Intergenerational Persistence In Educational Attainment In .

3y ago
304.21 KB
40 Pages
Last View : 1m ago
Last Download : 6m ago
Upload by : Rafael Ruffin

DISCUSSION PAPER SERIESIZA DP No. 3622Intergenerational Persistence in EducationalAttainment in ItalyDaniele ChecchiCarlo V. FiorioMarco LeonardiJuly 2008Forschungsinstitutzur Zukunft der ArbeitInstitute for the Studyof Labor

Intergenerational Persistence inEducational Attainment in ItalyDaniele ChecchiUniversity of Milan and IZACarlo V. FiorioUniversity of MilanMarco LeonardiUniversity of Milan and IZADiscussion Paper No. 3622July 2008IZAP.O. Box 724053072 BonnGermanyPhone: 49-228-3894-0Fax: 49-228-3894-180E-mail: iza@iza.orgAny opinions expressed here are those of the author(s) and not those of IZA. Research published inthis series may include views on policy, but the institute itself takes no institutional policy positions.The Institute for the Study of Labor (IZA) in Bonn is a local and virtual international research centerand a place of communication between science, politics and business. IZA is an independent nonprofitorganization supported by Deutsche Post World Net. The center is associated with the University ofBonn and offers a stimulating research environment through its international network, workshops andconferences, data service, project support, research visits and doctoral program. IZA engages in (i)original and internationally competitive research in all fields of labor economics, (ii) development ofpolicy concepts, and (iii) dissemination of research results and concepts to the interested public.IZA Discussion Papers often represent preliminary work and are circulated to encourage discussion.Citation of such a paper should account for its provisional character. A revised version may beavailable directly from the author.

IZA Discussion Paper No. 3622July 2008ABSTRACTIntergenerational Persistence in Educational Attainmentin Italy*In this paper we show that there is a reduction in the correlation coefficient between fatherand children schooling levels over time in Italy. However, focusing on equality ofcircumstances, we show that there is still a persistent difference in the odds of attaining acollege degree between children of college educated parents and children of parents withlower secondary education attainment. The explanation of these trends lies in differentialimpact of liquidity constraints and risk aversion. Some descriptive evidence on the persistentdifferential in returns to college education depending on father’s education is also provided.JEL Classification:Keywords:J62, I38educational attainment, Italy, family backgroundCorresponding author:Carlo FiorioDepartment of EconomicsBusiness and StatisticsUniversity of Milan20122 MilanItalyE-mail:*We thank participants to the 2007 SIRE conference on Mobility in Edinburgh and to the 2006 Lowerconference on “Intergenerational mobility” held in Annency, the 2007 SIEP conference held in Pavia,seminar participants at University of Cagliari, the University of Milan and the Université Paris 1Panthéon-Sorbonne. We also thank Massimiliano Bratti, Helena Holmlund and David Margolis forfruitful discussions. Usual disclaimers apply.

1IntroductionItaly has often been depicted as a country with low intergenerational mobility,given the strong association existing between the socio-economic outcomes ofparents and their children as adults (Checchi, Ichino, and Rustichini 1999).In his review of existing cross-country comparative evidence, Corak (2006)laments the scarcity of Italian data. This paper aims to expand our knowledge on intergenerational mobility in Italy over the last century. Given theabsence of longitudinal data that span a su cient time interval, we focus oneducational outcomes based on children recall of parental education.The absence of longitudinal data sets allowing the measurement of intergenerational persistence in incomes has pushed some authors to followBjörklund and Jäntti (1997) in imputing incomes for the parents generation. For example, Mocetti (2008) adopts a two-sample two-stage strategyto estimate intergenerational correlation in incomes for Italy. He uses theSurvey of Household Income and Wealth data set conducted by the Bankof Italy (SHIW hereafter) nding that Italy is one of the most immobilecountry according to this methodology of measurement (with an intergenerational correlation in incomes as high as 0.84). When decomposing intergenerational mobility channels between returns to education and liquidityconstraints (preventing children from poor families to achieve higher education), he claims that 60.7% of persistence is attributable to the educationalchannel, i.e. the dependency of children education onto parental income.Piraino (2007) adopts a similar strategy to predict parental income in theSHIW data set and nds a high intergenerational persistence (in the orderof 0.48), where less than one third (28%) is attributable to the educationalchannel.1 However this procedure has limitations, as pointed out by Grawein Corak (2006): on one hand, measurement errors, related to both the imputation procedure and the imperfect recall of children, tend to bias downwardthe estimated income elasticity; on the other hand, the impossibility to control for varying age distance between the two generations make it impossibleto assess the direction and the extent of the bias.1Using ECHP (European Community Household Panel) data, Comi (2004) provides estimates of intergenerational mobility in educational attainment, nding that Italy exhibitsa quite low level of mobility. However, the sample of children is rather young, because avast majority of them is still cohabiting. On the contrary, Chevalier, Denny, and McMahon (2007) using IALS (International Adult Literacy Survey) survey ranks Italy high interms of intergenerational mobility in education.2

In the present paper we exploit information available in the SHIW oneducational attainment of children and parents to obtain a view on the longrun evolution of intergenerational persistence in Italy. Educational attainment has advantages and disadvantages with respect to income data. On thepositive side, it proxies the human capital endowment, which is positivelycorrelated to permanent income; in addition, it is less subject to imperfectrecall. On the negative side, it is unevenly distributed in the population,the probability mass being concentrated around the attainment of relevantdegrees (sheepskin e ects). However, given the absence of proper incomedata for Italy, we hold that advantages exceed disadvantages in providing anoverview of the Italian evolution across age cohorts.The use of data on educational attainment by parental background isnot new. In the 1990s Shavit and Blossfeld (1993) produced one of the rstcomparative studies of intergenerational persistence in education by studying the correlation of children attainment with parental background by agecohorts and claimed that the expansion of higher education gave no contribution to improving intergenerational mobility. While most of their chapterswere based on data sets where parental information originated from childrenrecall, Blanden and Machin (2004) use longitudinal data for the UK, ndingthat the recent higher education expansion has not been equally distributedacross people from richer and poorer backgrounds. Rather, it has disproportionately bene ted children from relatively rich families. Holzer (2006) studies the evolution of the association between college attendance and parentalincome over di erent age cohorts in Sweden, pointing out that new openingof local colleges has not improved the degree of intergenerational mobility.Similarly, Heineck and Riphahn (2007) nd that the association of childrenand parents educational attainment has not declined in Germany over thelast half of previous century.The frequent nding of a non declining association between children educational attainment and parental background has strengthened the idea ofsome sort of genetic link underlying educational choices. The idea of intergenerational transmission of ability, originally introduced by Becker andTomes (1986), has frequently reappeared as one potential explanation of thispersistence (see for example Cameron and Heckman 2001). However, moreaccurate tests of the "nature vs. nurture" hypothesis, based on data on IQtests, show that the relative impact of cognitive abilities is limited, and cannot account for the entire e ect of parental background (see the contributionscollected in Arrow, Bowles, and Durlauf (2000), and more recently in Bowles3

and Gintis (2002)). When the richness of data allows for the decompositionof intergenerational correlation of incomes into ability (further decomposedinto cognitive and non cognitive abilities), education and labour market attachment (as in Blanden, Gregg, and Macmillan 2007 for the UK), the main nding is that abilities account for a limited fraction of social immobility,while most of the e ect still passes through the educational attainment inthe children generation.2Due to the lack of data, we cannot test the extent of association betweenskill formation and parental background for Italy.3 In the sequel we studythe evolution of intergenerational persistence in educational attainments forItaly, and we decompose this correlation into a "liquidity constraint/riskaversion" component (children from poor families are prevented by enteringhigher education by lack of resources and/or di erent degree of risk aversion)and a "labour market" component (children from poor families have lowerexpected incomes, and therefore less incentive to get educated).The plan of the paper is as follows. In Section 2 the data are introduced and some descriptive evidence about trend of educational attainmentsis provided. In Section 3 a simple statistical model for the study of intergenerational transmission of education is discussed and the rst empirical resultsare presented, showing the decrease of the correlation between children andfather education over children age cohorts. In Section 4 we isolate the roleof intergenerational transmission of education as a component of the childfather education correlation and analyse its temporal evolution. Finally inSection 5 we provide some explanations and in Section 6 we conclude.2Data and background analysisFor analysing intergenerational transmission of education one needs to relyon data sets that collect information on the education of children and their2"The dominant role of education disguises an important role for cognitive and noncognitive skills in generating persistence. These variables both work indirectly through in‡uencing the level of education obtained, but are nonetheless important, with the cognitivevariables accounting for 20% of intergenerational persistence and non-cognitive variablesaccounting for 10%." (Blanden, Gregg, and Macmillan 2007).3Checchi and Flabbi (2006) make use of PISA test scores (as proxy for cognitive abilities) to analyse the relative contribution of ability and parental income in sorting intodi erent tracks at high school level. They nd that while in the case of Germany abilityis more relevant than parental education, the opposite situation occurs in Italy.4

parents across time. In Italy there are di erent data sets reporting this information (from international surveys like IALS or ALL to national surveys likeILFI (Indagine Longitudinale sulle Famiglie Italiane) or ISFOL-Plus), butthere is only one dataset with a su cient number of observations that allowsfor sample splits according to age cohorts. This is the Survey on HouseholdIncome and Wealth (SHIW) conducted biannually on a representative sample of the Italian population; since 1993 the surveys contain a section askinginformation on the householder’s and spouse’s parents when they were ofthe same age as the interviewees, including education, occupation and industry. In order to increase the degrees of freedom available, we pool SHIWwaves from 1993 to 2004, selecting only the householder and -when presenthis/her partner: we refer to it as the ‘children’generation, while informationon the ‘parent’ generation is obtained from their recall. After eliminationof repeated observations which belong to the panel section of the data,4 weremain with 45,682 children (21,241 males and 24,441 females) and 41,134fathers.5 Finally, the data set is organised by 5-year cohorts by children’sbirth years.Table 1 reports the highest education attainment of fathers and childrenorganised by children-birth-year cohorts. The percentage of children withno degree decreased constantly across time, the percentage of children withonly primary education increased over 50% for cohorts born during the 1920sand then it started to decrease in favour of lower secondary schooling. Anincreasing proportion of children attains a high school or a college degree: inthe last cohorts, over 40% of Italians have high school degree, slightly lessthan 40% have lower secondary degree and over 10% have a college degree.Although also fathers’education increased across time the average years ofeducation of fathers remains well below the average years of education ofchildren, the former being between two and ve years smaller than the latter.The increase of average education induced a reduction of inequality ofeducation as measured by any common inequality measure computed overthe years of completed education. However, these measures of inequalitymight blur the picture of intergenerational transmission of education across4The panel section of the SHIW data set was not considered as the attrition rate isvery large and we focus on education of adult population, which is in most cases constant(recall that we call children only householders and their spouses).5Information on mothers are also available (and we exploit them in Table 3) but giventhe gender discrimination in family educational choices in the grandparent generations,we prefer not to rely on them excessively.5

time, also because education has an upward bounded measure. Hence, werevert to the joint analysis of education of children and of their fathers, by agecohort of the child. Figure 1 represents the joint frequency of highest degreeattained by children conditional on his father’s using a plot where biggercircles means higher frequency. It clearly emerges that for children born in1911-1920 most of the mass was concentrated in the cell characterised bychild with no or primary education and father with no education, while ftyyears after most of the mass had moved to the cell where a child holds alower secondary or high school education and his/her father has primaryeducation. This movement of frequency mass was due partly to economicdevelopment and the increasing demand for educated labour and partly tothe accomplishment of compulsory education reforms.6The dashed line shows the interpolation of average years of child’s education conditional on father’s educational title, where no education, primary,lower secondary, high school and college degree are replaced with 0, 5, 8,13, 18 years of education, respectively. The interpolated line shows that theaverage child’s education conditional on father’s is almost linear and thatacross time it ‡attened but remained positively sloped, i.e. that the positivecorrelation of child’s-father’s education remains also in the younger generations although lower than for older ones. This descriptive evidence aims atanalysing the issue in more detail.3A conceptual framework and rst empiricalresultsThere is a vast literature on the intergenerational correlation of educationalachievements and/or incomes. Among the reasons for this correlation the literature considers genetic transmission, access to pre-school facilities, parentalcare, parental income and/or wealth, parental role model and out-of-school6Five years of compulsory education were actually introduced in 1862 (Legge Casati)but they were never accomplished, since it relied on local municipalities taking responsibility of school building, which they never did due to lack of resources. It was in theaftermath of WWII that the Italian government devoted earmarked resources to schoolbuilding, and this opened the way to school mass attendance. In 1962 three additionalyears of compulsory education were added, while postponing the allocation to tracks atthe age of 14. Two additional years, taking compulsory education to ten years, wereintroduced in 2007.6

cultural environment. Due to the frequent lack of retrospective informationin data, these studies are limited to the correlation between parents’schoolingand children schooling. This strategy is open to the criticism that parents’education is an inadequate measure of familiar background because it doesnot properly take into account the presence of liquidity constraints and ofthe out-of-school cultural environment. It also neglects the presence of peere ects and the quality of schooling. Unfortunately data often do not indicate the individuals’ birth place or the location of the school attended northey provide information on parents’income. Here we are forced to considerthat the intergenerational transmission of education achievement partiallyincludes all these aspects.To analyse the intergenerational transmission of education, one mightwant to estimate a regression such asSic Sif "i for i 1; :::; N(1)where, Sic ; Sif are education of child i and of his/her father i, respectively, "iis an error term and is the parameter of interest. The OLS estimate ofisc cf cf2ffwhere j ; cf are the standard deviation of errors for j c; f generationsand the correlation coe cient between child’s and father’s education. Onemay interpret a decreasing as a reduced intergenerational transmission ofeducation, however it might be solely due to a reduction in c f . As theratio of standard deviations decreased through time in Italy (see Table 2), wealso normalised years of schooling of child and father by the correspondingstandard deviation and estimate separately for each cohort the followingequation:7Sic cSif "i(2)fThe temporal evolution of the coe cient can be interpreted in terms ofcorrelation of child’s and father’s education and as a measure of inequality7In this equation we neglect assortative mating which should reinforce the e ect ofparents’education and the so called children quantity-quality tradeo according to whichmore educated parents have lees children but give them a better education. We alsoabstract from gender di erences in intergenerational persistence.7

of circumstances, which are independent on child’s e ort. A high estimateof would indicate that children schooling is heavily in‡uenced by parents’schooling (which may capture cultural or nancial constraints, as well as peerand network e ects), whereas an estimate close to zero would indicate thatchildren schooling is independent of family background. The main di erencebetween the coe cient in (1) and the coe cient in (2) is that the former,by considering the ratio of variances, takes into account also a change ofinequality of educational outcomes in children and fathers generations, providing a relative measure of intergenerational mobility. The latter providesan absolute measure of intergenerational transmission, i.e. depurated frompossible evolution of the distribution of educational attainments, for instancedue to school reforms that increased the average schooling of the population,reducing its variance. International evidence Hertz, Jayasundera, Piraino,Selcuk, Smith, and Verashchagina (2008) shows that in several countriesand coe cients behaved di erently.The review of the literature on the intergenerational transmission of education by Haveman and Wolfe (1995) concludes that parents’education is themost important factor in explaining children success at school. The pervasivequestion in the literature is whether the high correlation between parents’and children schooling is attributable to the genetic transmission o

run evolution of intergenerational persistence in Italy. Educational attain-ment has advantages and disadvantages with respect to income data. On the positive side, it proxies the human capital endowment, which is positively correlated to permanent income; in addition, it is less subject to imperfect recall.

Related Documents:

educational attainment data in 1940. A larger proportion of women than men had completed high school or more education. A larger proportion of men had received at least a bachelor's degree. Differences in educational attainment by race and Hispanic origin existed. Attainment for non-Hispanic Whites and Asians was higher than attainment

The Effect of Intergenerational Educational Attainment on Cardiometabolic Health in United States Latinos Emma G. Crenshaw1, Lindsay Fernández-Rhodes2,3 PhD, Chantel L. Martin 2,4 PhD, Penny Gordon-Larsen2,5 6PhD, Mary N. Haan DrPH, MPH, Allison E. Aiello2,4 PhD 1Department of Biostatistics, Gillings School of Global Public Health, University of North

tional attainment were based on data primarily from the CPS. 1. The ACS is now used as the main source of educational attainment data because it has a larger sample and provides more reliable statis - tics for small levels of geography. The report also provides estimates of educational attainment in the United . States, including comparisons by .

Ohio's educational attainment exceeds the national average at the high school level. In 2000, Ohio's high school credential attainment percentage was 83%, compared to 80.4% for the nation. Ohio's positive gap in high school attainment was largest for younger adults: 89% (OH) to 83.9% (US) in the 25-34 age group.

schooling, attainment of Amerindians and other ethnic/racial groups in the United States. He finds that indigenous people have among the lowest schooling attainment levels. Maoris, the natives of New Zeland, receive less schooling than whites (Brosnan 1984). Similarly, Gerber (1990) documents the low educational attainment

The 10 Indiana counties with the lowest educational attainment (averaging 10.2% of residents with a bachelor's degree or higher) have an average per capita income of 20,660. By comparison, the 10 Indiana counties with the highest educational attainment (aver-Figure 1 illustrates the relationship between educational attain-ment and earned income.

Table 4.4: Trend in educational attainment by race in percentages in Limpopo Province . 59 Table 4.5: Trend in educational attainment in Limpopo Province by gender in percentages . 60 Table 4.6: Probit results, reporting marginal effects for highest educational level of the

Annual Book of ASTM Standards now available at the desktop! Tel: 877 413 5184 Fax: 303 397 2740 Email: Online: Immediate access to current ASTM Book of Standards is available through our Online Version, which includes: Fast direct access to the most up-to-date standards information No limit on the number of users who can access the data at your .