This project aims to identify how changes in gender relations influence fertility, child rearing, couple stability, or even the likelihood of marriage. There is some evidence that emerging family demographics are associated with social polarization along several key dimensions simultaneously (McLanahan, 2004). I propose a multiple equilibrium model to address three basic empirical questions: 1) what are the determinants of changing (family) equilibrium dynamics? 2) The determinants of new socially stratified patterns of fertility and marital behaviour? And 3) how does rising family polarization affect parenting behaviour and child outcomes? The project, then, has three distinct sub-components.

Recent research has brought to light a number of puzzling changes that seem to contradict the core convergence hypothesis in the Second Demographic Transition thesis. Firstly, the correlation between female employment and fertility has reversed. Forerunner countries in the transition have experienced a rise in birth rates while Southern Europe has slid into a prolonged 'lowest-low' fertility equilibrium (Billari and Kohler, 2004). Secondly, we see radical reversals in the social correlates of family behaviour precisely in those countries that have been the vanguards of the demographic transition. Fertility in Scandinavia is now greater among higher than lower educated women. Similarly, in the Nordic countries as well as in the US, marriage rates are now higher and partnerships more stable among the highly educated; vice-versa, divorce and single motherhood is increasingly concentrated among the low educated (McLanahan and Percheski, 2008; Harkonen and Dronkers, 2006). And we observe a parallel trend towards increased assortative marriage, in particular at the top (Blossfeld and Timm, 2003). Again, this trend appears stronger in the vanguard demographic transition countries.

These trends are closely associated with the revolution in women's roles. But the nature of the correlation is complex. Low fertility, non-marriage, and divorce have typically been biased towards higher educated career women. This is still the case in many countries. One obvious explanation lies in the steeper life-time opportunity costs associated with motherhood (Hotz et.al., 1997; Boeri et.al., 2005). But changing demographic behaviour in the Nordic countries would suggest that the opportunity costs should have diminished substantially. How? Many demographers now argue that fertility recovery requires more gender egalitarian social policies (McDonald, 2002). Family behaviour will, however, also depend on couple arrangements with regard to labour supply and specialization in domestic work. Recent studies support this view. In some countries, men's contribution to domestic work has increased sharply in recent years, albeit mainly among the higher educated (Esping-Andersen, 2009).

The point of departure for the proposed research is that families are increasingly stratified both in terms of demographic behaviour (divorce and lone parenthood is increasingly concentrated among the less educated) and in terms of couple relations (egalitarian arrangements are predominantly adopted by the highly educated). Worse, the same stratified pattern carries over to parenting. To illustrate, my own work suggests that higher educated parents dedicate more than twice as much time to their children as to the lower educated. Mechanisms of social selection are indisputably at the roots of not only family structural polarization, but also of widening gaps in the conditions that influence children's life chances. The same social selection process that produces low income and more unstable families also produces less nurturing and stimulus.

A multiple equilibrium approach

These ongoing trends display features characteristic of multiple equilibria. Multiple equilibrium models have a longstanding tradition in economic analysis, but despite their affinities to sociological norm theory they have not found their way into sociological research.  Building on some of my recent work, I argue that a multiple equilibrium approach can be both theoretically and empirically powerful for identifying the dynamics and consequences of family change.

It is theoretically assumed that stable equilibria produce Pareto optimal outcomes. Any departure from equilibrium is likely to yield inefficient and/or inequitable outcomes. The economists' approach has one major advantage over conventional sociological norm theory, namely that it rests on explicitly dynamic hypotheses. The emergence of multiple equilibria depends, in the first instance, on exogenous triggers that recast expectations. An alternative equilibrium will consolidate to the extent that the external 'shock' is followed by a self-reinforcing endogenous process of adaptation. The endogenous dynamics should display an accelerating momentum: initially slow and subsequently with increased rapidity (Fukao and Benabou, 1993). The logic is very similar to the 'critical social-mass' effect that Breen and Cooke (2005) describe. Multiple equilibria will include at least oneunstable equilibrium. It is of course misleading to use the concept of equilibrium here. It represents in a sense a normative flux that is likely to yield non-Paretian outcomes, i.e. inefficiencies and inequities. One may think of it as a half-way house between a fading old equilibrium and an embryonic new one.

As the conventional 'Parsonian' family is eroding, multiple family equilibria emerge. The concluding decades of the 20^th Century saw the emergence of a rival 'egalitarian' equilibrium, constructed around the dual career couple. Here we must be careful about the associated fundamentals. It does not generally include a renunciation of motherhood; the two-child norm remains basically intact as a preference. And couple specialization should become blurred as the partners' market productivities and labour supply converge. In other words, the 'ideal type' family would be based on double-earner couples with children, and gender symmetry in the domestic sphere.

In ongoing work, I have applied this framework to the division of childcare and housework in Denmark, the UK, and Spain. The traditional equilibrium is still strong in Spain, minoritarian in the UK, and has for all purposes disappeared in Denmark. The latter, in turn, is the only case where a new egalitarian equilibrium is becoming dominant (52% of couples, compared to 15% in Britain and less than 1% in Spain). And, in all three cases we identify a large (especially in Spain and the UK) unstable equilibrium in which couples only very partially adapt to wives' relative employment status, and in which domestic specialization appears both inefficient and inequitable.

Multiple equilibrium dynamics and family behaviour

The proposed research in this sub-component of the project is largely explorative. The existing literature provides no clear hypotheses as to how such family-equilibria dynamics should unfold with regard to the endogenous process.  Indeed, herein lies one reason why the project may be ground-breaking. There are, however, a number of evident hypotheses that must guide analyses. Generally speaking, normative adherence is guided by peer-group effects. Clearly, what is required for the endogenous dynamics to accelerate is the emergence of a critical social mass that adopts a novel normative framework.

A first logical hypothesis is that the erosion of the traditional equilibrium will parallel women's entry into paid employment; in turn, an accelerating momentum towards the egalitarian equilibrium will depend on the degree to which women's employment patterns converge with men's and their contribution to household income becomes decisive for joint welfare. We know that highly educated women generally represent the vanguard, so the key lies in changes within lower levels of education.

Secondly, the endogenous accelerator-effect must require a reorganization of domestic life. This means that a critical mass of men must adopt more gender egalitarian behaviour in home production. Recent time-use research suggests that the accelerator effect obtains. Following decades of stagnation, men's contribution in some countries suddenly underwent a quantum leap (Bonke and Esping-Andersen, 2010; Esping-Andersen et.al., forthcoming). What might trigger such sudden leaps?

This leads to a third hypothesis: adaptation is, no doubt, driven by socialization and social origins. The mother's status should be key. If so, the timing of the equilibrium shifts is likely to coincide with the arrival of cohorts within which a significant mass of mothers worked. Finally, institutional factors will clearly influence equilibrium shifts. The move towards an egalitarian equilibrium is likely to be influenced by policies that help reconcile family and working life. Hence, the process of endogenous acceleration is more likely to unfold where parental leaves are generous, parents enjoy job guarantees, and where child care (and elderly care) services are affordable and available - i.e. where the career penalties of motherhood are minimized. We must obviously assume that such policies are themselves endogenous with respect to family adaptation to women's changing roles, but only up to a point.

Fertility and marriage

The classical model of Becker and its empirical applications focused heavily on the male's breadwinner capacity in order to explain both the quantity and timing of fertility (Hotz et.al, 1997). Contemporary fertility models, in contrast, focus primarily on the opportunity costs related to motherhood. Not surprisingly, the conditions that help reconcile motherhood and careers have assumed centrality. Hence, one would predict more postponement, fewer births, and also childlessness among women with higher education and/or stronger life-time career commitments. 

The puzzle, then, is why fertility is now rising among the highly educated and falling among less educated women in some countries. There is substantial evidence that policies that support working mothers matter greatly (Sleebos, 2003). They seem to influence tempo more than quantum. But couple relations also matter. The propensity to remain childless, or to have children outside a union, has fallen among the higher educated while it has risen among the low educated in Scandinavia and the US. It is not yet clear whether this reversal is occurring in other countries. Such intra-national differences are unlikely to be driven by welfare state effects. Building on a few recent studies there are reasons to believe that male adaptation to gender symmetry is crucial (Duvander et.al., 2010; Brodmann et.al., 2007; Cooke, 2008).  

A first hypothesis is that fertility should be higher within the traditional and the egalitarian family equilibrium. In the traditional equilibrium, fertility will depend primarily on the male breadwinner's earnings capacity. In the egalitarian, fertility should be associated with the male partner's contribution to domestic work. The highly educated may face steeper opportunity costs of children but, as all recent data show, they also have exceptionally strong child preferences (Bonke, 2009). In contrast, I hypothesize that lower fertility among the less educated has to do with the eroding position of less skilled males as well as their greater disinclination to adopt more gender-symmetric roles. Women who face unfavourable marriage markets may also face fertility constraints. But it is in the unstable equilibrium that we should expect systematically lower fertility. For one, these are predominantly couples where the woman does have a career commitment - even if less binding - and where the male partner only reluctantly (if at all) contributes to home production. Low fertility in the unstable equilibrium should be particularly pronounced among women with higher levels of education.

As regards marriage behaviour and, in particular, couple stability the empirical hypotheses are quite parallel. We see a tendency for delayed marriage (as for fertility) and a stronger propensity to co-habitate.  As a general rule, delayed couple formation promotes greater stability, although the effect is non-linear. Very early marriages tend to be unusually unstable and the risk of rupture declines for later marriages (Martin, 2004). We know that the difference in mean age of marriage between low and high educated women has widened over the past decades. But if we are concerned with couple stability we need to understand the mechanisms that select people into a union in the first place. In the past, higher educated women were hugely over-represented among the never-married; this has now reversed itself in some countries. The logic is similar to that of fertility: the eroding position of low-skilled males and their disinclination to embrace more gender equality. Vice versa, the rise in marriage among the highly educated (especially for cohorts born since the 1960s) should be related to the leap in gender egalitarianism that we observe for higher educated men. The stratified profile of men's adaptation implies that marriage markets have improved for high educated women and deteriorated for the less educated.

Greater couple instability is to be expected within unstable equilibria since this is where inequities and inefficiencies are likely to be most pronounced. Controlling for social selection into a union in the first place we would, in comparison, expect greater stability among traditional and egalitarian couples. But the life course dynamics are crucial in this regard. A major focus of the research will be on changes in either of the partners' employment status (for example from inactive to active, or vive-versa) and how this affects couple specialization. We would expect greater couple volatility where a major change in job status is not matched by partner adjustments with regard to home production.

Family Stratification and Parental Investments in Children

In the classical Becker model (Becker and Tomes, 1986), parent-child correlations in terms of life chances depend primarily on inherited abilities, on monetary and time investments. Institutional factors, such as the education system, can modify the impact of the latter two. Very recent research suggests that genetic inheritance accounts for between 40 and 50 percent of the total parental effect on child outcomes (Bjorklund and Jantti, 2009; Nolan et.al., forthcoming).

McLanahan's (2004) diverging destinies thesis builds on the widening distance between families in terms of family structure, economic resources, and of parental time investment. Preliminary evidence from the US and Scandinavia seems to support this hypothesis. But it has, so far, not been systematically and cross-nationally tested with empirical data. To be sure, inter-generational mobility research has for decades examined how family structural characteristics (particularly lone motherhood and divorce, but also marital homogamy) influence child outcomes. This research has almost exclusively examined family-effects within one country only. Comparing results obtained from country-specific research does suggest that not only family structure, but also financial resources, play a stronger role in inter-generational transmission in some countries than in others (Bjorklund et.al., 2009; Nolan et.al., forthcoming). Such effects are particularly strong in the US and, vice-versa, weakest in Scandinavia.

The influence of parental socioeconomic status is rivalled by other family characteristics, in particular those related to the 'learning culture'. The PISA studies, for example, demonstrate that the volume of books in the home is a powerful predictor of children's cognitive performance. We have exceedingly few studies that identify the direct effects of parenting time inputs.

The overriding question I pose is whether changes in family structure and their correlates contribute to greater polarization in parental child investments, measured in terms of monetary but especially time allocation. Cross-sectionally, we expect that the variance in time and money allocation will widen the greater is the social concentration of at-risk families, lone parents in particular. Dynamically, we should expect to find significant changes in child investment associated with key events such as divorce and union formation. The educational gradient of parental child investments is well-documented. Bonke and Esping-Andersen (2010) suggest additionally that the logic of child care differs sharply between low and high educated parents, primarily because the latter espouse strong preferences for maximizing time with the children and because they prioritize 'developmental' inputs, such as reading with the children.

I plan also to analyze how variations in time inputs in early childhood influence later child outcomes with respect to one variable, namely educational attainment. The availability of long panels for some countries should make such outcome estimation possible in terms of completed upper-secondary level education and transitions into tertiary level.

Parental child investments occur of course within institutional contexts. Child care institutions may compensate for, or supplement, parental inputs and also influence later child outcomes. There is strong evidence that high quality early care (beyond age 1) has positive effects on child cognitive learning and school performance. The effect is especially positive for less-privileged children (Heckman and Lochner, 2000; Waldfogel, 2006; Esping-Andersen, 2009). What, however, is not well-documented is whether (and among whom) external care is supplementary or compensatory. There is some evidence that it tends to be supplementary among highly educated parents due to their strong pro-child preferences (Bonke and Esping-Andersen, 2010).

Data and Methodology

The three sub-components of the project all make similar data demands. We need panel-type micro-data that allow us to follow individuals and families over a long time. We require detailed information on family-related events and, perhaps most importantly, on spousal time allocation. For the study of parental investment we require, additionally, that we can follow the children into early adulthood. We also need to compare across countries. These requirements put severe limits on what is possible.

Fortunately, such analytical requirements are met by the availability of long panel studies for Denmark (the time use panels for 2001 and 2010 coupled to registry data), Germany (GSOEP), Britain (BHPS) and the U.S. (PSID). These countries offer excellent contrasts in terms of both family structural change and women's employment profile. But there are limitations: the Danish data will only permit limited analyses of child outcomes (due to right-censoring) but this is compensated for by the inclusion of a special module on parental expenditures on their children. The drawback of the PSID is that detailed time dedication data is available only for the mother.

The common analytical approach is longitudinal and will apply event-history modelling techniques, in particular since durations and life course events are crucial to model. The study will be centred at the Universitat Pompeu Fabra, but will include a number of international collaborators over the duration of the project: Jens Bonke and Anders Holm (Denmark); Lynn Prince Cooke (Britain); Juho Harkonen (Stockholm); Francesco Billari and Arnstein Aassve (Milano); Amy Hsin (US). Internal UPF collaborators include: Pau Baizan, Marta Dominguez, Maria Jose Gonzalez, Jorge Rodríguez, and Sebastián Sarasa.