The Private Rate of Return to Education Analysis

NUOVA SECONDARIA RICERCA
The Private Rate of Return to Education Analysis
Andrea Cegolon
il saggio prende in esame il problema delle differenze tra lavoratori salariati nel mercato del lavoro. viene dato ampio spazio alla teoria del “capitale umano” che, in questo ambito, attribuisce valore prevalente alle abilità maturate dal soggetto nel suo percorso scolastico prima che al tipo di attività svolta nel lavoro. le pagine hanno l’obiettivo di rispondere alla domanda: quale il peso e l’importanza
della scuola riguardo alla produttività del lavoratore e di conseguenza quali sono i vantaggi economici che da essa derivano al soggetto.
The essay examines problem of differences between salaried workers in the labor market. Wide space Is given to the theory of “human
capital” which, in this context, assigns a prevailing value to the skills gained by the subject in his schooling period if compared to the
type of activity carried out in the work. The pages are intended to answer this question: what is the weight and importance of the school
with regard to worker productivity and consequently what are the economic benefits that flow from it to the subject.
W
hat explains wage differences between workers in the labour market has recently got a lot
of attention from researchers. From a certain
point of view, the issue can intuitively be attributed
both to the type of economic activity workers are involved in and to the level of their skills and knowledge
with which they perform their work. The “human capital” theory tends to emphasize especially the latter side
of the problem: the body of skills and knowledge acquired by an individual through training and especially
through schooling (broadly termed as “human capital”)
thanks to which a worker is more productive (Schultz,
1961; Becker, 1962; Mincer, 1974). as a result, the
wage differentials in the labour market can be explained
by the different level of human capital between workers. given that, a rewarding line of research has focused on estimating the impact of education on an individual’s labour market productivity and wage premia.
This strand of literature is known as rate of return to education analysis.
The Methodology
There are two main methods of estimating rates of return
to investment in education.
The first approach, defined by Carnoy (1995) the “traditional method”, follows an algebraic definition of the
rate of return, that is the rate of discount equating the net
present value of life-time benefits of education of the individual, to the net present value of costs of education. in
other words, according to this approach the internal rate
of return consists in setting the discounted value of costs
(
6
∑
Ci
(1 + r )
i
) and benefits (
∑
Bi
(1 + r )
i
), over the time
equal to zero and solving for the implicit discount rate, r.
0=∑
Ci
(1 + r )
i
+∑
Bi
(1 + r )
i
The above equation shows that the individual spends for
education or other cost incurred (C) are negative whereas
the additional income or other benefits the individual
gains from the education (B) are positive.
From the individual’s standpoint, the pecuniary benefits
of additional education are the additional income the individual earn as a consequence; the nonpecuniary consumption benefits educational investment provides over
a person’s life, such as a greater enjoyment of cultural activities or higher social status, and the direct consumption
derived from taking education. however, owing to the difficulty to assign a value to non-pecuniary benefits in
measuring private rates of return, , economists have
checked out nonpecuniary benefits. in particular, according to the oeCD (2002), the benefits taken into account
in computing the rate of return are the gains in post-tax
earnings adjusted for higher employment probability, less
the repayment of public support during the period of
study.
Private costs of education include the income incurred by
students while attending school or other educational activities; the additional expenditures associated with taking education, namely direct costs (such as tuition fees,
books, transport); and non-monetary-costs (such as the
distaste for learning). of course, as seen above for the
schooling benefits, it is hard to measure some of the cost
components, namely the non-monetary ones. as a result,
usually the costs of schooling investment equal tuition
fees, foregone earnings net of taxes adjusted for the prob© Nuova Secondaria - n. 4, dicembre 2014 - Anno XXXII
NUOVA SECONDARIA RICERCA
Figure 1: Potential
Earnings Streams
Faced by a High
School Graduate and
College Graduate
(Borjas, 2010).
ability of being in employment, less the resources made
available to students in the form of grants and loans
(oeCD, 2002).
essentially, according to this method, individuals undertake a cost-benefit analysis to determine a quantifiable
economic rate of return to education and, consequently,
whether education would be obtained. as an acceptable
criterion, future earnings need at least to compensate individuals for the direct and indirect costs of education –
as can be seen in Figure 1.
Borjas (2010) shows the direct and indirect costs of acquiring education, which will hopefully be counteracted
by the higher earnings faced by the college graduate,
thus justifying his decision to acquire more education. although not primarily interested in whether a pupil is a college or high school graduate, the same principle can be applied to a child on the decision of whether to stay in
class, after compulsory school, or not. The foregone earnings of attending school, rather than perhaps working in
the agricultural sector, can be seen by the shaded area 2
in the figure. in addition to this area the direct costs of acquiring education – such as books and tuition fees - as
shown in area 1. if these costs (the two areas, 1 and 2) are
lower than the benefits of increased earnings obtainable
in the future (area 3), then an individual will attend school.
an issue related to the traditional method is the impossibility of taking into account all the costs and benefits associated to the schooling investment. Furthermore it is
quite demanding because it needs complete longitudinal
life histories of the earnings of individuals, beginning with
© Nuova Secondaria - n. 4, dicembre 2014 - Anno XXXII
their entry age into the labor force and ending with their
retirement, with additional information about the costs of
education. unfortunately, such ideal data are seldom
available. For this reason most empirical literature estimates the private returns to education using the Mincerian earnings function which refers to cross-sectional data.
This method, originally proposed by Jacob Mincer (1972),
typically adopts an ordinary least squares (olS) regression, where log earnings is regressed on years of study and
age/experience in the labour market, as follows:
ln Wi = α + ∑ β ik Sik + γ 1 E1 + γ 2 E22 + δ X i + u1
k
i = 1,…, N
W, the dependent variable, is an earnings measure for an
individual: typically the hourly net wage obtained by dividing the total wages and salaries receivable for work by
the total number of working hours. Regarding the independent variables, S represents a measure of schooling attainment: it is generally measured by the number of years
spent at school. Tough, often data set do not contain information about this number of years, but only on the
highest degree attained by individual. in this case the educational attainment of the individual is calculated by imputing the number of years required to complete her/his
reported level of educational attainment1. Practically, the
1. Standard, not actual, years of formal schooling are recorded. Since students
who fail to reach a standard have to repeat the year, the actual number of years
is likely to be underestimated.
7
NUOVA SECONDARIA RICERCA
ment in education raises worker productivity, and the
continuous years of schooling variable S is converted
market wage received by a worker at any point in the
into a series of dummy variables, to denote the fact that
life cycle is equal to the value of his current marginal
a person has completed the corresponding level of eduproductivity. Considered as the human capital intercation (Psacharopoulos, 1994). E is the labour market expretation of education (Becker, 1962), it provides the
perience and it equals to age, minus years of schooling,
economic rationale for individuals to invest in educaminus age of starting schooling (5), while the potential extion. in this case, β is really a measure of the private reperience E2 captured the non-linear relationship between
turns to the individual investment in education. anyW and E in the experience-earnings profile. X is a vector
way there is another theory which attempts to reverse
of observed exogenous explanatory variables which could
the causal relationship between education and earnings,
be assumed to affect earnings, such as gender, race/eththe so called “signalling” theory (Spence, 1973). acnicity, location, industry affiliation, etc. U is a disturcording this approach, education is instrumental to an
bance term representing other forces which may not be
employer in order to distinguish the quality of stuexplicitly measured, assumed independent of X and S.
dents, so selecting the most able. in the theoretical
in this semi-log specification the coefficient on years of
model of Spence, then, more years of study insure a
schooling β can be interpreted as the marginal private rate
higher income in comparison to a lower level of eduof return to education. however, there is a number of key
cation. in fact, education acts as a signal about the abilassumptions involved in accepting the interpretation of β
ity of the individual, and not because it has some effect
as a return to education:
on its productivity: the inherent human capital, not
1) The earnings captures the full benefits of the education
education itself, is what increases productivity and
investment (Björklund, Kjellström, 2002, p. 196). This
leads to higher wages (Kjelland, 2008, p. 70). in conassumption means that wages, measured net of taxes
trast to the human capital theory, in this latter case the
and transfers, also include non-monetary benefits of edrate of return to education loses its meaning because no
ucation to the individual. These private non-pecuniary
longer considered a proper indicator for evaluating the
benefits could be direct such as better health, spousal
profitability of investment in education.
health, longevity, happiness (or well being), the enjoyment from learning, but also indirect such as lower 4) The length of working life is the same independent of
the length of schooling. This is a very strong and quesinfant mortality, child health, child education, child
tionable assumption because it involves that “every adcognitive development (McMahon, 2008).
ditional year of schooling postpones retirement by one
2) The only costs of schooling are foregone earnings
year.
So, if a person who leaves school at age 19 to
(Björklund, Kjellström, 2002, p. 196). Many studies
start work retires at age 65, we must assume that a perabout the investment in education use cross-section
son who leaves school at age 20 retires at age 66”
data to estimate the rate of return to education. Such
(Björklund, Kjellström, 2002, p. 196).
data typically contain information on current earnings
of those in the labour force, age, and years of education, 5) Schooling precedes work. it is possible, as a matter of
fact, that an individual, after having been out in the
but no information on direct costs of education, such as
labour market for a few years, attends school as an
tuition fees, expenditure on books, transportation, priadult, but the earning function is based on a linear
vate lessons etc. Because of these limitations the only
time scan for which people first go to school, then they
category of cost taken into account is the opportunity
start to work and finally they retire after a fixed numcost, that is the non-income due to taking schooling inber of years. a possible limitation is suggested by
stead of working (foregone earnings). The Mincerian
Björklund, Kjellström: the advantages of taking eduearnings function automatically assigns foregone earncation in young ages are not revealed by the estimated
ings to the rate of return to education but, as
β coefficient since, given the assumption n. 4, the inPsacharoupoulos argues, primary school children could
ternal rate of return to education for people starting
not be expected to work full-time and so they cannot
school as adults is the same as the ones starting school
forego earnings during their school career. Thus “it is
at the age of 20 (Björklund, Kjellström, 2002).
a mistake to mechanically assign to them six years of
foregone earnings as part of the cost of their educa- 6) The economy is in a steady state without any wage and
tion” (1994, 610, p. 1326). as a result, the returns to
productivity growth. Since the rate of return to educaprimary education level turns out underestimated.
tion is estimated ex ante, the earning function should
take into account a wage and productivity growth.
3) The Mincerian equation captures the casual effect of
however, for simplicity reasons, it is assumed that
schooling. according to this assumption, the invest-
8
© Nuova Secondaria - n. 4, dicembre 2014 - Anno XXXII
wages are constant over time throughout the all working life (Björklund, Kjellström 2002).
a large literature does not state all these conditions under
which the coefficient of schooling β in an earnings regression can be interpreted as a rate of return to education
(Carneiro and heckman 2003).
General issues in estimating the returns to
education
The olS estimation of the standard wage equation involves some methodological problems. indeed it leads to
biased estimates of the rate of return to education, linked
to the “endogeneity” of schooling. The necessary condition for the correctness of the olS estimates is that the
schooling variable S is exogenous or not correlated with
the error term U. however, if the number of years of
school is not randomly assigned to the individual, but is
subject to its decision, the schooling variable S will be endogenous and, therefore, the estimation by ordinary least
squares (olS) will yield biased estimates of the return to
schooling β. The correlation between schooling variable
and the error term causing the endogeneity may result
from (Card, 1995; griliches, 1977):
1) omitted variables;
2) measurement error;
3) heterogeneity of rate of return to education within the
population.
in the first case the endogeneity arising from omitted
variables essentially refers to the omission in the Mincerian function of the unobserved innate ability (griliches
Z., Mason W. 1972; griliches 1977). The problem of
ability bias is basically as follows: more able people
could get exra schooling and thus being favored to earn
more not because of the additional schooling but just because they are more able. So if the individual’s ability and
educational attainments are correlated, estimation of the
wage equation would give biased results. according to
griliches (1977), innate ability may have opposite effects
on the rates of return to education, and there is no good
a priori reason to expect ability bias to be positive. on the
contrary it may turn out to be small or negative. From a
certain point of view, individuals with higher ability are
likely to convert schooling into human capital more effectively compared to the less able ones and this, consequently, raises the returns for individuals with higher
ability. on the other side, if ability to progress in school
is positively correlated with capacity to earn, this may reduce the returns: higher able persons may have been able
to earn more in the labour market and due to higher opportunity cost in attending school, they may end up leaving the school earlier (harmon et al., 2003).
© Nuova Secondaria - n. 4, dicembre 2014 - Anno XXXII
NUOVA SECONDARIA RICERCA
The traditional way of dealing with ability bias for labour
economists has been:
a) adding better controls for innate ability to the set of independent variables
b) instrumental variables
c) twins studies
Some studies have allowed to include proxy measures of
the ability such as iQ scores or test scores designed to
measure innate ability (Card, 1999; griliches, 1977).
These empirical studies tend to find the ability bias to be
small since the estimation of rate of return to education
falls only by about 10% after controlling the effect of ability. This strategy is theoretically easy to formulate, but not
so much to put into practice. Firstly, not many data sets
with labour market information contain intelligence or
ability information. Secondly, it is very difficult to get a
good measure for innate ability not in themselves determined by schooling; so in that case the use of ability proxies will bias estimated rates of return downward (ashenfelter et al., 2000). Thirdly, there may be other factors
besides ability, not correlated with schooling and affecting wages, that cannot be observed by researchers, such
as the family background (those aspects do not work
through ability). Some authors, indeed, (haveman et al.,
1991; Card, 1999) show that parental education, determinant on educational attainment of children, is also
highly correlated with wages. in particular, the family
background works in two ways: both by providing a better learning environment and by aiding better contacts or
connections. individuals with more educated parents are
more likely to get better information about employment
and, therefore, they often can obtain better paying or
more secure jobs in the formal sector (Krishnan, 1996;
Siphambe, 2000). generally these studies show that the
rate of return to education falls very slightly (by around
0,01), after controlling some family characteristics
(ashenfelter and Rouse, 1998; ashenfelter and Zimmerman, 1997).
The second strategy to deal with ability bias consists in using institutional aspects of the educational system to generate instrumental variables (iv) in some way correlated
with the experience of education but not with the individual’s ability. examples of iv, successfully used, include: college proximity (Card, 1995), quarter of birth
(angrist and Krueger, 1991; Staiger and Stock, 1997),
raising of the minimum school leaving age (harmon and
Walker, 1995; Meghir and Palme, 1999), educational reforms (Brunello and Miniaci, 1999), differences in educational attainment due to war (ichino and Winter-ebmer,
2004; angrist and Krueger, 1995). often the rate of return
estimates yielded by using iv estimation overcomes those
9
NUOVA SECONDARIA RICERCA
provided by olS. “The main criticism of IV estimates revolves around the concern that the instrument may not, in
fact, be truly independent of the earnings residual. If, for
example, the instrument is positively correlated with earnings, the IV estimator may be upward biased” (ashenfelter et al., 2000, p. 5).
another way to control the ability bias is to use data of
monozygotic twins since they probably have the same natural ability, while the differences in schooling are due to
random factors. in such a way this procedure approaches
the ideal random-assignment technique for estimating
treatment effects, namely the effects of more education.
The methodology was first used by Taubman (1976) who
found strong evidence for a positive ability bias in olS
estimates. More recent natural experiments on twins, reviewed by ashenfelter and Rouse (1998) and Miller et al.
(2003), typically suggest lower rate of return to education
than olS. Two questions have emerged in relation to
twins studies. First, even if identical twins have the same
genes, this kind of people could not manifest identical
ability or they could have different motivations and other
different features affecting education and earnings. in
that case the twin estimates would be completely biased.
The second problem relates to measurement error:
griliches (1979) points out that the use of estimates obtained from differencing between twins increases the extent of measurement error in schooling and so the tendency for estimates to be attenuated.
The distortion due to measurement errors is to be detected
in the calculation of the years of schooling. as rightly emphasized by Card (1999), schooling can differ from true
schooling by an additive error. This problem can be traced
to the fact that people reporting their schooling level
could make inaccurate statements or overstate/understate
their true level of schooling. For example, people with low
level of education may be inclined to overstate their actual education while more educated people may be more
accurate in their reporting. Moreover an individual holding a given qualification could have repeated a school year
and thus the years spent in school could be more than
those stated. even if, in theory, the biases from measurement error in schooling can go in either directions, Card
(1999) concludes, based on his assessment of the literature, that this kind of error leads to underestimate the rate
of return to education.
another source of bias to the estimates is the heterogeneity of returns to education in the population. Most
empirical studies overlook the chance that individuals
differ from each other by concentrating on populationwide estimates of rate of return to education, obtained
with a olS regression. But in this way it is not taken into
10
account that gains from schooling depend upon several individual’s features, such as the family background, innate
ability, motivation, the quality and type of school, etc. For
this reason, since the impact of an additional year/level of
education may not be the same across the earnings distribution, it is possible to use the quantile regression techniques to analyze heterogeneous patterns of return to education across the conditional wage distribution
(Buchinsky, 1998).
Finally the estimation of the wage equation could also suffer from the problem of ‘sample selection bias’. This occurs when wage functions are estimated using only the individuals working and consequently earning a wage. This
might be a selective group, but not a representative sample. as a matter of fact workers constitute only a short percentage of all the population: unemployed people, for example, are not taken into account. in such a situation, the
olS estimates maybe biased. To manage this problem, estimations, based on the method of joint maximum likelihood (JMl) suggested by heckman (1974), are usually
applied.
Estimates of the rate of return to education
The figures reported in the Table 1 refer to the returns to
education in italy as the result of different empirical survey. The estimated rate of return to an additional year of
schooling considerably vary across studies, also for the
method used in the estimate. Considering regional data
antonelli (1985) estimates that an additional year of
schooling increases annual net earnings by 4.6%. Cannari
et al. (1989) use a larger sample from the 1986 wave of
the Bank of italy, finding a similar result with a return
around 4%. While lucifora and Reilly (1990) estimate the
Mincerian earnings function using the eNi special survey
on earning and they find that the marginal return to
schooling is slightly higher for men than for women, but
again around 4%.
For the 1993 wave of Bank of italy Cannari and D’alessio
(1995), using family background variables as instruments
of educational outcomes, find that the marginal return to
education is around 7%, much higher than previous results. also Colussi (1996) arrives at a similar result, using the same wave and an analogous set of instruments.
Focusing on the returns to education separately for men
and women for 1991 and using an instrumental variable
approach based upon the identification of exogenous
changes in the schooling system wave, Flabbi (1997)
finds that the marginal effect of education is 6.2% for men
and 5.6% for women, so confirming the gender gap in
earnings. For the 1993 and 1995 waves, Brunello and
Miniaci (1999) estimate a return to education equal to
© Nuova Secondaria - n. 4, dicembre 2014 - Anno XXXII
NUOVA SECONDARIA RICERCA
Tab. 1 - A summary of the estimated return of an additional year of schooling in Italy
obtained in different surveys (Fiaschi and gabriellini, 2013, p. 7)
Author
antonelli (1985)
Cannari, Pellegrini, and Sestito (1989)
lucifora and Reilly (1990)
Cannari and D’alessio (1995)
Colussi (1996)
Flabbi (1997)
Brunello and Miniaci (1999)
Method
Years of observation
Marginal return to education
olS
1986
4.0
olS
olS
iv
iv
4.0 (men) / 3.6 (women)
1993
7.6
6.2 (men) / 5.6 (women)
1995
6.2 (men) / 7.7 (women)
1987-2000
6.9
olS
1987-2000
olS
1987-2000
olS
7.0
1991
1993 and 1995
Ciccone, Cingano, and Cipollone (2006)
Cingano and Cipollone (2009)
1985
iv
olS
4.6
1993
iv
Brunello, Comi, and lucifera (2000)
Ciccone (2004)
1977
5.7
6.1
6.0
5.7% (taking into account the endogeneity of schooling). Tab. 2 - OLS estimates of rate of return of an
additional year of schooling from 1985
The estimated coefficient on the Mincerian rate of return
through 1995 in some countries
to schooling is around 6% both in Ciccone (2004) and
(Trostel, Walker, Wooley, 2002, p. 5)
Cingano and Cipollone (2009). on average, the table
shows that the private rate of return associated with a year
Country
Males
Females
of schooling in italy is around 5%, implying that a person
7.4
9.6
with a master degree earns almost 25% more than an in- uSa
dividual holding a high school diploma.
great Britain
12.7
13.0
Comparing these estimates for italy with those of other Russia
4.4
5.3
developed countries, as shown in the Table 2, in our Norway
2.3
2.5
Country the rate of return to education is among the low5.1
5.2
est, along with Norway, Netherlands and Sweden. in australia
3.1
1.9
other words, studying in italy yields lower benefits than Netherlands
3.8
6.4
other countries, such as u.S, great Britain, Japan, ireland. austria
Rate of return analysis limitations
From a policy point of view, the rate of return analyses
can be useful in informing the trends in the supply and demand for different types of skill in the labour market. at
the same time, however, the private rates of return can inform individuals as to which type of education or training investment will yield them the highest return, i.e. the
highest future gain in wages. More specifically: private
rates of return can inform individuals as to the average
rate of return experienced by individuals who have made
that kind of investment in the past.
Neverthless this tool entails also some limitations:
i) often the multiple dimensions of human capital are not
taken into account: investment in schooling is not the
only form of human capital investment undertaken by
people. according to Thomas and Strauss (1997), for
example, health, affecting wage determination, is also
© Nuova Secondaria - n. 4, dicembre 2014 - Anno XXXII
Poland
7.3
10.0
italy
3.7
5.3
New Zealand
ireland
Japan
hungary
3.3
8.5
7.5
7.5
2.9
9.0
9.0
7.7
N. ireland
17.4
14.6
Slovenia
8.0
10.1
4.0
5.7
Sweden
israel
Bulgaria
Canada
Spain
Switzerland
Philippines
2.4
5.3
3.8
4.6
4.5
11.3
3.3
6.1
4.5
3.8
4.8
19.2
11
NUOVA SECONDARIA RICERCA
earnings of individuals who acquired different
an important component of human capital, thus omityears/levels of education in the past. Because of this
ting the health status in the rate of return to education
temporal gap, this too will not necessarily predict the
may lead to biased estimates. Though to really measure
future labour market value of education acquired tohealth condition is not so easy, Thomas and Strauss,
day (Powdthavee and vignoles 2006, 4). Thus, since
however, proxy for it with height, body mass index
rate of return are computed ex ante, caution is essen(BMi = Weight/height measured in kg/m), and various
tial when evaluating them too prescriptively.
calorie intake measures.
ii) The wage equation can use as a schooling variable the iv) Due to the lack of information of the data set, the most
serious weakness of Mincerian earning function
individuals’ highest levels of education, instead of the
method relates to its inability to account for the exmore common measure of years of schooling, though,
ternal benefits of education (non-pecuniary benefits).
from a policy perspective, it is probably preferable to
in fact it simply considers the earnings of the indiestimate the labour market value of different types of
viduals (post-tax earnings as private benefits). Relequalifications. Policy makers, but also students and
vant policy implications are a matter-of-fact, since the
their parents, need more information than merely the
rate of return to education is an important tool used
average return to a year, if they have to make decisions
by policy makers to decide on the allocations of pubabout different types of educational investment. in
lic investment on different levels of education. in
this regard Dearden et al. (2002) represents an interother words, since private returns are not sufficient to
esting paper for the uk, because it provides a comtake important policy decision, you also need to conprenhensive analysis of returns to specific British qualsider the social rates of returns to education which
ifications, containing a distinction of the returns to both
take into account the gains accruing to the society as
academic and vocational qualifications. The authors,
a whole. indeed, social rates are better for policy
thanks to the richness of two data sets used - the
making because they give info about the external
British National Child Development Study (NCDS)
benefits of education, generally referred to as “exterand the labour Force Survey (lFS) - estimate the renalities”, such as improved health of children (Currie
turns to all the qualifications held by an individual.
and Moretti, 2003), improved civic participation
This approach has a particular advantage if compared
(Dee, 2004, Milligan, Moretti, and oreopoulos,
to others based on highest level of education or on
2004), lower crime rates (lochner and Moretti, 2004),
years of education. in fact it enables to estimate the
etc. Because not measurable with any acceptable
wage effects from taking differing routes through the
level of precision, these externalities are rarely
education system, plus any given combination of qualbrought into the calculus of rate of return, and
ifications. another problem related to the approach denonetheless they are very important. indeed they convised to use the highest level of education is that,
stitute one of the main reasons of the State funding of
given different qualifications at a similar level, it is
education, especially for the lowers education levels.
questionable which one is the highest. “For example,
if an individual with a degree subsequently spends v) Finally, another limitation of this method is that it retwo or three months acquiring a professional qualifiports only estimates of the rate of return to years of edcation, which should be ranked as the highest? If we
ucation, while what determines individuals’ level of
assume that individual always progress to a higher
productivity is the level and appropriateness of their
level, then it would be the professional qualification.
skills. That is to say: education affects earnings not
On the other hand, the degree required considerably
only through the quantity of schooling, but also
more study time, and may be of more general value in
through the quality of that schooling. Much of the
the labour market. By analyzing the effect of all qualwork of economists focuses on the economic returns
ifications held by an individual, we avoid having to
to differing levels of school attainment for individuals,
make these somewhat arbitrary rankings of qualificabut, as suggested by hanushek and Wößmann (2007),
tions” (Dearden et al., 2002, pp. 251-252).
what matters for individuals’ performance in the labour
market is not only the quantity of school, but espeiii) a general limitation of rate of return to education recially the “quality”, namely the level of cognitive
lates its use to predictions about future scenarios of
skills mastered by students and measured with stanthe labour market. indeed the estimates of the rate of
dardize test (i.e. PiSa, TiMMS and PilS). Now, nureturn of schooling are based on retrospective data. in
merous empirical studies, both for developed (Mulliother words, a researcher estimates the rate of return
gan, 1999; Murnane et al. 2000; lazear, 2003,
to education by looking at the current labour market
12
© Nuova Secondaria - n. 4, dicembre 2014 - Anno XXXII
Mcintosh and vignoles, 2001) and developing contries
(Behrman et al., forthcoming; glewwe, 1996; Moll,
1998) document the positive impact of higher achievements on standardized tests on earnings. The basic approach used typically involves estimating a standard
Mincerian earnings function which relates the logarithm of earnings to years of schooling, experience, a
measure of individual cognitive skills, plus other factors affecting individual earning differences. The main
problem with this approach is the difficulty to get
data set on cognitive skills along with the subsequent
wages. This type of analyses generally requires tracking individuals over time.
Conclusion
Besides the previous methodology limitations, there is a
considerable unwilingness to take into account the rate of
return analysis, particularly in the form of cost-benefit
analysis. Such analysis would distract us from “tragic
questions” and moral concerns, by confining our decisions
along pragmatic lines (Nussbaum, 2001); it is “stupid”,
and precludes deliberative political decision-making
(Richardson, 2001); it is too weak, and rarely enforced
(hahn, 2001); and it is often not appropriate (Frank,
2001). ultimately, the strongest criticism of cost-benefit
analysis is the ‘saliency mismatch’: some costs and some
benefits are simply too difficult to calculate (herrnstein,
1997). Not everything is quantifiable, or has a price; and
what is quantifiable will drown out what cannot be quantified.
© Nuova Secondaria - n. 4, dicembre 2014 - Anno XXXII
NUOVA SECONDARIA RICERCA
Neverthless, there are also numerous advantages both in
calculating rates of return and using this for policy evaluation in education. Primarily, a rate of return is a standard metric, with a general meaning (Weale, 1993): returns to one investment can be directly compared with
returns to another (Cohn and hughes, 1994). Decision
rules about program development are relatively straightforward: once a critical threshold interest rate is posited,
a benchmark is provided, lower-bound for any rate of return. a precise rate is not necessary, only that the returns
are (substantially) above this threshold. and inference is
easier using rates of return: ceteris paribus, more investment will reduce the rate of return (as shown by Schultz,
1999). To sum up, even if the rate of return to education
analysis is obviously an incomplete methodology and,
such as it is, it requires sophisticated econometric techniques to be applied to the basic structure for correcting
different sources of bias affecting the estimates obtained
(as stressed by Posner 2001), no alternative evaluative criterion of sufficient merit has been proposed. after all, economics of education and the rate of return to education in
particular have been useful in unraveling several phenomena in education. While the rate of return method is
likely to stay, its value and relevance would certainly get
enhanced considerably, if attempts are made to capture for
example non- monetary benefits of schooling and bring
them into the rate of return calculations.
Andrea Cegolon
Università di Macerata
13
NUOVA SECONDARIA RICERCA
BIBLIOGRAFIA
J. D. Angrist, - A. B. Krueger, A. B., Split-Sample Instrumental Variables Estimates of the Return to Schooling, Journal of Business &
economic Statistics, american Statistical association, 13(2), (1995), pp. 225-235.
J. D. Angrist, - A. B. Krueger, Does Compulsory School Attendance Affect Schooling and Earnings?, The Quarterly Journal of
economics, 106(4), (1991), pp. 979-1014.
G. Antonelli, Risorse Umane e Redditi da lavoro, Franco angeli, Milano 1985.
O. Ashenfelter - C. Harmon - H. Oosterbeek, A Review of Estimates of the Schooling/Earnings Relationship, with Test for Publication
Bias, NBeR Working Paper No. 7457, (2000).
O. Ashenfelter - C. Rouse, Income, schooling and ability: Evidence from a new sample of identical twins, Quarterly Journal of
economics, 113(1), (1998) pp. 253–284.
G. S. Becker, Human Capital: A Theoretical and Empirical Analysis, 3rd edition, university of Chicago Press, Chicago and london
(1993 [1964]).
J. R. Behrman - D. Ross - R. Sabot, Improving the quality versus increasing the quantity of schooling: Estimates of rates of return from
rural Pakistan, Journal of Development economics, vol. 85, n. 1-2, (2008) pp. 94-104.
A. Bjorklund - C. Kjellstrom, Estimating the return to investments in education: how useful is the standard Mincer equation?,
Economics of Education Review, 21(3), (2002) pp. 195–210.
G. Borjas, Labor Economics, 5th ed., Mcgraw hill/ irwin, New York 2010.
G. Brunello - S. Comi - C. Lucifera, The Returns to Education in Italy: A new look at evidence, IZA DP n. 130, (2000).
G. Brunello - R. Miniaci, The Economic Returns to Schooling for Italian Men. An Evaluation Based on Instrumental Variables, labour
economics, (6)4, (1999) pp. 509-519.
L. Cannari - G. D’Alessio, Il Rendimento dell’istruzione: alcuni problemi di stima, Bank of italy, Temi di Discussione n. 253, Roma
1995.
L. Cannari – G. Pellegrini – P. Sestito, Età, Istruzione e Retribuzioni, Contributi all’analisi economica, Bank of italy, Roma, 1989.
D. Card, The causal effect of education on earnings, in o. ashenfelter - D. Card. (eds), Handbook of Labour Economics, 3, North
holland Publishing Co., NY and amsterdam 1999, pp. 1801-1863.
D. Card, Using geographic variation in college proximity to estimate the return to schooling, in L. F. Christofides - E. K. Grant - R.
Swidinsky (eds.), Aspects of labour market behaviour: essays in honor of John Vanderkamp, university of Toronto Press, Toronto 1995,
pp. 201-222.
P. Carneiro – J. Heckman, Human Capital Policy, IZA Discussion Papers n. 821, institute for the Study of labor (iZa), 2003.
M. Carnoy, Rate of return to education, in M. Carnoy (eds), International encyclopedia of economics of education, Pergamon, oxford
1995, pp. 364-369.
A. Ciccone, Human Capital as a Factors of Growth and Employment at the Regional Level. The Case of Italy. Report for the European
Commission, Dg for employment and Social affairs, 2004.
A. Ciccone - F. Cingano - P. Cipollone, The private and social return to schooling in Italy, Temi di Discussione, n. 569, Bank of italy,
Roma 2006.
F. Cingano - P. Cipollone, I rendimenti dell’istruzione, Questioni di Economia e Finanza (Occasional Papers), n. 53, Bank of italy,
Roma 2009.
A. Colussi, Il tasso di rendimento dell’istruzione in Italia: un’analisi cross-section, Politica Economica, 13(2), (1997), pp. 273-294.
E. Cohn - W. W. Hughes, A benefit–cost analysis of investment in college education in the United States: 1969–1985, economics of
education Review, 13(2), (1994), pp. 109–123.
J. Currie - E. Moretti, Mother’s Education and the Intergenerational Transmission of Human Capital: Evidence from College Openings,
Quarterly Journal of economics, 118(4), (2003), pp. 1495-1532.
L. Dearden – S. McIntosh – M. Myck - A. Vignoles, The Returns to Academic Vocational and Basic Skills in Britain, Bulletin of of
economic Research, 54(3), (2002), pp. 249-274.
T. S. Dee, Are there civic returns to education?, Journal of Public economics, 88 (9-10), (2004), pp.1697-1720.
D. Fiaschi - C. Gabbriellini, Wage Functions and Rates of Return to Education in Italy, Dipartimento di economia e Management,
university of Pisa, 2013.
L. Flabbi, Returns to Schooling in Italy: OLS, OV and Gender Differences, Working paper of Bocconi university, Serie di econometria
ed economia applicata, 1999.
R. H. Frank, Why is cost-benefit analysis so controversial?, in M. D. Adler - E.A. Posner (Eds). Cost–Benefit Analysis. Legal,
Economic, and Philosophical Perspectives, university of Chicago Press, Chicago 2001, pp. 77-94.
P. Glewwe, The relevance of standard estimates of rates of return to schooling for educational policy: A critical assessment, Journal of
Development economics, 51(2), (1996), pp. 267-290.
Z. Griliches, Sibling models and data in economics: Beginnings of a survey, Journal of Political economy, 87(5), (1979), pp. 37–64.
Z. Griliches, Estimating the Returns to Scholling: Some Econometric Problems, Econometrica, 45(1), (1977), pp. 1-22.
Z. Griliches - W. Mason, Education, Income and Ability, Journal of Political economy, 80(2), (1972), pp. 74-103.
M. Gunderson - P. Oreopoulos, Returns to education in Developed Countries, in D. J. Brewer - P. J. McEwan (ed.), Economics of
Education, elsevier, oxford 2010.
14
© Nuova Secondaria - n. 4, dicembre 2014 - Anno XXXII
NUOVA SECONDARIA RICERCA
R. W. Hahn, State and federal regulatory reform: A comparative analysis, in M. D. Adler - E.A. Posner (Eds), Cost–Benefit Analysis.
Legal, Economic, and Philosophical Perspectives, university of Chicago Press, Chicago 2001, pp. 109-123.
E. A. Hanushek - J. Wößmann, The Role of School Improvement in Economic Development, World Bank Policy Research Working
Paper 4122, 2007.
C. Harmon - H. Oosterbeek - I. Walker, The Returns to Education: Microeconomics, Journal of economic Surveys, 17(2), (2003), pp.
115-156.
C. Harmon - I. Walker, Estimates of the Economic Return to Schooling for the United Kingdom, The Economic American Review,
85(5), (1995), pp. 1278-1286.
R. Haveman - B. Wolfe - J. Spaulding, Childhood Events and Circumstances Influencing High School Completion, Demography, 28(1),
(1991), pp. 133-157.
J. Heckman - S. Polachek, The Functional Form of the Income- Schooling Relation, Journal of the american Statistical association,
69(346), (1974), pp. 350-354.
R. J. Herrnstein, The Matching Law, Russell Sage Foundation, Beverly hills Ca 1997.
A. Ichino - R. Winter-Ebmer, The Long-Run Educational Cost of World War II, Journal of labor economics, 22(1), (2004), pp. 57-86.
G. C. Kingdon, Does the labour explain lower female schooling in India?, Journal of Development Studies, 35(1), (1998), pp. 39-65.
J. Kjelland, Economic Returns to Higher Education: Signaling v. Human Capital Theory; An Analysis of Competing Theories, The Park
Place Economist: vol. 16, (2008).
P. Krishnan, Family Background, Education and Employment in Urban Ethiopia, Bulletin of Economics and Statistics, 58(1), (1996), pp.
167-183.
E. P. Lazear, Teacher incentives, Swedish economic Policy Review, 10(3), (2003), pp. 179-214.
L. F. Lee, Generalized econometric models with selectivity, econometrica, 51(2), (1983), pp. 507-512.
L. Lochner, - E. Moretti, The Effect of Education on Crime: Evidence from Prison Inmates, Arrests, and Self-Reports, american
economic Review, 94(1), (2004), pp. 155-189.
C. Lucifora - B. Reilly, Wage Discrimination and Female Occupational Intensity, labour, (4), (1990), pp. 147-168.
S. McIntosh - A. Vignoles, Measuring and assessing the impact of basic skills on labor market outcomes, oxford economic Papers, 53,
(2001), pp. 453-481.
W. McMahon, The External Social Benefits of Education, Discussion paper given to llaKeS, ioe, 2008.
C. Meghir - M. Palme, Assessing the effect of schooling on earnings using a social experiment, iFS Working Papers W99/10, institute
for Fiscal Studies, 1999.
K. Milligan - E. Moretti - P. Oreopoulos, Does education improve citizenship? Evidence from the United States and the United
Kingdom, Journal of Public economics, 88(9-10), (2004), pp. 1667-1695.
J. Mincer, Schooling, Experience and Earnings, Columbia university Press, New York 1974.
P.G. Moll, Primary schooling, cognitive skills, and wage in South Africa, economica, 65(258), (1998), pp. 263-284.
C. B. Mulligan, Galton versus the human capital approach to inheritance, Journal of Political economy, 107(6), (1999), pp. 184-224.
R.J. Murnane - J. B. Willett - Y. Duhaldeborde - J. H. Tyler, How important are the cognitive skills of teenagers in predicting
subsequent earnings?, Journal of Policy analysis and Management, 19(4), (2000), pp. 547-568.
M. C. Nussbaum, The costs of tragedy: some moral limits of cost-benefit analysis, in M. D. Adler - E.A. Posner (Eds.), Cost–Benefit
Analysis. Legal, Economic, and Philosophical Perspectives, university of Chicago Press, Chicago 2001, pp. 169-200.
Oecd, Education at a Glance 2002: OECD Indicators, oeCD Publishing, 2002.
R. A. Posner, Cost-benefit analysis: Definition, justification, and comment on conference papers, in M. D. adler - e.a. Posner (eds.),
Cost–Benefit Analysis. Legal, Economic, and Philosophical Perspectives, university of Chicago Press, Chicago 2001, pp. 317-341
G. Psacharopoulos, The profitability of investment in education: concepts and method, hCo Working Papers, 1995.
h. S. Richardson, The stupidity of the cost-benefit standard, in M. D. Adler - E.A. Posner (Eds.). Cost–Benefit Analysis. Legal,
Economic, and Philosophical Perspectives, university of Chicago Press, Chicago 2001, pp. 135-168.
T. P. Schultz, Health and schooling investments in Africa, Journal of economic Perspectives, 13(3), (1999), pp. 67–88.
T. W. Schultz, The Economic Value of Education, Columbia univ Press, New York 1963.
H. K. Siphambe, Rates of Return to Education in Botswana, economics of education Review, 19(3), (2000), pp. 291-300.
M. Spence, Job Market Signalling, Quarterly Journal of economics, 87(3), 1973, pp. 355-74.
D. Staiger - J. H. Stock, Instrumental Variables Regression with Weak Instruments, econometrica, 65(3), (1997), pp. 557-586.
P. Taubman, The Determinants of Earnings: Genetics, Family, and Other Environments; A Study of White Male Twins, The american
economic Review, 66(5), (1976), pp. 858-870.
D. Thomas, - J. Strauss, Health and wages: Evidence on men and women in urban Brazil, Journal of econometrics, 77 (1), (1997), pp.
159–185.
P. Trostel - I. Walker, - P. Wooley, Estimates of the economic return to schooling for 28 countries, labour economics, 9, (2002), pp. 116
J. Unni, Returns to education by gender among wage employees in urban India, Working paper n. 63, 1995.
M. Weale, A critical evaluation of rate of return analysis, economic Journal, 103(418), (1993), pp. 729–737.
© Nuova Secondaria - n. 4, dicembre 2014 - Anno XXXII
15