With Age Comes ...? An Examination of Gendered Differences in the Resource Advantages Associated with Parental Age in Norway

Using high-quality register data, this paper constructs an empirical portrait of older parenthood in contemporary Norway and explores gender differences in the extent to which older parents are better-resourced parents. Like most family issues, academic and policy discussions of older parenthood have tended to focus on the experiences of women. Although motherhood at older ages was not uncommon in previous generations, rapid social and family changes in recent decades mean that today’s older mother is far more likely to be having her fi rst child. She may have focused on obtaining a good education and then on establishing her career and fi nding a supportive partner. When viewed through this motherhood lens, older parents are often portrayed as being relatively well-off fi nancially and enjoying stable family lives. Viewed through the fatherhood lens, however, the family literature suggests that age might not be as strong a marker of socio-economic advantage. Our fi ndings show that this is the case: While older fathers are, on average, more socio-economically advantaged than younger fathers around the time their children are born, their relative advantage is narrower than what we observe when older mothers and younger mothers are compared. Gender differences in the family history profi les of older parents appear to explain some of the differentials we document. The children of older fathers are more likely to be born into blended families and less likely to born to a mother who postponed her fi rst birth than are the children born to older mothers. The results suggest that the meaning of paternal age and maternal age differs: paternal age is not as strong a marker of socio-economic advantages and resources for children as maternal age is.


Introduction
In many countries of Northern and Western Europe, the period since the late 1960s has been marked by dramatic and interrelated changes in fertility and family life. Cohabitation increasingly replaced marriage as the fi rst union, and unmarried childbearing lost much of its stigma -particularly in contexts in which unmarried parents tended to co-reside. At the same time, opportunities for women to participate in higher education and the labour market created incentives for women to make use of new contraceptive technologies and to delay the transition to motherhood (Goldin 2006). Divorce and dissolution displaced widowhood as the predominant pathway into lone parenthood, and the subsequent formation of new partnerships resulted in growing numbers of blended families. These changes and their consequences, which have been collectively described in the demographic literature as the Second Demographic Transition (SDT) (Lesthaeghe 1995;van de Kaa 1994), have attracted considerable international interest amongst academics, policymakers, and social commentators. For a variety of practical, pragmatic, and sometimes problematic reasons, social scientists have historically tended to approach the study of family changes by focusing almost exclusively on the experiences of women (Goldscheider/Kaufman 1996;Kneale/Joshi 2008;Lappegård et al. 2011;Sigle-Rushton 2005). Although demographers and family scholars increasingly recognise the theoretical and practical importance of carrying out research on men and fathers (Sigle-Rushton et al. 2013), the legacy of this focus on the experiences of women might mean that prevailing conceptual frameworks are less well-suited for depicting what has been going on with men. As feminist scholars confronted with androcentric bias in their fi elds of study have long noted, excavating the experiences of women is often challenging because existing frameworks and models do not always represent or explain their lived experiences well. In the study of family and fertility, we might encounter the same problem in reverse: When turning our attention from mothers to fathers, it might not be suffi cient to simply "add men and stir". This may well be the case when we attempt to understand trends towards older parenthood in post-SDT countries. The infl uential Diverging Destinies framework (McLanahan 2004) -and particularly the relationship between parental age and family resources that is central to it -may not fi t the experiences of men as well as it fi ts the experiences of women. Moreover, the Norwegian context might differ from the Anglo-Saxon settings in pertinent ways. With lower income inequality and more generous parental leave policies the Norwegian economy may not differentially reward highly skilled women who delay childbearing to the same extent (Sigle 2016;Mandel/ Shalev 2009).
Although motherhood at older ages was not uncommon in previous generations, rapid social and family changes in recent decades mean that today's older mother is far more likely to be having her fi rst child. She may have focused on obtaining a good education and then on establishing her career and fi nding a supportive partner. When viewed through this motherhood lens, older parents are often portrayed -in the social scientifi c literature at least -as being relatively well-off fi nancially and enjoying stable family lives (Goisis et al. 2017;Goisis/Sigle 2014). Viewed through the fatherhood-lens, however, age might not be as strong a marker of socio-economic advantage and resources for children. Some older fathers will be having children at an advanced age because they are forming or completing their fi rst family with a mother who postponed her fi rst birth, and so resemble older mothers as described in the Diverging Destinies framework. However, there are reasons to think that a smaller percentage of older fathers than older mothers will have followed this family trajectory. Evidence has shown that the likelihood of forming a second family is higher for men than for women (Van Bavel et al. 2012;Ivanova et al. 2013) and we expect to see more older fathers than older mothers in second and higher order partnerships. Given that divorce, particularly when children are involved, is a "stratifi ed and stratifying life event" (Raley/Sweeney 2020: 81; see also Sweeney 2010; for Norway, see Lappegård/Rønsen 2013;Poortman/Lyngstad 2007;Steele et al. 2009), this suggests that more older fathers than older mothers will be socioeconomically disadvantaged. As men tend to form partnerships with women who are younger than they are (Ní Bhrolcháin 2005) and wide age differences are often acceptable to both older men and younger women (Ní Bhrolcháin /Sigle-Rushton 2005), age differences are greater in subsequent partnerships (e.g. Ní Bhrolcháin 2005;Gustafson/Fransson 2015). We therefore expect older fatherhood to be less strongly associated with maternal postponement (having a partner who postponed her fi rst birth) -and all its associated advantages -than older motherhood. Taken together, these gender differences suggest that older fatherhood will not be the same marker of resources for children as older motherhood is (McLanahan 2004). In other words, what measures of parental age represent -their meaning -may be gendered. We are, however, not aware of any studies that have systematically compared gender differences in the meaning of parental age.
Using high-quality Norwegian register data, this paper will begin to address this research gap. Our aims are to describe what distinguishes men registered as fathers at an older age (40+) from men registered as fathers at a younger age in the same time period, to explore whether the differentials and trends are similar to what is observed when older (aged 36+) and younger mothers are compared, and to document whether these gender differences have changed over the course of the SDT. Our fi ndings represent a fi rst step towards understanding whether advanced paternal age is, on balance, a similar marker of higher parental resources (McLanahan 2004) as maternal age is.

Background and research questions
Efforts to document, describe and explain older childbearing, which is typically defi ned as childbearing after age 35 or 40, have, like fertility studies in general, tended to analyse the experiences of women (for an excellent recent exception, see Beaujouan 2020). In post-SDT countries, the increasing share of older mothers is often understood as being driven by economic or other incentives for a woman to have her fi rst child at a later age (Billari et al. 2007). Historically, many older mothers were completing larger-than-average families that they started at younger ages. When fertility fell to low levels, most women who started their childbearing at young ages completed their families while they were still relatively young, and older motherhood became increasingly rare. In recent years, the number and share of older mothers has increased as women started and completed their childbearing closer to the biological limit (Goisis et al. 2017). Consequently, recent trends in older motherhood have been understood as a consequence of the process of maternal postponement.
The predominant focus on women in fertility studies and the emphasis on maternal postponement as the main driver of the growing share of older mothers has shaped the way the needs of families with older parents are represented and understood. Economists and demographers have tended to conceptualise contemporary, postponement-driven older motherhood as the avoidance and reversal of the disadvantages associated with early motherhood. Compared to early motherhood, postponement is associated with higher socio-economic status, greater maturity and a lower likelihood of dissolution, and so older motherhood is frequently described as both a marker and driver of advantage, with important implications for children and their life chances (Gustafsson 2001;McLanahan 2004). Consequently, on average, "an increase in mother's age is viewed as an increase in parental resources" (McLanahan 2004: 609).
In the demographic and sociological literature, concerns about early motherhood have tended to focus on socio-economic disadvantages, while studies of late motherhood, if they focus on risks at all, tend to focus on issues associated with biological constraints, health risks (Sobotka 2010), and the economic costs to wider society: subfecundity leading to eventual childlessness and fuelling low fertility rates in ageing societies, and the medical costs of providing ART to older women. Although some recent work has demonstrated that late motherhood may violate wider social norms and so attract some level of opprobrium (Billari et al. 2011), the assumption that older mothers are, on average, better-resourced mothers underpins much of the social scientifi c literature (Goisis/Sigle 2014). Even if evidence suggests that advanced maternal age means higher resources, should we assume that advanced paternal age can be interpreted similarly? This is our fi rst research question.
Because men tend to be slightly older than their wives and partners, many men will become older fathers because they are having children with women who postponed their fi rst birth (Henwood et al. 2011). Patterns of assortative mating and hypergamy suggest that these men will be similarly socio-economically advantaged. They may also have benefi ted from their partner's postponement by having more time to invest in their own education and career before assuming the responsibilities of fatherhood. However, family changes that accompanied the SDT suggest that the family and partnership histories may be more varied for older fathers than for older mothers, and in ways that suggest less socio-economic advantage. Previous studies have shown that men repartner more often and more rapidly than women do, especially when they are older and have had children in a previous union (Van Bavel et al. 2012;Ivanova et al. 2013). Consequently, we expect that older fathers will be more likely than older mothers to have to provide time and resources to dependent children from previous partnerships.
As age differences tend to be narrowest in fi rst partnerships (Gustafson/ Fransson 2015), a greater share of second and higher order partnerships may lead to childbearing where the father, but not the mother, is becoming a parent at an older age. Taken together, this evidence suggests that some older fathers will be having children with mothers who have not postponed their fi rst birth and, importantly, many of these men will not be fi rst-time fathers. The literature on childbearing across partnerships suggests that these fathers are often more socio-economically disadvantaged than fathers who have children with only one partner (Lappegård/ Rønsen 2013;Ginther et al. 2019). However, few studies of multiple partner fertility link it to the rise in older parenthood more generally. Our second research question asks whether and how the families of older fathers and older mothers have differed in recent decades. To answer this question, we examine how older fathers and older mothers are distributed across fi ve family history profi les that take into account maternal postponement and childbearing across partnerships. Our third research question asks whether the fi ve family profi les are associated with differences in socioeconomic and health resources.
In contrast to the academic literature on older and late parenthood which has focused extensively on the postponement of fi rst births to women, the "start-over" or "do it again" dad -an older man in a higher order partnership who starts a second family after his children from previous partnerships attain adulthood -fi gures prominently in media depictions of older fathers (Rubin 2016). This older father is often represented as a deeply committed family fi gure who regrets choices made the fi rst time around, such as prioritizing work and earnings when his now adult children were younger. Like the mother who postpones, he is often assumed to be economically advantaged. However, he is consciously and actively embracing family life the second time around, often with a much younger partner who is starting her fi rst family with him (Beyette 1985;Hartley 2014). Because media representations of this kind tend to be anecdotal, they provide little information about how typical this type of older father actually is. Furthermore, the start-over father has not attracted much attention in the academic literature, perhaps because of its (until recently) almost exclusive focus on women. "Start-over" parenthood -which requires that children from previous partnerships have all reached adulthood before starting a second family -represents a pathway that, for biological reasons, few women could follow into older motherhood, especially in recent decades as the proportion of women having their fi rst birth in their early 20s has declined substantially. This might explain why non-overlapping serial families have received limited attention in the extant literature so preoccupied with the experiences of women. Children born to such fathers might face less competition for time and resources from halfsiblings when they are young. However, they might also be born to the oldest of older fathers, and so we expect them to be more likely to experience the death of their father before they research adulthood.
Evidence that start-over fathers are, in line with media depictions, a distinct, more socio-economically advantaged group than other fathers with childbearing across partnerships would resonate with previous studies that found that, for Norwegian fathers, multiple partner fertility is associated not just with socioeconomic disadvantage (as shown in previous literature) but also with socioeconomic advantage (Lappegård/Rønsen 2013). When we answer our second and third research questions, we are specifi cally interested in examining the following: How typical and how distinct -from other kinds of blended families and other families with older parents -are families with older, start-over fathers?

Data and Measures
To examine differences between older fathers and mothers, we use data extracted from the Norwegian Population Register which covers everyone who has ever lived in the country after 1964. From this register we identifi ed the mothers and fathers of children born in Norway in each of three time periods : 1991-95, 2001-05, and 2011-15. Individuals who became older parents in the fi rst period would have reached adulthood during the early stages of Norway's SDT (Frejka 2017;Lesthaeghe 2011), while many of those who became older parents in the most recent period -except perhaps the very oldest fathers -became adults after the fertility and family changes that comprise the SDT were fairly well-advanced. For 98.2 percent of the children born, both the mother and father can be identifi ed in the register. We excluded those cases where at least one parent was born before 1935 (less than 0.1 percent in the fi rst period, otherwise 0.0 percent) because the birth histories of those parents may not be complete. Because birth events (deliveries) rather than children or parents are the unit of analysis, multiple births are treated as one event, and some parents are associated with more than one birth event in the analytic sample. This is the case for 12.9 percent of older fathers and 8.6 percent of older mothers. For each delivery, we extracted information about the parents' ages at the time of delivery, the parents' ages at fi rst birth, whether they had previously registered a birth with someone else, and the dates of any previous deliveries. For those deliveries that were recorded in the fi rst time period, we extracted information about whether the parent died in the next 18 years, thus depriving the child of parental resources before they were legally independent. Furthermore, we added information about the highest educational level each of the parents had achieved, measured in the October before the year of birth, and the total pre-tax, pensionable income earned by each parent in the calendar year two years before the year of birth. This income measure includes wages and salaries, any income from self-employment, and unemployment benefi ts, temporary sickness benefi ts, and maternity or paternity benefi ts, but neither pensions nor permanent disability benefi ts.
For 7.1 percent of all deliveries in our sample, information about education was missing for either the mother or the father. This is often because the parent was born or educated abroad. In these cases, the delivery (the information for both parents) was excluded when average education levels were calculated.
Another limitation was that income was missing for many mothers or fathers. The missing value typically indicates that the parent did not earn any pensionable income that year. When there is no evidence that income was missing because the parent had been living abroad (3.8 percent of fathers and 8.8 percent of mothers), missing values were set to zero. When at least one parent was not living in Norway at the start of the previous two calendar years (8.1 percent of all deliveries), we excluded the delivery from all analyses of income differences. Finally, if at least one parent emigrated from Norway within 18 years of the delivery, that delivery is excluded from the analysis of survival/death.
We fi rst consider whether, over the past three decades in Norway, older fathers can be described as better-resourced than younger fathers. We also consider whether the extent of any resource advantage is similar to what is observed when older mothers are compared to younger mothers. Older mothers are defi ned as those aged 36 and older. This is one year above the age threshold that is often used in the medical literature to classify a pregnancy as "geriatric", and so all of our older mothers were "geriatric" at conception. The literature offers little guidance about where to set the age threshold for older fathers. In their study of older fi rsttime fathers in Norway, Nilsen and colleagues (2013) used age 40 as the threshold for "very advanced age". In a cross-national comparison of late and "latest-late" parenthood in wealthy countries, Beaujouan (2020) used age 40/48 as the threshold for late/extremely late motherhood and 45/55 as the threshold for late/extremely late fatherhood. In the absence of a clear convention, we relied primarily on information about the proportion of older mothers to set the age threshold for older fathers. Figure 1 shows the share of deliveries to all older parents (the height of the bars), and the share of all deliveries to "late" and "very late" parents. In the 2011-15 period, about 17 percent of mothers were aged 36+ and therefore classifi ed as older mothers. In the same period, a similar share of fathers was aged 40+ (about 16 percent of fathers) and so we used this as the threshold for older fathers.
While our decision ensures that older parenthood, as we defi ne it, is equally rare for both genders, there are differences and trends in the age distribution of older parents which may have implications for our results. To see these differences more clearly, Figure 2 shows how the older mothers and fathers are distributed across the late and extremely late age categories We use similar thresholds to those in Beaujouan (2020), with some slight adjustments so that the number of years of age included in the "late" category is the same for mothers (age 40-47) and fathers (age 44-51). While the majority of older parents in Norway are at the younger end of the age range, the share of older parents who are late or very late parents has increased over the time period our data covers: from 39 percent to 43 percent of deliveries to older fathers and from 19 percent to 25 percent of deliveries to older mothers. For men, the increase in the share of late fathers is observed mostly at ages 52+ ("extremely late"). In contrast, almost no Norwegian mothers have children at extremely late ages (48+) and so the 6 percentage point increase in births to late mothers is observed almost exclusively in the 40-47 age group.
To measure fathers' and mothers' socio-economic position, we use information about their educational attainment and income. The fi rst measure identifi es individuals with higher tertiary education: an undergraduate degree and at least one year of graduate education. We focus on this relatively high level of attainment, equivalent to a Master's degree in many countries, because we cannot be certain that those recorded as having lower tertiary education completed their undergraduate degree. Because we are interested in whether older parents are better-resourced and whether the children born to older parents have greater access to economic resources, we examine both pensionable income and equivalised pensionable income (pensionable income divided by 1+0.3k where k represents the number of own biological children under the age of 18). We construct indicators for whether these income measures were in the top or bottom quartile of the respective income distribution of all fathers and all mothers who became parents in the same year (regardless of their age).
To account for blended families, we create an indicator that identifi es fathers and mothers who have had children with more than one person. For parents who have registered a previous birth with another parent, we also calculated the time elapsed between the last child with a different parent and the fi rst child with the current parent. We use this information to identify those fathers who completed their last family before starting their current family with a mother who had not previously Source: Norwegian Population Register had children. This means the youngest child with another parent was at least 18 before the father started his current family with a fi rst-time mother. As noted above, for biological reasons, few women could become start-over mothers, especially in a context where few fi rst births occur in the late teens and early 20s, and so we examine start-over father families only. 1 To answer our second and third research questions, we move beyond individual circumstances of older parents and compare the distribution of older parents across fi ve family history profi les (research question 2) and examine how the characteristics and resources of parents vary across those profi les (research question 3). These profi les take into account whether the mother of the child postponed her fi rst birth (already established as a clear marker of high resources e.g.  We fi rst divide the sample into two groups: those children born with at least one half-sibling (maternal or paternal) and those children born to parents who have only had children with each other. From the fi rst group, we identify the Start-Over Dad families: there are only paternal half-siblings, the youngest of which was at least 18 when the parents had their fi rst child together (the fi rst full sibling or the delivery in our sample). Setting aside this Start-Over Dad group, the remaining two groups are each divided into families with and without maternal postponement. We consider motherhood to have been postponed when the fi rst birth occurs after the age of 30. In recent decades, time spent in education has increased and other events that mark the transition to adulthood, including the age at fi rst birth, have been delayed to older ages. The average age at fi rst birth climbed from 26.1 in the early 1990s to 28.2 in 2010. By 2016, the average age at fi rst birth for women in Norway was 29. Although the age 30 threshold is a bit lower than what has been used in recent Scandinavian medical studies (Waldenström 2016), it identifi es women who, even in the most recent period, were having their fi rst birth later than average and who had suffi cient time to complete high levels of education and get established in their careers before transitioning to motherhood. This classifi cation approach results in fi ve mutually exclusive family history profi les.
The fi rst family history profi le, Diverging Destinies, identifi es families where the mother postponed her fi rst birth (beyond age 30) and there are no maternal or paternal half-siblings. This profi le most closely resembles the high resource life course trajectory described by Sara McLanahan (2004) in her Diverging Destinies framework. The second family history profi le, pre-SDT, most closely resembles the normative family patterns of the mid-20 th century. It identifi es those families where the mother did not postpone her fi rst birth and there are no half-siblings. In many of these families, both parents were relatively young when their fi rst child was born. All of the children born to older mothers in the pre-SDT profi le will therefore be higherorder births. In contrast, some of older fathers in this group will be having their fi rst child with a much younger mother (8 percent, 11 percent and 14 percent in 1991-95, 2001-05, and 2011-15, respectively). The third family history profi le identifi es the Start-Over Dad families. There are only paternal half-siblings, and the youngest halfsibling was at least 18 before the father and mother had their fi rst child. The fourth and fi fth family history profi les contain all the remaining families with at least one half-sibling. The fourth profi le, Early Sequential Parenting, includes mothers who did not postpone their fi rst birth, and the fi fth profi le, Late Sequential Parenting, includes mothers who did.
As both early parenting and family complexity have been shown to be associated with greater disadvantages, we expect the Diverging Destinies families to be most advantaged and the Early Sequential Parenting profi le to be most disadvantaged. We expect the pre-SDT and Late Sequential Parenting family history profi les to fall in between the two extremes. Because media depictions of Start-Over Dad families as highly resourced are anecdotal, we examine whether, in a representative sample, these families have greater resources available to children than other kinds of stepfamilies.

Results
Our fi rst research question asks whether older fatherhood is as strong a marker of high (individual-level) resources as older motherhood is. Table 1 presents information on the socio-economic position of older and younger parents as well as information about the economic resources they can provide for their children. For each of the three time periods, we fi rst compare older fathers and mothers to their younger counterparts and then examine whether, drawing on what we know about gendered differences in family patterns and trends, those age-related differentials are narrower for fathers than for mothers at different points in time.
The fi gures in Table 1 indicate that both older fathers and older mothers are relatively more educated than younger fathers and mothers respectively. Since the early 1990s, however, the share of all parents with a tertiary qualifi cation has increased, and educational differences between older and younger parents have become less pronounced, with the largest change occurring between the fi rst and second period. In the early 1990s, older fathers were more than twice as likely as younger fathers to have higher tertiary education. Ten years later, they were just 35 percent more likely than younger fathers to have a postgraduate qualifi cation. In all three periods, the relative educational advantage is greater for older mothers than for older fathers. For the 2011-15 period, older mothers were 71 percent more likely than younger mothers to have obtained a postgraduate qualifi cation. In the same period, older fathers were only 20 percent more likely than younger fathers to have higher tertiary education.
Both older fathers and older mothers are over-represented in the top quartile and under-represented in the bottom quartile of pensionable income in all three time periods. However, the over-representation is most marked in the earliest period for older fathers, when 41.5 percent had pensionable income in the top quartile of the income distribution (of all fathers of newborns in the same year) and in the most recent period for older mothers, when 43.5 percent had a pensionable income in the top quartile of the mothers' income distribution.
Because income tends to increase with age, the differences in pensionable income between older and younger parents are not surprising. However, as the next rows of Table 1 show, older parents are less likely than younger parents to be having their fi rst birth and more likely to have larger families. While declining over time, in the most recent period, 20.4 percent of the older fathers had at least three previous children, compared to only 3.2 percent of fathers younger than 40. Older mothers are also more likely than younger mothers to have large families. However, the percentage of older mothers with three or more previous births is, in all periods, substantially smaller than the percentage of older fathers with at least three previous births. In the most recent period, for example, only 13 percent of older mothers have already had three or more children. These fi gures suggest that Source: Norwegian Population Register and registers on education and income age-related economic advantages observed using pensionable income will narrow when equivalised income is used. They also suggest that the results will change more for fathers than for mothers. When income is equivalised to account for the number of dependent children, older parents are still over-represented in the top quartile, but, as expected, the difference narrows. At the bottom end of the distribution, older mothers are less likely to have a bottom-quartile income in all three periods, and this is most pronounced in the most recent period. In contrast, older fathers are under-represented in the bottom quartile of the equivalised income distribution in the fi rst period only.
To the extent that higher educational attainment or a top-quartile equivalised income are valid indicators of high levels of resources for children born in Norway, the fi gures in Table 1 suggest that older fathers are better-resourced than younger fathers, but the association of parental age and high resources is stronger for mothers than for fathers. Moreover, the strength of the relationship between paternal age and resources for children declined over the three periods as enrolment in higher education has increased and childbearing has shifted to older ages more generally. For older mothers, their income advantages -but not their educational advantages -have actually widened. We interpret older mothers' improved position in the equivalised income distribution across the three time periods as an upward trend in the economic resources they have available for their children. The results for older fathers suggest an opposite trend.
Regardless of their age, few parents of children born in the early 1990s died within 18 years. Nonetheless, older parents were more likely to die than younger parents. Unsurprisingly, given their higher average age (Table 1) and more skewed age distribution (Fig. 2), the absolute and relative risk of death was highest for older fathers, who were 3.2 times as likely to die as fathers who were under age 40. Around 6.8 percent of older fathers who did not emigrate (and so were excluded from the analysis) did not survive. The risk of death for younger fathers and older mothers was similar: around 2 percent died within 18 years.
The individual characteristics of older parents suggest that parental age is a marker of resources for children, but that maternal age is a stronger marker than paternal age is. When considering the resources available to children, and the extent to which they vary by parental age, the characteristics of both parents are relevant. To consider whether and how the parental characteristics of children born to older fathers and older mothers differ in ways that might suggest higher or lower resources for their children, we next look at the distribution of older fathers and older mothers across the fi ve family history profi les. As described in the data and methods section, these fi ve profi les are constructed by cross-tabbing maternal postponement (the older mother postponed or the older father is having a child with a mother who postponed her fi rst birth) and family complexity (the presence of one or more half-siblings), and then separating out, as a fi fth profi le, those deliveries to blended families with start-over fathers. Figure 3 shows how older fathers and older mothers are distributed over the fi ve family history profi les and how the distributions have changed over time. The share of Diverging Destinies families (the mother postponed her fi rst birth, and neither parent has had a child with another person), which we hypothesised would contain the most advantaged older parents, has increased over time for both fathers and mothers. This profi le became the modal profi le of older mothers in the second period (41.3 percent of deliveries to older mothers), and by 2011-15, this profi le accounted for the majority of the older mothers -53.5 percent. While increasing over the three time periods, the share of older fathers in such families is notably smaller. It is only in the most recent period that the Diverging Destinies profi le becomes the modal family history profi le for older fathers, accounting for 33.8 percent of the deliveries.
The pre-SDT family history profi le was, in 1991-95, the modal family history profi le for both older fathers (37.9 percent of birth events) and older mothers (36 percent of birth events). In subsequent periods, the share of children born into pre-SDT families declined, but more for older mothers than older fathers. In the fi nal period, the Pre-SDT family history profi le still accounted for 30.4 percent of the older fathers, but only 19.7 percent of older mothers.
The share of parents in the remaining three profi les -all of which involve some childbearing across partnerships -has declined over the three time periods. The  1991-1995 2001-2005 2011-2015 1991-1995 2001-2005 2011-2015 Older Fathers Older Mothers Source: Norwegian Population Register latter trend is more pronounced for older fathers than older mothers. Nonetheless, in the most recent period, these three profi les contained a larger share of older fathers than older mothers, and so the children born to older fathers are more likely to have half-siblings than the children born to older mothers. Although the share of older fathers in the Late Sequential Parenting profi le is larger, the percentage of older mothers in the Diverging Destinies profi le is far larger, suggesting that older motherhood is more strongly associated with maternal postponement than older fatherhood is.
In the 2011-15 period, at least 63.2 percent of older mothers postponed their fi rst birth and at least 45.2 percent of older fathers were having a child with a mother who postponed her fi rst birth. We say "at least" because the Start-Over Dad profi le contains mothers who postponed and mothers who did not postpone their fi rst birth. For the 2011-15 period, only 4.3 percent of the older fathers were start-over fathers and 1.8 percent of the older mothers were having a child with a start-over father. 2 Around 63 percent of these start-over dads are having a child with a mother who postponed her fi rst birth (2.7 percent of deliveries to older fathers), while 89 percent of older mothers in Start-Over Dad families postponed their fi rst birth (1.6 percent of deliveries to older mothers). Adding the percentage of Start-Over Dad families where the mother postponed her fi rst birth thus has little impact: In total, 64.8 percent of children born to older mothers had a mother who postponed her fi rst birth and 47.9 percent of children born to older fathers had a mother who postponed her fi rst birth. That older motherhood is more strongly associated with maternal postponement than older fatherhood and that older fatherhood is more strongly associated with family complexity than older motherhood suggests differences in their family histories should partly explain why older motherhood is a stronger marker of socio-economic advantage and resources for children.
Turning to our third research question, Tables 2 and 3 present information on how parental characteristics and resources available to children with older fathers (Table 2) and older mothers (Table 3) vary across the fi ve family history profi les. The Diverging Destinies profi le often, but not always, includes the most highly resourced older parents. In all three time periods, older parents who meet the criteria for this family history profi le are more likely to be highly educated and to be having a child with someone who is highly educated. These educational differentials are most pronounced in the most recent period. For two of the three time periods, the Diverging Destinies profi le contains the smallest percentage of older fathers with a bottom-quartile, pensionable income. Differences from older fathers in the Late Sequential Parenting profi le are, in all periods, small, and in the most recent period, the gap in bottom-quartile pensionable income is reversed. While the Diverging Destinies fathers were most likely to have a top quartile pensionable income (42 percent) in the most recent period, the Late Sequential Parenting profi le was,

Tab. 2:
Socio-economic and health resources of older fathers by family profi le no half-siblings half-siblings other childbearing across no maternal partnerships mother mother did half siblings mother did Mother postponed not postpone not postpone Postponed Contrary to what we predicted, in all three time periods, the Late Sequential Parenting older mothers are least likely to have a bottom-quartile pensionable income and most likely to have a top-quartile pensionable income, although, at the top end of the earnings distribution, differences with the Diverging Destinies mothers narrowed substantially over time. In the 2011-15 period, the percentage with top-quartile pensionable income is roughly the same for both profi les. The relative economic advantage enjoyed by older mothers in Start-Over Dad families declined over the three time periods we consider, and by the fi nal period, their pensionable income distribution was substantially different from older mothers in the Diverging Destinies and Late Sequential Parenting profi les.
When we consider economic resources available to children by equivalising older fathers' pensionable income, the Diverging Destinies and Start-Over Dad profi les are the most advantaged in all three periods, and the Late Sequential Parenting profi le more closely resembles the two less-advantaged profi les (pre-SDT and Early Sequential Parenting

Tab. 2: Continuation
Notes: When a fi gure associated with profi le 2, 3, 4, or 5 is underlined, the difference with profi le 1 is statistically signifi cant at the 5 percent level. When the fi gure associated with profi le 4 or profi le 5 is in boldface, the difference with profi le 3 is statistically signifi cant at the 5 percent level. a Amongst those who did not emigrate.
Source: Norwegian Population Register and registers on education and income top end of the distribution, older mothers in Start-Over Dad families do not become less advantaged over time relative to the Diverging Destinies and Late Sequential Parenting profi les as we observed with pensionable income. In all periods, the Diverging Destinies profi le contains a lower share of top-quartile observations than the other two profi les, but the gap has narrowed substantially over time to around 5 percentage points in the most recent period. Taken together, these fi gures suggest that maternal postponement has remained persistently associated with socioeconomic advantage and resources for children, but the stratifying effects of family complexity depend to a greater extent on the measure of economic advantage that we use. In contrast to media depictions of start-over dads, the results suggest higher resources for children not because their fathers have markedly higher incomes but because they have markedly fewer dependents than fathers in other kinds of stepfamilies. They are less likely to be having a fourth or higher order birth than other older fathers in complex families, and so have fewer dependents. Even so, we might ask whether we have underestimated the claims on their resources. Although not legally dependent, adult children might nonetheless make a substantial claim on their fathers' resources. This would not be refl ected in our equivalised income measure, which did not count adult children as dependents. When we equivalise income using a count of all children regardless of their age (results not shown, but available upon request), older fathers in the Start-Over Dad profi le no longer stand out as having greater resources for their children. 3 They most resemble those older fathers in the Early Sequential Parenting profi le. Start-over fathers are not more highly qualifi ed than older fathers in other types of blended families either, and given their higher average age, they are the most likely to die within 18 years of the delivery (based on deliveries in the 1991-95 period only). Hence, it is not clear that older fathers in Start-Over Dad families provide, on average, greater resources for children.

Discussion and Conclusion
This paper takes as its point of departure previous research on the impact of family changes which have accompanied the SDT. Focusing predominantly on women's fertility and family experiences, and dramatic trends in the postponement of the transition to motherhood, the demographic literature has tended to use advanced maternal age as a marker of resources which benefi t parents and children (McLanahan 2004). Drawing on previous research which has questioned whether demographic indicators like maternal age or marital status mean the same thing 3 Because few women who become older mothers have adult children, their income results do not change much when we count all children. Only those mothers in the Early Sequential Parenting profi le (half-sibling(s) and the mother did not postpone) show any real change: They are slightly more likely to have a bottom-quartile income and slightly less likely to have a topquartile income when their income is equivalised using all children, dependent or otherwise.
across different contexts or different population groups (Sigle/Goisis 2019), and inspired by feminist research which has noted that conceptual frameworks that were developed to explain the experiences of one gender (usually men) are often inadequate when the focus is shifted to the other gender, we ask whether and to what extent advanced paternal age means the same thing as advanced maternal age does. In other words, is it a similar marker of resources and advantages? We also examine how the resources associated with advanced paternal and maternal age have changed over the course of the SDT in Norway.
Using high-quality Norwegian register data covering three time periods, we fi nd that around the time their children are born, older fathers (aged 40+) tend to be better resourced than younger fathers are. They are more highly educated and, even taking into account the number of biological dependent children that require their support, their fi nancial position tends to be better. However, their resource advantages have narrowed in recent decades. For most measures, the largest changes were observed from the fi rst (1991-95) to the middle period (2001-05). An economic recession beginning in 1988 and affecting younger people more than older people might have exacerbated income differences in the fi rst period. In addition, many older fathers in the early 2000s will have come of age prior to the rapid expansion of higher education (Aamodt/Arnesen 1995) which many of the younger parents they are compared to would have benefi tted from.
When we examined whether the differentials and trends were similar to what is observed when older (aged 36+) and younger mothers are compared, we found that maternal age tends to be consistently associated with more socio-economic advantage and more resources for children than paternal age is. Moreover, the age-related earned income advantage increased for older mothers and narrowed for older fathers over the three time periods. In the early 1990s, Norway was still often described as having a housewife model. However, policy developments in the 1980s had already initiated a move towards a dual-earner model (Lewis 1992;Frønes 1997). Older parents of children born in fi rst period transitioned to adulthood at a time when policies and the wider culture did not support mothers' full-time employment (Ellingsaeter/Jensen 2019). If highly educated older mothers in the later periods faced less statistical discrimination in the labour market than older mothers in the early period, highly educated mothers would be more likely to attain positions and pay commensurate with their education in recent years.
Differences in the family history profi les of older fathers and older mothers also appear to have contributed to the observed gender differences and trends. The more advantaged family history profi les contain a larger share of older mothers than older fathers, and these gender differences have widened over time. In the most recent period, the majority of older mothers in Norway followed what Sarah McLanahan (2004) described as a high resource trajectory: they had postponed their fi rst birth (facilitating investments in education and career) and neither they nor their partner had children with someone else. Only about one-third of older fathers lived in such families. Older mothers are more likely to have postponed their fi rst birth than are the partners of older fathers.
We fi nd weaker and more qualifi ed evidence for the stratifying effects of family complexity. The older parents who followed the Late Sequential Parenting and Start-Over Dad trajectory are less educated than the Diverging Destinies parents, but their pensionable income distribution (Late Sequential Parenting) or equivalised income distribution (Start-Over Dad) is similar. When we take into account the number of dependent children, older fathers -but not older mothers -in the Late Sequential Parenting profi le are less economically advantaged. Their relatively large family size means they have fewer economic resources available for their children. The Start-Over Dad fathers do not have higher pensionable incomes but, with fewer dependent children, they have more economic resources available for their child. When we consider all children rather than dependent children, older fathers in the Start-Over Dad profi le no longer appear similar to the Diverging Destiny fathers. For older mothers, the results are less dependent on the indicator of economic resources we use, in part because older mothers in the Late Sequential Parenting profi le do not have much larger families than the other profi les with relatively high pensionable incomes, and they are less likely to have adult children.
Taken together, the results illustrate important differences between older fathers and older mothers in contemporary Norway. While our fi ndings suggest that regardless of gender, older parents are better resourced parents, they also indicate that the resource advantage is more pronounced for older mothers than older fathers. Some family history profi les are less socio-economically advantaged or well-resourced than others, and these profi les are more typical of older fathers than older mothers. If the family histories of older fathers and older mothers were more similar, gender differences in the extent to which parental age is a marker of socioeconomic advantage and resources for children would likely narrow. Nonetheless, some family history profi les are more strongly associated with advantage for older mothers than for older fathers, in part because the characteristics of the families differ when the focus is on older fathers. For example, older fathers in the Late Sequential Parenting profi le are more likely than older mothers in the same profi le to have at least four children, and so have greater claims on their resources.
Our results highlight the importance of examining the resources and family histories of both parents. The more complex family experiences of older fathers may have resource implications for the children born to older mothers. Although older mothers are less likely than older fathers to have had children with a previous partner, the share of older mothers who give birth to children with a half-sibling increases substantially when the family histories of the fathers are considered as well. In recent years , about 15 percent of older mothers have had a child with a previous partner, but nearly 27 percent are having children with at least one half-sibling (results not shown but available upon request). The extent to which older parents must negotiate the costs and challenges associated with having had a previous family may be under-estimated when older parenthood is viewed exclusively through the lens of the mother's experiences (see also Cancian et al. 2011). At the same time, for older mothers, maternal postponement appears to be linked to more (maternal) resources even when half-siblings are present. Differences in the family histories of older fathers and older mothers underscore the need to attend carefully to what demographic indicators represent in a given context and for a given group.