SEARCH

SEARCH BY CITATION

Abstract

  1. Top of page
  2. Abstract
  3. I. INTRODUCTION
  4. II. REVIEW OF SELECTED STUDIES
  5. III. DATA
  6. IV. DESCRIPTIVE EVIDENCE OF PERFORMANCE DIFFERENCES
  7. V. EVALUATING THE INHERENT RISKINESS OF WAIVERED RECRUITS: THE EFFECT OF WAIVER STATUS ON ATTRITION AND TIME-TO-E5 PROMOTION
  8. VI. CONCLUSIONS
  9. REFERENCES

In this study, we evaluate the performance of waivered recruits in the U.S. military. Unlike the private sector, the military has formal standards for identifying ideal recruits and uses a formal screening process to determine those within risky populations who are most likely to succeed. (Recruits who make it through the screening process are issued a waiver.) The military's establishment of waiver categories and its tracking of waiver status provide us with a case study for determining whether such risk-identification strategies work. Using FY99–FY08 service-level waiver and personnel data, we evaluate whether the military recruiting strategy has been successful and whether firms should consider adopting similar screening mechanisms. We estimate the effect of waiver status on attrition and promotion, our primary performance indicators, after controlling for other quality indicators. We find that waivered recruits, on the whole, are not particularly poor performers, although their inherent riskiness does vary by service and by waiver type. (JEL J45, M51, J23)


ABBREVIATIONS
AFQT

Armed Forces Qualification Test

DAT

Drug Alcohol Test

DEP

Delayed Entry Program

DMDC

Defense Manpower Data Center

DoD

Department of Defense

DTM

Directive-Type Memorandum

GED

General Educational Development Test

MEPCOM

Military Entrance Processing Command

MEPS

Military Entrance Processing Station

I. INTRODUCTION

  1. Top of page
  2. Abstract
  3. I. INTRODUCTION
  4. II. REVIEW OF SELECTED STUDIES
  5. III. DATA
  6. IV. DESCRIPTIVE EVIDENCE OF PERFORMANCE DIFFERENCES
  7. V. EVALUATING THE INHERENT RISKINESS OF WAIVERED RECRUITS: THE EFFECT OF WAIVER STATUS ON ATTRITION AND TIME-TO-E5 PROMOTION
  8. VI. CONCLUSIONS
  9. REFERENCES

Each year, the services must recruit thousands of new people in order to support the military's capabilities and to promote long-term force health.1 Their primary recruiting pool consists of 17- to 24-year-old men and women from a cross section of society. Unlike the private sector, the military has formal, published standards for identifying the ideal recruit. They formally consider the recruit's height and weight, and whether he or she has a history of drug use, serious legal problems, or family issues. In addition, they administer aptitude exams. An increasing number of young adults, however, do not qualify for service based on these criteria. Roughly one in four young Americans is too overweight to serve and another third have other health problems that keep them from qualifying, such as diabetes, asthma, or hearing impairments (Mission: Readiness 2009). Others are disqualified because they lack a high school diploma, they have legal problems, or they are single custodial parents (Mission: Readiness 2009).

The services, however, do not entirely eliminate these individuals; instead, they have instituted a formalized screening process to identify those within a “risky” population who have otherwise demonstrated that they are likely to excel. Under this process, the services can waive their usual requirements in such areas as physical fitness (e.g., weight waiver), family status (e.g., dependent waiver), or legal matters (e.g., misdemeanor or felony waiver). In this paper, we evaluate this process, with the aim of determining whether similar processes could be applied in the private sector.

The services' published standards defining the “ideal recruit” and resulting waiver categories and screening processes provide a unique opportunity to evaluate the performance of “risky” individuals. Because the services require waivers for these recruits, we can track them over time. That is, we can identify those “new hires” considered risky upon entry and track their performance; this is not possible in the private sector. Thus, we are able to evaluate the effects of a formalized screening tool in the military. There are arguably both direct and indirect effects from allowing select members of these “at-risk” populations to serve: if productive, there are direct, positive gains to the U.S. military. Indirect effects would include helping troubled individuals improve their lives. This paper focuses on the direct effects.

As much of the human resource literature identifies, only large firms (which hire in large groups and have predictable hiring cycles) have implemented formalized hiring and screening processes to weed out individuals who are not a good match (Barber et al. 1999). The screening processes used by the services, however, may be applicable to firms of all sizes. In addition, just as it is important for firms to predict who they expect will quit or be fired shortly within their tenure, the military tries to minimize its first-term attrition (Huo et al. 2002). All enlisted recruits sign up for initial terms of 3–6 years, depending on specialty and service; if they separate before completing the obligation, we call that “attrition.” First-term attrition is particularly costly, as returns on training may not be realized and replacing the lost labor is a lengthy and involved process.

Depending on the gravity of the waiver, different levels of authority are required—this allows lower level staff to make waiver decisions when appropriate. In the Marine Corps, for example, admission to prior marijuana use requires a waiver approved by the Recruiting District's Commanding Officer. However, waivers for recruits who show up at the Military Entrance Processing Station (MEPS) with marijuana in their system must be approved by the Commanding General of the Marine Corps Recruiting Command. This waiver process allows the services to apply extra scrutiny to higher-risk recruits.

Past research has shown that in some cases waivered recruits do as well as, or better than, those who enter without waivers (Baldor 2008; Quester and Morse 2007).2 For those considered risky ex ante, the military evaluates more of the “whole person,” which some firms have cited as an effective hiring strategy (Bowen, Ledford, and Nathan 1991). The success of the military's strategy, as illustrated in the remainder of this study, suggests that firms should consider adopting a similar screening mechanism for historically risky groups, in lieu of simply eliminating them from the hiring process. Screening in this way would allow firms to expand their recruiting pool to include high-risk populations. Beyond following the military's example in screening, firms should also consider removing screening information from employees' permanent records, as the military does. That is, waiver status and pre-accession “risky behavior” do not remain on a servicemember's personnel file. Once a servicemember accesses, his or her waiver status is struck from the file, so that the occupation-assignment process and determination of career opportunities are blind to waiver status.3

In the remainder of this paper, we evaluate whether hiring out of “risky” populations has proved a successful recruiting strategy for the military. We focus on the attrition and promotion behavior of waivered recruits as compared with their non-waivered counterparts, and we examine whether they are more likely to attrite or less likely to promote quickly. We evaluate attrition behavior to determine whether waivered recruits are more likely to be problematic recruits, and we look at promotion behavior to determine whether waivered recruits provide a net gain to the services.

In our analyses, we have no way of separating individual behavior from organizational behavior. When we observe that a particular population is less likely to attrite, we do not know whether this is because that population was screened extensively, and the riskier members of the population did not access or whether, as a whole, individuals who require that waiver type are simply less risky. Soldiers with adult felony waivers, for example, are less likely to attrite early than non-waivered soldiers. We cannot decisively determine whether this is because the Army screened the adult felony waiver population extensively, or whether all applicants for adult felony waivers were inherently lower-risk individuals. Firms that implement this screening process would have an advantage; in determining the success of their programs over time, they would have perfect information on their organizational processes and could separate the two effects. That is, they would be able to determine which populations are inherently less risky and which require extra screening, and they would be able to create organizational policies to minimize risk.

II. REVIEW OF SELECTED STUDIES

  1. Top of page
  2. Abstract
  3. I. INTRODUCTION
  4. II. REVIEW OF SELECTED STUDIES
  5. III. DATA
  6. IV. DESCRIPTIVE EVIDENCE OF PERFORMANCE DIFFERENCES
  7. V. EVALUATING THE INHERENT RISKINESS OF WAIVERED RECRUITS: THE EFFECT OF WAIVER STATUS ON ATTRITION AND TIME-TO-E5 PROMOTION
  8. VI. CONCLUSIONS
  9. REFERENCES

Findings on the performance of waivered recruits vary depending on the service, the time period, and the performance metric. A study of first-term attrition in the Army, for example, finds that attrition rates for waivered soldiers are lower at the beginning of the first term, but higher at the end of the same term (Distifeno 2008). This study also finds that waivered soldiers have more disciplinary problems and face more courts-martial than non-waivered soldiers. In addition, it finds that the pattern of pre-enlistment offenses matters—recruits with numerous minor offenses are more likely to misbehave than those with only a single major offense. A similar study of performance in the Army finds that waivered soldiers have higher rates of desertion, misconduct, court-martial appearances, and alcohol rehabilitation failure (Baldor 2008). The same study also finds, however, that waivered soldiers are more likely to reenlist, promote faster to sergeant, and have lower rates of dismissal for personality disorders or unsatisfactory performance.

Other studies on performance of waivered recruits focus on the Navy and Marine Corps. One evaluation of Navy recruits from FY95– FY96 revealed that waivered sailors have significantly higher rates of “unsuitability attrition.” In addition, the study found that those who are both waivered and have not completed a high school degree are the most likely to attrite for “unsuitable” reasons (Hall 1999). Likewise, Wenger and Hodari (2004) find that although waiver status does predict attrition for sailors, it is not nearly as strong a predictor as being a heavy smoker, having a General Educational Development (GED) Test, or being a high school dropout. They also find that time spent in the Delayed Entry Program (DEP) is negatively correlated with 36-month attrition.4Etcho (1996) evaluates the likelihood of “unsuitable attrition” in the Marine Corps at a time when about 60% of Marine Corps accessions were receiving waivers.5 He finds that recruits with felony, drug/alcohol, or serious offense waivers were the most likely to attrite, as were those who were non-high school graduates or in aptitude categories IV, IIIB, or IIIA (in that order).6

Finally, there is one Department of Defense (DoD)–wide study of waiver status on attrition and performance, although it focuses on felony and serious waivers only. The authors find higher 18-month attrition rates for waivered recruits, with service-by-service variation in the types of waivered recruits that are most likely to attrite (Putka et al. 2004). In all services, however, waivers for testing positive for drug or alcohol use (Drug Alcohol Test [DAT] waivers) are highly predictive of attrition behavior.7

Overall, previous research suggests that we must distinguish between different waiver categories, evaluate attrition rates at multiple points in time (as short-term and long-term results may differ), and evaluate more than one performance metric. These studies also highlight the importance of comparing waivered recruits with other “risky” recruits, such as those who lack a traditional high school diploma.

III. DATA

  1. Top of page
  2. Abstract
  3. I. INTRODUCTION
  4. II. REVIEW OF SELECTED STUDIES
  5. III. DATA
  6. IV. DESCRIPTIVE EVIDENCE OF PERFORMANCE DIFFERENCES
  7. V. EVALUATING THE INHERENT RISKINESS OF WAIVERED RECRUITS: THE EFFECT OF WAIVER STATUS ON ATTRITION AND TIME-TO-E5 PROMOTION
  8. VI. CONCLUSIONS
  9. REFERENCES

We use data from four source files. The Defense Manpower Data Center (DMDC) provides the accession file, which is based on Military Entrance Processing Command (MEPCOM) transactions. DMDC also provides the active-duty master file (quarterly snapshots) and the active-duty transaction file. We supplement these data with data from CNA's military personnel files.

In the past, the services had heterogeneous waiver criteria, making it difficult to perform cross-service comparisons. In 2008, the DoD released a Directive-Type Memorandum (DTM) requiring that the services standardize their waiver reporting. The memorandum, however, created confusion among recruiting and accession commands, and a reliable source of cross-service waiver data is not yet available. For these reasons, we conduct intra-service analysis only, and focus on FY99–FY08.8

From the MEPCOM accession file, we extract information on servicemembers that is observable at accession, including Armed Forces Qualification Test (AFQT) score, months spent in DEP, race/ethnicity, age, and waivers. The MEPCOM file contains separate DEP and accession waivers, and a maximum of three data fields for each. MEPCOM advised us that some of the waivers recorded in a recruit's DEP section also appear in his or her accession waiver section—resulting in double-counting for that recruit. For example, consider recruits A and B. Recruit A accesses with a dependent waiver, which appears in both the DEP and accession waiver fields. Recruit B has two DEP waivers—drug/alcohol and dependent—which both appear in the DEP waiver fields, but only the dependent waiver appears in the accession waiver field. We count recruit A as having one type of waiver and recruit B as having two. That is, we define multiple waivers as having more than one waiver type. Table 1 provides a generic description of each waiver type (although they vary by service), as well as gross and relative population sizes; Figure 1 shows the service time trends in waivered accessions.

Table 1.  General Waiver Categories and Service-Specific Population Sizesa
Waiver TypeNeeded if the recruit:Army N%Navy N%Marine Corps N%Air Force N%
  1. aThe waiver descriptions provided here are only a “general” description encompassing all four services. In reality, there is significant variation in how each service defines each waiver type.

PhysicalIs outside height/weight standards47,1137.3%23,6206.0%42,05613.3%14,0664.7%
 or has a medical condition        
AptitudeHas a below-standard AFQT score (or sub-scores)1,2170.2%1,2110.3%1,8650.6%3,7941.3%
Drug/AlcoholAdmits to a history of drug or alcohol abuse1,3810.2%7,3221.8%109,82234.7%3420.1%
DAT WaiverFails a drug test upon arriving at MEPS9,9401.5%1,2750.3%4,9411.6%130.0%
DependentHas two or more non-spousal dependents5,2510.8%11,3522.9%10,2343.2%3,1131.0%
Adult FelonyWas convicted of a felony at 18 years of age or older7,0951.1%8510.2%1,6000.5%2,9571.0%
Juvenile FelonyHas a juvenile felony conviction4,2620.7%7930.2%2,8100.9%2280.1%
SeriousHas committed serious offenses (e.g., misdemeanor, adultery, unlawful entry)41,3186.4%38,7979.8%28,6629.0%12,4584.2%
OtherHas any other waiver (including waivers for age, being from a hostile country, etc.)4,8740.8%35,9569.1%28,9989.2%8,1802.7%
EducationHas particular combinations of education credentials and AFQT scores (e.g., Homeschool diploma and scored 50–64 on AFQT).N/A 14,9893.8%N/A N/A 
Total accessions 644,112 396,217 316,856 300,144 
image

Figure 1. Trends in Percent of Accessions with at Least One Waiver: FY99–FY08, by Service

Download figure to PowerPoint

The active-duty master file tracks non-prior-service accessions—those with no previous military experience—quarterly through time. We use this file to identify whether (and when) a recruit attrited. This file also provides loss reasons, which we use to define attrition. We consider servicemembers to have attrited if they leave for any negative reason, including drug or alcohol use, personality disorder, fraudulent entry, or misconduct (regardless of when the loss occurred). Some loss categories are never considered attrition. If, for example, a servicemember retires early, is lost because of force reduction, or transitions to officer status, he or she is not included in our attrition counts. Finally, some loss reasons are deemed negative (and thus counted as attrition) only if they occur more than 3 months before the end of a servicemember's obligation.9 These include weight control failure, failing physical standards, parenthood, or unsatisfactory performance. We use these loss codes to calculate attrition by 6 and 48 months (henceforth referred to as “early” and “first-term” attrition), and we use attrition as our primary performance metric.

Our other key performance metric is time-to-E5 promotion.10 We consider time-to-E5 promotion because promotions to E5 do not happen “automatically,” as they do through E4. In addition, the process is consistently competitive, and a sufficient number of servicemembers promote to E5 within their first term; this guarantees that we are not confounding the reenlistment decision with the promotion result.11 Servicemembers compete for promotion within their occupation and accession-year cohort (e.g., Marine Corps E4 riflemen compete against other E4 riflemen who entered in the same year). In addition, promotion rates are determined not only by a servicemember's relative quality but also by the demand for and supply of personnel within that occupation. (Servicemembers promote to vacancies—if there are no available E5 billets in a particular occupation, no E4s promote.) As a result, for the fast-to-E5 metric, we compare each individual's time-to-E5 promotion with those of other servicemembers in the same occupation-accession year cohort. We identify a servicemember as being fast-to-E5 if he or she is among the fastest 25% of promoters in that cohort.12

IV. DESCRIPTIVE EVIDENCE OF PERFORMANCE DIFFERENCES

  1. Top of page
  2. Abstract
  3. I. INTRODUCTION
  4. II. REVIEW OF SELECTED STUDIES
  5. III. DATA
  6. IV. DESCRIPTIVE EVIDENCE OF PERFORMANCE DIFFERENCES
  7. V. EVALUATING THE INHERENT RISKINESS OF WAIVERED RECRUITS: THE EFFECT OF WAIVER STATUS ON ATTRITION AND TIME-TO-E5 PROMOTION
  8. VI. CONCLUSIONS
  9. REFERENCES

In this section, we evaluate the relative riskiness of each waivered population, as measured both by the frequency of fast promotions and by average first-term attrition rates. For each service, we highlight those waiver populations that perform well and are therefore not a particularly risky accession group. We also highlight those who are among the riskiest. Table 2 shows the average first-term attrition and fast-to-E5 promotion rates for each waiver group, by service. The table also shows the relative performance of the Tier II/III population (defined as those with GEDs, alternative credentials, or without a high school diploma), as this population, like the waivered recruit population, is typically viewed as a high risk.

Table 2.  Percent First-Term Attrition and Fast-to-E5, by Waiver Status and Service
 Percent Attrition by End of First TermPercent Fast-to-E5
Army (%)Navy (%)Marine Corps (%)Air Force (%)Army (%)Navy (%)Marine Corps (%)Air Force (%)
Physical waiver38.2234.1125.3828.2524.4227.7519.9527.43
Aptitude waiver47.9237.4630.8036.0521.4416.8824.6514.79
Drug/alcohol waiver51.9838.3024.8628.1824.2429.6224.5635.04
DAT waiver49.3254.8635.06N/A23.1121.9020.10N/A
Dependent waiver41.8738.5427.7927.1634.7434.6931.6541.90
Adult felony waiver36.0635.3230.9630.2130.3429.6131.4630.49
Juvenile felony waiver37.8838.6828.1029.9423.9631.2027.1533.33
Serious waiver38.7842.8427.4531.4529.1028.6126.1427.34
Education waiverN/A51.15N/AN/AN/A24.35N/AN/A
Other waiver42.6440.0826.5730.2624.9527.2526.4531.01
No waiver38.0231.6219.5428.0924.7425.5826.3727.33
Tier II/III recruit44.8852.4338.2042.2018.9925.3817.7023.92

In the Army, those with adult felony, dependent, or serious offense waivers are low-risk recruits. They are more likely than their non-waivered counterparts to promote fast-to-E5, and only those with dependent or serious offense waivers are slightly more likely to attrite by 48 months. Conversely, those with drug/alcohol, DAT, or aptitude waivers are high risk, as are Tier II/III recruits. Similarly, in the Navy, those with dependent or adult felony waivers are low-risk populations, while the DAT waiver, education waiver, and Tier II/III populations are higher risk. In the Marine Corps, there is a large cluster of waivered populations with near-average risk levels; only those with a dependent waiver or no waiver stand out as being lower risk. The Tier II/III, DAT waiver, and aptitude waiver populations are among the riskiest. Finally, in the Air Force, there are a number of low-risk groups. These populations are small, as most Air Force recruits require no waiver at all. Those in the Tier II/III and aptitude waiver populations are the Air Force's riskiest accessions. The numbers presented in Table 2 are simply means; they do not represent the independent effects of waiver status.

In the remaining sections, we attempt to capture the independent waiver effects, in order to determine which waiver groups are most likely to be better (or worse) performers. We do this by controlling for characteristics observable to the services at the time of enlistment. Because waivers represent risky behavior prior to enlistment, and our performance measures focus on time in the military, there are no concerns of reverse causality.

V. EVALUATING THE INHERENT RISKINESS OF WAIVERED RECRUITS: THE EFFECT OF WAIVER STATUS ON ATTRITION AND TIME-TO-E5 PROMOTION

  1. Top of page
  2. Abstract
  3. I. INTRODUCTION
  4. II. REVIEW OF SELECTED STUDIES
  5. III. DATA
  6. IV. DESCRIPTIVE EVIDENCE OF PERFORMANCE DIFFERENCES
  7. V. EVALUATING THE INHERENT RISKINESS OF WAIVERED RECRUITS: THE EFFECT OF WAIVER STATUS ON ATTRITION AND TIME-TO-E5 PROMOTION
  8. VI. CONCLUSIONS
  9. REFERENCES

In this section, we estimate the effect of having a certain waiver type on the probability of attrition (at 6 and 48 months) and of promoting fast-to-E5, after controlling for other factors known to determine the performance of recruits. We estimate probit regressions, controlling for demographic and military characteristics. Specifically, we estimate the following equation:

  • image

In our first estimations, Y is attrition, i is the individual, and j is either 6 or 48 months (depending on whether we are estimating early or first-term attrition). X is a vector demographic variables, including gender, racial/ethnic group, a gender× racial/ethnic group interaction term, age at accession, age-squared, the number of non-spousal dependents, marital status, geographic region of origin, and an indicator for Tier II/III education status. There are ω waiver types, α is the intercept, d is the year variable, and ɛ is the error term. M is a vector of military characteristics and includes paygrade on the servicemember's first snapshot file, AFQT score, trimester of accession, and a dummy indicating whether the servicemember spent longer than average in DEP. We also include a variable indicating whether a recruit has multiple types of waivers, implying multiple “discrepancies” from the services' definitions of an ideal recruit.13 We are interested in the δ coefficients, representing the independent effect of waiver status on performance outcomes, after controlling for demographic and military characteristics. The same equation is estimated for time-to-E5, changing the left-hand-side variable to fast-to-E5 promotion.

Screening aside, we expect most waiver types to be associated with negative behaviors (more likely to attrite; less likely to promote quickly), as these servicemembers have previously exhibited negative behaviors, have less than the desired intellectual aptitude, or have medical/physical conditions that may interfere with their service. The expected effect of dependent waivers is ambiguous: their need for financial stability might encourage commitment to their military career; juggling familial and military responsibilities might create insurmountable pressures.

A. Attrition

Table 3 presents the effects of each waiver type on early- and first-term attrition for each of the services; these are the marginal effects from being in a particular waiver group (or having a particular characteristic) on the probability of attrition, all else equal.14

Table 3.  Attrition Marginal Effects from Probit Regressions
 ArmyNavyMarine CorpsAir Force
EarlyFirst termEarlyFirst termEarlyFirst termEarlyFirst term
  1. * and ** denote statistical significance at the 5% and 1% levels, respectively. Standard errors in parentheses.

Physical waiver0.0100.0040.0100.0060.0150.0240.002-0.003
 (0.002)**(0.004)(0.002)**(0.004)(0.002)**(0.003)**(0.002)(0.005)
Aptitude waiver0.0080.0210.0050.019-0.0010.0000.0110.025
 (0.008)(0.021)(0.007)(0.014)(0.005)(0.010)(0.004)*(0.008)**
Drug/alcohol waiver−0.0170.126−0.0060.051−0.0020.024−0.030−0.008
 (0.007)*(0.019)**(0.003)*(0.007)**(0.001)*(0.002)**(0.009)**(0.031)
DAT waiver−0.0130.136−0.0020.1700.0130.0980.1180.118
 (0.003)**(0.009)**(0.007)(0.016)**(0.004)**(0.009)**(0.114)(0.173)
Dependent waiver−0.020−0.0790.0200.0360.0050.0190.0020.025
 (0.004)**(0.014)**(0.004)**(0.007)**(0.003)(0.007)**(0.005)(0.012)*
Adult felony waiver−0.0230.004−0.0010.0300.0030.047−0.0160.035
 (0.003)**(0.011)(0.008)(0.023)(0.006)(0.013)**(0.004)**(0.011)**
Juvenile felony waiver−0.0180.042−0.0020.0460.0010.046−0.0110.004
 (0.004)**(0.015)**(0.009)(0.021)*(0.005)(0.010)**(0.015)(0.037)
Serious waiver−0.0180.0340.0040.083−0.0040.037−0.0080.039
 (0.001)**(0.005)**(0.001)**(0.003)**(0.002)*(0.004)**(0.002)**(0.005)**
Education waiverN/AN/A0.0100.055N/AN/AN/AN/A
   (0.002)**(0.006)**    
Other waiver0.0150.0480.0050.0580.0010.011−0.0020.031
 (0.005)**(0.013)**(0.002)**(0.004)**(0.002)(0.004)**(0.003)(0.007)**
Multiple waivers−0.003−0.040−0.007−0.0440.000−0.0130.003−0.004
 (0.004)(0.010)**(0.002)**(0.006)**(0.002)(0.004)**(0.006)(0.011)
Tier II/III0.0230.0580.0290.1350.0350.0930.0420.145
 (0.001)**(0.003)**(0.002)**(0.005)**(0.003)**(0.006)**(0.005)**(0.009)**
Feb–May trimester0.0150.042−0.0060.0180.0080.0150.0020.014
 (0.001)**(0.002)**(0.001)**(0.002)**(0.001)**(0.003)**(0.001)(0.003)**
Oct–Jan trimester0.0050.034−0.0120.007−0.0080.004−0.0130.000
 (0.001)**(0.003)**(0.001)**(0.002)**(0.001)**(0.002)(0.001)**(0.003)
Long DEP−0.014−0.051−0.016−0.075−0.010−0.042−0.009−0.042
 (0.001)**(0.003)**(0.001)**(0.002)**(0.001)**(0.002)**(0.001)**(0.002)**
Male−0.075−0.267−0.038−0.072−0.069−0.117−0.027−0.086
 (0.001)**(0.003)**(0.002)**(0.003)**(0.003)**(0.005)**(0.001)**(0.003)**
Average attrition rate0.100.360.070.300.070.190.060.27
N 617,517256,109372,137261,646302,304193,880285,513193,009
Pseudo R20.040.050.030.040.020.030.020.02

For the Army, the effect of waiver status on early attrition is usually negative. Soldiers accessed with an adult felony waiver are 2.3 percentage points less likely to attrite early than their non-waivered counterparts. Those with a juvenile felony, serious offense, dependent, drug/alcohol, or DAT waiver are also significantly less likely to attrite early. Conversely, those with physical or other waivers have higher short-term attrition rates.

The most significant contributions of waiver status on attrition probabilities, however, are for the first-term attrition rates of those with a dependent, drug/alcohol, or DAT waiver. Although those with a dependent waiver are less likely to attrite at all intervals, those with DAT or drug/alcohol waivers are more likely to attrite by 48 months. For example, soldiers accessed with a DAT waiver are 13 percentage points more likely to attrite by 48 months than the non-waivered.

In addition to the waiver effects, Table 3 shows the effects of a few other variables, including being a Tier II/III recruit (relative to Tier I) and accessing in the October–January or February–May trimesters (relative to the most popular period of June–September).15 These effects show how waiver effects compare with other variables related to attrition. In the Army, the effects of having a dependent, drug/alcohol, or DAT waiver on first-term attrition are significantly greater than the other effects. This suggests that, on average, Army recruits with a dependent waiver have inherently lower risk, whereas those with drug/alcohol or DAT waivers are riskier.16

In the Navy, the marginal effects of waivers on attrition probabilities are almost always positive. The only negative effect is for early attrition of those with drug/alcohol waivers, and the effect is small—0.6 percentage points. The most sizable effects occur for those with DAT or serious waivers; they are 17.0 and 8.3 percentage points more likely to attrite in the first term. We conclude that these two populations have inherently high attrition risk. Those with a DAT waiver are more likely to attrite in the first term than Tier II/III recruits.

We find similar results in the Marine Corps— DAT waivers have the greatest effect on attrition probabilities. Recruits with DAT waivers are 1.3 and 9.8 percentage points more likely to attrite early and by the end of their first term. Recruits in the Tier II/III population have an equally high attrition risk. Excluding those who require DAT waivers from the Marine Corps accession pool will reduce the overall attrition risk of the waivered population, as the effect of all other waiver types is much smaller.17

Overall marginal effects for the Air Force are small—less than 5 percentage points for all waivers and for both attrition rates. The effect of having a DAT waiver is large, but insignificant, because the Air Force granted only 13 DAT waivers during the sample period. The effect of being a Tier II/III recruit is much greater than any waiver effect: Tier II/III recruits are 4 and 14 percentage points more likely to attrite early and by the end of the first term than a comparable Tier I recruit. This is their riskiest population.

The effects of other covariates on attrition are consistent with previous literature. Laurence, Ramsberger, and Arabian (1996), Laurence (1984), Buddin (1984), Thompson (2011), and Hattiangadi and Brookshire (2006) find that attrition probabilities decrease with more years of education and with higher AFQT scores. McIntosh and Sayala (2011) and Hattiangadi and Brookshire (2006) find that longer DEP times decrease attrition probabilities and that attrition rates are highest for those accessed from February–May, followed by October–January and June–September. A number of these studies also report that age has a positive effect on attrition and that women and minorities are more likely to attrite.

In this section, we identified waiver categories with the greatest independent effect on attrition probability. In most cases, the independent effects of waiver status are small, usually under 5 percentage points, and are smaller than the effects of being a Tier II/III recruit or accessing between February and May. These results indicate that the services have effectively screened these populations to minimize risk. This suggests that private firms could gain from expanding their recruiting pools to include traditionally high-risk populations; they would not necessarily be putting themselves at risk of increased attrition. Thoughtful decisions on the most appropriate characteristics to screen for and proper implementation of the process will, of course, be key in making such a screening mechanism successful.

B. Fast-to-E5

In this section, we compare the performance of waivered and non-waivered recruits, using time-in-service prior to E5 promotion as our metric.18 We rely on the same probit regressions, changing the left-hand-side variable from attrition to promotion.19

The marginal effects of each waiver group, as well as the effects of spending longer than average in DEP and of being a Tier II/III, February–May, or October–January recruit are presented in Table 4. There is, again, a significant variation by waiver type. In the Army, we find that recruits with a physical waiver are less likely to promote fast-to-E5, whereas those with an adult felony, serious, or “other” waiver are more likely to promote quickly than their non-waivered counterparts. The positive effects are sizeable when compared with other typically important indicators—the probability of promoting fast-to-E5 is increased by 4.4, 3.8, and 3.6 percentage points, respectively, for adult felony, serious, and “other” waivers, while the effect of being Tier II/III is −3.0 and the effect from spending a longer than average time in DEP is 1.7.

Table 4.  Fast-to-E5 Marginal Effects from Probit Regressions
 ArmyNavyMarine CorpsAir Force
  1. FMAM, February-March-April-May; ONDJ, October-November-December-January.

  2. * and ** denote statistical significance at the 5% and 1% levels, respectively. Standard errors in parentheses.

Physical waiver−0.020−0.004−0.067−0.016
 (0.003)**(0.004)(0.003)**(0.006)*
Aptitude waiver−0.016−0.0430.004−0.029
 (0.018)(0.018)*(0.014)(0.011)*
Drug/alcohol waiver0.0070.011−0.0070.053
 (0.019)(0.007)*(0.002)**(0.043)
DAT waiver−0.012−0.039−0.0460.115
 (0.007)(0.019)(0.010)**(0.225)
Dependent waiver−0.0130.0060.0210.002
 (0.012)(0.008)(0.008)*(0.013)
Adult felony waiver0.044−0.0040.035−0.010
 (0.008)**(0.024)(0.017)*(0.013)
Juvenile felony waiver0.0110.0480.0180.073
 (0.011)(0.025)(0.013)(0.051)
Serious waiver0.0380.0090.002−0.015
 (0.004)**(0.003)*(0.004)(0.006)*
Education waiverN/A−0.020N/AN/A
  (0.007)*  
Other waiver−0.036−0.0200.006−0.001
 (0.009)**(0.003)**(0.004)(0.008)
Tier II/III recruit−0.0300.001−0.070−0.037
 (0.002)**(0.005)(0.007)**(0.010)**
FMAM recruit0.003−0.0130.0120.042
 (0.002)(0.002)**(0.003)**(0.003)**
ONDJ recruit−0.009−0.0210.0030.001
 (0.002)**(0.002)**(0.002)(0.003)
Long DEP0.0170.021−0.0050.008
 (0.002)**(0.002)**(0.002)(0.003)**
Male0.0120.0320.005−0.009
 (0.003)**(0.003)**(0.006)(0.004)*
Average fast-to-E5 rate0.250.260.250.26
N260,952176,449137,362104,114
Pseudo R20.040.060.030.07

In the Navy, there are negative effects from having an aptitude, education, or “other” waiver, but positive effects from having a serious or drug/alcohol waiver. There are also significant variations in waiver effects in the Marine Corps, where those with physical or DAT waivers are 6.7 and 4.6 percentage points less likely to promote quickly. Those with an adult felony waiver are 3.5 percentage points more likely. None of the Marine Corps waiver effects, however, are as large in magnitude as the effect from being Tier II/III. Finally, in the Air Force, we find only negative effects—for physical, aptitude, and serious waivers—and the waiver effects are all smaller than those from being a February–May recruit, suggesting that this is a riskier population.

The effects of other covariates on promotion are consistent with the sparse military-promotion literature. Both Schmitz et al. (2011) and Schmitz et al. (2008) find that recruits with higher AFQT scores and longer DEP times are more likely to advance to E4 and E5, while Tier II/III recruits are less likely.

Overall, we do not find that waivered recruits are systematically slower promoters and, hence, poorer performers. In contrast, we have shown that, after controlling for demographic and military characteristics, some waiver groups are inherently more likely to be fast promoters. Once again, our analysis indicates that the private sector could benefit from introducing a screening process similar to the services' process and that—if implemented carefully—they may find that some members of traditionally high-risk populations become some of their best performers.

VI. CONCLUSIONS

  1. Top of page
  2. Abstract
  3. I. INTRODUCTION
  4. II. REVIEW OF SELECTED STUDIES
  5. III. DATA
  6. IV. DESCRIPTIVE EVIDENCE OF PERFORMANCE DIFFERENCES
  7. V. EVALUATING THE INHERENT RISKINESS OF WAIVERED RECRUITS: THE EFFECT OF WAIVER STATUS ON ATTRITION AND TIME-TO-E5 PROMOTION
  8. VI. CONCLUSIONS
  9. REFERENCES

After controlling for various military and demographic characteristics, we find that waivered recruits, on the whole, are not a particularly poorly performing population. They do not, on average, stand out as an inherently risky accession source, although identified, ex ante, as such by the services. Our attrition and promotion analyses indicate that the riskiness of waivered recruits varies significantly by waiver type and by service. In many cases, education tier and trimester of accession have more predictive power of performance than a recruit's waiver status. Performance in a particular waiver group may be tied to the services' intense screening of recruits in that category; the military has illustrated that such screening reduces risk in traditionally risky populations.

It could be argued that our results are at least partially being driven by more flexible wartime attrition standards. However, even in an era of stop-loss (involuntary extensions) and war, involuntary separations still occur; it is the voluntary separations that come to a halt. We would therefore still expect the attrition that does occur to be among the riskiest individuals. And what we have determined here is that waivered recruits are not systematically more likely to be in this riskier population. In addition, there is no reason to expect stop-lossed personnel (or those retained solely for wartime needs) to be among the fastest promoters. This suggests that our findings on the performance of the waivered population are not contingent on flexible wartime standards.

In the services, the prevalence of waivers is a function of the recruiting market—when potential recruits who meet the quality standards are plentiful, the cost of turning away an individual who requires a waiver is low. When recruiting is more difficult, however, the cost is higher. Analogous trends are observed in the private sector—when unemployment rates are high and there is sufficient supply of qualified applicants, the need to overlook particular characteristics and recruit from traditionally riskier populations is low. It is in times of low qualified-labor supply (or high demand) when screening high-risk populations for the most qualified individuals could be fruitful in the private sector. Just as military recruiters must meet their missions, firms hiring to meet production “surges” must hire sufficient personnel. It is under such conditions that firms should look closely at what the military has learned about hiring from risky populations and apply the successful components of their hiring processes. This is clearly not a cost-free process—the services devote significant time and resources to screening applicants that require waivers. The military's success in screening waivered recruits illustrates, however, that there are gains from this process, as—in many cases—the services have been able to screen out members of these populations who may have been more likely to attrite or less likely to promote quickly.

REFERENCES

  1. Top of page
  2. Abstract
  3. I. INTRODUCTION
  4. II. REVIEW OF SELECTED STUDIES
  5. III. DATA
  6. IV. DESCRIPTIVE EVIDENCE OF PERFORMANCE DIFFERENCES
  7. V. EVALUATING THE INHERENT RISKINESS OF WAIVERED RECRUITS: THE EFFECT OF WAIVER STATUS ON ATTRITION AND TIME-TO-E5 PROMOTION
  8. VI. CONCLUSIONS
  9. REFERENCES
  • Baldor, L. “Study: Recruits on Waivers Get Promoted Faster.” Army Times, 2008. Accessed June 2011. http:// www.armytimes.com/news/2008/04/ap_militarywaive rs_042908/.
  • Barber, A., M. Wesson, Q. Roberson, and M. Taylor. “A Tale of Two Job Markets: Organizational Size and Its Effects on Hiring Practices and Job Search Behavior.” Personnel Psychology, 52(4), 1999, 84167.
  • Bowen, D., G. Ledford Jr., and B. Nathan. “Hiring for the Organization, Not the Job.” Academy of Management Executive 5(4), 1991, 3551.
  • Buddin, R. “Analysis of Early Military Attrition Behavior.” RAND, AD-A145 548, 1984.
  • Distifeno, C. “Effects of Moral Conduct Waivers on First-Term Attrition of U.S. Army Soldiers.” Naval Postgraduate School Thesis, 2008.
  • Etcho, L. “The Effect of Moral Waivers on First-Term, Unsuitability Attrition in the Marine Corps.” Naval Postgraduate School Thesis, 1996.
  • Golfin, P. “Manning Under AIP.” CNA Annotated Briefing D0014440.A1/Final, 2006.
  • Hall, L. “Analyzing Success of Navy Enlistees with Moral Waivers.” Naval Postgraduate School Thesis, 1999.
  • Hattiangadi A., and D. Brookshire. “Emerging Issues in USMC Recruiting: Comparing Relative Attrition Risk among Marine Corps Recruits.” CNA Research Memorandum D0014200.A2/Final, 2006.
  • Huo, Y., H. Huang, and N. Kapier. “Divergence or Convergence: A Cross-national Comparison of Personnel Selection Practices.” Human Resource Management, 41(1), 2002, 3144.
  • Laurence J. “Education Standards for Military Enlistment and the Search for Successful Recruits.” Human Resources Research Organization Final Report 84-4, FR-PRD-84-4, 1984.
  • Laurence J., P. Ramsberger, and J. Arabian. “Education Credential Tier Evaluation.” Human Resources Research Organization Final Report 96-19, FR0EADD-96-19, 1996.
  • McIntosh, M., and S. Sayala. “Non-Citizens in the Enlisted U.S. Military.” CNA Research Memorandum D00257 68.A2/Final, 2011.
  • Mission: Readiness (Military Leaders for Kids). “Ready, Willing, and Unable to Serve.” 2009. Accessed March 2010. http://cdn.missionreadiness.org/NATEE1109.pdf.
  • Putka, D. J., C. L. Noble, D. E. Becker, and P. F. Ramsberger. “Evaluating Moral Character Waiver Policy against Servicemember Attrition and In-Service Deviance through the First 18 Months of Service.” FR-03-96. Alexandria, VA: Human Resources Research Organization, 2004.
  • Quester A., and J. Morse. “Bibliography of CNA Manpower Work: Marine Corps Focus.” CNA Information Memorandum D0017311.A1/Final, 2007.
  • Schmitz, E., A. Clemens, C. Hiatt, and D. Reese. “Recruit Quality and Performance Indicators: Evidence from the Navy and Marine Corps.” CNA Research Memorandum D0025156.A2/Final, 2011.
  • Schmitz, E., M. Moskowitz, D. Gregory, and D. Reese. “Recruiting Budgets, Recruit Quality, and Enlisted Performance.” CNA Research Memorandum D001703 5.A2/Final, 2008.
  • Thompson, E. “Effect of State Unemployment Rate on Attrition for First-Term U.S. Navy Enlisted Attrition.” Naval Postgraduate School Thesis, 2011.
  • Wenger, J., and A. Hodari. “Predictors of Attrition: Attitudes, Behaviors, and Educational Characteristics.” CNA Research Memorandum D0010146.A2/Final, 2004.
Footnotes
  • 1

    Throughout this report, the “services” include the U.S. Army, the U.S. Navy, the U.S. Marine Corps, and the U.S. Air Force; we exclude the Coast Guard because of its small size.

  • 2

    CNA, as the federally funded research and development center for the Navy and Marine Corps, has conducted a fair amount of research on the performance of waivered recruits. This work, however, was done in support of military commands and is therefore not releasable to the public.

  • 3

    There is one exception—waiver information is transferred to the services for those applying for jobs that require a secret or top-secret clearance. Individuals with past drug behavior, felonies, or misdemeanors would not qualify for clearances, and the provision of waiver information in these cases eliminates a duplication of effort.

  • 4

    Individuals often enlist first in DEP before shipping to bootcamp. They spend these months interacting with recruiters, learning about military culture, and deciding whether they are fit for military life. They are often completing their senior year of high school and have DEP activities on the weekend. There is no binding contract while in DEP; these recruits are part of the inactive reserve until they ship to bootcamp.

  • 5

    This is mostly because the Marine Corps requires a drug waiver for admission of any previous marijuana use. Roughly 30% of Marine Corps accessions have required a drug/alcohol waiver over time.

  • 6

    The aptitude categories are based on the Armed Forces Qualification Test (AFQT) scores. Category IV recruits score 10–30, while IIIB recruits score 31–49, and IIIA recruits score 50–64. Category II recruits score 65–92; category I recruits score 93–100. The overall AFQT is percentile-based.

  • 7

    DAT waivers are required for recruits who do not admit drug use to a recruiter and then fail the drug test administered at the MEPS.

  • 8

    We include only non-prior-service accessions as we want to track the servicemembers' careers and estimate their service lengths.

  • 9

    We cannot be certain that servicemembers within 3 months of their end-of-obligation date have actually violated their contract.

  • 10

    In the military, there are nine enlisted paygrades (E1–E9); E1s are the most junior, E9s the most senior. Paygrades E1–E4 are generally considered “junior enlisted,” E5–E7 are mid-level, and E8–E9 comprise the senior enlisted force.

  • 11

    Individuals who have no intention of reenlisting can request to be removed from a particular promotion board.

  • 12

    We use 3- instead of 1-year accession cohorts, as the latter results in an insufficient sample size in many occupations. For those occupations that still have an insufficient sample size to create a reliable “comparison group,” we use the 25th percentile cutoff for the entire service, by 3-year cohort.

  • 13

    Among waivered recruits, the percentage with two waivers in our sample is 8.5 for the Army, 11.6 for the Navy, 21.3 for the Marine Corps, and 6.6 for the Air Force. The percentages with three or more waivers are 0.4, 1.3, 5.1, and 0.3.

  • 14

    These were calculated using Stata's dprobit command. Complete regression results available upon request.

  • 15

    The June–September trimester has the most accessions and the best quality, because most high school graduations occur in June.

  • 16

    Throughout the sample period (and prior to DoD's release of the 2008 DTM), dependent waivers were required for recruits with two non-spousal dependents in all services except the Marine Corps. In the Marine Corps, those married with two children were ineligible for a waiver, while those separated/divorced would be considered, based on custody/support obligations.

  • 17

    As of FY09, the Marine Corps no longer offers DAT waivers.

  • 18

    This methodology was first adopted by Golfin (2006).

  • 19

    On request, the author will provide complete regression results.