect on juvenile arrest rates between 1980 and 2008.
There is also an argument for the opposite direction of this relationship to be true since crime can also
affect educational attainment if arrest and incarceration incentivise school drop outs (Lochner, 2004,
2011; Sweeten, 2006; Hjalmarsson, 2008), the accumulation of criminal capital replaces the need to
invest in education (Lochner, 2004; Ward and Williams, 2015) or interactions with the criminal justice
system harm non-cognitive skills of adolescents which can reduce motivation or aspiration and hence
influence educational outcomes negatively (Behncke, 2012). However, while this literature review
revealed that evidence for the negative causal effect of education on crime is growing, it has also
highlighted the lack of empirical evidence for the opposite direction which is mainly due to the difficulty
of identifying exogenous variations in criminal involvement. One exception is a recently published
paper by Rud et al. (2018), which uses administrative data from the Netherlands to examine the effect
of early criminal activity on school dropouts and to analyse underlying factors of this relationship. They
use an OLS regression set-up with a large number of observable family and individual characteristics
as well as school, class, sibling and twin fixed effects and find that criminal activity is associated with
an 11 percentage point higher probability of school dropout with a stronger effect size if juveniles are
involved in severe criminal activities. However, the magnitude reduces when unobserved and observed
heterogeneity is controlled for and, therefore, they conclude that, if present, the treatment effect is
relatively small. A very similar strategy is employed by Webbink et al. (2013) who use Australian
cohort data on twins in a fixed effect estimation. They find that early arrests decrease the probability of
completing high school for fraternal twins and decrease educational attainment in general. By using a
male sample from the US NLSY 1997 in a multivariate mixed proportional hazard framework to control
for common unobserved confounders and address reverse causality, Ward et al. (2015) find that arrest
and delinquency both incentive school dropouts and that the effect of delinquency is driven by income
generating crimes.26 Aizer and Doyle (2015) use an IV approach and exploit exogenous variation in
juvenile incarceration as a result of the random assignment of juveniles to judges with different
tendencies to sentence youths to imprisonment. They use uniquely linked administrative data for over
35,000 juveniles over 10 years from a juvenile court in Chicago and find that juvenile incarceration
decreases the probability of graduating from high school.
26 They estimate in a baseline scenario with a reference male that was neither involved in delinquency nor was
arrested, that he is 35 per cent more likely to leave school at age 18 or earlier, which increases to 42 per cent if he
initiates delinquency at age 16 and is not arrested, and to 55 per cent if he has been delinquent at age 16 and has
been arrested at age 17. As a result, being delinquent at age 16 increases the probability of school leaving by age
18 by 20 per cent, while subsequently being arrested at age 17 increases it by a further 37 per cent.
56
Although evidence for a negative relationship between education and crime is already found in very
early studies, they are often based on standard regression models that only control for individual and
family characteristics (see, for instance, Ehrlich, 1975 and Witte, 1997). These results do not necessarily
imply causality because unobserved individual characteristics, such as the previously mentioned risk
aversion, are likely to impact on both schooling and the choice to commit a crime. More recent studies
address these issues by exploiting policy changes to schooling ages to create exogenous variation in
educational attainment. While our review has revealed that significant research progress has been made
in establishing a causal link between average educational attainment and arrest, conviction or
incarceration rates, most of the empirical evidence comes from studies carried out in the US (most
commonly known is the work by Lochner and Moretti, 2004 but most recent data up to 2015 has been
used by Bell et al., 2016, 2018) or Scandinavian countries (for example, Meghir et al., 2012; Brugård
and Falch, 2013; Hjalmarsson et al., 2015; Landersø et al., 2016). Another strand of literature adopts an
experimental research design based on US interventions, such as the 1960s High/Scope Perry Preschool
Program or the Moving to Opportunity experiment, which allocate people randomly into programs
aimed at improving skills. Cost benefit-analyses have shown long-run crime reductions through
improved male child development (Belfield et al., 2006; Heckman et al., 2013) and short-run reductions
in violent crime due to improved neighbourhood conditions (Kling et al., 2005; Sciandra et al., 2013).
The few exceptions of work carried out in the UK are studies conducted by Machin et al. (2011), Machin
et al. (2012) and Sabates and Feinstein (2008). Similar to most US based work, Machin et al. (2011)
exploit the increase in the minimum schooling age from 15 to 16 in England and Wales in the school
year of 1972/1973 to estimate the effect of education on property and violent crimes convictions
between 1972 and 1996. They used an IV and regression discontinuity setting to study cohort-level
changes for cohorts turning 15 just before and after the change in the law. They find that the schooling
reform increased educational attainment in the reduced-form and first-stage by generating a significant
causal estimate for the effect of education on crime, which was almost twice the size of the
corresponding OLS estimate. More precisely, Machin et al. (2011) find a decrease in conviction rates
for men committing property crimes by 20 to 30 per cent with every one-year increase in average
schooling levels. The effect on violent crime is also sizeable but statistically insignificant and education
affected neither of these crime types when committed by women. In the US, schooling effects were
smaller on property crime and larger on violent crime (Lochner and Moretti, 2004). Machin et al. (2012)
examined a similar relationship but adopt a different identification strategy. In this case they exploit the
increase in education levels across the whole education distribution, which was the result of a large
expansion of the UK post-compulsory education system that occurred in the late 1980s and early 1990s.
Their IV setting reveals that education reduces adolescent crime but also affects qualification attainment
and wages suggesting that additional time spent in school, i.e. the incapacitation effect, is not the sole
57
mechanism through which education impacts on crime participation. More precisely, they demonstrate
that a one per cent increase in the proportion of male (female) students reduces male (female) youth
crime by around 1.9 per cent (1.1 per cent) and a one per cent increase in the proportion of men (women)
staying on at school after the compulsory school leaving age reduces male (female) youth crime by
around 1.7 per cent (1.3 per cent). Similar to Machin et al.’s (2011) observation, these causal effects
are considerably larger than those from the OLS estimates. Sabates and Feinstein (2008) exploit two
contemporaneous events in a DiD approach to examine burglary activities in the UK. Between 1999
and 2002 the Educational Maintenance Allowance was piloted which provided a financial reward to
incentivise school attendance for low-income 16 to 18-year old students in 15 local areas with
particularly low attendance. At the same time, the Reducing Burglary Initiative was established to fund
a number of local burglary reduction schemes. They find that in areas where both schemes were
implemented, burglary rates decreased by about 5.5 per cent relative to matched comparison areas. The
effects were slightly lower when only considering the subsidy.
However, education not only has a direct effect on crime but also an indirect one through earnings,
since low income may incentivise crime by shifting the relative preferences associated with
participating in legitimate work and committing crimes. Machin and Meghir (2004) find empirical
support for this argument when examining the effect of wage changes at the bottom end of the wage
distribution of the retail trade sector on crime rates, using data on the police force areas of England and
Wales between 1975 and 1996. They find that the marginal effect of a 10 per cent increase in the wages
corresponds to a 0.7 percentage point fall in the crime rate (where the baseline crime rate is 8 per cent
or 80 crimes per 1000 people). This is robust to controlling for the conviction rate (additionally
instrumented by sentence lengths) and the inclusion of lagged dependent variables to account for
persistence in crime rates. Therefore, relative wage changes across regions can explain relative changes
in crime rates. Albeit the outdated data use in this research, these findings are consistent with other
studies which find that declining labour market prospects shape criminal careers because they are likely
to shift incentives towards criminal activities and away from labour market participation (Gould et al.,
2002 using US data; Fougere et al., 2009 using French data; Grönqvist, 2013 using Swedish data). Bell
et al. (2018), for instance, demonstrate that a recession (proxied by an unemployment rate of 5
percentage points higher than a normal unemployment rate) in the UK is associated with a 5.7 per cent
increase in the probability of ever being arrested (the effect is larger for individuals with fewer years of
schooling). This study draws from both panel data on year of birth cohorts over space and time and
individual data from the Offenders Index Database and the Police National Computer between 1980
and 2010. They estimate the long-run effect of initial labour market conditions by exploiting the
regional variation in entry unemployment rates across cohorts.
58
2.5.2. Family and crime
There is extensive literature that suggests that family life, including parental management, family size
and marital status, are strong predictors for children’s offending and criminal participation (Leschied et
al., 2008; Derzon, 2010). A meta-analysis based on longitudinal studies and carried out by Derzon
(2010) finds that the strongest predictors of criminal behaviour are in descending order parental
education, child-rearing skills and parental discord. For violent behaviour parental supervision and
family size were the most important predictors. With increasing numbers of children, parents’ attention
given to each of the children is reduced and this overcrowding can lead to conflict. An extensive
summary of the role of family size is provided in the Handbook of Crime Correlates by Ellis et al.
(2009).
There is also the argument that crime runs within families because criminal parents tend to have
delinquent children. The most comprehensive research on the concentration of offenders within families
was carried out in the prospective longitudinal Cambridge Study in Delinquent Development which
followed more than 400 males aged 8 to 48 living in South London. It revealed that criminal parents or
siblings predicted the boys’ own convictions (Farrington, 1996; Farrington et al., 2006; 2009) where,
for instance, 63 per cent of the boys with criminal fathers were criminal themselves and older siblings
were stronger predictors. A good summary of the relationship between family characteristics and crime
is provided in the book Crime and Public Policy by Wilson and Petersilia (2011, chapter 2).
2.5.3. Educational/vocational programs and recidivism
Duwe and Clark (2014) examine the role of obtaining secondary and post-secondary degrees during
imprisonment in Minnesota on recidivism and post-release employment. They apply propensity score
matching to reduce observable selection bias and observe offenders, who were released between 2007
and 2008, through the end of 2010. Although Duwe and Clark (2014) cannot control for any work
history or post-release educational attainment, they make contrasting observations about the impact of
qualification: While they find positive effects of obtaining secondary degrees on securing post-release
employment, these do not necessarily lead to higher wages (or more hours worked) and more consistent
employment. Post-secondary degrees, in comparison, did not improve the chances of entering
employment post-release but individuals earned higher income as a result of longer working hours.
These differential observations are likely to be the result of the different nature of the degrees, with the
secondary degrees focusing on basic skill development and post-secondary degree focusing on
discipline-specific knowledge.
59
Since the 2002 report by the Social Exclusion Unit on reducing recidivism, attention has moved to
improving literacy and numeracy in UK prisons, and the role of education and training has been
reemphasised by policy makers after the White Paper on Prison Reform was published in November
2016 (Ministry of Justice, 2016). However, many UK based studies so far have demonstrated the need
for educating prisoners rather than evaluating the efficiency of prison education. The exception is a
report commissioned by the Prison Education Trust from the Justice Data Lab, which examined
imprisoned offenders in England between January 2002 and March 2013 who applied to participate in
distance learning courses and buy learning materials (Ministry of Justice, 2015). The report reveals that
general participation in the program reduced re-offending by between 6 and 8 percentage points.
Participation in an academic course reduced it by 4 to 8 and in a vocational course by 6 to 9 percentage
points. Although the administrative data employed in this study is rich (e.g. it includes age, gender,
ethnicity as well as criminal, benefit and employment history), the potential influence of unobserved
characteristics, such as motivation, or other intervention programs cannot be accounted for.
2.6. Learning
2.6.1. Education environment
It is possible to explain some of an individual’s knowledge and skills accumulation through their actions
and interactions. Gurria (2016) released the results of the triennial OECD Programme for International
Student Assessment (PISA). Although not specific to the UK, numerous associations were made
between student activities and outcomes relating to school and the individual’s wellbeing. These
included:
• frequent bullying is associated with worse educational outcomes;
• physically active students are less likely to truant, feel very anxious, or be frequently bullied;
• students who spent the most time online outside school are more likely to be less satisfied with
their life, feel lonely and be less proficient in school;
• students who have a job alongside school are more likely to feel like an outsider at school, have
low expectations of further education, arrive late to school and truant.
Antonio (2004) analyse the spill-over effect among social groups during education, finding that students
who have best friends with relatively high levels of intellectual self-confidence or greater expectation
to attend university tend to exhibit the same qualities. Mora and Oreopoulos (2011) additionally assess
the impact of social groups on education. Assessing the marginal effects of a probit model it is found
that a 10-percentage point increase in the proportion of reciprocating friends intending to drop out
increased the likelihood of also dropping out by 1-2 percentage points.
60
Student’s traits and well-being levels have been found to have an impact on educational attainment.
Andrews and Wilding (2004) use longitudinal data to assess how depression and anxiety were related
to life stress and student achievement. A sample of UK graduates were asked to complete a Hospital
Anxiety and Depression Scale as well as fill out if they had been exposed to a list of threatening
experiences before university and mid-course. The authors find depression and financial difficulties are
the only two factors significantly related to subsequent exam performance, showing that both account
for an additional 1.8% of the variance in exam marks. However, only depression has a significant
regression coefficient, the authors suggest that depression mediate the relationship between financial
difficulties and exam results. Owens et al. (2012) supports this idea, finding greater levels of anxiety
and depression are related to lower academic performance. Hattie (2009), however, found anxiety to
be more significantly related than depression. Through the author’s meta-analysis, it is found that
reducing student anxiety had above average effect size (as measured by Cohen’s d) on students learning
outcomes, whereas student depression has a smaller effect size. It should be pointed out that meta-
analysis is conducted across literature from all countries, and therefore some factors may have more of
an effect in some countries than others.
Students personality traits are deemed to have an impact on a person’s education in many studies. One
such trait investigated is Emotional intelligence, which is the capacity for recognising our own feelings
and those of others, managing emotions in ourselves and in our relationships (Goleman, 1998). It is
important to recognise that emotional intelligence is not distinct from other personality traits but a
combination of them, which include self-awareness, motivation, self-regulation, empathy, and
adeptness in relationships (Cherniss and Goleman, 1998, Boyatzis, 2008, and Goleman, 2000). Many
studies show emotional intelligence is associated with academic achievement (Parker, Summerfeldt,
Hogan, & Majeski, 2004, Petrides, 2004). Petrides, Frederickson, and Furnham (2004) examine the
relationship of 650 year 11 students among EI, cognitive ability, and academic performance. They find
emotional intelligence had moderated the relationship between academic performance and cognitive
ability. Petrides et al. (2004) also find that emotional intelligence is negatively associated with deviant
school behaviours, such as unauthorized absences or being expelled from school, likely to influence
their academic performance.
Attention is referred to as “all those aspects of human cognition that the subject can control and to all
aspects of cognition having to do with limited resources or capacity, and methods of dealing with such
constraints” (Shiffrin, 1988, p. 739). McClelland et al (2013) use the Colorado Adoption Project (CAP),
a longitudinal study containing 245 adopted children, their biological parents, and their adoptive
families. The authors found that children’s age 4 attention span-persistence significantly predicted math
and reading achievement at age 21 after controlling for achievement levels at age 7 and social-
61
demographic factors. Furthermore, the authors find that those age 4 attention span-persistence skills
significantly predicted the odds of completing college by age 25, with children who were rated one
standard deviation higher on attention span-persistence at age 4 had 48.7% greater odds of completing
college by age 25.
Creativity is seen as the ability to make or bring something new to existence, providing a solution to a
new problem, a new method, device, or an artistic object to form (Nami et al, 2014). Ai (1999) examine
students who were randomly selected from 68 schools, with (2,264 students, 38% were boys and 62%
were girls). A number of tests were administrated to students around creativity. Using a self-reported
achievement scores in four areas, which included English, natural science, mathematics, and social
science, it is found that creativity was related to academic achievement for both boys and girls. Other
studies show a relationship between creative and academic achievement (Coyle and Pillow, 2008;
Palaniappan, 2005; Palaniappan, 2007a; Steinmayr and Spinath, 2009).
Resilience is defined as the “process of effectively negotiating, adapting to, or managing significant
sources of stress or trauma” (Windie, 2011). According to Alvord & Grados (2005), resilience indicates
the possession of several skills, in varying degrees, that help a person to cope, with Bernard, (1993,
1995) suggesting that resilient children have five attributes: social competence, problem-solving skills,
critical consciousness, autonomy, and a sense of purpose. Using data from the British Household Panel
Survey, (Windie, 2011) find that those with increased scores on a mental status measure (GHQ_12)
after exposure to adversity, such as bereavement, poverty, among others, returned to their pre-exposure
level after 1 year, bouncing back from their adversity. Schoon et al (2004) examines the influence of
socioeconomic adversity from school adjustment, on children age 16 and the long-term consequences
of school adjustment on those aged 33, considering factors such as parental, individual resources,
teacher expectations that might alleviate such adversity. The authors find that socioeconomic adversity
was a significant risk factor for educational failure and has an influencing impact on health-related
outcomes of age 33.
The environment the students learn in, including the structure and the quality of the learning
environment, is shown to influence student’s knowledge and skills in many studies. Stewart (2008) use
hierarchical linear modelling to show that school structural characteristics such as school size have
relatively small effects on student achievement when compared with individual-level characteristics
including school commitment and parent-student discussions. Ou and Reynolds (2008) find that
changes in the learning environment, more specifically frequent school mobility is linked to
significantly lower levels of educational attainment. Oishi and Schimmack (2010) show an additional
adverse effect of school moves, finding that these are also negatively associated with well-being.
62
Hughes and Karp (2004) survey literature around the topic of availability of school-based career advice,
finding that students benefit vocationally and academically from career course participation. Another
structural factor that is found to have an effect was standard school hours. Wheaton et al. (2016)
investigates this element in greater detail by reviewing literature around the subject, finding that later
start times tended to improve attendance, reduce tardiness, reduce falling asleep in class and improve
grades. Being a literature review, it is not clear as to what additional variables were additionally
controlled for.
There is much literature around the role that teachers play in knowledge and skills transfer and the
factors that affect these. Darling-Hammond (2000) use Schools and Staffing Surveys data to show that
measures of teacher quality (including certification status, degree class, and whether they had a master's
degrees) are strong correlates of students’ achievement in reading and maths in the US. Clotfelter et al.
(2007) reinforces the idea of teacher’s previous educational outcomes having an effect on student’s
outcomes, finding that test scores and regular licensure have positive effects on student achievement.
Teachers’ experience is also cited as having a significant effect in this study. It is important to
acknowledge that as overall school quality is not controlled for, this result could’ve arisen because
better-qualified teachers may go to better achieving schools. Atkinson et al. (2009) assesses an
additional measure of teacher quality by assessing the impact of performance-related pay schemes for
teachers in England. The authors employ a difference in difference model, controlling for pupil effects,
school effects and teacher effects. Results show that schemes did improve test scores on average by half
a grade per pupil. This effect is only observed in English and science teachers, maths teachers show no
significant improvement.
Rice (2010) examines the impact of teacher experience in more depth via an evidence review. The key
finding is that student achievement increased with teacher experience, with experience being more
valuable when teaching younger age groups. Teacher experience also shows a diminishing return after
4 years’ experience, implying a nonlinear relationship. The author cited a possible reason for this being
that inexperienced teachers are more likely to teach in high poverty schools. Harris and Sass (2011)
reinforce the idea of experience being an advantage. Using panel data, the authors show content-focused
teacher professional development and teacher experience has a positive effect on children’s educational
outcomes, however pre-teaching (undergraduate) training and teachers’ school test scores when they
were students has none. Hanushek et al. (1999) investigated the idea of salary as a measure of teacher
quality, finding that a significant relationship between salary and student achievement only exists for
experienced teachers. However, it is not clear if the salary is a signal of a better teacher or if it is received
as a result of being one.
63
Teachers have been found to work some of the longest hours, more than police officers and nurses
during an average working week which can create pressure and stress (Worth et al, 2018). Arens and
Morin (2016) investigate the effect this had on students, showing there is a significant negative
relationship between teachers’ emotional exhaustion and student achievement. Klem. and Connell
(2004) show how student reported teacher support for students is linked to student engagement in school
which in turn is associated with higher attendance and test scores. Lee (2012) supports this idea further,
showing supportive teacher–student relationships is significantly related to behavioural and emotional
student engagement, as well as being a significant predictor of reading performance. Gurria (2013)
argues that the mechanism through which teacher-student relations benefits students is greater student
engagement at school.
Carpenter (2006) analyse how different teaching methods affected educational outcomes using a
repeated measures ANOVA procedure. Findings suggest that moderately-active teaching methods such
as the jigsaw method are more effective than the more passive styles such as lecture, lecture/discussion,
and case study methods. Yoder and Hochevar (2005) support this result, finding that students scored
higher and with less within-class variation on tests where the information is presented via active learning
in comparison to lecture, autonomous readings, or video without discussion coverage. Freeman et al.
(2014) support evidence further, conducting a large meta-analysis to determine how effective active
learning (defined as learning through activities and/or discussion in class) is in comparison to traditional
lecture style learning (passively listening to an expert). Results show that student performance increased
by approximately half a standard deviation with active learning compared to lecturing.
2.6.2. Adult learning
Using the longitudinal panel dataset, the NCDS, Galindo-Rueda et al. (2003) identified the effects of
lifelong learning on wages and employment, controlling for other factors that affect labour market
outcomes. Their results found that those in the labour market in 1991 were more likely to be in work in
2000 if they had undertaken lifelong learning within the period. They also found convincing evidence
that learning leads to more learning. Those undertaking one episode of lifelong learning increased the
probability of the individual undertaking more learning. The results also show that, for individuals with
no qualifications in 1991, those who undertook lifelong learning between 1991 and 2000 were earning
higher wages in 2000 than those who had not engaged in lifelong learning over this period. In contrast,
Silles (2007) reported no returns to earning from adult schooling for men, so the definition of adult
learning is important in this mechanism. For example, Livingstone (2001) documents the large extent
of self-directed informal adult learning, whether relating to paid work, unpaid work, general interest, or
‘community work-related’ activities such as developing interpersonal and communication skills.
64
Desjardins (2011) implement a literature review to show the differing benefits of adult learning.
Evidence suggest that learning in adulthood is associated with positive labour market outcomes such as
being less likely to be unemployed and being more likely to experience wage growth. Wider benefits
include being associated with reduced criminal activity, increased social cohesion, savings in welfare
and medical costs, and increased voter participation. The Government Office for Science (Hyde and
Phillipson, 2014) looks into lifelong learning, including continuous training within the UK labour
market. They find at the individual level, participating in learning activities has shown improvement in
life satisfaction, well-being and self-confidence. At the macro-level, they find increasing the skills of
the workforce could generate an additional £80 billion for the economy, while also improving
employability of older workers. However, they also find barriers that might prevent people from
undertaking lifelong learning, highlighting three particular barriers: attitudinal, situational and
institutional. Attitudinal barriers include older individuals responding they were less likely to want to
take work-related training, but they were also less likely to be offered it. Situational barriers include
financial and time constraints, which have been consistently shown to be the most important reasons
why people do not participate in learning or training, as well as Institutional factors, such as the
availability of workplace training. In addition, Midtsundstad (2019) produced a review looking into the
research literature on adult learning and employability. The author found evidence to support that
acquiring higher education has a positive impact on employment and earnings while also having a
positive impact on upward mobility. Social skills and continuous learning ability influenced the career
success and career satisfaction of older workers, with older workers being more satisfied with their
career and hence continued to be in employment.
65
3. Determinants of earnings and their use in valuing human capital
In this section we review the literature on the determinants of earnings, and focus on the role of
education, skills, family background, health, and job mobility and on-the-job training. While we
acknowledge that there are a host of other factors that may affect earnings including individual factors,
such as ability, unique talents and effort, they are unobservable. Other earnings determinants also
include institutional factors, such as seniority wages and barriers to entry, restricting the supply of e.g.
professionals (doctors, lawyers) but they are likely to be fixed over long periods of time in any one
country due to union wage bargaining and collective agreements and other labour market institutions.
Therefore, we focus on the ones listed above as they are measurable and allow a more granular
representation of existing human capital stock values. For instance, while the current methodology
employed by ONS considers the impact of qualifications on earnings, skills gained through learning
and on-the-job training are arguably the true earnings determinant.
The first part of this section considers the role of education, which is widely recognised as one of the
key determinants of earnings. We depart from the standard analysis of returns to education (years of
schooling) and instead consider more specific types of educational attainment. This is followed by a
review of the literature on skills. In this sub-section, we focus on cognitive and non-cognitive skills
(personality skills), both of which have gained considerable attention (especially personality traits)
among policy makers and researchers. The importance of family background followed by the role of
health are then considered. The last two sub-sections review the literature on job mobility and on-the-
job-training.
Overall, findings from the review are as follows. In terms of the role of education as a determinant of
earnings, subject degree and educational institution quality is found to be of key importance. For
instance, individuals who studied or graduated with degrees in any of the STEM subjects, economics
and law, were found to receive wage premiums compared to individuals who graduated in other
humanities and social science degrees (Walker and Zhu, 2011; Belfield et al., 2018). Educational
institution quality also matters for earnings (Chevalier and Conlon, 2003; Hussain et al., 2009;
Chevalier, 2014; Britton et al.,2016) in that students, who graduated from high-quality institutions,
received a wage premium. While the relationship between educational attainment, including
institutional characteristics, clearly indicates the existence of positive returns, a further understanding
of causal mechanisms are needed. There is consistent evidence of positive effects between family
background and a child’s future earnings (Bjorklund and Jantti, 1997, 2009; Chadwick and Solon, 2002;
Solon, 2002; Corak, 2006; Chetty et al., 2015; Jerrim, 2017), regardless of country of study, although
the intergenerational elasticity is found to be higher for the UK compared to other countries. A key
66
concern with studies in this area relates to how family background is measured – earnings or income
and the time length over which these earnings are observed (Mazumder, 2005).27 The findings from the
skills sub-section show that higher ability during childhood (in terms of test scores) translates into
higher wages in adulthood (McIntosh and Vignoles, 2001; Vignoles et al., 2011; Vignoles, 2016). Non-
cognitive skills are also seen to raise earnings (Heckman et al., 2006b; Heineck, 2011; Heckman and
Kautz, 2012), although the effect depends on the type of skills examined, such as social skills and the
Big Five personality traits. The findings on the effects of health on earnings are mixed. For instance,
there are negative effects on earnings for those that experience a health shock (Lenhart, 2019), while
other studies show that the main effects of health manifest through preventing people from working;
when selection into employment is taken into account, poor health is not found to reduce wages (Brown
et al., 2010). Others, such as Kidd et al. (2000), found wage and participation rate differences between
disabled and able-bodied men, with disabled men characterised by lower wages and lower rates of
participation. Lastly, the review of the effects of job mobility as well as training and earnings show
relatively fewer evidence for the UK, and the results are often mixed. While parts of the literature show
that workers that change occupation mostly benefit in terms of higher wages relative to those who do
not experience such change (Longhi and Brynin, 2010), others such as Carrillo-Tudela et al. (2016)
show instead low wage growth.
3.1. Returns to education
For the past 40 years, considerable attention has been given to the role of human capital in economic
growth and development (Wössmann, 2016). It is now widely recognised that education (years of
schooling) plays a fundamental role in determining future earnings of individuals (Psacharopoulos,
1985, 1989, 1994; Psacharopoulos and Patrinos, 2004, 2018). The relationship between education and
earnings originated mainly from the work on human capital theory (Smith, 1776; Marshall, 1890).
Human capital theory emphasises that, investment in education (in particular higher education) will
increase future productivity and lead to greater earnings (Becker, 1964; Lucas, 1988). This implies that
different degrees add differing amounts of productivity and therefore will differ in their impact on
employment and wages. A number of explanations have been given to explain the positive relationship
between earnings and (higher) education, most notable among them suggesting that there is a signalling
effect in the labour market (Chevalier, 2004). High-productivity individuals may signal their superior
productivity by acquiring more education than individuals who are less productive. See Arteaga (2018)
for more on this. There is also the issue of selection since more able individuals, that is, those with
higher levels of prior academic achievement, stay longer in school.
27 This strand of the literature largely focuses on the relationship between parental earnings (or income) and
children’s earnings and does not control for genetic factors.
67
Studies that model the returns to education date back to the mid-20th century (Schultz, 1960, 1961;
Becker, 1962, 1964; Becker and Chiswick, 1966; Mincer, 1974) with interest in this broad area of
research persisting right into the 21st century (Card, 2001; Duflo, 2001; Heckman et al., 2006a;
Heckman et al., 2016; Psacharopoulos and Patrinos, 2018). The focus of most of these studies has been
on dimensions such as the relative returns to education by gender, subject, qualification, institution,
ethnicity, and social class (Belfield et al., 2018). For example, in relation to the returns to educational
qualification, private returns seem to be greatest for primary education and education of women
(Psacharopoulos and Patrinos, 2018). The most widely-used method when estimating the returns to
education is the Mincer earnings function (Mincer, 1958, 1974), which models the natural logarithm of
wages as a function of education, experience, and a range of additional observable characteristics that
have the potential to impact earnings.
This section will deviate from the usual examination/review of the general literature on returns to
education, mainly years of schooling (see Psacharopoulos, 1985, 1989, 1994; Psacharopoulos and
Patrinos, 2004; Walker and Zhu, 2008), and focus instead on more specific types of educational
attainment, namely higher education, degree subject, and institutional characteristics. This direction is
motivated by the fact that estimates of a single rate of return may not be very informative if returns to
education differ by education level or differ across populations (including by social strata). The
importance of this for policy can by no means be understated, although methodologically there are
concerns about these estimates.
3.1.1. Educational qualification, degree subject, and institutional quality
A number of studies have analysed the returns to education, most popular amongst them being the recent
global update by Psacharopoulos and Patrinos (2004) (most recent is the 2018 version). There are also
numerous studies for the UK, with recent evidence indicating a high level of variation in the earnings
of graduates who study different subjects or who attend different higher education institutions (Britton
et al., 2016; Walker and Zhu, 2017, Espinoza and Speckesser, 2019). Starting at relatively lower levels
of educational attainment, the Department for Education (2014) used the LFS over the period 2006 to
2013 to examine the returns to different educational qualifications. They found the estimated returns to
GCSEs and A-levels are lower at earlier years of an individual’s career, increase and peak at
approximately age 35 and then fall over the course of an individual’s lifecycle. In contrast to academic
qualifications, the earning returns of vocational level 2 and 4 apprenticeships were considered highest
in the first few years and then diminished at higher years. The authors also discounted productivity
gains, where the productivity gains are with respect to society and defined as lifetime wage returns for
an individual of obtaining some higher-level qualification relative to a lower level qualification. The
68
lifetime productivity gains to GCSEs, A-levels, and apprenticeships were higher compared to similar
individuals with lower qualifications. Despite the gains in higher educational qualifications, evidence
suggests that some of these differences are gendered (Psacharopoulos and Patrinos, 2018). More
specifically, the average lifetime productivity gains for individuals with five or more GCSEs including
mathematics and English amounts to £63,000 for males and £54,000 for females compared to those
with qualifications below level 2. Moreover, the marginal lifetime productivity gains in comparison to
those of no qualifications are around £283,000 for males and £232,000 for females. A very recently
published working paper by Espinoza and Speckesser (2019) compare earnings of people with higher
vocational/technical qualifications to those of degree holders using the earliest cohort of English
secondary school leavers with newly available Longitudinal Education Outcomes data. They find that
in the mid-twenties the initially higher earnings of people with higher vocational education disappear
but that high returns to higher vocational/technical education in STEM subjects exist, which remain
significantly above those of many degree holders by age 30.
3.1.2. Higher education and earnings
The issue of higher education participation continues to remain a matter of policy concern, primarily
due to the higher associated earnings and the lack thereof of individuals from certain socio-economic
background not being able to participate meaningfully. Interestingly, existing global evidence suggests
that the proportion of young people enrolling into higher education has increased (Becker et al., 2010).
Focus has now shifted to examination of the returns to different educational degrees. For example,
Belfield et al. (2018) uses administrative data (Longitudinal Education Outcomes) to investigate the
returns to higher education in the UK five years after graduation. They find that labour market returns
to different degrees vary considerably even after accounting for the differences in student composition.
Specifically, their study shows that individuals that studied for a degree in Medicine, Maths, and
Economics earned at least 30 per cent more compared to the average graduate. In contrast, creative art
graduates earned 25 per cent less. In terms of gender differences, the authors show returns to be greater
for females that studied Medicine or English whereas for males, returns were higher for Computer
Science graduates. There are also differences in the returns to educational degrees based on family
background. For instance, returns were greater for individuals that studied Medicine but from lower
socio-economic background, compared with Economics where returns are highest for students from a
higher socio-economic background. In addition, they show that after conditioning on all other
characteristics, such as degree subject and institution, graduates from independent schools and the top
quintile earn around seven per cent to nine per cent more than those graduates from the lowest SES
backgrounds. Interestingly, the authors also found that some of the differences in relative earnings could
be explained by the university attended (there is limited research on this for the UK), whilst
acknowledging that these differences may be underpinned by the demand for various skills in the labour
69
market and quality of higher education. In particular, high-status universities, such as the Russell Group
and universities established before 1992, typically have higher earning graduates, although one must
bear in mind that high ability students have a higher likelihood of attending higher-status universities.
Walker and Zhu (2011) also provide additional evidence on the relationship between earnings and
different degree subjects in England and Wales using data from the LFS. A key consideration of their
analysis is the inclusion of postgraduate degrees. Men with Law, Economics, and Management degrees
experienced larger returns as well with Science, Technology, Engineering, and Math degrees. The
returns were however lower for graduates that studied for a degree in other social sciences subjects and
humanities, and for those with combined degrees. The authors also provided analysis of the effect of a
rise in tuition fees on earnings and found that although this led to an overall decrease in earnings of
around one to three per cent, the differentiated effects on different degree subjects still remained. That
is, degrees that offer higher returns remained better off and degrees with lower returns continued to
offer poor returns. There is further evidence for the UK on the returns to different degrees (see O’Leary
and Sloane, 200528; Chevalier, 201129; Walker and Zhu, 201330; Britton et al., 201631). Evidence of the
variation by subject of study and institution has also been found elsewhere, such as the US (Altonji et
al., 2012) and Norway (Kirkeboen et al., 2016).
3.1.3. Institutional characteristics
A few studies provide evidence on how earnings vary by institutional characteristics in the UK.
Exceptions are Chevalier and Conlon (2003), Hussain et al. (2009), Chevalier (2014) and Britton et al.
(2016). These authors use an indicator of university quality, including measures for research intensity
and student intake (aggregated), and found that earnings were higher among individuals that attended
higher-quality institutions, with the effects strongest at the top end of the earnings distribution. Britton
et al. (2016) also analysed the variation in graduate earnings by higher education institution attended
using linked administrative tax data. The authors found variations in earnings both within and across
different institutions, although most of the variations observed were explained by student background
and subject differences. Chevalier (2014) uses data from the Longitudinal Destination of Leavers from
Higher Education linked to administrative data from HESA to provide estimates of the effect of
institutional quality on earnings of early career graduates in the UK. A strength of this study is that the
authors use different identification strategies, including, generalised propensity score matching to
28 O’Leary and Sloane (2005) make use of data drawn from the LFS over the time period of 1994 to 2002. 29 Chevalier (2011) utilise data from the Longitudinal Destination of Leavers of Higher Education. This contains
a sample of graduates from the 2003 cohort. 30 Walker and Zhu (2013) exploit data from a number of sources including the LFS covering the period from 1981
to 2010, the BHPS starting from 1991 over 18 waves, and the Destination of Leavers from Higher Education. 31 Britton et al. (2016) employ tax and student loan administrative data and data from the UK’s HESA.
70
address issues of selection into different institutions. For more different techniques to address selection
and other causal effects, see (1) Black and Smith (2004, 2006), who use propensity score matching
methods to match on the predicted probability of attending a high-quality university, (2) Long (2008),
who uses the average quality of colleges within a certain radius of a student as an instrument for the
quality of the college at which the student attends, (3) Hoekstra (2009), who exploits a large
discontinuity in the probability of enrolment at the admission cut-off, (4) Saavedra (2008), who uses
scores on a national college entry test creating exogenous peer and resource quality variation near
admission cut-offs, and finally (5) Dale and Krueger (2002), who match students who applied to and
were accepted by similar colleges to eliminate the bias that colleges may admit students based on
characteristics related to future earning, for more on different techniques to address selection and other
causal effects. Their findings suggest evidence of heterogeneity in the returns to institutional quality,
which they measured using a set education inputs, namely research assessment score, academic
expenditures per student, mean entry grade and graduate prospect, and student-staff ratio. In particular,
the authors find almost no returns for below median quality institutions and rather large returns for those
that attend high-quality institutions. Their econometric approach also indicated that OLS estimates were
biased upwards and larger than the estimates obtained from the generalised propensity score method.
Hussain et al. (2009) also provide further examination of the earnings effects of institutional quality for
four graduate cohorts (1985, 1990, 1995 and 1999) for the UK using data from HESA to obtain
measures of institutional quality. The authors measure institutional quality using information on
research assessment exercise, faculty-student ratio, retention rate, total tariff score (based on A-levels
or other qualifications), mean faculty salary, and expenditure per student. While not addressing issues
related to selection and unobserved heterogeneities, the authors find similar to previous studies that
there is a positive return to attending higher quality institution. They also found differences in earnings
along the quality distribution. For instance, students that attended institutions in the top quartile of the
research assessment exercise score, retention rate or total tariff, were associated with increases in
earnings between 10 and 16 per cent.
A recent study by Belfield and Erve (2018) deviates from the usual examination of the effects of higher
education on earnings and considers the implications on later adulthood income of women. The authors
show that gaining a higher education qualification increases women's net family income in adulthood
by around 20 per cent relative to leaving school at 18. There are also indirect effects on earnings through
the number of hours that women work, in addition to increasing the likelihood that a woman would
have a partner with a higher education qualification. The authors further note that higher gross earnings
associated with higher educational qualifications could reduce net returns to higher education due to tax
payments. In a further examination of some of the causal mechanisms, the authors show that the
71
mechanism through which higher education affects incomes varies over the life cycle. They show that
at earlier ages, the most important mechanism is the woman’s own earnings. However, the impact of
woman’s higher education becomes less relevant at around age 42 when instead the partner’s earnings
become the most important determinant, i.e. associative mating becomes more important.
Delaney and Devereux (2019) also deviate from the traditional analysis of returns to education by
focusing instead on the variation in earnings over time using three different measures of earnings
volatility, namely earnings variability, degree of earnings cyclicality, and the probability of receiving a
pay cut over a 5-year period. The authors used data from the New Earnings Survey Panel Dataset to
estimate a regression discontinuity based on the change in the compulsory schooling reform that took
place in the UK in the 1970s. The authors find that each extra year of schooling reduced earnings
variability particularly at younger age and reduces earnings cyclicality. Their finding also show that
educated men tend to be affected less by wage cyclicality and the probability of receiving pay cut is
found to be reduced almost 3.5 percentage points.
3.1.4. Identifying causal effects
A fundamental concern with estimates of returns to education is that measures of education are often
considered to be exogenous. However, the literature on returns to education shows that education is
endogenous due to certain unobserved factors that influence the individual’s ability to acquire more
years of schooling (while not having an independent effect on an individual’s earnings) as well as
measurement errors (Harmon et al., 2003; Dickson and Harmon, 2011; Heckman et al., 2018).
Previous studies have addressed these concerns of endogeneity mainly by employing IV estimation
techniques (Angrist and Krueger, 1991; Harmon and Walker, 1995; Buscha and Dickson, 2015).
Changes in compulsory schooling laws, which occurred in the UK in 1947 and 1973 have been used
extensively as a candidate when addressing issues of endogeneity (Harmon and Walker, 1995;
Oreopoulos, 2006; Grenet, 2009; Devereux and Hart, 2010). For instance, Devereux and Hart (2010)
exploited variation in the school leaving age reform that occurred in Britain in 1947 as an instrument
for education and found the returns to education to be around six per cent. Devereux and Fan (2011)
also used the school leaving age policy reform to uncover causal effects of education on earnings. The
authors showed large increases in educational attainment with larger number of students staying longer
in school post the reform, as well as hourly wages and weekly earnings rising over the expansion period.
See also Dickson and Smith (2011) who use the school leaving age policy reform as well in addition to
another institutional rule, the Easter Leaving Rule, to examine the returns to education. Grenet (2013),
on the other hand, found returns close to nil in France using a similar policy reform that took place in
France in 1967 as an instrument for education. Building upon the Mincer equation (1974), Heckman et
72
al. (2016) developed a robust dynamic discrete choice empirical framework that highlights the
importance of accounting for the dynamics of schooling choices at different stages of educational
decision-making in estimating the causal impact on earnings. This work conformed with the earlier
work of Becker (1964), strongly suggesting that education has a significant causal effect on earnings.
As a robustness exercise, Dolton and Sandi (2017) provide a reassessment of a number of studies
(Harmon and Walker, 1995; Oreopoulos, 2006; Devereux and Hart, 2010) that previously addressed
causal effects of education on earnings for the UK, using data from the General Household Survey and
the Family Expenditure Survey. The authors found that previous estimates of the returns to education
from these studies were sensitive to equation specifications.
Despite IV techniques being the most frequently used, there are concerns about their appropriateness
(Heckman and Urzua, 2010), most notable among them being weak instruments which can lead to
biased estimates, and the fact that IV estimates rely on strong, a priori data assumptions and are likely
to give different results depending on the instrument used and assumptions made. See Heckman and
Vytlacil (2005) for more on alternative estimation techniques.
3.2. Skills and personality traits
The pursuit of quantifying labour market outcomes, such as wage premiums under the traditional
approach where one makes use of human capital variables, such as education and experience, is argued
to be insufficient in relatively recent years. It is argued that cognitive skills may result in enhanced
performance and productivity differences which could potentially result in wage differentials (Anger
and Heineck, 2010). Recent work identified cognitive skills, such as numeracy (Hanushek et al., 2015),
problem-solving and ICT skills (Falck et al., 2016), key in explaining most of the variation in earnings
differences in technology-rich environments. At the same time, the increase in automation also
highlighted the financial returns to non-cognitive skills, such as social skills and personality traits (e.g.
Heckman et al., 2006b; Lindqvist and Vestman, 2011; Heckman and Kautz, 2012; Deming and Kahn,
2018) and their complementarity with cognitive skills (Weinberger, 2014; Deming, 2017; Deming and
Kahn, 2018). These findings are discussed in detail below.
3.2.1. Cognitive skills
Most of the work investigating the influence of cognitive abilities has focused on the UK and the US,
with much evidence of a positive relationship between cognitive abilities and earnings (Cameron and
Heckman, 1993; Green and Riddell, 2003; Bronars and Oettinger, 2006; Anger and Heineck, 2010;
Lindley and McIntosh, 2015). Others noted that this may not be the case, and in fact, cognitive ability
73
is a much weaker predictor than education, family background characteristics or the environment
(Cawley et al., 2001; Zax and Rees, 2002; Anger and Heineck, 2010). An alternative view is offered by
Bowles and Gintis (2000), who compare two wage equations, one with education as a predictor and the
second including a measure of cognitive skills as a variable and argue that approximately 20 per cent
of the schooling effect on earnings is cognitive. A more recent study conducted by Hanushek et al.
(2013) modified the benchmark Mincer equation to reflect cognitive skills - measured across numeracy,
literacy, and problem-solving skills - and employed data from PIAAC to analyse the value of these
skills for the whole of the labour force spanning 22 countries. They found that a one-standard-deviation
increase in numeracy skills is suggested to be linked with higher wages of 18 per cent for prime-aged
members of the labour force. However, it is noteworthy that there is a significant amount of
heterogeneity among these 22 countries, with the US having the greatest return of 28 per cent which is
approximately double the return observed for the countries with the lowest returns; Sweden, Czech
Republic, and Norway. Another finding is that the return estimates obtained for workers between the
age of 35 and 54 consistently indicated greater and robust returns to these respective cognitive skills
than for the members of the labour force, aged between 25 and 34, on average a 4 percentage point
difference. Moreover, the empirical results for the wage returns of literacy and numeracy are
consistently higher than the returns to problem-solving skills. A similar study that makes use of PIAAC
investigated the returns to ICT skills across 19 developed countries. Falck et al. (2016) acquired
empirically sound results and highlight that a one-standard-deviation increase in ICT skills is associated
with an eight per cent return, with the effect being substantially stronger in Germany. Nonetheless,
Vignoles (2016) notes that these studies, that examine skills utilising datasets like PIAAC, are more
correlational studies and thus may not necessarily show the true causal impact of cognitive abilities.
Nevertheless, they offer insightful evidence since they show that, given a similar level of education,
those with a higher quality skillset earn greater wage premia.
With respect to the UK, earlier work that utilises UK birth cohort longitudinal datasets, which provide
information on cognitive ability through tests taken in primary and secondary school, show evidence
that for a given level of initial ability (test scores) during childhood, those with better basic skills in
adulthood tend to earn higher wages and have greater rates of employment (McIntosh and Vignoles,
2001; Vignoles et al., 2011; Lindley and McIntosh, 2015; Crawford and Cribb, 2015; Vignoles, 2016).
These studies provide more robust evidence on the impact of basic literacy and numeracy skills on
earnings. For instance, findings by Vignoles (2016) suggest that a one-standard-deviation in literacy
skills is associated with a wage premium of approximately 14 per cent and numeracy skills with a
premium of around 11 per cent obtained from a cohort of 34-year-olds in 2004 via the 1970 BCS. The
same dataset is used by Crawford and Cribb (2015) who find that reading skills are associated with
significant increases in gross hourly wages and gross weekly earnings, particularly at the ages 38 and
74
42 and amongst those from poor backgrounds. Lindley and McIntosh (2015) study the rising graduate
wage inequality observed in the UK and argue that degree subjects provide variation across graduates
which can be exploited to explain potential causes of the growth in graduate wage inequality if
inequality has increased more within specific subjects and if such variation is related to graduates’
characteristics in those subjects. They use a combination of datasets (NCDS, the BCS and the
Longitudinal Survey of Young People in England) and find that the greater cognitive skills dispersion,
measured in terms of maths and reading test scores at the age of 10, are associated with increased
graduate earnings inequality over time.
3.2.2. Personality traits
A number of reasons are associated with why personality traits can be important in explaining labour
market outcomes. Most prominent among them is that personality can affect individual’s performance
on the job (Thiel and Thomsen, 2013) and can be used as a sorting mechanism, whereby individuals
with certain traits sort into certain occupations. Personality trait as a source of individual differences in
achievements have been examined thoroughly in the psychology literature (see Roberts et al., 2007 for
a review), which is now receiving attention in the economics literature (Heckman and Rubinstein, 2001;
Cawley et al., 2001) where it is often referred to as non-cognitive skills. Others include Borghans et al.
(2008), Heckman and Kautz (2012) and Heckman et al. (2006b). Views from the economics literature
indicate a positive association between personality traits and earnings, and also with education. For
instance, individuals with certain traits tend to stay longer in school which often demonstrates higher
ability and acts as a signal to employers. These individuals are then rewarded with higher wages in the
labour market (Weiss, 1995). A number of studies provide support for this of which Anger and Heineck
(2010) is an example. Heckman and Kautz (2012) also show the importance of personality traits in
determining individual success.
In the economics research community, data limitation explains part of the lack of empirical examination
of the role of non-cognitive skills in the labour market. Compared to cognitive skills, which are easily
available in many surveys (NCDS for the UK), access to data that captures information on non-cognitive
skills are limited and economists are unfamiliar with psychometric measures (Heineck, 2011). Existing
empirical evidence shows that non-cognitive skills raises earnings directly through their effect on
productivity, while also indirectly raising schooling and other outcomes (Heckman et al., 2006b;
Heckman and Kautz, 2012). Cobb-Clark and Tan (2011) provide an assessment of gender differences
in non-cognitive skills (measured by the Big Five Personality traits - (McCrae and Costa, 1997) and
locus of control) and show how this affects occupational choice and consequently wages for Australia.
The authors find non-cognitive skills to impact on employment probability. Specifically, they observe
75
that despite men and women having similar non-cognitive skills, men and women tend to enter
occupations at different rates. The authors further find that non-cognitive skills give women slight wage
advantages. In Sweden, Kajonius and Carlander (2017) examined how much of life outcomes were
explained by personality traits once childhood socioeconomic status was controlled for. Interestingly,
the authors found that personality traits were just as important or even more important in determining
adulthood outcomes, including annual income and life satisfaction. Childhood socioeconomic
conditions were instead more related to educational attainment.
Personality traits and economic success have also received attention in the UK. Using data from the
BHPS, Heineck (2011) provides evidence on the extent to which the Big Five traits, namely, openness
to experience, conscientiousness, extraversion, agreeableness, and neuroticism, affect wages. The
author finds that agreeableness and neuroticism (females only) are penalised, whereas openness is
rewarded in the labour market. Nandi and Nicoletti (2009) provide a decomposition analysis for the UK
also using data from the BHPS. Similar to previous studies, the authors find that openness to experience
is rewarded positively in the labour market whereas agreeableness and neuroticism are associated with
lower wages. Lindley (2018) uses data collected from an online survey of undergraduate students from
the King's College London Business School and uses the Big Five personality traits to generate
psychopathy, narcissism and Machiavellianism scores. She finds an hourly wage premium to
Machiavellianism of around two per cent with every one‐point increase on the Machiavellianism scale.
This premium is largest at the 90th percentile, over and above all productivity‐related explanations.
Continuing with the theme of the big five personality traits, individuals who are more agreeable tend to
be more accommodating and cooperative in their attitude within the social environment, agreeable with
others, leading to higher academic motivation and behaviours which improve academic performance
(Vermetten et al, 2001). While agreeableness is linked to an improvement in academic performance, its
impact on wages is ambiguous (Vedel, Anna & Poropat, 2017). Some studies have shown agreeableness
having a negative relationship with earnings, with reasons being down to workers helping the interests
of others, being poor wage negotiators, having an egalitarian attitude towards work and pay (Nyhus and
Pons, 2005). On the other hand, agreeable individuals might be rewarded by employers because they
are more likely to respond to incentives by firms (Nyhus and Pons, 2005). On an undesirable note,
people with low score on agreeableness may have something in common with Machiavellian
intelligence, which has been found to have a positive effect on earnings and occupational attainment
(Turner & Martinez, 1977).
In addition, conscientiousness has also been found to be closely related to motivation to learn (Colquitt
et al., 2000) and reliability (Mount and Barrick, 1995). Duncan and Dunifon (1998) examined the long-
76
term effects of workers motivation on wages, exploring the relationship between social psychological
traits observed in men aged between 21 and 29 and their labour market attainments 15 to 25 years later.
Controlling for schooling, parental background, and cognitive skills, they find that individuals posing
both an orientation toward challenge and a belief that their own actions are effective, earn higher wages
20 to 25 years later. Within this work, Duncan and Dunifon (1998) also acknowledged that those
individuals with an orientation to achievement and self-efficacy earned considerably higher wages over
the same period. Self-efficacy is the belief that an individual has to perform the behaviours needed to
achieve a desired outcome (Bandura, 1997). Van Dam, Oreg, & Schyns (2008) found that employees
with a high role breadth self-efficacy are more open to organizational change, with Van Dam & Seijts
(2007) also finding that employees are more likely to report more learning and innovative behaviours.
Schnitker & Emmons (2007) suggested that patience is positively correlated with an individual’s
personal well-being, positive coping virtues, and thriving. John, Donahue, & Kentle (1991) found that
patience was correlated with higher Agreeableness, higher Openness, and lower Neuroticism, as well
as finding a relationship between patience and well-being, decreased depression, lower incidence of
health problems, and pro-social traits, such as empathy. Being more agreeable is also related to
individuals being more co-operative. Co-operation has been expressed as the individual differences in
identification with and the acceptance of other people, with individuals with a high-level of co-operation
described as empathetic, tolerant, compassionate, supportive, and fair individuals, who get pleasure
from trying to co-operate with others (Kose, 2003).
A very recent literature strand highlights the role of social skills and their complementarity with
cognitive skills, but they are all US based (Weinberger, 2014; Deming, 2017; Deming and Kahn, 2018).
For instance, Weinberger (2014) uses data from two National Center for Education Statistics
longitudinal studies of high school students (National Longitudinal Study of the High School Class of
1972 and the National Education Longitudinal Study of 1988) which provide information on senior year
math scores and comparable questions about extracurricular participation, leadership roles, and earnings
seven years after the senior year of high school. Deming (2017) uses the NLSY 1979 and 1997 and
finds evidence that social skills (measured as a standardized composite of sociability in childhood and
adulthood and participation in high school clubs and in team sports) and cognitive skills (measured
through test scores) are complements in a Mincerian wage equation. Deming and Kahn (2018) use
employment vacancy data provided by Burning Glass Technologies to examine changes in skill
requirements demanded by firms and labour markets. To capture cognitive skills, they search for terms,
such as problem solving, research, analytical, critical thinking, math, science, or statistics while social
skills are comprised of communication, teamwork, collaboration, negotiation and presentation skills.
These pieces of work generally find that self-organisation or advanced numeracy skills in digital
77
intensive industries have particularly high returns. A cross-country comparison using PIAAC shows
that workers endowed with a combination of these two skills sets or numeracy skills, managing and
communication skills receive an additional wage premium (Grundke et al., 2018). These findings are
consistent with the literature strand on dynamic skill complementarities in the production of human
capital, which argues that the return to investments in skills, such as ICT skills, is higher when the initial
level of skills is already high (e.g. Cunha and Heckman, 2007; Cunha et al., 2010; Aizer and Cunha,
2012).
3.3. Family background and earnings
A strong relationship between family background, often measured as parental income, and children’s
economic success is often considered an indicator of equality of opportunity. If most individuals’
socioeconomic outcomes are strongly related to those of their parents, the implication is that children
from poor families/backgrounds are more likely to be relatively poor as adults.
The question of how family background influences children’s lifetime economic status has been a
subject of considerable interest, beginning with the work of Becker and Tomes (1986). The
underpinning hypothesis of this argument is based on the notion of parental altruism towards children
and the investment in children’s human capital. Becker and Tomes (1986) also present a number of
findings: first, earnings adjust to the mean at a faster rate in richer families than in poorer ones because
of capital constraints suffered by poor families; second, consumption, differently from earnings, regress
to the mean more rapidly in poorer households than richer households in the case where there is no
correlation between fertility and parents’ wealth; third, fertility is positively correlated to family’s
wealth but in richer families the regression to the mean occurs as for the relation between consumption
per child and of parents.
The economics literature on the role of family background (parental income) on children’s future
economic success (intergenerational mobility) often relies on monetary measures, with income or
earnings of the child often used as the outcome variable. In addition, most parts of the literature
measures mobility (how children’s earnings are shaped by family background) using the correlations
between earnings of sons and their fathers (Blanden, 2013). This section of the review will present
evidence of the relationship between family background, mainly parental income, and children’s
earnings with focus on the UK where possible. Depending on the type of data and cohort used, the few
empirical studies that analysed the trend in intergenerational mobility in the UK have produced mixed
results.
78
Blanden et al. (2004) compare two birth cohorts (1958 and 1970) in the UK using NCDS, a survey of
all children born in the UK between 3 and 9 March 1958, and the BCS, a survey of all children born
between 5 and 11 April 1970. They showed that the link between parental income32 and children’s
(sons) earnings was stronger for the recent cohort compared to the older one. In particular, they found
an IGE of 0.12 for the cohort in 1958 and around 0.25 for the cohort in 1970, indicating that children
whose parent’s income is higher in the parent’s generation expect their own earnings to be 25 per cent
above the average of their own generation. In a similar exercise, Dearden et al. (1997) used the NCDS
for a sample of individuals born in 1958 and found IGEs ranging between 0.40 to 0.60 for sons and 0.45
to 0.70 for daughters, depending on the econometric method used.33 Blanden et al. (2005) uses survey
data and registers for Europe and North America, the UK and the US and found the same order of
intergenerational mobility (0.27 and 0.29 respectively). Jäntti et al. (2006) found similar results.
Utilising the NCDS, similar to Blanden et al. (2004) and Blanden et al. (2005), Belfield et al. (2017)
confirm the strength of the link between parental income and son’s labour market earnings. In Belfield
et al. (2017), the authors use net family income rather than gross individual earnings. This enabled the
authors to include individuals that were out-of-work. In doing so, they found that for the population of
males born in 1970 an increase in parental income at the age of 16 increased net family income of the
child by 0.35 per cent at age 42. This increased by 0.12 percentage points from the previous cohort.
According to Haider and Solon (2006), both the elasticity and the correlation coefficients are affected
by a life-cycle bias, which depends on the son’s age. This implies that IGE estimates will be
underestimated for younger children (sons) and overestimated for older ones. Past studies have
attempted to address this by using earnings data for children at older ages and also by averaging parental
earnings over a longer period of time rather than using one time period estimates (Solon, 1992; Corak
and Heisz, 1999; CBS, 2011). According to Mazumder (2005), using average earnings over a longer
period of time mitigates concerns of biasedness in estimated IGEs. Chetty et al. (2014) however argue
that the baseline measures do not suffer from substantial life-cycle or attenuation bias as mobility
estimates are said to stabilise by the age of 30 and are not significantly sensitive to the number of years
used to measure parental income (see Mazumder, 2005; Haider and Solon, 2006; Gregg et al., 2016).
In the same study, the authors attempt to address the concern of individuals with zero income. They do
this by dropping individuals with zero reported income and as a result they find that estimates are
32 The NCDS parental income data comes from separate measures of father’s and mother’s after-tax earnings, and
other income, whereas the BCS contains data on only combined parental income. 33 The different estimation methods used are: OLS using the estimated residuals; OLS using the predictions
approach; and IV estimation with different combinations of instruments (father’s socioeconomic status, e.g. social
class; father’s education). OLS estimates were found to be downward-bias.
79
sensitive to different specifications. Overall, they find IGEs ranging from 0.26 to 0.70. See also Corak
(2013), Jerrim (2017), OECD (2018), World Bank (2018) for recent updates on IGE.
The concern of how parental earnings are measured is a contentious one. Previous literature often relied
on only paternal earnings, although more recently mother’s earnings have been considered, owing
primarily to the recent increase in female labour force participation. In addition, earlier studies
considered only father-son correlations. Concerns have also been raised in particular on how family
background should be measured, with some arguing that using family income, rather than parental
income, may be more appropriate and may have different effects on children’s earnings (Mazumder,
2005; CBS, 2011). Family income could better capture socioeconomic transition and may provide
further insight into the effects of family resources on adulthood outcomes. Mazumder (2005)
experiment with this and find that family income is less prone to error and can capture a lot more
compared to earnings which are mostly observed for individuals in work. Specifically, Mazumder
(2005) find that the IGE between family income and children’s earnings is higher than the IGE between
father’s earnings and children’s earnings. Chadwick and Solon (2002) observed similar effects.
3.4. Health as an earnings determinant
The association between health and labour market outcomes is theoretically embedded in the notion of
human capital investment with most of the early literature studying returns to education rather than
health (Mincer, 1958). However, with increasing recognition of health as a significant contributor to
human capital (Becker, 1962; Mushkin, 1962; Grossman, 1972a, 1972b), focus shifted and scholars
examined empirically its role on the willingness to supply labour and on wages (Grossman and Benham,
1973; Luft, 1975; Bartel and Taubman, 1979; Berkowitz et al., 1983). Grossman (1972) describes health
as a “durable capital stock that produces an output of health time” which can be spent on leisure and
work. However, poor health reduces the amount of time that can be spent in generating income, which
is referred to as ‘absenteeism’ from work. At the same time poor health also affects the quality of the
time available and therefore can result in a productivity loss, which is called ‘presenteeism’ in the
occupational health literature (Burton et al., 1999). Since hourly wages are assumed to reflect marginal
productivity, economists often use earnings as the closest proxy for productivity. However, other
reasons put forward as to why health and earnings are associated include that employers may
discriminate against individuals in poor health or they may believe that health is correlated with
unobserved characteristics which are positively associated with productivity.
Early empirical evidence of presenteeism was limited to the use of cross-sectional data but recently
scholars started exploiting longitudinal data to disentangle the fact that health contributes to human
80
capital stock (Grossman, 1972), which at the same time determines wages (Becker, 1964) and the
amount of time spent in the labour market. However, addressing all sources of the endogenous nature
of health in the health-employment nexus remains challenging and UK based work is limited. First,
there is the issue of reverse causality, which was introduced earlier. Second, there is unobserved time-
invariant heterogeneity, where factors, such as personality traits and individual-specific circumstances,
can drive the relationship between health and earnings. For instance, very motivated individuals can
have higher wages and have better mental health. Lastly, a lot of the existing work relies on self-reported
health information but a review of existing literature on the validity of these reveals that this measure
is often plagued by measurement error and ‘reporting bias’ (Kerkhofs and Lindeboom, 1995) because
individuals are heterogeneous and different sub-groups use different reference points when answering
the same health-related question.
Recently published work by Lenhart (2019) examined the health-earnings nexus in the UK using
different time periods of the BHPS. By exploiting the longitudinal nature of the survey and using
propensity score matching combined with a DiD approach, he matched individuals who experience a
worsening in self-assessed health (treated group) to others that are similar but maintain their health
status (control group) based on characteristics prior to the health shock and first differenced the
outcomes of the treated group and the control group to eliminate any unobservable fixed effect that
influence the selection into the groups as well as the employment outcomes. He found that hourly wages
decrease by more than £2 when analysing individuals over a 9-year period suggesting that individuals
who stay at work after a negative health shock experience lower wage growth. The estimates are
comparatively small and imprecise when observing the 3-year period. When examining annual labour
income, which implicitly accounts for absenteeism as well as presenteeism, he detects a decline of
around £1,200 for the year after the shock when observing individuals over a 3-year period. This figure
increases to up to approximately £4,400 when following people over a 9-year period which examines
the effects for up to 4 years after the health shock. Therefore, Lenhart argues that people do not
necessarily adapt to their health status and that they have difficulties being reintegrated. Since these
effect sizes are even larger in magnitude for overall household income, individual health effects are
likely to spillover to other household members who take time off from work for caregiving activities.
More severe rather than mild health declines lead to higher income reductions with larger effects
observed when workers are followed over a longer period, indicating that labour market effects are
persistent and not temporary. These findings are replicated when using the onset of a health condition
as an adverse health shock. These findings are partly consistent with early and widely cited work by
Contoyannis and Rice (2001) who also used the BHPS and self-assessed health, but they identify
gender-specific effects on earnings. While they find poor general health to reduce hourly wages only
for women, they identified mental health to be the significant determinant of males’ wages.
81
The problem is that people out of work often differ systematically from individuals who are in work.
Since they are more likely to have lower levels of education and a greater incidence of health conditions,
they are expected to be on average less productive and, as a result, earn lower wages were they to work
according to human capital theory. Brown et al. (2010) addresses this complexity by estimating health
effects in earnings in an endogenous switching model which predicts reservation wages for the
unemployed and market wages for the employed using the same dataset as Lenhart (2019). They use
usual net pay per month converted into weekly wages and the weekly reservation wage of non-working
but job seeking individuals provided in the BHPS. They construct their own health variable, based on
work on health and retirement behaviour by Bound et al. (1999) and Disney et al. (2006), and predict
self-assessed health categories using socio-economic characteristics and health problems in a
generalized ordered probit model. This approach allows individuals with the same underlying level of
health to choose different health thresholds on the assumed underlying continuous scale (Kerkhofs and
Lindeboom, 1995; Lindeboom and van Doorslaer, 2004; Rice et al., 2011). The predicted dummy
variables for excellent, good or very good and fair health (poor is the base category) were then used in
a full information maximum likelihood approach to estimate the endogenous switching regression
model. Using this approach, they come to a different conclusion than Lenhart. They find that the main
effect of health is the prevention of people from working but, once they account for selection into the
labour market, be it currently employed or unemployed, poor health does not reduce the market or the
reservation wage.
Some scholars argue that employment exits of individuals in poor health can be a result of incentives
created by disability benefits. However, UK schemes are not linked to previous earnings as it is often
the case in other countries and provide benefits at a flat rate instead, which is unlikely to incentivise
unnecessary employment exits. For the UK, Walker and Thompson (1996) used the first three waves
of the BHPS and found that, after controlling for the endogeneity of schooling, disability has a greater
effect on labour market participation than on wages. In a similar vein, Kidd et al. (2000) compare the
labour market outcomes of disabled and able-bodied men in the UK using the LFS and find substantial
wage and participation rate differences between the two groups with the disabled characterised by lower
wages and lower rates of participation with productivity-related characteristics explaining
approximately 50 per cent of the differentials. More recently, Jones et al. (2006a) stated that, although
the average wages of the disabled are over 85 per cent of their non-disabled counterparts, the rate of
participation of disabled people is approximately half that of the non-disabled. Similarly Jones et al.
(2006b) provides an extensive review of empirical evidence on how disability has affected labour
market outcomes by gender since the introduction of the Disability Discrimination Act in 1995 and
82
conclude that differences in employment and wage prospects exists for those with mental health
problems.
3.5. Job mobility
There is by now a general agreement that human capital plays a major role in determining earnings. The
initial focus of the literature on firm-specific capital motivated the study of the relative importance of
specific capital, by examining the effects of firm tenure on earnings profiles. Some of this literature
provides conflicting evidence on the magnitude of tenure effects (see e.g., Topel, 1991; Abowd et
al.,1999). Parrado et al. (2007) find that in the US occupational movement is associated with lower
earnings, even controlling for selection effects (once personal characteristics, such as schooling, work
experience, job tenure, marital status, and race are controlled for). They find that older and less educated
workers are less likely to shift occupation or industry, as are better paid men. This result is not found
for better-paid women; they also stress that the tendency for younger and more educated workers to
change occupations and industry more frequently has risen over time.
Longhi and Brynin (2010) find that, both in Germany and Britain, workers who change occupation
mostly benefit from the change in terms of wage increases, relative to those who do not experience such
change, and while this is no different to those that change job only, they achieve higher satisfaction.
Among the benefits associated with job and occupation changes, increase in life satisfaction is the
largest. Their analysis shows that change is on average beneficial, in terms of both wages and job
satisfaction, with those who change occupation generating a wage premium as high as those who change
jobs but do not change occupation. This is different to some of the evidence for the US (see Kambourov
and Manovskii, 2008 for the US) which indicates that high occupational turnover results in relatively
low growth in wage. Using the LFS, Carrillo-Tudela et al. (2016) distinguishes wage effects in relation
to occupational mobility, depending on the level of skills and finds that UK workers at the bottom of
the wage distribution, who changed occupations, experienced a fall in real wages, while switchers at
the top of the wage distribution are the ones that engendered the largest increases in wages following
the transition.
Parent (2000) estimated earning equations using US data and included industry experience in the
regressors of wage equations. This paper finds that what matters most for the wage profile in terms of
human capital is industry-specificity, not firm-specificity. Including total experience in the industry as
an additional explanatory variable, this paper shows that the return to seniority is markedly reduced
83
using Generalised Least Squares (GLS) while it virtually disappears using IV34- GLS, at both the one-
digit and three-digit industry levels. The latter two techniques correct for the fact that unobserved
components are likely to be correlated with tenure, total labour market experience, and total experience
in the industry.
Zangelidis (2008) is one of the first empirical papers looking at the specificity of work experience,
drawing from data from the BHPS for the period 1991 to 2001. This paper finds that occupational
experience makes a positive contribution to wage growth, with not much evidence of industry effects,
in contrast to earlier evidence. Zangelidis (2008) finds these positive returns vary across occupations,
and that longevity in occupations is important. This finding is consistent with evidence based on the
US; when occupational experience is taken into account, tenure within an industry or employer has
relatively little importance in accounting for the wage one receives (Kambourov and Manovskii, 2009).
Kambourov and Manovskii (2009) find that five years of occupational tenure are associated with an
increase in wages ranging between 12 and 20 per cent. These findings are consistent with studies
supporting the idea that human capital is largely occupation-specific.
For the US, Sullivan (2010) shows that workers accumulate skills that are specific to both occupations
and industries, while truly firm specific skills contribute little to the growth of wages over the career.
Sullivan (2010) uses an IV approach35 to estimate the effects of firm tenure, occupation specific work
experience, industry specific work experience, and general work experience on wages, using data from
the 1979 cohort of the NLSY (for a sample of young men). These estimates indicate that both occupation
and industry-specific human capital are key determinants of wages but there is variability. Human
capital appears to be primarily occupation-specific in occupations such as craftsmen, where workers
benefit from a 14 per cent increase in wages after five years of occupation-specific experience; these
workers do not seem to benefit from industry-specific experience. In contrast, human capital appears to
be primarily industry specific in other occupations such as managerial employment where workers
realise a 23 per cent wage increase after five years of industry-specific work experience. In other
occupations, such as professional employment, both occupation and industry specific human capital are
chief determinants of wages.
Drawing form high-quality German administrative data, Fitzenberger et al. (2015) distinguishes the
wage effects of job mobility from those of occupational mobility. This study provides causal estimates
34 Tenure is instrumented with its deviations from job-match means whereas experience is instrumented with its
deviations from individual means. Total experience in the industry is instrumented with its deviations from
industry-match means. The instruments for tenure and experience in the industry are, by construction, uncorrelated
with their respective match quality components, while the instrument for experience is, also by construction,
uncorrelated with the individual component. 35 Similar to Parent (2000)
84
of the wage effects of mobility among graduates from apprenticeship in Germany.36 They use the IAB37
Employment Sample regional file from 1975 to 2004, a two per cent random sample of all employees
paying social security taxes. The analysis distinguishes between pure firm-switchers, within-firm
occupation switchers, and across-firm occupation switchers. To identify the causal effect of mobility
after apprenticeship on wages, they exploit variation in the local labour market conditions in the year
of graduation. The instruments utilised include both push and pull factors, such as indicators regarding
tightness of the local labour markets, and group specific mobility rates (which is assumed provide
exogenous variation in mobility conditional on the sorting of apprentices by two-digit training
occupations, which they account for by including occupation fixed effects). Due to the likely presence
of selection based on unobservables, OLS estimates are likely to be biased and they employ an IV
approach. The IV results suggest that pure firm changes after apprenticeship lead to wage losses, but
the conclusions regarding the wage effects of occupational mobility after apprenticeships are somewhat
more positive. The results of the study suggest that the skills acquired through apprenticeship training
in a specific occupation are sufficiently general to be useful when working in another occupation.
Using data from the BHPS, Dorsett et al. (2011) find that lifelong learning appears to provide a one-off
boost to wages growth for those in stable employment (the study refers only to women). It also
influences the probability of being in work and thereby indirectly increases earnings for movers. These
results are robust to controlling for the selection in terms of the characteristics of people who undertake
lifelong learning. They find that those qualifications, which result in women’s educational status being
upgraded, have a clear positive effect on their earnings.
3.6. On-the-job training
Compared to many EU countries, the UK ranks high in the proportion of workers receiving on-the-job
training (O’Mahony, 2012). While during the period 2000-2007 the proportion of workers receiving
training in the UK was comparable to that in Scandinavian countries and the Netherlands, Green et al.
(2016) has identified for the UK a downward trend in training volumes over the period 1997-2012.
Squicciarini et al. (2015) present a novel methodology for the measurement of investment in human
capital that is related to training. They use an expenditure approach that encompasses investment both
in formal and on-the-job training, as well as in informal learning, and estimate how much this broader
measure of training impacts on a country’s GDP. In order to arrive at these estimates, they link the
incidence of training to PIAAC data and other sources of information. This paper finds that during the
period 2011-2012, the UK has invested more in training than many OECD countries, to a level
36 In Germany, vocational training in an apprenticeship involves a job in the training firm and training in a specific
occupation. 37 Institut für Arbeitsmarkt- und Berufsforschung der Bundesagentur für Arbeit (Germany)
85
comparable to that in countries like Australia, Denmark, the Netherlands and Canada. It is worth
pointing out that the observed cross-country heterogeneity in training intensity is mainly driven by
differences in formal training, while the differences in on-the-job training are less pronounced.
Squicciarini et al. estimate that in the UK formal training represents six per cent of total GDP, on-the
job-training represents less than three per cent, and the contribution of informal training is below one
per cent of total GDP. Arguably, the investment in formal training in countries like the UK (and also
countries such as US, Australia and Canada), where private education is more common and tuition fees
are generally higher, the figures for formal training may be overestimated as the cost per student is
comparatively higher.
There is a large literature showing that the accumulation of human capital through the formal education
system plays a crucial role in determining productivity and earnings of workers (Moretti, 2004), but
also that more educated individuals have a higher probability of receiving training (Blundell et al.,
1996). Overall there is not much evidence on the effects of training provided by firms. Blundell et al.
uses data from the NCDS (a continuous longitudinal survey for people living in Britain with detailed
information on individual family background and labour market experience) for the period 1981 to 1991
to look at the wage effects of training on both men and women at age 33. On the returns to training, the
paper finds that being involved in employer-provided training courses confer significant wage
advantage for men, but interestingly this is more pronounced if this takes place off-the job. Also, the
returns are higher if the training results in a high-level vocational qualification. A surprising finding of
this study is that the training provided by a previous employer yields comparatively marginal smaller
returns. For women, employer-provided training results in smaller returns than for men, but a relatively
larger impact is found for those women undertaking courses that lead to qualifications. The results
show, moreover, that the returns to training are complementary to formal qualifications but are higher
for those with intermediate-level qualifications (which is recognised as one of the groups that
traditionally has received less training).
A number of papers use observational (panel) data, usually on industries or firms, to investigate the
causal impact of firm-provided training on productivity and wages. Dearden et al. (2006) examine for
the UK the effects of work-related training on direct measures of productivity during the period 1983-
1996. Dearden et al. assemble a dataset that aggregates individual-level data on training and
establishment data on productivity and investment into an industry panel, and control for unobserved
heterogeneity and the potential endogeneity of training using a generalised method moments estimation.
They find that there is a significant impact of training on labour productivity for UK industries
(measured as value added per hour) and, second, that the effects of training on productivity are larger
than the effects of training on wages. They argue that the focus on wages as the only relevant measure
86
of productivity, that many studies take, ignores additional productivity benefits for firms. They estimate
that one percentage point increase in training is associated with an increase in value added per hour of
about 0.6 per cent, and an increase in hourly wages of about 0.3 per cent. The authors argue however
than the broken link between wages and productivity may be related to imperfect competition prevailing
in the labour market as employees may find themselves be paid less than their marginal revenue product.
There are other more recent contributions that estimate the returns of employer-provided training. Using
company accounts Belgian data, Konings and Vanormelingen (2015) also find that, after controlling for
the possible endogeneity of training, that training boosts marginal productivity of an employee more
than it increases its wage. They estimate that the productivity premium for a trained worker is estimated
at 23 per cent, while the wage premium of training is estimated to be 12 per cent. There exists
considerable sector heterogeneity in the impact of training on both productivity and wages, and they
estimate a slightly higher impact of training in non-manufacturing compared to manufacturing sectors.
Among different manufacturing sectors, the largest productivity gains can be found in the chemicals
and rubber and plastic industries. This type of firm-level analysis is possible because Belgian firms are
obliged by law to submit a supplement to their annual accounts statement, which contains information
on a number of elements related to training. These include the proportion of workers that receive
training, the number of hours they are trained for, and the cost of training to the firm, and these data can
then be linked to both wages and productivity at the firm level. They estimate a production function
using an estimation strategy that controls for the endogeneity of input factors.
Using unique matched establishment-employee data from Germany, Berg et al. (2017) focus on the role
of training for the retention of workers, in particular of older workers. This study is based on the linked
German IAB Establishment Panel, an annual nationally representative survey of German
establishments, with individual-level administrative data for the employees of participating
establishments. They find that when establishments offer special training programs targeted at older
workers, women, and especially lower wage women, are less likely to go into retirement. The research
provides evidence that training programs are likely to play an important role in retaining and advancing
careers in particular of low wage older women. Results suggest this relationship may be related to
greater wage growth. For men, the findings suggest that the inclusion in training programs may improve
retention particularly of low wage men, but the analysis of pre-existing differences in establishment
retirement patterns suggests this relationship may not be causal.
87
4. Conclusion
It is of major interest to policy makers to understand what drives a country’s human capital
accumulation. This literature review makes a contribution to this goal by providing an accessible and
comprehensive literature review on (1) the factors that affect human capital accumulation and their
measurement and (2) the determinants of earnings, and their use in valuing human capital.
This review has outlined a number of factors which impact on education and earnings, such as family
background, cognitive and non-cognitive skills and health, many of which are themselves affected by
education and earnings. What emerges is a sense that countries might be able to get on a virtuous cycle
of human capital accumulation through better education and training, which increases earnings and
impacts future generations allowing them to enjoy better health, greater job mobility in the face of major
technological changes which in turn raises productivity and living standards, financing more education
and so on. On the other hand, there are many factors which might prevent such an occurrence, such as
institutional constraints, cultural conventions and market restrictions.
Much of the literature has focused on identifying causal impacts of individual factors such as family
background on education, leading to considerable improvements in our understanding. Nevertheless,
finding convincing identification strategies are often difficult, and these are frequently based on
experiments which may not be easily generalizable. The use of large scale administrative data offers
great potential, but access to such data is only available in a few countries. Far less effort has been
devoted to understand how these various factors impact on each other, or interact with the institutional
environment, primarily due to lack of good data, but also due to the persistent nature of institutions and
culture.
For example, the review of the literature on the role of family background as a determinant of human
capital highlighted a lack of detailed and rich data for the UK, comparable to Pronzato’s (2012)
Norwegian study and other similar studies on causal effects (using twins, siblings and adopted samples)
of parental education. It is also important to understand how education impacts children’s development
along the distribution of parental education. Parents, usually mothers with lower levels of educational
attainment, often aim to invest more in their children’s education but of course, challenges such as lack
of resources and poverty sometimes makes this highly unlikely. This potentially leads to a cycle of
generational poverty. Further research is therefore needed to understand some of the causal
mechanisms. Similarly, there is dearth of evidence on family background and children’s cognitive skills
development. The issue of cognitive and non-cognitive development has also received large attention
88
by governments and policy-makers. There are concerns that despite higher levels of educational
attainment, a considerable proportion of the UK population have low literacy and numeracy skills.
Family background again becomes crucial for addressing this issue. It is therefore important for future
studies to examine the extent to which family background plays a role, but of course, such an exercise
will require rich data that captures not only numeracy and literacy but also on other soft skills, including
personality traits.
A growing body of literature suggests that education has benefits beyond increases in labour market
productivity which extend to health. The literature review has highlighted that the majority of UK based
studies exploit the increase in the minimum school-leaving age in 1947 and/or 1973 as an exogenous
change in the amount of schooling received to examine the causal effect of education on health
conditions (e.g. biomarkers), subjective general health and health behaviour, such as smoking. Albeit
the common strategy, scholars draw different conclusions and find mixed evidence for the causal effect
of education on health (Oreopoulos, 2006; Blanchflower and Oswald, 2008; Silles, 2009; Powdthavee,
2010; Devereux and Hart, 2010; Clark and Royer, 2013; Jürges et al., 2013). Although evidence for the
opposite direction of the relationship suggests a health selection process which generates socioeconomic
inequalities in child- and young adulthood, the large majority of work is carried out in the US and
mainly relies on standard logit regression models. When looking at the context of labour supply, the
majority of UK based evidence comes from the BHPS and the Understanding Society Survey and the
indirect role of informal care provision associated with third parties’ health rather than from the role of
individual health (Michaud et al., 2010; Carmichael and Ercolani, 2016; Carr et al., 2016). The same
data sources are used in studies on the impact of earnings on health, which mainly use the minimum
wage legislation implemented in 1999 as an instrument (Kronenber et al., 2017; Lenhart, 2017; Reeves
et al., 2017). Here administrative data that links health records to levels of education and earnings would
be particularly useful, but face difficult issues relating to data security.
We have investigated the extent to which job mobility and worker turnover is associated with a loss of
human capital, as early theories emphasised the importance of firm- and or industry specific capital.
With the rise of the digital economy the degree of worker mobility is on the rise, and the degree of
transferability of human capital is an important factor in evaluating the costs of job displacement and
the re-training requirements of workers. Our review of the most recent literature has provided a few
interesting insights. First, a strand of literature has stressed that a major component of human capital is
occupation-specific or task-specific. Nevertheless, some contributions highlight that firm-specific
human capital can arise from the different uses that firms make of the general and transferrable pool of
skills available in the market. A number of studies estimate that the transition costs across occupations
can be substantial, and argue that these depend on the degree of dissimilarity and the mix of
89
requirements needed to perform a job. But an important factor, which is often overlooked, refers to the
occupation-specific entry cost such as qualification credentials and professional training, which are
independent of task content. The study of transferability of certain types of human capital has been
facilitated by the availability of high-quality data on complete job histories and wages of workers,
containing information on tasks performed in a number of occupations, in countries like Germany. In
contrast, most of the evidence on occupational mobility for the UK uses panel information from survey
data, mainly the BHPS or the LFS.
The degree of occupational mobility (not just changing jobs within an occupation) appears to be larger
in the UK than in many European countries. In theory the process of changing occupation should not
necessarily be associated with loss of human capital, as to some extent occupational change can be the
result of natural career progression. Evidence for the UK, however, suggests that occupational change
in the UK mostly occurs as a response to a job mismatch, and occurs less as a natural career progression
due to promotion (at least compared to other countries like Germany). Occupational change in the UK
is also more likely for the low-skilled than the high-skilled workers. Other recent evidence shows that
occupational change tends to be enhanced in the case of individuals that enjoy certain financial security,
who want to explore an alternative life choice. This is observed also for individuals at later stages of
their career, even if in general mobility decreases with age. There is evidence of a deterioration in the
deployment of skills after an occupation change, with better results in terms of job satisfaction. We have
also explored evidence on whether occupational movement is associated with lower earnings. Research
using UK data suggests that workers who change occupation mostly benefit from the change in terms
of wage increases and to a larger extent in terms of job satisfaction. This evidence is contradictory to
some of the evidence available for other countries like the US that finds that mobility is detrimental for
wage growth. Evidence for the UK suggests that the effects may not be uniform across the wage
distribution with the most positive wage effects for those workers at the top of the wage distribution.
Again the most recent literature suggests that occupation-specific experience matters the most for wage
growth, compared to industry-specific or firm-specific experience alone. Some of these recent studies
implement IV approaches to derive causal estimates of mobility on wages. The use of administrative
data provides a rich and reliable source of information to distinguish the effects on wages of pure firm-
switches, within-firm occupation switches, and across-firm occupation switches. This type of studies is
emerging for countries such as Germany.
Evidence for the UK suggests that regional variation in human capital stocks may originate from
training and education of resident populations, but is also highly affected by spatial flows, that is, the
migration of the highly educated. Highly educated individuals are found to exhibit highest levels of
spatial mobility (Faggian et al., 2015), which contribute to the process of knowledge transfer. IV-based
90
results show that while the innovativeness of a region is one of the major factors encouraging university
graduates to seek employment in that region, the large inflows of highly mobile university graduates
also contributes to enhance the innovation performance of a region. The majority of UK-based studies
are based on regional data. Thus, they are not able to follow the spatial trajectories of individuals, as is
possible using administrative sources, such as recent empirical studies for the Nordic countries. These
types of data allow a much richer analysis of patterns and behavioural aspects of interregional migration.
Our review of the existing literature on crime and its effect on human capital has found only a few
studies based on the UK, with the majority of studies undertaken for other countries. Most studies use
a common methodology by exploiting school age reforms to instrument educational effects on crime.
The review highlights the role of early educational intervention in reducing criminal activity. Although
different features of crime incidence affect human capital, the most widely studied area involves
education which reveals that human capital increases the opportunity costs of crime from foregone work
and expected costs with incarceration (Lochner, 2004; Lochner and Moretti, 2004). It is important to
note that this area of research is mainly carried out using US and Scandinavian data. The few UK based
exceptions use the increase in the minimum schooling age in 1997 as an identification strategy (Sabates
and Feinstein, 2008; Machin et al., 2011; 2012). The review has further highlighted the lack of empirical
evidence for the opposite direction of criminal activity on educational outcomes which is mainly due to
the difficulty of identifying exogenous variations in criminal involvement.
The review on returns to education looked also at the importance of institutional quality, with most
studies indicating that high quality education institutions impact positively on individual earnings. This
is a matter of potential interest to government, policy makers, and also the higher education institutions
themselves. Further studies on how to improve the quality of low performing universities could be
beneficial for future generations, through improved learning and higher labour market rewards. This of
course requires use of detailed data. As for the issue of subject degrees, the review highlighted stark
differences in earnings for different degree programmes. In terms of intergenerational transmission of
inequality, more robust evidence is needed for the UK, and there is the need to use rich administrative
data, such as the Longitudinal Education Outcomes dataset. A further key consideration will be to
acquire data on parental income, not just parental earnings. The use of parental income will be able to
capture other sources of income, such as transfer payments, all of which adds to wealth.
Econometrically, there is also the need to use techniques that address how people with unreported
income and earnings are dealt with.
Recent literature highlights the importance of skills gained through learning and on-the-job training as
the true earnings determinant. Empirical evidence suggests that cognitive skills, such as numeracy
91
(Hanushek et al., 2015), problem-solving and ICT skills (Falck et al., 2016), are key in explaining most
of the variation in earnings differences in technology-rich environments. At the same time, the increase
in automation also highlighted the financial returns to non-cognitive skills, such as social skills and
personality traits (e.g. Heckman et al., 2006b; Lindqvist and Vestman, 2011; Heckman and Kautz, 2012;
Deming and Kahn, 2018) and their complementarity with cognitive skills (Weinberger, 2014; Deming,
2017; Deming and Kahn, 2018). However, while information on cognitive skills is easily available in
many surveys (NCDS for the UK), access to data that captures information on non-cognitive skills are
limited which explains part of the lack of empirical examination of the role of non-cognitive skills in
the labour market. Existing work on the UK exploits information on the Big Five traits in the BHPS
(Nandi and Nicoletti, 2009; Heineck, 2011) and more recent work uses PIAAC in cross-country
comparisons to fill this gap. A very recent literature strand highlights the role of social skills and their
complementarity with cognitive skills, but they are all US based (Weinberger, 2014; Deming, 2017;
Deming and Kahn, 2018).
The empirical literature exploring the returns of employer-provided training, in terms of wages and
productivity, is not large. Earlier studies focused on individual wage returns (Blundell et al., 1996) and
subsequently the literature stressed the importance of looking at overall productivity effects of firms.
Several studies found significantly larger productivity effects of training than those on wages. This
‘wedge’ between wages and productivity pointed to a number of reasons including the existence of
important externalities of training at the firm level, and imperfect competition in the labour market.
Many early studies faced important data and methodological limitations as training was often measured
at single points in time, and thus could be picking up many unobservable firm-specific factors correlated
with both training and productivity. The empirical literature attempted to estimate causal effects by
relying on panel data of measures of training and firm performance, although aggregated to industries.
Dearden et al. (2006) was the first study combining the use of a production function with a wage
equation to estimate the effects of training. In this type of industry-level study one would pick up
spillovers to training that are taking place within an industry; these effects would not be observed in
firm-level analyses, which would potentially miss out on these linkages, underestimating the return to
human capital. But there could be important aggregation biases that would lead to potentially conflicting
results. A number of recent papers (for other countries than the UK mainly) draw from information of
performance and training measures collected at the firm level. Literature is emerging using matched
employer-employee data that enables linking firm outcomes to individual characteristics and behaviour.
Overall, the review points to the need to exploit large administrative databases where available. This
requires a concerted effort by those gathering the data, the ONS and research users to address concerns
92
relating to data security, but also requires significant resources in matching datasets and ensuring they
are fit for research purposes.
5. References
Abowd, J. M., Kramarz, F. and Margolis, D. N. (1999). High wage workers and high wage
firms. Econometrica, 67(2), pp.251-333.
Acemoglu, D. and Restrepo, P. (2017). Robots and Jobs: Evidence from US Labor Markets. NBER
Working Paper No. 23285.
Acemoglu, D. and Restrepo, P. (2018). Artifical Intelligence, Automation and Work. NBER Working
Paper No. 24196.
Adda, J., Banks, J. and von Gaudecker, H.M. (2009). The impact of income shocks on health: evidence
from cohort data. Journal of the European Economic Association, 7(6), pp.1361-1399.
Agee, M.D. and Crocker, T.D. (2002). Parents’ discount rate and the intergenerational transmission of
cognitive skills. Economica, 69(273), pp.143-154.
Ahlroth, S., Björklund, A. and Forslund, A. (1997). The Output of the Swedish Education Sector.
Review of Income and Wealth, 43(1), pp.89–104.
Ai, X. (1999). Creativity and Academic Achievement: An Investigation of Gender Differences.
Creativity Research Journal, 12(4), 329-337.Dingledine, R. (2003). Creativity: Environment and
Genetic factors.
Aizer, A. and Cunha, F. (2012). The production of human capital: Endowments, investments and
fertility, NBER Working Paper No. w18429.
Aizer, A. and Doyle, J. J. (2015). Juvenile incarceration, human capital, and future crime: Evidence
from randomly assigned judges. Quarterly Journal of Economics, 130(2): pp.759 – 803.
Alhola, P. and Polo-Kantola, P., (2007). Sleep deprivation: Impact on cognitive
performance. Neuropsychiatric disease and treatment.
Allen, T.D., Herst, D.E., Bruck, C.S. and Sutton, M., (2000). Consequences associated with work-to-
family conflict: A review and agenda for future research. Journal of Occupational Health
Psychology, 5, pp. 278-308.
Almlund, M., Duckworth, A., Heckman, J. and Kautz, T. (2011). Personality Psychology and
Economics.
Al-Qaisy, L.M. (2011). The relation of depression and anxiety in academic achievement among group
of university students. International journal of psychology and counselling, 3(5), pp.96-100.
Altonji, J. G., E. Blom and C. Meghir (2012). Heterogeneity in human capital investments: high school
curriculum, college major and careers. Annual Review of Economics, 4(1), pp.185–223.
Alvord,M. K., & Grados, J. J. (2005). Enhancing resilience in children: A proactive approach.
Professional Psychology: Research and Practice, 36(3), 238–245, http://dx.doi.org/10.1037/0735-
7028.36.3.238.
Amin, V., Lundborg, P. and Rooth, D. O. (2015). The intergenerational transmission of schooling: Are
mothers really less important than fathers?. Economics of Education Review, 47(1), pp.100-117.
Anders, J., D., Henderson, M., Moulton, V. and Sullivan, A. (2017). Socio-economic status and subject
choice at 14: do they interact to affect university access. London: Nuffield Foundation.
93
Andersen, R.M., (1995). Revisiting the behavioral model and access to medical care: does it
matter?. Journal of health and social behavior, pp.1-10.
Anderson, D. M. (2014). In school and out of trouble? The minimum dropout age and juvenile
crime. Review of Economics and Statistics, 96(2), pp.318-331.
Andrews, B. and Wilding, J.M. (2004). The relation of depression and anxiety to life‐stress and
achievement in students. British journal of psychology, 95(4), pp.509-521.
Andrieu, E., Jamet, S., Marcolin, L. and Squicciarini, M. (2018). Occupational transitions: the cost of
moving to a “safe haven”. OECD Working Paper No. DSTI/CIIE/WPIA(2018)3
Anger, S. (2011). The Intergenerational Transmission of Cognitive and Non-Cognitive Skills during
Adolescence and Young Adulthood. IZA Discussion Paper 5749.
Anger, S. and Heineck, G. (2010). Do smart parents raise smart children? The intergenerational
transmission of cognitive abilities. Journal of Population Economics, 23(3), pp.1105-1132.
Anger, S. and Schnitzlein, D. (2017). Cognitive skills, non-cognitive skills and family background:
evidence from sibling correlations. Journal of Population Economics 30(2), pp.591-620.
Angrist, J. D. and Krueger, A. B. (1991). Does compulsory school attendance affect schooling and
earnings?. The Quarterly Journal of Economics, 106(4), pp.979-1014.
Angrist, J., Lavy, V. and Schlosser, A. (2010). Multiple experiments for the causal link between the
quantity and quality of children. Journal of Labor Economics, 28(4), pp.773-824.
Angrist, N., Djankov, S., Goldberg, P. K. and Patrinos, H. A. (2019). Measuring Human Capital. World
Bank Policy Research Working Paper No. 8742.
Antonio, A.L. (2004). The influence of friendship groups on intellectual self-confidence and
educational aspirations in college. The Journal of Higher Education, 75(4), pp.446-471.
Apouey, B. and Clark, A. E. (2015). Winning big but feeling no better? The effect of lottery prizes on
physical and mental health. Health Economics, 24(5), pp.516–538.
Appleby, P.N. and Key, T.J., (2016). The long-term health of vegetarians and vegans. Proceedings of
the Nutrition Society, 75(3), pp.287-293.
Arens, A.K. and Morin, A.J. (2016). Relations between teachers’ emotional exhaustion and students’
educational outcomes. Journal of Educational Psychology, 108(6), p.800.
Arntz, M., Gregory, T. and Zierahn, U. (2017). Revisiting the risk of automation. Economics
Letters, 159(1), pp.157-160.
Arteaga, C. (2018). The effect of human capital on earnings: Evidence from a reform at Colombia's top
university. Journal of Public Economics, 157, pp.212-225.
Artuç, E., Chaudhuri, S. and McLaren, J. (2010). Trade shocks and labor adjustment: A structural
empirical approach. American Economic Review, 100(3), pp.1008-45.
Ascherio, A., Rimm, E.B., Giovannucci, E.L., Spiegelman, D., Meir, S. and Willett, W.C., (1996).
Dietary fat and risk of coronary heart disease in men: cohort follow up study in the United
States. Bmj, 313(7049), pp.84-90.
Ásgeirsdóttir, T.L., Corman, H., Noonan, K., Ólafsdóttir, Þ. and Reichman, N.E. (2014). Was the
economic crisis of 2008 good for Icelanders? Impact on health behaviors. Economics and Human
Biology, 13, pp.1-19.
94
Atkinson, A., Burgess, S., Croxson, B., Gregg, P., Propper, C., Slater, H. and Wilson, D. (2009).
Evaluating the impact of performance-related pay for teachers in England. Labour
Economics, 16(3), pp.251-261.
Atkinson, T. (2005). Atkinson Review: Final Report: Measurement of Government Output and
Productivity for the National Accounts. Basingstoke: Palgrave MacMillan
Autor, D. H., Levy, F. and Murnane, R. J. (2003). The skill content of recent technological change: An
empirical exploration. The Quarterly Journal of Economics, 118(4), pp.1279-1333.
Babitsch, B., Gohl, D. and von Lengerke, T., (2012). Re-revisiting Andersen’s Behavioral Model of
Health Services Use: a systematic review of studies from 1998–2011. GMS Psycho-Social-
Medicine, 9.
Bachmann, R., Bechara, P. and Vonnahme, C. (2017) Occupational mobility in Europe: Extent,
determinants and consequences. Ruhr Economic Papers 732, RWI - Leibniz-Institut für
Wirtschaftsforschung, Ruhr-University Bochum, TU Dortmund University, University of
Duisburg-Essen
Bandura, A. (1997). Self-efficacy: The exercise of control. New York: W.H. Freeman and Company
Barro, R. J. (1991). A cross-country study of growth, saving and government. In: National saving and
economic performance (pp. 271-304) University of Chicago Press.
Barron, J.M., Ewing, B.T. and Waddell, G.R., (2000). The effects of high school athletic participation
on education and labor market outcomes. Review of Economics and Statistics, 82(3), pp.409-421.
Bartel, A. and Taubman, P. (1979). Health and labor market success: The role of various diseases. The
Review of Economics and Statistics, 61(1), pp.1-8.
Bassi, A., Dodd, S., Williamson, P. and Bodger, K., (2004). Cost of illness of inflammatory bowel
disease in the UK: a single centre retrospective study. Gut, 53(10), pp.1471-1478.
Bauer, J.M. and Sousa-Poza, A. (2015). Impacts of informal caregiving on caregiver employment,
health and family. Journal of Population Ageing, 8(3), pp.113-145.
Bechichi, N., Grundke, R., Jamet, S. and Squicciarini, M. (2018). Moving between jobs: An analysis of
occupation distances and skill needs. OECD Working Paper No. 52.
Becker, G. S. (1962). Investment in human capital: A theoretical analysis. Journal of political
economy, 70(5), pp.9-49.
Becker, G. S. (1964). Human Capital: A Theoretical and Empirical Analysis with Special Reference to
Education. New York: National Bureau of Economic Research.
Becker, G. S. (2007). Health as human capital: synthesis and extensions. Oxford Economic
Papers, 59(3), pp.379-410.
Becker, G. S. and Chiswick, B. R. (1966). Education and the Distribution of Earnings. The American
Economic Review, 56(1/2), pp.358-369.
Becker, G. S. and Tomes, N. (1986). Human capital and the rise and fall of families. Journal of Labor
Economics, 4(3), pp.1-39.
Becker, G. S., Hubbard, W. H. and Murphy, K. M. (2010). The market for college graduates and the
worldwide boom in higher education of women. American Economic Review, 100(2), pp.229-33.
95
Begley, C.E. and Beghi, E., 2002. The economic cost of epilepsy: a review of the
literature. Epilepsia, 43, pp.3-9.
Behncke, S. (2012). How Do Shocks to Non-Cognitive Skills Affect Test Scores?. Annals of Economics
and Statistics, pp.155-173.
Belfield, C. and van der Erve, L. (2018). The impact of higher education on the living standards of
female graduates. IFS Working Paper No. W18/25.
Belfield, C. R., Nores, M., Barnett, S. and Schweinhart, L. (2006). The high/scope perry preschool
program cost–benefit analysis using data from the age-40 followup. Journal of Human
resources, 41(1), pp.162-190.
Belfield, C., Britton, J., Buscha, F., Dearden, L., Dickson, M., Van Der Erve, L., Sibieta, L., Vignoles,
A., Walker, I. and Zhu, Y. (2018). The relative labour market returns to different degrees. London:
DoE and IFS.
Belfield, C., Crawford, C., Greaves, E., Gregg, P. and Macmillan, L. (2017). Intergenerational income
persistence within families. London: IFS.
Bell, B., Bindler, A. and Machin, S. (2018). Crime scars: recessions and the making of career
criminals. Review of Economics and Statistics, 100(3), pp.392-404.
Bell, B., Costa, R. and Machin, S. (2016). Crime, compulsory schooling laws and education. Economics
of Education Review, 54(1), pp.214-226.
Bennett, P. (2018). The heterogeneous effects of education on crime: Evidence from Danish
administrative twin data. Labour Economics, 52(1), pp.160-177.
Benson, R., von Hippel, P. T., and Lynch, J. L. (2018). Does more education cause lower BMI, or do
lower-BMI individuals become more educated? Evidence from the National Longitudinal Survey
of Youth 1979. Social Science & Medicine, 211(1), pp.370-377.
Bender S, Bloom N, Card D, Van Reenen J and Wolter S (2016), ‘Management Practices, Workforce
Selection, and Productivity’, CEP Discussion Paper Number 1,416, March 2016
Berg, P. B., Hamman, M. K., Piszczek, M. M. and Ruhm, C. J. (2017). The relationship between
employer‐provided training and the retention of older workers: Evidence from
Germany. International Labour Review, 156(3-4), pp.495-523.
Berkowitz, M., Fenn, P. and Lambrinos, J. (1983). The optimal stock of health with endogenous wages:
application to partial disability compensation. Journal of Health Economics, 2(2), pp.139-147.
Bernard, B. (1993). Fostering resiliency in kids. Educational Leadership, 51(3), 44–48.
Bernard, B. (1995). Fostering resilience in children. Retrieved from ERIC database.
Björklund, A. and Jäntti, M. (1997). Intergenerational Income Mobility in Sweden Compared to the
United States. American Economic Review, 87(5), pp.1009–1018.
Björklund, A. and Jäntti, M. (2009). Intergenerational economic inequality, in W. Salverda, B. Nolan
and T. Smeeding (eds), The Oxford Handbook of Economic Inequality, Oxford: Oxford University
Press.
Björklund, A., K.H. Eriksson and Jäntti, M. (2010). IQ and family background: Are associations strong
or weak?. B.E. Journal of Economic Analysis and Policy, 10(1).
96
Björklund, A., Lindahl, M. and Plug, E. (2006). The origins of intergenerational associations: Lessons
from Swedish adoption data. The Quarterly Journal of Economics, 121(3), pp.999-1028.
Black D. and Smith, J. (2004). How robust is the evidence on the effects of college quality? Evidence
from matching, Journal of Econometrics, 121(1), pp.99-124.
Black D. and Smith, J. (2006). Estimating the Returns to College Quality with Multiple Proxies for
Quality, Journal of Labor Economics, 24(1), pp.701-729.
Black, C.M. (2008). Working for a healthier tomorrow: Dame Carol Black's review of the health of
Britain's working age population. The Stationery Office.
Black, S. E., Devereux,P. J. and Salvanes, K. (2010). Small Family, Smart Family? Family Size and the
IQ Scores of Young Men. The Journal of Human Resources, 45(1), pp. 33-58.
Black, S.E. and Devereux, P.J. (2010). Recent developments in intergenerational mobility. NBER
Working Paper No. w15889.
Black, S.E., Breining, S., Figlio, D.N., Guryan, J., Karbownik, K., Nielsen, H.S., Roth, J. and Simonsen,
M. (2017). Sibling spillovers. NBER Working Paper No. w23062.
Black, S.E., Devereux, P.J. and Salvanes, K.G. (2005). Why the apple doesn't fall far: Understanding
intergenerational transmission of human capital. American Economic Review, 95(1), pp.437-449.
Black, S.E., Devereux, P.J. and Salvanes, K.G. (2009). Like father, like son? A note on the
intergenerational transmission of IQ scores. Economics Letters, 105(1), pp.138-140.
Blanchflower, D.G. and Oswald, A.J. (2008). Hypertension and happiness across nations. Journal of
health economics, 27(2), pp.218-233.
Blanden, J. (2013). Cross‐country rankings in intergenerational mobility: a comparison of approaches
from economics and sociology. Journal of Economic Surveys, 27(1), pp.38-73.
Blanden, J. and Gregg, P. (2004). Family income and educational attainment: a review of approaches
and evidence for Britain. Oxford Review of Economic Policy, 20(2), pp.245-263.
Blanden, J., Goodman, A., Gregg, P. and Machin, S. (2004). Changes in intergenerational mobility in
Britain. Generational income mobility in North America and Europe, 122-46.
Blanden, J., Gregg, P. and Machin, S. (2005). Intergenerational mobility in Europe and North
America. Report supported by the Sutton Trust, Centre for Economic Performance, London School
of Economics.
Blanden, J., Gregg, P. and Macmillan, L. (2007). Accounting for intergenerational income persistence:
noncognitive skills, ability and education. The Economic Journal, 117(519), pp.43-60.
Bloom N, Brynjolfsson E, Foster L, Jarmin R, Patnaik M and Saporta-Eksten I (2017), ‘What Drives
Differences in Management?’, Centre for Economic Performance, Discussion Paper Number 1,470
Bloom N, Brynjolfsson E, Foster L, Jarmin R, Saporta-Eksten I and Van Reenen J (2013), ‘Management
in America’, Working Papers 13 to 01, Center for Economic Studies, US Census Bureau
Bloom, D.E., Canning, D. and Fink, G. (2010). Implications of population ageing for economic
growth. Oxford review of economic policy, 26(4), pp.583-612.
Blundell, R., Dearden, L. and Meghir, C. (1996). The Determinants of Work Related Training in Britain,
London: Institute of Fiscal Studies.
97
Booth, A.L. and Kee, H.J. (2009). Birth order matters: the effect of family size and birth order on
educational attainment. Journal of Population Economics, 22(2), pp.367-397.
Borghans, L., Meijers, F. & ter Weel, B. (2006). The role of noncognitive skills in explaining cognitive
test scores. Economic Inquiry, 46(1), 2-12.
Borghans, L., Ter Weel, B. and Weinberg, B. A. (2008). Interpersonal styles and labor market
outcomes. Journal of Human Resources, 43(4), pp.815-858.
Bottazzi, R., Jappelli, T. and Padula, M. (2006). Retirement expectations, pension reforms and their
impact on private wealth accumulation. Journal of Public Economics, 90(12), pp.2187-2212.
Bound, J., Schoenbaum, M., Stinebrickner, T. R. and Waidmann, T. (1999). The dynamic effects of
health on the labour force transitions of older workers. Labour Economics, 6(2), pp.179–202.
Bowles, S. and Gintis, H. (2000). Does schooling raise earnings by making people smarter?.
Meritocracy and Economic Inequality, p.118.
Boyatzis, R. (2008). Competencies in the 21st century. Journal of Management Development 27 (1), 5-
12.
Breining, S. (2014). The presence of ADHD: Spillovers between Siblings. Economics Letters, 124(3),
pp. 469–473.
Breining, S. N., Doyle, J. J., Figlio, D. N., Karbownik, K., and Roth, J. (2017). Birth order and
delinquency: Evidence from Denmark and Florida. NBER Working Paper No. w23038.
Breining, S., Daysal, N.M., Simonsen, M. and Trandafir, M. (2015). Spillover effects of early-life
medical interventions. IZA Working Paper No. 9086.
Brenøe, A. A. (2018). Origins of Gender Norms: Sibling Gender Composition and Women's Choice of
Occupation and Partner. IZA Discussion Paper No. 11692.
Britton, J., L. Dearden, N. Shephard and A. Vignoles (2016). How English domiciled graduate earnings
vary with gender, institution attended, subject and socio-economic background. IFS Working
Paper No. WP16/06.
Bronars, S. G. and Oettinger, G. S. (2006). Estimates of the return to schooling and ability: evidence
from sibling data. Labour Economics, 13(1), 19-34.
Brown, S., McIntosh, S. and Taylor, K. (2011). Following in your parents’ footsteps? Empirical analysis
of matched parent–offspring test scores. Oxford Bulletin of Economics and Statistics, 73(1), pp.40-
58.
Brown, S., Roberts, J. and Taylor, K. (2010). Reservation wages, labour market participation and
health. Journal of the Royal Statistical Society: Series A (Statistics in Society), 173(3), 501-529.
Brugård, K. H. and Falch, T. (2013). Post-compulsory education and imprisonment. Labour
Economics, 23(1), pp.97-106.
Brugiavini, A. (1997). Social security and retirement in Italy. NBER Working Paper No. 6155.
Broszeit S, Fritsch U, Görg H and Marie-Christine L (2016), ‘Management practices and productivity
in Germany’, IAB Discussion Paper Number 32/2016, Nuremberg: Institute for Employment
Research
98
Burhan, N. A. S., Yunus, M. M., Tovar, M. E. L. and Burhan, N. M. G. (2017). Why are cognitive
abilities of children so different across countries? The link between major socioeconomic factors
and PISA test scores. Personality and Individual Differences, 105, pp.95-106.
Burton, W. N., Conti, D. J., Chen, C. Y., Schultz, A. B. and Edington, D. W. (1999). The role of health
risk factors and disease on worker productivity. Journal of Occupational and Environmental
Medicine, 41(10), pp.863-877.
Buscha, F. and Dickson, M. (2015). The Wage Returns to Education over the Life-Cycle: Heterogeneity
and the Role of Experience. IZA Discussion Paper No. 9596.
Cade, J.E., Burley, V.J., Greenwood, D.C. and UK Women's Cohort Study Steering Group, (2004). The
UK Women's Cohort Study: comparison of vegetarians, fish-eaters and meat-eaters. Public health
nutrition, 7(7), pp.871-878.
Cai, L., Mavromaras, K. and Oguzoglu, U. (2014). The effects of health status and health shocks on
hours worked. Health Economics, 23(5), pp.516-528.
Cameron, S. V. and Heckman, J. J. (1993). The nonequivalence of high school equivalents. Journal of
Labor Economics, 11(1, Part 1), 1-47.
Cameron, S.V. and Heckman, J.J. (1998). Life cycle schooling and dynamic selection bias: Models and
evidence for five cohorts of American males. Journal of Political Economy, 106(2), pp.262-333.
Card, D. (2001). Estimating the return to schooling: Progress on some persistent econometric
problems. Econometrica, 69(5), pp.1127-1160.
Carers UK (2014).Facts about carers. Available at: https://www.carersuk.org/for-
professionals/policy/policy-library?task=download&file=policy_file&id=4762
Carmichael, F. and Ercolani, M.G. (2016). Unpaid caregiving and paid work over life-courses: Different
pathways, diverging outcomes. Social Science and Medicine, 156, pp.1-11.
Carmichael, F., Charles, S. and Hulme, C. (2010). Who will care? Employment participation and
willingness to supply informal care. Journal of Health Economics, 29(1), pp.182-190.
Carmichael, F., Hulme, C., Sheppard, S. and Connell, G. (2008). Work–life imbalance: Informal care
and paid employment in the UK. Feminist Economics, 14(2), 3-35.
Carneiro, P., Heckman, J.J. and Krueger, A.B. (2004). Inequality in America: What Role for Human
Capital Policies?. Human Capital Policy, pp.77-240.
Carneiro, P., Meghir, C. and Parey, M. (2013). Maternal education, home environments and the
development of children and adolescents. Journal of the European Economic Association, 11(1),
pp.123-160.
Carpenter, J.M. (2006). Effective teaching methods for large classes. Journal of Family & Consumer
Sciences Education, 24(2).
Carr, E., Murray, E. T., Zaninotto, P., Cadar, D., Head, J., Stansfeld, S. and Stafford, M. (2016). The
association between informal caregiving and exit from employment among older workers:
prospective findings from the UK Household Longitudinal Study. The Journals of Gerontology:
Series B, 73(7), pp.1253-1262.
Carrillo-Tudela, C., Hobijn, B., She, P. and Visschers, L. (2016). The extent and cyclicality of career
changes: Evidence for the UK. European Economic Review, 84(1), pp.18-41.
99
Carroll, J. M., Humphries, M., and Muller, C. (2018). Mental and physical health impairments at the
transition to college: Early patterns in the education-health gradient. Social science
research, 74(1), pp.120-131.
Case, A., Fertig, A. and Paxson, C. (2005). The lasting impact of childhood health and
circumstance. Journal of health economics, 24(2), pp.365-389.
Case, A., Lin, I. F. and McLanahan, S. (2000). How hungry is the selfish gene?. The Economic Journal,
110(466), pp.781-804.
Cawley, J., Heckman, J. and Vytlacil, E. (2001). Three observations on wages and measured cognitive
ability. Labour Economics, 8(4), pp.419-442.
CBS (Statistics Netherlands) (2011). Measuring Intergenerational Income Mobility, The Hague:
Statistics Netherlands.
Cesarini, D., Lindqvist, E., Östling, R. and Wallace, B. (2016). Wealth, health and child development:
Evidence from administrative data on Swedish lottery players. The Quarterly Journal of
Economics, 131(2).
Chadwick, L. and Solon, G. (2002). Intergenerational mobility among daughters. American Economic
Review, 92(1), pp. 335–344.
Checchi, D. (2006). The economics of education: Human capital, family background and inequality,
Cambridge University Press.
Cherniss, C., Goleman, D., Emmerling, R., Cowar, K. and M. Adler (1998). Bringing emotional
intelligence to the workplace. The Consortium for Research on Emotional Intelligence in
Organizations.
Chetty, R., Hendren, N., Kline, P. and Saez, E. (2014). Where is the land of opportunity? The geography
of intergenerational mobility in the United States. The Quarterly Journal of Economics, 129(4),
pp.1553-1623.
Chevalier, A. (2004). Parental Education and Child’s Education: A natural experiment. IZA Discussion
Paper No. 1153.
Chevalier, A. (2011). Subject choice and earning of UK graduates. Economics of Education Review,
30(6), 1187–1201.
Chevalier, A. (2014). Does higher education quality matter in the UK?. In Factors Affecting Worker
Well-being: The Impact of Change in the Labor Market (pp. 257-292). Emerald Group Publishing
Limited.
Chevalier, A. and Conlon, G. (2003). Does it pay to attend a prestigious university?. IZA Discussion
Paper No. 848.
Chevalier, A. and Lanot, G. (2002). The relative effect of family characteristics and financial situation
on educational achievement. Education Economics, 10(2), pp.165-181.
Chevalier, A., Harmon, C., O’Sullivan, V. and Walker, I. (2013). The impact of parental income and
education on the schooling of their children. IZA Journal of Labor Economics, 2(1), 8.
Chowdry, H., Crawford, C., Dearden, L., Goodman, A. and Vignoles, A. (2013). Widening participation
in higher education: analysis using linked administrative data. Journal of the Royal Statistical
Society: Series A (Statistics in Society), 176(2), pp.431-457.
100
Christian, M. (2010). Human Capital Accounting in the United States, 1994–2006. Survey of Current
Business, 90(6), pp.31–36.
Clark, C., Crombie, R., Head, J., Van Kamp, I., Van Kempen, E. and Stansfeld, S.A. (2012). Does
traffic-related air pollution explain associations of aircraft and road traffic noise exposure on
children's health and cognition? A secondary analysis of the United Kingdom sample from the
RANCH project. American journal of epidemiology, 176(4), pp.327-337.
Clark, C., Martin, R., Van Kempen, E., Alfred, T., Head, J., Davies, H.W., Haines, M.M., Barrio, I.L.,
Matheson, M. and Stansfeld, S.A. (2005). Exposure-effect relations between aircraft and road
traffic noise exposure at school and reading comprehension: the RANCH project. American
journal of epidemiology, 163(1), pp.27-37.
Clark, D. and Royer, H. (2013). The effect of education on adult mortality and health: Evidence from
Britain. American Economic Review, 103(6), pp.2087-2120.
Clotfelter, C.T., Ladd, H.F. and Vigdor, J.L., (2007). Teacher credentials and student achievement:
Longitudinal analysis with student fixed effects. Economics of Education Review, 26(6), pp.673-
682.
Cobb-Clark, D. A. and Tan, M. (2011). Noncognitive skills, occupational attainment and relative
wages. Labour Economics, 18(1), pp.1-13.
Coelli, M. (2011). Parental job loss and the education enrollment of youth. Labour Economics, 18(1),
pp.25-35.
Colquitt, J. A., LePine, A., & Noe, R. A. (2000). Toward an integrative theory of training motivation:A
meta-analytic path analysis of 20 years of research. Journal of Applied Psychology, 85, 678–707.
doi:10.1037//0021-9010.85.5.678
Comunian, R., Faggian, A. and Jewell, S. (2016). Talent on the move: creative human capital migration
patterns in the UK. In Higher Education and the Creative Economy (pp. 134-154). Routledge.
Connolly, S. and Gregory, M. (2002). The national minimum wage and hours of work: implications for
low paid women. Oxford Bulletin of Economics and Statistics, 64, pp.607-631.
Connolly, S. and Gregory, M. (2008). Moving down: women's part‐time work and occupational change
in Britain 1991–2001. The Economic Journal, 118(526), pp.52-76.
Conti, G., Heckman, J. and Urzua, S. (2010). The education-health gradient. American Economic
Review, 100(2), pp.234-38.
Contoyannis, P. and Rice, N. (2001). The impact of health on wages: evidence from the British
Household Panel Survey. Empirical Economics, 26(4), pp.599-622.
Cools, A. and Patacchini, E. (2017). Sibling Gender Composition and Women's Wages. IZA Discussion
Paper No. 11001.
Coppola, M. and Wilke, C. B. (2014). At What Age Do You Expect to Retire? Retirement Expectations
and Increases in the Statutory Retirement Age. Fiscal Studies, 35(2), pp.165-188.
Corak, M. (2006). Do poor children become poor adults? Lessons from a cross-country comparison of
generational earnings mobility. IZA Discussion Paper no. 1993
Corak, M. (2013). Income inequality, equality of opportunity and intergenerational mobility. Journal
of Economic Perspectives, 27(3), pp.79-102.
101
Corak, M. and Heisz, A. (1999). The intergenerational earnings and income mobility of Canadian men.
Journal of Human Resources, 34(1), pp.504-533.
Cortes, G. M. and Gallipoli, G. (2017). The costs of occupational mobility: An aggregate
analysis. Journal of the European Economic Association, 16(2), pp.275-315.
Covay, E. and Carbonaro, W. (2010). After the bell: Participation in extracurricular activities, classroom
behavior, and academic achievement. Sociology of Education, 83(1), pp.20-45.
Covay, E. and Carbonaro, W. (2010). After the bell: Participation in extracurricular activities, classroom
behavior, and academic achievement. Sociology of Education, 83(1), pp.20-45.
Coyle, T. R., & Pillow, D. R. (2008). SAT and ACT predict college GPA after removing g. Intelligence,
36(6), 719-729.
Crawford, C. and Cribb, J. (2015). The link between childhood reading skills and adult outcomes:
analysis of a cohort of British children, IFS Briefing Note BN169.
Cullen, J., B. Jacob and Levitt, S. (2006). The Effect of School Choice on Participants: Evidence from
Randomized Lotteries. Econometrica, 74(1): 1191–1230.
Cunha, F. and Heckman, J. (2007). The technology of skill formation. American Economic
Review, 97(2), pp.31-47.
Cunha, F. and Heckman, J.J. (2008). Formulating, identifying and estimating the technology of
cognitive and noncognitive skill formation. Journal of human resources, 43(4), pp.738-782.
Cunha, F., Heckman, J. J. and Schennach, S. M. (2010). Estimating the technology of cognitive and
noncognitive skill formation. Econometrica, 78(3), pp.883-931.
Currie, A., Shields, M.A. and Price, S.W. (2007). The child health/family income gradient: Evidence
from England. Journal of health economics, 26(2), pp.213-232.
Currie, J. and Almond, D., 2011. Human capital development before age five. In Handbook of labor
economics (Vol. 4, pp. 1315-1486). Elsevier.
Cutler, D. and Lleras-Muney, A. (2008). Education and Health: Evaluating Theories and Evidence. In
Making Americans Healthier: Social and Economic Policy as Health Policy, edited by J. House,
R. Schoeni, G. Kaplan, and H. Pollack. New York: Russell Sage Foundation.
D'Addio, A. (2007). Intergenerational Transmission of Disadvantage: Mobility or Immobility Across
Generations?. OECD Social, Employment and Migration Working Paper No. 52.
Dale, S. B. and Krueger, A. B. (2002). Estimating the payoff to attending a more selective college: An
application of selection on observables and unobservables. The Quarterly Journal of
Economics, 117(4), pp.1491-1527.
Darling-Hammond, L., (2000). Teacher quality and student achievement. Education policy analysis
archives, 8, p.1.
Dawson, A. E., Bernstein, B. L. & Bekki, J. M. (2015). Providing the psychosocial benefits of mentoring
to women in STEM: CareerWISE as an online solution. New Directions for Higher Education,
171, 53–62. doi:10.1002/he.20142
Day, H. I. (1982). Curiosity and the interested explorer. Performance and Instruction, 21, 19-22.
de Coulon, A., Meschi, E. and Vignoles, A. (2011). Parents' skills and children's cognitive and non-
cognitive outcomes. Education economics, 19(5), pp.451-474.
102
Dearden, L. (1998). Ability, family, education and earnings in Britain. Institute of Fiscal Studies
Working Paper, (14).
Dearden, L., Machin, S. and Reed, H. (1997). Intergenerational mobility in Britain. The Economic
Journal, 107(440), pp.47-66.
Dearden, L., Machin, S. and Reed, H. (1997). Intergenerational mobility in Britain. The Economic
Journal, 107(440), pp.47-66.
Dearden, L., Machin, S. and Reed, H. (1997). Intergenerational mobility in Britain. The Economic
Journal, 107(440), pp.47-66.
Dearden, L., Reed, H. and Van Reenen, J. (2006). The impact of training on productivity and wages:
Evidence from British panel data. Oxford bulletin of economics and statistics, 68(4), pp.397-421.
Deci, E. L., & Ryan, R. M. (1985). Intrinsic motivation and self-determination in human behavior. New
York: Plenum.
Delaney, J. M. and Devereux, P. J. (2019). More education, less volatility? The effect of education on
earnings volatility over the life cycle. Journal of Labor Economics, 37(1), pp.101-137.
Delaney, L., Harmon, C. and Redmond, C. (2011). Parental education, grade attainment and earnings
expectations among university students. Economics of Education Review, 30(6), pp.1136-1152.
Deming, D. and Kahn, L. B. (2018). Skill requirements across firms and labor markets: Evidence from
job postings for professionals. Journal of Labor Economics, 36(1), pp.337-369.
Deming, D. J. (2011). Better schools, less crime?. The Quarterly Journal of Economics, 126(4),
pp.2063-2115.
Deming, D. J. (2017). The growing importance of social skills in the labor market. The Quarterly
Journal of Economics, 132(4), pp.1593-1640.
Derzon, J. H. (2010). The correspondence of family features with problem, aggressive, criminal and
violent behavior: A meta-analysis. Journal of Experimental Criminology, 6(3), pp.263-292.
Desforges, C. and Abouchaar, A. (2003). The impact of parental involvement, parental support and
family education on pupil achievement and adjustment: A literature review (Vol. 433). London:
DfES.
Devereux, P. J. and Fan, W. (2011). Earnings returns to the British education expansion. Economics of
Education Review, 30(6), pp.1153-1166.
Devereux, P. J. and Hart, R. A. (2010). Forced to be rich? Returns to compulsory schooling in
Britain. The Economic Journal, 120(549), pp.1345-1364.
Dex, S., and Smith, C., (2001) Employee Commitment as an Outcome of Family-Friendly Policies?
Analysis of the Workplace Employee Relations Survey, Research Papers in Management Studies,
WP20/2001, Judge Institute of Management, University of Cambridge.
Dex, S. and Smith, C. (2002) The Nature and Pattern of Family-friendly Employment Policies in
Britain. Bristol: Policy Press.
DFE. (2019). Outcomes for children looked after by local authorities in England, 31 March 2018.
Dickens, R. and Manning, A. (2004). Has the national minimum wage reduced UK wage
inequality?. Journal of the Royal Statistical Society: Series A (Statistics in Society), 167(4),
pp.613-626.
103
Dickson, M. and Harmon, C. (2011). Economic returns to education: What we know, what we don’t
know and where we are going—some brief pointers. Economics of Education Review, 30(6),
pp.1118-1122.
Dickson, M. and Smith, S. (2011). What determines the return to education: an extra year or a hurdle
cleared?. Economics of Education Review, 30(6), pp.1167-1176.
Dickson, M., Gregg, P. and Robinson, H. (2016). Early, late or never? When does parental education
impact child outcomes?. The Economic Journal, 126(596), pp.184-231.
Disney, R., Emmerson, C. and Wakefield, M. (2006). Ill health and retirement in Britain: A panel data-
based analysis. Journal of Health Economics, 25(4), pp.621-649.
Dix‐Carneiro, R. (2014). Trade liberalization and labor market dynamics. Econometrica, 82(3), pp.825-
885.
Dolton, P. and Sandi, M. (2017). Returning to returns: revisiting the British education evidence. Labour
Economics, 48(1), pp.87-104.
Dorsett, R., Lui, S. and Weale, M. (2011). Estimating the effect of lifelong learning on women’s
earnings using a switching model. LLAKES Research Paper No.30.
Draca, M. and Machin, S. (2015). Crime and economic incentives. Economics, 7(1), pp.389-408.
Drummond, S.P., Brown, G.G., Gillin, J.C., Stricker, J.L., Wong, E.C. and Buxton, R.B., (2000).
Altered brain response to verbal learning following sleep deprivation. Nature, 403(6770), p.655.
Duflo, E. (2001). Schooling and labor market consequences of school construction in Indonesia:
Evidence from an unusual policy experiment. American Economic Review, 91(4), pp.795-813.
Duncan, G. and Dunifon, R. (1998) Soft-Skills and Long-Run Labor Market Success. Research in
Labor Economics, 123-49.
Duncan, G.J. and Magnuson, K.A. (2005). Can family socioeconomic resources account for racial and
ethnic test score gaps?. The future of children, 15(1), pp.35-54.
Duwe, G. and Clark, V. (2014). The effects of prison-based educational programming on recidivism
and employment. The Prison Journal, 94(4), pp.454-478.
Ehrlich, I. (1975). The deterrent effect of capital punishment: a question of life and death. American
Economic Review, 65(3), pp.397–417.
Eide, E.R. and Showalter, M.H., (2012). Sleep and student achievement. Eastern Economic
Journal, 38(4), pp.512-524.
Ellis, L., Beaver, K. M. and Wright, J. (2009). Handbook of crime correlates. Academic Press.
Ermisch, J. and Francesconi, M. (2000). Educational choice, families and young people's
earnings. Journal of Human Resources, 35(1), pp.143-176.
Ermisch, J. and Francesconi, M. (2004). Intergenerational mobility in Britain: new evidence, in M.
Corak (ed.), Generational Income Mobility in North America and Europe, Cambridge University
Press, Cambridge.
Ermisch, J. and Peter, F. H. and Spiess, C. K.(2012). 'Early childhood outcomes and family structure.'
In: Ermisch, John and Jäntti, Markus and Smeeding, Timothy M., (eds.) From parents to children:
the intergenerational transmission of advantage. New York: Russell Sage Foundation.
104
Ermisch, J. and Pronzato, C. (2011). Causal effects of parents’ education on children’s education.” In:
Persistence, Privilege and Parenting (Tim Smeeding, Robert Erikson and Markus Jantti). New
York: Russell Sage Foundation.
Ermisch, J., Francesconi, M. and Pevalin, D.J. (2001). Outcomes for children of poverty (No. 158).
Leeds: Corporate Document Services.
Espinoza, H. and Speckesser, S. (2019). A comparison of earnings related to higher level
vocational/technical and academic education. NIESR Discussion Paper No. 502.
Evans J M. (2001) Firms’ Contribution to the Reconciliation between Work and Family Life. OECD
Occasional Papers No. 48. Labour Market and Social Policy. OECD, Paris.
Evans, W. N. and Garthwaite, C. L. (2014). Giving mom a break: The impact of higher EITC payments
on maternal health. American Economic Journal: Economic Policy, 6(2), pp.258-90.
Faggian, A. and McCann, P. (2009). Human capital, graduate migration and innovation in British
regions. Cambridge Journal of Economics, 33(2), pp.317-333.
Faggian, A., J. Corcoran and Partridge, M. (2015). Interregional Migration Analysis. In: C. Karlsson,
M. Andersson and T. Norman, eds., Handbook in the Research of Methods and Applications in
Economic Geography (pp. 468–490). Cheltenham: Edward Elgar Publishing.
Falck, O., A. Heimisch and Wiederhold, S. (2016). Returns to ICT Skills. OECD Education Working
Paper No. 134.
Farrington, D. (1996). Later life outcomes of truants in the Cambridge Study. In I. Berg and J. Nursten
(Eds.), Unwillingly to school (pp. 96-118). London, England: Gaskell/Royal College of
Psychiatrists.
Farrington, D. P., Coid, J. W., and West, D. J. (2009). The development of offending from age 8 to age
50: Recent results from the Cambridge Study in Delinquent Development. Monatsschrift für
Kriminologie und Strafrechtsreform, 92(2-3), pp.160-173.
Farrington, D. P., Coid, J. W., Harnett, L., Jolliffe, D., Soteriou, N., Turner, R., and West, D. J.
(2006). Criminal careers up to age 50 and life success up to age 48: New findings from the
Cambridge Study in Delinquent Development (Vol. 94). London, UK: Home Office Research,
Development and Statistics Directorate.
Felfe, C. 2012. “The motherhood wage gap: What about job amenities?”, in Labour Economics, Vol.
19, No. 1, pp. 59–67.Amuedo-Dorantes, Catalina & Kimmel, Jean. (2008). New Evidence on the
Motherhood Wage Gap.
Fischer, J. and Lipovská, H. (2013). How Does the Parents’ Attained Level of Education Influence
Lifelong Learning of Children. Efficiency and Responsibility in Education, pp.128-135.
Fitzenberger, B., Licklederer, S. and Zwiener, H. (2015). Mobility across firms and occupations among
graduates from apprenticeship. Labour Economics, 34(1), pp.138-151.
Flynn, R. J., Tessier, N. G., & Coulombe, D. (2013). Placement, protective and risk factors in the
educational success of young people in care: crosssectional and longitudinal analyses, European
Journal of Social Work, 16(1), 70-87.
Fougère, D., Kramarz, F. and Pouget, J. (2009). Youth unemployment and crime in France. Journal of
the European Economic Association, 7(5), pp.909-938.
105
Freeman, S., Eddy, S.L., McDonough, M., Smith, M.K., Okoroafor, N., Jordt, H. and Wenderoth, M.P.
(2014). Active learning increases student performance in science, engineering, and
mathematics. Proceedings of the National Academy of Sciences, 111(23), pp.8410-8415.
Frey, C. and Osborne, M. (2017). The future of employment: How susceptible are jobs to
computerisation?. Technological Forecasting and Social Change, 114(C), pp.254-280.
Friedman, E.M. and Mare, R.D. (2014). The schooling of offspring and the survival of parents.
Demography, 51(4), pp.1271–1293.
Gadie, A., Shafto, M., Leng, Y. and Kievit, R.A., (2017). How are age-related differences in sleep
quality associated with health outcomes? An epidemiological investigation in a UK cohort of 2406
adults. BMJ open, 7(7), p.e014920.
Gagliardi, L. (2015). Does skilled migration foster innovative performance? Evidence from British local
areas. Papers in Regional Science, 94(4), pp.773-794.
Galindo-Rueda, Fernando & Vignoles, Anna & Jenkins, Andrew & Wolf, Alison. (2003). The
Determinants and Labour Market Effects of Lifelong Learning. Applied Economics. 35. 1711-
1721. 10.1080/0003684032000155445.
García-Gómez, P., Jones, A. M. and Rice, N. (2010). Health effects on labour market exits and entries.
Labour Economics, 17(1), pp.62-76.
Gardner, J. and Oswald, A. J. (2007). Money and mental wellbeing: A longitudinal study of medium-
sized lottery wins. Journal of Health Economics, 26(1), pp.49–60.
Gathmann, C. and Schonberg, U. (2010). How general is human capital? A task-based approach.
Journal of Labor Economics, 28(1), pp.1-49.
Gennaioli, N., R. La Porta, F. Lopez-De-Silanes and Shleifer, A. (2013). Human capital and regional
development. The Quarterly Journal of Economics, 128(1), pp.105–164.
Ginther, D. K., and Pollak, R. A. (2004). Family structure and children’s educational outcomes: Blended
families, stylized facts, and descriptive regressions. Demography, 41(4), pp.671-696.
Glass, J. and Estes, S. (1997) The Family Responsive Workplace. Annual Review of Sociology, 23,
289-313.
Goetzel, R. Z., Hawkins, K., Ozminkowski, R. J. and Wang, S. (2003). The health and productivity cost
burden of the “top 10” physical and mental health conditions affecting six large US employers in
1999. Journal of occupational and environmental medicine, 45(1), pp. 5-14.
Goleman, D. (1998). The emotional intelligence of leaders. Leader to Leader, 1998(10), pp.20-26.
Goleman, D. (2000). Leadership that gets results, Harvard Business Review MarchApril, 2-17.
Gould, E. D., Weinberg, B. A. and Mustard, D. B. (2002). Crime rates and local labor market
opportunities in the United States: 1979–1997. Review of Economics and statistics, 84(1), pp.45-
61.
Gould, E., Simhon, A. and Weinberg, B.A. (2019). Does parental quality matter? Evidence on the
transmission of human capital using variation in parental influence from death, divorce and family
size, NBER Working Paper No. w25495.
Green, D. A. and Riddell, W. C. (2003). Literacy and earnings: an investigation of the interaction of
cognitive and unobserved skills in earnings generation. Labour Economics, 10(2), 165-184.
106
Green, F., Felstead, A., Gallie, D., Inanc, H. and Jewson, N. (2016). The declining volume of workers’
training in Britain. British Journal of Industrial Relations, 54(2), pp.422-448.
Greenhaus, J.H. and Beutell, N.J., (1985). Sources of conflict between work and family roles. Academy
of management review, 10(1), pp.76-88.
Gregg, P., Kelly, G., Dowden, J., Kendrick, F., O'Grady, F., Robb, A., Unwin, J., Watt, J. and Wickham,
R. (2016). Closing the Gap:: A Living Wage that means families don't go short. APR Working
Paper.
Grenet, J. (2009). Is It Enough to Increase Compulsory Education to Raise Earnings? Evidence from
French and British Compulsory Schooling Laws. CEP Working Paper.
Grenet, J. (2013). Is extending compulsory schooling alone enough to raise earnings? Evidence from
French and British compulsory schooling laws. The Scandinavian Journal of Economics, 115(1),
pp.176-210.
Grönqvist, E., Öckert, B. and Vlachos, J. (2017). The intergenerational transmission of cognitive and
noncognitive abilities. Journal of Human Resources, 52(4), pp.887-918.
Grönqvist, H. (2013). Youth unemployment and crime: Lessons from longitudinal population
records. Swedish Institute for Social Research, mimeo.
Grossman, M. (1972a). The Demand for Health: A Theoretical and Empirical Investigation. New York:
Columbia University Press
Grossman, M. (1972b). On the Concept of Health Capital and the Demand for Health. Journal of
Political Economy, 80(2), pp.223-255.
Grossman, M. (2000). The human capital model. In Handbook of health economics (Vol. 1, pp. 347-
408). Elsevier.
Grossman, M. (2006). Education and nonmarket outcomes. Handbook of the Economics of
Education, 1, pp.577-633.
Grossman, M. and Benham, L. (1973). Health, hours and wages ‘, in: M. Perlman,(ed.) The Economics
of Health and Healthcare (New York).
Grundke, R., Marcolin, L. and Squicciarini, M. (2018). Which skills for the digital era?: Returns to
skills analysis. OECD Science, Technology and Industry Working Papers,2018(9), 0_1-37.
Gu, W. and Wong, A. (2010). Estimates of Human Capital in Canada: The Lifetime Income Approach.
Economic Analysis Division, Statistics Canada Catalogue no. 11F0027M, no. 062.
Gupta, R.D. and Guest, J.F., (2002). Annual cost of bipolar disorder to UK society. The British Journal
of Psychiatry, 180(3), pp.227-233.
Gurria, A. (2013) . PISA 2012 results in focus.
Gurria, A. (2016). PISA 2015 results in focus.
Guryan, J. (2004). Desegregation and Black Dropout Rates. American Economic Review, 94(1),
pp.919–943.
Haas, S. A. and Fosse, N. E. (2008). Health and the educational attainment of adolescents: Evidence
from the NLSY97. Journal of Health and Social Behavior, 49(2), pp.178-192.
107
Hafner, M., Stepanek, M., Taylor, J., Troxel, W.M. and van Stolk, C., (2017). Why sleep matters—the
economic costs of insufficient sleep: a cross-country comparative analysis. Rand health
quarterly, 6(4).
Haider, S. and Solon, G. (2006). Life-cycle variation in the association between current and lifetime
earnings. American Economic Review, 96(4), pp.1308–1320.
Hanushek, E. A. and Wössmann, L. (2012). Do Better Schools Lead to More Growth? Cognitive Skills,
Economic Outcomes and Causation. Journal of Economic Growth, 17(4), pp.267-321.
Hanushek, E. A., Schwerdt, G., Wiederhold, S. and Wössmann, L. (2015). Returns to skills around the
world: Evidence from PIAAC. European Economic Review, 73(1), pp.103-130.
Hanushek, E. A., Schwerdt, G., Wiederhold, S., and Wössmann, L. (2013). Returns to Skills around the
World: Evidence from PIAAC. NBER Working Paper No. w19762.
Hanushek, E.A., Kain, J.F. and Rivkin, S.G. (1999). Do higher salaries buy better teachers? (No.
w7082). National bureau of economic research.
Harmon, C. and Walker, I. (1995). Estimates of the economic return to schooling for the United
Kingdom. The American Economic Review, 85(5), pp.1278-1286.
Harmon, C., Oosterbeek, H. and Walker, I. (2003). The returns to education: Microeconomics. Journal
of economic surveys, 17(2), pp.115-156.
Harris, D.N. and Sass, T.R. (2011). Teacher training, teacher quality and student achievement. Journal
of public economics, 95(7-8), pp.798-812.
Hattie, J. (2015). The applicability of Visible Learning to higher education. Scholarship of Teaching
and Learning in Psychology, 1(1), 79.
Heckman, J. J. and Kautz, T. (2012). Hard evidence on soft skills. Labour Economics, 19(4), pp.451-
464.
Heckman, J. J. and Rubinstein, Y. (2001). The importance of noncognitive skills: Lessons from the
GED testing program. American Economic Review, 91(2), pp.145-149.
Heckman, J. J. and Urzua, S. (2010). Comparing IV with structural models: What simple IV can and
cannot identify. Journal of Econometrics, 156(1), pp.27-37.
Heckman, J. J. and Vytlacil, E. (2005). Structural equations, treatment effects and econometric policy
evaluation 1. Econometrica, 73(3), pp.669-738.
Heckman, J. J., Humphries, J. E. and Veramendi, G. (2018). Returns to education: The causal effects
of education on earnings, health and smoking. Journal of Political Economy, 126(1), pp.197-246.
Heckman, J. J., Humphries, J. E., and Veramendi, G. (2016). Dynamic treatment effects. Journal of
Econometrics, 191(2), pp.276-292.
Heckman, J. J., Lochner, L. J. and Todd, P. E. (2006a). Earnings functions, rates of return and treatment
effects: The Mincer equation and beyond. Handbook of the Economics of Education, 1, pp.307-
458.
Heckman, J. J., Stixrud, J. and Urzua, S. (2006b). The effects of cognitive and noncognitive abilities on
labor market outcomes and social behavior. Journal of Labor Economics, 24(3), pp.411-482.
108
Heckman, J., Pinto, R. and Savelyev, P. (2013). Understanding the mechanisms through which an
influential early childhood program boosted adult outcomes. American Economic Review, 103(6),
pp.2052-86.
Heineck, G. (2011). Does it pay to be nice? Personality and earnings in the United Kingdom. ILR
Review, 64(5), pp.1020-1038.
Heitmueller, A. (2007). The chicken or the egg?: Endogeneity in labour market participation of informal
carers in England. Journal of Health Economics, 26(3), pp.536-559.
Hjalmarsson, R. (2008). Criminal justice involvement and high school completion. Journal of Urban
Economics, 63(2), pp.613-630.
Hjalmarsson, R. and Lochner, L. (2012). The impact of education on crime: international
evidence. CESifo DICE Report, 10(2), pp.49-55.
Hjalmarsson, R., Holmlund, H. and Lindquist, M. J. (2015). The Effect of Education on Criminal
Convictions and Incarceration: Causal Evidence from Micro‐data. The Economic
Journal, 125(587), pp.1290-1326.
Ho, D.E., Imai, K., King, G. and Stuart, E.A., 2007. Matching as nonparametric preprocessing for
reducing model dependence in parametric causal inference. Political analysis, 15(3), pp.199-236.
Hoekstra, M. (2009). The effect of attending the flagship state university on earnings: A discontinuity-
based approach. The Review of Economics and Statistics, 91(4), pp.717-724
Holmes, J., Gear, E., Bottomley, J., Gillam, S., Murphy, M. and Williams, R., (2003). Do people with
type 2 diabetes and their carers lose income?(T2ARDIS-4). Health Policy, 64(3), pp.291-296.
Holmlund, H., Lindahl, M. and Plug, E. (2011). The causal effect of parents' schooling on children's
schooling: A comparison of estimation methods. Journal of Economic Literature, 49(3), pp.615-
51.
Hopkins, K. (2012). The pre-custody employment, training and education status of newly sentenced
prisoners. Results from the Surveying Prisoner Crime Reduction (SPCR) longitudinal cohort study
of prisoners. Ministry of Justice Analytical Services, Ministry of Justice Research Series.
Hoynes, H., Miller, D. and Simon, D. (2015). Income, the earned income tax credit and infant
health. American Economic Journal: Economic Policy, 7(1), pp.172-211.
Hoynes, H., Page, M. and Stevens, A. H. (2011). Can targeted transfers improve birth outcomes?:
Evidence from the introduction of the WIC program. Journal of Public Economics, 95(7-8),
pp.813-827.
Hughes, K.L. and Karp, M.J.M. (2004). School-based career development: A synthesis of the literature.
Hussain, I., McNally, S. and Telhaj, S. (2009). University quality and graduate wages in the UK. IZA
Discussion Paper No. 4043.
Huttunen, K., Møen, J., and Salvanes, K. G. (2018). Job loss and regional mobility. Journal of Labor
Economics, 36(2), pp.479-509.
Hyde, M. and Phillipson, C. (2014). How can lifelong learning, including continuous training within
the labour market, be enabled and who will pay for this? Looking forward to 2025 and 2040 how
might this evolve?. [online] Dera.ioe.ac.uk. Available at: https://dera.ioe.ac.uk/23660/1/gs-15-9-
future-ageing-lifelong-learning-er02.pdf [Accessed 2 Sep. 2019].
109
Imison, C., Sonola, L., Honeyman, M., Ross, S., (2014). The reconfiguration of clinical services: what
is the evidence?. The Kings Fund.
Jacob, B. A. and Lefgren, L. (2003). Are idle hands the devil's workshop? Incapacitation, concentration
and juvenile crime. American Economic Review, 93(5), pp.1560-1577.
Jäntti, M., Bratsberg, B., Røed, K., Raaum, O., Naylor, R., Österbacka, E., Björklund, A. and Eriksson,
T. (2006). American Exceptionalism in a New Light: A Comparison of Intergenerational Earnings
Mobility in the Nordic Countries, the United Kingdom and the United States. IZA Discussion Paper
No.1938.
Jerrim, J. (2017). The Link between Family Background and Later Lifetime Income: How Does the UK
Compare with Other Countries?. Fiscal Studies, 38(1), pp. 49–79.
John, O. P., Donahue, E. M., & Kentle, R. L. (1991). The Big Five Inventory - Versions 4a and 54.
Berkeley, CA: University of California, Berkeley, Institute of Personality and Social Research.
Jones, A. M., Rice, N. and Contoyannis, P. (2006b). The dynamics of health. Elgar Companion to
Health Economics, (Part I).
Jones, A.M., Rice, N. and Zantomio, F., 2016. Acute health shocks and labour market
outcomes. University Ca'Foscari of Venice, Dept. of Economics Research Paper Series, (9).
Jones, M. K., Latreille, P. L. and Sloane, P. J. (2006a). Disability, gender, and the British labour
market. Oxford Economic Papers, 58(3), pp.407-449.
Jones, R. and B. Chiripanhura (2010). Measuring the UK’s Human Capital Stock. Economic and Labor
Market Review, 4(11), pp.36–63.
Jorgenson, D. W. and Fraumeni, B. M. (1989). The accumulation of human and nonhuman capital,
1948–1984. In: Lipsey, R. E. and Tice, H. S. ed. The Measurement of Savings, Investment and
Wealth. Chicago, I.L.: The University of Chicago Press, pp. 227–286.
Jorgenson, D. W. and Fraumeni, B. M. (1992). The output of the education sector. In: Griliches, Z. ed.
Output Measurement in the Services Sector. Chicago, I.L.: The University of Chicago Press, pp.
303–338.
Jürges, H., Kruk, E. and Reinhold, S. (2013). The effect of compulsory schooling on health—evidence
from biomarkers. Journal of Population Economics, 26(2), pp.645-672.
Kahn, R.L., Wolfe, D.M., Quinn, R.P., Snoek, J.D. and Rosenthal, R.A., (1964). Organizational stress:
Studies in role conflict and ambiguity.
Kajonius, P. J. and Carlander, A. (2017). Who gets ahead in life? Personality traits and childhood
background in economic success. Journal of Economic Psychology, 59(1), pp,164-170.
Kalil, A., Mogstad, M., Rege, M. and Votruba, M.E. (2016). Father presence and the intergenerational
transmission of educational attainment. Journal of Human Resources, 51(4), pp.869-899.
Kambourov, G. and Manovskii, I. (2008). Rising occupational and industry mobility in the United
States: 1968–97. International Economic Review, 49(1), pp.41-79.
Kambourov, G. and Manovskii, I. (2009). Occupational specificity of human capital. International
Economic Review, 50(1), pp.63-115.
110
Kelly, C., Hulme, C., Farragher, T. and Clarke, G., (2016). Are differences in travel time or distance to
healthcare for adults in global north countries associated with an impact on health outcomes? A
systematic review. BMJ open, 6(11), p.e013059.
Kerkhofs, M. and Lindeboom, M. (1995). Subjective health measures and state dependent reporting
errors. Health Economics, 4(3), pp.221-235.
Key, T.J., Thorogood, M., Appleby, P.N. and Burr, M.L., (1996). Dietary habits and mortality in 11
000 vegetarians and health conscious people: results of a 17 year follow up. Bmj, 313(7060),
pp.775-779.
Kidd, M. P., Sloane, P. J. and Ferko, I. (2000). Disability and the labour market: an analysis of British
males. Journal of health economics, 19(6), pp.961-981.
Kirkeboen, L. J., E. Leuven and M. Mogstad (2016). Field of study, earnings and self-selection. The
Quarterly Journal of Economics, 131(3), pp.1057–1111.
Klem, A.M. and Connell, J.P. (2004). Relationships matter: Linking teacher support to student
engagement and achievement. Journal of school health, 74(7), pp.262-273.
Kling, J. R., Ludwig, J. and Katz, L. F. (2005). Neighborhood effects on crime for female and male
youth: Evidence from a randomized housing voucher experiment. The Quarterly Journal of
Economics, 120(1), pp.87-130.
Konings, J. and Vanormelingen, S. (2015). The impact of training on productivity and wages: firm-
level evidence. Review of Economics and Statistics, 97(2), pp.485-497.
Kooiman, N, Latte, J. and Bontje, M. (2018) Human Capital Migration: A Longitudinal Perspective.
Journal of Economic and Social Geography, 109(5), pp.644-660.
Korhonen, M., Luoma, I., Salmelin, R. and Tamminen, T. (2012). A longitudinal study of maternal
prenatal, postnatal and concurrent depressive symptoms and adolescent well-being. Journal of
affective disorders, 136(3), pp.680-692.
Kose, Samet. (2003). A Psychobiological Model of Temperament and Character: TCI. Yeni
Symposium. 41.
Kretchmer, N., Beard, J.L. and Carlson, S., (1996). The role of nutrition in the development of normal
cognition. The American journal of clinical nutrition, 63(6), pp.997S-1001S.
Kronenberg, C., Jacobs, R. and Zucchelli, E. (2017). The impact of the UK National Minimum Wage
on mental health. SSM-population health, 3, pp.749-755.
Krueger, A. B. (2018). Inequality, too much of a good thing. In The Inequality Reader (pp. 25-33).
Routledge.
Landersø, R., Nielsen, H. S. and Simonsen, M. (2016). School starting age and the crime‐age
profile. The Economic Journal, 127(602), pp.1096-1118.
Lazear, E. P. (2009). Firm-specific human capital: A skill-weights approach." Journal of Political
Economy, 117(5), pp.914-940.
Le, T., Gibson, J. and L. Oxley (2006).A Forward-Looking Measure of the Stock of Human Capital in
New Zealand. The Manchester School, 74 (5): 593–609
Lechner, M. and Downward, P.M., (2013). Heterogeneous sports participation and labour market
outcomes in England.
111
Lechner, M. and Sari, N., (2015). Labor market effects of sports and exercise: Evidence from Canadian
panel data. Labour Economics, 35, pp.1-15.
Lee, J.S. (2012). The effects of the teacher–student relationship and academic press on student
engagement and academic performance. International Journal of Educational Research, 53,
pp.330-340.
Lehmann, J.Y.K., Nuevo-Chiquero, A. and Vidal-Fernandez, M. (2018). The early origins of birth order
differences in children’s outcomes and parental behavior. Journal of Human Resources, 53(1),
pp.123-156.
Leitch, S. (2006). Prosperity for all in the global economy-world class skills. The Stationery Office.
Lenhart, O. (2017). Do higher minimum wages benefit health? Evidence from the UK. Journal of Policy
Analysis and Management, 36(4), pp.828-852.
Lenhart, O. (2019). The effects of health shocks on labor market outcomes: evidence from UK panel
data. The European Journal of Health Economics, 20(1), pp.83-98.
Leschied, A., Chiodo, D., Nowicki, E. and Rodger, S. (2008). Childhood predictors of adult criminality:
A meta-analysis drawn from the prospective longitudinal literature. Canadian Journal of
Criminology and Criminal Justice, 50(4), pp.435-467.
Lessof, C., Ross, A., Brind, R., Harding, C., Bell, C., & Kyriakopoulos, G. (2018). Understanding KS4
attainment and progress: evidence from LSYPE2. Department for Education.
Li, H., Liang, Y., Fraumeni, B., Liu, Z. and Wang, X. (2013). Human Capital in China. Review of
Income and Wealth, 59(2), pp.212-234.
Lievens, F., Ones, D. S., & Dilchert, S. (2009). Personality scale validities increase throughout medical
school. Journal of Applied Psychology, 94(6), 1514-1535.
Lindeboom, M. and Van Doorslaer, E. (2004). Cut-point shift and index shift in self-reported
health. Journal of Health Economics, 23(6), pp.1083-1099.
Lindley, J. and McIntosh, S. (2015). Growth in within graduate wage inequality: The role of subjects,
cognitive skill dispersion and occupational concentration. Labour Economics, 37(1), pp.101-111.
Lindley, J. K. (2018). Are There Unexplained Financial Rewards for the Snakes in Suits? A Labour
Market Analysis of the Dark Triad of Personality. British Journal of Industrial Relations, 56(4),
pp.770-797.
Lindqvist, E. and Vestman, R. (2011). The labor market returns to cognitive and noncognitive ability:
Evidence from the Swedish enlistment. American Economic Journal: Applied Economics, 3(1),
pp.101-28.
Lipovská, H. And Fischer, J. (2016). Gifted students and human capital accumulation. Journal on
Efficiency and Responsibility in Education and Science, 9(3), pp.60-69.
Litaker, D., Koroukian, S.M. and Love, T.E., (2005). Context and healthcare access: looking beyond
the individual. Medical care, pp.531-540.
Liu, G. and Greaker, M. (2009). Measuring the Stock of Human Capital for Norway: A Lifetime Labor
Income Approach. Statistics Norway Research No 2009/12.
Livingstone, D.W. (2001). Adults' Informal Learning: Definitions, Findings, Gaps and Future
Research.
112
Lochner, L. (2004). Education, work and crime: A human capital approach. International Economic
Review, 45(3), pp.811-843.
Lochner, L. and Moretti, E. (2004). The effect of education on crime: Evidence from prison inmates,
arrests and self-reports. American Economic Review, 94(1), pp.155-189.
Long M. (2008). College quality and early adult outcomes. Economics of Education Review, 27(1),
pp.588-602.
Longhi, S. and Brynin, M. (2010). Occupational change in Britain and Germany. Labour
Economics, 17(4), pp.655-666.
Lounsbury, J. W., Sundstrom, E., Loveland, J. M., & Gibson, L. W. (2003). Intelligence, "Big Five"
personality traits, and work drive as predictors of course grade. Personality and Individual
Differences, 35(6), 1231-1239.
Luallen, J. (2006). School's out… forever: A study of juvenile crime, at-risk youths and teacher
strikes. Journal of urban economics, 59(1), pp.75-103.
Lucas, R. E. (1988). On the mechanics of economic development. Journal of monetary economics,
22(1), pp.3-42.
Luft, H. S. (1975). The impact of poor health on earnings. The Review of Economics and Statistics,
57(1), pp.43-57.
Lundborg, P. and Majlesi, K. (2018). Intergenerational transmission of human capital: Is it a one-way
street?. Journal of health economics, 57(1), pp.206-220.
Lundborg, P., Nilsson, A. and Rooth, D.O. (2014). Parental education and offspring outcomes: evidence
from the Swedish compulsory School Reform. American Economic Journal: Applied
Economics, 6(1), pp.253-78.
Luppa, M., Heinrich, S., Angermeyer, M.C., König, H.H. and Riedel-Heller, S.G., 2007. Cost-of-illness
studies of depression: a systematic review. Journal of affective disorders, 98(1-2), pp.29-43.
Lynch, J.L. and von Hippel, P.T. (2016). An education gradient in health, a health gradient in education,
or a confounded gradient in both?. Social Science and Medicine, 154(1), pp.18-27.
Machin, S. and Meghir, C. (2004). Crime and economic incentives. Journal of Human
Resources, 39(4), pp.958-979.
Machin, S., Marie, O. and Vujić, S. (2012). Youth crime and education expansion. German Economic
Review, 13(4), pp.366-384.
Machin, S., O. Marie and Vujic, S. (2011). The Crime Reducing Effect of Education. Economic Journal,
121(1), pp.463–84.
Maniadakis, N. and Gray, A., (2000). The economic burden of back pain in the UK. Pain, 84(1), pp.95-
103.
Mankiw, G., Romer, D. and Weil, D. (1992). A contribution to the Empirics of Economic Growth.
Quarterly Journal of Economics, 107(2), pp.407-437.
Manoli, D. S. and Weber, A. (2016). The Effects of the Early Retirement Age on Retirement Decisions.
NBER Working Paper No. 22561.
Marshall, A. (1890). Principles of Economics, 8th ed. London: Mcmillan.
113
Maslach, C., (2003). Job burnout: New directions in research and intervention. Current Directions in
Psychological Science, 12, pp. 189-92.
Mastrogiacomo, M. (2004). Retirement, Expectations and Realizations: Essays on the Netherlands
and Italy. Tinbergen Institute Research Series No. 336.
Mazumder, B. (2005).Fortunate sons: new estimates of intergenerational mobility in the United States
using social security earnings data. The Review of Economics and Statistics, 87(1), pp.235–255.
McClelland, M., Acock, A., Piccinin, A., Rhea, S. and Stallings, M. (2013). Relations between
preschool attention span-persistence and age 25 educational outcomes. Early Childhood Research
Quarterly, 28(2), pp.314-324.
McCrae, R. R., and Costa Jr, P. T. (1997). Personality trait structure as a human universal. American
psychologist, 52(5), p.509.
McInerney, M., Mellor, J.M. and Nicholas, L.H. (2013). Recession depression: mental health effects of
the 2008 stock market crash. Journal of health economics, 32(6), pp.1090-1104.
McIntosh, S. and Vignoles, A. (2001). Measuring and assessing the impact of basic skills on labour
market outcomes. Oxford Economic Papers, 53(3), pp.453-481.
McLanahan, S. and Sandefur, G. (1994). Growing Up with a Single Parent. What Hurts, What Helps.
Cambridge: Harvard University Press.
McLanahan, S., Tach, L. and Schneider, D. (2013). The causal effects of father absence. Annual Review
of Sociology, 39(1), pp.399-427.
Meghir, C., Palme, M. and Schnabel, M. (2012). The effect of education policy on crime: an
intergenerational perspective. NBER Working Paper No. w18145).
Meredith, C.N. and Dwyer, J.T., 1991. Nutrition and exercise: effects on adolescent health. Annual
review of public health, 12(1), pp.309-333.
Michaud, P.C., Heitmueller, A. and Nazarov, Z. (2010). A dynamic analysis of informal care and
employment in England. Labour Economics, 17(3), pp.455-465.
Midtsundstad, T. (2019). A review of the research literature on adult learning and employability.
European Journal of Education.
Milligan, K. and Stabile, M. (2011). Do child tax benefits affect the well-being of children? Evidence
from Canadian child benefit expansions. American Economic Journal: Economic Policy, 3(3),
pp.175-205.
Mincer, J. (1958). Investment in human capital and personal income distribution. The journal of
political economy, 66(4), pp.281-302.
Mincer, J. (1974). Schooling, Experience and Earnings. Human Behavior and Social Institutions No.
2.
Ministry of Justice (2015). Justice Data Lab Re-offending Analysis: Prisoners Education Trust.
Available at:
https://assets.publishing.service.gov.uk/government/uploads/system/uploads/attachment_data/file
/459470/prisoners-education-trust-report.pdf
114
Ministry of Justice (2016). Prison Safety and Reform. Available at:
https://assets.publishing.service.gov.uk/government/uploads/system/uploads/attachment_data/file
/565014/cm-9350-prison-safety-and-reform-_web_.pdf
Mora, T. and Oreopoulos, P. (2011). Peer effects on high school aspirations: Evidence from a sample
of close and not-so-close friends. Economics of Education Review, 30(4), pp.575-581.
Moretti, E. (2004). Workers' education, spillovers and productivity: evidence from plant-level
production functions. American Economic Review, 94(3), pp.656-690.
Moretti, E. (2012). The New Geography of Jobs. New York: Houghton Mifflin Harcourt
Muffels, R. and Luijkx, R. (2008). Labour market mobility and employment security of male employees
in Europe:trade-off'orflexicurity'?. Work, Employment and Society, 22(2), pp.221-242.
Mulligan, C.B. (1999). Galton versus the human capital approach to inheritance. Journal of political
Economy, 107(6), pp.184-S224.
Murray, R. (2016). What is Happening to NHS Waiting Times? The King’s Fund. Available at: https
://www.kings fund.org/blog/2016/-2/nhs-waiti ng times.
Mushkin, S.J. (1962). Health as an Investment. Journal of political economy, 70(5), pp.129-157.
Nami, Yaghoob & Marsooli, Hossein & Ashouri, Maral. (2014). The Relationship Between Creativity
And Academic Achievement. Procedia - Social and Behavioral Sciences. 114. 36-39.
10.1016/j.sbspro.2013.12.652.
Nandi, A. and Nicoletti, C. (2009). Explaining personality pay gaps in the UK. ISER Working Paper
Series. No. 2009-22.
Narayan, A., Van der Weide, R., Cojocaru, A., Lakner, C., Redaelli, S., Gerszon Mahler, D.,
Ramasubbaiah, R. G. and Thewissen, S. (2018). Fair Progress?: Economic Mobility Across
Generations Around the World. The World Bank.
Neal, D. (1995).Industry-specific human capital: Evidence from displaced workers. Journal of labor
Economics, 13(4): pp.653-677.
NHS (2018). Benefits of exercise. Available at: https://www.nhs.uk/live-well/exercise/exercise-health-
benefits/
Nicoletti, C. and Rabe, B. (2017). The effect of school spending on student achievement: addressing
biases in value added models. Journal of the Royal Statistical Society: Series A (Statistics in
Society), 181(2), pp.487-515.
Nyhus, E. and Pons, E. (2005). The effects of personality on earnings. Journal of Economic Psychology,
26(3), pp.363-384.
O’Leary, N. C. and P. J. Sloane (2005). The return to a university education in Great Britain. National
Institute Economic Review, 193(1), pp.75–89.
O’Mahony, M. (2012). Human capital formation and continuous training: Evidence for EU
countries. Review of income and wealth, 58(3), pp.531-549.
O’Mahony, M. and Samek, L. (2019). The Impact of Health Status on Human Capital. The Wellness
Journal, Vol3., Global Perspectives.
OECD (1998). Human Capital Investment: An international Comparison, Paris: OECD Publishing.
115
OECD (2018). A Broken Social Elevator? How to Promote Social Mobility, Paris: OECD.
Ohayon, M.M., Carskadon, M.A., Guilleminault, C. and Vitiello, M.V., (2004). Meta-analysis of
quantitative sleep parameters from childhood to old age in healthy individuals: developing
normative sleep values across the human lifespan. Sleep, 27(7), pp.1255-1273.
Oishi, S. and Schimmack, U. (2010). Residential mobility, well-being, and mortality. Journal of
personality and social psychology, 98(6), p.980.
ONS (2018). Sickness absence in the UK labour market. Available at:
https://www.ons.gov.uk/file?uri=/employmentandlabourmarket/peopleinwork/employmentandem
ployeetypes/datasets/sicknessabsenceinthelabourmarket/2017/sicknessabsencetables.xlsx
ONS (2017), ‘Management practices and productivity among manufacturing businesses in Great
Britain: experimental estimates for 2015’, Office for National Statistics
Oreopoulos, P. (2006). Estimating average and local average treatment effects of education when
compulsory schooling laws really matter. American Economic Review, 96(1), pp.152-175.
Orpen, C. (1997), "The effects of formal mentoring on employee work motivation, organizational
commitment and job performance", The Learning Organization, Vol. 4 No. 2, pp. 53-60.
https://doi.org/10.1108/09696479710160906
Orpen, C. (1995). The effects of mentoring on employees’ career success. Journal of Social Psychology,
135, 667–668.
Ou, S.R. and Reynolds, A.J., (2008). Predictors of educational attainment in the Chicago Longitudinal
Study. School Psychology Quarterly, 23(2), p.199. Oishi, S. and Schimmack, U., 2010. Residential
mobility, well-being, and mortality. Journal of personality and social psychology, 98(6), p.980.
Owens, M., Stevenson, J., Hadwin, J.A. and Norgate, R. (2012). Anxiety and depression in academic
performance: An exploration of the mediating factors of worry and working memory. School
Psychology International, 33(4), pp.433-449.
Palaniappan, A. K. (2005). Creativity and Academic Achievement: A Malaysian Perspective. Shah
Alam: Karis Publications.
Palaniappan, A. K. (2007a). Creative Perception and Academic Achievement: Implications for
education in Malaysia. Kuala Lumpour: Inreach Edition.
Palloni, A. (2006). Reproducing inequalities: Luck, wallets, and the enduring effects of childhood
health. Demography, 43(4), pp.587-615.
Panos, G. A., Pouliakas, K. and Zangelidis, A. (2014). Multiple job holding, skill diversification and
mobility. Industrial Relations: A Journal of Economy and Society, 53(2), pp.223-272.
Papageorge, N. W. and Thom, K. (2018). Genes, education and labor market outcomes: evidence from
the health and retirement study. NBER Working Paper No. w25114.
Parent, D. (2000). Industry-specific capital and the wage profile: Evidence from the national
longitudinal survey of youth and the panel study of income dynamics. Journal of Labor
Economics, 18(2), pp.306-323.
Parker, J. D. A., Summerfeldt, L. J., Hogan, M. J., & Majeski, S. (2004). Emotional intelligence and
academic success: Examining the transition from high school to university. Personality and
Individual Differences, 36, 163–172.
116
Parrado, E., Caner, A. and Wolff, E. N. (2007). Occupational and industrial mobility in the United
States. Labour Economics, 14(3), pp.435-455.
Pavan, R. (2016). On the production of skills and the birth-order effect. Journal of Human
Resources, 51(3), pp.699-726.
Peeters, M.C.W., Montgomery, A.J., Bakker, A.B. and Schaufeli, W.B., (2005). Balancing work and
home: How job and home demands are related to burnout. International Journal of Stress
Management, 12, pp. 43-61.
Petrides, K. V., Frederickson, N., & Furnham, A. (2004). The role of trait emotional intelligence in
academic performance and deviant behavior at school. Personality and Individual Differences, 36,
277–293
Plug, E. (2004). Estimating the effect of mother's schooling on children's schooling using a sample of
adoptees. American Economic Review, 94(1), pp.358-368.
Poletaev, M. and Robinson, C. (2008). Human capital specificity: evidence from the Dictionary of
Occupational Titles and Displaced Worker Surveys, 1984–2000. Journal of Labor
Economics, 26(3), pp.387-420.
Powdthavee, N. (2010). Does education reduce the risk of hypertension? Estimating the biomarker
effect of compulsory schooling in England. Journal of Human Capital, 4(2), pp.173-202.
Pronzato, C. (2012). An examination of paternal and maternal intergenerational transmission of
schooling. Journal of Population Economics, 25(2), pp. 591–608.
Propper, C., Rigg, J. and Burgess, S. (2007). Child health: evidence on the roles of family income and
maternal mental health from a UK birth cohort. Health Economics, 16(11), pp.1245-1269.
Psacharopoulos, G. (1985). Returns to education: a further international update and
implications. Journal of Human Resources, 20(4), pp.583-604.
Psacharopoulos, G. (1989). Time trends of the returns to education: Cross-national
evidence. Economics of Education Review, 8(3), pp.225-231.
Psacharopoulos, G. (1994). Returns to investment in education: A global update. World
Development, 22(9), pp.1325-1343.
Psacharopoulos, G. and Patrinos, H. A. (2004). Returns to investment in education: a further
update. Education economics, 12(2), pp.111-134.
Psacharopoulos, G. and Patrinos, H. A. (2018). Returns to investment in education: a decennial review
of the global literature. Education Economics, 26(5), pp.445-458.
Public Health England (2017) Research and analysis, Chapter 5: inequality in health. Available at:
https://www.gov.uk/government/publications/health-profile-for-england/chapter-5-inequality-in-
health
Qureshi, J. (2018). Siblings, Teachers and Spillovers in Academic Achievement. Journal of Human
Resources, 53(1), pp. 272-297.
Rao, N. and Chatterjee, T. (2017). Sibling gender and wage differences. Applied Economics, 50(15),
pp.1725-1745.
117
Reeves, A., McKee, M., Mackenbach, J., Whitehead, M. and Stuckler, D. (2017). Introduction of a
national minimum wage reduced depressive symptoms in low‐wage workers: a quasi-natural
experiment in the UK. Health Economics, 26(5), pp.639-655.
Rice, J.K. (2010). The Impact of Teacher Experience: Examining the Evidence and Policy Implications.
Brief No. 11. National center for analysis of longitudinal data in education research.
Rice, N. E., Lang, I. A., Henley, W. and Melzer, D. (2011). Common health predictors of early
retirement: findings from the English Longitudinal Study of Ageing. Age and Ageing, 40(1), pp.54-
61.
Roberts, B.W., Kuncel, N.R., Shiner, R.L., Caspi, A. and Goldberg, L.R. (2007). The power of
personality: The comparative validity of personality traits, socioeconomic status, and cognitive
ability for predicting important life outcomes. Perspectives on Psychological Science, 2(4),
pp.313–345.
Rud, I., van Klaveren, C., Groot, W., and van den Brink, H. M. (2018). What drives the relationship
between early criminal involvement and school dropout?. Journal of Quantitative
Criminology, 34(1), pp.139-166.
Saavedra, J. E. (2008). The returns to college quality: A regression discontinuity analysis. Unpublished
Harvard Mimeo.
Sabates, R. and Duckworth, K. (2010). Maternal schooling and children’s relative inequalities in
developmental outcomes: evidence from the 1947 School Leaving Age Reform in Britain. Oxford
Review of Education, 36(4), pp. 445–61.
Sabates, R. and L. Feinstein (2008). Effects of Government Initiatives on Youth Crime. Oxford
Economic Papers, 60(1), pp.462–83.
Sacerdote, B. (2007). How large are the effects from changes in family environment? A study of Korean
American Adoptees. Quarterly Journal of Economics, 122(1), pp.119-57.
Sainsbury Centre for Mental Health (2017). Mental health at work: The business costs ten years on.
Available at: https://www.centreformentalhealth.org.uk/sites/default/files/2018-
09/CentreforMentalHealth_Mental_health_problems_in_the_workplace.pdf
A. Silles, Mary. (2007). Adult Education and Earnings: Evidence from Britain. Bulletin of Economic
Research. 59. 313-326. 10.1111/j.0307-3378.2007.00268.x.
Sammons, P., Toth, K., & Sylva, K. (2015). Pre-school and early home learning effects on A-level
outcomes. Department for Education.
Schaufeli, W.B., Langelaan, S., Bakker, A.B., Van Rhenen, W. and Van Doornen, L.J.P. (2006), “Do
burned-out and work-engaged employees differ in the functioning of the hypothalamic-pituitary-
adrenal axis?”, Scandinavian Journal of Work, Environment, and Health, Vol. 32, pp. 339-48.
Schaufeli W.B., Salanova. M, Gonzalez-Roma, V and Bakker A.B., (2002) ‘The measurement of
engagement and burnout and: A confirmative analytic approach’, Journal of Happiness Studies, 3,
2002, 71-92
Schnitker, S. A., & Emmons, R. A. (2007). Patience as a virtue: Religious and psychological
perspectives. Research in the Social Scientific Study of Religion, 18, 177-207.
Schuller, T, & Desjardins, R. (2011). Wider benefits of adult learning. In Rubenson, K. (ed.),
International Encyclopedia of Adult Learning and Education (pp. 294-298). Oxford: Elsevier.
118
Schultz, T. W. (1960). Capital formation by education. Journal of Political Economy, 68(6), pp.571-
583.
Schultz, T. W. (1961). Investment in human capital. American Economic Review, 51(1), pp.1-17.
Schunk, D. H., & Pajares, F. (2002). The development of academic self-efficacy. In A. Wigfield, & J.
S. Eccles (Eds.), Development of achievement motivation (pp. 15–31). San Diego, CA: Academic
Press
Sciandra, M., Sanbonmatsu, L., Duncan, G. J., Gennetian, L. A., Katz, L. F., Kessler, R. C., Kling, J.
F. and Ludwig, J. (2013). Long-term effects of the Moving to Opportunity residential mobility
experiment on crime and delinquency. Journal of Experimental Criminology, 9(4), pp.451-489.
Scullin, M.K. and Bliwise, D.L., (2015). Sleep, cognition, and normal aging: integrating a half century
of multidisciplinary research. Perspectives on Psychological Science, 10(1), pp.97-137.
Sebba, J., Berridge, D., Luke, N., Fletcher, J., Bell, K., Strand, S., Thomas, S., Sinclair, I. and
O’Higgins, A. (2015). The educational progress of looked after children in England: Linking care
and educational data. University of Oxford Department of Education/University of Bristol.
Seuring, T., Archangelidi, O. and Suhrcke, M., (2015). The economic costs of type 2 diabetes: a global
systematic review. Pharmacoeconomics, 33(8), pp.811-831.
Shiffrin, R. M. (1988). Attention. In R. C. Atkinson, R. J. Herrnstein, G, Lindzey, & R. D. Luce (Eds.),
Stevens' handbook of experimental psychology (2nd ed., Vol. 2, pp. 863-952). New York: Wiley.
Sibley, B.A. and Etnier, J.L., (2003). The relationship between physical activity and cognition in
children: a meta-analysis. Pediatric exercise science, 15(3), pp.243-256.
Sibley, L.M. and Weiner, J.P., (2011). An evaluation of access to health care services along the rural-
urban continuum in Canada. BMC health services research, 11(1), p.20.
Silles, M. (2009). The causal effect of education on health: Evidence from the United
Kingdom. Economics of Education review, 28(1), pp.122-128.
Silles, M. (2010). The intergenerational effects of parental schooling on the cognitive and non-cognitive
development of children. Economics of Education Review, 30(2), pp. 258–68.
Smith, A. (1776). An Inquiry into the Nature and Causes of the Wealth of Nations. 2nd. London: W.
Strahan and T. Cadell
Solon, G. (1992). Intergenerational income mobility in the United States. The American Economic
Review, 82(3), pp.393-408.
Solon, G. (2002). Cross-country differences in intergenerational earnings mobility. Journal of
Economic Perspectives, 16(3), pp.59-66.
Squicciarini, M., Marcolin, L. and Horvát, P. (2015). Estimating Cross-Country Investment in Training:
An Experimental Methodology Using PIAAC Data, OECD Science, Technology and Industry
Working Papers, 2015/09.
Steinmayr, R., & Spinath, B. (2009). The importance of motivation as a predictor of school
achievement. Learning and Individual Differences, 19(1), 80-90.
Stern, D. N. (1973). The interpersonal world of the child. New York: Basic Books.
Stern, S. (1995). Estimating family long-term care decisions in the presence of endogenous child
characteristics. Journal of Human Resources, 30(3), pp.551-580.
119
Stewart, E.B. (2008). School structural characteristics, student effort, peer associations, and parental
involvement: The influence of school-and individual-level factors on academic
achievement. Education and urban society, 40(2), pp.179-204.
Stewart, M.B. (2004). The employment effects of the national minimum wage. The Economic Journal,
114(494), pp.110–116.
Stiglitz, J., Sen, A. K. and Fitoussi, J. P. (2009). The measurement of economic performance and social
progress revisited: reflections and overview. Available at: https://hal-sciencespo.archives-
ouvertes.fr/hal-01069384/document
Suglia, S.F., Gryparis, A., Wright, R.O., Schwartz, J. and Wright, R.J. (2007). Association of black
carbon with cognition among children in a prospective birth cohort study. American journal of
epidemiology, 167(3), pp.280-286.
Sullivan, P. (2010). Empirical evidence on occupation and industry specific human capital. Labour
Economics, 17(3), pp.567-580.
Sullivan, S.E. and Bhagat, R.S., (1992). Organizational stress, job satisfaction and job performance:
where do we go from here?. Journal of management, 18(2), pp.353-374.
Sweeten, G. (2006). Who will graduate? Disruption of high school education by arrest and court
involvement. Justice Quarterly, 23(4), pp.462-480.
Tauchen, H., Witte, A. and Griesinger, H. (1994). Criminal Deterrence: Revisiting the Issue with a Birth
Cohort. The Review of Economics and Statistics, 76(3), pp. 399- 412.
Thiel, H. and Thomsen, S. L. (2013). Noncognitive skills in economics: Models, measurement and
empirical evidence. Research in Economics, 67(2), pp.189-214.
Thomas, L. T., & Ganster, D. C. (1995). Impact of family-supportive work variables on work-family
conflict and strain: A control perspective. Journal of Applied Psychology, 80(1), 6-15.
Tong, C., & Kram, K. E. (2013). The efficacy of mentoring – The benefits for mentees, mentors, and
organizations. In J. Passmore, D. B. Peterson, & T. Freire (Eds.), The Wiley-Blackwell handbook
of the psychology of coaching and mentoring (pp. 217–242). Oxford: Wiley
Topel, R. (1991). Specific capital, mobility and wages: Wages rise with job seniority. Journal of
political Economy, 99(1), pp.145-176.
Torssander, J. (2013). From child to parent? The significance of children’s educationfor their parents’
longevity. Demography, 50(2), pp.637–659.
Travis, D.J., Lizano, E.L. and Mor Barak, M.E., (2015). ‘I’m so stressed!’: A longitudinal model of
stress, burnout and engagement among social workers in child welfare settings. The British Journal
of Social Work, 46(4), pp. 20176-1095.
Turner, Charles F., and Daniel C. Martinez. “Socioeconomic Achievement and the Machiavellian
Personality.” Sociometry, vol. 40, no. 4, 1977, pp. 325–336. JSTOR,
www.jstor.org/stable/3033481.
United Nations (UN) (2009). World Population Prospects: The 2008 Revision. Edition - Extended
Dataset United Nations.
United Nations International Children's Emergency Fund (UNICEF) (2018). Levels and Trends in Child
Mortality, Report 2018. Available at: https://childmortality.org/wp-content/uploads/2018/12/UN-
IGME-Child-Mortality-Report-2018.pdf
120
Vallerand, R. J., Pelletier, L. G., Blais, M. R., Briere, N. M., Senecal, C., & Vallieres, E. F. (1992). The
academic motivation scale: A measure of intrinsic, extrinsic, and amotivation in education.
Educational and Psychological Measurement, 52, 1003−1017.
Van Dam, K., & Seijts, G. H. (2007, April). Measuring goal orientation climate. Paper for the
symposium ‘Goal orientation research across levels: The role of motives and context’, at the 22nd
Annual Conference of the Society of Industrial and Organizational Psychology, New York, USA.
Van Dam, K., Oreg, S., & Schyns, B. (2008). Daily work contexts and resistance to organizational
change: The role of leader–member exchange, development climate, and change process
characteristics. Applied Psychology, An International Review, 57, 313–334.
Vedel, Anna & Poropat, Arthur. (2017). Personality and Academic Performance. 10.1007/978-3-319-
28099-8_989-1.
Vermetten, Y., Lodewijks, H. and Vermunt, J. (2001). The Role of Personality Traits and Goal
Orientations in Strategy Use. Contemporary Educational Psychology, 26(2), pp.149-170.
Vignoles, A. (2016). What is the economic value of literacy and numeracy?. IZA World of Labor.
Vignoles, A., De Coulon, A. and Marcenaro-Gutierrez, O. (2011). The value of basic skills in the British
labour market. Oxford Economic Papers, 63(1), pp.27-48.
Walker, I. and Thompson, A. (1996). Disability Wages and Labour Force Participation: Evidence from
UK Panal Data. Department of Economics, Keele University, No. 96/14.
Walker, I. and Y. Zhu (2013). The impact of university degrees on the lifecycle of earnings: some further
analysis. BIS Research Paper No. 112.
Walker, I. and Y. Zhu (2017). University selectivity and the graduate wage premium: evidence from
the UK. IZA Discussion Paper No. 10536.
Walker, I. and Zhu, Y. (2008). The college wage premium and the expansion of higher education in the
UK. Scandinavian Journal of Economics, 110(4), pp.695-709.
Walker, I. and Zhu, Y. (2011). Differences by degree: Evidence of the net financial rates of return to
undergraduate study for England and Wales. Economics of Education Review, 30(6), pp.1177-
1186.
Ward, S., Williams, J., and Van Ours, J. (2015). Bad behavior: delinquency, arrest and early school
leaving. IZA Discussion Paper No. 9248.
Webbink, H. D., Koning, P., Vujić, S., and Martin, N. G. (2013). Why are criminals less educated than
non-criminals? Evidence from a cohort of young Australian twins. The Journal of Law, Economics,
and Organization, 29(1) pp.115 – 144.
Wei, H. (2004). Measuring the Stock of Human Capital for Australia. Australian Bureau of Statistics
No. 1351.0.55.001.
Wei, H. (2008). Measuring Human Capital Flows for Australia: A Lifetime Labor Income Approach.
Australian Bureau of Statistics No. 1351.0.55.023.
Weinberger, C. J. (2014).The Increasing Complementarity between Cognitive and Social Skills, Review
of Economics and Statistics, 96(5), pp.849-861.
Weiner, D., B. Lutz and Ludwig, J. (2009). The Effects of School Desegregation on Crime. NBER
Working Paper No. 15380.
121
Weiss, A. (1995). Human capital vs. signalling explanations of wages. Journal of Economic
Perspectives, 9(4), pp.133-154.
Wheaton, A.G., Chapman, D.P. and Croft, J.B. (2016). School start times, sleep, behavioral, health, and
academic outcomes: a review of the literature. Journal of School Health, 86(5), pp.363-381.
Wilson, J. Q. and Petersilia, J. (2011). Crime and public policy. Oxford University Press.
Windle, Gill. (2011). What is resilience? A review and concept analysis. Reviews in Clinical
Gerontology. 21. 152 - 169. 10.1017/S0959259810000420.
Witte, A.D. (1997). “Crime”, in J. Behrman and N. Stacey, eds., The Social Benefits of Education,
University of Michigan Press, Ann Arbor, 219–46.
Wohlwill, J. F. (1987). Introduction. In D. Gorlitz & J. F. Wohlwill (Eds.), Curiosity, imagination, and
play (pp. 1-21). Hillsdale, NJ: Erlbaum
World Bank (2006). Where is the Wealth of Nations?. Washington, D.C.: World Bank.
World Bank (2018). Fair Progress? Economic Mobility across Generations around the World,
Washington: World Bank Group.
Worth, J., Lynch, S., Hilary, J., Rennie, C. and Andrade, J. (2018). Teacher Workforce Dynamics in
England: Nurturing, Supporting and Valuing Teachers. National Foundation for Educational
Research.
Wössmann, L. (2016). The economic case for education. Education Economics, 24(1), pp.3-32
Yoder, J.D. and Hochevar, C.M. (2005). Encouraging active learning can improve students'
performance on examinations. Teaching of psychology, 32(2), pp.91-95.
Zangelidis, A. (2008). Occupational and industry specificity of human capital in the British labour
market. Scottish Journal of Political Economy, 55(4), pp.420-443.
Zax, J. S. and Rees, D. I. (2002). IQ, academic performance, environment and earnings. Review of
Economics and Statistics, 84(4), 600-616.
Zimmer, Z., Martin, L.G., Ofstedal, M.B., Chuang, Y.L. (2007). Education of adult children and
mortality of their elderly parents in Taiwan. Demography, 44(2), pp.289–305.