Skip to main content

Gone to war: have deployments increased divorces?


Owing to the armed conflicts in Iraq and Afghanistan, members of the US military have experienced very high rates of deployment overseas. Because military personnel have little to no control over their deployments, the military setting offers a unique opportunity to study the causal effect of major disruptions on marital dissolution. In this paper, we use longitudinal individual-level administrative data from 1999 to 2008 and find that an additional month in deployment increases the divorce hazard of military families, with females being more affected. A standard conceptual framework of marital formation and dissolution predicts a differential effect of these types of shocks depending on the degree to which they are anticipated when a couple gets married. Consistent with this prediction, we find a larger effect for couples married before 9/11, who clearly expected a lower risk of deployment than what they faced post 9/11.

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

Fig. 1
Fig. 2
Fig. 3
Fig. 4
Fig. 5


  1. Karney and Crown (2007) find that time spent in deployment decreased the probability of divorce among military families, concluding that military couples are resilient and that the benefits deployed service members receive might have compensated for the negative aspects of deployment for both spouses. However, their time frame is limited, as they only focus on individuals who entered the military between 2002 and 2005, married and deployed while married. Given that over this period the average time between military enlistment and first marriage was about 2 years, and deployments were about a year in many cases, the time over which service members are at the risk of divorce may be too short for the estimation of a deployment effect.

  2. It may be the case that when the gains of the military couple decrease, service members leave the military and become civilians in order to avoid divorce. We do not model explicitly this possibility. However, we find empirical evidence to the contrary, that is, individuals who reenlist in the military are more stable than the overall population of enlistees (see Section 6 and Appendix B in the ESM).

  3. Using unit deployment as an instrument for individual deployment, Lyle (2006) and Savych (2008) conduct a Hausman test, which rejects the hypothesis that the individual-level deployment is not orthogonal to the error.

  4. An extensive treatment of this framework is provided by Singer and Willett (1993) and Willett and Singer (1995).

  5. All military members have strong incentives to update their family status in DEERS online or in paper form, as their updated status determines the conditions under which they have access to military health care (TRICARE) and other military family benefits.

  6. To ensure that we include only couples that could form expectations regarding military service at the time of marriage formation, we exclude individuals already married at the time of enlistment.

  7. The list of hostile-fire pay areas as of March 2008 included Afghanistan, Algeria, certain areas of the Arabian Peninsula and adjacent sea areas, Azerbaijan, Bahrain, Burundi, Chad, Colombia, Cote d’Ivoire, Cuba (Guantanamo), Democratic Republic of Congo, Djibouti, East Timor, Egypt, Eritrea, Ethiopia, certain areas of Greece, Haiti, Indonesia, Iran, Iraq, Israel, Jordan, Kenya, Kosovo, Kuwait, Kyrgyzstan, Lebanon, Liberia, Malaysia, Montenegro, Oman, Pakistan, Philippines, Qatar, Rwanda, Saudi Arabia, Serbia, Somalia, Sudan, Syria, Tajikistan, certain areas of Turkey, Uganda, United Arab Emirates, Uzbekistan, and Yemen (Office of the Under Secretary of Defense [Comptroller] (2009), cited in Hosek and Martorell, 2009).

  8. As a result, we also exclude those individuals for whom all available observations correspond to deployment time. This is particularly the case for more recent entrants.

  9. We also estimated full sample models in which we interacted the deployment time variables with a female dummy variable. The predicted divorce hazards and cumulative divorce hazards that we generated for males and females using the estimates from this model are very similar to the ones presented in Fig. 5 below.

  10. Given that marriage markets are largely confined to one’s race (e.g., Qian 1997, 1998), and that marriage and divorce patterns may therefore vary by race, we also estimate our main empirical specification on subsamples of Whites, Blacks, and Hispanics, respectively (estimates not reported). Over our time frame, we find no differential pattern in the effect of deployment shocks by race.

  11. We investigated whether our main deployment effects are robust to alternative measures of tenure and rank in the military. This is a valid concern, because tenure and rank variables may subsume some of the impacts of deployments. Using tenure and rank at the time of marriage generates the same estimates as in Tables 3 and 4. Finally, the positive sign on the variable measuring time elapsed since last deployment and the negative sign on its squared term indicate that divorce is more likely to occur sooner after a return from deployment than later.

  12. The divorce hazard, also known as the “instantaneous” divorce probability, is defined as the probability of a couple to get divorced in the current period conditional on not having divorced in any of the previous periods.

  13. In order to offer a simple comparison across the divorce hazards of nondeploying individuals and individuals deploying for various durations, for Fig. 2, we assume that the deployment episode starts immediately after the service member gets married. We nevertheless experimented with other scenarios, in which individuals deploy some time later in their marriage. For the same value of time married and deployment duration, the postdeployment divorce hazard of an individual deploying later in marriage is very similar to the postdeployment divorce hazard of somebody who deploys immediately after marriage.

  14. The cumulative divorce hazard is calculated using the following formula: \(H_{t}=1-\prod \nolimits _{s=1}^{t}{({1-h_{s}})}\), where \(h_{s}\) represents the divorce hazard in period s after marriage. As all predicted divorce hazards shown in Fig. 2 have a concave profile, the cumulative divorce hazards have higher growth rates in the first few post-deployment periods and lower growth rates over the time married when the divorce hazard declines.

  15. All predictions presented in this paper are generated for nondual couples.

  16. For instance, only as of February 2013 were women allowed to be involved in direct combat.

  17. As our predictions are generated only for nondual couples, the gender difference in the effect of deployment is not driven by the fact that some families are dual military couples.

  18. The unit branches are Airborne Division/Brigade, Armor Cavalry, Army Corps, Aviation, Chaplain, Chemical, Civil Affairs, Composite Service, Engineers, Field Artillery, Finance, General, Heavy Division/Brigade, Infantry, Judge Advocate, Military Police, Military Intelligence, Ordnance, Psychological Operations, Public Information, Quartermaster, Security, Signal, Special Operations, Transportation and Training Division.

  19. First terms are usually 4-year contracts.

  20. In this estimation, “divorce” indicates that the service member divorced during the first term prior to the reenlistment decision date.

  21. The estimates of the hazard model, as well as those from the Heckman probit model predict a similar effect on the cumulative probability of divorce. In the sample-selection probit model, one additional month in deployment during the first term increases the probability of divorce by 0.75 %. This translates into a cumulative probability of divorce of 11.8 % for an enlisted service member who deployed for 6 months. Similarly, the discrete hazard model estimates predict that the cumulative probability of divorce for service members that experience 6-month deployments during the last 3 years of their first term is slightly smaller (10.8 %), but of the same order of magnitude.


  • Angrist JD, Johnson JH IV (2000) Effects of work-related absences on families: evidence from the Gulf War. Ind Labor Relat Rev 54:41–58

    Article  Google Scholar 

  • Army Regulation doc. 614-200 (2013) Assignments, details and transfers—enlisted assignments and utilization management., Accessed 5 July 2013

  • Becker GS (1973) A theory of marriage: part I. J Polit Econ 81:813–846

    Article  Google Scholar 

  • Becker GS (1974) A theory of marriage: part II. J Polit Econ 82:S11—S26

    Article  Google Scholar 

  • Becker GS, Landes EM, Michael RT (1977) An economic analysis of marital instability. J Polit Econ 85:1141–1187

    Article  Google Scholar 

  • Bergstrom TC (1997) A survey of theories of the family. In: Rosenzweig MR, Stark O (eds) Handbook of population and family economics. Kluwer, Norwell, pp 21–79

    Chapter  Google Scholar 

  • Castro CA (2008) 10 unpleasant facts about combat and what leaders can do to change them. Briefing at the Walter Reed Army Institute of Research, January 15, 2008. Accessed 1 March 2010

  • Cesur R, Sabia JJ, Tekin E (2013) The psychological costs of war: military combat and mental health. J Health Econ 32:51–65

    Article  Google Scholar 

  • Chandra A, Lara-Cinisomo S, Jaycox LH, Tanielian T, Han B, Burns RM, Ruder T (2011) Views from the homefront. RAND Corporation, Santa Monica

    Google Scholar 

  • Charles KK, Stephens M Jr (2004) Disability, job displacement and divorce. J Labor Econ 22:489–522

    Article  Google Scholar 

  • Department of Defense (1997) Occupational conversion manual, enlisted/officer/civilian. Office of the Assistant Secretary of Defense, Force Management & Personnel, DoD 1312.1-M

  • Doiron D, Mendolia S (2012) The impact of job loss on family dissolution. J Popul Econ 25:367–398

    Article  Google Scholar 

  • Engel RC, Gallagher LB, Lyle DS (2010) Military deployments and children’s academic achievement: evidence from Department of Defense education activity schools. Econ Educ Rev 29:73–82

    Article  Google Scholar 

  • Greene W (2010) Testing hypotheses about interaction terms in nonlinear models. Econ Lett 107:291–296

    Article  Google Scholar 

  • Hattiangadi AU, Ackerman D, Kimble TH, Quester AO (2004) Cost-Benefit analysis of lump sum bonuses for Zone A, Zone B, and Zone C reenlistments: final report. Center for Naval Analyses, Alexandria, Accessed 6 April 2009

  • Holley P, Yabiku S, Benin M (2006) The relationship between intelligence and divorce. J Fam Issues 27:1723–1748

    Article  Google Scholar 

  • Hosek JR, Martorell F (2009) How have deployments during the war on terrorism affected reenlistment? RAND Corporation, MG-873, Santa Monica

  • Hosek JR, Kavanagh J, Miller L (2006) How deployments affect service members. RAND Corporation, MG-432-RC, Santa Monica

  • How the Army Works (2011) A senior leader reference handbook. Accessed 5 July 2013

  • Karney B, Crown J (2007) Families under stress. An assessment of data, theory, and research on marriage and divorce in the military. RAND Corporation, Santa Monica

    Google Scholar 

  • Lyle DS (2006) Using military deployments and job assignments to estimate the effect of parental absences and household relocations on children’s academic achievement. J Labor Econ 24:319–350

    Article  Google Scholar 

  • Office of the Under Secretary of Defense (Comptroller) (2009) Special pay or bonus-qualified members extending duty at designated locations overseas. In: DoD financial management regulation vol 7A military pay policies and procedures—active duty and reserve pay. Accessed 1 March 2010

  • Qian Z (1997) Breaking the racial barriers: variations in interracial marriage between 1980 and 1990. Demography 34:263–276

    Article  Google Scholar 

  • Qian Z (1998) Changes in assortative mating: the impact of age and education, 1970–1990. Demography 35:279–292

    Article  Google Scholar 

  • Savych B (2008) Effects of deployments on spouses of military personnel. PhD dissertation, Pardee RAND Graduate School. RAND Corporation, Santa Monica

    Google Scholar 

  • Singer JD, Willett JB (1993) It’s about time: using discrete-time survival analysis to study duration and the timing of events. J Educ Behav Stat 18:155–195

    Article  Google Scholar 

  • Tanielian T, Jaycox LH (2008) Invisible wounds of war—psychological and cognitive injuries, their consequences, and services to assist recovery. RAND Corporation, Santa Monica

    Google Scholar 

  • Weiss Y, Willis R (1997) Match quality, new information, and marital dissolution. J Labor Economics 15:S293–S329

    Article  Google Scholar 

  • Willett JB, Singer JD (1995) It’s déjà vu all over again: using multiple-spell discrete-time survival analysis. J Educ Behav Stat 20:41–67

    Google Scholar 

Download references


We especially thank Beth Asch, Benjamin Karney, David Loughran, Linda G. Martin, Francisco Martorell, Juergen Maurer, Amalia R. Miller, Sonia Oreffice, John T. Warner and other colleagues at RAND and seminar participants at the Western Economic Association Annual Conference, US Naval Academy, University of Alicante and Aarhus University for their thoughtful suggestions. All remaining errors are our own.

Author information

Authors and Affiliations


Corresponding author

Correspondence to Sebastian Negrusa.

Additional information

Responsible editor: Erdal Tekin

Electronic supplementary material

Below is the link to the electronic supplementary material.

(DOCX 953 KB)

Rights and permissions

Reprints and Permissions

About this article

Cite this article

Negrusa, S., Negrusa, B. & Hosek, J. Gone to war: have deployments increased divorces?. J Popul Econ 27, 473–496 (2014).

Download citation

  • Received:

  • Accepted:

  • Published:

  • Issue Date:

  • DOI:


  • Divorce
  • Work-related absences
  • Unanticipated deployment shocks

JEL codes

  • J12
  • D10
  • C41