Divorce and the Birth Control Pill in the US, 1950–85

ABSTRACT

This paper explores the relationship between the advent of the birth control pill and divorce rates. Women using the pill can decide when and whether to have children and whether to maintain their attachment to the labor force. This ability may increase women's autonomy, making divorce more feasible. The pill's effects are identified through a quasi-experiment exploiting differences in the language of the Comstock anti-obscenity statutes approved in the late 1800s and early 1900s in the United States. Empirical evidence from state-level data on US divorce rates 1950 to 1985 shows that sales bans of oral contraceptives have a negative impact on divorce. These findings are robust to alternative specifications and controls for observed (such as women's labor force participation) and unobserved state-specific factors, and time-varying factors at the state level. Results suggest that the impact of women's control of hormonal contraception on their autonomy is important in divorce decisions.

KEYWORDS:

• Divorce rate
• birth control pill
• sales bans
• unilateral divorce

JEL Codes:

• J12
• J13
• J18

NOTES ON CONTRIBUTOR

Miriam Marcén received her PhD in Economics from the Universidad de Zaragoza (Spain) in 2010. Currently, she is working at the Universidad de Zaragoza. Her main research topic is household economic behavior, with particular interest in marital relationships and health economics. Her work has been published in Demography, Journal of Economic Behavior and Organization, European Journal of Health Economics, Labour Economics, and Economic Modelling, among others.

ACKNOWLEDGMENTS

The author is grateful for the financial support from the Spanish Ministry of Economics and Competitiveness (Project ECO2012-34828). I also thank the editors and the anonymous reviewers for their constructive comments.

Notes

1 Female sterilization is the most effective contraceptive method. However, a surgical procedure is necessary, and women cannot have children after using this method, without further surgery. Of 100 women who have a sterilization procedure, less than one would expect to become pregnant each year (Bullough 2001). The birth control pill is less effective than female sterilization although, if used correctly, the probability of failure is less than 0.1 percent (Bullough 2001), but much more effective than the other two most common barrier methods, the diaphragm and the condom (Bailey 2010). Women controlled the diaphragm, which experienced failure rates of 16 percent, including all users (James Trussell 2007). The diaphragm was more effective among experienced users (Judith Bruce and S. Bruce Schearer 1979).

2 For example, as Christensen (2012) explained, use of the pill was 36–40 percent greater among 17–19-year-old women who lived in states with less restrictive laws regarding contraception.

3 Those states explicitly considering contraception an obscenity, and introducing sales bans, are: Arkansas, Arizona, California, Colorado, Connecticut, Delaware, Iowa, Idaho, Illinois, Indiana, Kansas, Massachusetts, Minnesota, Missouri, Mississippi, Montana, Nebraska, Nevada, New Jersey, New York, Ohio, Oregon, Wisconsin, and Wyoming. Those states explicitly considering contraception an obscenity, but not introducing sales bans, are: Louisiana, Maine, Michigan, Pennsylvania, South Dakota, and Washington. These laws were coded in Bailey (2010).

4 In 1957, the pharmaceutical company G. D. Searle marketed the first pill (Enovid) for gynecological and menstrual disorders. This became possible after development by several researchers from a range of countries, such as Gregory Pincus and colleagues, who were supported by the birth control advocate Margaret Sanger with financial backing by Katharine McCormick (Barbara Seaman and Gideon Seaman 1977; Marks 2010).

5 There is no information on Comstock laws in the case of Alaska, District of Columbia, and Hawaii; see Bailey (2010). Thus, those states are not considered in the analysis.

6 Estimated using weights. Results do not vary substantially without weights.

7 The number of married people in years in which the decennial US census was conducted was obtained to calculate this rate (Integrated Public Use Microdata Series; Steven Ruggles, Trent Alexander, Katie Genadek, Ronald Goeken, Matthew B. Schroeder, and Matthew Sobek 2010). Yearly data was reckoned by linear interpolation. The annual number of divorces comes from CDC (2014), and Wolfers (2006).

8 The standard definition of divorce, known as the crude divorce rate, is defined as the annual number of divorces per 1,000 individuals. The evolution of the crude divorce rate is similar to that described in this work. Both are plotted in Figure 1.

9 As Halla (2013) suggests, Nevada is excluded since the evolution of the divorce rate in this state is quite different to the rest of the states. As a robustness check, all of the analysis is run with Nevada included, and results are maintained.

10 To test whether results are sensitive to the definition of divorce, a simple robustness check was run. The analysis was also repeated using an additional dependent variable, the crude divorce rate. Results are displayed in the Supplemental Online Appendix Table A.1 showing that, despite the small change in the magnitude of the impact of the variable of interest (this is not surprising since the denominator of the dependent variable has changed), a negative and significant effect of sales bans on divorce can still be observed.

11 Nevada is excluded in this case as well.

12 Data on women's LFP at the state level come from the Current Population Survey (Miriam King, Steven Ruggles, J. Trent Alexander, Sarah Flood, Katie Genadek, Matthew B. Schroeder, Brandon Trampe, and Rebecca Vick 2010) and from the Integrated Public Use Microdata Series (Ruggles et al. 2010; gaps filled by linear interpolation).

13 Data come from the US decennial Census from 1950 to 1990 (Ruggles et al. 2010; gaps filled by linear interpolation).

14 This information is obtained from the Panel Income of Study Dynamics (1970).

15 The inclusion of all these omitted controls may be problematic since, as Douglas W. Allen (2002) suggests, many of the measures of economic performance, or other demographic attributes, have not been truly exogenous. As can be seen in Table 2, the coefficients capturing the effect of the controls included in this subsection are not as expected, although the signs change when all the same specifications are added. Indeed, Gary S. Becker (1981) explained that causality between these variables and the divorce rate may run in both directions. Thus, because of endogeneity concerns, I prefer not to include them in the rest of the analysis, even though, as shown previously, the results on the variable of interest are unchanged after adding these controls. Note that the analysis has also been run including these omitted variables and results are maintained.

16 Findings remain unchanged even if only state and year fixed effects or state-specific linear trends are introduced.

17 All of the analyses were repeated adding controls for the variations in early legal access to the pill, and results do not vary substantially.

18 The main distinction in marital regimes is between separation of property and community property (see Carmen Diana Deere and Cheryl R. Doss [2006]). In separation of property, known as Common Law, all property acquired prior to marriage and any property acquired during the marriage belongs to the person who acquired it. In (partial) community property, all property acquired during the marriage belongs equally to the couple, but property acquired prior to marriage belongs to the individual; and in full community property regimes, all property, irrespective of when it was acquired, belongs to the couple once they marry. The “equitable distribution” laws give judges discretion to distribute property in a manner different than what would be forthcoming under the given marital regime. As Gray (1998) notes, in the case that men own most of the property, under Common Law, most of the property would go to the husband, and under community property, wives would be better off since they are automatically entitled to half of the marital property (anything acquired during the marriage). With equitable distribution, the judge can modify this distribution, but would be more likely to do so in Common Law states, since the division of property (the 50–50 rule for marital property) is statutory in community property states.

19 Note that, since the entire population at risk of divorce is considered, any change in the marriage population, such as post-war or baby boomer marriages, is taken into account in the dependent variable.

20 The methodology used earlier only identifies discrete series break.

21 Although the first large-scale clinical trial, developed in Puerto Rico in 1956, generated controversy on the safety of the birth control pill because of the unpleasant side effects such as nausea, headaches, and dizziness (Jay Katz, Alexander Morgan Capron, and Eleanor Swift Glass 1972; Marks 2010), differences in the divorce rate as a consequence of that oral contraception are observed in the short term. This is not surprising since, as Marks (2010) claims, once the pill reached the market, opposition to it disappeared. The promotion of the pill by the lobby for population control was also significant, and it provided the pharmaceutical industry with enormous business. Note that those clinical trials would not meet the ethical standards of today, but they were the norm in the 1950s (Marks 2010).

22 As in González-Val and Marcén (2012), I introduce two main policy changes that have swept the US since the late 1970s: the approval of the joint-custody regime, and the Child Support Enforcement program. Under a joint-custody regime, decisions affecting the child must be jointly made by parents, requiring discussion and collaboration (González-Val and Marcén 2012). The Child Support Enforcement program is aimed at enforcing support obligations to prevent poverty among children and to reduce welfare costs (González-Val and Marcén 2012). Data come from Teng Wah Leo (2006) and González-Val and Marcén (2012).

23 Alaska, District of Columbia, Hawaii, and Nevada are also excluded here.

24 Despite the clear differences across countries, the UN Conference on Human Rights in Tehran, in 1968, recognized family planning as a basic human right for all.