Although the teen birth rate has been declining in recent years, the United States still has the highest teen birth rate of any industrialized country in the world (United Nations, 2017). As such, there is substantial policy interest in understanding the causes of teen pregnancy and, more generally, adolescent sexual behavior.3 Previous work has investigated the relationship between demographics (such as race, religion, education, and income) and teen pregnancy, the impact of policies regarding access to abortion, sex education, and family planning services on sexual behavior, and the role of media influences on teen childbearing.4 However, much less is known about the role of peer influence from teen pregnancy. In this paper, we explore whether sexual behavior is affected by observing the teen pregnancy of a peer (which we define as a friend or older sibling) and the specific ways in which their behavior changes. Understanding this relationship will enable the design of policies and programs that are most effective at reducing the risk of teen pregnancies.

Peers have been shown to play an important role in the risky behaviors of teenagers (Altonji et al., 2017, Ouyang, 2004, Card and Giuliano, 2013, Ajilore, 2015, Heissel, 2017, Heissel, 2021), but little is known about the impact of a sibling or friend’s teen pregnancy on one’s own sexual behavior. Teen pregnancy as an influence on a friend or sibling is particularly interesting because it may have large, visible consequences that are difficult for a teenager to fully imagine, ex ante. On the one hand, a teen might mimic a peer’s sexual behavior if they admire the peer, desire the same attention or view the experience as legitimizing of teen pregnancy. On the other hand, directly observing a pregnancy, abortion or baby might make an individual more cautious, deterring one from engaging in sexual behaviors that put them at risk of their own teen pregnancy.5 For siblings, changes in parenting style may also play a role, with parents either becoming more strict towards all offspring or instead focusing more attention on the sibling with the pregnancy.

Identifying the causal impact of a peer’s teen pregnancy on sexual behavior is complicated by the fact that environmental factors contributing to teen pregnancies will be common across the peer groups. Thanks to the extremely rich National Longitudinal Study of Adolescent to Adult Health dataset (Harris, 2018), we can exactly pinpoint the timing and characteristics of all sexual relationships and pregnancies of respondents. We therefore exploit the sharp timing of a peer’s teen pregnancy and analyze the evolution of an individual’s sexual behavior before and after the event, with approximately a third of the events being sibling teen pregnancies and the remainder friend teen pregnancies. Our approach identifies the causal impact of a peer’s pregnancy on own sexual behavior if peer groups with pregnancies would have been on a similar age trajectory in terms of their sexual behavior, absent the peer pregnancy. Identification is threatened if, for example, groups with teen pregnancies would have accelerated their sexual activity during their teenage years, relative to groups without pregnancies, even absent the pregnancy. Naturally, we cannot fully rule this pattern out; however, we note that we see no evidence of differential pre-trends for those with peer pregnancies. Further, we find similar results when allow for differential age trends as a function of observables and when we use as a control group those who will eventually have a peer teen pregnancy but have not yet experienced one by a certain age. A remaining concern, of course, is that peer groups experience contemporaneous shocks driving the pregnancy and any changes in sexual behavior (Manski, 1993). However, we expect such a bias would go in the opposite direction of our findings.

Indeed, we find that those who observe a friend or older sibling’s teen pregnancy are less likely to engage in sexual behavior that puts them at risk of their own teen pregnancy after the event compared to those who have not experienced a peer’s teen pregnancy by that age. In particular, we find that respondents are 2.3 to 4.6 percentage points (18 to 35 percent) less likely to have unprotected sex (primarily through their use of birth control, rather than through abstaining altogether) and have fewer sexual partners after observing a peer’s teen pregnancy. These effects on sexual behavior are evident primarily after the end of a peer’s teen pregnancy, suggesting that observing the lifestyle change from being a teen parent causes a larger behavioral change than the pregnancy itself.

In heterogeneity analyses, we find that effects are larger when the focal observation is female and when the peer is a teen mom. Both these effects seem intuitive given that girls have access to long-acting reversable contraception and the experience and visibility of a teen mom may differ from that of a teen dad. In addition, effects are entirely accounted for by peer pregnancies that result in a live birth, which makes sense given that these would be the most visible. These intuitive patterns are reassuring; if omitted variables were driving our results, we might expect those to be fairly similar across live births as those that terminate early and across females and males.

Lastly, we explore whether peer pregnancy impacts the likelihood of one’s own teen pregnancy using a hazard analysis. We find negative effects that are large in magnitude for the sibling sample and marginally significant, but for the sample as a whole, effects are smaller and insignificant, though still negative. We may lack the power to say much about teen pregnancy as an outcome given that it is relatively rare in this time period. That makes our analyses on sexual behaviors especially useful since effects can be esimated with more precision. Understanding the specific ways in which youth respond to a peer’s teen pregnancy is important for designing teen pregnancy prevention programs that target youth in the ways that they are most likely to respond. For example, if youth are likely to respond to a teen pregnancy by increasing birth control use, programs may be more effective if they provide information on different methods of birth control and access to family planning services.

Our paper contributes to the existing literature in several important ways. First, the existing literature on the impact of a peer’s teen pregnancy on sexual behavior has primarily focused on own teen pregnancy as an outcome.6 Their findings have been mixed: some papers have found that a teen pregnancy in the family (Monstad et al., 2011) or among grademates in the same school (Fletcher and Yakusheva, 2016) positively impacts one’s own probability of a teen pregnancy, suggesting that a desire to mimic the peer with the pregnancy will outweigh any deterrent effect for exposure to real consequneces, while others (Yakusheva and Fletcher, 2015, Kapinos and Yakusheva, 2016) have found that a friend’s teen birth is associated with a reduction in the likelihood of own teen pregnancy.7 Focusing exclusively on teen pregnancy as an outcome, as most of the previous literature has done, may miss some of the nuance of the effects on sexual behavior. Our paper seeks to fill this gap by studying a range of sexual behavior measures, including sexual activity, number of partners, and birth control use, and the timing of the response.

Another contribution of our paper is methodological. Previous papers that have looked at social learning from teen pregnancy or childbearing have attempted a range of clever identification strategies that are useful, but not without their own strong assumptions. Yakusheva and Fletcher (2015) and Kapinos and Yakusheva (2016) use the absence of a miscarriage as an instrument for the probability of teen childbearing, arguing that conditional on pregnancy, miscarriage probability is random. This is a nice approach that was first proposed by Hotz et al. (1997) and Hotz et al. (2005), but assumes that a friend’s miscarriage has no effect on behavior and may suffer from differential measurement error to the extent that youth may only selectively recall miscarriages. Fletcher and Yakusheva (2016) use the proportion of grademates whose mother had a teen pregnancy and the average age of menarche of grademates as instruments, neither of which fully rule out other environmental factors correlated with these instruments. Monstad et al. (2011) use a change in compulsory school laws in Norway as a source of variation in teen pregnancy, but cannot fully isolate the teen pregnancy effect from the main effects of increased schooling. Our paper develops a new identification strategy that does not depend on these strong assumptions and allows us to control for differences in the level of sexual activity across peer groups of the same age with and without teen pregnancies.

Finally, our paper contributes to the literature by looking at the impact of a sibling’s teen pregnancy in addition to a friend’s teen pregnancy. There are many reasons why we might expect a sibling’s teen pregnancy to have a larger impact on sexual behavior than a friend’s behavior. Youth have more exposure to the consequences of a sibling’s teen pregnancy than a friend’s teen pregnancy given that they are living in the same household. Furthermore, sisters of teen mothers have been shown to spend more time on childcare than those whose siblings do not have a teen pregnancy (Heissel, 2021), leaving them with less time to engage in other activities. Their parents may also become more strict after their older child’s teen pregnancy, possibly by monitoring their behavior more carefully, imposing rules like a curfew, or providing them with contraception. By including siblings in our definition of peers, our paper is able to capture a more complete picture of how youth respond to observing a peer’s teen pregnancy.

We find that youth respond to the experiences of their peers by reducing their own likelihood of having a teen pregnancy. Our results have the direct policy implication that education campaigns that provide a realistic portrayal of teen parenthood may be effective in reducing behaviors that may result in teen pregnancy. This is consistent with previous findings that the introduction of MTV’s reality tv show 16 and Pregnant, which shows viewers the realities of teen motherhood, led to a 4.3 percent reduction in teen births (Kearney and Levine, 2015).

The organization of the paper is as follows. Section 2 describes the data and provides some summary statistics. Section 3 presents the empirical strategy used to identify the effects of a peer’s teen pregnancy on sexual behavior. Section 4 presents the empirical results and discussion. Section 5 concludes.