top of page
Search
Writer's pictureZachary Van Winkle

The complexity of employment and family life courses across 20th century Europe

More evidence for large cross-national differences but little change across 1916-1966 birth cohorts



Unstable life courses moving between different jobs and unemployment, or recurrently changing family situations are often thought to be detrimental for individuals and their family members. But moving between different jobs and family situations can also be seen as a hallmark of liberal societies, where individuals are free to choose and re-adjust life paths. Life courses have been found most stable and uniform in the regulative communist societies of Eastern Europe and the dictatorships in Southern Europe in the 1950s and 1960s. This can hardly be seen as an indication of a generally desirable life course outcome. Before answering the empirical and normative question, whether life course instability is associated with desirable or undesirable outcomes, one has to establish whether life course instability has really increased over the past decades.

Most studies on life course complexity are motivated by the perception among scholars and the general public that lives have indeed become more complex across time. Contrary to common conjectures two recent studies demonstrated that although employment and family lives became moderately more complex across birth cohorts, differences across countries are considerably larger. Van Winkle and Fasang (2017) used life history data from the third wave of the Survey of Health, Ageing, and Retirement in Europe (SHARE) to follow individuals employment lives from ages 15 to 45. They show that only 2 percent of the variance in employment complexity is attributable to cross-temporal differences, while 15 percent could be accounted for by differences across countries. Van Winkle (2018) used the same data source following individuals from ages 15 to 50, as well as the Generations and Gender Survey (GGS), and found that less than 2 percent of family life course complexity variation can be traced back to cohort differences, but cross-national differences could account for 10 percent of the variance. Moreover, both studies found little evidence for country-cohort interactions. In other words, few birth cohorts within single countries deviated from the average trend for all countries towards more complex employment and family life courses.

However, both studies were based on a limited sample of countries (N = 14) and birth cohorts (N ≈ 13 from 1924 – 1956). A core criticism of the original studies was that they missed younger birth cohorts born in the 1960s that were among the most affected by the structural and normative changes assumed to increase life course instability, including economic restructuring and skill biased technological change. If indeed employment and family complexity sharply increased for the cohorts born in the 1960s who experienced their early to mid-adult life courses between the 1980s and early 2000s, the argument would be limited to the earlier historical period covered in their original studies.

A second core criticism was that the selection of countries was heavily skewed towards North- and Southwest Europe (10 of 14 countries): Austria, West Germany, the Netherlands, France, Switzerland, Belgium, Sweden, Denmark, Spain, and Italy. East and Central Europe was only represented by the Czech Republic, Poland, and East Germany. Greece was the only representative of Southeast Europe or the Balkans.

In this study we propose a replication and extension of Van Winkle and Fasang (2017) and Van Winkle (20018) to assess whether the core argument of more life course variation across countries than across time holds for younger birth cohorts and in a boarder range of countries.

First, we add one decade of new birth cohorts (up to 1966) that includes birth cohorts assumed to be particularly affected by structural driving forces increasing life course complexity. Our study follows in the steps of Van Winkle and Fasang (2017) and Van Winkle (20018) in questioning the common sense messages in sociology and broader society that life courses have become more complex, unstable, and unpredictable than in the early 20th century.

In addition, we extend the analysis including the seventh wave of SHARE, that allows us to include 15 additional countries. Next to Luxembourg, Portugal, and Finland as additional representatives of North and Southwest Europe, we add Slovakia and Hungry to the sample of East and Central European countries, as well as Cyprus, Malta, and Israel. Most importantly, we are for the first time cover two areas of Europe that are nearly non-existent in the current literature on employment and family life course change: former Yugoslavian and Southeast Balkan countries – Croatia, Slovenia, Romania, and Bulgaria – as well as the former Soviet Baltic countries – Lithuania, Latvia, and Estonia. Individual life courses in these two regions will likely be inevitably different from other East European and especially West European countries. Data & Methods As always, detailed information on the data and methods can be found in the published study including a replication package. We follow Van Winkle and Fasang (2017) and Van Winkle (2018) to define sequence states, calculate sequence complexity, and decompose sequence complexity variation across countries and birth cohorts.

Our analyses are based on the Survey of Health, Ageing, and Retirement (SHARE). The target population of SHARE is individuals age 50 and older at the time of data collection and refreshment sampling, but SHARE also collects information from and on respondents’ spouses as well as other household and non-household members. We include respondents and their partners who conducted a life history interview in the 3rd wave of SHARE (SHARELIFE) collected between autumn of 2008 and 2009 in 14 countries used in the original studies by Van Winkle and Fasang (2017) and Van Winkle (2018). We add SHARELIFE interviews conducted in the 7th wave of SHARE collected in 2017 and 2018.

We conceptualize individual employment trajectories by combining the school-to-work transitions with moves between employers and transitions in and out of employment. Each individual sequence is composed of 35 consecutive years. States are defined either as 1) in education, 2) in full-time employment, 3) in part-time employment, 4) unemployed, 5) inactive, or 6) in retirement. Employment states additionally include a job spell number to distinguish mobility between jobs from the first, second to nth job.

Family sequences are also composed of 35 consecutive annual states. Each sequence state is either 1) in the parental home, 2) single, 3) cohabiting, or 4) married. Further, each state element can be extended by the presence of at least one child: for example, married with at least one child. Note that “single” indicates that the respondent was neither in the parental home nor cohabitating; it does not specify the relationship status of the respondent in terms of living apart together relationships.

We use a composite measure developed in sequence analysis to assess the complexity of sequences of categorical states: the sequence complexity index. Complexity is minimal in sequences composed of a single state and maximal in sequences that contain each state element with equal durations and have the maximum number of transitions.

In the following section, we first decompose the variance of employment and family sequence complexity using additive and interacted cross-classified random effects regressions. This allows us to quantify the proportion of variance attributable to country differences versus change across cohorts. In a second step, we assess average levels of employment and family complexity across countries and cohorts using empirical Bayes estimates of the country and cohort random effects. Finally, we use the empirical Bayes estimates of the interacted country-cohort random effect to determine whether countries deviate substantially from the average cohort trend. Results Overall findings substantiated the conclusions from Van Winkle and Fasang (2017) and Van Winkle (2018) also including twice as many countries and a decade of younger birth cohorts: considerably more variation in the complexity of employment and family trajectories was attributable to cross-national differences compared to change over time. For employment trajectories, 14.6 percent of the variance in sequence complexity could be ascribed to differences across countries (15 percent in Van Winkle and Fasang 2017), and 5.5 percent to change across birth cohorts (2 percent in the original study) (see column 1 of Table 1). Accordingly, while variation across cohorts is still substantially smaller, it increased moderately for the youngest cohorts included in this update. Findings thereby support that the structural changes noted above indeed moderately increased employment complexity across Europe. For family trajectories, cross-temporal differences could account for less than 2 percent of the variance of sequence complexity, while roughly 10 percent are due to cross national differences (equally 2 and 10 percent Van Winkle 2018).

Comparing the left and right panel of Figure 1 again underscores how substantial country differences are compared to cohort change. As can be seen in Figure 1, countries broadly map on to welfare state regime types in terms of employment complexity. Southern European countries – Portugal, Greece, Cyprus, Malta, Spain, and Italy – had the least complex trajectories. Somewhat more complex but still below average were the Balkan countries – Romania, Croatia, Slovenia, and Bulgaria. Countries with average complexity included Eastern European countries – Hungary, Poland, and the Czech Republic – but also countries classified in the Western European conservative-corporatist regime – Luxembourg, Austria, Belgium, West Germany, and France. Countries with the highest average complexity were from the Scandinavian social democratic regimes – Denmark, Sweden, and Finland – as well as conservative Western European countries – the Netherlands and Switzerland – and East Germany. East Germany shows relatively high employment complexity, which is an unexpected outlier from the perspective of welfare state regimes and might be related to the distinct mobility regime during communism and reunification process in East Germany. Among the Baltic States, Estonia resembles its Scandinavian neighbours, while Latvia and Lithuania are closer to West Germany and France. Figure 1: Empirical Bayes Estimates of Employment Complexity by Cohort and Country


Figure 2: Empirical Bayes Estimates of Country-Specific Deviations from Cohort Employment Complexity


Our results highlight that changes in the two decades between 1980 and 2000, when the 1960s cohorts were entering and establishing themselves on the labour market, lead to an overall trend of increasing employment complexity that is substantively significant albeit moderate. The average trend across our sample of European countries increases from below average levels typical of Southern Europe to above average levels typical of East Germany, Finland, the Netherlands, and Estonia. Moreover, the trend towards increasing complexity is approximately linear: there is no evidence that a certain birth cohort or cohorts were suddenly affected by a period event that increased only their average complexity levels. Figure 2 shows the empirical Bayes estimates of the country-cohort random effects, which are presented as country-specific deviations from the cohort trend shown in Figure 1. However, we find no deviations from the overall cohort trend within countries.

Again, when comparing the left and right panels in Figure 3, cross-national differences are substantially larger than change over time. The order of the countries from least to most complex in Figure 3 also matches common welfare state groupings, although to a lesser degree than for employment trajectories. The least complex family sequences could be found in Eastern Europe – Slovakia, Poland, Hungary, and the Czech Republic – Southern Europe – Malta, Portugal, Spain, Italy, Greece, and Cyprus – as well as countries in the Balkans – Bulgaria, Croatia, Slovenia, and Romania. Another tight group of countries with slightly above average family complexity were mainly members of the Western European conservative-corporatist welfare regime – Belgium, Austria, the Netherlands, and Luxembourg – as well as East Germany, Ireland, and Lithuania. Sweden and Denmark are two countries with the most complex family sequences. Between them and the former group of Western European countries lie Estonia, Finland, and Switzerland on the upper end and Latvia, France, and West Germany on the lower end. Figure 3: Empirical Bayes Estimates of Family Complexity by Cohort and Country


Figure 4: Empirical Bayes Estimates of Country-Specific Deviations from Cohort Family Complexity


Although the birth cohort estimates indicated a trend towards more complex family life courses, as shown in the right panel of Figure 3, that upward trend is less pronounced than for employment trajectories. In fact, the results demonstrated that average complexity was relatively stable for cohorts born between 1916 and 1936, before continually increasing between the 1934 and 1954 cohorts. After 1954 there was no increase in the complexity of family sequences across our countries. However, in contrast to employment complexity, we found numerous country-specific deviations from that trend, especially located among the youngest birth cohorts. The empirical Bayes estimates of the country-cohort random effects for family complexity are displayed in Figure 4. Younger cohorts from the Scandinavian social democratic countries – Sweden, Denmark, and Finland – as well as some Western European countries – France, Switzerland and Belgium – have considerably higher average complexity levels than the general cohort trend (roughly 2 points or 15 percent above the cohort mean). This indicates that there may be a polarizing trend in the complexity of family life courses in Europe: while most of Europe experienced no increases in complexity following cohorts born in the mid-1950s, the complexity of family trajectories continues to increase in Scandinavian countries and a few Western European countries.


Discussion

This study replicated and extended two recent articles on the complexity of employment life courses (Van Winkle & Fasang, 2017) and the complexity of family life courses (Van Winkle, 2018). Specifically, we addressed a core criticism of both studies: we expanded the limited sample of countries and birth cohorts by adding more than 15 new countries and a new decade of younger birth cohorts born in the 1960s. Findings substantiated the original conclusions and added information on cross-country and cross-temporal variation in employment and family life course complexity.


The inclusion of new countries and birth cohorts contributes to one of the most central debates in European family demography. Are patterns of family formation converging or diverging over time, and do cross-national differences persist or even widen? Most research has contended that cross-national differences are stable or growing rather than converging as suggested by the SDT thesis. Our results suggest that cross-national differences in the complexity of family life courses are indeed stable on average for a sample of 30 countries. However, our findings also support divergence in cohort change across countries. While the complexity of family trajectories continues to increase across more recent cohorts in a number of Nordic and Western European countries, it has stagnated across most countries, especially in Southern and Eastern Europe as well as the Balkans.


In contrast to country-specific trends for family complexity, employment complexity continues to increase across more recent cohorts for all countries. This is in line with scholars highlighting increasing employment precarity among younger cohorts. However, even if the trend towards more complex employment life courses continues, cross-temporal change would not be as large as cross-national differences for decades to come. While moderate in size against the benchmark of stable cross-national differences, this universal increase is notable. Yet one should not jump to conclusions about similar universal driving forces underlying this trend. It is possible that global economic developments and less employment security in many countries play a role in this increase.


Our study underlines the potential of cross-national comparisons to understand the drivers of both employment and family life course complexity. The bulk of the variation in these outcomes lies in stable differences across countries. For example, studies show that employment protection legislation and wage protection rates are associated with intra-generational mobility and employment complexity. Family policies that incentivise a male breadwinner female homemaker division of labour stabilize family life courses. However, cultural differences, such as the quality of democracy, trust in institutions, attitudes towards work and family, may drive cross-national and cross-temporal differences in employment and family complexity. To disentangle the combined effects of different institutional features on life course complexity cross-national comparisons therefore seem particularly promising.


References



110 views0 comments

Comments


bottom of page