EFFECTS OF DIVORCE ON CHILDREN'S EDUCATIONAL ATTAINMENT IN A MEDITERRANEAN AND CATHOLIC SOCIETY: Evidence from Italy

The aim of this paper is to contribute to the research on the effects of parental divorce on children's educational attainment in Catholic and Mediterranean countries. Many scholars from diverse disciplines have reported on the intergenerational effects of parental divorce: for instance on the educational and occupational level attained by children from divorced families, and on the subsequent risk of divorce. But there are indications that the intergenerational effects of divorce are not the same across all Western societies. As an example, Pong et al. (2003), comparing the educational results of pupils with divorced parents in 11 developed countries, found that the negative influence of single-parenthood is less detrimental in countries with favourable family policies. They also found a significant relationship between the prevalence of single-parent families and the achievement gap between children from two- and single-parent families. The achievement gap was greater in countries where single-parent families were more common. Another example of the differences in the intergenerational effects of divorce between countries is evident in the study by Engelhardt et al. (2002). Comparing West and East Germany, they found that the intergenerational effects of parental divorce were lower in East Germany despite its much higher divorce level, but that this difference disappeared by controlling for the differences in the religious composition of these two parts of Germany. In particular, the negative effect of parental divorce on children was less strong in the less religious and less traditional DDR than in the more religious and traditional FRD.

Despite the wealth of literature on the intergenerational effects of parental divorce, this research has predominantly been undertaken in countries with a strong Protestant history and culture – where divorce has long been a possibility – and in countries characterized by ‘weak’ family ties and a low degree of welfare familialization. Thus, most of this research has been done in the USA and in the north-western part of Europe: USA (McLanahan and Sandefur 1994; Amato 2001); Germany (Wagner 1997; Engelhardt et al. 2002), The Netherlands (Dronkers 1999; Goede et al.1999), Sweden (Murray and Sandquist 1990; Jonsson and Gähler 1997), and the UK (Cherlin et al.1991; Kiernan et al.1998; Ní Bhrolcháin et al.2000). Only recently the first studies on the intergenerational effects of parental divorce have been published for Catholic societies, e.g., for France (Martin 1997: 121; Archambault 2002) and Italy (Tomassini et al.2007; Albertini and Saraceno 2008). Thus, analysing the intergenerational effects for a country, such as Italy, characterized by the great relevance of the Catholic religion and a Mediterranean type of welfare state, may be fruitful, in order to avoid generalization based on just a few, not necessarily representative, societies.

There are many aspects, particularly relevant to the study of the intergenerational consequences of divorce, of Italian society that are quite different from Western and Northern European countries: the structural, cultural and institutional characteristics of its ‘family regime’; the relevant role played by the Catholic church and Catholic values in influencing family legislation, policies and culture; and also, more specifically, the legislation and the frequency and characteristics of separation and divorce.

Peculiarities of the Italian ‘family regime’ can be found in all of the three dimensions along which we can decompose it. Furthermore, they are intrinsically connected with the specificities of the Mediterranean welfare regime (Naldini 2003). Firstly, when we consider the structural characteristics of Italian families, or in other words the prevalent family forms in the country, we find that family models and living arrangements in Italy are deemed to have remained traditional and stable, especially when compared with the changes in household forms in other countries (Blossfeld et al.1995; Lesthaeghe 1995; Pinelli, 1995). As such, in Italy there continues to be a high percentage of women who are not in the labour market, a low rate of non-marital cohabitation, a relatively high rate of marriage, a low rate of divorce and a low number of births outside of wedlock (Crouch 1999; Rosina 2002; Saraceno 2003). The second dimension is that of family culture and values. Italy, according to the well-known distinction made by Reher (1998), belongs to the group of ‘strong’ family countries. Recent empirical evidence has also shown that young Italians consider their family to be one of the most important things in their lives (De Lillo 2002) and that Italian elderly population attitudes towards family solidarity are much more positive and ‘familialistic’ than in other European countries (Kohli and Albertini 2007). Finally, the traits of the Italian ‘family regime’ are also quite distinctive in its institutional dimension – i.e., that of family policies and, thus, of institutions such as family allowances, parental leave and care services. Clearly these institutions can also be considered to be part of the welfare state and in this sense the specificities of Italian family and welfare regime overlap. The Italian welfare system is characterized by a high degree of individuals’ welfare familialization, or what Leitner (2003) calls implicit familialism (see also Bahle and Pfenning 2000). The burden of supporting individuals’ economic and care needs is left, to a very large extent, on the shoulders of families. Legal obligations of solidarity within the family and kinship are highly extended (Esping-Andersen 1999; Saraceno 2000; Naldini 2003). In more general terms, with the relevant exceptions of the pension and health care system, the Italian welfare state results to be comparatively rather weak and ungenerous.

Another peculiar characteristic of Italian society is the important role played by the Catholic church, especially in family legislation and family policies, and in attitudes and behaviour towards family-related issues and, in particular, separation and divorce. The influence of the Catholic church and of its social doctrine on the design of the welfare state and of social policies has been very relevant (Castles 1994). The emphasis of this doctrine on the principle of subsidiarity has led the state to identify the family as the main actor which has the duty of providing social and financial support to dependent family members, thus limiting state intervention to cases in which the family ‘fails’ to provide such support (Trifiletti 1995). In addition, the Catholic church has strongly influenced family law, and particularly the legislation on civil marriage, divorce, and family solidarity obligations (Naldini 2003). Its influence also extends to the domain of individuals’ values and attitudes, particularly those regarding family and family life. According to a recent research report, more than 46 percent of young people have complete trust in priests, and a large majority of young people think that there is a strong social stigma against those people who separate/divorce, have an abortion or cohabit before marriage (Faccioli and Altieri 2002; La Valle 2002). Therefore, the ‘traditional’ approach of the Catholic church to the family – expressed, for example, in the Pope's letters Familiaris Consortio; Mulieres Dignitatem and Lettera alle Famiglie – is likely to play a stronger role in Italian society than in other Western countries. The ethical and moral values promoted by this institution certainly influence individuals’ behaviour; in particular it has been argued that religiosity has a strong effect on the separation and divorce rate in Italy (Barbagli and Saraceno 1998).

Closely connected to the above-mentioned specificities are the peculiarities of Italy as regards the legislation, frequency and characteristics of separation and divorce. The possibility of civil divorce, for example, has only recently been introduced in Italy. With the exception of a very short period under the Napoleonic regime, until 1970 only legal separation was allowed. Furthermore, unlike the majority of other Western countries, in Italy divorce is still a two-step process. Indeed, before being allowed to divorce spouses must undergo a 3-year period of legal separation – and about 40 percent of separations do not end in divorce (ISTAT 2001, 2002, 2005). The length and importance of legal separation in Italy is also exceptional among Catholic and Southern European countries (Naldini and Jurado 2008). It is worth noting, however, that most of the legal, social and economic effects of divorce – including decisions about child custody – take place at the moment of separation.¹ Italy also represents a peculiar case with respect to the frequency of separation and divorce. Although divorce and separation rates have markedly increased in Italy over the last two decades, these phenomena are still less frequent than in many other Western societies, including those in Southern Europe (ISTAT 2001, 2002, 2005; Naldini and Jurado 2008).

Finally, and also as a consequence of the specific Italian family and welfare regime and the role of Catholic church, the characteristics of the phenomenon of separation are quite different in Italy in contrast to other Western and Northern European countries. One major difference between Italy and other countries relates to the socio-economic status of those couples who separate. In fact, while the positive relationship between the rate of divorce and social class, which held during the nineteenth century, has changed in other countries during the twentieth century (Harkonen and Dronkers 2007), in Italy this relationship remained stable during the 1980s and only started to change very slowly during the 1990s.² A second peculiar characteristic derives directly from this late ‘democratization’ of separation: it is weakly associated with economic disadvantage. As a matter of fact, the over representation of employed women and of families from the highest social classes among the separated/divorced population explains why the risks of poverty associated with marriage break up in Italy are much weaker than those found, for example, in Anglo-Saxon countries (Del Boca 2002). This phenomenon is particularly striking if one also takes into account that the welfare provision for lone separated/divorced mothers is almost non-existent in Italy – whereas the welfare provision to lone widowed mothers is comparatively rather generous (Trifiletti 2007).³ Thirdly, in Italy children represent a ‘disincentive’ to separation. Indeed, separation rate is inversely – and strongly – related to the number of children. This negative relation is much weaker in other European countries. Finally, the high level of familialization of individuals’ well being, and the fact that separation nearly always affects the bonds within the extended family (whether the biological family or the family-in-law) negatively, can probably explain the finding that informal help received from the family in the case of economic difficulties following separation is less generous than the help afforded following a comparable decline in economic resources not related to marriage break-up (Sabbadini 2006; Trifiletti et al.2006).

Considering all the above-mentioned peculiarities, therefore, we might wonder if the results found in other European countries, in relation to the intergenerational effects of divorce, also apply in the Italian case.

The intergenerational effects of divorce on the educational attainment of children in European societies are generally negative and significant. Nevertheless, it is important to stress that they are not usually very strong. As mentioned above, our first aim in this article is to explore if Italian parents’ separation has negative effects on pupils educational attainment.

Another important aim of our research concerns the causes of the negative effect of parents’ separation on the educational attainment of their children. In considering possible explanations for this effect, we can identify two different perspectives that are not necessarily mutually exclusive. Firstly, many scholars have argued that economic deprivation, which is often a consequence of divorce (see for example McLanahan and Sandefur 1994; Sörensen 1994), is the main cause of the intergenerational effects of divorce. According to this perspective, family disruption dramatically reduces the economic resources available to the children, and consequently affects their educational attainment. In particular, it is often the case that the economic resources of the non-custodial parent are no longer accessible to the children. This is particularly harmful if the non-custodial parent, generally the father, was the main income earner. The second perspective on the lower educational attainment of children of divorced parents relates to the lower quality of the socialization process. According to this perspective, family disruption negatively affects parent–child relationships by reducing parental control and authority over the children, and thus the quality of the socialization process. This lower quality is partly a consequence of the time and energy devoted to the separation, but mostly because of the parental conflict before and after it (Dronkers 1999; Sun 2001). Here it should be stressed that while economic deprivation and lower parental control also affect, in principle, other types of lone parent families (i.e., unmarried and widowed mothers), the latter factor – parental conflict – is typical of separated parent families. Thus, if negative intergenerational effects of separation are also found in Italy, the second aim of this article will be to test whether economic poverty is the main factor explaining them. If the intergenerational effects of divorce are largely due to economic poverty resulting from the separation of partners, we would expect to find that once we control for income and wealth, the negative effects become smaller, or even insignificant. On the other hand, if poverty were not the main cause of intergenerational effects of separation, controlling for wealth and income would not cancel out these effects. It is worth noting, however, that due to data limitations we cannot test the hypothesis that the negative effects of separation on children's education arise not from poor economic conditions per se, but from the worsening of the economic well-being of the family after separation.

The third aim is to test whether the intergenerational effect of separation is different from the analogous effect of widowhood on the educational attainment of children, once income and wealth have been controlled for. If the former effect is larger than the latter, then we find support for the conflict explanation of the divorce effect. If, on the other hand, both effects are equal, it is likely that one common explanation underlies the problem of single-parenthood.⁴

The fourth aim is to test whether the Catholic religion and church have a significant effect on the consequences of divorce. However, since we do not have individual information on the religious attitudes of the parents, we have to rely on macro indicators of differences in religious behaviour and attitudes at the regional level. In line with the findings of Engelhardt et al. (2002), we assume that the stigma towards divorced people is stronger in the more religious and traditional regions of Italy and thus that the negative effect of separation on children is greater in these regions. Thus, our analysis will try to clarify if the degree of religiosity of the social context in which separation takes place has on impact on its negative intergenerational consequences.

Our research aims and the relevant literature lead us to the following ceteris paribus hypotheses:

• 1.

  The educational level of children, living with their separated/divorced mother, is lower than the educational level of children of married parents.

• 2.

  The educational level of children of widowed mothers is equal or lower to the educational level of children of married parents, but higher than the educational level of children of separated/divorced parents.

• 3.

  The negative effect of parental separation/divorce on the educational level of their children is equal for both lesser and highly educated mothers.

• 4.

  The greater the religious participation in a region, the stronger the negative effect of parental separation/divorce on the educational level of their children.

• 5.

  The higher the level of non-traditional attitudes in a region, the weaker the negative effect of parental separation/divorce on the educational level of their children.

This article uses data from the Bank of Italy Survey of Households’ Income and Wealth (SHIW); more specifically we have used the data provided in the Historical Archive (HA) of the surveyfor the following years: 1989, 1991, 1993, 1995, 1998 and 2000. Each wave of the survey consists of two sub-parts: a panel section and a cross-sectional one. The survey focuses predominantly on information regarding the economic status of households. This proves to be particularly useful when evaluating the hypothesis that poor economic conditions are the main cause of the intergenerational effects of separation. On the other hand, the SHIW-HA also provides information regarding other characteristics relevant to our study, for example the marital status of the household members and the educational attainment of each member.⁵

The unit of analysis we use is the child – all statistics reported below are corrected for clustering within the same household. In particular, we selected all children aged between 13 and 35 years who are still living in the parental home. Due to the low number of children of separated/divorced parents in the survey, we were forced to merge all individuals interviewed in the different waves irrespective of the moment that they entered or exited the survey and of the number of times they had participated in it. If one individual participates in more than one wave, the information considered is that provided by the last wave in which s/he participated. This solution has certain advantages: (a) it does not suffer from a possible bias due to non-random panel-dropout of the households (Quintano et al.1996; Brandolini 1999); and (b) it provides a much greater number of children living with separated/divorced, unmarried or widowed parents.⁶

Despite the merging of different waves of the survey, there are still very few children living with their separated/divorced, widowed or unmarried fathers. Consequently, they could not be analysed reliably as a separate group. Thus, our analyses have been further restricted to children living with their separated/divorced, widowed or unmarried mothers. Moreover, we do not use fathers’ characteristics in the analysis, because they are only available for married mothers and the very few cases in which the mother separated, or became a widow, while participating in the panel part of the survey.

It is worth noting that, due to data limitations, our analyses will suffer from two main limitations: (a) we only have information about the family condition after separation, whereas we do not have data about changes produced by separation. Therefore, for example, we cannot control our results for the change in the economic condition after separation; and (b) we cannot control our result for the father's educational and occupational status.

The educational attainment of the child is our dependent variable. The educational attainment of each family member is registered in the SHIW-HA by asking people which is the highest diploma they have obtained. Therefore, this variable does not take into account any current education or any course that they attended without achieving a diploma. The variable runs from ‘no degree or elementary school’, ‘middle school’, ‘high school’ to ‘bachelor's degree or higher’.⁷ The variable measures the child's qualification in the last survey in which s/he participated.

Our main independent variable is, of course, the marital status of the mother; this was registered in the SHIW-HA in all of the years considered. The data do not allow us to take into account desertion without a corresponding change in the legal marital status of the couple. Looking at those who participate more than once in the survey, the marital status of mothers can take a variety of different forms: some remain married for the whole period covered by the surveys, some separate, divorce and remarry, some separate and then declare themselves – or, at least are registered in the data set as being – unmarried, and so on. Therefore, we have constructed a new variable, labelled ‘family type’, to refer to the answers given in the surveys to the question of marital status. It is coded using the following procedure: when a mother indicates once that she is separated or divorced, she is coded as separated. If a mother indicates once that she is widowed and she never indicates that she is separated or divorced, she is coded as widowed. If a mother indicates once that she is unmarried and never indicates that she is separated, divorced or widowed, she is coded as unmarried. If a mother indicates that she is married and she never indicates that she is separated, divorced, widowed or unmarried, she is coded as married. We use married mothers as the reference category in the analyses.

Among other controlling variables we take into consideration household wealth and income. The total net wealth of a household is measured in the SHIW-HA as the sum of real assets and financial assets minus financial liabilities. Net household income is measured in the SHIW-HA as the sum of compensation of employees, pensions and other transfers, net income from self-employment, and property income. It should be stressed that the high rate of inflation in Italy in some of the years considered, as well as the fact that we are merging data from different years, necessitates that we correct wealth and income values in order to avoid any possible selection bias. To this end, we have utilized the relevant coefficients provided by ISTAT (the official Italian statistics office), which are based on the national index of prices (FOI), to compute the value of household wealth and income in the various sample years into 2002 tens of thousands of Euro. Consequently, the year in which these variables are registered does not bias our results.

In order to take account of the different cultural contexts in which children of separated and divorced parents are living we include in the analyses some macro level indicators. Italy is divided into 20 administrative regions. We use these 20 regions to characterize the social and cultural context of the households analysed. Three different variables are used to distinguish the social and cultural characteristics of the living environment of the child: the number of separations per 1,000 inhabitants; the percentage of inhabitants of the region aged over 5 who declared that they participate in religious activities at least once a week; and the number of voluntarily interrupted pregnancies per 1,000 women aged between 14 and 49 years. The values of these variables have been registered by ISTAT for the year 1998.

As reported in Table 1, the children of separated/divorced and unmarried mothers only represent, respectively, 3.5 and 1 percent of our sample while children of widows are more than double this figure. The death of the father, rather than separation, is still the most frequent cause of lone parenthood in Italy.

Females represent less than 50 percent of children in all types of families considered.⁸ The average age of the child is very similar in all types of families, with the exception of widowed families where the average age of the child is higher. The level of education of the mother varies considerably according to the family type, with separated/divorced and unmarried mothers being generally better educated than the others, while widows have, on average, a much lower educational level. Despite the fact that separated/divorced mothers are on average the best educated, they also head the poorest families, whereas widowed mothers, with the lowest educational levels, enjoy a better economic situation than separated/divorced mothers. The children of separated/divorced mothers are over represented in the child population of the northern regions. On the other hand, children living with widowed mothers are most likely to live in the southern regions. Finally, it should be noted that, generally speaking, children of separated/divorced mothers live in bigger municipalities and in regions where the number of separations is higher and the religiosity is lower.

The cross tabulation of children's educational attainment and the type of family is presented in Table 2. First, it should be noted that in all lone parent families there is an over representation of children with only an elementary school diploma. A further consideration is that the percentage of those living in separated/divorced families who get a diploma higher than that of middle school is clearly lower than that of children living with married parents or with a widowed mother.

The results of the ordered logistic regression analyses are presented in Table 3. The analysis begins with a model that only considers the effect of the mother's marital status on the educational attainment of the child. The results of this first model are fully comparable with those of Table 2. With respect to the reference category – married mothers – all of the other marital status categories considered are negatively associated to the child's educational attainment. However, only the children of separated/divorced mothers have a significantly lower educational level than those with married parents.

Model 2 includes the child's characteristics. Therefore, our basic hypothesis about the negative effect of divorce is tested against compositional effects. As expected, the age of the child is strongly and positively related to the level of diploma obtained. However, in this second model, there is an unexpected but important change in the coefficient related to the widowhood status of the mother: it becomes more negative and significant. This negative effect of widowhood was probably hidden by the age composition of the children of widowed mothers who are, on average, much older than other children. At this point in the analysis, widowhood seems to have a worse effect on a child's educational attainment than does separation. The effect of the latter is little changed by the introduction of the child's characteristics.

In model 3, variables accounting for the economic conditions of the family, family-size, and the educational level and occupational status of the mother are added to the equation. As expected, the educational level of the mother and income are positively related to the educational attainment of the child. This model can also be seen as a test for the hypothesis that the economic disadvantage associated with separation is the main cause of its negative intergenerational effects. If this hypothesis explains the negative effect of separation that we found in our first two models, this effect should become insignificant. On the contrary the results of our third model suggest that, once income, wealth and other mother's characteristics have been controlled for, the negative relation between the separated/divorced status of the mother and children's educational attainment becomes even stronger. Interestingly, the negative effect of widowhood hardly changes between model 3 and 2 and, unexpectedly, it is about the same size as that of separation.

The fourth step is to add to the control variables the region of residence of the household, and the social and cultural characteristics of the region of residence. The coefficients of the different marital status categories of the mother do not change significantly by adding these geographical characteristics. We added to this fourth model all interactions between regional religiosity, abortion rate and marital status. None of these interactions are significant, thus refuting our fourth and fifth hypotheses on the differential effects of divorce in regions with different levels of religiosity or non-traditional attitudes.

In model 5, we test the significance of the interaction between mothers’ educational level and their marital status. The two interactions, between mother's educational level and separated/divorced and widowed status prove to be significant. The evidence thus suggests that mothers’ educational level has a much more positive effect on the educational level of the child when the mother is separated or widowed. And yet, it should be noted that – considering the size of the coefficients – the mother has to have more than a high school diploma in order to neutralize the negative effect of separation. Figures 1 and 2 illustrate this result in another way. Specifically, children of lesser educated, separated/divorced mothers have a higher chance of ending up with a low educational level, while children of separated/divorced mothers with a bachelor v's degree or more have higher educational levels than children of married mothers with a bachelor degree or more. Similarly the children of widowed mothers with a high school education or more do not seem to suffer from the negative effect of the death of the father.

Thus, our first hypothesis (that the educational level of children, living with their separated/divorced mother, is lower than the educational level of children of parents still married) is accepted, but does not apply to children of separated/divorced mothers with a bachelor's degree. The second hypothesis (that the educational level of children of widowed mothers is equal or lower to the educational level of children of married parents, but higher than the educational level of children of separated/divorced parents) is only correct for widows with a middle school diploma or less. The third hypothesis (the negative effect of parental separation on the educational level of their children is equal for lesser and highly educated mothers) is clearly not correct. Our fourth and the fifth hypotheses on the regional differences in the effect of marital status on children's educational achievement are not supported by our data.

We summarize our results with a discussion of our hypotheses. The first hypothesis, which assumes that the educational level of children of separated/divorced parents living with their mother is lower than the educational level of children of parents still married, is fully endorsed by our results. This negative result cannot be explained by the more (at present) difficult economic and social situation of separated/divorced families, nor by the negative impact of the cultural climate of their region. Controlling for these factors increases the negative effect of separation instead of decreasing it.⁹ The main explanation for this unusual increase is that, in Italy, separation with children mostly occurs among parents with sufficient resources. This ‘bias’ of separation by the higher social strata in Italy partly ‘hides’ its negative effect on the educational outcomes of their children. The acceptance of our first hypothesis suggests that economic deprivation is one, but not the most important, cause of the negative effects of marriage break-up on children.

The second hypothesis is not fully supported by our data. Children of widowed mothers with less than a middle school diploma have lower educational outcomes than comparable children with married parents with the same economic and social resources. However, the educational levels of children with widowed mothers with a bachelor degree or more are higher than those of comparable children living in a household with their married parents. So, living without a father does not need to be negative for children, but the mother needs to have sufficient socio-economic resources and human capital to compensate for the weak public welfare provision and, probably, the loss of a part of the support of her wider family. The weakness of the Italian welfare system, particularly in care services provision, and the importance of family support for individual well-being in Italian society, might also explain the negative effect of widowhood on children's educational outcomes. A comparable negative effect of widowhood is not found in other European societies with more generous and less familialized welfare regimes. The only exception is a Spanish analysis of the effect of widowhood and divorce ( Martinez 2002), but Spain also exhibits a weak welfare system and a strong emphasis on the family as a means of support in difficult times (Kohli and Albertini 2007).

The third hypothesis, which assumes that the negative effect of parental separation on the educational level of their children is equal for lesser and highly educated mothers, must be rejected. Separated and divorced mothers with at least a bachelor degree are able to fully neutralize the substantive negative effect of separation on the educational level of their children. This also means that the children of the least educated separated/divorced mothers have the largest educational arrears compared with children of married but similarly less educated mothers. In other words, the negative effect of separation on children is for the separated families with the lowest social and cultural resources. One explanation for this might be that parents who already have fewer social and cultural resources find it more difficult to cope with the negative aspects of separation – such as: parental conflict before and after separation; the partial loss of parental authority; the lower quality of socialization; the partial loss of support from the wider family – and have more difficulties in accessing the limited public welfare resources available. The stronger effect of the educational level of widowed mothers and the possibility of highly educated widows who can fully neutralize the negative consequences of the father's death can be interpreted as an indication of the more generous support afforded to widows mothers (over separated ones) by families, community, and the welfare system (Trifiletti 2007). Separation represents a stronger break with the family-oriented character of Italian society, while it denies the eternal bonds within the (extended) family and thus produces a larger loss of social and cultural resources than separation/divorce in more individualistic European societies or widowhood in Italy.

The rejection of the two last hypotheses on the importance of regional indicators of religiosity or non-traditional attitudes indicates that the effect of separation is equal in all regions of Italy. This result suggests that the family-oriented character of Italian society is not restricted to some backward regions or to their particular characteristics. It also suggests that the negative effect of separation will not decrease as soon as an Italian region becomes less traditional or religious and more modern or individualistic. Nor do the results support the thesis that religious differences in Catholic practice and attitudes are important for understanding the different outcomes of separation; from this point of view, the Catholic character of Italy seems to be less important for these outcomes.

The clear intergenerational effect of parental separation shows that Italy, in this respect, does not deviate strongly from other European societies. The Catholic character of the country does not seem to be a reason for a deviant Italian case. The only deviation might be as the result of the Mediterranean type of welfare state, with its strong familialization of individuals’ welfare and poor levels of benefit compared to the more individualistic northern European welfare states. This makes the consequences of separation more negative in Italy, because it often also means the partial loss of the help of the family.

Marco Albertini is Junior Researcher at the Instituto Juan March in Madrid, Spain. He has a PhD from the European University Institute. His research interests include family sociology, income inequality, and intergenerational relations and transfers. Current research projects are on health inequalities, intergenerational effects of divorce and the generational contract.

Jaap Dronkers is Professor of Social Stratification and Inequality at the European University Institute in Italy. He published mostly on causes and consequences of unequal educational and occupational attainment, effect-differences between public and religious schools, education of elites, educational and occupational achievements of migrants, and the effect of parental divorce on their children.