Life-course-sensitive analysis of group inequalities: Combining Sequence Analysis with the Kitagawa-Oaxaca-Blinder decomposition

Abstract

This paper improves the understanding of how prior life-courses that unfold over several decades relate to inequalities between groups in old age. Economic inequality in old age results from many processes that unfold over individuals’ prior life-courses in given institutional settings. With population aging, such economic inequalities in mature adulthood and old age become increasingly relevant. We propose a step wise process combining the strengths of Multichannel Sequence Analysis to analyse life-courses with the Kitagawa-Oaxaca-Blinder decomposition, which has long been established to analyse group inequalities in outcomes. To illustrate the proposed procedure, we apply it to the case of Gender Pension Gaps in Italy and West Germany. Our analysis reveals that a large part of the gap in both countries is due to highly gender-specific work-family patterns; male-typical life courses receive higher average pension income while female-typical life courses are worse rewarded. On the other hand, only a minor share of the gap is associated with different pension rewards of men and women with the same work-family life course pattern. Thus, we do not find evidence for redistributional measures of pension systems (e.g. care benefits) leading to higher pension incomes of women compared to men within the same life course clusters and thus to a significant reduction of the Gender Pension Gap. We contribute to the literatures on gender inequality, life-course processes, and comparative analysis of old age pensions by showing how combined work and family life-courses lead to different gender pension gaps in two countries.

Housewives never retire!? Gender biases in popular sample definitions for studies on the elderly

“…One-quarter of women who do not report being retired continuously engaged in unpaid care work from age 18-65. Another frequent work life-course pattern is characterized by women from full-time employment to continuous unpaid care work in their early twenties, which might be linked to family formation as shown previously (e.g., Rowold et al, 2023). For men, the work trajectories are more heterogeneous: the fifteen most frequent trajectories only cover 18.7% of those who do not self-identify as 25 retirees.…”

“…However, this depends on the pension system. Pension systems such as in the Netherlands guarantee relatively broad access to public pensions through flat-rate pensions, whereas high minimum contribution years such as 20 years in Italy constrain access, particularly for individuals with interrupted careers (Leitner, 2001;Rowold, Struffolino, & Fasang, 2023). The latter applies disproportionally to women and can lead to large differences between men and women in access to pensions (Gender Coverage Gap, e.g., Tinios, Bettio, Betti, & Georgiadis, 2015).…”

Uncovering what matters: Family life course aspects and personal wealth in late working age

“…First, the SOEP respondents' cohabitation histories were not recorded retrospectively. Although we argue that the absence of cohabitation data in our sequences is not a significant concern for our studygiven the social undesirability and discouragement of cohabitation within the studied cohort and context (Le Goff, 2002;Rowold et al, 2022)-it should be noted that cohabitation may be a crucial factor in future research, particularly when examining more recent cohorts or more liberal contexts. Second, our methodological approach is, at least for the time being, limited in the number of life course channels it can consider.…”

“…Consequently, such states are likely to be rare in our data. Reinforcing this perspective, Rowold et al (2022) demonstrated that in a cohort of slightly older West Germans and Italians, less than one % of person-year spells between the ages of 18 to 65 were spent in non-married cohabitation.…”

“…First, family life courses were operationalised using a small range of summary indicators, such as age at first marriage, times married, the duration within first marriage, and the number of children, and these indicators were used to predict wealth levels (Halpern-Manners et al, 2015). Although research that followed this methodological tradition provided critical first impulses, the approach is highly selective and limited regarding the indicators and number of indicators that can be considered due to multicollinearity issues (Rowold et al, 2022). This means that the approach may not capture all relevant indicators of family life courses and provide an incomplete picture of the association of interest.…”