Choices regarding sexual behavior and contraceptive use can mitigate the risks of acquiring a sexually transmitted infection (STI) and becoming pregnant. A key feature of these choices is that risk mitigation strategies entail trade-offs. While some choices, such as abstinence, reduce both STI and pregnancy risk, others decrease one risk but increase or leave the other unchanged. For example, condoms and monogamy both reduce STI risk, but can potentially increase the likelihood of pregnancy. This is especially true if condoms are used to substitute for more reliable forms of contraception, such as oral contraceptives, or if women with only one sexual partner have condomless sex more frequently or view pregnancy as more desirable.

I estimate the effect of increases in STI risk and resulting behavioral changes on birth rates. Because of the trade-offs between STI and pregnancy prevention strategies, the effects of increases in STI risk are theoretically ambiguous. I empirically examine the effect of STI risk on birth rates by exploiting variation in the spread of AIDS across U.S. cities in the 1980s and 1990s. The AIDS epidemic created a large and plausibly exogenous increase in the cost of contracting an STI. During this period, the spread of AIDS was largely driven by male same-sex contact and the average time between HIV infection and AIDS diagnosis was 10 years. Thus, within a city, the timing of AIDS arrival and the extent of the epidemic was unrelated to pre-existing trends in birth rates.

I find that local AIDS incidence has a positive and statistically significant effect on both birth rates and abortion rates. I estimate that the birth rate increased by 0.55 percent due to women adopting behaviors associated with lower risk of contracting HIV/AIDS, for a total of 191,776 additional births between 1981 and 1996. I show that the increase in births is not due to a broader cultural shock associated with the AIDS epidemic, nor is it due to increases in “risky” sexual behavior. Women adjust their behavior in response to their specific risk of infection, as measured by local AIDS incidence among those with only opposite sex partners. Further, there is a corresponding decrease in the incidence of other STIs. Women adopted behaviors that decreased their likelihood of contracting AIDS and other STIs, but at the expense of heightened pregnancy likelihood.

I provide evidence that the increase in births is in part due to women entering monogamous partnerships to lower their risk of contracting HIV/AIDS. The increase in births is accompanied by an increase in marriages and an improvement in infant health as measured by birth weight and prenatal care. These results suggest that some women choose both monogamy and pregnancy in response to the AIDS epidemic. This mechanism is consistent with survey data showing that 16 percent of unmarried women decided to stop having sex with more than one man in response to the AIDS epidemic, and that this was the most commonly reported behavioral change (Mosher and Pratt, 1993). The increase in abortions may be the result of unplanned pregnancies in monogamous partnerships or of women switching from more effective contraceptives to condoms to lower their risk of contracting HIV/AIDS. There is also evidence of this in survey data – between 1988 and 1995 decreases in use of the oral contraceptive pill were fully offset by increases in condom use (Piccinino and Mosher, 1998).

Previous studies have shown that people adjusted their behavior in response to the AIDS epidemic using both contraceptive technology and choices about sexual partners. In the United States, Francis (2008) shows that those who had a relative with AIDS adjusted their choice of sexual partners to lower their risk of infection: Men shifted to opposite-sex partners and women shifted to same-sex partners. Ahituv et al. (1996) show that men adopted condoms in response to local AIDS prevalence, but did not find a similar effect among women. In Brazil, Hakak and Pereda (2021) show that the AIDS epidemic increased women’s marriage market value and increased the marriage rate.

I show that these behavioral changes had a spillover effects on birth rates. This is the first paper to examine the effect of increased STI risk on birth rates in a developed country. There are multiple studies examining the effects of the AIDS epidemic on fertility in countries in sub-Saharan Africa, with mixed results (Fortson, 2009, Magadi and Agwanda, 2010, Kalemli-Ozcan and Turan, 2011, Karlsson and Pichler, 2015, Duflo et al., 2015, Chin and Wilson, 2018).2 These studies are in part motivated by the potential effects of higher child mortality rates on fertility. This mechanism is unlikely to explain findings in the U.S. context, given the much lower child mortality rate overall and the very low incidence of AIDS in children in the U.S. in this time period. Nevertheless, my results do correspond to those of Duflo et al. (2015), which finds that pregnancy and STI in adolescent girls in Kenya are not determined by unprotected sex, but by choices over casual versus committed relationships.

Studies in developing countries also document the importance of bargaining power in women’s ability to lower their risk of contracting STIs. For example, Gertler et al. (2005) show that sex workers with more bargaining power can charge a higher premium for condomless sex. Anderson (2018) shows that female HIV rates are higher when women have less bargaining power and are less able to negotiate safe sex practices, such as condom use. Cassidy et al. (2021) also shows that condom use is lower in households with lower female bargaining power. It is unclear if these results would apply in the U.S. context, particularly given differences in the accessibility of contraceptives other than condoms. However, there is existing evidence that bargaining power in sexual partnerships affects women in the U.S.: Akerlof et al. (1996) argue that greater availability of contraception and abortion decreased women’s bargaining power in marriage decisions.

This paper highlights that decisions about number of partners are an important risk mitigation strategy for women, particularly given women’s lesser control over condom use. In further support of this conclusion, I find that overall results are driven by an increase in births to white women and that there is no effect of local AIDS incidence on births to Black women. This is consistent with results from Charles and Luoh (2010) and Johnson and Raphael (2009) showing that the high incarceration rate among Black men during the 1980s and 1990s decreased the marriage rate and increased AIDS incidence in Black women. Absent the option to mitigate STI risk via monogamy, women may have few other accessible options available to lower their risk of contracting an STI.

Understanding the effects of increases in STI risk is relevant to current trends in public health. The emergence of sexually transmitted monkeypox cases and the rapid rise of drug-resistant gonorrhea highlight the rising costs associated with STI risk and the continual importance of understanding how people respond to these risks (Kupferschmidt, 2022, US DHHS, 2017, Bodie et al., 2019). Understanding how people adjust their behaviors in response to STI risk can inform future public health interventions to address the spread of disease. My research provides insights into the unique constraints women face in mitigating STI risk given the concurrent risk of pregnancy. Decisions about sexual partnerships can be an important risk mitigation strategy, but at the cost of increased pregnancy likelihood.