The effect of different enlistment ages on first-term attrition rate

Calhoun: The NPS Institutional Archive DSpace Repository Theses and Dissertations Thesis and Dissertation Collection 2014-03 The effect of different enlistment ages on first-term attrition rate Seker, Erdal Monterey, California: Naval Postgraduate School http://hdl.handle.net/10945/41442 Downloaded from NPS Archive: Calhoun

NAVAL POSTGRADUATE SCHOOL MONTEREY, CALIFORNIA THESIS THE EFFECT OF DIFFERENT ENLISTMENT AGES ON FIRST-TERM ATTRITION RATE by Erdal Seker and Emrah Ibis March 2014 Thesis Advisor: Co-Advisor: Ryan Sullivan Jeremy Arkes Approved for public release; distribution is unlimited

THIS PAGE INTENTIONALLY LEFT BLANK

REPORT DOCUMENTATION PAGE Form Approved OMB No. 0704 0188 Public reporting burden for this collection of information is estimated to average 1 hour per response, including the time for reviewing instruction, searching existing data sources, gathering and maintaining the data needed, and completing and reviewing the collection of information. Send comments regarding this burden estimate or any other aspect of this collection of information, including suggestions for reducing this burden, to Washington headquarters Services, Directorate for Information Operations and Reports, 1215 Jefferson Davis Highway, Suite 1204, Arlington, VA 22202-4302, and to the Office of Management and Budget, Paperwork Reduction Project (0704-0188) Washington, DC 20503. 1. AGENCY USE ONLY (Leave blan 2. REPORT DATE 3. REPORT TYPE AND DATES COVERED March 2014 Master s Thesis 4. TITLE AND SUBTITLE 5. FUNDING NUMBERS THE EFFECT OF DIFFERENT ENLISTMENT AGES ON FIRST-TERM ATTRITION RATE 6. AUTHOR(S) Erdal Seker and Emrah Ibis 7. PERFORMING ORGANIZATION NAME(S) AND ADDRESS(ES) Naval Postgraduate School Monterey, CA 93943-5000 9. SPONSORING /MONITORING AGENCY NAME(S) AND ADDRESS(ES) N/A 8. PERFORMING ORGANIZATION REPORT NUMBER 10. SPONSORING/MONITORING AGENCY REPORT NUMBER 11. SUPPLEMENTARY NOTES The views expressed in this thesis are those of the author and do not reflect the official policy or position of the Department of Defense or the U.S. Government. IRB Protocol number N/A. 12a. DISTRIBUTION / AVAILABILITY STATEMENT Approved for public release; distribution is unlimited 13. ABSTRACT: 12b. DISTRIBUTION CODE A This thesis analyzes the effects of different enlistment age on the first-term attrition for U.S. Army, Navy, Air Force, and Marine Corps enlisted personnel with non-prior service and prior service that attrited between fiscal years of 1995 and 2013. Two separate probit models were used to analyze attrition behavior. The first model was formed to analyze the effect of age at entry on the first-term attrition for four forces. The second model focuses on the attrition based on the character disorder and analyzes the effect of different enlistment ages on this attrition. Both attrition models are conditional and analyzing attrition behavior for four terms at six months, between 6 and 12 months, between 12 and 24 months, and between 24 and 45 months. The independent variables for the two models types included demographic variables, such as Black, White, Hispanic, other race, and unknown; education level; different enlistment age dummies between 18 and 42; female or male; and AFQT Cat. Unemployment rates by states were included in the regressions. The study concluded that enlistment ages do significantly affect the attrition of enlisted personnel. This effect varies across different time periods the first six months, the second six months, the second year, and 45-months and different forces. 14. SUBJECT TERMS Attrition, Enlistment age, Character disorder, Demographics 15. NUMBER OF PAGES 89 16. PRICE CODE 17. SECURITY CLASSIFICATION OF REPORT Unclassified 18. SECURITY CLASSIFICATION OF THIS PAGE Unclassified 19. SECURITY CLASSIFICATION OF ABSTRACT Unclassified 20. LIMITATION OF ABSTRACT NSN 7540 01-280-5500 Standard Form 298 (Rev. 2 89) Prescribed by ANSI Std. 239 18 UU i

THIS PAGE INTENTIONALLY LEFT BLANK ii

Approved for public release; distribution is unlimited THE EFFECT OF DIFFERENT ENLISTMENT AGES ON FIRST-TERM ATTRITION RATE Erdal Seker Captain, Turkish Army B.S., Turkish Army Academy, 2005 Emrah Ibis 1 st Lt, Turkish Army B.S., Turkish Army Academy, 2005 Submitted in partial fulfillment of the requirements for the degree of MASTER OF SCIENCE IN MANAGEMENT from the NAVAL POSTGRADUATE SCHOOL March 2014 Authors: Erdal Seker Emrah Ibis Approved by: Ryan Sullivan Thesis Advisor Jeremy Arkes Co Advisor William R. Gates Dean, Graduate School of Business and Public Policy iii

THIS PAGE INTENTIONALLY LEFT BLANK iv

ABSTRACT This thesis analyzes the effects of different enlistment age on the first-term attrition for U.S. Army, Navy, Air Force, and Marine Corps enlisted personnel with non-prior service and prior service that attrited between fiscal years of 1995 and 2013. Two separate probit models were used to analyze attrition behavior. The first model was formed to analyze the effect of age at entry on the first-term attrition for four forces. The second model focuses on the attrition based on the character disorder and analyzes the effect of different enlistment ages on this attrition. Both attrition models are conditional and analyzing attrition behavior for four terms at six months, between 6 and 12 months, between 12 and 24 months, and between 24 and 45 months. The independent variables for the two models types included demographic variables, such as Black, White, Hispanic, other race, and unknown; education level; different enlistment age dummies between 18 and 42; female or male; and AFQT Cat. Unemployment rates by states were included in the regressions. The study concluded that enlistment ages do significantly affect the attrition of enlisted personnel. This effect varies across different time periods the first six months, the second six months, the second year, and 45-months and different forces. v

THIS PAGE INTENTIONALLY LEFT BLANK vi

TABLE OF CONTENTS I. INTRODUCTION... 1 A. BACKGROUND... 2 1. Recruiting... 2 B. SCOPE AND METHODOLOGY... 4 C. ORGANIZATION... 5 II. LITERATURE REVIEW... 7 A. INTRODUCTION... 7 B. ATTRITION AND REENLISTMENT... 9 C. CONCLUSION... 13 III. THE ENLISTMENT AGE POLICY... 15 A. INTRODUCTION... 15 B. DIFFERENT ENLISTMENT AGE POLICIES IN THE HISTORY... 15 1. The Ottoman Empire... 16 2. The British Army... 16 3. The U.S. Army... 17 C. ENLISTMENT AGES FOR COUNTRIES TODAY... 18 IV. DATA... 21 A. INTRODUCTION... 21 B. DATA DESCRIPTION... 21 C. SAMPLE... 22 1. Sample Characteristics... 22 2. Descriptive Statistics... 24 3. Dependent Variables... 30 4. Independent Variables... 30 5. Enlistment Characteristics... 31 V. METHODOLOGY AND RESULTS... 35 A. INTRODUCTION... 35 B. RESEARCH DESIGN... 35 C. THEORETICAL FRAMEWORK... 36 1. Model Specification... 36 D. RESULTS... 36 1. THE ARMY... 55 2. THE NAVY... 56 3. THE AIR FORCE... 57 4. MARINE CORPS... 58 VI. CONCLUSION... 59 APPENDIX. AGE REQUIREMENTS AND RECRUITMENT... 61 A. MINIMUM AND MAXIMUM AGE REQUIREMENTS OF COUNTRIES HAVING ONLY VOLUNTARY SYSTEMS... 61 vii

B. MINIMUM AND MAXIMUM AGE REQUIREMENTS OF THE COUNTRIES HAVING ONLY COMPULSORY SYSTEMS... 64 C. MINIMUM AND MAXIMUM AGE REQUIREMENTS OF THE COUNTRIES HAVING BOTH RECRUITMENT SYSTEMS... 65 LIST OF REFERENCES... 67 INITIAL DISTRIBUTION LIST... 71 viii

LIST OF FIGURES Figure 1. Prior Service Enlistment Rates for Services (1997 2009)... 22 Figure 2. Male Prior Service Enlistment Age Distribution... 23 Figure 3. Female Prior Service Enlistment Age Distribution... 23 ix

THIS PAGE INTENTIONALLY LEFT BLANK x

LIST OF TABLES Table 1. Enlistment Ages around the World... 8 Table 2. Different Enlistment Ages and Numbers of the Countries... 18 Table 3. Enlistment Age Distribution... 19 Table 4. Non-Prior Service (NPS) Active Component Enlisted Accessions Percent Married by Service FYs 1995 2011... 25 Table 5. Descriptive Statistics of Army and Navy... 26 Table 6. Descriptive Statistics of Air Force and Marine Corps... 28 Table 7. Army Attrition Rates by Recruit Characteristics... 39 Table 8. Navy Attrition Rates by Recruit Characteristics... 43 Table 9. Air Force Attrition Rates by Recruit Characteristics... 47 Table 10. Marine Corps Attrition Rates by Recruit Characteristics... 51 Table 11. Countries Having Only Voluntary System. Adapted from The world factbook, 2013. Retrieved from https://www.cia.gov/library/publications/the-worldfactbook/fields/print_2024.html... 61 Table 12. Counties Having Only Compulsory System... 64 Table 13. Countries Having both Compulsory and Voluntary System... 65 xi

THIS PAGE INTENTIONALLY LEFT BLANK xii

LIST OF ACRONYMS AND ABBREVIATIONS AFQT BLPM CAT COS DEP DMDC DOD GED JROTC ISC MOSs NPS OLS PS SRBs Armed Forces Qualification Test Binary Linear Probability Model category Character of Service Delayed Entry Program The Defense Manpower Data Center Department of Defense General Education Development Junior Reserve Officers Training Corps Inter-service Separation Code military occupational specialties non-prior service Ordinary Least Squares prior service Selective Re-enlistment Bonus xiii

THIS PAGE INTENTIONALLY LEFT BLANK xiv

ACKNOWLEDGMENTS We sincerely and gratefully thank Professor Ryan Sullivan and Professor Jeremy Arkes, our thesis advisors, for their patience, guidance, and unrelenting assistance throughout this process. With their professional insights they helped us overcome many problems in completing this thesis. Without their assistance this thesis would have not been completed. Especially, we would like to thank our beautiful wives, and our lovely families, for their constant love, enduring support, and sacrifice. Lastly, we would like to thank to the Turkish Army for giving us the opportunity to study at Naval Postgraduate School. xv

THIS PAGE INTENTIONALLY LEFT BLANK xvi

I. INTRODUCTION United States Armed Forces have recruited a great numbers of enlistees for each year since the beginning of the all-volunteer force in 1973. Although these enlistees are required to stay in the military until the end of their contracts, it is found that many do not fulfill this commitment. The goal of this thesis is to determine the effects of different enlistment ages on first-term attrition rates in the United States Marine Corps, Air Force, Navy, and Army enlistees between 1995 and 2013. In addition, this study focuses on developing an empirical model to calculate the real effect of enlistment age on attrition. The United States Armed Forces have engaged in many more operations worldwide since the September 11, 2001, attacks on the Pentagon and World Trade Center. Operation Iraqi Freedom, Operation Noble Eagle, Operation Enduring Freedom are the three vital operations of the critical period that has increased the manpower needs of the Services. Additionally, state annual unemployment rates that are acquired by The Bureau of Labor Statistics are examined to calculate the true effect of enlistment age on first-term attrition rates. Many variables are used in our analysis for four branches of the United States Armed Forces, including demographics, education, Armed Forces Qualification Test (AFQT) category variables, the unemployment rate variable and regional differences. Our study also aims to determine the optimum enlistment age whether there is, thus it will help recruiters to select the most suitable candidates from the large applicant pools for the forces and will reduce the training and recruiting cost. In this paper, we study a new approach that intends to reduce recruitment cost of the U.S. Armed Forces by focusing on the enlistment age. 1

A. BACKGROUND 1. Recruiting According to DOD officials, recruiting is the military services ability to bring new members into the military to carry out essential tasks in the near term and to begin creating a sufficient pool of entry-level personnel to develop into future mid-level and upper-level military leaders. Since the military has to choose recruits who are the best candidates among peer groups, they invest large amounts of money to recruit and keep qualified applicants. United States Armed Forces have had difficulty in meeting the manpower requirements to support the military operations launched against terrorism after September 11 (GAO, 2005). Although the Navy has been confronted with some contingencies, recruiting goals succeeded during the fiscal years 2000 2009. However, Armed Forces recruiting policy must continue to focus on developing a more skillful and dynamic applicant pool to support operations. Recruiting competent enlistees is vital to national security, but is a costly process. For example, each year the U.S. Department of Defense (DOD) must replace approximately 11 percent of military personnel (about 160,000 troops), due to normal workforce turnover. In this endeavor, they are spending approximately $11,000 to recruit each new soldier (Department of Defense, 2013). The DOD defines attrition as the failure of an enlistee to complete his or her contractual obligation. Attrition is one of the most serious and costly personnel problems faced by the U.S. military. Before basic training begins, recruits are obliged to take an enlistment oath and sign a 2 6 year contract to serve for one of the services. Some attrition occurs during basic training, which lasts from 6 to 12 weeks, depending upon the service. Some attrition occurs during initial skill training, which can last from a few weeks to more than one year, depending upon the enlistee s occupation. Finally, some attrition occurs after enlistees have reported to their first duty assignments. By the six-month point in an enlistee s first term, most have completed both basic and initial skill 2

training, and have been assigned to their first duty stations though this is not the case for enlistees whose occupations require longer and more extensive training. Most attrition occurs during basic training or the first six months of active duty service (GAO, 2000). Enlisted attrition involves direct costs (training costs) and indirect costs (damage to force stability). The first term of enlistment is usually a 4-year term, but some terms may be two-six years in length (GAO, 2000). Early separation of enlistees can happen at any time during the first term because of physical, medical, and performance problems, or fraudulent enlistment. Cost of recruiting new service members averages about $11,000 for each recruit, combined with an average cost of initial entry training at $35,000. As more than 200,000 of America s youth are recruited for active military service each year, the DOD s investment in military recruit accessions and training is enormous. The latest available data from the DOD resources explains that the Navy has the highest first-term attrition rates with 13.6 percent in all of the active duty enlistees. The Army and Marines follow with 13.5 percent and 11.7 percentage points. In contrast, the Air Force has the lowest first-term attrition rate with 8.8 percent among the active duty enlistees. This study initially focused on that the fact that attrition and recruiting are parts of an unbreakable chain, which cannot be thought of separately. In their previous studies, researchers have discovered that those who enlisted before turning 18-years-of-age were more likely than any other age group to attrite. This study will try to investigate recruiting and attrition as a whole, and determine whether enlistment age should be taken in consideration in recruiting. 3

2. Research Questions The primary questions of this study are focused on the effect of enlistment ages on the first-term attrition rates for the Army, Navy, Marine Corps, and Air Force. 1. What is the optimal enlistment age to increase the productivity in the Navy, Army, Air Force, Marine Corps? 2. Does the age-effect on attrition differ for men and women in the firstterm? 3. Does the age-effect on attrition differ for race categories in the firstterm? B. SCOPE AND METHODOLOGY While this study involves some overall historical attrition data, it also focuses on the detailed analysis of enlistees who entered the services between Fiscal Year 1994 and Fiscal Year 2013. This study focuses on Army, Marine Corps, Navy, and Air Force enlistees were screened from Fiscal Years 1995 to 2013.The scope of the thesis includes a review of studies about attrition without the effect of optimal enlistment age, an in-depth review of the effect of optimal enlistment age on of this section, and the evaluation of this effect in short-term and long-term attrition for U.S. Armed Forces. The thesis concludes with a recommendation for the current Recruitment System of U.S. Armed Forces and provides attrition data for enlistees by educational background, AFQT score, and age at the time of enlistment, gender, and race/ethnic group. To determine historical first-term attrition rates, we obtained the data from the Defense Manpower Data Center (DMDC), whose primary purpose is to support the management needs of the Office of the Under Secretary of Defense for Personnel and Readiness. The data covered all enlistees with prior service who entered the military from Fiscal Year 1994 through 2013 and included the gender, educational background, age at enlistment, race, AFQT category, marital status, and separation code (for those who left the services). Because the majority of first-term contracts are for four 4

years, we made our calculation at the 48 months point to include the most recent data available (Fiscal Year 2013). Our method does not include the attrition of enlistees with 5- or 6-year contracts who were separated in their final one to two years of service. C. ORGANIZATION The thesis consists of six chapters. Chapter I is an introduction and brief information about attrition related to the United States Armed Forces. Chapter II is a literature review of previous attrition studies, and any modeling studies that focus on the age variable. Chapter III is the explanation of all the variables that are discussed in the analysis. Chapter IV provides an analysis of the data with summary statistics. Chapter V explains methodology and presents the findings of the probit models to predict the effects of mainly enlistment age variables and other variables on attrition. Chapter VI concludes the thesis with a conclusion and recommendations from the findings. 5

THIS PAGE INTENTIONALLY LEFT BLANK 6

II. LITERATURE REVIEW A. INTRODUCTION Eighteen is thought to be the optimal draft age in most countries to use the enlistee s power, quick reaction, and tendency to learn. However, family environment and the lack of life experience can negatively affect the success of armed forces. Furthermore, politicians still discuss the morality and suitability of the early enlistment age. Enlistment age varies among countries based on whether their armies are AFV. Most of the countries continue to accept enlistees at the age of 18. The lowest enlistment age is 16, with parental consent, in the following countries: Bangladesh, Egypt, El Salvador, Guinea-Bissau, Guyana, India, Iran, Korea, North, Mexico, Pakistan, Papua New Guinea, Tonga, United Kingdom, and Zambia. Within a specific country, minimum enlistment age differs according to forces and gender (CIA, 2013). Table 1 demonstrates some examples of different enlistment age policies around the world. 7

Table 1. Enlistment Ages around the World Australia Canada China France Germany Russia Turkey United Kingdom United States 17 years of age for voluntary military service (with parental consent); no conscription; women allowed to serve in most combat roles, except the Army special forces (2013) 17 years of age for voluntary male and female military service (with parental consent); Canadian citizenship or permanent residence status required; maximum 34 years of age; service obligation 3 9 years (2012) 18-24 years of age for selective compulsory military service, with a 2-year service obligation; no minimum age for voluntary service (all officers are volunteers); 18 19 years of age for women high school graduates who meet requirements for specific military jobs. 17-40 years of age for male and female voluntary military service (with parental consent); no conscription; 1-year service obligation. 17-23 years of age for male and female voluntary military service; conscription ended 1 July 2011; service obligation 8 23 months or 12 years. 18-27 years of age for compulsory or voluntary military service; males are registered for the draft at 17 years of age; service obligation is 1 year (conscripts can only be sent to combat zones after 6 months of training). 21-41 years of age for male compulsory military service; 18 years of age for voluntary service; 12 months conscript obligation for non-university graduates, conscripts are called to register at age 20, for service at 21 16-33 years of age (officers 17 28) for voluntary military service (with parental consent under 18); no conscription; women serve in military services, but are excluded from ground combat positions 18 years of age (17 years of age with parental consent) for male and female voluntary service; no conscription; maximum enlistment age 42 (Army), 27 (Air Force), 34 (Navy), 28 (Marines); service obligation 8 years, including 2 5 years active duty (Army), 2 years active (Navy), 4 years active (Air Force, Marines) Adapted from The World Factbook, 2013. Retrieved from https://www.cia.gov/library/publications/the-worldfactbook/fields/print_2024.html 8

B. ATTRITION AND REENLISTMENT Although most studies include the effect of average age on attrition, promotion, and reenlistment, they have not defined the effect of enlistees age during the contract year on attrition, reenlistment, and promotion. Golan, Greene, and Perloff (2010) conducted an important study about U.S. Navy promotion and retention by race and gender. They determined that promotion rates depend on an individual s characteristics, war, economic conditions and factors that Navy policymakers can control. Their primary focus was on the question of whether Navy promotion rates differ across gender and race, and whether the Navy can alter its promotion and retention policies to maintain their sailors. They estimated a bivariate probit model to examine the Navy s promotion, and an individual s reenlistment decision. They developed two equations, one of which directly related to individual s promotion rates based on ability and individual performance, and the other one represented the individual s decision to reenlist or to leave based on their promotion. Although they have used many variables to define the demographics and individual characteristics, they excluded the age variable in their research. Greene and Perloff study determined that the probability of promotion differs across races, despite all the measures taken by the Navy. In addition, it was determined that promotion rates vary for men and women across pay grades. If civilian economic conditions improve, it will be more difficult to keep personnel in their current positions. They recommended that the Navy should raise promotion rates, pay higher selective re-enlistment bonuses (SRBs) to continue to increase the Navy s relative wage. Buddin (2005) examined the effect of recruiting practices and recruit characteristics against the first-term success in his study. He developed a model to estimate the success of enlistees and recruitment programs by using first-term attrition, promotion, and reenlistment rates between the years of 1995 2001. 9

Because the primary focus of the study was the first-term success, the recent data was not used. His research examined recruit progress at various steps. In his study, he used three different models; attrition, promotion and reenlistment models. In the attrition model, he used age at time of contract variable and found a statistically significant effect on attrition. Buddin s model tried to predict probability of a recruit in the Army to reach the E5 rank in his/her first-term service. Among other things, the control variables included race, gender, military occupational specialties (MOSs), AFQT scores, and educational background. The study found that Blacks, older recruits, those who had completed some college credits, and those with higher AFQT scores, had higher probabilities of promotion. On the other hand, being female or Asian, and having obtained a general education development (GED) instead of a high school diploma had a negative effect on promotion. Reenlistment is another potential indicator of the success in the armed forces. Roy (2007) used enlisted data from cohorts between 1994 and 2002 to analyze reenlistment. He used econometric analysis and found that participation in JROTC increases the probability of reenlistment. Females who participate in JROTC reenlist at higher rates than both males and females that participate in no youth programs. Although his study provided beneficial results about the effect of JROTC on reenlistment, promotion, and attrition, he did not mention the effect of enlistment age, nor did he include the age variable into the model. Three basic studies have analyzed variables that affect and individual s reenlistment decisions. One study looks at the effects of pay, selective reenlistment bonuses (SRB)/incentive pays, pay grade, and marriage on a military member s decisions to reenlist and remain in the armed forces. The second group of studies looks at aptitude test scores, race, gender, and educational background of recruits prior to entering the armed forces. The third and last group of studies analyzes the effects of youth programs (e.g., JROTC) on an individual s career decisions in and out of the armed forces. 10

The study that is the most similar to our work is Greenamyer s Master s Thesis (2009). His focus was on the effect of AFQT percentile scores and age on the Navy Delayed Entry Program (DEP) attrition. Men and women were analyzed separately, because of the differences in attrition by gender and sample size. He examined the effect of the enlistment age on DEP attrition, by using a Binary Linear Probability Model (BLPM). Greenamyer categorized the enlistees ages into six groups: under 18, ages 18 20, ages 21 23, ages 24 26, ages 27 29, and over 30. He found that enlistees under 18-years-old were less likely to attrite, whereas the other enlistees in the different age cohorts are more likely to leave the service. In his thesis, Jag (2009) analyzed the effect of selected demographic characteristics on first-term enlisted attrition from the U.S. Navy. The characteristics included age, marital status, dependency status, gender, race, AFQT score, and education credentials. Probit regression models were used to determine the true effect of demographic variables on first-term attrition. The unrestricted model constructed by Jag indicated that increasing age was correlated with a higher likelihood of attrition. On the other hand, the restricted model showed that this effect was reversed once the first 90-days of service were controlled. Older recruits, who made it through initial training, were less likely to attrite than their younger counterparts. Mehay and Arkes (2013) conducted a study that investigated the effects of home-state unemployment rates on attrition behavior of Navy enlistees, for successive career windows during the first-term of service. Analysis included the first six months, the second six months, the second year, and the third year of service. The results indicated that attrition is negatively associated with changes in the local unemployment rate during the first three career windows covering two years of service. 11

Their data set was drawn from Navy enlisted master files, and contained demographic information and service history on active-duty personnel who entered the Navy between 1999 and 2009. In their attrition model, they used the entry ages as demographic variables, in addition to economic variables. As a result, they learned that in the first six months and one year period, the older the enlistees in six month, the more likely to attrite. However, result was just the opposite for the other three terms. Older personnel were less likely to attrite in second six month, second year and third year periods. Perger (2011) analyzed the effects of combat exposure on reenlistment attrition in his Master s Thesis. This study attempted to find a correlation between the battlefield experience and the retention and attrition decisions of first-term enlistees. Probit models were used to estimate effects. Age was also included in his models to estimate the true effect. Consequently, although the age variable has shown an insignificant effect on attrition and reenlistment for the Air Force, and the reenlistment model for Marine Corps and Army, it has had a significant effect on Navy attrition and reenlistment models, along with Marine Corps and Army attrition models. Cox (2003) analyzes relationship between the size of the enlistment bonus offered for a recruit and the likelihood of the recruits attrition. His study focuses on a subset of first-term enlistees with five- and six-year initial contract. The study estimates separate models for each year between 1993 and 1997. In his cross sectional analyses, he finds that age at the service entry is a statistically significant covariate. According to results of his study enlistees with five year contract are more likely to attrite when they get older one more year for the first year and less likely to attrite for the second year. The same effects are seen for the enlistees with six year contract in his survival analysis. However, Cox does not include six-month or forty-five month attrition in his analyses, and he only focuses on Navy enlistees. 12

Moore and Reese (2001) track the sailors from street to fleet analyzing the factors which influence their attrition behavior during initial skill training. Their study indicates that one of the determinants of attrition is age at the entry. Logit models are used in their analyses to determine the important predictors of the attrition. Results for his study about boot attrition and post boot attrition show that age at the entry is statistically significant. While it has a positive marginal effect, on average holding all else constant on boot camp attrition, it has negative marginal effect on post boot camp attrition. Although this study shows a similarity to our study in terms of including age at the entry in the Navy, it differs in not comparing enlistment ages between themselves for all forces (CNA, 2001). Wenger and Hodari (2004) made a research to find out how non-cognitive factors affect attrition. They primarily focused on 36 months attrition rate by using large service-wide survey. According to this study, non-cognitive factors such as education credentials, smoking, marital status, expelling from a school have an effect on attrition. They also include the accession age in their regressions and find statistically significant results. Namely, all recruits who enlist before turning 18 have higher attrition rates, but enlistees with non-high school degree at age 20 or more have lower attrition rate than their younger colleagues. Their study does not include the age variable more than 23 separately. Furthermore, the attrition models are not conditional (CNA, 2004). C. CONCLUSION Though there are many parameters measuring the success in organizations, we have chosen attrition as the indicator of productivity in our study. We searched a great number of studies about first-term attrition rate; however, few studies taking the effect of entry age into account, especially the age-at-the-contract year (enlistment age) on productivity. In our study, we will focus on the effect of enlistment age on attrition in the first term for four armed forces. 13

This study focuses on providing beneficial information to the armed forces command to use in future recruitment systems from the enlistment age perspective. 14

III. THE ENLISTMENT AGE POLICY A. INTRODUCTION In this chapter, we will provide information about different enlistment age policies used in the past and the present. A historical part of this chapter focuses on three major armies: The Ottoman Army, The British Army, and The United States Army. Current enlistment age systems are explained in tables and summarized in this chapter. During the study, 175 countries were examined. A detailed table contains results is displayed in the Appendix. B. DIFFERENT ENLISTMENT AGE POLICIES IN THE HISTORY Countries have armies for both peace and war time. Due to the probability of war, countries increase or decrease their military strength. There are different ways to improve the strength. One is to add more manpower. To raise the number of soldiers, countries usually modify their enlistment policies and decrease the enlistment age to enlarge applicant population. Following are historical examples for different enlistment policies: Between AD 600 1453 (Byzantine Empire), the minimum age for recruitment was 18 and the maximum was 40 (Haldon, 2002). During Napoleonic Wars in 1764, Russia prohibited enlisting any youth before age 15 (Mikaberidze, 2005). On June 14, 1775, the second continental congress formed the Continental Army, raising 22,000 troops from the Boston area, and 5,000 from New York. Of these troops, most had little training or military experience, and the minimum enlistment age was 16 (U.S. History, Volume I, 2013). In 1939, enlistment age limits for Australian Imperial Forces were 20 35, and in 1943, limits were 18 40. For the militia, the limits were extended to 18 60. This is a great example of how the enlistment age policy was being revised to increase the number of soldiers by decreasing minimum enlistment age, and increasing maximum age (Mitchell, 2013). 15

1. The Ottoman Empire The Ottoman Army was one of the largest and the most powerful armies of its time. In order to sustain to this power, the Ottoman Army used different recruitment systems. One of the systems, the Pencik system, was used from 1363 until the 15th century. According to this system, the Army enlisted 20 percent of the prisoners of war (POWs) between the ages of 10 to 17-years-old. After the 15th century, Ottoman used the Devshirme System until the 17th century. Young Christian boys between 10 and 20 years old were enlisted accordingly. The boys were usually enlisted from Balkans; most of them accepted Islam. After enacting the military service law in 1826, healthy and physically strong males between 15 and 40-years old were enlisted (Sakin, 2011). In 1846, the service law was changed, consequently males between 20- to 25-years-old were enlisted. In accordance with 1886 and 1908 laws, 18-year-old males were obligated to do compulsory military service (Yıldız, 2009). Before WWI, the enlistment age was 18, due to increased manpower needs in the Army. Prior to this, the enlistment age was 20. During WWI, males between 18 and 45 were enlisted. In Dardanelles War, the need for soldiers was enormous, which prompted 15-year-old males to be enlisted to war. After the Independence War in 1922, the need for soldiers was diminished. Hence, a new military service law was enacted in 1927, which increased the enlistment age to 20. The maximum compulsory military service age was 46 (Aslan, 2005). 2. The British Army At the beginning of the 18th century, the strength of the British Army was cut off, and stood at 7,000 troops at home. After the Treaty of Ryswick, 14,000 soldiers were based overseas, with recruits ranging from 17 to 50 years of age. This situation remained the same until the start of WWI (Neuman & Brenner, 2001). 16

At the start of 1914, the British Army had a reported strength of 710,000 men including reserves of which 247,432 were regular troops and 80,000 regular troops formed as the British Expeditionary Force. The Military Service Bill, which determined that single men between 18 and 41 were liable to be called up for service, was put into effect. By the end of WWI, over five million men nearly a quarter of the total male population of the United Kingdom of Great Britain and Ireland had registered for the military (Chandler & Beckett, 2003). In early 1939, Conscription was accepted to meet the threat of Germany, with the Military Training Act of April 27, 1939. This act required all men aged 20 and 21 to take six months of military training. This act was extended to include all fit men between the ages of 18 and 41 as the war continued (BBC, 2003). 3. The U.S. Army Prior to 1862 in America, white men were required to attend local militia units, while combat duty was always voluntary. During the Civil War, both the North and South implemented the conscription, in order to meet their increasing military need. On April 16, 1862, the Confederate Congress passed an act requiring military service for three years from all males aged 18 35. In 1863, the Law of Congress extended the age range from 20 to 45 (Poinsett, 1840). During WWI, Woodrow Wilson started to implement the Conscription after enacting the Selective Service Act of 1917. This law had determined that men between the ages of 21 and 31 would be called up for military service, and prohibited all forms of bounties, substitutions, or purchase of exemptions (Moore, 1924). 17

C. ENLISTMENT AGES FOR COUNTRIES TODAY In this section, we provide some information about the current recruitment ages all over the World. In Table 2, different enlistment ages of the countries having voluntary and compulsory service and their frequency are listed. In Table 3, the enlistment age distribution among sixty counties both having compulsory and voluntary service is shown. Table 2. Different Enlistment Ages and Numbers of the Countries VOLUNTARY SERVICE (165 COUNTRIES) COMPULSORY SERVICE (70 COUNTRIES) ENLISTMENT AGE NUMBERS OF COUNTRIES PERCENTAGE 16 14 8% 16,5 1 1% 17 23 14% 17,5 2 1% 18 119 72% 19 3 2% 20 3 2% 21 0 0% 17 2 3% 18 56 80% 19 4 6% 20 6 9% 21 2 3% Adapted from The World Factbook, 2013. Retrieved from https://www.cia.gov/library/publications/the-worldfactbook/fields/print_2024.html 18

Table 3. Enlistment Age Distribution COMPULSORY & VOLUNTARY SERVICE SYSTEMS (60 COUNTRIES) COMPULSORY SYSTEM VOLUNTARY SYSTEM ENLISTMENT AGE NUMBERS OF COUNTRIES PERCENTAGE 16 6 10% 16,5 1 2% 17 8 13% 18 43 72% 19 1 2% 20 1 2% 21 0 0% 16 0 0% 16,5 0 0% 17 0 0% 18 50 83% 19 4 7% 20 5 8% 21 1 2% Adapted from The World Factbook, 2013. Retrieved from https://www.cia.gov/library/publications/the-worldfactbook/fields/print_2024.html There are three recruitment systems among 175 countries in the World. One of them is voluntary, another one is compulsory, and the last one is the system which contains both compulsory and voluntary recruitment at the same time. Sixty percent of the countries have voluntary systems, 6 percent have compulsory systems, and 34 percent have both systems. As seen in Table 2, 119 countries having voluntary systems are recruiting at the age of 18, which compose 72 percent of the voluntary systems, among 165 countries. Fifty-six countries have compulsory services and are recruiting at the age of 18, which constitutes 80 percent of the compulsory system among 70 countries (Central Intelligence Agency, 2013). 19

THIS PAGE INTENTIONALLY LEFT BLANK 20

IV. DATA A. INTRODUCTION This chapter discusses the data source and defines the dependent and independent variables used in the regression. B. DATA DESCRIPTION We acquired the data for our study from the Defense Manpower Data Center (DMDC). The data belongs to soldiers and enlistees who have been followed between the years of 1995 and 2013. It includes cohorts from three different year periods. Cohort 1 covers the data of Fiscal Years 1995 and 1999. Cohort 2 is tracked between 2000 and 2012; Cohort 3 was monitored in 2013. These three files have been merged into Stata (a kind of statistical program) to analyze the continuous first-term attrition decisions of the enlistees. The annual state unemployment rate was added to increase the accuracy of the models. The first raw data file involved 3,545,241 observations, and 17 data variables. The data elements consisted of personal demographics and military background information, such as SSN Number, date of birth, gender, race, accession date, separation date, education level, marital status, character of service (COS), Inter-service separation Code (ISC), AFQT score and category, military service, and ethnicity. Records having separation dates earlier than the accession date were dropped, and enlistees older than 42-years-old, and younger than 17-years-old. Additionally, third gender categories and Coast Guard service H categories were extracted from our data. Because we wanted to only follow attrition decisions of enlistees, 221,983 records belonging to officers were removed from the sample. Before studying our model, in the first phase, 368,137 observations were deleted, in order to clean and harmonize our data. In the second phase, we balanced our data to determine the six-month attrition. In 2013, enlistees having accession dates after March 30, 2013, and a missing separation date (because these samples do not have enough time to follow all 21

possible outcomes) were struck out. For the next phases we repeated the same extraction for 12-month attrition rates (September 30, 2012), 24-month attrition rates (September 30, 2011), and 45-month attrition rates (December 30, 2009). Ultimately, we calculated 3,177,104 for the six-month regression. C. SAMPLE 1. Sample Characteristics Our data includes both non-prior service (NPS) and prior service (PS) enlistees, that s why the average values of the characteristics, especially age and marital status, represent the average values of only NPS enlistees ( Bureau of Labor Statistics,2011). Figure 1 shows prior service accession rates between the fiscal years of 1997 and 2009 for each service. Regarding to these statistics, on average, 5 percent of the enlistees in DOD have prior service, while this ratio has an uneven distribution among four forces. For instance, it is 10 percent for the Army, 2 percent for the Navy and Marine Corps, and the Air Force has the lowest prior service enlistment, with 1 percent. Adapted from Population Representation in the Military Services. Retrieved from http://prhome.defense.gov Figure 1. Prior Service Enlistment Rates for Services (1997 2009) 22

In addition to the statistics mentioned previously, Figure 2 and Figure 3 show the age distribution of the prior service enlistees by gender. In Figure 2 and Figure 3, most of the prior service enlisted personnel, in all services, are 25 years and above at entry. These statistics support our results. Male Age Distribution Adapted from Population Representation in the Military Services. Retrieved from http://prhome.defense.gov Figure 2. Male Prior Service Enlistment Age Distribution Female Age Distribution Adapted from Population Representation in the Military Services. Retrieved from http://prhome.defense.gov Figure 3. Female Prior Service Enlistment Age Distribution 23

2. Descriptive Statistics Table 4 and Table 5 provide summary statistics of the enlisted attrition dataset used to estimate the parameters for two models. As shown in both tables, the predominant enlistment age is 18, which is true for all forces. This includes 18 percent for the Army and Navy, and 29 percent for Air Force and Marine Corps. The average age of an entrant for the Army is 22.4, for Navy 23.96, for Air Force 19.8, for Marine Corps, 19.99. Female enlistees are accounted for 16 percent for Army, 14 percent for Navy, 23 percent for Air Force, and 7 percent for Marine Corps. The marriage rate for the Army is 28.8 percent, 30.8 percent for the Navy 30.8, 10.4 for the Air Force, and 10.8 percent for the Marine Corps. As can be seen in Table 3, these numbers are higher than the expected average number in that people are monitored throughout their military service. Thus, the marriage rate should be higher, since people tend to get married as their career progresses. 24

Table 4. Non-Prior Service (NPS) Active Component Enlisted Accessions Percent Married by Service FYs 1995 2011 YEARS ARMY NAVY MARINE CORPS AIR FORCE 1995 14.5 4.9 4.1 11 1996 14.6 5.1 4.2 10.7 1997 15.9 5.2 4.2 10 1998 14.4 5.2 4.3 9.3 1999 14 6.1 3.9 9.7 2000 13.2 6 3.3 8.6 2001 12.9 5.7 3.1 9.2 2002 13.8 6 3 9.2 2003 13.6 5.5 4.6 8.6 2004 13.5 5.2 2.4 8.4 2005 14.2 5.1 2.6 8.4 2006 14.8 5.7 2.7 8.6 2007 16.2 6.4 3 8.9 2008 22.9 1.7 3.2 10.7 2009 18.3 2.1 3.3 11.7 2010 17.7 4.5 2.6 11.2 2011 16 4.9 2.1 10.3 Average % 15.32 5.02 3.33 9.68 Adapted from Population Representation in the Military Services. Retrieved from http://prhome.defense.gov 25

Table 5. Descriptive Statistics of Army and Navy ARMY(n=1,543,047) 26 NAVY(n=458,718) Mean Std. Dev. Mean Std. Dev. Overall attrition in 6 months n=1,528,537 0.0675 0.2509 n=458,099 0.0995 0.2994 Overall attrition in 12 months n=1,399,181 0.0466 0.2109 n=411,653 0.0374 0.1896 Overall attrition in 24 months n=1,307,633 0.0874 0.2825 n=391,381 0.0675 0.2509 Overall attrition in 45 months n=921,229 0.2895 0.4535 n=357,242 0.1369 0.3438 Character disorder attrition in 6 n=1,528,537 0.0008 0.0275 n=458,099 0.0009 0.0301 months Character disorder attrition in 12 n=1,399,181 0.0034 0.0584 n=411,653 0.0029 0.0538 months Character disorder attrition in 24 n=1,307,633 0.0131 0.1136 n=391,381 0.0077 0.0875 months Character disorder attrition in n=921,229 0.0270 0.1620 n=357,242 0.0146 0.1198 45months ENLISTMENT AGES age17 0.0750 0.2634 0.0339 0.1809 age18 0.1812 0.3852 0.1894 0.3918 age19 0.1397 0.3467 0.1339 0.3405 age20 0.0976 0.2968 0.0799 0.2711 age21 0.0791 0.2699 0.0583 0.2343 age22 0.0656 0.2475 0.0551 0.2281 age23 0.0548 0.2275 0.0481 0.2141 age24 0.0458 0.2091 0.0417 0.1999 age25 0.0383 0.1919 0.0335 0.1798 age26_27 0.0590 0.2355 0.0565 0.2309 age28_29 0.0438 0.2045 0.0522 0.2223 age30_34 0.0750 0.2633 0.1234 0.3289 age35_42 0.0453 0.2079 0.0942 0.2922 age 22.4504 5.3227 23.9633 6.3877 TIS 1669.4400 1373.1710 2092.0930 1614.1280 female 0.1673 0.3732 0.1412 0.3482 cat1 0.0528 0.2237 0.0470 0.2117 cat2 0.2615 0.4395 0.3322 0.4710 cat3 0.4861 0.4998 0.5124 0.4998 cat4 0.0255 0.1577 0.0271 0.1622 cat5 0.0000 0.0053 0.0000 0.0026 cat_unknown 0.1740 0.3791 0.0813 0.2733

ARMY(n=1,543,047) NAVY(n=458,718) Mean Std. Dev. Mean Std. Dev. MARITAL STATUS other_marital status 0.0248 0.1556 0.0003 0.0161 married 0.2883 0.4530 0.3081 0.4617 single 0.6850 0.4645 0.6868 0.4638 mar_stat_unknown 0.0018 0.0425 0.0048 0.0689 EDUC_LEVEL 21.8727 5.4897 20.9491 4.2068 some_college (no diploma) 0.0402 0.1965 0.0154 0.1230 some high school(no diploma) 0.0195 0.1381 0.0251 0.1566 High school degree 0.7638 0.4248 0.8994 0.3008 education unknown 0.0513 0.2205 0.0101 0.1001 alternate education 0.0761 0.2652 0.0302 0.1710 college_degree 0.0368 0.1882 0.0181 0.1332 doctorate,master and above 0.0123 0.1104 0.0018 0.0422 RACE black 0.2320 0.4221 0.2087 0.4064 white 0.6898 0.4626 0.6939 0.4609 race_unknown 0.0108 0.1035 0.0060 0.0775 race_others 0.0233 0.1507 0.0103 0.1008 hispanic 0.0441 0.2053 0.0811 0.2730 hispan_95_99 0.0441 0.2053 0.0811 0.2730 CHARACTER OF SERVICE cos_missing 0.2140 0.4101 0.3792 0.4852 badconduct 0.0012 0.0352 0.0043 0.0653 honorable 0.4276 0.4947 0.4871 0.4998 uncharacterized 0.0811 0.2730 0.0096 0.0975 under_honorable 0.0761 0.2652 0.0516 0.2212 cos_unknown 0.1999 0.3999 0.0683 0.2522 character disorder 0.0424 0.2015 0.0363 0.1870 27

Overall attrition in 6 months Overall attrition in 12 months Overall attrition in 24 months Overall attrition in 45 months Character disorder attrition in 6 months Character disorder attrition in 12 months Character disorder attrition in 24 months Table 6. Descriptive Statistics of Air Force and Marine Corps AIRFORCE(507,496) MARINE CORPS(667,843) Mean Std. Dev. Mean Std. Dev. n=495,783 0.0970 0.2960 n=664,392 0.0676 0.2510 n=434,970 0.0335 0.1800 n=606,253 0.0418 0.2002 n=394,576 0.0597 0.2369 n=566,072 0.0560 0.2298 n=232,927 0.1749 0.3799 n=412,797 0.1148 0.3187 n= 495,783 0.0153 0.1225 n=664,392 0.0009 0.0293 n=434,970 0.0050 0.0704 n=606,253 0.0054 0.0736 n=394,576 0.0103 0.1011 n=566,072 0.0145 0.1194 Character disorder attrition in 45months n=232,927 0.0197 0.1391 n=412,797 0.0311 0.1736 ENLISTMENT AGES age17 0.0406 0.1973 0.1842 0.3877 age18 0.2961 0.4565 0.2958 0.4564 age19 0.2301 0.4209 0.1704 0.3760 age20 0.1446 0.3517 0.0912 0.2878 age21 0.0931 0.2906 0.0594 0.2363 age22 0.0658 0.2480 0.0456 0.2086 age23 0.0452 0.2078 0.0327 0.1778 age24 0.0315 0.1746 0.0233 0.1507 age25 0.0211 0.1438 0.0171 0.1297 age26_27 0.0264 0.1603 0.0235 0.1515 age28_29 0.0030 0.0547 0.0133 0.1146 age30_34 0.0021 0.0459 0.0242 0.1537 age35_42 0.0003 0.0186 0.0194 0.1378 age 19.8740 2.2950 19.9991 3.9565 TIS 1,438.8840 1,162.2200 1,595.4690 1,091.7350 female 0.2331 0.4228 0.0728 0.2598 cat1 0.0241 0.1534 0.0356 0.1854 cat2 0.1795 0.3838 0.2908 0.4541 cat3 0.1928 0.3945 0.4631 0.4986 cat4 0.0006 0.0241 0.0079 0.0887 cat5 0.0000 0.0000 0.0000 0.0030 cat_unknown 0.6030 0.4893 0.2025 0.4019 28

MARINE AIRFORCE(507,496) CORPS(667,843) Mean Std. Dev. Mean Std. Dev. MARITAL STATUS other_marital status 0.0048 0.0688 0.0097 0.0978 married 0.1047 0.3062 0.1081 0.3105 single 0.8903 0.3125 0.8823 0.3223 mar_stat_unknown 0.0002 0.0154 0.0000 0.0039 EDUC_LEVEL 21.6610 5.3148 21.7189 3.0238 some_college (no diploma) 0.0441 0.2054 0.0112 0.1054 some high school(no diploma) 0.0033 0.0573 0.0044 0.0665 High school degree 0.8704 0.3358 0.9278 0.2588 education unknown 0.0648 0.2462 0.0132 0.1139 alternate education 0.0066 0.0812 0.0318 0.1756 college_degree 0.0083 0.0905 0.0078 0.0882 doctorate,master and above 0.0024 0.0491 0.0036 0.0602 RACE black 0.1585 0.3652 0.1154 0.3195 white 0.7504 0.4328 0.7785 0.4153 race_unknown 0.0191 0.1369 0.0440 0.2052 race_others 0.0430 0.2028 0.0234 0.1510 hispanic 0.0291 0.1681 0.0387 0.1929 hispan_95_99 0.0291 0.1681 0.0387 0.1929 CHARACTER OF SERVICE cos_missing 0.2247 0.4174 0.1567 0.3635 badconduct 0.0043 0.0652 0.0162 0.1262 honorable 0.3333 0.4714 0.4719 0.4992 uncharacterized 0.0517 0.2215 0.0565 0.2308 under_honorable 0.0549 0.2278 0.0648 0.2461 cos_unknown 0.3311 0.4706 0.2340 0.4234 character disorder 0.0416 0.1997 0.0484 0.2146 29