Institutional Archive of the Naval Postgraduate School

Calhoun: The NPS Institutional Archive DSpace Repository

Theses and Dissertations 1. Thesis and Dissertation Collection, all items


An analysis of the effect of reserve participation and training on civilian employment and earnings

McGuire, John A.

Monterey, California. Naval Postgraduate School http://hdl.handle.net/10945/39871

This publication is a work of the U.S. Government as defined in Title 17, United States Code, Section 101. Copyright protection is not available for this work in the United States.

Downloaded from NPS Archive: Calhoun

Calhoun is the Naval Postgraduate School's public access digital repository for

D U DL EY research materials and institutional publications created by the NPS community. «iit Calhoun is named for Professor of Mathematics Guy K. Calhoun, NPS's first

NY KNOX appointed -- and published -- scholarly author.

LIBRARY Dudley Knox Library / Naval Postgraduate School 411 Dyer Road / 1 University Circle

http://www.nps.edu/library Monterey, California USA 93943

31 nt i ii i




March 1993

Principal Advisor: Stephen L. Mehay Associate Advisor: Gregory G. Hildebrandt

Approved for public release; distribution is unlimited.


OTIC 3" 2 7 caf) 2:

| 93-11969 9 ces


Security Classitication of this page REPORT DOCUMENTATION PAGE

Ja Report Security Classification: Unclassified Ib Restrictive Markings

3 Distribution/Availability of Report Approved for public release; distribution is unlimited.

5 Monitoring Organization Report Number(s)

6a Name of Pertorming Organization 6b Office Symbol Ta Name of Monitoring Organization ¥ Naval Postgraduate School (if applicable) 36 Naval Postgraduate School

6c Address (city, state, and ZIP code) To Address (city, state, and ZIP code} Monterey CA 93943-5000 Monterey CA 93943-5000

8a Name of Funding/Sponsoring Organization 6b Office Symbol 9 Procurement Instrument Identification Number (if applicable)

10 Source of Funding Numbers Program Element No Work Unit Accession No


12 Personal Author(s} John A. McGuire

13a Type of Report 13b Time Covered 14 Date of Report fyear, month, day) 15 Page Count Master's Thesis From To 1993, March 64

16 Supplementary Notation 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.

17 Cosati Codes 18 Subject Terms (continue on reverse if necessary and idenufy by block number) Subgroup Reserve Training, Reserve Affiliation, Two-Stage Least Squares, Reserve Components aT

19 Abstract (continue on reverse if necessary and identify by block number)

Utilizing data from the 1986 Reserve Components Surveys, this thesis implements a test of the hypothesis that a positive relationship exists between occupational training received in the reserves and increased benefits and wages on reservists’ civilian jobs. The null hypothesis was that no such relationship exists, or that it exists for relatively few reservists, so that reserve participation is mainly a form of moonlighting with few spillover benefits to the individual or society in the form of increased worker productivity. Log-earnings regression equations were specified to test the basic hypothesis. The two-stage least squares (2SLS) estimating technique was utilized to estimate the models due to the existence of simultaneity bias in the regression equations. It was determined that affiliating with the reserves to receive training results in an increase in civilian benefits and wages. Therefore the null hypothesis was rejected, leading to the conicusion that reserve training does appear to provide important benefits to some enlistees, namely those who are motivated to seek skill training that can be used on their civilian job or used to find a better civilian job.

20 Distribution/Availability of Abstract 21 Abstract Security Classification a _X_ unclassified/unlimited __ same as report _. DTIC users | Unclassified 228 Name of Responsible Individual 22b Telephone (include Area Code) 22¢ Office Symbol Stephen L. Mehay (408) 656-2643 AS/MP DD FORM 1473,84 MAR 83 APR edition may be used until exhausic security classification of this page Al! other editions are obsolete Unclassified


Approved for public release; distribution is unlimited. AN ANALYSIS OF THE EFFECT OF RESERVE


by John A. McGuire Lieutenant, United States Navy B.S.E., Florida State University, 1987

Submitted in partial fulfillment of the requirements for the degree of


from the



John A. McGuire

Approved by:

David R. Whipple, Chaiyman Department of Administrativé Sciences



Utilizing data from the 1986 Reserve Components Surveys, this thesis implements a test of the hypothesis that a positive relationship exists between occupational training received in the reserves and increased benefits and wages on reservists’ civilian jobs. The null hypothesis was that no such relationship exists, or that it exists for relatively few reservists, so that reserve participation is mainly a form of moonlighting with few spillover benefits to the individual or society in the form of increased worker productivity. Log- earnings regression equations were specified to test the basic hypothesis. The two-stage least squares (2SLS) estimating technique was utilized to estimate the models due to the existence of simultaneity bias in the regression eyuations. It was determined that affiliating with the reserves to receive training results in an increase in civilian benefits and wages. Therefore the null hypothesis was rejected, leading to the conclusion that reserve training does appear to provide important benefits to some enlistees, namely those who are motivated to seek skill training that can be used on their

civilian job or used to find a better civilian job.

Accession For

a = e -





INTRODUCTION . . . -.- 2...) eee ee eee A. RESERVE HISTORY ... .












wow O90 AQ wD p




B. RECOMMENDATIONS ... . REFERENCES . . . . . 2 e- sw ee






The United States military reserve system is deeply rooted in American history. From early colonial days, up to and including the Spanish American War of 1898, the reserves coexisted with their professional counterparts, the continental or standing army. There were relatively few changes made to the reserve system during this period primarily due to the success of the militia. Those changes that were made were done so with no regard for the resultant efficiency of the militia. Military reformers at the time of the Spanish American War were well aware of the conscription systems in use by Furopean powers. These systems drafted men into the active army and then involuntarily assigned them to reserve mobilization billets. This provided a readily available pool of trained soldiers for periods of mobilization. (Sullivan, 1985]

The success of the reserves in the Spanish American War made changing the reserve system a difficult task. America was fast becoming a world superpower and took on a more internationally political role during the first few years of the nineteenth century. This active role in international politics required a larger, more effective military force.

Subsequently, this also required a more centralized and

better trained reserve force than the pre-1898 reserve system could provide. Beginning in 1903 Congress enacted three acts that provided for a large, voluntary standing force, a reserve force to be used primarily for support missions, and a national guard for combat and civil disturbance missions. These three acts, the Dick Act of 1903, and the National Defense Acts of 1916 and 1920, while amended several times over the years, have provided the structural framework found in America’s military system



There are basically two classifications of reservists: those who belong to the Selected Reserves (SELRES) and those who belong to the Individual Ready Reserves (IRR). The IRR is made up of individuals who have served less than six years in the active or selected reserve forces and have residual military service obligation (MSO). Most first term enlistments require the merber to obligate him/herself for eight years from the date of enlistment. The IRR are not organized into units and members do not receive periodic training or pay.

The Selected Reserves are organized into specific units whose primary mission is to provide combat, combat support,

and combat service support units that can be mobilized

quickly in wartime. The same reserve units may be used for

civil disturbance missions during peacetime. These "weekend

warriors" are divided into the Army and Air National Guard and the reserve components of the Army, Navy, Air Force, and Marines. The Selected Reserves are required to drill one weekend per month and one two-week period annually. Selected reservists are paid for both their monthly drills and the annual training concurrent with their paygrade. Analysts have often referred to selected reservists as "moonlighters" since they normally hold a primary full- or part-time job in addition to their reserve duty. Moonlighters are characterized as individuals who participate in the secondary labor market in an effort to increase their earnings. The amount of moonlighting hours a worker provides is directly related to the worker’s primary hours. If a worker is unable to work the amount of hours at his/her primary occupation that he/she desires, then the worker is "underemployed." Disregarding any additional costs of securing a second job, an underemployed worker will moonlight to enhance his/her total earnings. Since reserve pay is essentially a fixed wage per training day, reservists are not able to choose the amount of moonlighting hours they wish to work. This study will focus primarily on the Selected Reserves since there is no moonlighting benefit derived from being a member of the Individual Ready



Members of the Selected Reserves are afforded the same initial training opportunities as their active duty counterparts upon enlistment. A member is offered a choice of several military occupational specialties (MOS) depending on his/her scores on the Armed Services Vocational Aptitude Battery (ASVAB). The reservist first goes to boot camp or basic training and, if qualified for a technical occupation, moves on to advanced training in a formal service school to learn his/her military occupational specialty. There is no difference in the training provided the reservist and the

active duty member at the time of enlistment.


Analysts have often argued that the reserves benefit from enlisting members wicse primary motivation for participating in the reserves is to moonlight and to earn extra income. Some of these reservists are already trained in what eventually becomes their military occupational specialty. Both those with and those without prior training participate in the reserves as a way of supplementing their earnings.

On the other hand, there may be members, mostly non- prior service, who enlist in the reserves in order to receive training in an occupational skill to enhance their

opportunities to obtain civilian employment and to boost

their long-run civilian earnings potential. Those who use the reserves in this way become better trained, more productive, and more employable civilian workers. As a result, there is a spillover benefit to society. If, however, the primary reason for enlistment in the reserves is to augment one’s civilian income, reserve participation is simply a second job or a moonlighting activity with few indirect, or secondary, benefits to the individual and to society. Of coures, the increase in earnings for those who moonlight in the reserves may initially exceed the increase in earnings for those who participate for traininy. But moonlighting tends to be a temporary phenomenon and ends when one’s enlistment ends. For those who are trained in the reserves, benefits on their primary civilian job may continue throughout their worklife.

This thesis will attempt to test the alternative hypothesis that a positive relationship exists between undergoing reserve training and benefits and wages on one’s primary civilian job for some reservists. The null hypothesis is that no such relationship exists, or that it exists for relatively few reservists so that reserve participation is mainly a form of moonlighting with few derivative benefits to the individual or to society. If a positive relationship is observed between reserve training

and civilian occupations and compensation, this research

will also measure the magnitude of the benefits of joining

the reserves to non-py.1-c¢ service high school graduates

entering the labor force, which should provide valuable

information to manpower planners and recruiters.



Very few prior studies have treated reserve participation as a means of obtaining a civilian job or the training needed to obtain a civilian job. It is generally accepted that the reserves benefit by enlisting members from the civilian sector who are already trained in an occupational specialty. But, what about those reservists who had no prior skill training and were employed in an unskilled occupation or were unemployed when they enlisted?

In the 1986 Reserve Components Survey (RCS) over 18,000 (27 percent) of the more than 65,000 respondents stated that they joined the reserves to obtain skill training to help them get a civilian job. Presumably, respondents that either were not employed, or were employed only part-time, would be most likely to cite this reason for enlistment; they also would be the most likely to obtain a "Spillover" benefit from reserve participation. However, others who were employed in an occupation in which they did not intend to remain would also stand to gain by receiving training that would allow them to switch to a better occupation. In the absence of previous studies in this area, this study examines the hypothesis that a positive relationship exists

between training obtained in the reserves and increased

wages and benefits to reservists in their primary civilian



There are virtually no other second jobs which come to mind that parallel the characteristics of reserve participation. Burright, Grissmer and Doering (1982) found three requirements of reserve participation that set it apart from other second jobs and voluntary activities. First, during ennual training, reservists must spend 14 days of full-time work during the summer. This requires their absence from home and from their primary civilian full-time job. Non-prior service rese:vists must, upon entry, train full time for four months in their occupational specialty. Additionally, during periods of national crisis, such as Operation Desert Storm, or civil emergencies, reservists may be called-up to full-time duty.

The obligation to serve full time during summer ACDUTRA does not necessarily represent a cost to reservists, especia’’y if their full-time military pay exceeds their civilian pay. David Grissmer, Richard Buddin, and Sheila Nataraj Kirby (1989) found that over 50 percent of the respondents in the RCS would face moderate or serious decreases in total income if mobilized for thirty days or more. If the training recaived in the reserves is transferable to a job opportunity in the civilian sector, then any costs associated with full-time duty are reduced.


Second, the reservist is legally obligated for up to eight years of service. Civilian second jobs do not normally require such an employment contract. This requirement provides job security to some reservists, while for others it represents an opportunity cost because it reduces the possibility of holding other secondary jobs.

Finally, Burright et al. determined the inflexible work schedule of reserve participation differs significantly from the work schedules of most moonlighting jobs. In 1982 veservists were paid for either 8 or 16 hours per month with no opportunity for increasing paid hours. Most mandatory drills are scheduled on weekends with no flexibility for alternative schedules to accomodate civilian employer concerns. Burright et al. found that civilian employer attitudes toward the reservist’s participation were major

factors in the reenlistment,enlistment decision.

C. SECONDARY LABOR MARKET PARTICIPATION THEORY Moonlighting has traditionally been treated as a decision to participate in the secondary labor market as a means of supplementing one’s primary job income. Most of the prior studies on reserve participation have treated it as a labor force decision similar to civilian moonlighting {Mehay 1990]. Linda Gorman and George Thomas (1991) hypothesized that reserve membership is part-time employment that competes with leisure time and will usually have lower priority than the member's primary occupation. Stephen L.


Mehay (1990) hypothesized that, contrary to the assumptions

of prior studies that reserve participation and moonlighting are influenced by similar economic factors, different criteria are used in the decisions of reservists and civilians. If the decision to participate in the reserves is distinctly different from the decision to moonlight, then previous studies that have treated them the same will be affected by specification bias [Mehay 1990}. Robert Shishko and Bernard Rostker (1976) simply defined anyone who holds two or more jobs as a moonlighter and thus participates in the secondary labor market. They estimated the moonlighting supply curve with data from the Panel Study of Income Dynamics using the Tobit technique for estimating relationships with limited dependent variables. Applying their definition of moonlighting to a person who holds two part-time jobs would seem to violate the principles of secondary labor market participation. Which job would be considered the primary occupation?

Grissmer et al. found that approximately three-quarters of Army reservists hold full-time civilian jobs in addition to participating in the reserves. They observed that reservists are drawn from the competitive labor market and, as such, the reserves compete with other employers who

provide more flexible hours, perhaps a better wage, and

occasional overtime. If an individual is able to obtain

overtime hours on his/her primary job, he/she will be less likely to either moonlight or participate in the reserves.

Grissmer et al. outlined four components every prospective reservist must consider in the decision to participate: the present and future monetary benefits from reserve service; the non-monetary benefits of reserve service; the monetary opportunity costs from reserve service; and the non-monetary opportunity costs of reserve service. An implicit assumption is that the prospective reservist is able to differentiate between monetary and non- monetary benefits as well as other available alternatives. This may not necessarily be true.

The decision to moonlight is based on several economic factors. Shishko and Rostker theorized that an individual's decision to moonlight is based on whether he/she can work enough hours at his/her primary wage rate to satisfy desired income goals. They identified hours worked on the primary job, the primary wage, the secondary wage, and non-labor income as the most important variables in the moonlighting decision. They proposed that "changes in the primary wage alter the minimum wage necessary to induce people to take a second job." An increase in the primary wage may result in an increase or a decrease in the minimum acceptable secondary wage rate (i.e., the second job reservation wage).

Due to substitution and income effects, an increase in

the secondary wage could result in an increase or decrease


in moonlighting hours worked. This is especially true when

the secondary wage is greater than the primary wage. If

there are constraints on the number of hours a worker may

work on the primary job, secondary jobs may be accepted even - if the secondary wage is less than the primary wage.

Non-labor income only affects hours worked if the desired hours of employment fall below the actual hours. If an individual is working more hours than he desires, a small increase in non-labor income will result in an attempt to reduce the number of hours worked in either the primary or secondary job.

Gorman and Thomas proposed that along with economic factors such as extra income and additional training for future income, there are also psychic factors such as patriotism and camaraderie that are associated with reserve participation. Mehay also challenged the traditional "moonlighting hypothesis" stating that, along with extra income, reserve participation offers the member dynamic training and learning experiences, extensive fringe benefits, camaraderie and other unique features not normally found in civilian moonlighting jobs. Like Mehay, and Gorman and Thomas, Burright, et al. found reserve participation provides non-pecuniary rewards such as camaraderie and a sense of team accomplishment. Burright, et al. also found fringe benefits such as health care, life insurance,

educational benefits, tax benefits, and a cost-of-living-


adjusted pension at the age of 60 after 20 years of service to be major attractions to reserve participation. In an empirical moonlighting study conducted by Compton (1979), it was concluded that the supply of labor for second jobs will increase if: 1. Wages in the second job increase significantly. 2. The person is black. 3. The person is non-urban. 4. Wages in the primary job decrease significantly. 5. The person is not a high school graduate.

6. The number of hours required for the primary job decreases.

7. The person’s spouse is not working or quits working.

8. The person’s non-labor income (interest, dividends, etc.) decreases or is nonexistent.

SOURCE: Motivation For First Term Reserve Reenlistment, Naval Postgraduate School Master’s Thesis by James S. Sullivan Jr., 1985.

In his empirical analysis, Mehay constructed a choice- based sample consisting of reservists and civilians who were already working full-time in a primary occupation and who chose to participate either in the reserves or the secondary labor market. He developed a trichotomous model whereby an individual could moonlight, participate in the reserves, or hold one primary job. He hypothesized that if the model

collapsed to a dichotomous one then the decisions to


participate in the reserves or to moonlight would be assumed to be the same and the individual would be indifferent between reserve affiliation and moonlighting. Mehay concluded that the two decisions are not equivalent. His results did support previous research in that he found participation to be strongly influenced by members’ economic status and job market factors such as unemployment rates and prevailing wage rates in the local geographic area. Mehay also concluded that individuals join the reserves to Supplement their income, obtain skill training, receive fringe benefits and serve their country.

Gorman and Thomas observed many college students who join the U.S. Army Reserve (U.S.A.R.) to help finance their education. The U.S.A.R. is capable of accommodating people who pursue goals not fully compatible with service in the active Army [Gorman, et al. 1991].

Gorman and Thomas were surprised to find that almost 25 percent of their sample joined the Army Reserve intending to transfer to the active Army. It seemed that many high school graduates joined the reserves in order to obtain an "education" in Army life. Without obligating themselves to a full-time active Army job they could see what it would be like and decide if they wanted to transfer to the active Army or remain in the reserves until their commitment was

up. This behavior provides an opportunity to increase the


pool of active Army members by increasing the pool of Army


Gorman and Thomas utilized log-linear models to estimate the probability that a person 18 years of age or younger will choose one of three (author-established) "motives" for enlisting in the reserves. These motives were "earn more money", "self-improvement", and "serve." The serve category included those members who responded they wanted to serve their country or that family tradition warranted serving.

Gorman and Thomas found that college-age recruits in mental categories 1 and 2 joined the reserves to earn more money. Almost 70 percent of those seeking more money did so to pay college expenses.

Gorman and Thomas concluded that the Army Reserve may be an invaluable source of high-quality recruits for the active Army. More than half of their respondents were in mental categories 1 and 2 and, of these, 25 percent stated that they planned to transfer to the active Army. Gorman and Thomas’ findings suggest that reservists enlist fora variety of reasons including self-improvement, the opportunity to earn extra money, and patriotism. Of these, a large percentage enlist with the intent to transfer to the active Army.

Burright, et al. identified five variables that influence the reenlistment decision. For the purpose of

this study these variables also could be applied to the


initial enlistment decision. The variables included net reserve pay, net required days of reserve service, civilian wage rate, number of hours worked on the civilian job, and frequency of overtime opportunities on the civilian job.

Utilizing a logistic regression model, Burright, et al. defined the reenlistment decision by a dichotomous variable assuming the value of one for reenlistment and zero for separation. As expected, higher net reserve wages and fewer net reserve days would increase reenlistment rates. Presumably, these same factors would increase initial enlistment rates. Likewise, higher civilian wages, longer civilian hours, and increased civilian overtime opportunities decreased the probability of reenlistment. Presumably, these factors would also decrea:-: initial enlistment rates.

Finally, Regets (1990) argues that reserve participation may be simply a form of "compensated leisure." He theorized that "moonlighting and compensated leisure models of reserve service generate very different predictions of labor supply behavior". Utilizing two sets of data, a Naval Reserve data set created from personnel records, and an all-service extract of reservists from the Survey of Income and Program Participation, Regets empirically tested both models. He found the compensated leisure model predicted a positive income effect (i.e., increases in non-labor income increase

the quantity of labor supplied to Reserve service). The


moonlighting model however, predicts a negative income effect. With the exception of low-income reservists, Regets’ empirical results supported the compensated leisure model.

In summary, most prior studies have utilized the economic labor market theory of moonlighting to explain reserve participation and reenlistment decisions. However, based on significant differences in the characteristics of reserve jobs and civilian moonlighting jobs, other analysts have questioned whether the formal economic model of moonlighting applies to reserve decisions, or whether reserve participation is a labor market activity at all.

One purpose of this study is to turn the question around and to inguire whether reserve participation is motivated by a desire to upgrade earnings capacity on one’s primary job.

If so, economic factors, such as the level of reserve pay, would not have a strong impact on participations decisions. Rather, the opportunity for skill training, and the prospect that such training would augment future long-run earnings on the primary job, would be the principal motivating factor. The next section of the thesis proposes an empirical test of

this hypothesis.



In January, 1983, the Deputy Secretary of Defense mandated a survey of military families, active and reserve, who were increasingly recognized as important to the retention and preparedness of the United States’ armed forces. Each of the services had previously conducted small-scale studies of its own member families. However, no Single consistent cross-service data set was available to permit the study of emerging DoD family issues. The DoD also needed to assess the various impacts of numerous personnel policies that had been implemented in the early 1980’s.

The Assistant Secretary of Defense for Manpower, Reserve Affairs and Logistics established a DoD-wide committee, the Family Survey Coordinating Committee, to assess the information requirements and data sources needed to survey both the active and reserve components of the military (Hunt, et al. 1986]. Due to the complexity of surveying both components, the Committee initiated active force surveys in 1985 but temporarily postponed the reserve surveys. The Reserve Components Surveys (RCS) were not

completed until 1986. The Defense Manpower Data Center

(DMDC) was contracted to conduct both surveys.

The data provided by the 1986 Reserve Components Surveys made possible xesearch on patterns of previous active and reserve service, financial issues that would face Guard and Reserve families during periods of mobilization, reserve compensation and career intentions, the relationships between civilian and military occupations for reserve members, and numerous other topics. The RCS is also the

primary data source for this thesis.


The Reserve Components Common Personnel Data System (RCCPDS) as of 30 October 1985 was used to initially define the population on which the samples were based. This data system contained administrative information on all members of the reserves. The 1986 RCS consisted of Selected Reserve trained officers and enlisted personnel who had already completed training. Members in the training pipeline were not included. As a result, the target population was 9 percent smaller than the total population of the Selected Reserve. [Hunt, et al. 1986]

Survey packages containing questionnaires and related materials were mailed directly to approximately 15,000 reserve units in the United States and Puerto Rico. Each unit had, on average, 7-10 survey participants. The number of survey participants per unit ranged from one or two to 50

or more.


The basic sample selected for the RCS consisted of a

total of 109,067 officer and enlisted personnel. Individuals who participated in the 1979 Reserve Force (RF) Follow-up Survey were included which brought the total number of participants to 120,787.

Data collection began in February, 1986 with the mailing of the initial notification letters to units containing sampled individuals. The last questionnaires were not received for processing until June 1986. Most of the

questionnaires were received in March and April 1986.


The Family Survey Coordinating Committee consisted of representatives from each of the reserve components, the office of the Deputy Assistant Secretary of Defense (Guard/Reserve Manpower and Personnel) and technical experts from DMDC. The Committee identified various subject areas from previous studies which would be important to reevaluate as well as new areas for which survey data would be helpful.

After the Committee reached agreement on the content of the survey questionnaires, DMDC prepared draft questionnaires. Numerous pretests were conducted in iterative fashion. These tests were administered to selected officers, enlisted personnel and spouses. Any problems or deficiencies found in previous tests were

corrected or modified prior to the next test.


As a result of numerous pretests, the questionnaire underwent considerable refinements. In final form it contained questions pertaining to military background,

{(l.e., reserve component, paygrade, number of active duty years, etc.), "future" military plans, military training, benefits and programs, individual and family characteristics, civilian work, and military life. All of the survey respondents were provided with the opportunity to make additional comments or recommendations on all topics, whether or not the topic was included in the survey

questionnaires. {Hunt, et al. 1986]


The most logical approach to assess response rates would be to compare the number of questionnaires mailed out with the final numbers received. Table 3.1 provides a breakdown of response rates by reserve component. This table is adapted from the 1986 RCS User’s Manual and Codebook. In the table, the column labeled "Frame Count" refers to the number of reservists in the population, the column labeled "Selected" is the number chosen to participate using the RCCPDS, "Eligible" is the number of reservists still assigned to the same unit they were assigned in 1985, and "Responding" is the actual number of reservists who responded to the survey. The unadjusted rates do not account for the fact that some individuals who had been selected for participation from the 30 October 1985


TABLE 3.1 1986 RESERVE COMPONENTS SURVEYS RESPONSE RATES FOR MILITARY MEMBERS, BY RESERVE COMPONENTS Unadjusted Adjusted Reserve Frame Response Response Component Court Selected Eligible Responding Rate Rate

Rank Group «= Officer

USAR 53567 6006 5056 3608 60.1 71.4 USAFR 15710 1809 1611 1331 73.6 82.6 ARNG 42139 4421 3922 2810 63.6 71.6 ANG 13027 1393 1333 1124 80.7 84.3 USMCR 3279 1363 1225 965 70.8 78.8 USNR 22838 2456 2126 1685 68.6 79.3 Subtotal 150560 17448 15273 11523 66.0 75.4

Rank Group = Enlisted

USAR 204321 25391 19704 9640 38.0 48.9 USAFR $7955 5783 4960 3565 61.6 71.9 ARNG 356982 42300 36636 21034 49.7 57.4 ANG 92574 9251 8593 6991 75.6 61.4 USMCR 32853 6562 5414 3333 50.8 61.6 USNR 100653 9898 8132 4893 49 .4 69.2 Subtotal 845338 99185 83439 49456 43.9 59.3

Reserve Components

USAR 257888 31397 24760 13248 42.2 53.5 USAFR 73665 7592 6571 4896 64.5 74.5 ARNG 399121 46721 40558 23844 51.0 58.8 ANG 105601 10644 9926 8115 76.2 81.8 USMCR 36132 7925 6639 4298 54.2 64.7 USNR 123491 12354 10258 6578 §3.2 64.1 Subtotal 995898 116633 98712 60979 $2.3 61.8

Source: Description of Officers and Enlisted Personnel in the U.S. Selected Reserve: 1986, Defense Manpower Data Center, Washington, D.C.


RCCPDS were no longer members of the unit to which the

questionnaires were sent at the time of actual data

collection. Individuals may have separated, transferred to

an active component, or transferred to another reserve unit.

Upon examination of Table 3.1 it can be seen that the

unadjusted response rates for all components except the Army

are over 50 percent. Since the Army comprises 65 percent of

the total DoD sample selected, its response rate lowers the

overall unadjusted DoD response rate to approximately 52


Of the 120,787 individuals originally selected, only

102,267 were still in an active drilling status in the

reserves. After adjusting for the "missing" members, the

adjusted response rates, shown in Table 3.1, were

substantially higher tha. the unadjusted response rates.

The unadjusted response rate is calculated by dividing the

responding members by the number of members selected for the

survey. The adjusted response rate is calculated by

dividing the responding members by the number of eligible

members (i.e., those still in a drilling status). The

overall DoD response rate increased to 62 percent. For this

reason, the dataset available for analysis consisted of

60,120 observations from the officer and enlisted

communities of guard and reserve units from the Army, Air

Force, Navy, Marine Corps and Coast Guard. For the purposes

of this study all members of the Coast Guard were dropped


from the analysis. The Army Reserve response rates were considerably lower than the other components perhaps due to

the greater mobility of Army reservists.


In order to eliminate those members who may have been motivated to enlist in the reserves to avoid the draft, only members who joined after the end of conscription in 1972 were included in this study. In addition, two other limitations were placed on the sample. To study the effects of reserve training on civilian earnings, only non-prior service members holding full-time civilian jobs were included. Prior service members were omitted because it is less likely that they will view the reserves as a source of valuable training because they will have already been trained in an occupation through their active duty service. After imposing these limitations on the models, and allowing for observations with missing variables, the sample size decreased to 7,377 observations of enlisted members only. Finally, officers were deleted from this study because most, if not all, are college graduates and have far different earnings capacities than enlisted personnel.

A standard human capital earnings equation was specified and estimated using Two-Stage Least Squares (2SLS) techniques. The natural log of the individual's annual income was used as the dependent variable. The coefficients

of the independent variables could then be interpreted as


the percentage change in the income of the individual for a unit change in the independent variable. The models were

specified as:

LNENGS=a+ZB ,X,+B,+e€ (1)

LNENGS=a+L6B ,X;+B +e (2) where,

LNENGS = log of annual earnings = vector of explanatory variables summarized in Table 3.2 = dummy variable representing those who cite reserve training as reason for reserve participation S = dummy variable representing those who cite supplementing their income as reason for reserve participation e = a random error term that is normally distributed with with mean zero and a constant variance

xX; R

Because members who enlist in the reserves do so voluntarily, and for various reasons, regression models estimated by Ordinary Least Squares (OLS) may be affected by selectivity bias. Members may "self select" themselves to obtain reserve training to increase their civilian earnings. Previous studies have shown that the probability of reserve participation decreases as civilian earnings increase. People with lower civilian earnings may be more likely to join the reserves to receive training in order to increase their potential to obtain better jobs and to increase their civilian earnings. On the other hand, others may join the reserves with the intent of supplementing their income. OLS


models assume one-way causality in which acquisition of

reserve training causes an increase in civilian earnings.

If two-way causality exists, using OLS to estimate an earnings model would violate the classical assumption of the regression model which states that all explanatory variables are uncorrelated with the error term. However, the models, as specified, may involve correlation of "reserve training" (RECTNG) or "Supplement income" (SUPPINC) with the error term because of the two-way causality between LNENGS and RECTNG or SUPPINC. This interaction between earnings and the reason for reserve participation creates a simultaneity bias. If LNENGS and RECTNG or SUPPINC are simultaneously determined, the expected values of the OLS-estimated structural coefficients are not equal to the true coefficients (b’s). Two-way causality may occur because RECTNG is a function of all the other explanatory variables including LNENGS. It is likely the lower one’s annual income, the higher the probability that one will choose to receive training in order to increase human capital and subsequently annual income. This relationship is shown in equation (3). On the other hand one’s annual income is a function of many other explanatory variables including the extent of one’s training. Those who have had formal training are more likely to have a higher annual income than those who have not. This relationship is shown in equation



In the empirical section, a test for simultaneity is performed.

When simultaneity bias is present, the Two-Stage Least Squares (2SLS) technique can be used to generate efficient estimates of the parameters in the models. The 2SLS method eliminates simultaneity bias by substituting an instrumental variable for the endogenous variable that is correlated with the error term, in this case, RECTNG or SUPPINC. An endogenous variable is any variable that is simultaneously determined with any otier variable. The instrumental variable must be a good proxy for the endogenous variable and be independent of the error term.

One’s earnings on the civilian job are hypothesized to be affected by the desire to receive training in the reserves, or one’s desire to supplement his or her income, as displayed in equations (1) and (2). However, it may also be true that the desire to obtain training or supplement one’s income will be greater the lower one’s annual income. This simultaneity suggests the following two-equation system

(written for RECTNG) :

RECTNG=0.+C,LNENGS+2D,Z,+€, (3)


The instrumental variables approach involves estimating each

endogenous variable, RECTNG and LNENGS, as a function of the


exogenous variables in the system, Z; and X; and substituting the fitted values of RECTNG and LNENGS back into equations (3) and (4). In order to test the hypothesis that reservists join to supplement their income, the same approach is taken and the variable SUPPINC is substituted in

place of RECTNG in equations (3) and (4).

E. VARIABLE DEFINITION Table 3.2 contains the definitions of all of the

variables used in the model as well as the occupational variables utilized. Two dummy variables were created for the purpose of this study to test the hypothesized effect of reserve training. These variables were "participate to