Skip to main content

Alcohol consumption and risky sexual behavior among young adults: evidence from minimum legal drinking age laws

Abstract

This paper exploits the discrete jump in alcohol consumption at the minimum legal drinking age (MLDA) in the USA and uses a regression discontinuity design to investigate the relationship between drinking and risky sexual behaviors among young adults. Using confidential data from the National Longitudinal Survey of Youth (1997 Cohort), we document that young adults tend to drink up to 2.1 days more once they are granted legal access to alcohol at age 21. Although the discrete jump in alcohol consumption at the MLDA is associated with an increase in the probability of having sex by up to 7.8 percentage points, it does not have a significant impact on the probability of engaging in risky sexual behaviors among young adults. We also document that the effect of the MLDA on the probability of using several different birth control methods is not significant for those who had sex in the past 4 weeks. These results are robust under alternative specifications and imply that although the MLDA law is quite effective in reducing alcohol consumption among young adults, spillover effects of this law on risky sexual behaviors are relatively limited.

This is a preview of subscription content, access via your institution.

Fig. 1
Fig. 2
Fig. 3

Notes

  1. By 1988, all states had set 21 as the MLDA. Since then, it has been illegal for youths under age 21 to purchase or consume alcohol in the USA. In some states, alcohol consumption under 21 may be allowed under certain circumstances. These include religious or educational purposes and the presence of a parent. We do not address these exceptions directly because their impact is likely to be minor in the context of our study.

  2. Wagenaar and Toomey (2002) and Carpenter and Dobkin (2011) provide an extensive review of the literature on the effects the MLDA laws. The main focus of most of the existing studies is the impact of the MLDA laws on alcohol-related traffic accidents. See, for example, Dee (1999), Lovenheim and Slemrod (2010), Carpenter and Dobkin (2009), and Kreft and Epling (2007).

  3. The exception is Miron and Tetelbaum (2009) who use state-level panel data to show that any nationwide impact of the MLDA is driven by states that increased their MLDA prior to any inducement from the federal government.

  4. The main identifying assumption in these studies is that the observable and unobservable determinants of alcohol consumption and alcohol consumption-related outcomes are likely to be distributed smoothly across the cutoff of age 21. Hence, the changes in alcohol consumption and alcohol consumption-related outcomes at age 21 can solely be attributed to the MLDA law itself. This assumption is partially testable. We discuss this identification problem in Section 4 in detail.

  5. As discussed by Yörük and Ertan Yörük (2011), without information on exact birth dates, it is not possible to identify the treatment and control groups around the MLDA cutoff. Suppose that one has information only on the month and year of the birth date of each respondent and her interview date. A respondent who was born on January 30, 1980 and interviewed on January 1, 2001 will be mistakenly coded as a 21-year-old and placed in a treatment group (those who are 21 and older). But, this respondent is actually in the control group since she is 29 days younger than 21 at the time of the interview. Furthermore, by definition, the RD approach estimates the local treatment effect, which calls for a very detailed information around the cutoff of age 21.

  6. These studies include those of Butcher et al. (1991), Cooper et al. (1994), Bentler and Newcomb (1986), Mott and Haurin (1988), and Staton (1999).

  7. This corresponds to a bandwidth of 732 days on either side of the age cutoff of 21. We followCarpenter and Dobkin (2009) and Yörük and Ertan Yörük (2011) in order to choose the age bandwidth. The NLSY was also employed by other researchers to study the determinants of teenage alcohol consumption and sexual behavior. See, for example, Arkes and Klerman (2009) and Keng and Huffman (2010). We also consider alternative age bandwidths and inclusion of married respondents to our sample as a part of robustness checks. These results are presented in Section 5.

  8. Since the empirical results are based on self-reported survey data, young adults under 21 may also be more likely to underreport their alcohol consumption since alcohol consumption is illegal for those who are under this cutoff age. This could generate a discrete jump in reported level of alcohol consumption at age 21 even if there is no true change in actual behavior. However, Yörük and Ertan Yörük (2011) argue that in the NLSY97, the alcohol consumption patterns of 21-year-olds are quite similar compared with those of 20- and 22-year-olds, which suggests that the empirical results documented in this paper are not subject to a underreporting bias.

  9. Yörük and Ertan Yörük (2011) use the same alcohol consumption outcomes in their analysis. Our results are similar but slightly different than their results because their estimation sample includes only those who have consumed alcohol at least once since the DLI. In our analysis, we also include those who did not consume alcohol since the DLI. Naturally, for these respondents, the number of days that they consumed alcohol or engaged in binge drinking over the past month is zero.

  10. The wording of the question is as follows: “About how many times have you had sexual intercourse in the past four weeks?”. We exclude those who reported to have had sex more than 50 times in the past month (505 observations).

  11. In the 2000 survey year, there was a single question about the birth control method used in the most recent sexual intercourse. Hence, respondents were allowed to choose only one of the birth control methods listed. However, in the 2001 to 2005 survey years, respondents were first asked whether they used condom in their most recent sexual intercourse. If the answer is a yes, then they were asked whether they used any other birth control method in their most recent sexual intercourse. Therefore, in the 2001 to 2005 survey years, respondents were allowed to choose two birth control methods instead of one. Hence, the percent of respondents who reported to have used condom, pill, or any other birth control method exceeds 100 % for the full sample.

  12. Young adults may also tend to drink more in their birthdays. Therefore, it is hard to distinguish the birthday celebration effect in the 21st birthday from the effect of the MLDA law. Carpenter and Dobkin (2009) argue that a dummy variable, which is equal to 1 if the respondent was interviewed during the first month after turning 21, also controls for the potential impact of the birthday celebration at the MLDA.

  13. Imbens and Lemieux (2008) and Lee and Lemieux (2009) provide a detailed discussion of the RD design.

  14. Following Yörük and Ertan Yörük (2011), we first calculate household income in 2006 prices. Next, we create dummy variables for different income ranges (less than 20,000; 20,000 to 40,000; 40,000 to 60,000; 60,000 to 80,000; 80,000 to 100,000; more than 100,000; and a dummy for missing observations) and include these dummies to our regressions as controls for income.

  15. Our selection control variables follows Yörük and Ertan Yörük (2011, 2013), who use NLSY97 to investigate the spillover effects of the MLDA law on smoking and marijuana use among young adults. Yörük and Ertan Yörük (2011) also test the possibility that these control variables exhibit a discrete change at the MLDA. They show that there is no evidence of a significant change in any of the control variables at the MLDA. Since we employ the same data set and consider the same age bandwidth, this finding also applies to our analysis. This result also reduces the concerns about omitted variable bias and suggests that parametric models estimated with or without controls should yield similar results.

  16. Although not reported here, we also estimate nonparametric models following Hahn et al. (2001). In these models, following Malamud and Pop-Eleches (2011), we use triangular kernel which has been shown to be boundary optimal by putting more weight on observations closer to the cutoff point () and employ the bandwidth selection procedure suggested by Imbens and Kalyanamaran (2012). In general, similar to parametric models, nonparametric models also suggest that the effect of the MLDA on risky sexual behaviors among young adults is statistically insignificant.

  17. We do not include sexual activity as a control variable when estimating the effect of the MLDA on alcohol consumption. However, including this variable does not affect our results. For instance, our original estimate of the effect of the MLDA on alcohol consumption in a model that contain a quartic age profile as reported in Table 2 is 2.117 with a standard error of 0.620. If we include the number of times that the respondent had sex during the past 4 weeks as an additional control variable to our model, we get an estimate of 2.212 with a standard error of 0.627, which is virtually the same compared with our original estimate.

  18. Other birth control methods include foam, jelly, creme, sponge, or suppositories, withdrawal, diaphragm, safe time, intrauterine device, norplant, depo-provera or injectables, and any other method.

  19. We use the full set of control variables in all regressions. Although not reported in this paper, the results from models that contain quadratic or cubic polynomial of age yield similar results and are available from the authors upon request.

References

  • Arkes J, Klerman JA (2009) Understanding the link between the economy and teenage sexual behavior and fertility outcomes. J Popul Econ 22:517–536

    Article  Google Scholar 

  • Bentler PM, Newcomb MD (1986) Personality, sexual behavior and drug use revealed through latent variable methods. Clin Psychol Rev 6:363–-385

    Article  Google Scholar 

  • Butcher AH, Thompson T, O’Neal E (1991) HIV-related sexual behavior of college students. J Am Coll Health 40:115–188

    Article  Google Scholar 

  • Carrell SE, Hoekstra M, West JE (2011) Does drinking impair college performance? Evidence from a regression discontinuity approach. J Public Econ 95:54–62

    Article  Google Scholar 

  • Carpenter C (2005) Youth alcohol use and risky sexual behavior: evidence from underage drunk driving laws. J Health Econ 24:613–628

    Article  Google Scholar 

  • Carpenter C, Dobkin C (2009) The effect of alcohol consumption on mortality: regression discontinuity evidence from the minimum drinking age. Am Econ J: Appl Econ 1:164–182

    Google Scholar 

  • Carpenter C, Dobkin C (2011) The minimum legal drinking age and public health. J Economic Perspect 25:133–156

    Article  Google Scholar 

  • Cheng MY, Fan J, Marron JS (1997) On automatic boundary correction. Ann Stat 25:1691–1708

    Article  Google Scholar 

  • Chesson H, Harrison P, Kassler W (2000) Sex under the influence: the effect of alcohol policy on sexually transmitted disease rates in the United States. J Law Econ 43:215–238

    Article  Google Scholar 

  • Cooper ML, Peirce RS, Huselid RF (1994) Substance use and sexual risk-taking among black adolescents and white adolescents. Health Psychol 13:251–262

    Article  Google Scholar 

  • Dee TS (1999) State alcohol policies, teen drinking and traffic fatalities. J Public Econ 72:289–315

    Article  Google Scholar 

  • Dee TS (2001) Minimum legal drinking ages and teen childbearing. J Hum Resour 36:823–828

    Article  Google Scholar 

  • Ertan Yörük C, Yörük BK (2012) The impact of drinking on psychological well-being: evidence from minimum drinking age laws in the United States. Soc Sci Med 75:1844–1854

    Article  Google Scholar 

  • Grossman M, Markowitz S (2005) I did what last night? Adolescent risky sexual behaviors and substance use. East Econ J 31:383–405

    Google Scholar 

  • Hahn J, Todd P, van der Klaauw W (2001) Identification and estimation of treatment effects with a regression-discontinuity design. Econometrica 69:201–209

    Article  Google Scholar 

  • Imbens G, Lemieux T (2008) Regression discontinuity designs: a guide to practice. J Economet 142:615–635

    Article  Google Scholar 

  • Imbens G, Kalyanamaran K (2012) Optimal bandwidth choice for the regression discontinuity estimator. Rev Econ Stat 79:933–959

    Article  Google Scholar 

  • Keng S, Huffman WE (2010) Binge drinking and labor market success: a longitudinal study on young people. J Popul Econ 23:303–322

    Article  Google Scholar 

  • Kreft SF, Epling NM (2007) Do border crossings contribute to underage motor-vehicle fatalities? An analysis of Michigan border crossings. Can J Econ 40:765–781

    Article  Google Scholar 

  • Lee DS, Lemieux T (2009) Regression discontinuity designs in economics. NBER working paper no. 02138.

  • Lovenheim MF, Slemrod J (2010) The fatal toll of driving to drink: the effect of minimum legal drinking age evasion on traffic fatalities. J Health Econ 29:62–77

    Article  Google Scholar 

  • Malamud O, Pop-Eleches C (2011) Home computer use and the development of human capital. Q J Econ 126:987–1027

    Article  Google Scholar 

  • Miron JA, Tetelbaum E (2009) Does the minimum drinking age save lives? Econ Inq 47:317–336

    Article  Google Scholar 

  • Mott FL, Haurin RJ (1988) Linkages between sexual activity and alcohol and drug use among American adolescents. Fam Plann Perspect 20:15–30

    Article  Google Scholar 

  • Rashad I, Kaestner R (2004) Teenage sex, drugs and alcohol use: problems identifying the cause of risky behaviors. J Health Econ 23:493503

    Article  Google Scholar 

  • Reece M, Herbenick D, Schick V, Sanders SA, Dodge B, Fortenberry JD (2010) Condom use rates in a national probability sample of males and females ages 14 to 94 in the United States. J Sex Med 7:291–304

    Article  Google Scholar 

  • Rees D, Argys L, Averett S (2001) New evidence on the relationship between substance use and adolescent sexual behavior. J Health Econ 20:835–845

    Article  Google Scholar 

  • Sen B (2002) Does alcohol-use increase the risk of sexual intercourse among adolescents? Evidence from the NLSY97. J Health Econ 21:1085–1093

    Article  Google Scholar 

  • Staton M (1999) Gender differences in substance use and initiation of sexual intercourse. Popul Res Policy Rev 18:89–100

    Article  Google Scholar 

  • Wagenaar AC, Toomey TL (2002) Effects of minimum drinking age laws: review and analyses of the literature from 1960 to 2000. J Stud Alcohol 14:206–225

    Google Scholar 

  • Yörük BK, Ertan Yörük C (2011) The impact of minimum legal drinking age laws on alcohol consumption, smoking, and marijuana use: evidence from a regression discontinuity design using exact date of birth. J Health Econ 30:740–752

    Article  Google Scholar 

  • Yörük BK, Ertan Yörük C (2013) The impact of minimum legal drinking age laws on alcohol consumption, smoking, and marijuana use revisited. J Health Econ 32:477–479

    Article  Google Scholar 

Download references

Acknowledgments

This paper uses confidential data provided by Bureau of Labor Statistics (BLS). The views expressed in this paper are those of the authors and do not necessarily reflect those of the BLS.

Author information

Authors and Affiliations

Authors

Corresponding author

Correspondence to Ceren Ertan Yörük.

Additional information

Responsible editor: Erdal Tekin

Rights and permissions

Reprints and Permissions

About this article

Verify currency and authenticity via CrossMark

Cite this article

Ertan Yörük, C., Yörük, B.K. Alcohol consumption and risky sexual behavior among young adults: evidence from minimum legal drinking age laws. J Popul Econ 28, 133–157 (2015). https://doi.org/10.1007/s00148-014-0520-1

Download citation

  • Received:

  • Accepted:

  • Published:

  • Issue Date:

  • DOI: https://doi.org/10.1007/s00148-014-0520-1

Keywords

  • Alcohol consumption
  • Minimum legal drinking age
  • Risky sexual behavior
  • Sexual activity

JEL Classifications

  • I10
  • I18
  • J13