Most young children, even those with stay-at-home mothers, experience some nonmaternal care throughout the preschool years, ranging from an afternoon with grandparents to a few hours with a babysitter at home or at a play group in a local center. But these limited episodes are not sufficient to facilitate full-time participation in the labor force. Extended child care is the cornerstone of family-friendly policy. Without alternate care for thirty hours or more a week during the preschool years, two parents employed in regular full-time jobs would be unable to raise two children unless one of the working parents withdrew from the labor force for eight years.4 Even then, a few hours of before — and after-school care would be necessary if both parents worked a regular nine-to-five shift or the longer hours of some high-powered professional firms.5
After-school programs that serve older children have excited less passion and received less attention in family-policy deliberations than programs that deliver extended, nonmaternal day care to preschoolers. Because the provision of extended preschool care diminishes the traditional role of motherhood during the early and critical childrearing years, it is hard to make a substantive assessment—positive or negative—of this service without hitting a sensitive nerve somewhere. Working mothers are distraught by studies that suggest they might be shortchanging their children by placing them in day care for long periods of time during the preschool years. And stay-at — home mothers are equally pained by studies suggesting that children in out-of-home care thrive more intellectually and socially than their own children, in whom they invest so much personal time and energy each day.
Research on the touchy subject of nonmaternal care and its interpretation in public discourse are further complicated by the blurring of two agendas, each with its own idea of what day care should achieve. One agenda is to furnish day care that serves as an equivalent substitute for the nurturing and socialization provided by an average mother. Following this course, day care serves mainly as a public provision to harmonize work and family life by permitting mothers to shift their labor to the market while ensuring that their children receive decent care that resembles acceptable standards of mothering. These programs are structured to accommodate the schedule of working parents, with early drop-offs, long hours, and late pick-ups.
The other agenda goes beyond supplying the care of an average substitute mother. It involves the provision of an enriching day-care experience that promises to improve on the socialization and academic skills that preschool children would otherwise acquire through maternal care. Here nonmaternal care aims not so much to help women balance motherhood and work as to help young children develop more fully than they might under the daily supervision of their mothers.
The agenda of developmental enhancement is usually associated with nonmaternal care directed toward children from disadvantaged families. These services are structured more to meet the social and educational needs of children than the work schedules of their parents. The largest and best known preschool program is Head Start, which usually runs for part of the day during nine months of the year.6 One of Head Start’s predecessors, the legendary Perry Preschool, was a model early-intervention program that ran thirty weeks of the year, with two-and-a-half-hour classes on weekday mornings and weekly one-and-a-half-hour home visits with mothers and children in the afternoon—a schedule hardly intended to facilitate the mothers’ full-time employment.7 One of the reasons sometimes given for the success of this program is that part of the intervention was aimed at parents.8
The two agendas of child care—enhanced development and substitute mothering—blur because they represent a continuum rather than two mutually exclusive alternatives. Substitute mothering seeks to provide an experience beyond mass-produced custodial care by offering the kind of intellectual stimulation and emotionally supportive interaction that children are expected to receive under healthy maternal care. On this side of the continuum, however, there are no doubt some services that approach custodial care, offering little more than warmth and safety. But child-care providers and parents rarely admit to it, claiming instead that these services fully meet the nurturing standards of substitute mothering. Developmental enhancement moves beyond substitute mothering. These services aim to deliver a more structured and advanced level of socialization and academic preparation than children would otherwise receive at home—giving them a “head start.” All sorts of day-care arrangements along this continuum lay some claim to developmental benefits and insist that children thrive under their daily supervision. To do otherwise would put them out of business. Thus, the research and public discourse on these arrangements tend to concentrate on issues of developmental enhancement. Once the central question about the impact of nonmaternal care on child development is posed, there is always the incendiary possibility of uncovering detrimental effects.
Given the explosive potential of the subject, it is perhaps fortunate that much of the day-care research has been plagued by methodological flaws, resulting in a corpus of findings muffled by a host of qualifications that acknowledge selection bias, inconsistency of findings, lack of data on longterm effects, inability to measure subtle characteristics, and difficulty in establishing causality. In addition, there have been problems disentangling the complex interactions of numerous mediating variables—quality of care, number of hours in day care each day, maternal sensitivity, home environment, child’s temperament, child’s age, social class, father’s involvement, mother’s work schedule, child’s sex, and number of children in the family, to mention the most obvious. A careful review of forty studies in the late 1970s concluded that knowledge about the effects of day care was exceedingly limited and that what was known could not be generalized to the kind and quality of care available throughout the country.9
Buffered by methodological qualifications, the first wave of child-care research in the 1970s and early 1980s avoided rattling the cage by making vague summary statements about the consequences of nonmaternal care and maternal home care, such as day care “typically does no harm,” “children of working mothers do as well as those of mothers who stay at home,” and “children of employed and non-employed mothers do not differ on various child adjustment measures.”10 Years later, in 1999, this theme was echoed in an American Psychological Association press release, which reported a study’s conclusion that “having a working mother does no significant harm to children.”11 The findings attracted considerable media attention and were covered by the CBS Evening News, the Washington Post, and the Atlanta Journal-Constitution, among others. The media paid little attention to the fact that the findings were based on a sample of disproportionately low-income, single, young, minority mothers; it also ignored other details, including that the definition of “working” mothers lumped together those who worked a few hours a week with others who worked up to forty hours, and that the emotional assessments of the children were based primarily on what the mothers told government interviewers.12
The “does no harm” refrain was occasionally reinforced by child development experts, who asserted, for example, that “children are usually better off with a satisfied substitute care-giver and a happy part-time mother than with an angry frustrated full-time mother.”13 Statements along these lines conveyed the sense that nonmaternal care was an equivalent substitute provision for traditional mothering—while implicitly encouraging bored, frustrated mothers suffering from Friedan’s problem with no name to join the female exodus into the labor market, which was in high gear at the time.
These quantitative studies in the 1970s and 1980s were usually bland and reassuring; however, qualitative assessments of nonmaternal care, which looked at how children experience life in these settings on a daily basis, offered a more discomforting view. Deborah Fallows’s appraisal of the day-care experience found a cheerless, desultory milieu in most of the facilities she observed around the country. Fallows described the quality of a child’s daily existence in a typical day-care center, where the main activity was filling time: “He didn’t do badly—he roamed independently, joining in when he felt like it, taking off when he didn’t. He got no individual attention, because he didn’t demand any. He got no special instruction, because none was offered. No one talked to him or hugged him, because there weren’t enough adults to go around.”14 There were exceptions, of course, the best of which were very pricey.
Seeking to address some of the weaknesses in earlier studies, a new wave of rigorous empirical research was launched in the early 1990s, starting with the National Institute of Child Health and Human Development’s (NICHD) Study of Early Child Care. This large-scale longitudinal investigation followed an initial sample of 1,364 children from birth through age seven in ten vicinities across the United States. The study was conducted by the NICHD Early Child Care Research Network—a team of twenty-five researchers from ten universities. Members of the research network and other scholars given access to the NICHD data generated a voluminous body of empirical work analyzing connections among developmental outcomes and a wide range of variables such as the quality of the childcare environment, the quality of the school environment (as the children entered kindergarten and the primary grades), and the quality of home and family life.15
The NICHD study is one of the most exacting and comprehensive of its kind to date, described by child development experts as the gold standard in this area of research.16 Despite such praise, however, the survey methods used to produce the findings discussed below are subject to the type of criticism often brought to bear on nonexperimental studies (that is, studies that do not employ a rigorous experimental design, which involves random assignment of subjects to experimental and control groups).17 The NICHD researchers studied child care from an instrumental perspective, focusing on the results of nonmaternal care. The outcomes most frequently brought to the public’s attention analyzed the effects of quality care on attachment relationships with mothers, behavioral problems, and cognitive development. Although the numbers were clear and the statistics conveyed a voice of scientific authority, their meaning was open to interpretation.
Regarding the emotional bond between mothers and children, psychologists believe that the security of the mother — child attachment in the early years exerts a measurable influence on the child’s feelings about self and others, and on the child’s future capacity to form relationships and regulate aggression. Prior to the NICHD research there was heated discussion about child care’s potentially negative impact on the mother-child attachment. In the late 1970s, for example, Selma Fraiberg expressed concern that this bond would be seriously impaired by placing youngsters in extended child care. She worried that children in day care would be subjected to the rough justice of the preschool playground, while learning “that all adults are interchangeable, that love is capricious, that human attachment is a perilous investment, and that love should be hoarded for the self in the service of survival.”18 But in an initial systematic review of the empirical research literature, Jay Belsky, a highly respected psychology professor, and his colleagues found little evidence that day care for infants had detrimental effects on the mother-child bond.19 Later, however, Belsky reexamined the evidence and found it more troubling. He concluded, with painstaking caution, “that if one does not feel compelled to draw only irrefutable conclusions, a relatively persuasive circumstantial case can be made that early infant care may be associated with the increased avoidance of mother, possibly to the point of greater insecurity in the attachment relationship, and that such care may also be associated with diminished compliance and cooperation with adults, increased aggressiveness, and possibly even greater social maladjustment in the preschool and early school-age years” (emphasis in original).20 This statement was made in the mid-1980s, when the research evidence rarely controlled for child-care quality and family differences.
By the mid-1990s, the NICHD study was able to conduct more sophisticated analyses using information on the quality of care and family differences. Revisiting the issue of mother — child attachment, the NICHD found that extended nonmaternal child care was not linked to increased insecurity, except in those instances where maternal sensitivity and quality of child care were judged to be low. Maternal sensitivity was measured in terms of the mothers’ expressions of supportiveness, intrusiveness, positive regard, and hostility observed in videotaped interactions with their children. Quality of child care was assessed according to structural characteristics, such as the caregiver-child ratio and the number of children under supervision, as well as the caregivers’ training, education, and behavior, such as language stimulation and emotionally supportive interactions. The NICHD research showed that the likelihood of insecure mother-child attachment at fifteen months increased when children were in nonmaternal care more than ten hours per week only for those children whose mothers were judged as highly insensitive and who received a low quality of care.21 A similar finding was observed at thirty-six months.22
The good news is that when the data are controlled for quality and maternal sensitivity, extended child care does not appear to diminish the security of attachment. But this assessment starts to waver a bit in light of further analyses. First, when evaluating the quality of nonmaternal child care in centers against the guidelines of the American Public Health Association and the American Academy of Pediatrics, the NICHD research team found that “the results were not encouraging.”23 Other studies of 628 centers gave an overall average quality rating of a little less than halfway between minimal and good.24 Not only is most child care of mediocre quality, but it was also found that mothers who were judged as less sensitive were more likely to have children in centers where the quality of care was not highly rated.25 Finally, mothers who were judged low on the measures of maternal sensitivity tended to have children who spent more hours in nonmaternal care.26
The web of cause-and-effect relationships among mother — child attachment, family characteristics such as maternal sensitivity, and quantity and quality of nonmaternal care is difficult to untangle. The issue depends in part on whether maternal insensitivity is seen as an inherent personal characteristic or as a reaction to broader circumstances. Women employed full-time necessarily have children who spend extended periods in nonmaternal care. Under the stress of trying to balance full-time employment and motherhood, it is possible that these mothers will be less relaxed, patient, tolerant, and responsive in interactions with their children—thus scoring lower in maternal sensitivity—than mothers who are not as pressed in their daily lives. In these circumstances the association between maternal insensitivity and extended time in child care might well stem from the antecedent condition of fulltime employment. Since the quality of most day-care services tends to be no more than mediocre, even those who want and can afford high-quality care have a hard time finding it. Thus, although child care per se poses little threat to mother-child attachment, given the current attributes of child-care offerings in the United States, children under three years of age in extended nonmaternal care on average probably have a somewhat higher risk of developing insecure attachments than those cared for by mothers at home.
In contrast to the weak link between the amount of time spent in nonmaternal care during the preschool years and the security of the mother-child attachment, the NICHD study revealed a comparatively robust relationship between the amount of time spent in nonmaternal care during the preschool years and the likelihood of behavioral problems, particularly aggressive behavior toward other children, showing up in kindergarten.27 Specifically, the research found that children in regularly scheduled nonmaternal care more than thirty hours per week during the first four and a half years of life were almost three times more likely to behave aggressively toward other children than children who had been in nonmaternal care less than ten hours per week. Although the difference sounds quite large, in absolute terms it represents a gap of 11 percentage points—between problematic behavior expressed by 17 percent of children who were in extended nonmaternal care and by 6 percent of children who were in nonmaternal care for shorter periods of time. Another way of interpreting the results is to highlight the good news that behavioral problems were not evident in 83 percent of children in extended care.28
The apparent impact of extended care was moderate, but not trivial. Further analyses lend increasing credibility to these findings. The data showed not merely a simple correlation between quantity of care and behavioral problems but a more compelling “dose-response” relationship. That is, the average scores on problem behavior rose consistently with the amount of time children spent in nonmaternal care.29 Moreover, the relationship between quantity of care and problem behavior remained significant even after controlling for the quality and type of care, and for family background.30 It is possible, however, that selection bias might have accounted for the higher levels of aggressive behavior among those in extended care, if rambunctious and difficult-to-manage children were more likely to be placed in care centers for longer hours than placid and easy-to-manage kids.
Does it really make much difference if the behavior of children who spend an extended period in day care is more problematic than that of other children when they enter kindergarten? The answer depends, in part, upon whether this result is judged from an existential or instrumental perspective. From an existential perspective, how children experience life on a daily basis is as important as future outcomes. In this view, the understanding of a good life is based on what takes place during the journey as much as the eventual destination. Thus, the existentialists’ answer is yes, the higher level of aggressive behavior matters, since it would make kindergarten a less pleasant experience in the voyage of life for both the kids who act out and their classmates. From an instrumental perspective, concerns are more sharply focused on the long-term impact of extended day care on cognitive and behavioral outcomes. Instrumentalists want to know what the aggressive behavior in kindergarten will lead to—will the aggressive kids grow up to be thugs or captains of industry? Or will they simply mellow out over time?
In fact, the children did eventually mellow out. The data show that when the children were evaluated after kindergarten, the earlier relationship between the quantity of early child care and levels of problem behavior began to dissipate, and by the third grade it was no longer statistically significant. These findings at the third-grade level do not close the book, since effects associated with early child care that seemed to disappear have been found to reemerge later, particularly during periods of developmental transitions. As the study’s authors caution, “It remains to be determined if these relations with early child care remain, dissipate or grow in early adolescence, a critical transition period for many children."31
While behavioral problems associated with extended child care regardless of quality were no longer evident by the third grade, findings from the NICHD study showed that academic gains were linked to high-quality care and were sustained through the third grade.32 These findings are supported by other large-sample studies.33 In most cases, however, the gains were modest in size and had the largest impact on vulnerable children from disadvantaged families.34 And there is also some evidence that academic gains associated with preschool attendance fade over the early years of elementary school.35 Still, several of the model early-intervention programs, such as Perry Preschool and the Carolina Abecedarian program, showed promising long-term results. Coming from very deprived backgrounds, the Perry Preschool participants were more likely to have graduated from high school and to be employed by age nineteen, and to earn more, have fewer criminal arrests, and receive less welfare by age twenty-eight, than the control group.36
No doubt, there is a good case to be made for providing extended nonmaternal care to seriously disadvantaged children from homes that fail to offer sufficient cognitive stimulation and emotional support or that present high risks of abuse and neglect. Yet, for a variety of reasons, such as limited education, social problems, and psychological deficiencies, mothers from these homes are least likely to find employment that would pay for the costs of such care. (Ironically, some of these women could conceivably become child-care workers, earning a median hourly wage of $8.o6.)37
But the vast majority of young children in the United States are not from disadvantaged families. Their life experiences are shaped by competent mothers (over 80 percent of whom have at least completed high school) whose nurturing and personal interactions are vested with more devotion than money can buy. These children are least likely to gain substantial or persistent social and cognitive benefits from high — quality nonmaternal care. Indeed, cut through the professional jargon of “structure” (small groups with a high ratio of caretakers to children) and “processes” (language stimulation and positive reinforcement), and you find that the essential definition of high-quality day care approximates the warm, personal, and supportive interactions associated with average maternal care. There is much evidence that family characteristics, particularly mother’s education, have considerably more effect than day-care experiences on cognitive development and other outcomes.38 And there are some indications that children from well-functioning, middle-class families, particularly males, may be disadvantaged by nonmaternal care.39
What does all this tell us about the consequences of extended day care as a family-friendly policy? In trying to understand the impact of day care it is helpful to distinguish between programs that offer substitute mothering designed to mind preschool children during their parents’ working hours and those that offer special services designed to enhance children’s cognitive and emotional development beyond what they would normally receive at home. Overall, the research suggests that the emotional, behavioral, and cognitive effects of nonmaternal day care are mixed and vary depending on the intensity and quality of the service as well as the characteristics of the children and their families. Neither the advantages nor the detrimental effects appear to be large, and both fade over time to some extent. A few of the most costly and well-executed programs intended to enhance development have shown promising long-term results, predominantly for children from socially and economically disadvantaged families. It is important to recognize, however, that while these programs offer potential benefits for children in vulnerable circumstances, for the most part they do not provide the extended hours of care that are necessary to facilitate the parents’ full-time employment.
In sum, after more than thirty years of research we have gained a deeper appreciation of the complexities of gauging how extended child care affects children, but our conclusions remain in limbo. There is little evidence that these programs benefit the daily experiences or enhance the developmental capacities of most children who are not highly disadvantaged. At the same time there is no convincing evidence that these programs have long-lasting detrimental effects. Given the uncertain impact on children, all that can be said for extended day care as a family friendly policy is that it is most friendly to mothers who want or are required by dire necessity to work long hours outside the home.
Family Policy: Promoting Mothers or Markets?
What about the benefits of the broader package of conventional family-friendly policies generally available in the major industrialized democracies? To whom or what are they friendly? Western European countries are well known for having a more powerful arsenal of day care and other family-friendly benefits than the United States. For example, more than 70 percent of the children from age three to school age in Belgium, Denmark, France, Germany, Italy, the Netherlands, and the United Kingdom are in publicly financed child care.40 Paid maternity leave, long periods of vacation, and government allowances for children and families are also common.
But the high levels of public care and other benefits have not been matched by the private production of children. Indeed, in a number of European nations, the birthrate is perched on the brink of demographic suicide. At the current level of decline, Belgium’s native-born population is poised to fall by 12.5 percent between 2001 and 2020. Spain’s population is expected to plunge by almost 25 percent over the next forty-five years, and during that time Germany is also expected to suffer a huge population loss, comparable to the size of the population of the former East Germany.41 No member nation of the European Union currently has a fertility rate high enough to replace their existing population (replacement level is 2.1). And United Nations projections show the fertility rate remaining well below the replacement level through 2050.
The shrinking of native-born populations between 2005 and 2050 could be offset somewhat by the movement of an estimated 98 million international migrants from underdeveloped regions of the world.42 However, absorption of that many outsiders could result in the transformation, if not the collapse, of established cultures. Beyond jeopardizing the cultural heritage of nations, declining birthrates have a ripple effect on family life. Fewer people experience being fathers and mothers, having brothers and sisters, being grandparents or aunts and uncles. As the web of intergenerational family relationships withers, generating less social, emotional, and fiscal support, the state comes under greater pressure to provide public services for what was previously handled in private life.
It is possible that low fertility rates will start to climb without any outside intervention, particularly under conditions in which immigration is tightly controlled. The well-known Easterlin hypothesis suggests that as fertility declines, individual opportunities increase, along with personal welfare. Compared to the previous cohort, those born into a low-fertility society encounter, for example, fewer people competing for jobs in the labor market, more houses for sale to fewer available buyers, and a relatively higher number of slots in the educational system. The increasing level of opportunity and personal welfare presumably lowers the costs of having children.43 However, Easterlin’s economic perspective on fertility has been challenged on various fronts. Opposing arguments suggest that social attitudes may influence childbearing behavior more than economic costs and benefits, and that societal conditions have changed in recent decades, which may have diminished the self-correcting effect of cohort size on fertility.44
At the same time that fertility rates have declined, female labor-force participation and divorce rates have increased throughout the European Union. The temporal connection between women entering the labor force and having fewer children is open to various interpretations based on the time frames selected, the countries included in the analysis, the statistical manipulations applied to the data, and what one is looking to prove. The exactitude of measurement is also worth bearing in mind—the European Union labor-force survey defines as “employed” all respondents who report at least one hour of gainful employment in the previous week.
Contrary to what one might expect, sociologist Gosta Esping-Andersen found that in 1992 European countries with high levels of female employment tended to have higher fertility rates than those with low levels of female employment. This was a reversal of the negative relationship between fertility and employment found among those countries in earlier years. Based on a positive correlation from a cross-sectional analysis of nineteen countries, he concluded that in some contexts women’s careers and children can become fairly compatible.45 Similar moderately positive results emerged from a crosssectional analysis of fertility rates and female employment in twelve European countries in 1997.46 In both instances, it is worth noting that the countries that registered the highest fertility rates were still well below the replacement level.
In reality, these positive findings say more about the perspective afforded by cross-sectional analysis than about the overall long-term trend in the relationship between fertility and female labor-force participation rates. Repeated crosssectional analyses comparing different countries at various points in time over several decades show that the negative correlation between fertility rates and female employment was reversed by the 1990s. From a longitudinal perspective, however, fertility rates declined as female employment increased in each country—but the magnitude of this negative association became weaker over time.47
Among the forces that might affect the relationship between changes in fertility and female employment, Esping — Andersen suggests that we would be likely to find a positive influence in countries where day care and family services are offered to everyone. Sweden, Finland, and Denmark are among the countries where these services are most widely available. They also have among the highest fertility rates in the European Union. Yet, in each of these countries, female labor-force participation rates were higher in 2004 than in 1994 while fertility rates were lower in 2004 than in 1994 (see Appendix, Table 1). This inverse relationship presents a completely different picture than the positive one that emerges from a cross-sectional perspective. Indeed, taking a larger group of countries (seventeen) over a longer period of time (1980 to 2002), we find a substantial inverse relationship between the average fertility and female employment rates. But there are several ways to look at these data, and different results appear depending on how the analysis is framed (see Appendix, Figures 1,2, and 3).
Following the downward trend in fertility, a corresponding decline in marriage rates between 1980 and 2001 shows a strong inverse correlation with female labor-force participation (see Appendix, Figure 4). Much of the decline in marriage has been made up for by the historic increase in cohabitation across northern Europe. From Scandinavia to France the formality of marriage has fallen out of favor with a huge proportion of the current generation. By 2005 more than half of all first-born French children had unwed parents, most of whom were in an ongoing relationship. The normative acceptance of this shift away from the traditional commitment of marriage is reflected by the long-term cohabitations of such popular political figures as Segolene Royal, the French Socialist Party’s 2007 presidential candidate, and Francois Hollande, the party’s leader, who lived together for twenty-five years; and Defense Minister Michele Alliot-Marie, who is in a long-term relationship with a member of the National Assembly.48
As female labor-force participation rates climbed, public efforts were made to reduce the friction between work and family life in many of the advanced industrialized countries of the OECD. One way to estimate the effects of these efforts is to look at how patterns of public spending on family-friendly benefits such as day care, household services, maternity and parental leave, children’s allowances, and other social provisions vary with marriage and fertility rates. Although the pattern of spending on family-friendly benefits rises and falls, overall the average rates of public expenditure on these benefits as a percentage of GDP increased slightly between 1980 and 2001 (see Appendix, Figure 5). This spending had an inverse correlation with fertility rates and showed a similar relationship to marriage rates.
Findings based on aggregate data may, of course, mask large variations between individual countries and among groups of countries. Analysis conducted on countries grouped according to the widely cited classification of social-democratic, conservative, liberal, and southern European regimes showed some variance from the pattern that emerged when the results for all countries were averaged.49 Specifically, a positive correlation appeared between fertility rates and spending on family benefits in the social-democratic Scandinavian countries. The social-democratic countries also had a higher level of public expenditure on family-friendly benefits than other groups. After 1990, however, social-democratic expenditures began to decline while the levels of spending in other regimes increased or remained constant, causing the average spending on family — friendly benefits to converge.50 A case might be made that too much attention to levels of spending on family-friendly measures ignores important differences in the substantive configuration of these policies, which influence work and family relations.51 Indeed, even among the social-democratic Scandinavian countries, sharp differences emerge in the extent to which family policies promote gender equality and afford mothers the right to choose among alternative child-care arrangements.52
Family-friendly policies, of course, involve more than the OECD categories of expenditure represented by family benefits. For example, more than 70 percent of the employed women in the Netherlands work in part-time jobs that have benefits similar to those of full-time employment, and Dutch children spend more days per year in school than most elementary school students in the European Union. A thorough assessment of the measures that affect efforts to balance work and family life would include flexible work schedules, number and length of school days, paid vacation time, and other benefits, some of which are reflected in the larger scheme of total public social expenditures.53 Analyses of overall social — expenditure data reveal patterns that parallel the findings noted above—that is, rates of total public social expenditure between 1980 and 2001 are inversely related to both fertility and marriage rates (see Appendix, Figure 6).
Still, even when total public social expenditures are considered the picture is not complete, since there are different ways to count public spending. A caveat on comparative data is in order. It has long been recognized that, in addition to the checks written directly by government, comprehensive measures of social-welfare efforts should include other sources of social expenditure that promote individual well-being. In the 1930s, special tax deductions and exemptions were identified as a form of government aid in Arthur Pigou’s classic text Economics of Welfare. It was not until the mid-1970s, however, that data on tax expenditures became available and were introduced as a regular component of the president’s budget in the United States.54 In the 1980s it became increasingly clear that the conventional categories of social expenditure were at best a crude metric for comparisons among different countries.55 In the mid-1990s, as additional data came available, researchers at the OECD developed a new ledger for social accounting, which controls for the effects of costs and benefits from various sources.56 Yet even this highly sophisticated measure does not take into account the effects of government deficit spending, which can finance current benefits through the creation of debt that must be discharged sometime in the future.
Moreover, the most sophisticated measure of net social expenditure, which compares spending among countries as a percentage of their GDP, yields a different set of results than when assessments are based on spending per capita. An analysis of social-welfare spending among ten OECD countries reveals that when the measurement shifts from the size of expenditure as a percentage of GDP to the size of per capita expenditure, the United States’ rank jumps from near the bottom (just above Canada and Australia) to second from the top (just below Sweden).57 All of these cautions about comparative measurements are to say that the findings that fertility and marriage rates have often declined as spending on family benefits and total social expenditure have increased can only be taken as suggestive. But what do they suggest?