Evaluating the Effectiveness of the Job Corps in the South

Material Information

Evaluating the Effectiveness of the Job Corps in the South Does It Work Where It Matters the Most?
Mansoor, Abu
Place of Publication:
[Gainesville, Fla.]
University of Florida
Publication Date:
Physical Description:
1 online resource (56 p.)

Thesis/Dissertation Information

Master's ( M.S.)
Degree Grantor:
University of Florida
Degree Disciplines:
Food and Resource Economics
Committee Chair:
Flores-lagunes, Alfonso
Committee Members:
House, Lisa O.
Graduation Date:


Subjects / Keywords:
Control groups ( jstor )
Employment ( jstor )
Employment statistics ( jstor )
High schools ( jstor )
Hispanics ( jstor )
Job training ( jstor )
Net income ( jstor )
Poverty ( jstor )
School dropouts ( jstor )
Vocational education ( jstor )
Food and Resource Economics -- Dissertations, Academic -- UF
abu, alfonso, control, corps, difference, effectiveness, efficacy, experiment, flores, florida, intention, job, lagunes, linear, mansoor, nation, national, random, randomized, regression, social, southeast, treatment
Electronic Thesis or Dissertation
born-digital ( sobekcm )
Food and Resource Economics thesis, M.S.


This research evaluates the effectiveness of Job Corps in the state of Florida and the Southeast region of the United States. Job Corps is a federal job training program that provides academic and vocational training to disadvantaged youth. Using data from the National Job Corps Study, which is based on a randomized social experiment, we assess if there is a positive effect of the program on participants in comparison to non-participants. Specifically, we focus on the effects of the program on outcomes such as weekly earnings and employment rates. ( en )
General Note:
In the series University of Florida Digital Collections.
General Note:
Includes vita.
Includes bibliographical references.
Source of Description:
Description based on online resource; title from PDF title page.
Source of Description:
This bibliographic record is available under the Creative Commons CC0 public domain dedication. The University of Florida Libraries, as creator of this bibliographic record, has waived all rights to it worldwide under copyright law, including all related and neighboring rights, to the extent allowed by law.
Thesis (M.S.)--University of Florida, 2009.
Adviser: Flores-lagunes, Alfonso.
Statement of Responsibility:
by Abu Mansoor.

Record Information

Source Institution:
University of Florida
Holding Location:
University of Florida
Rights Management:
Copyright Mansoor, Abu. Permission granted to the University of Florida to digitize, archive and distribute this item for non-profit research and educational purposes. Any reuse of this item in excess of fair use or other copyright exemptions requires permission of the copyright holder.
Resource Identifier:
665167777 ( OCLC )
LD1780 2009 ( lcc )


This item has the following downloads:

Full Text




2009 Abu Isa Mansoor 2


To my Family 3


ACKNOWLEDGMENTS I would like to first thank my parents for bringing me so fa r in life but especially my mother who always believed in me. I would lik e to thank Dr. Lisa House for aiding me in my college career it was with her aid that I was ab le to obtain the funding from the United States Department of Agricultures National Needs Fello wship that allowed me to pursue a masters degree and complete my thesis. I thank Carlos Jauregui for his help in statistics and his knowledge of Stata. I thank the University of Floridas Food and Resource Economics Departments faculty and staff for guiding me in a ll my endeavors. Finally, I want to thank Dr. Alfonso Flores-Lagunes, were it no t for his insight, patience, a nd mentorship I would not have had a thesis topic and a soundi ng board for ideas and thoughts. 4


TABLE OF CONTENTS page ACKNOWLEDGMENTS ............................................................................................................... 4LIST OF TABLES ...........................................................................................................................6LIST OF FIGURES .........................................................................................................................7ABSTRACT ...................................................................................................................... ...............8 CHAPTER 1 INTRODUCTION ................................................................................................................ ....9A Brief History of Job Corps .................................................................................................. ..9Underlying Goal of Job Corps ................................................................................................12Previous Studies About Job Corps .........................................................................................14Problem Setting ............................................................................................................... .......18Research Question ..................................................................................................................19Study Overview ......................................................................................................................202 THE NATIONAL JOB CORPS STUDY, THE DATA EMPLOYED, AND THE REGIONS OF INTEREST .....................................................................................................22National Job Corps Study ...................................................................................................... .22The National Job Corps Study Sample Design .......................................................................22NJCS Sample Integrity ......................................................................................................... ..23The National Job Corps Study Estimation Techniques ..........................................................24National Job Corps Study Conclusions ..................................................................................25Demographic Background of the Unit ed States Relative to NJCS .........................................26Unemployment Rates Nationally, in the Southeast, and in Florida ........................................27Education and Income Levels of the Nation relative to the Southeast and Florida ................28Current Study ..........................................................................................................................323 MODEL ..................................................................................................................................404 RESULTS ..................................................................................................................... ..........435 CONCLUSION AND IMPLICATIONS ................................................................................51LIST OF REFERENCES ...............................................................................................................54BIOGRAPHICAL SKETCH .........................................................................................................56 5


LIST OF TABLES Table page 1-1 Annual Unemployment Rate: Years 1994-1995 ................................................................212-1 National Unemployment Rate: Years 1994-1995 By Ages, Sex, Race/Ethnicity ............342-2 Population Demographics for 1990 ...................................................................................342-3 Enrollment and Educational Attainment Poverty Status In 1994 of Persons 16 to 24 Years Old ...........................................................................................................................352-4 The National Population 18 to 24 Y ears Old by High School Graduate Status, Attainment, Sex, and Race/Ethnicity: Year 1995 .............................................................362-5 High School Status Dropout Rates and Percent in Poverty by Nation, Southeast, Florida In 1995 ............................................................................................................... ....362-6 Industry Employment Categories and Av erage Annual Earnings 1994 by National, Southeast, Florida ..............................................................................................................374-1 Summary Statistics for Control and Treatment Groups National, Southeast, and Florida ....................................................................................................................... .........474-2 Summary Statistics for Control and Treatment Groups National, Southeast, and Florida: Employment and Earnings ..................................................................................484-3 Linear Regression Results for Earnings in Quarter 16: National, Southeast, and Florida ....................................................................................................................... .........494-4 Linear Regression Results for Percent Empl oyed in Quarter 16: National, Southeast, and Florida .........................................................................................................................50 6


LIST OF FIGURES Figure page 2-1 Population Demographics for 1990 ...................................................................................382-2 The National Population 18 to 24 Years Old by High School Graduate Status, Attainment, Sex, and Race/Ethnicity: Year 1995 .............................................................382-3 Industry Employment Categories and Av erage Annual Earnings 1994 by National, Southeast, Florida ..............................................................................................................39 7


8 Abstract of Thesis Presen ted to the Graduate School of the University of Florida in Partial Fulfillment of the Requirements for the Degree of Master of Science EVALUATING THE EFFECTIVENESS OF THE JOB CORPS IN THE SOUTH: DOES IT WORK WHERE IT MATTERS THE MOST? By Abu Isa Mansoor May 2009 Chair: Alfonso Flores-Lagunes Major: Food and Resource Economics This research evaluates the effectiveness of Job Corps in the state of Florida and the Southeast region of the United States. Job Corps is a federal job training program that provides academic and vocational training to disadvantag ed youth. Using data from the National Job Corps Study, which is based on a randomized social experiment, we assess if there is a positive effect of the program on particip ants in comparison to non-partic ipants. Specifically, we focus on the effects of the program on outcomes such as weekly earnings and employment rates.


CHAPTER 1 INTRODUCTION A Brief History of Job Corps The Job Corps program was created as a majo r initiative of Presid ent Lyndon B. Johnsons War on Poverty. It was enacted through the Econ omic Opportunity Act, which was signed into law on August 20, 1964. The programs directives were to help underprivileged youth break the cycle of hopelessness they faced in their lives of menial work, frequent unemployment, crime, and deprivation (Levitan and Johnston, 1975). This was done by removing underprivileged youths aged, 16 to 21, from their poverty stricken e nvironments to distant re sidential centers. At these centers, they were provided educational and vocational training to improve and develop their skill set for employability. This program was not based on an innovative e ffort, but rather on prior social/economic initiatives and studies conducte d by the Civilian Conservation Corps (CCC) and the Farm Security Administration of the 1930s. In a ddition, community outreach and development programs established by the Ford Foundation in th e 1950s and the welfare reforms by the U.S. Department of Health, Education, and Welfare helped lay a foundation for the fledgling Job Corps. Furthermore, previous studies, social experiments, social policie s, and agencies helped establish frameworks and parameters for the ne w agency. Job Corps was also influenced by scholarly studies that found variab les such as economic growth, mental health, racial and ethnic biases, illiteracy, and family lifestyles were in extricably linked to the development and wellbeing of constructive members of society (L evitan and Mangum, 1969). Job Corps shares similarities with the CCC, which was used during the Great Depression as a job-creation tool that put 1.5 million enrollees from a cross-section of so ciety to work. However, one main difference between CCC and Job Corps is the focus on training. The enrollees in CCC were youthful and 9


were involved in projects that worked toward conservation, such as improving millions of acres of federal and state lands and parks during a time of unemploymen t. They also built roads, telephone lines, and planted over a million trees. Job Corps genesis did not come during a time of great unemployment, but rath er during a time when overall unemployment was low, however, youth unemployment was rising (Levitan and Mangum, 1969). Awareness of the problem of youth unemploymen t and its associated problems kept the idea of Job Corps alive. In 1963, Presiden t John F. Kennedys Task Force on Manpower Conservation published a report, One-Third of a Nation which stated that the armed forces rejected one out of three dr aftees during World War II becau se of mental and physical deficiencies. Most telling of th e report was that the rejected came from underprivileged areas and homes. This gave strength and interest in making Job Corps a viable way to improve disadvantaged youths chances of entering the labor market (Levitan and Johnston, 1975). There was much debate on what type of centers w ould be established. Ultimately, three were established, two for men and one for women. Rural centers were catered to youths with educational difficulties, who would receive basic educational instruction and work experience on conservation projects. Urban cen ters provided vocational training to youths that had at least a sixth grade education. The third type of center for young women provided training that was clerical in nature (L evitan and Johnston, 1975). In 1966, Job Corps came under fruition under th e control of the Office of Economic Opportunity (OEO). The program consisted of six large mens centers each serving 1,000-3,000 individuals, more than 80 small conservation centers (exclusively for men), and 17 womens centers of medium size where each served 300-1,000 individuals. These centers, in addition to providing training and housing, prov ided health care and counseling. The combined centers had 10


a peak of 42,000 enrollees in 1966 (Levitan and J ohnston, 1975). Job Corps was never without controversy. During Johnsons presidency it was continuously challenged by problems such as having a high dropout rate, misbehavior at th e centers, managerial disputes, community hostility toward nearby centers, difficulty in find ing suitable locations for centers, high financial costs, and congressional and political opposition (Levitan and Mangum, 1969). When President Richard M. Nixon took office in 1968, he began a process that cl osed many Job Corps Centers and cut the program's budget. In 1969, he a nnounced the closure of 50 of 82 conservation centers, two of six mens urban centers and seven of 17 women s centers, which reduced overall enrollment to 22,000. The agency was later abso rbed into the Manpower Administration of the Department of Labor. After the absorption, there was a shift from one of the founding assumptions: disadvantaged youths have to be removed from an environment that was not conducive to their personal growth. Job Corps Ce nters that had distant residential areas were closed and replaced with centers that were with in commuting distance of the participants homes. Over the next four years, Job Corps added 25 new centers to replace those which had been closed. New centers were dubbed Residentia l Manpower Centers (RMCs) and Residential Support Centers (RSCs). Residential Manpower Centers provided educational and vocational training, with enrollees drawn from the local ar eas. Whereas Residentia l Support Centers were located in urban residential areas, and offere d counseling without in-house educational and vocational training (Levitan and Johnston, 1975). In addition, vocational and technical training largely replaced general education or re medial education classes (McCarron, 2000). Currently, Job Corps provides food, shelter, job training, work clothes, stipend, and healthcare to teenagers and young adults at 123 campus centers in the United States. It is broken into six regions: Atlanta, Boston, Chicago, Dalla s, Philadelphia, and San Francisco. The Job 11


Corps provides general and vocatio nal education, technical training, and useful work experience at residential centers for youths from impoverished backgrounds (McCarron, 2000). While 46 states have at least one center, the Job Corps progr am capacity differs among the states because the number and size of Job Corps cente rs vary from state to state. For example, 19 states have one Job Corps center, whereas four states have six or more centers. Moreover, Job Corps program capacity in a state is not rela ted to the number of state residents enrolled in Job Corps and participants are not necessarily trained or assigned to a center in their home state (U.S. Government Accounting Office, 1996). Within these six regions, the U.S. Depart ment of Labor (DOL) contracts out center operations, recruiting and screening of new students, and placement of students in jobs and other educational opportunities afte r they leave the program. The U. S. Departments of Agriculture and Interior operate 30 centers, cal led Civilian Conservation Centers (CCCs). About another 80 centers are operated by private contractors a nd are administered thr ough contracts with Job Corps regional offices. Recruitment and placem ent are also administer ed within the regional offices (Burghardt et al., 1999). Underlying Goal of Job Corps The aim of Job Corps is to impart undereducated and low-skilled youths the necessary skills to be marketable in the work place (U.S. Department of Labor, 2006). This, in turn, should lead to positive outcomes for Job Corps participan ts be it in increased earnings, reductions on the dependence of public assistance, or a decline in antisocial behavior (L ong et al., 1981). Job Corps participants are selected on some of the following eligibility criteria: age (1624), residency status (citizen and/or permanent resident), low income, a high school dropout, requires additional education and/or vocational training, homeless, runawa y, foster child, or a parent (U.S. Department of Labor, 1999). 12


Each year, it serves more than 70,000 new pa rticipants at a cost of about $1.5 billion, which is more than half of all funds spent by the U.S. Department of Labor (DOL) on youth training programs. The typical J ob Corps student is a minority ( 70 percent of all students), 18 years of age, has dropped out of high school (80 percent), and reads at a seventh grade level. The services that Job Corps provides are basic education, vocational training, job placement assistance, residential living (including social skil ls training), counseling, and healthcare. Each of these components are tailored to each participant. The following descriptions draw heavily from the National Job Corps Study: Report on Study Implementation (Burghardt et al., 1999). Education: The goal of the education co mponent is to enable students to achieve educational attainment as fast as their individual abilities permit. Education programs in Job Corps are individualized and self-paced and operate on an open-entry and open-exit basis. The programs include remedial education (emphasizing reading and mathematics), world of work (including consumer educati on, driver education, home and family living, health education, and programs designed for individuals whose primary language is not English), and a General Education Deve lopment (GED) program of high school equivalency for students who are academically qualified. Some centers also offer some students the opportunity to atte nd postsecondary education while enrolled in Job Corps. Students are assigned to classes based on the results of diagnos tic tests administered during the first few weeks (Burghardt et al., 1999). Vocational Training: The vocational traini ng programs at Job Corps are individualized and self-paced and operate on an open-entry and open-exit ba sis. Each Job Corps center offers training in several vocational trades typically including business and clerical occupations, health occupations, constructi on trades, culinary arts, and building and apartment maintenance. National labor a nd business organizations provide vocational training at many centers. In many trades, students gain hands-on experience by working on supervised work projects, such as the cons truction or rehabilitati on of buildings either on center or in the community (Burghardt et al., 1999). Health Care and Education: Students rece ive comprehensive health services, including medical examinations and treatment; immunizat ions; dental examinations and treatment (for participants who remain in the program at least 90 days); counseling for emotional and other mental health problems; and instructi on in basic hygiene, preventive medicine, and self-care (Burghardt et al., 1999). Residential Living: Residenti al living is the most distinc tive component of the Job Corps program and distinguishes it from most ot her employment and training programs. The 13


idea behind residential living is that, given the disadvantaged environments from which most participants come, the students requi re a new and more supportive environment to derive the maximum benefits from educati on and vocational traini ng. All students must participate in five formal social skills traini ng activities. The residential living component also includes meals, dormitory life, entert ainment, sports and recreation, center government, center maintenance, and other re lated activities (Bur ghardt et al., 1999). Counseling and Other Ancillary Services: Job Corps centers provide counselors and residential advisers. These staff help st udents plan their educational and vocational curricula, offer motivation, and create a supporti ve environment. Support services are also provided during recruitment, placement, and the tr ansition to regular life and jobs after Job Corps (Burghardt et al., 1999). Placement: The final step in the Job Co rps process is placement. The placement component focuses on helping students find jobs in training-relate d occupations with prospects for long-term employment and adva ncement. Placement contractors are state employment offices or private contractors, a nd some centers perform placement activities. Placement agencies help students find jobs by providing interviewing and resume-writing assistance and job development and referral services. They are also responsible for distributing the readjustment allowance, a st ipend students receive af ter leaving Job Corps (Burghardt et al., 1999). Previous Studies About Job Corps There have been several studies of the J ob Corps program. The most comprehensive study, before the National Job Corps Study in 1999, was the one conducted by Mallar, Kerachsky, Thornton, and Long (1982). This st udy provides the impact of Job Corps on a participants post-program employ ment and earnings by comparing the experiences of a sample of participants with th e experiences of a sample of disadva ntaged youth who did not attempt to enroll in the Job Corps. Their findings indicate that Job Corps students worked an average of three additional weeks per year and had higher earnings than non-participants by approximately $600 per year during the four-year post-program observation period. The baseline surveys in this study were conducted in May 1977 on two groups. The two groups of comparison were youths that were participants in Job Corps a nd those who never applied. The group of youths that never applied to Job Corps did not know about the program, however, they were similar in socio-economic status to particip ants. The baseline survey obtain ed detailed information on the 14


youths concerning: demographic characteristics, so cio-economic backgrounds, work histories, and related activities beginning six months before Job Corps enrollment and continuing up to the date of the interview. Additionally, thr ee follow-up surveys conducted 9, 24, and 54 months after the baseline survey, which contained detail ed information on work histories and related activities during the post-program period after Job Corps partic ipants had been out of the program from 42 to 54 months. In a related study utilizing the same data, L ong et al. (1981) suggest s that the Job Corps program yielded net benefits to Job Corps part icipants and to society. The increase in postprogram output of $3,896/year for each participant and the reduction in crime were the most important benefits. The stipend and room and board that Job Corps pa rticipants received $1,208, greatly aided in their post-progr am output. Although, society had to bear the cost of the transfer payments to Job Corps participants, the reductions in Job Corps participants criminal activities were substantial. Social benefits per partic ipant from reduced criminal activity was $1,962. Through his methodology, they estimated Job Corps to be socially efficient in its use of resources to society and particip ants. In addition, Long et al. (1981) found the estimated value of the reductions in criminal activity, dependen ce on social programs, and other activity in measured benefits were approximately $7,300 pe r Job Corps participan t in 1977 dollars. The total cost of the program to soci ety was estimated to be $5,070 per participant. Therefore, the net benefit was worth approximately $2,230. In 1995, the U.S. Government Accounting O ffice sent a report that questioned the effectiveness of Job Corps in its efforts to use independent National contractors to provide training to the U.S. Senate (U.S. Government Accounting Office, 1995). The study was based on a 1993 survey administered to a random sample of 413 Job Corps students from six centers, 15


who had previously obtained jobs. The study found that, almost half the jobs obtained by students were low-skill jobs not related to th e training provided. However, the students who completed vocational training at these centers were five times more likel y to obtain a trainingrelated job and wages that were 25 percent higher than the average wage paid to individuals who did not obtain training-related jobs, $6.60 versus $5.28 per hour in 1993 dollars. About twothirds of the jobs obtained by students who did not complete their training were in low-skill positions, such as fast food worker, cashier, laborer, assembler, and janitor. Johnson and Troppe (1992) conducted regression analysis on the employability of Job Corps participants who terminated in 1988, based on the effect of educa tional gains and other related outcomes of Job Corps enrollment. The sample consisted of 24,594 students who had data on initial and follow-up reading and math tests. The study found that 26.7 percent of the sample had been employed prior to Job Corp s participation as compared to 72.5 percent employed within 6 months after Job Corps enrollm ent. With the attainment of a GED, the likelihood of employment increased by 6 percentage points. It also found that increases in reading ability increased the likelihood of job placement by 1 percentage point, but the relationship between math gains and employment were weaker. Greenberg et al. (2004) found that the Job Corps program increased earnings by $808, on average, in the year after the training took place. Greenberg evaluated training programs based on 19 evaluations of 13 voluntary government-funded training programs fo r the disadvantaged that operated in the United States between 1964 and 1998. In order to appropriately weigh the estimates to account for the varying statistical precision across the studies, the study weighted each estimate by the inverse of its estimated variance. 16


Another study by Rawlins (1971) focuses on the operations of an urban Job Corps Center in Pleasanton, California. The study was c onducted on 239 participants of Job Corps. Multivariate regression analysis was utilized to estimate the impact of training on earnings and employment. The sample used in this analysis co nsists of a set of individuals whose files were selected at random from those of the participants enrolled at Pleasanton, who terminated before December 1967. From the 450 files that were selected, a sample of 239 remained after incomplete files missing critical information were eliminated. Personal and training characteristics of the participants such as education, entry into the program, length of training, and completion of the training program were utili zed as independent variables. The dependent variable is earnings over the six-month period immediately following the training experience. This paper found that there was a highly si gnificant relationship between earnings and participation in the Job Corps program. On aver age, completion of training translated into an additional $375.00 in earnings over the six-month follow-up period. Williams and Cooper (2004) found on average, fo r every dollar a Job Corps Center spends in its local area, $1.91 in economic activity results within the local area (t he input is local Job Corps Center spending plus local student spending ). Also, for every dollar a Job Corps Center spends overall, an average of $1.53 in economic activity results within its lo cal area (the input is total Job Corps Centers and student spending). Th is can be seen as a multiplier effect of Job Corps Centers. These monetary values were obtained through economic tools called inputoutput models (Williams and Cooper, 2004). These models define a communitys economy as collections of industries and gauge how active thos e industries are by measuring the historic flow of money, goods, and services within a nd between communities throughout the Nation. 17


Further, Williams and Cooper (2004) conducted an economic impact study that used a non-probability sampling procedure in order to obtain a cross-section of Job Corps Centers. Out of the 118 Job Corps Centers, 20 were chosen to achieve a cross-section along the dimensions of size (student capacity), location (urban or rural setting), region (geographi c area of the country), and age (older or newer than 10 years in opera tion). On average, for every job a Job Corps Center provides, the economic activity it stimulat es supports 66 percent of another full-time job in its local area. For every million dollars a Job Corps Center spends overall per year, an average of 23.28 workers are employed in its local area. In summary, most studies that assess the effectiveness of th e Job Corps find a trend that favors it. However, it is hard to synthesize th e evidence given the multiplicity of methodologies and samples employed. More importantly, none of the studies employ a randomized approach that, by construction, makes participants and nonparticipants truly comparable. The National Job Corps Study undertaken by Mathematica Policy Research and the Department of Labor tried to address some of these issues, which will be discussed further in later chapters. Furthermore, we employ data from the National Job Corps Study to analyze the performa nce of the Program in the Southeast and Florida. Problem Setting The main issue this study investigates: Is th ere a causal relationshi p between participating in Job Corps and positive outcomes in the Southeas t United States and the state of Florida? The National Job Corps Study was carri ed out during late 1994 and 1995 to answer this question at the National level, which was a mandate by Congress. Results show positive overall effects such as increase in earnings and employ ment (Schochet et al., 2001). However, does it hold true in the Southeast and in the state of Florida? 18


This study adds to the literature by focusing on the regional differences that Job Corps participants face Nationally, in the Southeast, and in Florida. The impact of program participation will likely be affected regionally by socioeconomic differences, ethnic/racial concentrations, and levels of industry employ ment. The National Job Corps Study results may not be necessarily indicative for the Southeast and Florida, sin ce labor market differences may dilute the effects of training programs on employme nt and earnings at different regional levels. Research Question The primary objective of this thesis is to determine whether Job Corps has appreciable impacts on outcomes such as weekly earnings an d employment rates. Training programs might affect outcomes other than earnings and employment, such as welfare and unemployment compensation payments, crime rates, and feelings of satisfaction. There is a focus on earnings and employment because a major objective of a ll government-sponsored training programs is to increase the economic well-being of participants. This will determine the effectiveness of the program on participants in comparison to non-partic ipants at the National le vel, at the Southeast region, and at the stat e of Florida level. In the National Job Corps Study, National is defined as the 48 contiguous states and the District of Columbia. For the purpose of our anal ysis, the Southeast region consists of Alabama, Florida, Georgia, Louisiana, and Mississippi. The justification for focusing on the Southeast and Florida is due to the un ique characteristics they share in c ontrast to the Nation. For example, during the intake period for pa rticipants of the study during 1994 and 1995, the National annual unemployment rate was 6.1 percen t and 5.6 percent, respectively. On the other hand, it was higher in the Southeast for both years and margin ally lower for Florida in 1995 see Table 1-1. There are higher levels of poverty, high school dr opout rates, and differing race demographics which adversely affect the earnings and employ ment characteristics of the Southeast United 19


States and Florida. In addition, the number of Job Corps Centers va ries by state, Florida has four and Alabama has two, while Georgia, Louisiana, and Mississippi each have three centers. This difference in the number of centers may have im plications regarding th e effectiveness of Job Corps to serve its clients. These characteristics w ill be discussed further in subsequent chapters. Study Overview The current chapter has summarized Job Corps program elements and the characteristics of youths who are eligible for the program. In addition, it discussed some previous research about Job Corps. Chapter two will discuss th e elements of the National Job Corps Study, the data employed, and the regions of interest. The third chapter presents the economical and statistical model used to analyze the data. Chapter four provides results and discussion. Finally, chapter five provides the c onclusions and thoughts on the out come of the research. 20


21 Table 1-1. Annual Unemploy ment Rate: Years 1994-1995 Year National Southeast Florida 1994 6.1% 6.3% 6.7% 1995 5.6% 5.8% 5.5% Source: U.S. Bureau of Labor Statistics, 1994-1995.


CHAPTER 2 THE NATIONAL JOB CORPS STUDY, THE DAT A EMPLOYED, AND THE REGIONS OF INTEREST National Job Corps Study Within the public sphere, there has always b een debate regarding the effectiveness of public assistance programs. Critics argue whether public dollars are spent wisely. Therefore, in order to respond this criticism it is important to show that public programs result in positive measurable outcomes to participants (and ideally to society) beyond just an ecdotal discussions of their impact. However, to conduct such an a ssessment is quite costly and monitoring input productivity and metering the outcome is not easy. In the 1990s, Job Corps faced a number of challeng es, threats, and critic s. Critics charged that the program was wasteful because it wa s spending $26,000 per student, and fewer than 15 percent of participants were completing the prog ram. A 1995 Congressional bill wanted to turn administration of Job Corps over to the states and close a number of centers. Instead, Congress voted that the federal government retain control and close a few of the centers. To determine Job Corps credibility, Labor Secretary Robert Reich commissioned to Mathematica Policy Research, Inc (MPR), the Nationa l Job Corps Study. One controversial element of the study was the intentional denial of admission to one in ever y twelve eligible applicants to use them as a control group. It then paid them $10 each for fo llow-up interviews to study their subsequent fate. The cost of the study was $17.9 million and was conducted over nine years (McCarron, 2000). The National Job Corps Study Sample Design The data used for the National Job Corps St udy (NJCS) comes from a randomized social experiment carried out during the mid-90s. Applicants to Job Corp s between November 1994 through December 1995 were considered for the st udy. The study is considered a fully National 22


sample because of its deployment in the 48 con tiguous states and the District of Columbia. Youths were sampled from all outreach and admi ssions (OA) agencies na tionwide (Burghardt et al., 1999). The sample intake period began on November 17, 1994, and continued for 16 months, until February 29, 1996. During this period, MPR pro cessed information on 113,803 cases in total, of which 80,883 were eligible applicants that met Job Corps criteria. The sample was further narrowed and consisted of 5,977 control group partic ipants (7.4 percent of those randomized), 9,409 treatment group participants (11.6 percent of those randomized), and 65,497 program nonresearch Job Corps members (Burghardt et al ., 1999). Those in the treatment group were allowed to receive Job Corps training and bene fits, while those in the control group were not permitted to receive Job Corps training or benefits for a period of three years, although they were able to enroll in other programs. The outcome s of the control group, th erefore, represent the outcomes of the treatment group if they had not been given the opportunity to enroll in Job Corps. In the study, the sampling rate was se t lower for females who had a high likelihood of being residential students because residential females are difficult to recruit and Job Corps staff were concerned that the study would cause slots for residential females to go unfilled (Burghardt et al., 1999). NJCS Sample Integrity In order to draw valid inferences from th e random assignment study about the effects of Job Corps on post-program outcomes, MPR, DO L, OA had to implement procedures that maintained the integrity of the sampling. In order to do so, randomization procedures were undertaken to maintain the cr edibility of the study. Randomization procedures were determined by Job Corps staff and a process analysis using telephone surveys to OA counselors, mail surveys to Job Corps Centers (JCC), and visits 23


to 23 centers were used to monitor and confirm the outcome (Flores-Lagunes et al., 2008). The outcome of these procedures were that 265 enrollees of the sample frame enrolled in Job Corps before randomization and only 0.6 percent of progr am group participants did not enroll in Job Corps because they were not eligible or not ac cepted into the program. During the intake period, a total of 68 control gr oup members enrolled at Job Corps Centers, which only represents 1.14 percent of all control gr oup members. The crossover of c ontrol group members and participation of ineligible enrollees was mini mized during the sample frame. The Job Corps staff implemented the random a ssignment procedures successfully over the 16 month sample intake period (Burghardt et al ., 1999). Through the end of February 1999, only 1.4 percent of control group member s enrolled in Job Corps before the end of the three-year period during which control group members were not supposed to enroll. As a result, the research sample is representative of the youths in the intended sample frame and the bias in the impact estimates due to contamination of the control group is likely to be small (Burghardt et al., 1999). Further details about the analysis of ra ndomization process can be found in the National Job Corps Study: Report on Study Implem entation (Burghardt et al., 1999). The National Job Corps Study Estimation Techniques The original NJCS program evaluation is based on a differences-in-means estimator, modified to account for non-compliance: individuals in the treatment group who never enroll in Job Corps and individuals in the control group th at enroll in Job Corps before the three-year exclusion (Flores-Lagunes et al., 2008). The NJCS employed difference of means esti mator, which was modified to account for non-compliance such as dropout and crossover of individuals in the trea tment group and control group. This can be interpreted from the following equation: 24


25 (2-1) The term outcome DMcomp refers to the difference in mean outcomes between the treatment group and the contro l group, with respect to th e proportional differences in participation. In general, = 1 0, ( 1) refers to the average out come of the treatment group, ( 0) represents the average outcome for the control group, and the refers to the difference between the two. The non-compliance of individuals is a ddressed by dividing by the proportion of those individua ls in the treatment group who enroll in Job Corps (Ppart) minus the proportion of those individuals in th e control group that enroll in Job Corps before the end of the three-year embargo (Pcross) (Flores-Lagunes et al., 2008). From the 15,386 Job Corps participants in the NJCS, about 11,313 youths (4,485 control group and 6,828 treatment group members) were analyzed that completed the 48th month interview (Schochet et al., 2001). In addition, re gression analysis was used to estimate the impact of Job Corps participation. The multivariate regression model was used to control for any unbalances in factors measured at baseline that affect the outcomes. This form of analysis increases the precision of the estimated program participation and the strength of the significance tests can be measured against the differences-in-means estimations. The study found the differences-in-means estima tion and the regression analysis provided similar results. Qualitatively the same conclu sions can be drawn from both sets of estimates (Schochet et al., 2001). This was a testament to the sample selection process and procedures that enabled quality inferences to be made. National Job Corps Study Conclusions The National Job Corps Study found that J ob Corps was a good investment based on increasing employment and earnings of participants. At the 48th month observation period, Job


Corps participants were 5 percent more likely to be employed and earned an average of about 22 cents more per hour (Schochet et al., 2001). The employment and earning gains were not larger at the inception of the program for treatment group members. They forewent employment and earnings to gain an advantage in education, training, and/or vocational preparation. In year 4 after random assignment, average weekly earnings for program group members we re $16 higher than for control group members ($211, compared to $195). The difference was f ound to be statistically significant at the 1 percent level (Schoche t et al., 2001). Job Corps participation had positive effects on the employ ment rate and time spent employed beginning in year 3 after random assignment. In year 4 after random assignment, the percentage of employment for Job Corps participants was 71.1 percent for the treatment group, compared to 68.7 percent for the control group. This was statistically signif icant at the 5 percent level. Demographic Background of the Unit ed States Relative to NJCS The demographic focus will be primarily on wh ites, blacks, and Hispanics. The U.S. Census Bureau provides statistics on the com position of the U.S. population. In 1990, the white population accounted for 80.3 percent, black was 12.1 percent, and Hispanic 9 percent. Males accounted for 48.7 percent, while females accounted for 51.2 percent (U.S. Bureau of the Census, 1990). The U.S. Census Bureau (2000) repor ts that the percent of the population that is white is 75.1, blacks are 12.3, and Hispanic is 12.5. Males accounted for 49.1 percent, whereas females accounted for 50.9 percent. Since the study took place between late 1994 and 1995, we take an average of the two census years of 1990 and 2000 to make a rough estimate of the U.S. population relative to the Job Corps applican ts. Whites may account for 77.7 percent, blacks 12.2 percent, and Hispanics 10.7 percent. The male to female composition would be 48.9 26


percent and 51.1 percent, respectivel y. Of the eligible applicants to Job Corps, 60 percent were male, whites represented 26 percen t, 50 percent were bl acks, 18 percent were Hispanic, 4 percent were Native American, and Asian or Pacific Islander represented 2 percent. Some employment characteristic s of applicants to Job Corps are that 65 percent of them were employed in the previous year before they were deemed eligible for Job Corps. On average, the hourly wage they received was about $5.10, for both males and females. Overall, average weekly earnings equaled about $180. Thes e characteristics speak to the disadvantage that these youths faced (Schochet, 1998). Unemployment Rates Nationally, in the Southeast, and in Florida The National unemployment rate is telling wh en we categorize the rates by age, gender, and race/ethnicity. From the figures in Tabl e 2-1, the National unemployment rate in 1994 and 1995 was 6.1 and 5.6, respectively. Unemployment fi gures for ages 16 to 19, which is a target segment for Job Corps participation, is quite high at 17 percent. When labor market participants are segmented by ages greater than or equal to 20 and by gender, qualitatively there is no difference. When we segment the labor particip ants by ages greater than or equal to 16 by race/ethnicity, Hispanics and blacks have much greater unemployment rates than whites from 4 to 6 percentage points. Following the National unemployment rates, we present demographic characteristics of the Nation, Southeast, and Florida in Table 2-2 a nd in Figure 2-1. The S outheast has a higher percentage of blacks (22.3 percent) versus th e National level (12.1 per cent) and lower numbers of Hispanics (6.0 percent) than nationally (9.0 percent). Meanwhile, Fl orida has higher numbers of Hispanics (12.2 percent) and marginally highe r levels of blacks (13.6) in comparison to the National figures. These differences in racial/eth nic concentrations are ma de clearer in Figure 227


1, as blacks are predominant in the Southeast and Hispanics are predominant in Florida. From Table 1-1, blacks and Hispanics fare worse in unemployment levels relative to whites. Education and Income Levels of the Nati on relative to the Southeast and Florida Income inequality has been on the rise in th e U.S. since the 1970s. Evidence of income inequality is the percentage of total income earned by the highest income families and the percentage of total income earned by the lowest income families. According to recent Census Bureau data, the 25 percent highest income fam ilies now receive 44.6 percent of U.S. income. The 25 percent lowest income families earn 4.4 percent. This is the widest rich-poor gap since the Bureau revised their data collection methods in 1947 (Robison and Siles, 1999). The relationship between education and income ha s been one of the clearest differences in America (Robison and Siles, 1999). The higher th e level of education, the better the job one obtains; the higher the income, th e lower the rates of poverty and the more money that can be devoted to education. The most important i ndicator for the declin e in poverty was the educational attainment of the head of households. In about 25 percent of families that lived in poverty, the head of household had less than 8 ye ars of schooling. When the head of household attended, but did not graduate fr om high school, the poverty rate was 20 percent. When the head of household graduated from high school, the poverty rate was only 9 percent. When the head of household had at least one year of college, the rate fell to 3.5 percent. Th is pattern was true of whites, blacks, and Hispanic s (Robison and Siles, 1999). Table 2-3 further supports this claim when we specifically look at the role educational attainment plays on poverty. In a ll age categories, the rates for individuals with No High School Diploma were nearly 40 percent a nd more likely to be living in pove rty. What is very salient, noticeable is that the levels of poverty are mu ch higher for Hispanics and blacks in comparison to whites. The importance of this observati on is significant because the National Job Corps 28


Study results do not clarify the regional differenc es in income levels and the race/ethnicity concentrations. Increased employment is one of the central goals of Job Corps, and the remedying factors to attain that are increa sed educational and technical training. One critical factor that deserves scrutiny is high school educational attainment. We present National high school graduation rates by segmenting the target demographic by race/ethnicity and sex seen in Table 2-4 and Figure 22. High school completion not only represents a cornerstone of educational achievement for young people, it is also a minimum requirement for pursuing most types of further education or training and for entering the labor force (U .S. Department of Education, 1997). A high school dropout is a student who was enro lled at the beginning of the year, but was not enrolled at the beginning of the next year or who did not gradua te from high school or completes some other districtor state-approved educational program. The event dropout rate measures the percentage of high school students who drop out in a given year. Another measure of the dropout rate is the status dropout rate, which is measur ed as the percentage of young people aged 18 through 24 who dropped out of grad es 10 through 12 in the past year. Status rates are higher than event rates because they incl ude all dropouts in this age range, regardless of when they last attended school (U.S. Department of Education, 1997). Our results will focus on status rates. We can see that at the National level with all ra ces/ethnicities and sex, the dropout rate is 13.9 percent. Once it is segmented by sex, the rates are marginally higher for males and lower for females. Once we have segmented the target group by race/ethnicity, Hispan ics are disparately larger in high school dropout rates; approximately 20 percenta ge points higher from the all race/ethnicity category rate. Blacks are a lit tle less than 2 percent higher than the all 29


race/ethnicity category rate. Whites have the lo west dropout rate, 4 percentage points less than the all race/ethnicity ca tegory rate. The Hispanic dropout rate may be accounted for by the language barrier they may face and/or the l ack of bilingual education to overcome their educational difficulties. The disp arities of the dropout rates may speak to the above mentioned demographic characteristics. We can also say that these disparities effect the Southeast and Florida adversely, as black and Hisp anic concentrations are greater than that of National levels. In addition, Figure 2-2, provides a graphical representation of the dropout rates by race/ethnicity. The rates of poverty in the Southeast and Florida, in compar ison to the Nation, reflect the overall disadvantage that individuals face in t hose regions. From Table 2-5, both poverty and high school dropout rates are greater for the Southeast regions by more than four percentage points from the National rate. The composition of the Southeast for poverty and high school dropout rates shows the disadvantage that indivi duals face in poverty with the exception of Georgia. Alabama, Louisiana, and Mississippi anecdotally have large populations of blacks, who appear to face higher rates of unemploy ment, poverty, and high sc hool dropout rates. Hispanic populations have been on the rise in Florida, with greater ra tes of high school dropouts (20 percent greater than National all category rate s) from Table 2-4. In order to delve further into the differe nces in market conditions, we will focus on industry composition and the corresponding annual pay rates, which can be viewed in Table 2-6 and Figure 2-3 is provided to graphically show the differing industry compositions throughout the regions. There are certain i ndustries that employ great number s of individuals in all three regions. Therefore, it is not necessary to examine those industries in greater detail. Our focus will be on the industries that differ from the National figures. In the Southeast, mining, construction, and retail trade are slightly larger employers of labor from National results, while, 30


finance and professional services show slight decreases as employers of labor. In addition, Florida construction, retail trade, business and repair services, fi nancial, personal services, and entertainment show almost 1 to 2 percen tage points higher than National figures. Average annual pay levels for the Nation varied widely by industry. The mining industry, which accounts for less than 1 percent of privat e sector employment, had the highest average annual pay at $43,652. The next highest pay level was in finance, insurance, and real estate at $36,062. The lowest average annual pay, $14,386, wa s in the retail trade industry, which employs a large proportion of part-time worker s (U.S. Department of Labor, 1995). In the Southeast, the pattern was the same as pay va ried by industry: mining ($36,807), transportation, communications and public util ities ($32,236), and re tail trade ($13,224). Florida follows a similar wage pattern: mining ($36,087), wholes ale trade ($32,683), and retail trade ($14,697). There are great differences in manufacturing as an employer of labor regionally, with Florida having a seven percentage -point negative differe nce relative to National figures. For the Southeast, this rate is similar to the National rate In Florida, what is salient is the number of retail trade positions that employ the labor force, which reflects the unique market conditions of the state, see Figure 2-3. Workers of all ages are employed in each indus try. However, certain industries tend to possess workers of distinct age groups. An exampl e of such industry would be retail trade that employs a relatively high proportion of younge r workers to fill part-time and temporary positions. In contrast, the manufacturing sect or employs individuals who are older because many jobs in this sector require a number of years of acquired skill s. In recent years, manufacturing employment has been declining, which means th ere are fewer opportunities for younger workers to gain entry into this sector (U.S. Department of Labor, 1995). 31


Current Study The National sample used in this study follo ws closely with the National sample of the NJCS. Nationally, the number of control group participants is 3,976 and the treatment group is comprised of 6,108 participants. The discrepanc y in the population samples is accounted for by the removal of ethnicity labeled as Other in the dataset of this study, which was comprised of Native American and Asian or Pacific Islander. At the 48th month survey, if the participant had missing information in his or her reported earnings in quarter 16, the data was not included in the data set. The percent of the sample that is white is 28.3, blacks account for 52.9 percent, and Hispanics comprise 18.6 percent. Employing confidential NJCS data that contains participants zip-code of residence, we are able to identify those living in the Southeast United States and in the state of Florida. The Southeast sample consists of 811 contro l group participants and 1,149 treatment group participants. The percent of the sample that is white is 14.6, blacks acco unt for 77.1 percent and, Hispanics comprise 8.3 percent. The Florida re gion consists of 233 control group participants and 351 treatment group participants. The percen t of the sample that is white is 15.9, blacks account for 62.3 percent, and Hispanics comprise 21.7 percent. From the race/ethnicity concentrations, we can see that there are clear differences in comparison to the National composition. Blacks account for 77 percent of the sample in the Southeast region and Hispanics are 10 percent less from the National makeup. In Florida, blacks comprise 62.3 percent, but that is still 10 percentage points gr eater than the National composition. Hispanics have a large concentration in Florida with a 21.7 percent composition. From the previous discussions about unemp loyment levels, educational attainment, and poverty/income levels, it is reas onable to say that there are cl ear differences in job market conditions for the different race/ ethnicity categories. These fi gures speak to the disadvantage 32


that blacks and Hispanics face. Our focus at th e regional level follows th e expectation that the Southeast and Florida do not have sim ilar trends with the National results. 33


Table 2-1. National Unemployment Rate: Years 1994-1995 By Ages, Sex, Race/Ethnicity Year Age 16 Age 16 to 19 Age 20 Male Age 20 Female Hispanic Age 16 White Age 16 Black Age 16 1994 6.1 17.6 5.4 5.4 9.9 5.3 11.5 1995 5.6 17.3 4.8 4.9 9.3 4.9 10.4 Source: U.S. Bureau of Labor Statistics, 1994-1995. Table 2-2. Population Demographics for 1990 Nation Southeast Florida Total population 248,709,873 30,249,918 12,937,926 Sex by Percentage Male 48.7% 48.3% 48.1% Female 51.3% 51.7% 51.9% Race/Ethnicity by Percentage Hispanic 9.0% 6.0% 12.2% White 80.3% 75.4% 83.1% Black 12.1% 22.3% 13.6% Source: U.S. Bureau of the Census, 1990. 34


Table 2-3. Enrollment and Educat ional Attainment Poverty Stat us In 1994 of Persons 16 to 24 Years Old All Races/Ethnicity Poverty Level Percent of Total White Below Poverty Level Percent of Total Black Below Poverty Level Percent of Total Hispanic Below Poverty Level Percent of Total Total 18.1 15.2 31.1 31.4 16 to 17 Years Old 18.2 13.6 37.7 36.0 Enrolled In School 16.3 12.1 34.9 32.8 Not Enrolled 45.0 36.9 68.2 55.6 No High School Diploma 47.1 39.1 69.2 58.1 18 to 21 Years Old 18.9 16.4 30.4 32.8 Enrolled In School 14.4 11.9 25.0 28.2 Not Enrolled 24.0 21.4 35.6 35.6 No High School Diploma 42.2 38.1 58.5 45.6 22 to 24 Years Old 17.0 14.8 27.1 27.0 Enrolled In School 13.9 11.9 21.0 23.4 Not Enrolled 18.0 15.7 28.6 27.7 No High School Diploma 38.9 36.1 53.3 39.3 Source: U.S. Bureau of the Census, 1996. Income, Poverty, and Labor Force Information. 35


Table 2-4. The National Population 18 to 24 Years Old by High School Graduate Status, Attainment, Sex, and Race/Ethnicity (N umbers in thousands): Year 1995 Total Percent High School Graduates Percent High School Status Dropouts All Race/Ethnicity Both Se xes 24,900 80.8% 13.9% Male 12,351 79.3% 14.5% Female 12,548 82.4% 13.4% Hispanic 3,603 58.6% 34.7% White 16,867 86.1% 9.8% Black 3,625 76.9% 14.4% Source: U.S. Department of Education, 1997. Table 2-5. High School Status Dropout Rates and Percent in Poverty by Nation, Southeast, Florida In 1995 (Numbe rs in thousands) All Ages Percent in Poverty Ages 18 to 24 Percent In HS Status Dropout Rates National 13.8 13.9 Southeast 18.3 18.3 Florida 16.2 19.3 Composite of Southeast Alabama 20.1 16.0 Florida 16.2 19.3 Georgia 12.1 19.7 Louisiana 19.7 19.5 Mississippi 23.5 16.9 Source: U.S. Department of Education, 1997. 36


Table 2-6. Industry Employment Categories an d Average Annual Earnings 1994 by National, Southeast, Florida National Percentage Employed Southeast Percentage Employed Florida Percentage Employed Total 26,496 100.00 23,008 100.00 23,310 100.00 Mining 43,652 0.60 36,807 1.10 36,087 0.20 Construction 28,308 6.20 23,703 7.00 24,236 7.80 Manufacturing 33,527 17.70 27,920 17.60 30,094 10.50 Transportation, communications, and public utilities 34,201 7.10 32,236 7.50 31,472 7.70 Wholesale trade 34,646 4.40 30,674 4.40 32,683 4.60 Retail trade 14,386 16.80 13,224 17.20 14,697 19.60 Finance, insurance, and real estate 36,062 6.90 28,887 6.00 31,097 8.10 Services 25,113 32.70 22,257 31.30 23,305 33.80 Government 29,202 4.80 24,589 5.10 27,260 5.00 Source: U.S. Bureau of Labor Statistics, Industries 1994-1995. 37


38 0% 10% 20% 30% 40% 50% 60% 70% 80% 90% MaleFemaleHispanicWhiteBlack Nation Southeast Florida Figure 2-1. Population De mographics Year 1990 Figure 2-2. The National Population 18 to 24 Years Old by High School Graduate Status, Attainment, Sex, and Race/Ethnicity Year 1995 0% 20% 40% 60% 80% 100% MaleFemaleHispanicWhiteBlack Percent High School Graduates Percent High School Status Dropouts


0% 5% 10% 15% 20% 25% 30% 35% 40%Min i ng C onstructi o n Manu fa ctu ri n g TCP Whol e sale Re t ail t r a d e FIR S e r v ic e s Government National Southeast Florida 39 Figure 2-3. Industry Employment Categories 1994 by National, Southeast, Florida


CHAPTER 3 MODEL This chapter gives a brief description of the empirical methods used to analyze whether the Job Corps program is effective in the Southeast and in the state of Florida. The main idea behind our methods is to exploit the availability of th e randomization employed in the NJCS. By this randomization, the individuals in the treatment a nd control groups are comparable, on average. This holds true for the National sample, as well as the subsamples of Southeastern states and Florida. Thus, randomization allows us to empl oy simple methods such as differences-in-mean outcomes. However, given the presence of non-compliance in the form of treatment-group members failing to participate in Job Corps (the reason the NJCS adjusted its estimates), the estimates presented below should be interpreted as intention-to-treat effects (Flores-Lagunes et al., 2008). Intention-to-treat (ITT) is a strategy for the analysis of randomized so cial experiments that compares participants in the groups to which they were originally randomly assigned. This is generally thought of as including all participants, whether they actually received the treatment and/or they withdrew from th e treatment (Hollis and Campbell, 1999). The ITT approach has two main purposes. First, it maintains treatment groups that are similar apart from random variation. This is the reason for randomization, a nd the proper results may be lost if analysis is not performed on the groups produced by the ra ndomization process. Second, ITT analysis allows for non-compliance and deviations from policy by the experimenter. Some types of deviations from randomized selection may occur only within the experi ment setting and would not be expected in routine practi ce. However, most types of devi ations from the experiment that could normally occur should be included in the es timated benefit of a treatment, such as the preceding example of non-compliance by treatment gr oup participants. ITT analysis is suitable 40


for pragmatic experiments of effectiveness ra ther than for explanatory investigations of efficacy (Hollis and Campbell, 1999). The difference-in-means model employed is: (3-1) = 0 1 STATA is the functional statistical software used to compute the difference-in-means estimator its standard error, and t-te sts of significance. The term refers to the difference-inmean outcomes between the treatment group an d the control group, with respect to the proportional differences in pa rticipation. In general, = 0 1, ( 0) represents the average outcome for the control group, ( 1) refers to the averag e outcome of the treatment group. Statas ttest performs the two-group mean-comparison test for variables that have continuous values. Statas prtest will be utilized for binary va riables to perform the two-group test of proportion. Each test will provide the difference-in-means estimate ( ) and the significance of each variable. Estimates will be obtained for the Southeast, Florida, and for comparison, Nationally. Given the smaller sample sizes for the Southeast and Florida relative to the National sample, we also present estimates that control for differences in pre-treatment covariates (those covariates measured in the baseline survey). Adjusting for such differences should result in estimates of the effects that have greater efficien cy (Flores-Lagunes et al., 2008). The regression model employed is: (3-2) Yi is the dependent or response variable i.e. the weekly earnings or employment in Quarter 16. 0, are the coefficients of the regression line, Ti is the dummy variable i.e. treatment status (0=control and 1=treatment), Xi is the set of pre-treatment c ovariates or predictor variables and is the residual. 41


In the linear regression model, we regress the dependent variab le earnings in quarter 16 on treatment status and additional 18 independent base line covariates. The pre-treatment covariates included in our baseline specification are: age, gender, whether the indi vidual has children, is married, is head of household, dummy variables for residence in MSA (small city) or PMSA (large city), whether the individu al speaks English as native langua ge, has ever been arrested, has a high school diploma, has a GED, has a vocational degree, employment status, ever worked before, his/her pre-treatment week ly earnings, and indicator variab les for whether the individual is Hispanic, black or white. 42


CHAPTER 4 RESULTS We start with a description of the summary statistics of the data employed. Table 4-1 reports summary statistics of the data for the National, Southeast, and Florida regions, each broken down by treatment status. The table also reports tests of differences-in-means between treatment and control groups. Given the randomization, we would expect that the observable characteristics at baseline are aligned between treatment and control groups. However, it is possible to observe imbalances given attrition, non-response, or due to chance. The extent to which imbalances are present in our samples are documented by using simple di fferences-in-means (or proportions) tests. Surprisingly, the National samples control a nd treatment groups show more statistically significant differences-in-mean characteristics th an the Southeast and Florida samples in Table 4-1. The variables at baseline that have sta tistically different means between treatment and control groups for the National sample are: ag e of individual (p>0.00), th e proportion of females (p>0.00), whether individual has children (p>0.0 2), and whether participant has obtained a high school diploma (p>0.04). The mean age of part icipants in the treatment group (18.90) of the National sample is greater than that of the control group (18.76) Moreover, the proportion of females in the control group (38 percent) is smalle r than that of the trea tment group (46 percent) and treatment group participants are more likely to have children (21 percent) versus the control group (19 percent). Finally, treatm ent group participants are more likely to have completed their high school diploma (19 percent) th an the control (18 percent). The Southeast region difference-in-means tests th at are significant are female participation (p>0.00), individuals that have children (p>0.05), and the percentage of Hispanics (p>0.07). Females have a larger participation proportion, which is also greater for treatment group (45 43


percent) than for control group (38 percen t). Treatment group members have a higher percentage of children (24 per cent) than the control group (20 percent). Hispanics compose a greater proportion in the control group (10 percent) versus the treatment group (7 percent). In Florida, the factors that are significant are: proportion of females (p>0.08), completion of a GED at baseline (p>0.02), and the percen tage of blacks (p>0.07) and Hispanics (p>0.01) that are participating. The proportion of female s is greater in the treatment group (44 percent) than the control (36 percent), a nd control group members are more likely to have completed their GED at 6 percent and for treatment participants (2 percent). Blacks are in greater numbers in the treatment group at 65 percent and 58 percent in the control group; while Hispanics are in greater proportion for the control group at 27 percent and 18 percent for the treatment group. To account for these imbalances across treatment and control groups, we will undertake a linear regression analysis. In order to compare the effect of Job Corps participation on ear nings and employment Nationally, the Southeast, and Florida regions, we start by presenting simple difference-in-means estimates in Table 4-2. This estimate (the di fference in means between treatment and control groups) tells us the return or loss that the individuals acquire on average as a result of undergoing Job Corps training. We concentrate on two outco mes: 1) the average weekly earnings of individuals in quarter 16 after ra ndom assignment and 2) the percen tage of individuals that are working in quarter 16 after the random assignment (employment). In similar order to Table 4-1, Table 4-2 reports the summary statistics of the data for the National, Southeast, and Florida regions, which are each broken down by treatment status. The table presents the average weekly earnings and employment percenta ges in quarter 16. Additionally, it reports the differences in means between treatment and control groups. 44


Two-tailed statistical tests were performed to test the null hypot hesis of no program impact. The null hypothesis is the same for all th ree tests: there is no difference in earnings of individuals in the control and treatment group. The interpretation of th e p-values for National (p>0.00) and Southeast (p>0.00) indicates that the probability of observing an earnings difference between the control group and the treatment group is statis tically significant. Florida, has a p-value of p>0.29. Therefore, we fail to reject the null hypothesis and the difference-inmeans of earnings is not statistically significant. Furthermore, the results show th at participation in Job Corps is effective Nationally and in the Southeast. The weekly earnings of individua ls 48 months after randomization are greater and statistically significant. Nati onally, participants in the tr eatment group, earn $13.88 per week more in earnings than the control group. In the Southeast, treatment group members earned $30.04 more than the control. Conversely, Flor ida does not follow this trend. Although, Job Corps participation is effectiv e in increasing weekly earnings, with the treatment group earning $17.70 more than the control group, this figur e is not statistica lly significant. In the case of Job Corps effectiveness on obtaining employment, referring back to Table 42, the p-values for National (p> 0.04) and Southeast (p>0.10) tells us that the probability of observing an employment difference between th e control group and the treatment group is statistically significant and slight ly significant, respectively. Fo r Florida, with a p-value of p>0.67, we fail to reject the null hypo thesis and the difference in th e means of employment is not statistically significant. In continuing the analysis of Table 4-2, we see the percentage of individuals that were employed in quarter 16 is higher with participat ion in Job Corps. This is reflected both Nationally and in the Southeast. The employment of individuals 48 months after randomization 45


46 is greater and statistically si gnificant Nationally and slightly significant for the Southeast. Nationally, participants in the treatment group are employed two pe rcent more than the control group. In the Southeast, treatm ent group participants were four percent more likely to be employed at quarter 16. In contra st, the treatment group in Florida is one percent more likely to be employed. However, this figure is not statistically significant. In continuing our analysis, we test the r obustness of our summary statistics results by utilizing linear regression testi ng for earnings and employment. In Table 4-3 and 4-4, we present the coefficients of our pre-treatment variables and the impact of J ob Corps participation. Based on our linear regression results, 48 mont hs after randomization, the earnings for National, Southeast, and Florid a regions showed that particip ation in Job Corps does have a effect on the earnings of indi viduals. From Table 4-3, ear nings Nationally ($18.38, p>0.00) and in the Southeast region ($36, p>0.00) are statistically significant. The effect of Job Corps on individuals earnings in quarter 16 is greater in Florida (p>0.10) than it is at the National level ($28.50), but only marginally st atistically significant. The linear regression results for the percentage of st udy participants that were employed in quarter 16 in Table 4-4 also follows a similar outcome to the employme nt difference-of-means results. While adjusting for the pre-treatme nt covariates, it seems Nationally and in the Southeast, Job Corps participation has a positive impact on employment in quarter 16. Job Corps participation increases the likeliness of being employed Nationally by two percent (p>0.06) and in the Southeast regi on by four percent (p>0.06); thes e results are significant. In Florida, Job Corps increases the likeliness of be ing employed by 4 percent. However, it is not statistically significant at (p>23).


Table 4-1. Summary Statistics fo r Control and Treatment Groups Na tional, Southeast, and Florida Characteristics Control Mean SE National Treatment Mean SE p-value Control Mean SE Southeast Treatment Mean SE. p-value Control Mean SE Florida Treatment Mean SE p-value Age 18.76 0.03 18.90 0.03 0.00 18.70 0.07 18.78 0.06 0.36 19.16 0.15 19.02 0.12 0.49 Percent Female 0.38 0.01 0.46 0.01 0.00 0.38 0.02 0.45 0.01 0.00 0.36 0.03 0.44 0.03 0.08 Percent Has Child 0.19 0.01 0.21 0.01 0.02 0.20 0.01 0.24 0.01 0.05 0.21 0.03 0.19 0.02 0.63 Percent who are married 0.02 0.00 0.02 0.00 0.67 0.02 0.00 0.02 0.00 0.82 0.01 0.01 0.02 0.01 0.52 Percent who are Household Heads 0.12 0.01 0.13 0.00 0.57 0.11 0.01 0.13 0.01 0.34 0.15 0.02 0.13 0.02 0.49 Percent living in a MSA 0.47 0.01 0.47 0.01 0.85 0.57 0.02 0.59 0.01 0.45 0.44 0.03 0.44 0.03 0.99 Percent living in a PMSA 0.32 0.01 0.32 0.01 0.58 0.13 0.01 0.14 0.01 0.48 0.46 0.03 0.47 0.03 0.82 Percent who speak English 0.88 0.01 0.88 0.00 0.87 0.93 0.01 0.93 0.01 0.84 0.78 0.03 0.81 0.02 0.41 Percent ever convicted 0.26 0.01 0.25 0.01 0.15 0.25 0.02 0.24 0.01 0.75 0.28 0.03 0.28 0.02 0.85 Highest Grade Completed 10.07 0.02 10.11 0.02 0.18 9.94 0.05 9.93 0.05 0.91 10.20 0.10 10.16 0.09 0.75 Percent with High School Diploma 0.18 0.01 0.19 0.01 0.04 0.14 0.01 0.16 0.01 0.26 0.19 0.03 0.19 0.02 0.78 Percent with GED 0.05 0.00 0.05 0.00 0.12 0.04 0.01 0.03 0.01 0.50 0.06 0.02 0.02 0.01 0.02 Percent with Vocational Degree 0.02 0.00 0.02 0.00 0.27 0.01 0.00 0.01 0.00 0.83 0.01 0.01 0.01 0.01 0.55 Percent unemployed at baseline 0.58 0.01 0.59 0.01 0.84 0.54 0.02 0.53 0.01 0.87 0.57 0.03 0.55 0.03 0.60 Percent ever worked before 0.79 0.01 0.80 0.01 0.45 0.74 0.02 0.75 0.01 0.71 0.82 0.03 0.79 0.02 0.44 Avg. weekly Pre-treatment Earnings $110.15 1.82 $115.45 4. 30 0.35 $102.86 3.97 $128.34 21.44 0.33 $122.34 7.51 $184.25 69.95 0. 48 Percent White 0.29 0.01 0.28 0.01 0.64 0.15 0.01 0.14 0.01 0.67 0.15 0.02 0.17 0.02 0.63 Percent Black 0.52 0.01 0.53 0.01 0.49 0.75 0.02 0.78 0.01 0.12 0.58 0.03 0.65 0.03 0.07 Percent Hispanic 0.19 0.01 0.19 0.00 0.73 0.10 0.01 0.07 0.01 0.07 0.27 0.03 0.18 0.02 0.01 Sample size 3976 6108 811 1149 233 351 47 P-value is the smallest level of significance for which the observ ed sample statistic tells us to reject the null hypothesis.


Table 4-2. Summary Statistics for Control and Treatment Groups National, Southeast, and Florida: Employment and Earnings Region At 48th Month Interview Average weekly earnings in Quarter 16 Percent who worked in Quarter 16 Sample size National Control Mean $201.20 69.00% 3976 S.E. 3.26 1.00% Treatment Mean $215.08 71.00% 6108 S.E. 2.81 1.00% Differ. of Mean $13.88 2.00% Diff. of S.E. 0.45 0.00% p-value 0.00 0.04% Southeast Control Mean $179.92 66.00% 811 S.E. 6.58 2.00% Treatment Mean $209.96 70.00% 1149 S.E. 6.63 1.00% Differ. of Mean $30.04 4.00% Diff. of S.E 0.05 1.00% p-value 0.00 0.10% Florida Control Mean $208.91 73.00% 233 S.E. 12.41 3.00% Treatment Mean $226.62 74.00% 351 S.E. 10.93 2.00% Differ. of Mean $17.70 1.00% Diff. of S.E 1.48 1.00% p-value 0.29 0.67% P-value is the smallest level of significance for which the observed sample statistic tells us to reject the null hypothesis. 48


Table 4-3. Linear Regression Result s for Earnings in Quarter 16: National, Southeast, and Florida Coef. National* S.E. t-stat P>|t| Coef. Southeast** S.E. t-stat P>|t| Coef. Florida*** S.E. t-stat P>|t| Treatment Variable 18.38 4.35 4.22 0.00 36.00 9.47 3.80 0.00 28.50 17.11 1.67 0.10 Baseline Pre-Treatment Variables Age of the Individual 1.45 1.31 1.10 0.27 1.96 2.90 0.67 0.50 -1.21 4.71 -0.26 0.80 Individual is Female -65.69 4.63 -14.20 0.00 -57.20 10.27 -5.57 0.00 -38.88 18.37 -2.12 0.04 Individual has Child(ren) 3.50 6.20 0.56 0.57 -4.85 12.80 -0.38 0.71 23.02 23.25 0.99 0.32 Individual is Married 7.94 15.22 0.52 0.60 -31.71 36.34 -0.87 0.38 -6.95 62.50 -0.11 0.91 Individual is Head of Household 10.58 7.02 1.51 0.13 -3.05 15.35 -0.20 0.84 -36.76 26.58 -1.38 0.17 Individual Resides in MSA 12.28 5.73 2.14 0.03 33.51 10.84 3.09 0.00 19.33 30.83 0.63 0.53 Individual Resides in PMSA 24.91 6.28 3.97 0.00 49.18 16.69 2.95 0.00 26.30 31.78 0.83 0.41 Individual Primary Language is English -10.05 8.68 -1.16 0.25 -13.89 23.24 -0.60 0.55 -3.55 25.88 -0.14 0.89 Individual has Ever Been Arrested -15.05 5.02 -3.00 0.00 -20.78 11.12 -1.87 0.06 -35.98 19.53 -1.84 0.07 Highest Grade Completed 7.89 1.99 3.96 0.00 4.07 4.10 0.99 0.32 1.00 6.99 0.14 0.89 Individual has High School Diploma 20.35 7.46 2.73 0.01 27.28 16.48 1.66 0.10 65.95 26.88 2.45 0.01 Individual has GED 22.41 10.16 2.21 0.03 4.43 25.05 0.18 0.86 -2.76 44.22 -0.06 0.95 Individual has Vocational Degree 25.09 15.18 1.65 0.10 -0.79 42.37 -0.02 0.99 -42.30 74.54 -0.57 0.57 Individual is Unemployed -41.31 5.49 -7.52 0.00 -27.47 12.33 -2.23 0.03 -0.97 21.11 -0.05 0.96 Individual has Ever Worked Before 84.43 7.00 12.06 0.00 79.22 14.63 5.42 0.00 36.15 26.42 1.37 0.17 Individual Weekly Earnings 0.04 0.01 4.74 0.00 0.01 0.01 1.45 0.15 0.00 0.01 0.39 0.70 Individual is Hispanic -36.65 7.99 -4.59 0.00 -62.74 24.01 -2.61 0.01 1.85 33.14 0.06 0.96 Individual is Black -60.88 5.28 -11.54 0.00 -85.51 13.92 -6.14 0.00 -53.04 25.16 -2.11 0.04 Constant 103.68 27.04 3.83 0.00 139.24 59.23 2.35 0.02 218.95 100.50 2.18 0.03 49 P-value is the smallest level of significance for which the observ ed sample statistic tells us to reject the null hypothesis. Number of observations = 9405 ** Number of observations = 1857 *** Number of observations = 543


Table 4-4. Linear Regression Results for Percent Employed in Quarter 16: National, Southeast, and Florida Coef. National* Rob. S.E. t-stat P>|t| Coef. Southeast** Rob. S.E. t-stat P>|t| Coef. Florida*** Rob. S.E. t-stat P>|t| Treatment Variable 0.02 0.01 1.85 0.06 0.04 0.02 1.92 0.06 0.04 0.04 1.20 0.23 Baseline Pre-Treatment Variables Age of the Individual 0.00 0.00 -0.50 0.62 0.00 0.01 0.57 0.57 -0.02 0.01 -1.44 0.15 Individual is Female -0.04 0.01 -4.26 0.00 -0.01 0.02 -0.51 0.61 0.01 0.04 0.29 0.77 Individual has Child(ren) 0.02 0.01 1.39 0.16 0.01 0.03 0.19 0.85 0.10 0.05 2.12 0.04 Individual is Married -0.02 0.03 -0.74 0.46 -0.10 0.09 -1.09 0.28 0.03 0.13 0.22 0.83 Individual is Head of Household 0.00 0.02 -0.26 0.80 -0.04 0.03 -1.09 0.28 -0.02 0.06 -0.42 0.68 Individual Resides in MSA 0.02 0.01 1.44 0.15 0.09 0.03 3.73 0.00 0.08 0.07 1.07 0.29 Individual Resides in PMSA 0.01 0.01 0.97 0.33 0.10 0.04 2.66 0.01 0.04 0.08 0.47 0.64 Individual Primary Language is English -0.03 0.02 -1.57 0.12 0.01 0.05 0.20 0.84 0.04 0.06 0.64 0.52 Individual has Ever Been Arrested -0.04 0.01 -3.97 0.00 -0.05 0.03 -1.88 0.06 -0.06 0.04 -1.34 0.18 Highest Grade Completed 0.01 0.00 2.97 0.00 -0.01 0.01 -1.44 0.15 0.00 0.02 -0.18 0.86 Individual has High School Diploma 0.05 0.02 3.13 0.00 0.06 0.04 1.66 0.10 0.08 0.06 1.40 0.16 Individual has GED 0.06 0.02 2.71 0.01 0.03 0.05 0.57 0.57 0.01 0.10 0.07 0.95 Individual has Vocational Degree 0.03 0.03 1.06 0.29 0.06 0.09 0.73 0.47 -0.06 0.17 -0.33 0.74 Individual is Unemployed -0.08 0.01 -7.52 0.00 -0.12 0.02 -5.11 0.00 -0.05 0.04 -1.08 0.28 Individual has Ever Worked Before 0.18 0.02 11.70 0.00 0.23 0.03 7.35 0.00 0.14 0.06 2.36 0.02 Individual Weekly Earnings 0.00 0.00 1.51 0.13 0.00 0.00 1.72 0.09 0.00 0.00 0.86 0.39 Individual is Hispanic -0.07 0.02 -4.36 0.00 -0.02 0.05 -0.40 0.69 0.07 0.07 0.95 0.34 Individual is Black -0.11 0.01 -10.09 0.00 -0.09 0.03 -2.90 0.00 -0.10 0.05 -1.97 0.05 Constant 0.59 0.06 10.17 0.00 0.61 0.13 4.54 0.00 0.91 0.23 3.92 0.00 50 P-value is the smallest level of significance for which the observ ed sample statistic tells us to reject the null hypothesis. Number of observations = 9405 ** Number of observations = 1857 *** Number of observations = 543


CHAPTER 5 CONCLUSION AND IMPLICATIONS The Job Corps program was created to take underprivileged youths aged 16 to 21 and remove them from their non conducive environments to reside ntial centers and campuses to provide them with educational and vocational tr aining to improve and develop their skill set for employability. The major goals of Job Corps are to increase the employment and earnings of the participants. At the 48th month interview, Job Corps participants Nationally were 2 percent more likely to be employed and earned $13.88 more per we ek (control-$201.20 versus treatment-$215.08). These figures were found to be statistically signi ficant in the difference-of-means estimations. In the Southeast, treatment gr oup participants were 4 percent more likely to be employed (slightly significant) and earned $30.04 more per week (control-$179.92 versus treatment$209.96), which is significant. In Florida, Job Corp s participants were 1 percent more likely to be employed and earned $17.70 more per week (control-$208.91 versus treatment-$226.62). Nevertheless, these figures were not found to be statisti cally significant. The subsequent linear regre ssion tests were conducted to examine the robustness of the results on employment and earnings at quarter 16 at the National, Southeast, and state of Florida levels, which are qualitatively similar to the diffe rence-of-means results. Nationally and in the Southeast, employment at the 48th month interview improved among Job Corps participants by two percent and four percent, re spectively; both are si gnificant. While in Florida, employment increased by four percent, which is not significan t. Earnings, Nationally and in the Southeast are improved by Job Corps participation. The current study has found that there are regi onal differences, which affect the outcomes of employment and earnings. The Nation and the Southeast follow a similar trajectory while 51


Florida deviates. However, we must mention th e small sample size of Florida relative to the Nation and the Southeast, and that Florida is a mo re heterogeneous state in terms of population. These observations are important to note because th ey may explain the differences in results. For instance, the fact that the estimated impacts of Job Corps for Florida are largely similar to those of the Nation and the Southeast, aside from their statistical signifi cance, call for future research that explores this issue in more detail. Also, we have discusse d the varying levels of unemployment, high school attainment, poverty le vels, racial concentr ations, and industry composition. These factors are tied directly to the differences in the years of schooling, poverty status, and socioeconomic conditions that are ex perienced by whites, blacks, and Hispanics. From the research, blacks and Hispanics face greate r wage and employment difficulties. Further, since Florida has higher percenta ges of blacks and Hispanics, th ese obstacles are magnified for these two groups in comparison to whites. This ultimately leads us to the conclusion that Job Corps is effective Nationally and in the Southeast, but we do not find this to be the case in the state of Florida. The above conclusions regarding Florida rais e important policy issues. What is the justification for Job Corps training if it is not serving the community effectively? One interpretation of the results in this thesis is that they cast doubt on Job Corps effectiveness in Florida. Following this interpretation, the fede rally-designed program may have to be re-tooled to better serve the states unique so cio-economic conditions. An interesting aspect that may contribute to th e lack of efficacy of Job Corps in Florida is the potential inability of state Job Corps Center s to serve their target demographic groups. In a 1996 report to Congress that covered Job Corps Ce nters 1993-1994 capacity to serve, the U.S. Government Accounting Office found that Job Co rps had the capacity to serve 81 percent of 52


program participants in their home states. A pproximately, 59 percent of participants were assigned to centers in their home state, while the remaining participants were sent to centers outside their home state and traveled an average of ove r four times as far. However, about 83 percent of those participants who obtained j obs were employed in their home state (U.S. Government Accounting Office, 1996). In relation to the difficulties in placing clients in their home state, there are also insufficient numbers of seats available to serve the community that may contribute to the discrepancy of the lack of effectiveness of Job Corps. Nationall y, there are 11,861 insufficien t number of seats with approximately 32 percent of those seats are account ed to the Southeast, and Florida accounts for nearly 16 percent of the insufficiency (U.S. G overnment Accounting Offi ce, 1996). It could be reasoned that in Florida the clie nts are underserved by Job Corps and need increased capacity to serve underprivileged groups more effectively. Th is may also address th e non-participation of Job Corps selectees, who chose to forego the pr ogram because of the lack of capacity or availability of program instructi on. More research is needed along these lines to explore the role of insufficient Job Corps resour ces in the state of Florida. This study has provided preliminary findings for fu rther research to reanalyze the role that Job Corps plays in the state of Fl orida. Greater efforts can be i nvested in this topic in hopes to find out the root causes of why certain groups are not making the same gains in earnings and employment in comparison to other groups. Perhaps there has to be a refinement in the individualized plan relating to training, c ounseling, and the type of skills provided. 53


LIST OF REFERENCES Burghardt, J., McConnell, S., Meckstroth, A., Schochet, P., Johnson, T., Homrighausen, J. (1999). "National Job Corps Study: Repor t on Study Implementation. Mathematica Policy Research, Inc.: 8140-8510. Flores-Lagunes, A., Gonzalez, A., Neumann, T. (2008). "Learning But Not Earning? The Impact of Job Corps Training on Hispanic Youth." forthcoming, Economic Inquiry Greenberg, D.H., Michalopoulos, C., Robins, P.K. (2004). What Happens to the Effects of Government-Funded Training Programs over Time? The Journal of Human Resources (39)1: 277-293. Hollis, S. and Campbell, F. (1999). What is meant by intention to treat analysis? Survey of published randomised controlled trials. British Medical Journal. 319: 670-674. Johnson, T. R. and Troppe, M. (1992). Imp roving Literacy and Employability among Disadvantaged Youth: The Job Corp s Model Youth Society. 23(335). Levitan, S. A. and Mangum, G. L. (1969). Federal Training and Work Programs in the Sixties, Ann Arbor: Institute of Labor and Industr ial Relations, University of Michigan. Levitan, S. A. and Johnson, B. H. (1975). The Job Corps: A Social Experiment That Works. Baltimore and London, The Johns Hopkins University Press. Long, D. A., Mallar, C. D., Thornton, C.V.D. (1981) "Evaluating the Benefits and Costs of the Job Corps. Journal of Policy Analysis and Management 1(1): 55-76. Mallar, C. D., Kerachsky, S., Thornton, C., Long, D. A. (1982). Evaluation of the Economic Impact of the Job Corps Program. Thir d Follow-up Report. Mathematica Policy Research, Inc. McCarron, K. M. (2000). "Job Corps, AmeriC orps, and Peace Corps: An Overview. Occupational Outlook Quarterly 44(3): 18-25. Rawlins, V. L. (1971). "Job Corps: The Urban Center as a Tr aining Facility." The Journal of Human Resources 6(2): 221-235. Robison, L. J. and Siles, M. E. (1999). "Social capital and household inco me distributions in the United States: 1980, 1990." Journal of Socio-Economics 28: 43-93. Schochet, P. Z. (1998). "Nationa l Job Corps Study: Characteristics of Youths Served by Job Corps. Mathematica Policy Research, Inc. Schochet, P. Z., Burghardt, J., Glazerman, S. (2001). "National Job Corps Study: The Impacts of Job Corps on Participants Employment and Related Outcomes." Mathematica Policy Research, Inc. 54


55 U.S. Government Accounting Office. (1995). Job Corps: High Costs and Mixed Results Raise Questions About Programs Effectiveness. Washington, DC: Government Accounting Office. U.S. Government Accounting Office. (1996). Job Corps: Where Participants Are Recruited. Trained, and Placed in Jobs. Washingt on, DC: Government Accounting Office. U.S. Bureau of the Census. (1990) Statistical Abstract of the United States. Washington, DC: Bureau of the Census. U.S. Bureau of the Census. (1996) Income, Pove rty, and Labor Force Information. Washington, DC: Bureau of the Census. U.S. Bureau of the Census. (2000) Statistical Abstract of the United States. Washington, DC: Bureau of the Census. U.S. Department of Education. (1997). Dropout Rates in the United States: 1995. Washington DC: National Center for Education Statistics. U.S. Department of Labor. (1999). "Job Corp s Fact Sheet. Retrieved June, 2008, from U.S. Department of Labor. (1995). 1994-95 Career Guide to Industries. Washington, DC: Bureau of Labor Statistics. Williams, L. and Cooper, J. (2004). The Economic Impact of Job Corps Centers on Their Local Communities. Decision Information Resources, Inc.


BIOGRAPHICAL SKETCH Abu Isa Mansoor is part of the Food and Re source Economics Department (FRED) at the University of Florida. Abu immigrated to the Un ited States at the age of five with his father, mother, and two sisters. They arrived in New York and traveled to Connecticut. They did not find the weather agreeable and headed south to Florida. Abu attended Suncoast Community High School and graduated in 2002. He then atte nded Florida Atlantic University and later transferred to the University of Florida in 2004, where he was accepted into FRED. He improved greatly in his academic studies and gra duated in 2006 with a Bachelors degree in FRE. In 2006, Abu received a Fellowship from the U.S. Department of Agriculture. He thought he was very fortunate and grateful to the people that believed in him. He received his M.S. from the University of Florida in the Spring of 2009.