Abstract
In this article we study some of the overlooked mechanisms of how social origin affects an individual’s occupational success at labour market entry. The empirical analysis draws on four Dutch retrospective life-course surveys. The analytical sample consists of 6,416 respondents born in the period 1931–80. The results show intergenerational transmission of occupational status, but the effect of father’s occupation on his child’s first job has declined over time. Part of the decreased impact of social origin is related to the increased age at first job. Educational expansion prolonged the school career of individuals and increased the age at which they first enter the labour market, when parental control and influence matter less. The effect of level of education has decreased over time too, at an even faster rate. Because of that, a trend from ascription to achievement cannot be confirmed. Furthermore, we found evidence of a weaker impact of social origin on occupational status attainment for the higher educated. The labour market for higher levels of education is more meritocratic, as employers use the degree to which various educational programmes at these levels provide occupation-specific skills to evaluate the labour productivity of potential workers.
Introduction
Social origin is still an important determinant of occupational success. Despite predictions from modernization theory (Treiman, 1970), it remains to be seen whether the direct impact of social origin has declined over time relative to own level of education. We argue that in order to reach more substantive conclusions with respect to trends in social origin effects, social stratification sociologists should pay more attention to how the effect of social origin depends on own educational achievement and how these trends could be explained. In this article, we scrutinize some of the overlooked mechanisms of modernization theory and test, first, whether they can explain change over time in the importance of social origin and, second, how the relevance of social origin depends on educational attainment levels.
Stratification scholars consider an individual’s occupational position as a very important (if not the most important) characteristic of social standing. The question how individuals attain their occupational position, therefore, has been on the sociological agenda for a long time now (Ganzeboom et al., 1991). The two most studied determinants of occupational success are social origin and educational attainment. Both are incorporated in the by now classic status attainment model of Blau and Duncan (1967). This path model makes clear how father’s occupational status (as a measure of social origin) is transmitted to that of their children. On the one hand, intergenerational transmission is direct by the immediate transfer of proprietorship or family business, by providing their children with occupational aspirations and by offering them the right social and cultural capital (Ganzeboom and Luijkx, 2004). On the other hand, there is an indirect intergenerational transmission of occupational status via level of education; children from high social origin achieve, on average, a higher level of education than those from low social origin. Primary and secondary effects of social stratification are responsible for this (Jackson et al., 2007). Therefore, children from high social origin attain, also indirectly via their education, a higher occupational status.
Next to the mechanisms of intergenerational transmission of occupational status, the focus in social stratification research is on historical developments in the paths by which social origin is converted into (dis)advantages for their children (Treiman and Ganzeboom, 1990). In particular, trends in the mediating role of education in the intergenerational transmission of occupational attainment are a central issue. To what extent has the inheritance of occupational positions channelled through education changed over time? Related to this is the question whether the effect of social origin on occupational attainment – over and above what is mediated by education – has altered. This change in the direct effect of social origin may be the result of a weakening intergenerational transmission of parental resources to their children and/or decreasing labour market returns of these resources, such as propriety, and social and cultural capital.
Modernization theory, widely used in social stratification research, has it that modern societies are characterized by efficient allocation and selection processes in the labour market. Selection less often takes place on the basis of ascribed characteristics, such as social origin and its related assets. Instead, occupational attainment is increasingly based on merit or achieved properties, in particular occupation-specific knowledge and skills. Education is, therefore, considered as the single most important characteristic in the selection and allocation process in modern labour markets (Blau and Duncan, 1967; Treiman, 1970).
It is claimed that this trend from ascription to achievement is caused by technological developments in modern labour markets, which have led to a shift in demand from a low-skilled to a high-skilled labour force (Kerr et al., 1960). This process of skills upgrading is known under the heading of ‘skill biased technological change’ (Levy and Murnane, 1992). In particular, the introduction of automation and computerization in post-industrial or ‘knowledge’ societies has resulted in a changing occupational structure. The changing occupational structure led to increasing educational requirements on the labour market. The skills acquired in education represent human capital (Becker, 1964). Investments in human capital are useful so long as they lead to higher productivity on the labour market. Employers value labour productivity by offering the best labour market positions (and wages) to those individuals who have obtained most human capital. When technological developments lead to a growing demand for high-skilled labour, employers increasingly select on, and individuals increasingly invest in, educational credentials. Employers no longer base their estimations of employees’ labour market productivity on social origin. As a consequence, economic modernization has resulted in a rapid increase in educational attainment in advanced societies.
Besides economic modernization, cultural modernization took place in the form of a shift in dominant value patterns. In modern societies, particularistic value patterns, which were predominant in pre-industrial societies, have been replaced by universalistic ones. Also values with respect to status attainment have changed (Blau, 1962). Nowadays, individuals no longer accept being judged on the basis of their social origin; they believe that status positions should be distributed based on achievement and merit. Moreover, modern societies have become bureaucratic. The rise of organizations has enlarged the possibilities of control and formal procedures of recruitment (Weber, 1922). For that reason, processes of selection and allocation in the labour market are more likely to follow universalistic criteria. Thus, as a result of cultural modernization and changing social norms, employers no longer want to select, and are no longer expected to select on social origin. Moreover, bureaucratization limited opportunities to select on social origin.
Cultural modernization has also led to institutional reforms aimed at making educational systems more meritocratic. In particular, the decreasing direct and indirect costs of education and the increasing age of compulsory education have further triggered educational expansion and reduced inequality of educational opportunity (Tieben and Wolbers, 2010). Not only has social origin become less decisive for one’s educational degree, at the same time educational expansion may have weakened the discriminatory power of qualifications as educational degrees lost their differentiating ability. Furthermore, the value of educational credentials, and hence education’s role in intergenerational inequality, depends on the balance between demand of skills, that is, the occupational structure, and the supply of graduates with these employable skills (Hout, 2012). For The Netherlands, there is evidence that the labour market has not been able to absorb the ever higher educated labour force population in jobs that require their skills (Wolbers et al., 2001).
In sum, a gradual decline of the direct effect of social origin on occupational attainment (arrow A, Figure 1) and on educational achievement (arrow B, Figure 1) is predicted. The direct impact of one’s education on occupational attainment (arrow C, Figure 1) is expected to increase. Hence, when the direct impact of social origin declines and of education rises, we find support for modernization theory. However, as stated above, the direct impact of education on occupational attainment may decrease, because – as a result of modernization and the accompanying educational expansion – educational degrees have lost their differentiating ability and the supply of highly educated with occupation-specific skills may have grown faster than the demand for these skills. When the impact of social origin declines relative to that of education – although both (absolute) trends of the direct effects may be negative – this is also seen as confirmation of modernization theory, as this would similarly signal a trend from ascription to achievement. 1
Top panel: Classic status attainment model. Bottom panel: Status attainment model including moderated direct social origin effect. Note: Closed arrows represent main effects. Dashed arrow represents moderation effect.
In addition to existing economic and cultural explanations for the (expected) trend from ascription to achievement, it is claimed that the labour market for higher educated individuals is more meritocratic than other labour markets (Torche, 2011). This would imply first that the influence of social origin on occupational attainment is weaker among the higher educated (arrow D, Figure 1, bottom panel) and, second, that for society at large the impact of social origin decreases because more and more higher educated are allocated to the ever-growing higher segments of the labour market (Hout, 1988; Mastekaasa, 2011). In other words, it needs to be acknowledged that education not only mediates the effect of social origin, but also that the impact of social origin is moderated by it. Educational expansion may thereby explain, in part, why the impact of social origin declines over time: there are more individuals with a higher education for whom social origin matters less (Breen and Jonsson, 2007).
In this article, we first of all investigate whether there (still) is a direct, positive effect of social origin on occupational success in The Netherlands over and above the effect of own level of education. Second, we determine whether the direct impact of social origin has declined over time and whether this has been accompanied by a trend from ascription to achievement, as modernization theory predicts. Third, we scrutinise to what extent the effect of social origin is conditional on one’s level of education, where we expect it to be weaker among those with higher education. Here, we also explore to what extent this moderation effect explains the trend in the impact of social origin. Fourth, we test several explanations for the expected weaker impact of social origin over time and for the higher educated.
To answer these research questions, retrospective life-course data from the Family Survey Dutch Population are analysed. We use four rounds of this survey (as held in 1998, 2000, 2003 and 2009), which involves random samples from all Dutch inhabitants and face-to-face interviews with respondents at home. Occupational success refers to the occupational status attained at labour market entry, that is, when individuals have equal (that is, no) labour market experience. This enables us to determine genuine historical developments in status attainment by comparing birth cohorts. Moreover, it can be expected that the effects of social origin and education on occupational success are most pronounced at the start of the working career (Shavit and Müller, 1998). In the multivariate analysis, linear regression models are applied to the data.
Apart from contributions to the Dutch case, as well as to the social stratification literature more generally, this article improves earlier research in at least one very important way. By empirically investigating the mechanisms that explain both a weaker impact of social origin over time and a smaller effect of social origin for higher levels of education (see our fourth research question), it offers direct tests of modernization theory. Prior research has only provided indirect support for modernization theory by presenting empirical evidence of a trend from ascription to achievement (see, for instance, de Graaf and Luijkx (1993) for The Netherlands, Grusky and DiPrete (1990) for the United States, Jonsson (1996) for Sweden, Luijkx et al. (2002) for Hungary and Marks (1992) for Australia). However, the basic explanatory elements of this theory leading to meritocracy (that is, the increased focus on productive skills in the process of technological change and the decreased relevance of social origin in recruitment procedures in the process of bureaucratization) have never been put to a serious empirical test (Bernardi, 2012). In this article, we improve upon this by adding operationalizations of these elements (firm size, age at labour market entry, acquired occupation-specific skills) to the classic status attainment model in order to test the predictive validity of modernization theory. In particular, the role of occupation-specific skills acquisition in explaining the moderation of the social origin effect on occupational success by educational attainment levels is innovative and, as it turns out in the empirical analysis, promising.
Hypotheses
For The Netherlands, several authors have found a direct, positive impact of social origin on occupational attainment, over and above the effect of education, even in recent years. This conclusion holds for both the individual’s first occupational position (de Graaf and Ultee, 1998; Tolsma and Wolbers, 2010a; Wolbers et al., 2011) and his or her current or last one (de Graaf and Luijkx, 1993; Hendrickx and Ganzeboom, 1998; Ganzeboom and Luijkx, 2004). So, the first hypothesis – based solely on prior research – simply reads:
Empirical evidence for a trend from ascription to achievement in The Netherlands is found in de Graaf and Luijkx (1993). These authors established that the direct impact of father’s occupation on the occupational status of his child(ren) declined. Also the indirect effect of father’s occupation, mediated by the level of education of his child(ren), decreased in the course of time, because the impact of social origin on educational achievement decreased faster than the direct impact of education on occupational attainment increased. In total, the intergenerational transmission of occupational status reduced by some 40 per cent between 1930 and the early 1990s. Other studies for The Netherlands confirmed the declining direct effect of social origin on occupational status attainment (Hendrickx and Ganzeboom, 1998; Ganzeboom and Luijkx, 2004).
The recent Dutch studies of Tolsma and Wolbers (2010a, b) described more nuanced results. They showed that the decrease in the direct impact of social origin on the occupational status of the current or last job was not significant and that the (direct) impact of social origin compared to education even increased, contrasting the predictions of modernization theory (Tolsma and Wolbers, 2010b). That said, in a different study they did observe decreasing direct effects of social origin on the occupational attainment of the first job, albeit no decrease in the relative impact of social origin compared to education for men (birth cohorts 1914–1969) (Tolsma and Wolbers, 2010a). Also Wolbers et al. (2011) did not find trends in the (direct and relative) effect of father’s occupational status on his son’s occupational status; nor at labour market entry, neither after 10 and 20 years of working experience. 2
Despite this mixed empirical evidence for The Netherlands, it is too early to draw a final conclusion about the predictive validity of modernization theory, in particular because the assumed mechanisms of modernization theory are hardly ever directly tested (Bernardi, 2012). In this article, we offer two interpretations for the predicted decline in the social origin effect on occupational attainment. First, more and more workers in The Netherlands are employed in larger, and hence, more bureaucratic organizations (Dekker et al., 1995). Larger organizations more often have formal, more bureaucratic, hiring practices (Weber, 1922) and we assume that, especially during formal hiring practices, educational qualifications are seen as the most reliable signals of future labour productivity, and social origin does not play a major role (anymore) (Breen and Jonsson, 2007). Second, as a consequence of skill biased technological change and, related to that, educational expansion, individuals start their occupational career at a later age. Just as the impact of social origin on each subsequent transition point of the educational career declines (see, for The Netherlands, for instance, de Graaf and Ganzeboom, 1993), in part because parental control and influence decrease during the school career (Müller, 1990), we expect it to decline with the final transition onto the labour market, which is made at ever later ages. In sum, we hypothesize that:
Similarly, it is to be expected that the impact of social origin on occupational attainment is weaker with increasing levels of education (Hout, 1988; Mastekaasa, 2011), because higher educated workers are more often employed in larger and hence more bureaucratic organizations than lower educated (Dekker et al., 1995) and because higher educated start their occupational career at a later age. An additional reason to expect a weaker impact of social origin for higher educated individuals is that they, more often than lower educated counterparts, attended occupation-specific educational programmes; there are no vocational tracks in elementary school or at lower general secondary education. Vocational tracks supply students with occupation-specific skills that are in demand on the labour market, as stakeholders determine in part the content of the curricula (van der Velden and Wolbers, 2007). We further assume that higher educated individuals have more recognizable skills (that is, followed more occupation-specific tracks) and more employable skills than lower educated. The more (recognizable and employable) skills are, the less relevant social origin, or rather the cultural, social and economic assets that come with social origin, is in the selection and allocation process of individuals in the labour market. Thus we expect:
We would like to point out that we offer (partly) the same mechanisms to explain both a weaker impact of social origin over time and a smaller effect of social origin for higher levels of education. Thus, even for each level of education, we may expect that the impact of social origin declines, as both the lower and higher educated alike have come to work in more bureaucratic organizations and entered the labour market at a later age. That said, previous researchers may have overestimated the declining impact of social origin on occupational attainment over time, because they did not take into account the moderating role of education. If the association between social origin and occupational attainment is smaller for higher levels of education, the gross association between origin and destination may be expected to decrease as the educational system expands, since this expansion is associated with more individuals reaching higher levels of education, and the destination (that is, occupational attainment) of these individuals is less influenced by their social origin. Breen and Jonsson (2007) refer to this as the compositional effect of educational expansion. To what extent the presumed, declining impact of social origin is, on the one hand, due to a weaker impact of social origin over time for each educational category and, on the other hand, due to a combination of educational expansion and a smaller effect of social origin for the higher educated remains to be investigated. However, we do expect that:
Data and measurement
To empirically test these hypotheses, retrospective life-course data from the Family Survey Dutch Population (FSDP) are analysed. We use four rounds of this survey (as held in 1998, 2000, 2003 and 2009), which involves random samples (de Graaf et al. 1998, 2000, 2003; Kraaykamp et al., 2009). The sampling population is the Dutch (speaking) population aged between 18 and 70 years. Primary respondents and their partner have been interviewed using the same structured questionnaire. Information was collected by means of face-to-face interviews with respondents at home. Response rates varied around 45 per cent. For the current analysis, we selected respondents from 25 years and older, as many (higher educated) individuals do not finish their educational career before this age. The analytical sample consists of 6,416 respondents.
In the multivariate analysis, occupational success at labour market entry, operationalized as the occupational status attained in the first job, is the dependent variable. Combined with the retrospective character of the FSDP, this enables us to determine genuine historical developments in status attainment by comparing birth cohorts, on the (likely) assumption that the retrospective data are accurate (de Vries and de Graaf, 2008). The occupational status is measured by assigning to each job title a score from the International Socio-Economic Index of Occupational Status (ISEI) (Ganzeboom et al., 1992). This scale ranges from 16 for occupations with the lowest to 90 for those with the highest status.
Various independent variables are included in the multivariate analysis. Level of education concerns the highest level obtained and is categorized in the following way: (1) primary education (Lo), (2) secondary education, general (Vo), (3) secondary education, vocational (Mbo), (4) tertiary education, vocational (Hbo) and (5) tertiary education, university (Wo). For descriptive purposes, we also defined level of education as the number of years in education (according to the nominal duration of the various educational programmes), ranging from 6 years for primary education (Lo) to 16 years for university education (Wo). Social origin is defined as father’s occupation when the respondent was 15 years old. Like the respondent, it is based on the ISEI scale. Gender differences are captured by an indicator variable that takes the value 1 for male respondents. To determine trends in the effects of social origin and education on occupational success at labour market entry, we compare birth cohorts. For the descriptive analysis, successive 10-year birth cohorts are distinguished: 1931–40, 1941–50, 1951–60, 1961–70 and 1971–80. In the multivariate analysis, we simply include year of birth. Size of first workplace is based on an interval scale and refers to the size of the firm where individuals were employed during their first job. This information is provided by the respondent him or herself. Age at first job is measured in months. To each educational programme (that is, field of study), the values of four scales that refer to cultural, economic, communicative and technical skills were assigned. These scales were constructed by van de Werfhorst (2001). In designing the scales, he asked respondents who participated in the 1998 round of the FSDP to indicate the extent to which their educational programme devoted attention to a list of 16 kinds of knowledge and skills. As we only have this individual-level information for the 1998 round, we use the aggregated scale values for each field of study and have to neglect possible heterogeneity in skills within each field of study and assume that the skills students acquired within each field of study are constant over time. Students with different levels of education, but with the same field of study are assigned the same level of occupation-specific skills. In this way, the skills variable cannot pick up an educational level effect. Higher scores indicate that students acquired more occupational skills. We acknowledge that this cannot directly be equated with the occupational specificity of the skills or with the employability of these skills. We decided to use these four skill variables and not the original field of study variable, because it allows for more parsimonious models and it is a more substantive measure; we see what it is of field of study that drives effects.
With the exception of gender, level of education and birth year, all variables (including the dependent variable) are z-standardized in the multivariate analysis. Birth year is centred around 1960. These centred scores are divided by 10 so that, when studying trends in the effects of social origin and education, we compare 10-year time spans.
Results
From Table 1, it becomes clear that in The Netherlands an enormous educational expansion has taken place. Whereas for the cohort born between 1931 and 1940 the average level of education is 9 years, for the youngest cohort, born in the period 1971–80, this level is 12 years. Although Tolsma and Wolbers (2010a, b) showed that intergenerational downward educational mobility is increasing (especially among men), there is no clear indication that the increase in the mean educational level has come to a halt. Also note that gender inequality in educational achievement has reversed in the course of time. Whereas women were in a disadvantaged educational position in the past, they are nowadays higher educated than men.
Level of education and occupational status by gender and cohort.
Source: Family Survey Dutch Population 1998, 2000, 2003 and 2009.
Due to an increasing demand for higher educated employees, the occupational structure has upgraded in The Netherlands over time. Table 1 shows that the average occupational status of the first job of respondents has considerably increased over time: from around 39 status points for the oldest birth cohort to somewhat less than 46 points for the youngest.
Despite the skills upgrading of the occupational structure, it becomes clear that this process could not keep up with the educational expansion The Netherlands witnessed (see also Wolbers et al., 2001). The decrease in the ratio between the mean level of education and the mean occupational status over time (from 4.40 for the oldest birth cohort to 3.80 for the youngest) suggests that the occupational returns to education have declined. This finding is a first indication that educational returns may have diminished.
To test our first hypothesis, we regressed the occupational success at labour market entry on social origin and level of education, while controlling for gender and birth year. In Model 1 of Table 2, we first of all observe intergenerational transmission of occupational status: father’s occupational status has a positive impact on his child’s. For every standard deviation increase in father’s occupational status his child’s occupational status increases by 0.129 standard deviation. This finding supports hypothesis 1. Next to father’s occupational status, we find a strong, positive effect of education on occupational success at labour market entry. To summarize the effects of the dummy variables for education, we calculated a sheaf coefficient, which is a standardized regression coefficient that treats multiple indicators as if they were one. The implied effect is 0.463 (not shown). So, the explanatory power of level of education is three and a half times greater than that of father’s occupational status (0.463 / 0.129 = 3.589). This finding supports modernization theory, which posits that education matters more than social origin with regard to occupational status attainment.
Results of linear regression analysis of occupational success at labour market entry: Testing the main hypotheses (N = 6,416).
*p < 0.05; **p < 0.01 (two-tailed tests).
Source: Family Survey Dutch Population 1998, 2000, 2003 and 2009.
To test our second hypothesis on the changing impact of social origin, we estimated Model 2, in which the statistical interaction term between father’s occupational status and birth year was added. We observe a negative interaction effect (b = –0.051), implying that the effect of father’s occupational status is smaller for those who were born more recently. Thus, we conclude that the impact of social origin has declined over time, which is in line with the general prediction of hypothesis 2.
The explanation for this declining impact is investigated in Table 3. The parameter estimates presented in Model 1 are a duplicate of the results of Model 2 in Table 2. In Model 2 of Table 3, it is shown that the size of first workplace has a positive effect on occupational success at labour market entry. Moreover, the correlation between the size of first workplace and birth year is, as assumed, positive (r = 0.126; p < 0.01; not shown). This means that our auxiliary assumptions are valid. Nevertheless, the explanatory power of the size of first workplace is limited. After all, the declining effect of social origin on occupational success at labour market is hardly reduced after controlling for the size of first workplace (and the interaction between father’s occupational status and the size of first workplace). The effect drops from –0.051 in Model 1 to –0.049 in Model 2. So, we conclude that hypothesis 2a cannot be corroborated.
Results of linear regression analysis of occupational success at labour market entry: Testing the explanatory mechanisms of hypothesis 2 (N = 6,416).
*p < 0.05; **p < 0.01 (two-tailed tests).
Source: Family Survey Dutch Population 1998, 2000, 2003 and 2009.
The second presumed explanation for the decreasing effect of social origin on occupational success at labour market entry is related to the age of first entry into the labour market. This explanation is tested in Model 3 of Table 3. 3 Given the increase in level of education over time and the fact that higher educated enter the labour market at a later age, the correlation between age at first job and birth year is positive, but not problematically high (r = 0.309; p < 0.01). Age at first job has a positive effect on the occupational status attained in this job (b = 0.238). Furthermore, the effect of father’s occupational status on occupational success at labour market entry is, as predicted, smaller for those who entered the labour market at a later age, as indicated by the significant interaction term (b = –0.033). Taken together, these results lead to a reduction of the interaction effect between father’s occupational status and birth year. The interaction term now is –0.033. In other words: the declining effect of social origin on occupational success at labour market entry over time is reduced by some 35 per cent (1 – (–0.033 / –0.051)) after taking the age at first job into account. This finding lends considerable support for hypothesis 2b.
In Model 4 of Table 3, both the size of first workplace and age at first job (and their interactions with father’s occupational status) are added. The results show that the declining impact of social origin on occupational success at labour market entry is not substantially further reduced compared to Model 3.
Our third hypothesis states that the impact of social origin on an individual’s occupational success at labour market entry is conditional on his or her own educational attainment, namely that it is weaker among those with higher educational levels. We tested this hypothesis by estimating Model 3 as presented in Table 2. The results of this model indicate that the effect of father’s occupational status is strongest for labour market entrants without any diploma (b = 0.267) and that it is significantly smaller for higher educated. For university graduates, for instance, the impact of father’s occupational status is 0.105 (0.267 – 0.162). This implies that the descriptive part of hypothesis 3 is supported by the data.
We expected that the impact of social origin on occupational success is weaker for the higher educated, because the higher educated more often work in larger firms, enter the labour market at a later age and have more occupation-specific skills. Controlling for these mechanisms should hence reduce the observed negative interaction between social origin and education. We test these mechanisms in Table 4. Although it is observed that the size of first workplace (see Model 2) and the age of first job (and its interaction with father’s occupational status) (see Model 3) matter with respect to occupational success at labour market entry, these characteristics do not explain the negative interaction between father’s occupational status and level of education. The size of the interaction terms have hardly changed after the introduction of these variables. This implies that hypotheses 3a and 3b cannot be confirmed.
Results of linear regression analysis of occupational success at labour market entry: Testing the explanatory mechanisms of hypothesis 3 (N = 6,416).
*p < 0.05; **p < 0.01 (two-tailed tests).
Source: Family Survey Dutch Population 1998, 2000, 2003 and 2009.
The picture completely changes when taking the knowledge and skills as generated by the obtained educational programme as an explanatory factor into account (see Model 4). All four parameters that capture the interaction between father’s occupational status and level of education are now insignificant. This finding implies that the labour market for higher levels of education (that is, [intermediate and higher] vocational education and university education) is meritocratic and that employers use skills to evaluate the productivity of potential workers. Note that all skill variables have a positive effect on occupational status attainment, which demonstrates the internal validity of our measures.
Cultural skills matter most (b = 0.200); technical skills are least relevant (b = 0.015). In addition, it is found that the impact of father’s occupational status is less for educational programmes in which the level of cultural and economic skills is higher. In additional models (not shown) we once replaced the four skills variables with the original field of study variable, which consisted of eleven categories, and once with a simply dummy variable capturing whether or not respondents reached a vocational track (including university education). In these alternative models, the finding that social origin matters less for higher educated (or, alternatively, that education matters more for those from lower social origin) could not be explained. We conclude that the proposed mechanism through occupation-specific skills is valid and that hypothesis 3c is supported: the stronger emphasis on (recognizable and employable) occupation-specific skills in higher levels of education explains why for individuals with these levels of education the role of social origin matters less in terms of occupational status attainment.
We test hypothesis 4, finally, by comparing Model 2 with Model 3 of Table 2. Model 3 shows that part of the declining effect of social origin on occupational status attainment over time is reduced when the interaction between social origin and education is taken into account (the estimated parameter drops from –0.051 in Model 2 to –0.042 in Model 3). The observation that father’s occupational status matters less for higher educated combined with the fact that, due to educational expansion, the labour market consists of ever more higher educated, means that social origin has become less important in determining occupational success at labour market entry. This is in line with hypothesis 4.
However, the story is more complex, as is shown in Model 5, in which we included the three-way interaction between father’s occupational status, birth year and level of education. The estimated interaction terms indicate that the decreased effect of social origin on occupational success at labour market entry over time is smaller for higher educated individuals. For those without a diploma, the decline in the social origin effect is –0.113 (with a base value of 0.192). For those with tertiary education, for instance, the effect of social origin has not changed at all (for Hbo the change in the social origin effect is –0.113 + 0.113; for Wo: –0.113 + 0.136). This implies that the overall established decline of the social origin effect, caused by the smaller impact of social origin among higher educated in combination with educational expansion, is suppressed by the smaller and even absent decline in the social origin effect over time for higher educated.
Another important result from Model 5 (and already shown in Model 4) is that the effect of level of education has become smaller over time. So, despite a reduced role of ascriptive characteristics such as social origin, we do not find evidence of an increased importance of achieved properties, such as education. Indeed, the effect of level of education has even declined faster than the social origin effect in the same time period (the calculated sheaf coefficient of the interaction between level of education and birth year is –0.056 in Model 4 [not shown], which is larger than –0.032 for the interaction between father’s occupational status and birth year), which has resulted in a rise in the relative impact of social origin compared to level of education. This finding contradicts an important aspect of modernization theory.
Conclusions
In the empirical analysis of this article, we first of all found that an individual’s occupational status depends upon his or her father’s occupational status. Apart from that, an individual’s level of education determines strongly his or her occupational status. The effect of education was found to be a few times stronger than that of social origin, indicating that achievement matters more than ascription.
Next, it was established that the effect of social origin has declined over time. By comparing birth cohorts, it was shown that for individuals who were born more recently the impact of father’s occupational status is weaker than for individuals originating from older birth cohorts. The effect of an individual’s level of education has, on average, decreased over time too. So, this aspect of modernization theory, that predicts a trend from ascription to achievement, cannot be confirmed. In fact, the impact of level of education has even decreased faster than the social origin effect, implying that the impact of social origin relative to that of education has risen in the course of time. This is a surprising result, given prior Dutch findings on trends in status attainment. The considerable credential inflation in The Netherlands (Wolbers et al., 2001) may explain this result. Around one-third of the declined impact of social origin is related to the increased age at first job. Given that the educational expansion prolonged the school career of individuals, they make the transition from school to work at ever later ages, where parental control and influence matter less.
Furthermore, the article found evidence of a weaker impact of social origin on occupational success at labour market entry for higher educated individuals, as suggested in earlier studies by Hout (1988), Mastekaasa (2011) and Torche (2011). The reason for this weaker effect is related to the fact that the educational programmes offered at higher levels of education are more focused on the acquisition of occupation-specific skills. Employers use these skills to estimate the expected labour productivity of (future) workers rather than sorting them on the basis of their social origin. Those with the best qualification are first selected for vacant jobs, and social origin matters less.
Finally, we found indications for the prediction that the declining impact of social origin on occupational success at labour market entry over time is partly due to a combination of a smaller effect of social origin for higher educated and educational expansion. This overall established decline of the social origin effect, however, is suppressed by the smaller reduction of the social origin effect over time for higher educated. Nevertheless, these findings do support Breen and Jonsson’s idea of a compositional effect of educational expansion in explaining the gross association between origin and destination (2007).
Discussion
Modernization theory is often used in social stratification research to deduce general hypotheses regarding occupational status attainment. The implicit mechanisms assumed by economic and cultural modernization have so far hardly been tested. We considered the empirical test of several explanations for the expected weaker impact of social origin over time and for the higher educated as the major contribution of our article. We predicted that, as a result of modernization, more individuals start to work in larger firms and enter the labour market at later ages over time and that this, in turn, could explain the declining impact of social origin. Especially the increasing age at first job turned out to be relevant in this respect. When one becomes older, the impact of social origin declines. However, at this point we cannot rule out that this finding is, for instance, in part due to higher social origin students being able to search longer to find their first job (that is, their reservation wages are higher) (see also note 3). That we did not find an impact of firm size may be because firm size is not a valid proxy for formal recruiting procedures in a bureaucratic context. We encourage scholars to test these hypotheses with more appropriate operationalizations, for instance, by looking at the contrast between the private and public sector with the latter being more bureaucratic.
In addition, we expected that the impact of social origin on occupational success is weaker for the higher educated, because higher educated individuals more often work in larger firms, enter the labour market at a later age and more often have recognizable occupation-specific skills that are in demand on the labour market. We found that the labour market for higher levels of education is more meritocratic. Our results lend support to the idea that employers use the degree to which various educational programmes provide occupation-specific skills to evaluate the labour productivity of potential workers. We ruled out an alternative explanation, namely that it is just because these people have followed a vocational track. Unfortunately, we had to rely on the field of study of the respondent to construct the skills variables. In future research an attempt should be made to use individual rather than aggregate measures of the knowledge and skills obtained during the educational career. In addition, a direct measure for track specificity should be incorporated within the theoretical framework to disentangle whether higher educated are better in choosing tracks that offer the most skills and/or the tracks that most clearly signal occupation-specific skills to potential employers.
Our results that education especially affects the occupational success of individuals from lower social origin can also be interpreted as corroborative evidence for the idea that new or non-traditional students benefit more from education than others (Hout, 2012). Lower social origin students have more skills to learn when they start their educational career. In theory, it is possible to test which explanation holds by including measures for the (occupation-specific) skills students already possess before they start their track of choice.
All in all, this article offers a test of some overlooked mechanisms of how social origin affects occupational success. The rather innovative way of doing this enables the existing knowledge gaps in social stratification research to be filled in. With our contribution we hope to stimulate other scholars in the field to continue in this direction.