The number of out-of-wedlock births in the United States reached a record high (1.71 million) in 2007, when nearly 40% of all births were to unmarried women.¹ Nonmarital childbearing is linked to a variety of negative outcomes for children, mothers and society,² and has therefore received a substantial amount of national attention. Most of this attention has gone to the exceptionally high nonmarital fertility ratio (i.e., nonmarital births as a percentage of all births, NMFR) of black women (72%).¹ However, nonmarital childbearing among Latinas is also high: Slightly more than half of all births to Latinas occur outside of marriage.

Teenage childbearing also is linked to poor outcomes for individuals, families and society,³ and in this regard too, Latinas are at particular risk. Although the rate of teenage childbearing among Latinas declined by 22% between 1991 and 2005, it remains at twice the national level (83 vs. 42 births per 1,000 women aged 15–19).¹

Among Latinos, especially recent immigrants and teenagers, high nonmarital fertility ratios may reinforce economic inequality and slow the pace of social and economic incorporation into American society. Indeed, as among other racial minorities, more than one-half (55%) of Latino children aged six or younger who live with a single mother fall below the government poverty line.⁴ Reducing the Latina NMFR arguably is a necessary (albeit insufficient) condition for improving the socioeconomic trajectories of young Latinos, and would help break the intergenerational cycle of out-of-wedlock childbearing, poverty and welfare dependence.⁵

Unfortunately, our understanding of the demographic foundation of recent trends and differentials in NMFRs among Latinas is limited. Previous population-based research on trends in out-of-wedlock childbearing has focused largely on U.S. women of reproductive age⁶ or on the high rates among blacks.^7,8 In this article, we decompose 1994–2005 trends in NMFRs among Latina women and teenagers, using rate standardization techniques devised by Kitagawa⁹ and adapted by Das Gupta.10 We then compare these results with ratios among non-Latina blacks and whites. Specifically, we partition 1994–2005 changes in the share of all births that occur out of wedlock into three demographic components: changes in nonmarital fertility, changes in marital fertility and changes in the proportion of women who were married.

Each demographic component represents a target for policy intervention.^* Obviously, NMFRs can change over time if fertility rates among unmarried women change. Presumably, successful interventions to reduce nonmarital fertility rates should be reflected in lower NMFRs. However, declines in nonmarital rates have not yielded declining NMFRs over the past decade. The paradox from a policy standpoint is that the other two components—changing marital fertility and changing marital status composition—also matter. Indeed, declining marital fertility rates and the "retreat from marriage" (which increases women's risk of nonmarital births) place upward or countervailing demographic pressure on the share of births that occur to unmarried women. The apparent demographic paradox is that NMFRs can increase even as nonmarital fertility rates decline.

This fact is not often fully appreciated. Among Latinas, high NMFRs coexist with familial and cultural norms that support traditional marriage, early childbearing, large families, and strong family ties and intergenerational support.^11–13 Our objective is to link recent trends in Latina nonmarital fertility ratios to shifts in marriage behavior and childbearing, both within and outside of marriage.

MARRIAGE AND FERTILITY

We update the 1960–1992 time series of Smith and colleagues,⁸ and focus on America's most rapidly growing minority population—Latinos. More than ever, an adequate understanding of fertility today—and of America's demographic future—requires some appreciation of recent marriage and fertility patterns among Latina women, especially those of reproductive age. In the mid-1960s, Latinos represented just 4% of the total U.S. population; by 2006, this proportion had grown to nearly 15%.¹⁴ In raw numbers, the Latino population increased from 8.5 million in 1966 to nearly 45 million in 2006, and this increase accounts for the largest proportion (36%) of U.S. population growth over this period. Significantly, the Latino population is overrepresented among America's adolescents and young adults—age-groups with high rates of fertility.¹⁵

More than a decade ago, Smith and colleagues demonstrated that an increase in the proportion of black women who were unmarried accounted for the overwhelming share of the increase in the black nonmarital fertility ratio.⁸ Between 1983 and 1992, for example, the black NMFR rose by 10 percentage points, from 59% to 69%. Roughly 85% of this increase could be attributed to the declining proportion of women who were married. Rising nonmarital fertility played a significant but more modest role, while an increase in the marital fertility rate and shifts in the age composition placed slight downward pressure on the NMFR. By contrast, among whites, a rise in the nonmarital fertility rate accounted for the largest share of growth in the NMFR (eight percentage points of a nine-point increase) between 1983 and 1992. The lesson is clear: The demographic mechanisms responsible for recent trends in NMFRs for different racial or ethnic groups may be widely divergent. In the case of black-white comparisons, the analysis by Smith and colleagues demonstrated in a compelling way that the retreat from marriage played a large and unprecedented role in the shaping of racial differences in nonmarital fertility ratios over the 1983–1992 period.

Do similar conclusions about the demographic sources of nonmarital fertility trends apply to the Latino population? To be sure, the time period studied by Smith and colleagues differs in significant ways from the period examined here, during which Latinos surpassed blacks as the nation's largest minority population. A rise in NMFRs that began in the mid-20th century stalled in the late 1990s and early 2000s, when roughly one-third of all births in the United States were to unwed mothers.¹ Since then, the NMFR has again begun to increase, reaching nearly 40% in 2007 (after averaging 33% between 1994 and 2002). The rate of nonmarital births also reached an all-time high in 2007—53 births per 1,000 unmarried women aged 15–44, a rate 21% higher than the 2002 level.

In general, Latinas have followed national trends in marital and nonmarital fertility, although patterns have differed markedly by nativity, national origin and residence (e.g., immigrant gateways, rural agricultural areas or inner-city ethnic enclaves).^16,17 Fertility rates were substantially higher among Latina than among non-Latina women (102 births per 1,000 Latina women of reproductive age, compared with 71 for blacks and 60 for whites in 2006).¹ Birthrates for unmarried Latinas were similarly high: 100 births per 1,000 unmarried women in 2005, compared with 68 for blacks and 30 for whites. Notably, the rate of teenage childbearing among Latinas surpassed that among blacks for the first time in the mid-1990s; in 2006, teenage Latinas had 83 births per 1,000 women, whereas blacks had 64. However, in 2002, the share of births to teenage Latinas that occurred within marriage (26%) was higher than that among black (4%) or all teenagers (20%).¹⁸

For Latinas, the implications of recent trends in marriage are ambiguous. Latinos share many of the socioeconomic disadvantages faced by blacks; yet, paradoxically, Latino marriage rates have remained high (except among Puerto Ricans) and are generally more similar to those of whites than to those of blacks.^12,19 However, the demographic and economic characteristics of Latina women have shifted over the past decade, perhaps in ways that reduce marriage rates (e.g., socioeconomic disparities among new immigrants may have grown). Meanwhile, the effects of acculturation on sexual and reproductive behaviors may now be promoting outcomes among unmarried new immigrants that mirror patterns among native-born populations.²⁰

METHODS

Data

We used annual birth data from the Natality Detail File for 1994–2005 and population estimates from the Current Population Survey (CPS) for 1993–2006 to generate the data for our analysis. The count of births to women by age-group (15–17, 18–19, 20–24, 25–29, 30–34 or 40–44), race or ethnicity (black, white or Latina),^† marital status (married or unmarried) and nativity (native-born or foreign-born) used to estimate annual fertility rates were derived from the natality statistics. These annual data provided the numerators for our marital and nonmarital fertility rates; corresponding population estimates taken from the CPS provided the denominators.

CPS data by nativity are available only after 1993, and machine-readable, public-use natality statistics—which provide the necessary degree of flexibility for estimating denominators—are available through 2005. Because of year-to-year variability in our population estimates for specific groups, our CPS population estimates were based on three-year moving averages (e.g., the 1995 estimate was based on the average of data for 1994, 1995 and 1996).

Demographic Decomposition

We used the statistical package R and rewrote the original FORTRAN code developed by Das Gupta for our decomposition analysis.^10,21 To decompose NMFRs for a population in any given year, we followed exactly the procedures described by Smith and colleagues.⁸ That is, we estimated the components of 1994–2005 changes in the NMFRs for Latinas, blacks and whites. We also disaggregated the results by nativity (foreign-born or native-born) for the Latina population. Latinas aged 15–19 were examined separately. This method of decomposition is mainly an accounting exercise; we are not aware of any valid procedures for testing whether the differences among the components are statistically significant.

RESULTS

Nonmarital Fertility Ratios

NMFRs for black, white and Latina women changed little between 1994 and 2002, but ratios for Latina and white women ticked upward after 2002 (Figure 1). Similarly the share of all U.S. births that occurred outside of marriage showed an increase over the 2002–2005 period (from 34% to 37%).²² Differentials in the NMFR by race and ethnicity were consistent over the study period: NMFRs for black women were the highest (averaging 69% over 1994–2005; not shown), followed by those for Latinas (43%) and whites (23%).

The modest changes in NMFRs between 1994 and 2005, however, do not eliminate the possibility (even likelihood) that the mix of underlying demographic components has shifted over time or differs by racial or ethnic group. Indeed, the various demographic components may have offsetting effects. For example, the impact of declines in nonmarital fertility rates may have been counterbalanced by similar declines in marital fertility rates and in the proportion of women who were married, with the latter adding to the share of women who are "at risk" of nonmarital childbearing.

Decomposition of Nonmarital Fertility Ratios

Between 1994 and 2005, the observed Latina NMFR increased by five percentage points, from 43% to 48% (Table 1). For each component, the difference between the standardized NMFRs in 1994 and in 2005 yields the independent effect that the component would have had on the ratio, all else being equal. A decrease in the standardized NMFR over the time period suggests that the given component exerted downward pressure on the nonmarital fertility ratio, and an increase suggests that it exerted upward pressure. For example, the NMFR standardized for all factors except the proportion of women who were married increased from 41% to 45%, indicating that as a result of shifts in marital status composition alone, the proportion of births that were nonmarital would have increased by four percentage points (4.6 points without rounding). The sum of the changes attributed to all of the components is equal to the total change in the observed NMFR over the study period.

Nearly the entire five-percentage-point rise in the Latina NMFR between 1994 and 2005 can be attributed to a decline in the proportion married. That is, the total increase in the observed NMFR was roughly the same as the increase due to changes in the married share alone. Each of the other components—population aging and shifts in fertility rates—contributed less than one percentage point to the overall change.

Decompositions for white and black women revealed somewhat different pictures. As in the Latina analysis, decreases in the proportions married placed upward pressure on NMFRs (four and five percentage points for whites and blacks, respectively). Shifts in the age structure also increased NMFRs to a small extent (1–2 percentage points). However, for white women, an increase in the nonmarital fertility rate put upward pressure on the NMFR over the 1994–2005 period (four percentage points), which was somewhat offset by the effect of increasing marital fertility (a decline of three percentage points). The net effect for whites was an increase in the observed NMFR of seven percentage points. Black women's pattern was exactly opposite that of whites: Decreasing marital fertility inflated the NMFR (four percentage points), and a large decline in nonmarital fertility more than offset the impact of declining marriage and marital fertility (11 percentage points). Overall, the NMFR for blacks was virtually unchanged over the study period.

Differences Between Latina Subgroups

The observed NMFRs were roughly seven percentage points higher among native-born than among foreign-born Latinas over the 1994–2005 period (Figure 2). As in the analysis for all women, the ratio among native-born and foreign-born Latinas changed very little from 1994 to 2002, before increasing between 2002 and 2005. The ratio increased by nearly six percentage points among foreign-born Latinas and by five points among native-born Latinas over the study period.

Despite the similar changes in the two subgroups, the demographic components underlying the changes differed (Table 2). Decreases in marital fertility over the 1994–2005 period placed upward pressure on NMFRs among women aged 15–44 (six percentage points for foreign-born women and three points for the native-born). Among foreign-born Latinas, an increase in nonmarital fertility resulted in a three-point increase in the NMFR over this period. By contrast, among native-born Latinas, a recent decline in nonmarital fertility—largely masked in the results presented in Table 1—depressed the NMFR by 14 percentage points. The benefit of declining nonmarital fertility among native-born Latinas, however, was completely offset by the effects of declining marriage rates, which boosted the NMFR by 14 points. An unexpected increase in marriage among foreign-born Latinas over 1994–2005 placed downward demographic pressure on their NMFR.

We replicated these analyses for 15–19-year-old Latinas. Native-born teenage Latinas exhibited higher NMFRs than their foreign-born counterparts (78% vs. 66%, on average, over the study period; not shown). However, the NMFR increased similarly for both groups—by 11–12 percentage points. Like the results presented above, the decomposition results for teenagers differed by nativity status. Among foreign-born teenagers, increases in the NMFR attributable to decreasing marriage rates (three percentage points), increasing nonmarital fertility (six points) and declining marital fertility (three points) accounted for most of the net increase in the NMFR. Among native-born teenagers, the increase in the NMFR could be attributed to a large decline in the proportion who were married, which placed unprecedented upward pressure on the NMFR (38 percentage points). An increase in marital fertility and a decrease in nonmarital fertility, however, negated some of this change (causing declines of 14 and 13 percentage points, respectively).

Differences Between Racial and Ethnic Groups

Although the NMFR for black women was consistently higher than those for Latinas and white women, the demographic sources of racial or ethnic differences in NMFRs may not be uniform and may have changed over time. We therefore present a decomposition of such differences in NMFRs at the beginning and end of our time series (Table 3).

These data reveal significant cross-sectional variation in the effect of specific demographic factors on racial and ethnic differences in nonmarital fertility ratios. For example, NMFRs were much higher among Latinas than among whites in both 1994 and 2005 (21- and 18-percentage-point differences, respectively). Virtually all of the difference was due to the much higher nonmarital fertility rates among Latinas.

The results were dramatically different for comparisons involving black women. In 1994, blacks' NMFR was 50 percentage points higher than whites'; most of this difference resulted from the much higher black nonmarital fertility rate (25 percentage points) and lower marriage rate (20 points). In 2005, the black-white NMFR difference of 41 percentage points was driven mostly by a smaller proportion of women who were married (20 percentage points) and lower marital fertility (13 points); the contribution from the difference in nonmarital fertility shrank by nearly 67%, to eight percentage points. Most of the black-Latina difference in NMFRs in 1994 and 2005 (29 and 23 percentage points, respectively) can be explained by higher rates of marriage among Latinas (26 points for both years). However, reductions due to differences in nonmarital fertility (three percentage points in 1994 and 16 points in 2005) and increases stemming from differences in marital fertility (six points and 12 points, respectively) are of increasing importance.

The underlying demographic factors responsible for differences in NMFRs within the Latina population became more pronounced over time and differed by age. For example, the counterbalancing effects of the proportion married and nonmarital fertility were responsible for much of the gap in NMFRs between native-born and foreign-born Latinas (of all ages) in both 1994 and 2005 (seven and six percentage points, respectively). Moreover, each of these standardized effects was larger in 2005 than in 1994. The differences in NMFRs between native-born and foreign-born teenage Latinas were similar in the two years (11 and 12 percentage points, respectively); change in the married share was mostly offset by change in the nonmarital fertility rate in 2005, but was only partially offset in 1994. At the beginning of the study period, a high rate of marriage among foreign-born teenagers in 1994 clearly placed downward demographic pressure on their NMFR and narrowed the difference between the groups' ratios, and a higher nonmarital fertility rate among native-born Latinas had the opposite effect. At the end of the period, the pattern was reversed.

DISCUSSION

Our fundamental goal was to identify the demographic sources of change in Latina nonmarital fertility ratios between 1994 and 2005. This is an important task in light of recent shifts in the size and composition of America's Latino population. The high rate of nonmarital childbearing arguably has been a significant deterrent to Latinos' full social and economic incorporation into American society.²³ Current estimates indicate that more than one-half of all teenage Latinas become pregnant at least once before age 20.²⁴ Moreover, as we have shown here, more than 40% of all Latina births occurred outside of marriage in the past decade, a figure that is well above the U.S. average.

NMFRs changed very little between 1994 and 2002, before increasing in recent years. Our results reveal that this comparatively flat trajectory concealed dynamic offsetting demographic effects. Indeed, the retreat from marriage played a large role, counterbalancing any positive trends associated with declining nonmarital fertility rates for black women and native-born Latinas. Except among foreign-born Latinas, declines in the proportion of women who were married generally accounted for much of the observed increases in NMFRs, as more single women overall were exposed to the risk of having a nonmarital birth. However, for foreign-born Latinas and whites, increases in nonmarital fertility rates played a significant role in explaining trends in the share of births occurring outside of marriage. A comparison of NMFRs showed that the bulk of the Latina-white difference was a result of increased nonmarital fertility rates. Notably, black-white and black-Latina comparisons for 2005 showed that most of the white and Latina "advantage" in having lower nonmarital fertility ratios now rests on rates of marriage and marital fertility, and not on childbearing among those who are unmarried.

Policymakers have been rightfully concerned about high nonmarital fertility rates, especially among teenagers. Our results indicate that the underlying components of changes in teenage Latinas' NMFRs between 1994 and 2005 have mirrored those of changes in adults' ratios. Indeed, teenage and unintended childbearing exact a substantial cost on the nation, women and children. For this reason, among others, the 2006 reauthorization of the 1996 Personal Responsibility and Work Opportunity Act explicitly included the reduction of out-of-wedlock childbearing as a major goal of the legislation.^‡ Our findings highlight the need to understand changes in marital and nonmarital fertility in the context of the current retreat from marriage and apparently growing differentials among population groups, including racial and ethnic groups. The conventional wisdom states that declining rates of marital fertility, coupled with high or increasing rates of nonmarital fertility, are mostly responsible for the growing share of births that occur to unwed women. But our analyses show that marriage is also an important, although often overlooked, demographic source of change in nonmarital fertility ratios. The implication is clear: Marriage promotion and relationship counseling should be important components of a comprehensive strategy aimed at reducing out-of-wedlock childbearing.^25,26

Our analyses were limited by the availability of data for the 1994–2005 period only. Since 2005, the United States has seen a continuation of the uptick in nonmarital fertility ratios that began in 2002, in part because nonmarital fertility rates have increased, even among teenagers.^1,21 This study, along with the analyses of Smith and colleagues,8 emphasizes the unappreciated role of declining marriage rates. But the recent uptick may mark the beginning of a return to an earlier era, when the proportion of births that occurred outside marriage was driven largely by changing rates of nonmarital fertility, mostly at younger ages. Our results provide a baseline for additional studies of U.S. and Latina nonmarital fertility.