Much of the research conducted on teenage childbearing in the developed world indicates that compared with older mothers, teenage mothers are more likely to have been brought up in a less-advantageous social environment.1 Moreover, growing up in a low-socioeconomic-status household, living with a single parent and having low educational and career aspirations have all been shown to be strongly related to the likelihood of teenage childbearing, at least in the United States.2 For example, research has shown that the daughters and sisters of teenage mothers are more likely to become teenage mothers themselves.3 Studies have also documented subsequent socioeconomic disadvantage in adulthood among women who gave birth as teenagers, as measured by their lower educational attainment, poorer economic well-being, marital instability and difficulties in achieving desired family size.4
Because teenage mothers tend to come from poor social environments, the question of whether the accumulation of socioeconomic problems among teenage mothers reflects prechildbearing disadvantage or is a consequence of teenage motherhood has been vigorously debated.5 Studies on the possible long-term socioeconomic effects of early motherhood have mostly been conducted among heterogeneous populations characterized by relatively large socioeconomic differences. The population in Sweden is fairly homogenous, however, and the virtual absence of severe poverty there provides an opportunity to evaluate the long-term consequences of early childbearing from a new perspective. This article presents the results of a national, population-based longitudinal study that examined the association between teenage childbearing and socioeconomic characteristics later in life in Sweden.
Materials and Methods
Our analysis consists of a follow-up of all women born in Sweden from 1941 to 1970 who were registered in the 1985 Swedish Population Census. We focus on the 888,044 women registered in this census who had their first child between 1954 and 1989, when they were younger than 30 years old (at ages 11-29).
We then identified the parents of each woman through linkages to the country's Population Register, a unique national registration system in which numbers are assigned to each Swedish resident. We obtained information on the socioeconomic position in 1960 of the parents of 93% of the study population by further record- linkage to the 1960 Swedish Census. Family socioeconomic position refers to the occupation and status of whichever parent held the higher status in 1960.
The following independent variables were treated categorically in the analysis: the woman's completed age at the time she first gave birth; her birth cohort; the year in which she first gave birth; and her family's socioeconomic background. We stratified the women's completed age at first delivery into the following groups: 15 years or younger, 16-17, 18-19, 20-24 and 25-29. We arranged the women into three birth cohorts—those born in 1941-1950, in 1951-1960 and in 1961-1970. The years in which these women's first child was born were grouped into the following periods: 1954-1963, 1964-1973, 1974-1983 and 1984-1989. We categorized the woman's parent's socioeconomic position into blue-collar, white-collar, self-employed (including farming) and not employed.
Our study outcomes were the woman's socioeconomic situation and living arrangement in 1980, 1985 or 1990, depending on her birth cohort; her educational attainment as of 1990; her parity as of 1993; and her receipt of a disability pension and welfare dependency in 1994.
Information about the women's socioeconomic status and family arrangement during adulthood was obtained from a series of Population Censuses. To fit the five-year census data as closely as possible to the adulthood of the women, we recast the three birth cohorts into five, so that the resulting socioeconomic outcome data pertain to women at ages 35-39 for one of these five groups, to women at ages 30-34 for three of the groups, and to women at ages 20-29 for one of these groups.
For example, women born in 1941-1945 were linked to data collected in the 1980 census (when they would have been 35-39 years old) and women born in 1946-1950 were also linked to that 1980 census (yielding data that reflect their situation at ages 30-34). Women born in 1951-1955 were linked with data collected in the 1985 census (when they would have been 30-34 years old). Finally, we linked the 1990 census data both to women born in 1956-1960 (yielding outcome data corresponding to 30-34-year-olds) and to women born in 1961-1970 (resulting in outcome data for 20-29-year-olds).
The living arrangement data were derived from the same census sources and thus apply to the five cohorts in the same pattern; this measure did not take into account marital status and instead was simply dichotomized as single (not living with a partner) or living with a partner. Socioeconomic status was categorized by specific occupation based primarily on the degree of education required for each occupation.6
Information on educational attainment was available from the 1990 census only, and our classification of completed levels of schooling corresponds to the Swedish public school system.* We coded women who had gone no further than compulsory school as having a low educational level. Women who had completed secondary school were classified as having an intermediate level of education, and those who went beyond secondary school were considered to have a high educational level.
We used the individual record linkages to the Population Register, which covers births to all Swedish citizens, to obtain data on the number of children born to each woman as of 1993. We designated four parity groups—one child, 2-3 children, four children and five or more. Information on whether the woman was collecting a disability pension and welfare in 1994 was obtained from the Total Enumeration Income Survey of that year. This database of all nationally registered persons is maintained by Statistics Sweden, which collects nationwide data on social benefits. We dichotomized the disability pension variable as receiving a pension or not, irrespective of the woman's part-time employment status. Women were coded as dependent on social welfare if any person in her household received this benefit.
We used odds ratios to estimate the effect of age at first birth among women who had given birth as teenagers on the eight socioeconomic outcomes, using women who were aged 20-24 when they first gave birth as the reference group. We entered as controls the woman's birth cohort and her family's socioeconomic situation in 1960. Since three of the independent variables—year of the first child's birth, the woman's age at her first birth and the year of the woman's birth—are linearly dependent, the effects of all three variables cannot be captured in the same multivariate model. Thus, we excluded the year of the first child's birth, and included in the models the women's birth cohort together with her age at first birth. We performed multiple logistic regression analyses using the logistic procedure in the SAS program package.
Among all Swedish women born in 1941-1970 who were younger than 30 when they first gave birth, nearly 16% were teenagers. (Twelve percent were aged 18-19 at their first birth, 3% were 16-17 years old and only 0.1% were aged 11-15.) Compared with older mothers, a higher proportion of teenage mothers came from families with a blue-collar background or from families with unemployed parents (Table 1). Among the women who gave birth as teenagers, there were no substantial differences in family background measures between those who gave birth at ages 18-19 and those who were even younger.
When they were adults, women who had their first child as teenagers were more likely than those who delayed childbearing to be unskilled blue-collar workers, to be unemployed and to have a low educational attainment. There was a consistent age-dependent relationship between maternal age at first birth and the two outcome variables of single living arrangements and high parity (four or more children): These outcomes were highest among the youngest teenage mothers and were lowest among women who were in the oldest maternal-age group (25-29 years).
For example, among the 11-15-year-old mothers, 9% had had five or more children by 1993, while the corresponding proportion among 25- 29-year-old mothers was only 0.5%. The proportion of women who were collecting a disability pension or welfare in 1994 was also far higher among women who gave birth as teenagers than among those who waited (Table 2). In addition, the proportion of women who depended on social welfare increased linearly with decreasing maternal age. These relationships between women's age at first birth and the receipt of welfare or disability were the same across all birth cohorts of women.
According to the multivariate analysis used to examine the associations between age at first birth and later socioeconomic status, after adjustment for family background and maternal birth cohort, the odds of less favorable outcomes increased almost consistently with decreasing maternal age (Table 3). For example, relative to women aged 20-24 at their first birth, those aged 11-15 had the highest risks of unfavorable outcomes, while women aged 25-29 consistently had the lowest odds. Net of controls for the woman's birth cohort and her socioeconomic background, teenagers aged 11-15 at their first birth faced increased odds of being unemployed later in life, compared with the odds among women aged 20-24 at their first birth (odds ratio, 1.3). Teenage mothers also had significantly increased odds (odds ratios, 1.4-1.5) of being an unskilled blue-collar worker as an adult compared with women who gave birth at ages 20-24. Moreover, Swedish women who gave birth as teenagers were at increased odds of low educational attainment relative to those who gave birth when they were aged 20-24 (odds ratios, 1.7-1.9).
Swedish teenage mothers also had significantly increased odds of having single living arrangements later in life compared with women who first gave birth at ages 20-24 (odds ratios of 2.3 for those first giving birth at ages 11-15, 1.8 for 16-17-year-old mothers and 1.5 for 18-19-year-old mothers).
The odds of having five or more children were especially elevated among adolescent mothers relative to older mothers: Those giving birth at ages 11-15 (odds ratio, 6.0) and at ages 16-17 (odds ratio, 4.0) were significantly more likely to have had at least five children. Being a teenage mother also significantly raised the odds of collecting disability (odds ratios, 1.6-1.9) or welfare (odds ratios, 1.9-2.6) later in life, compared with women who gave birth at ages 20-24. When we restricted our analysis to include only women living without a partner, the effect of young maternal age on the odds of welfare dependency in 1994 was lowered somewhat, but remained significantly increased for all subgroups of adolescent mothers (not shown).
While the impact of young maternal age on later socioeconomic status was evident in all maternal birth cohorts, the odds of being unemployed, of living without a partner, of having at least four children and of collecting welfare were highest among teenage mothers who were born in 1961-1970, and these odds were lowest among the adolescent mothers in the 1941-1950 cohort. In an additional analysis restricted to Swedish women whose parents were white-collar workers, the adjusted effects of young maternal age at first birth on the various indicators of socioeconomic status in adulthood were of essentially the same magnitude (not shown).
To indirectly control for other social background factors such as educational aspirations and abilities, we conducted additional multivariate analyses that included only women who had completed secondary school or had gone on to a university (Table 4). The odds of adverse socioeconomic outcomes related to age at first birth were essentially unchanged from the odds found among women of all educational backgrounds (Table 3), with the exception of the effect of very young age (11-15) on parity. That is, including only better-educated women in the analysis reduced the effect of very young age at first birth on high parity—the adjusted odds ratio fell from 6.0 to 3.8.
Our results suggest that, compared with women who give birth at ages 20-24, those who first give birth as teenagers are at increased risk of socioeconomic disadvantage in adulthood in terms of employment, living arrangements, parity and dependence on social welfare. The odds for each of these outcomes were elevated irrespective of the adolescent mother's parents' socioeconomic situation and of her own educational level.
Our longitudinal study was based on nearly 140,000 Swedish teenage mothers. Many of our analyses' odds ratios were of a relatively large magnitude and high precision. The significant relationships were generally dose-dependent, with the highest odds of each unfavorable outcome being among the youngest mothers. In the bivariate analysis, teenage childbearing was far more common among women from blue-collar backgrounds and from families in which the parents were not gainfully employed, indicating that socioeconomic standing discriminates well between maternal-age groups. However, adjusting for socioeconomic background only slightly decreased the estimated effects of an early first birth. We were able to minimize the impact of the variation in the data resulting from choosing a single year in which to measure parent's socioeconomic background—i.e., in 1960, when the women in our oldest cohort would have been 10-19 years old and those in the youngest had not yet been born—because we adjusted for maternal birth cohort and family background in the same model. However, we cannot exclude the possibility of residual confounding and we may, to some extent, have overestimated the effect attributable to age at first birth per se.
It is important to distinguish between causality and correlation. Since teenage mothers tend to come from poorer social environments than women who give birth later, socioeconomic problems in adulthood might reflect conditions that were already present before a birth.7 Measuring parents' socioeconomic status may not adequately capture extreme disadvantage or sociocultural factors in early childhood, and the estimated odds may be upwardly biased.8
Researchers have used a variety of study designs to more adequately estimate the independent effects of a teenage birth on the adult life course. Such techniques include comparing sisters who first gave birth at different ages; comparing teenagers who had a singleton birth with those who had twins; and comparing teenagers who gave birth with those who had a miscarriage.9 Studies based on these techniques have yielded a reduced effect of young age at first birth compared with studies that have tried to adjust for social selection into early motherhood. For example, in one such analysis of a sample of 428 sister-pairs, the negative social effects of teenage childbearing were reduced when family background factors were taken into account, although significant negative effects for outcomes such as high school graduation and economic well-being were still present.10
Our data set did not permit us to design a study with a control group of sisters. Family background, as measured in our study, might be too rough a measure to adequately describe social conditions before childbirth. However, socioeconomic background discriminated well between maternal-age groups. To further reduce the possible bias from differences in unmeasured confounders, such as educational aspirations, we stratified the analyses and found that negative social effects were also present among women who had intermediate and high levels of formal education.
The likelihood of giving birth as a teenager is far lower in Sweden than in other developed countries, such as the United States and Great Britain.11 We suggest that the long-term social consequences of teenage childbearing may be comparatively more serious in a society where teenage childbearing is rare. In Sweden, contraceptives have been readily available to teenagers since the mid-1970s, and the country has a long tradition of providing comprehensive sex education at school. Efforts to reduce Sweden's relatively low number of unintended adolescent births have been successful, and the abortion rate among teenagers has declined in tandem with the adolescent birthrate.12 The sharp decline in Swedish teenage fertility began in the early 1970s. The rate went from 42 births per 1,000 15-19-year-olds in 1966-1970 to seven per 1,000 in 1997; this rate is currently one of the lowest in Europe.13
In our analysis, women who first gave birth at ages 20-24 still had elevated odds of an unfavorable situation in adulthood compared with women who first gave birth at an even later age (25-29).† These differences cannot be explained by the fact that even older mothers have had better opportunities to reach higher educational levels and that education correlates with other outcomes.
Our study population covered women born over three decades (essentially, the 1940s, 1950s and 1960s) who had their first child between 1954 and 1989. Although increased odds of unfavorable social outcomes were evident in all three birth cohorts, the odds ratios were generally greater for the youngest birth cohort (those born in 1961-1970) than for older women. Specifically, the odds ratios of high parity by 1993 were highest among teenage mothers in the youngest birth cohort, despite the fact that these women were still well within childbearing age with many fertile years ahead when the parity data were collected in 1993. (These women were aged 22-32 in 1993, compared with those in the earliest birth cohort, who would have been 33-52 in 1993.)
The adult socioeconomic outcome data from the censuses also reflect the situation of much younger women for these teenage mothers born in the latest cohort. Such young women are also more likely than those born in earlier cohorts to still have small children in the household, a fact that probably influences their employment status and overall economic situation. Research has shown that women who have many children face a poorer socioeconomic situation than those with few children,14 and many women who begin childbearing at an early age ultimately go on to have larger families (and shorter interpregnancy intervals) than women who begin childbearing when they are older.15A small-scale Swedish study that investigated deliveries to teenagers in the 1970s, for example, found that about 10% of these young mothers became pregnant again during their first year postpartum.16Thus, women who become mothers during their teenage years are not only more likely to face a less favorable socioeconomic situation in later life, but must also support a larger family.
We found that adolescent mothers are more likely than older mothers to raise their children in a single-parent household, a finding similar to that of previous research.17 The youngest mothers in our sample probably lived with their own parents after they had their first child, while most adult mothers cohabited with the child's father. The increased likelihood of single parenthood among the adolescent mothers further worsens their economic disadvantage.
Thus, teenage motherhood constitutes a risk for a less favorable socioeconomic situation in adulthood among Swedish women born in the 1940s through 1960s. This independently higher risk for disadvantage in later life appears to apply even to adolescent mothers who come from a favorable family background and who have gone beyond elementary school.