1Non-marital childbearing has increased dramatically in the United States over the past three decades. The number of live births per thousand unmarried women grew from about 25 in the early 1970s to over 44 today. The percentage of births occurring outside of marriage has increased as well. Whereas in 1970, approximately 12% of all births were to unmarried mothers, today over a third occur outside marriage (Ventura and Bachrach, 2000).

2In the United States, only half of unmarried mothers are living in cohabiting unions at the time of birth, and, even taking into account the presence of children, cohabiting unions are more likely to dissolve than marriages. The association of unwed parenthood with father absence is important given that single mothers and their children have such high rates of poverty. In the year 2000, about 9% of all families and 25% of female-headed families (with no partner present) lived in poverty. Although poverty rates declined substantially from 1993 to 2000, a substantial gap in poverty between married and single mother households remained, and in 2000, 45% of children in female-headed households were living in poverty (US Census Bureau, 2001).

3In response to the trends in non-marital childbearing and its disturbing association with child poverty, policymakers have begun to put forward the promotion of marriage as a strategy to reduce poverty and are discussing ways in which they can promote marriage within the existing income support system and encourage more marriage among low-income parents. Citing well-known research findings that indicate the wide-ranging benefits of marriage for adults and children, these advocates want to divert cash assistance funds to the development and implementation of programmes that promote marriage (Fagan, 2001). Unfortunately, there is no strong evidence that such measures would have an appreciable impact on marriage. More importantly, it is not clear, from existing evidence and methodologies, that marriage would dramatically reduce poverty among single mothers.

4In this article, we use information from a new survey of unwed parents to assess whether the “marriage strategy” is likely to be as effective as these kinds of assessments imply. We present some descriptive information about the economic circumstances of new parents, both married and unmarried, and show the results of two simulations that examine how well each group of parents would fare in a variety of family circumstances. Our first results are based on their reported last year’s earnings. The second simulation uses hourly wages at the current or last job in order to assess how well parents would fare if they could (or did) work steady hours. Large differences between the married and unmarried groups provide some evidence that the benefits of marriage are overstated when married couples are used as a reference group.

I – US social welfare state. Background and recent reforms

5Social policy experts have described the US social welfare state as residualist (Titmuss, 1974) or liberal (Esping-Andersen, 1990), and, taking into account its treatment of women and families, male breadwinning (Sommestad, 1997). Because the liberal ideology is characterized by notions of individualism and self-determination, liberal welfare policy models reflect an underlying apprehension that generous state benefits will create perverse incentives that result in the able-bodied poor relying on state support when they could otherwise provide for themselves. As a consequence, means-testing is common and benefit levels tend to be low. Assistance is provided only to those deemed genuinely in need—those who, through no fault of their own, are unable to provide for themselves. But the preoccupation with incentives can often result in punitive welfare policies that stigmatize recipients.

6In a male breadwinning state, policies operate to exclude women from the labour market. Social security and tax systems presume and encourage a working husband and a dependent wife. Women are expected to remain in the home and specialize in unpaid domestic and caring work while men (and not the state) support them by specializing in wage labour (Lewis, 1992). Because male breadwinning states presume highly specialized families with two adults, the treatment of lone mothers in these welfare systems is particularly problematic. States must choose between wanting to support women in their role as mothers and not wanting to undermine men’s incentives to take private responsibility for their own dependents. Garfinkel and McLanahan (1986) show that policy-makers in the US have often vacillated over time, sometimes treating lone mothers primarily as workers and other times primarily as mothers.

7While the United States does conform fairly well to the liberal and male breadwinning typologies, it is important to keep in mind that these models are ideals and nowhere exist in their “pure” form. In reality, the US system is more pluralistic than the ideal liberal state. It offers a dual system of higher quality benefits in the form of social insurance and horizontally redistributive tax concessions to working families [1] and a lower quality catalogue of more traditionally “liberal” assistance programmes to the sporadically working or non-working poor. Consistent with a strong male breadwinning model, the former benefits are more likely to be provided to men and to married couples (via a male wage earner) while the latter go predominantly to unmarried mothers (Lewis, 1992). The joint taxation of married couples is normative, so that dual earner marriages are penalized relative to male breadwinning families [2]. To ensure that able-bodied men would provide for their families, prior to 1962, cash assistance benefits were denied to families in which the father was present and not disabled. After that, means-testing and stricter eligibility rules excluded many needy two-parent families, and most welfare recipients continued to be single mothers [3].

8Beginning in the 1960s and 1970s, dramatic changes in family structure and the economy began to occur. As a result of more frequent non-marital childbearing and divorce, the prevalence of lone mother families began to increase. Female labour force participation grew, and many married mothers entered paid (albeit often part-time) work. While in the first half of the twentieth century it was widely believed that the government should provide enough aid that lone mothers could stay at home and raise their children, increases in married mothers’ employment began to defy the logic of this position. Gradually, the consensus shifted from a conception of women as mothers to one of women as workers—at least those women who chose not to have a husband to support them. As lone parenthood increasingly resulted from non-marriage and family dissolution and not death, policymakers became less concerned with the incentives for men not to work and more concerned with the incentives for men to become absent fathers [4].

9In response to these demographic and attitudinal changes, the 1996 Personal Responsibility and Work Opportunity Act (PROWORA) dramatically altered the provision of income support in the United States. Replacing the existing federal programme with Temporary Aid to Needy Families (TANF), the legislation placed duration limits on the receipt of cash benefits, introduced much stricter work requirements, limited the duration of cash benefits to no more than five years in a person’s life, and penalized states for not reducing their welfare rolls. States were also encouraged to reduce the number of non-marital births and to “encourage the formation and maintenance of two-parent families”. With the continuation of the PROWORA legislation coming under review in the summer of 2002, many conservative politicians are suggesting that although states have increased self-sufficiency by moving mothers on income support into paid employment, they have ignored an integral goal of the reform—encouraging the formation of two-parent families. In this vein, the Bush administration is offering states up to $300 million in support of programmes that promote “healthy marriages” and child well-being. Proposed policies include “marriage bonuses” for couples that marry before their child is born, marriage preparation courses and mentoring programmes, and benefit cuts to cohabiting couples (Horn, 2001). These policies are motivated by the belief that the retreat from marriage—resulting, in part, from the marriage penalties in assistance programmes—is a root cause of poverty. However, some argue that funds might be better spent making poor parents, especially fathers, better able to support their families.

II – Policy arguments and existing literature

10To demonstrate the benefits of marriage, proponents of new, pro-marriage initiatives frequently compare the characteristics of married parents to those of single parents. Noting that “married adults…are happier, healthier and wealthier than their unmarried counterparts,” advocates assume that marriage is the cause of happiness, wealth and positive relationships (Horn, 2001). To explain why more people are not choosing to reap these benefits, they point to the disintegration of pro-marriage values, particularly in low-income communities (Dennis and Erdos, 1992).

11Dramatic income or poverty differentials between married couples and single mother households are often used as evidence that the absence of a male wage earner is a major cause of poverty among single mothers (Besharov and Sullivan, 1996; Fagan, 2001; Popenoe and Whitehead, 1999; Rector, 2001) [5]. While it makes sense that two parents and two potential wage earners are better than one, it does not follow that marriage will make unmarried parents as well off as currently married parents.

12Academic research provides reasons why arguments based on these kinds of comparisons may be flawed. Both qualitative and quantitative studies show that most unmarried parents value marriage (Carlson, McLanahan and England, 2001; Harknett and McLanahan, 2001). According to these studies, single mothers want to marry but they are reluctant to commit themselves to a man who cannot support a family (Edin, 2000). While marriage may improve economic status, research shows that deteriorating economic opportunities for males are associated with less and later marriage and that women from low socio-economic backgrounds are more likely to have children out-of-wedlock (Wilson, 1987; Oppenheimer, 2000; South and Lloyd, 1992; Rosenzweig, 1999). Consequently, unmarried individuals are likely to be very different from married individuals on a range of economic attributes. Moreover, patterns of assortative mating suggest that single mothers from disadvantaged backgrounds are likely to be bearing children with similarly disadvantaged men, so it is not clear that marriage alone will be the best anti-poverty strategy for unmarried parents (Glei, Garfinkel and McLanahan, 2002). These findings would appear to suggest that helping unmarried parents to become more economically self-sufficient should be an important policy goal.

13Identifying just how much better off unmarried parents would be if they were to marry has, thus far, been hampered by two problems. First, few data sources have information on the prospective marital partners of unmarried women. These data limitations have often meant that researchers were forced either to look at married couples for information or form assumptions about whom an unmarried woman would marry. One particularly useful method comes from the child support literature (Garfinkel and Oellerich, 1989; Miller, Garfinkel and McLanahan, 1997; Glei et al., 2002). Researchers estimate a model of husband’s earnings based on married women’s characteristics. Next, the parameters from the married couple model are combined with the characteristics of unmarried women to estimate the wages of unmarried fathers (see also Schultz, 1994). This method predicts average incomes rather well, but does not take into account the distribution of partner characteristics among the unmarried. Although most unmarried parents have partners of a similar age, compared to married couples, there is a higher percentage of unmarried couples in which the mother is older than her partner. Similarly, white, unmarried mothers are more likely to have non-white partners than their married counterparts (Glei et al., 2002). Predicted values based on married samples will tend to overstate estimated earnings for these atypical couples.

14The second problem is more challenging, and most probably, insoluble. Because marriage is a life-altering choice, changes in preferences and behaviour may make unmarried people behave differently after they marry. Bargaining positions both prior and subsequent to marriage may alter behaviour, so even an (impossible) experiment that randomly assigned marriage to one group of couples and non-marriage to another would yield less than perfect information on the benefits of marriage because it would not allow for the plausible negotiation between partners that real marriage (and the threat of exit) entails. For this reason, the actual benefits of marriage—economic or otherwise—will most probably never be estimated with certainty, and our estimates below should be interpreted with that caveat.

III – Data

15The Fragile Families and Child Well-being Study, a joint effort by researchers at Princeton and Columbia Universities, was designed to provide new information on the capabilities and relationships of new, unwed parents (particularly unwed fathers), as well as the effects of policies on family formation and child well-being. The study, funded by a host of government organizations and private foundations in the United States, follows a birth cohort of new (mostly) unwed parents over a four-year period. The mothers’ first interview took place in the hospital, within 48 hours of the delivery. Fathers were also interviewed, often in the hospital, a short time later [6]. The baseline sample contains information on 3,712 deliveries to unmarried parents in 20 large US cities [7], and is representative of all non-marital deliveries in US cities with populations exceeding 200,000. In each city, a comparison group of married parents was interviewed, providing a sample of 1,188 deliveries to married parents. These data contain the information needed to compare the characteristics and earnings capacities of married and unmarried parents. When we restrict our sample to those couples with all of the necessary socio-economic information, we are left with 3,334 unmarried couples and 1,141 married couples.

16Before encouraging marriage among unwed parents, we must ask whether or not the anticipated benefits of doing so are realistic. In this analysis, we focus on the economic benefits only, and ask “Would marriage make unmarried parents as economically well-off as currently married parents?”—the group whose economic outcomes are frequently used to justify marriage promotion programmes. First, we compare various socio-demographic characteristics of married and unmarried parents that are likely to be related to earnings capacities. Large differences between married and unmarried individuals would be suggestive of lower income security among unmarried couples even if they were to marry.

17Table 1 demonstrates important differences between married and unmarried parents. Compared to married parents, unmarried mothers and fathers are more than twice as likely to have dropped out of high school and less than half as likely to have attended university. To the extent that earnings are linked to education, these differences imply that unmarried parents have lower earnings capacities than married parents. In addition, unmarried parents are substantially younger.

Table 1

Characteristics of married and unmarried new parents

Mothers Fathers Married Unmarried Married Unmarried Median age (years) 29 22 31 25 Education (%): Below high school 17 40 18 37 High school 20 34 24 41 Beyond high school 64 26 58 22 Work-limiting illness (%) 3 7 Drug/alcohol problems (%) 1 3 2 6 Median hourly wages ($) 12.36 7.50 14.91 9.50 “Steady job” (%) 91 74 N 1,141 3,334 1,141 3,334 Source: Fragile Families and Child Well-being Study, 1998-2000.

Characteristics of married and unmarried new parents

18The median age of all unmarried mothers is only 22—seven years less than the median age of married mothers. Similarly, unmarried fathers have a median age of 25 compared to 31 for married fathers.

19In addition to these economic limitations, unmarried fathers are twice as likely as married fathers to suffer from a physical or psychological condition that interferes with their ability to work. Similarly, unmarried parents are several times more likely than married parents to report having ever had a drug or alcohol problem. While such problems are rare, the disparity between married and unmarried parents merits careful consideration before we aggressively and universally encourage marriage.

20Because age, education and other limiting conditions are only indicators of earnings, we also examine differences in hourly wages and labour supply. Not surprisingly, the hourly wages of the married and unmarried parents show a substantial gap [8]. Table 1 shows that the median hourly wages for married fathers are more than $5 per hour higher than those of unmarried fathers, while the median married mother earns $4.86 per hour more than the median unmarried mother.

21While wage rates are important in determining economic capacity, consistent labour market participation is equally important, especially for low-income families who have fewer savings and assets. As an indicator of employment stability, the Fragile Families survey asks whether or not the baby’s father was working at a “steady job” during the week before the child was born. Once again, large differences between married and unmarried couples emerge. Fewer than 10% of the married fathers are not working, whereas over a quarter of unmarried fathers are not steadily employed. The latter are likely to have unreliable incomes making it difficult to support a family. Clearly, unmarried mothers are more economically disadvantaged than married mothers. In addition, they are partnering with a pool of men who are more economically disadvantaged than married fathers.

IV – Earnings and earnings potential

22In the last section we presented several measures that indicate important differences in the earnings capacities of married and unmarried parents. In this section, we examine how well married and unmarried parents would fare assuming a variety of family structure and work scenarios. We first ask how the parents would fare, given their reported earnings in the past 12 months, in three different family scenarios. The results for each scenario are presented in Table 2, for all couples, by the mothers’ actual marital status. First, we assume that all mothers live alone and rely on their own reported earnings. In the second example, we assume that all parents are living as couples but that only the father’s earnings support the family. These families represent a traditional “breadwinner model” in which the mother specializes in domestic work and the father specializes in market work. In the third example, we assume that all parents live together and combine their earnings. Unfortunately, the information on earnings was only recorded in bands, the width of which varied in the fathers’ and the mothers’ questionnaires. When we report the distributions of earnings in Table 2 using just one person’s earnings, we report the quartiles as income bands, but when we attempt to combine earnings, as in the third example, we assume that both individuals fall within the midpoint of their respective earnings bands [9].

Table 2

Yearly earnings of married and unmarried parents and poverty ratios assuming three family types, by mother’s marital status at the time of birth

Earnings in the past 12 months Poverty ratios Earnings: Women Single parent family: Based on mother’s earnings only* quartile 1 quartile 2 quartile 3 0-100 100-150 150-200 over 200 Married 0 [$9,000-9,999] [$25,000 +] Married 58% 7% 7% 28% Unmarried 0 [$1,000-2,999] [$17,500-19,999] Unmarried 86% 6% 4% 5% Earnings: Men Male breadwinner family: Based on father’s earnings only* quartile 1 quartile 2 quartile 3 0-100 100-150 150-200 over 200 Married [$15,000-19,999] [$25,000-34,999] [$50,000-74,999] Married 21% 16% 11% 53% Unmarried [$5,000-9,999] [$10,000-14,999] [$20,000-24,999] Unmarried 58% 18% 10% 15% Combined Earnings* Dual earning family: Based on combined earnings* quartile 1 quartile 2 quartile 3 0-100 100-150 150-200 over 200 Married $26,250 $50,000 $88,638 Married 15% 9% 9% 67% Unmarried $8,250 $17,500 $30,500 Unmarried 46% 18% 12% 24% * Earnings data are categorical. For these calculations, we assume earnings at the mid-point of the range. Source: Fragile Families and Child Well-being Study, 1998-2000.

Yearly earnings of married and unmarried parents and poverty ratios assuming three family types, by mother’s marital status at the time of birth

23The left panel of Table 2 presents quartiles of reported earnings, by the mothers’ actual marital status, for each of the three family structure profiles. Because larger families have greater income needs, earnings alone do not reflect how well a family would fare in each of the three profiles. Therefore, the right panel of Table 2 presents the earnings data expressed as a percentage of the 2000 federal poverty line (FPL) [10], a collection of income thresholds that vary by family size and composition (the number of adults and children) and are used by the United States government to identify poor individuals. In order to construct family size in our data, we assume that a single mother lives with her own biological children and that a married woman lives with her partner and her own biological children [11]. When the ratio of earned, pre-tax income to the poverty threshold—the poverty ratio—is less than one, a family, and each person in it, is considered poor.

24The data in the first rows of Table 2 indicate that some of the poverty single mothers face can be attributed to their low earnings and not simply to their family structure. If the married women in our sample supported themselves and their children with only their own earnings, their median income would exceed the median of the unmarried mothers in the same family situation by $6,000-$8,999 per year. In addition, a much higher percentage of married women could maintain their families at more than twice the FPL. Only 5% of the unmarried mothers could do so with their current earnings. Although many married women would live in poverty if they had to use their current earnings to support their families on their own—indeed more than a quarter of married mothers have no earnings at all—these numbers should be interpreted with some caution because they may well underestimate the extent to which women could support their families. Women who are currently in partnerships (or who still live with their parents) may have reduced labour supply if their partner (or another family member) is providing some economic support. Faced with having to support their children on their own, many women may increase their working hours, and consequently, their earnings.

25The next set of estimates shows that unmarried male breadwinning families would have incomes that are about $10,000-$25,000 per year less than the married parents in the same situation. This represents a large part of the difference in median earnings between married parents in the male breadwinning scenario and unmarried mothers in the single mother scenario (about $22,000-$34,000 per year). In terms of poverty, about 58% of the unmarried women in our sample would be living below the FPL if they were to marry and rely on this traditional arrangement. Another 18% would be earning in the range of 100-150% of the FPL. Finally, the last rows of the panel demonstrate that under the “dual earnings” scenario, the difference in median earnings between the married and unmarried parents would be even larger—at more than $30,000 per year—than we saw under the male breadwinning example.

26In both married family examples in Table 2, we see that a substantial portion of the economic gain that would often be attributed to marriage is, in fact, due to differences in the earnings of the married and the unmarried populations. In the last scenario, we do see a large drop in poverty from 86% to 46% for unmarried mothers and from 58% to 15% for married mothers. While the reduction in poverty is, without a doubt, substantial, at current earnings, marriage would only succeed in lifting some mothers out of poverty—a higher percentage of whom are already living with their baby’s father. Among unmarried parents who were not co-residing at the time of birth (results not shown), a higher percentage (58% vs. 46% for the whole sample) would be living in poverty if they were to live together and rely on their own current earnings.

27While the data in Table 2 provide some information on the earnings capacities of married and unmarried parents, it is important to take into account that a comparison of current earnings combines differences in both wages and work patterns. The birth of a new child and marriage itself could arguably change work motivations and increase earnings (although the time required to care for a new infant could result in some countervailing earnings effects as well). Our next simulation isolates the differences between married and unmarried couples that are due only to differences in wages. We use reported hourly wages (presented in Table 1) for the mothers and fathers to estimate earnings capacities assuming both the same family structure, and a fixed set of work hours for each family structure. We also experimented with simulations that used the married parents’ wage schedule to predict the earnings of unmarried parents. This, in effect, assumes that unmarried parents can translate their age, education, and other characteristics into wages in exactly the same way their married counterparts can, and allows, to some extent, for a “marriage premium” in wages [12]. The results from those simulations are strikingly similar to those we report below, and are not presented here, but are available, on request, from the authors.

28Table 3 presents the yearly, pre-tax earnings for all of the mothers in our sample assuming three different family and work profiles. Similar to Table 2, the results from these simulations are presented, for all couples, by the mothers’ actual marital status. In other words, we hold family structure and work patterns constant and compare the earnings potential of married and unmarried parents. By potential we mean what they could earn if they were to work steady hours. First, we assume that all mothers live alone and work full-time (2,000 hours per year). In the second example, we assume that all mothers are married and not working while all fathers work full-time. In the third example, we assume that all mothers are married and work half-time (1,000 hours) and all fathers work full-time.

Table 3

Simulations of economic well-being assuming three family structure and labour supply scenarios using reported wages, by mother’s marital status at the time of birth

Predicted yearly earnings Poverty ratios Not married: Woman works full-time Not married: Woman works full-time quartile 1 quartile 2 quartile 3 0-100 100-150 150-200 over 200 Married $16,529 $24,711 $37,144 Married 17% 23% 19% 41% Unmarried $12,672 $15,000 $18,220 Unmarried 37% 45% 12% 6% Married: Husband full-time, wife doesn’t work Married: Husband full-time, wife doesn’t work quartile 1 quartile 2 quartile 3 0-100 100-150 150-200 over 200 Married $20,087 $29,823, $43,990 Married 16% 23% 18% 43% Unmarried $15,026 $19,001 $23,440 Unmarried 32% 40% 14% 15% Married: Husband full-time, wife half-time* Married: Husband full-time, wife half-time quartile 1 quartile 2 quartile 3 0-100 100-150 150-200 over 200 Married $30,576 $42,488 $62,067 Married 3% 12% 16% 68% Unmarried $22,389 $26,655 $32,615 Unmarried 6% 31% 31% 32% *The values for married parents in quartiles 2 and 3 are lower here than in Table 2 because many of the highest earning parents actually both worked full time. Source: Fragile Families and Child Well-being Study, 1998-2000.

Simulations of economic well-being assuming three family structure and labour supply scenarios using reported wages, by mother’s marital status at the time of birth

29Because the assumptions we make about working hours are neither realistic nor verifiable, the results that follow should not be interpreted as estimates of the economic benefits of marriage. First, we allow for generous changes in work patterns for many parents, but particularly for unmarried men, and our assumptions most probably err on the side of optimism. For this reason, our simulation is meant only to explore to what extent differences in wages between married and unmarried populations are responsible for differences in their potential income and poverty rates.

30The left panel of Table 3 presents quartiles of predicted family earnings, by the mothers’ actual marital status, for each of the three hypothetical family structure and work profiles. In each of the three scenarios, married women’s family earning potential is higher than unmarried women’s earning potential. Indeed, the third quartile of earnings for unmarried mothers falls, in all cases, below the median earnings for married women with the same family structure and work profile. The right panel of Table 3 presents the earnings data expressed as a percentage of the 2000 FPL.

31The estimates in the first rows of Table 3 indicate that if married women were forced to rely on their own wages and able to work 2,000 hours per year, their median incomes would exceed those of the unmarried mothers in the same situation by more than $9,000 per year. In addition, over 40% of the married women could sustain their families at more than twice the FPL. Only 6% of the unmarried mothers have a similar capacity. Although many married women would live in poverty if they worked full-time and had to support their families on their own (and at their current wages), their capabilities far exceed those of unmarried mothers.

32The next set of estimates shows that even if fathers worked full-time all year round, the unmarried male breadwinner families would have median incomes that are about $10,800 per year less than the married parents in the same scenario. This represents much of the difference in median earnings between the married parents in the male breadwinner scenario and the unmarried mothers in the single mother scenario. Nearly three-quarters of that income difference is not due to marital status but to differences in male wages. In terms of poverty, about 32% of the unmarried women in our sample would be living below the FPL if they were to marry and rely on this traditional arrangement. This is much lower than the 58% reported for the same family structure in Table 2 (and the two earner family example, as well), suggesting that full employment should not be ignored as an essential component of poverty reduction. Another 40% of unmarried women would be living somewhere between 100-150% of the FPL if they lived in this situation.

33Finally, the last rows of the panel demonstrate that under the “two working parents” scenario, the difference in median income between the married and unmarried parents would be around $16,000 per year. This represents more than half of the gap in median earnings between the married couples in the two working parents scenario and the unmarried women in the single mother scenario. In both married couple examples, a substantial portion of the difference in economic outcomes between married parents and unmarried mothers is due to important differences in their hourly wages.

34Under the last scenario, only 6% of the unmarried women and 3% of the married women would have earnings below the FPL. Comparing these numbers with the last rows of Table 2, it is apparent that this impressive result is due largely to our assumptions about working hours and the way in which the FPL takes into account economies of scale. For a single mother living with one child, the FPL in 2000 was $11,869. If another adult were to move into that household, the FPL would increase to $13,874. The additional income needed for one extra adult requires full-time wages of just over $1 per hour. Our assumption of full-time employment allows quite a few families close to the poverty line to be lifted out of poverty simply by adding a full-time worker. Nonetheless, there are many more single than married mothers who would live just above the poverty line (in the range of 100-150% of the FPL) even if they were married to the fathers of their babies and both parents were working. Unmarried parents would still face a good deal of economic uncertainty if they were to marry and would be far less well off than some marriage proponents have suggested.

35Comparing Tables 2 and 3, it is clear that poverty would be reduced nearly universally only if unmarried parents could obtain full-time employment in conjunction with getting married. With full-time work, there would be less poverty among lone mothers (left panel of Table 3) than there would be if both parents lived together and relied on their current, joint earnings (left panel of Table 2).

Discussion and conclusion

36Consistent with previous research, the Fragile Families and Child Well-being Study demonstrates that important differences exist between married and unmarried couples that cannot be altered with a marriage licence. Unmarried parents are vastly different from married parents when it comes to age, education, health status and behaviour, employment, and wage rates. These differences translate into important differences in actual earnings and earnings capacities, which, in turn, translate into differences in poverty. Proponents of marriage are substantially overstating its benefits when they compare the earnings or poverty rates of single mother families to those of married, two-parent families. Our results indicate that two incomes could lift many mothers out of poverty, but, at their current earnings, 46% of unmarried parents would continue to earn below the FPL (not including taxes and transfers) even if they were to marry. This is especially important because the economic benefits of marriage must be weighed against the loss of many means-tested benefits that may result from marriage. With welfare programmes that make it more difficult for two-parent families to obtain support when the market fails, marriage for unmarried couples might mean more rather than less vulnerability.

37Assuming the same family structure and labour supply, our estimates suggest that, even under the very best circumstances, much of the economic differential between married couples and unmarried parents can be attributed to factors other than marital status. Low employment and large wage gaps translate into large differences in earnings capacities. Nonetheless, we see dramatic reductions in poverty when we assume parents both live together and work regular hours, year round. Our results suggest that preparing parents for jobs and increasing their work hours—whether they marry or not—is an essential ingredient to poverty reduction. If only they could manage to work full-time at their current hourly wages, poverty among unmarried mothers and their children could fall by more than half. This is a larger drop in poverty than if parents lived together and shared their current last year’s earnings. While some might argue that marriage increases the work motivations of men and that Table 3 provides a more realistic picture of how parents would fare if they married, that argument assumes men can work as many hours as they choose. Given their low levels of human capital and the high percentage of unmarried fathers not working steady jobs, even during the best of economic times (1998-2000), it is not clear that motivations would quickly or easily turn into reality [13]. While our findings are far from conclusive, they do suggest that it might be more cost-effective for policymakers to focus on the structural causes of economic disadvantage—low wages and unemployment—than to divert resources to the promotion of marriage.