Working Paper Series
Skill Premia and Intergenerational Skill
Transmission: The French Case
B. Ben Halima
N. Chusseau
J. Hellier
ECINEQ WP 2013 – 285
ECINEQ 2013 – 285
January 2013
www.ecineq.org
Skill Premia and Intergenerational Skill
Transmission: The French Case
*
B. Ben Halima
N. Chusseau
EQUIPPE, Univ. of Lille 1 and MESHS
J. Hellier
LEMNA, University of Nantes
Abstract
In the case of France, we analyse the changes (i) in the skill premium linked to each level
of education and (ii) in the impact of parents’ skill and income upon the educational
attainment of their children. To this end, we build a theoretical model which is
subsequently estimated. Our calculations firstly reveal (i) a critical decline in the skill
premium of the Baccalaureate in relation to the lowest skill level, and (ii) an increase in the
skill premia of higher education in relation to the Baccalaureate, which however is not
large enough to avoid the decrease in all the skill premia relative to the lowest skill.
Secondly, we find (i) a significant increase in the impact of the family backgrounds upon
the individuals’ education from 1993 to 2003 which essentially derives from a higher
impact of parental income upon the educational attainment, and (ii) an increase in the
impact of public expenditure upon education. Consequently, if inequality has decreased
among the employed population, the slowdown in intergenerational mobility could reverse
this tendency in the longer term. This may however be offset by higher public educational
expenditure.
Keywords: Family backgrounds, intergenerational mobility, return to education, skill
premium.
JEL Classification: I2, J24, J31.
*
We wish to thank the French Research National Agency (ANR) for its financial support. In
addition, we thank Frédéric Séraphin for his contribution to an early version of this paper.
Contacts:
Joël Hellier: joel.hellier@wanadoo.fr
2
1 Introduction
The aim of this article is twofold. It firstly determines the variations in the skill premia linked
to each level of education in France over the period 1977-2003. From this first result, we
analyse the intergenerational skill transmission for the individuals surveyed in 1993 and 2003.
Since the seminal work of Mincer (1974), a large body of empirical literature has analysed
the impact of education upon wages. These works typically diagnose a large and significant
impact (see the reviews of Psacharopoulos, 1985 and 1994; Cohn and Addison, 1994; Card,
1999; Black and Devereux, 2011). In line with Mincer’s equation, most empirical studies
have measured the education level by the schooling years above a lower limit of 6 or 7 years
(Psacharopoulos and Patrinos, 2004). In their review of the empirical works on France,
Hanchane and Moullet (2000) present eight studies, all of them measuring human capital by
the number of schooling years. Nevertheless, this measure suffers several limitations. In
particular, it gives the same weight to all the schooling years regardless of the educational
level. Several studies have found decreasing returns to the schooling years (Psacharopoulos,
1994; Wössmann, 2003). In their analyses of the French case, Jarousse and Mingat (1986) and
Goux and Maurin (1994) have improved the measure based on schooling years by accounting
for repeated years and by distinguishing the certified from the non certified years. Introducing
qualitative variables, Jarousse and Mingat (1986) also find that, compared to the average
return to the related schooling years, the University 'Deuxième Cycle' (final two Degree
years) shows a wage deficit of 9% whereas the ‘Grandes Ecoles’ (Post Graduate Schools)
Degree benefits from a wage surplus of 30%. These results suggest that the wage value of one
schooling year can critically differ depending on the stage and the type of study considered.
Since Becker and Tomes (1976, 1979, 1986), the family background has been considered
as a key determinant of children’s education and income (Piketty, 2000; Chusseau and
Hellier, 2012). A sizeable impact indicates a low intergenerational mobility. The influence of
parents on their children’s human capital runs through several channels. Firstly, intra-family
human capital externalities and transfers impact upon both the children’s human capital and
their capacity to learn. In this respect, a number of empirical works have underlined the
influence of parental characteristics upon children’s performance at school (Acemoglu and
Pischke, 2001, for the US; Ermisch and Francesconi, 2001, for the UK; Lauer, 2003, for
Germany and France; Checchi et al., 2008, and Brunello and Checchi, 2005, for Italy, Liu et
al., 2000 and 2006, for Taiwan). A second channel of influence comes from the impact of
3
parents’ income on the funding of their children’s education. Highly skilled parents have
higher incomes and can thereby invest more in their offspring’s education, resulting in higher
skill and incomes of their children (Solon, 2004). Family funding is essential when credit
market imperfections prevent youngsters from borrowing for their education (Becker and
Tomes, 1979; Mulligan, 1997; Han and Mulligan, 2001; Grawe and Mulligan, 2002; Grawe,
2004). In addition, children with educated parents are more and better informed (Entwistle
and Alexander, 1992). Finally, the literature has pointed to the genetically transmitted
differences in ability (Miller et al., 1995; Ashenfelter and Krueger, 1994; Rouse, 1999;
Bowles and Gintis, 2001, 2002).
The impact of parents’ position upon children’s attainment has typically been estimated
by intergenerational elasticities of earnings and/or education.
An abundant empirical literature has analysed the impact of parent’s income upon the
child’s income through intergenerational earnings elasticities
1
. Intergenerational income
elasticities critically differ across countries, the lowest values (less than 0.3) being found in
Nordic countries (Björklund and Jäntti, 1997; Österberg, 2000; Jänti et al., 2006) and the
highest (between 0.4 and 0.6) in the US (Solon, 1992; Jänti et al., 2006; Mazumder, 2005), the
UK and France being in-between (Nicoletti and Ermisch, 2007 and Blanden et al., 2004 for
the UK; Lefranc and Trannoy, 2005 for France).
Intergenerational mobility has also been measured by the influence of parents’ human
capital on their children’s human capital (education). In these works, human capital is
typically measured by the number of schooling years, and OLS are used to estimate
intergenerational human capital elasticity. For the US, Mulligan (1997) finds an
intergenerational coefficient of 0.32 between father and son, and 0.33 between father and
child. For the UK, Dearden et al. (1997) find 0.424 for the father-son coefficient and 0.415 for
the father-daughter coefficient. Comparing the US and Germany, Couch and Dunn (1997)
find a father-son intergenerational coefficient of 0.42 in the US and 0.24 in Germany. This
reveals higher persistence in the former. Using the French database Formation Qualification
Professionnelle (FQP) in 1993, Fabre and Moullet (2004) find an intergenerational coefficient
of education of 0.31 between father and son and a mother-son coefficient of 0.29. By using
the number of schooling years to measure human capital, these empirical works are exposed
to the already mentioned critique of allocating the same weight to qualitatively different
schooling years.
1
Solon (2002), Grawe (2004), Jäntti et al. (2006), Mazumder (2005), Nicoletti and Ermisch (2007), Björklund
and Jäntti (2009), Blanden (2009), Corak (2006), Black and Devereux (2011) etc.
4
In this article, we firstly develop a theoretical model in which individuals choose their
skill level by maximising the return to education, with the different stages/cycles of education
providing uneven income gains and the educational attainment depending on the parents’ skill
and income. We subsequently estimate a slightly modified empirical version of the model on
the French database FQP, (i) to quantify the return (skill premium) to each level of the French
educational system, and (ii) to measure the impact of the parents’ human capital upon the
human capital of their offspring, by distinguishing the intra-family skill externalities from the
influence of parents’ income. Human capital is not measured by schooling years but by its
market value calculated from the results of a wage equation. As regards the first point, our
main findings are (i) that, compared to the lowest skill level, the skill premium generated by
the Baccalaureate (henceforth ‘bac’) has critically declined since the mid-seventies, and (ii)
that the premium generated by higher education in relation to the level of the bac has
increased, but this increase is not sufficient to offset the decline in the return to the bac.
Consequently, France has experienced a general decrease in its skill premia. Concerning the
second point, our calculations reveal that the impact of the family backgrounds upon the
individuals’ skill has critically increased from 1993 to 2003, this increase essentially deriving
from a higher impact of parental income upon the educational attainment. At the same time,
the impact of public educational expenditure has also increased.
In Section 2, we build the structural framework which is subsequently transformed into an
empirical model utilised for the estimations. Section 3 describes the data and Section 4 the
construction of the variables and the econometric methods. The results are exposed and
discussed in Section 5. We conclude in Section 6.
2 The model
We construct the theoretical framework from which we subsequently derive the empirical
model that is estimated in the following sections.
2.1. The theoretical model
We develop a theoretical framework which synthesises the two main channels through which
the parents influence their children’s skill, i.e., their income on the one hand and the human
capital intra-family externality and transfers on the other. These generate an intergenerational
transmission of human capital and a persistence of skill discrepancies across the dynasties.
