Home Articles Reader Opinion Editorial Book Reviews Discussion Writers Guide About TCRecord
transparent 13
Topics
Discussion
Announcements
 

Evaluating the Impact of the Post-9/11 GI Bill on College Enrollment for Veterans with Service-connected Disabilities


by Liang Zhang - 2020

Background/Context: The Post-9/11 GI Bill has provided educational benefits to millions of military service members and veterans since its adoption in August 2009. Recent studies indicate that the bill has significantly improved college enrollment and educational attainment among post-9/11 veterans. A significant proportion of veterans suffer from service-connected disabilities. While provisions of the Post-9/11 GI Bill may render education benefits that are appealing to veterans with service-connected disabilities, little is known with regard to how the bill has affected college participation among this venerable subpopulation of veterans.

Purpose/Objective: This study examines the effect of the Post-9/11 GI Bill on college enrollment rates among veterans with service-connected disabilities and unpacks potentially heterogeneous impacts across groups with different demographic characteristics (i.e., sex, age, race/ethnicity, educational attainment, and disability ratings).

Population: Post-9/11 veterans.

Research Design: Triple differences.

Data Collection and Analysis: Secondary data analysis based on American Community Survey 2005–2016.

Findings/Results: While the Post-9/11 GI Bill has increased college enrollment for veterans without service-connected disabilities by less than 1 percentage point, the increase is much larger—about 5 percentage points—for veterans with service-connected disabilities. Enrollment effects for veterans with service-connected disabilities are consistent and positive across sex, age, race/ethnicity, educational attainment, and disability ratings.

Conclusions/Recommendations: The results of this study provide strong evidence for the significant enrollment growth among veterans with service-connected disabilities after the adoption of the Post-9/11 GI Bill. While this result is reassuring, it is not clear whether this large effect for veterans with service-connected disabilities is due to favorable provisions in the Post-9/11 GI Bill or due to lower opportunity costs. In the future, researchers may want to identify appropriate sources for data on detailed educational benefits to examine the mechanisms behind the effect.



INTRODUCTION


The original GI Bill of 1944, together with various expansions and modifications over years, have played a pivotal role in shaping the United States since WWII. By providing educational benefits to millions of military service members and veterans for their college education, they improved the human capital stock and contributed to economic growth of the country (Bound & Turner, 2002). Built on the long legacy of America’s commitment to its veterans, the Post-9/11 Veterans Educational Assistance Act of 2008 (also known as the Post-9/11 GI Bill) represented another significant expansion and improvement of educational benefits provided to millions of military service members and veterans. In fiscal year 2017 alone, 755,476 veterans received educational benefits under the Post-9/11 GI Bill totaling $11 billion (U.S. Department of Veterans Affairs, 2018). Compared with its immediate predecessor, the Montgomery GI Bill (MGIB), the Post-9/11 GI Bill removes the buy-in requirement,1 offers more generous tuition benefits, provides a monthly housing allowance, and covers miscellaneous educational costs. Furthermore, the Forever GI Bill passed in 2017 eliminates the 15-year benefit expiration date for those who separated in 2013 or later. Recent studies of the Post-9/11 GI Bill indicate that the bill has significantly improved college enrollment and educational attainment among post-9/11 veterans (Barr, 2015, 2019; Zhang, 2018).


A significant proportion of veterans suffer from service-connected disabilities, which are disabilities due to injury or illness incurred or aggravated by military service. In August 2016, about 22% of all veterans in the United States had service-connected disabilities; that proportion was 36% amongst post-9/11 veterans (Bureau of Labor Statistics, 2017). Despite the initial evidence on the Post-9/11 GI Bill’s positive impact on college enrollment and educational attainment among the overall veteran population, little is known with regard to how this federal program affects a pivotal yet most venerable subpopulation of veterans—those with service-connected disabilities. From a policy perspective, provisions of the Post-9/11 GI Bill may render education benefits that are particularly appealing to veterans with service-connected disabilities, including for example the lower minimum time requirement for full benefits and the possibility of combining the Post-9/11 GI Bill and the Vocational Rehabilitation and Employment Program (VR&E). In addition, veterans with service-connected disabilities may face a lower opportunity cost to attend college than veterans without such disabilities.


Although it stands to reason that the Post-9/11 GI Bill may encourage college enrollment among veterans with service-connected disabilities, there exists no prior empirical research on whether and how the Post-9/11 GI Bill has improved college participation for veterans with such disabilities.2 This study sets out to address this critical void in the literature by examining the overall impact of the financial incentives provided by the Post-9/11 GI Bill on college participation among veterans with service-connected disabilities, and unpacking potentially heterogeneous impacts across groups with different demographic characteristics (i.e., sex, age, race/ethnicity, educational attainment, and disability ratings). This study contributes to two lines of inquiry in postsecondary education in notable ways. First, despite copious research on the effects of financial incentives on college access for members of the general population, very few studies have focused specifically on how such incentives affect college access for veterans. The current study extends the financial aid literature by investigating the impact of the Post-9/11 GI Bill on veterans’ college participation by disability status, and examining potentially heterogeneous impacts among veterans with service-connected disabilities. Second, this study adds to the body of research on students with disabilities, a particularly vulnerable population in higher education. Among students with disabilities, veterans with service-connected disabilities comprise a unique and under-served subgroup, many of whom are negotiating even more complex challenges than their non-veteran counterparts. New research is critical to better understand and serve this growing, yet under-studied student population. Situated at the intersection of two pivotal lines of literature—financial aid and students with disabilities—this study fills an important void in post-secondary education research.


BACKGROUND: THE POST-9/11 GI BILL AND SERVICE-CONNECTED DISABILITIES


The Post-9/11 GI Bill was implemented in August 2009, providing education benefits for military members who have served on active duty since September 11, 2001. The main provision of the Post-9/11 GI Bill includes: (a) full tuition and fees at in-state public schools, (b) a monthly housing allowance, and (c) up to $1,000 a year for books and supplies.3 Tiers of benefits are determined by the length of active duty service since 9/11. Full benefits are awarded to veterans who have served 36 cumulative months or received an honorable discharge for a service-connected disability after at least 30 days of active duty; 40% of full benefits are awarded to those who have served at least 90 days, and 50% to those who have served for at least 6 cumulative months, with each additional 6 months of service increasing the benefit tier by 10 percentage points. With these provisions and programs, the Post-9/11 GI Bill effectively provides free college tuition for most post-9/11 veterans. In addition, the bill offers a living stipend based on the institution’s location;4 the current rate (FY 2018, Grade E5 with dependents) ranges from just under $1,000/month for the least expensive places to $4,329/month for San Francisco. Students enrolled in online programs receive a flat allowance rate, with the current rate of $840.50/month during the 2017–2018 academic year. These benefits are much improved over the MGIB, which paid a flat amount of about $1,400/month in 2008 and $1,857/month in 2017. The difference in average benefits between the Post-9/11 GI Bill and MGIB could be much smaller than the difference in full benefits. For example, in fiscal year 2010, the average educational benefits received by veteran students were $13,954 under the Post-9/11 GI Bill and $7,592 under the MGIB, a difference of $6,362 (U.S. Department of Veterans Affairs, 2011).


Several provisions of the Post-9/11 GI Bill may render education benefits that are particularly appealing to veterans with service-connected disabilities. First, the minimum time requirement for full benefits is much lower for veterans with service-connected disabilities than those without such disabilities (U.S. Department of Veterans Affairs, 2018). A service member receives full educational benefits after serving at least 36 cumulative months of active duty; however, veterans with service-connected disabilities who were discharged honorably due to said disabilities may meet the minimum time requirement after 30 days of active duty service. According to the Veterans Supplement of Current Population Survey between 2009 and 2016, which collects data on the total length of time veterans served on active duty in the Armed Forces, approximately 37% of post-9/11 veterans without service-connected disabilities served for less than 3 years, hence not qualified for full benefits under the Post-9/11 GI Bill.5  


Second, the combination of the Post-9/11 GI Bill and the Vocational Rehabilitation and Employment Program (VR&E) creates substantial values for veterans with service-connected disabilities. Although VR&E provides benefits to disabled veterans after an evaluation process to determine their eligibility for full tuition, fees, books and supplies, the Post-9/11 GI Bill offers a more transparent eligibility rule and generous benefits. In addition, veterans participating in the VR&E program who qualify for Post-9/11 GI Bill benefits can elect to receive full GI Bill benefits with Basic Allowance for Housing (BAH) in most cases higher than the regular subsistence allowance, even if their Post-9/11 GI Bill eligibility is less than 100%. Additional benefits are still available through the VR&E program such as payment for all required books, fees, and supplies as well as other supportive services. In some cases, veterans with service-connected disabilities might qualify for both the VR&E program and the Post-9/11 GI Bill. For example, they may use VR&E benefits to complete a bachelor’s degree when required by their identified employment opportunities and then use the Post-9/11 GI Bill benefits to obtain an advanced degree if so desired. Indeed, when the Post-9/11 GI Bill became effective in 2009, the number of beneficiaries of the MGIB plunged, but the number of beneficiaries of the VR&E program kept rising steadily (U.S. Department of Veterans Affairs, 2018).   


In addition to these favorable provisions, veterans with service-connected disabilities may face a lower opportunity cost to attend college than veterans without such disabilities. For example, in August 2016, the employment-population ratio of veterans with service-connected disabilities was 44.2%, lower than the 48.3% ratio for veterans without such disabilities (Bureau of Labor Statistics, 2017).6 The lower labor force participation rate of veterans with service-connected disabilities means lower foregone earnings, thus making the GI Bill benefits a more attractive revenue source. In short, the increased financial incentives and lower opportunity costs may work in tandem to make college enrollment an especially attractive option for veterans with service-connected disabilities.


Because service-connected disability is a key variable in this study, a discussion of its definition is in order. The Americans with Disabilities Act (ADA) defines a person with a disability as someone who “has a physical or mental impairment that substantially limits one or more major life activities, has a record of such an impairment, or is regarded as having such an impairment.”7 Most federal data collection instruments (i.e., the U.S. Census, ACS, Current Population Survey) focus on a person’s functional abilities including sensory, physical, mental, ambulatory, and self-care.


Service-connected disability differs from general disability in some important ways. According to U.S. Department of Veterans Affairs (2018), a service-connected disability is defined as “current chronic disabilities diagnosed by a medical professional and determined by the Veterans Affairs to have been caused or aggravated by military service or secondary to an existing service-connected disability.” For example, one of the most common service-connected disabilities is tinnitus, the perception of noise or ringing in the ears. An individual with tinnitus, however, may or may not have serious difficulty hearing. Similarly, not all veterans’ disabilities are service-connected, because some disabilities may not have been caused or aggravated by military service. Veterans with service-connected disabilities are evaluated and assigned disability ratings, with the degree of disability ranging from 0% to 100% in 10% increments.8 The ACS data contain information on both general disabilities and service-connected disabilities, as well as disability ratings for veterans with service-connected disabilities.


LITERATURE REVIEW


Literature pertaining to both financial aid and students with disabilities informs the current study. College costs and financial aid figure prominently into college enrollment decisions, which in most studies are framed as being guided by the human capital framework. That is, financial incentives increase college participation by reducing the cost side of the cost-benefit analysis, making college participation a favorable decision. Recent studies on the effect of financial aid programs on college enrollment have confirmed that financial subsidies improve college participation, although effects may vary across types of financial aid (Angrist, Autor, Hudson, & Pallais, 2016; Castleman & Long, 2016; Dynarski, 2000; Goldrick-Rab, Kelchen, Harris, & Benson, 2016; Long, 2004; Sjoquist & Winters, 2012). Deming and Dynarski (2010) concluded that on average, an increase of $1,000 in financial aid improves the likelihood of college enrollment by 4 to 6 percentage points.


The positive enrollment effects of financial incentives were also found among veterans. Bound and Turner (2002) found that the original GI Bill adopted in 1944 increased college education by 0.23 to 0.28 of a year among World War II veterans, equivalent to an increase of 5–6 percentage points in college completion rates. Angrist (1993) estimated that veterans’ educational benefits increased veterans’ college education by approximately 1.4 years among a group of veterans in the 1987 Survey of Veterans. Using MGIB as the research setting, Simon, Negrusa, and Warner (2010) found an increase of 5 percentage points in the benefits usage rate for each $10,000 increase in educational benefits. In a recent study on the effect of the Post-9/11 GI Bill, Barr (2015) found that the expansion in veterans’ education benefits increased the college enrollment rate of veterans by approximately 5 percentage points soon after its adoption. However, over a longer period of time, Zhang (2018) found larger enrollment effects immediately after the bill’s adoption than in later years.


While college costs and financial incentives might be the primary consideration when students make college decisions, research in higher education has long recognized the benefit of adopting additional theoretical perspectives in explaining student college decisions (Hossler, Braxton, & Coopersmith, 1989; Perna, 2006, 2010). Perna (2006) presented an integrated conceptual model that drew on the economic model of human capital investment, the sociological concepts of habitus, cultural and social capital, and organizational context. This model suggests that students’ perceptions and responses to financial incentives could be moderated by their demographic characteristics. Following this line of literature, I consider both the overall enrollment effect of the Post-9/11 GI Bill and potentially heterogeneous impacts across groups with different demographic characteristics including sex, age, race/ethnicity, educational attainment, and disability ratings.


Among the fast-growing population of veteran students in higher education, veterans with service-connected disabilities represent an important subpopulation. Recent statistics indicate that over a third of post-9/11 veterans have service-connected disabilities (Bureau of Labor Statistics, 2017). Having disabilities may subject these veterans to financial burdens, positioning them as especially sensitive to costs and responsive to financial incentives for college attendance. Although financial stress is a challenge that most college students face and is not unique to students with disabilities, it may have a compounding detrimental effect for students with disabilities (Murray, Lombardi, Bender, & Gerdes, 2013; Newman et al., 2011). While there has been a fairly amount of research on the experiences of college students with disabilities, including college adaptation (e.g., Adams & Proctor, 2010), disability accommodation (e.g., Barnard-Brak, Davis, Tate, & Sulak, 2009), college persistence and success (e.g., Mamiseishvili & Koch, 2012), and instructional strategies (e.g., Orr & Hammig, 2009; Polloway, Patton, & Serna, 2008), very few scholars have examined the role of financial incentives on college outcomes for students with disabilities. Further, a review of extant empirical work in financial aid did not yield a single study focusing on veterans with service-connected disabilities, marking a notable gap in the literature.


According to human capital theory, enhanced educational benefits provided by the Post-9/11 GI Bill should lead to an increase in college enrollment among veterans, especially those with service-connected disabilities; however, it does not yield much insight into potentially heterogeneous effects within veterans with service-connected disabilities. The emerging body of inquiry on prospective or current veteran students with disabilities indicates that they are not a homogenous group; many live not only with service-connected physical disabilities, but also with mental health issues stemming from their experiences in combat and their status as non-traditional students (Ford & Vignare, 2014; Madaus & Miller, 2009; Steele, Salcedo, & Coley, 2010; Vacchi & Berger, 2014). Thus, depending on the nature and magnitude of these veterans’ disabilities and other socio-demographic characteristics, financial incentives may have heterogeneous impacts. Informed by and extending the lens of human capital, I aim to address this significant gap in the literature by generating new empirical insights into how the Post-9/11 GI Bill has influenced college participation decisions among veterans with disabilities.


DATA AND METHODS


DATA AND SAMPLES


The ACS uses a series of cross-sectional samples to produce annual estimates of housing units and individuals in the United States. Its public use microdata samples (PUMS) represented about 0.4% of the U.S. population between 2001 and 2004; since 2005, they have represented about 1%. Because military members represent a small proportion of the population, larger samples are preferred, especially when examining effects in subgroups. For example, among the 3.15 million individuals included in the ACS 2016 file, about 1.6% had served in military since 9/11. This proportion was lower in previous years, e.g., 0.7% in 2005. After considering the sample size of veterans in each ACS file and especially the availability of disability information from 2005 onwards, I selected all ACS files between 2005 and 2016.


For each ACS file, I extracted the following information: state of residence, place of birth, age, sex, race/ethnicity, educational attainment, school enrollment, disability status, and military status; for veterans, I also extracted whether they had served during the Gulf War era (i.e., August 1990 to August 2001) and/or the post-9/11 era, and whether they had service-connected disabilities, and if so, their disability ratings. In this study, I limited the sample to those individuals who were born in the United States and who were between 20 and 60 years old.9 Since college enrollment including all levels of post-secondary education was the main dependent variable, I further limited the analytic sample to individuals who had graduated from high school including GED and alternative credentials, regardless of whether or not they held college degrees. These restrictions resulted in the final analytic sample containing about 1 million individuals with 10,000 to 25,000 post-9/11 veterans every year.


VARIABLES


Table 1 presents descriptive statistics of main variables in this study. The dependent variable, college enrollment, is derived from both “school enrollment” and “grade level attending” data in the ACS. College enrollment is a binary variable (1 = yes; 0 = no) defined as whether an individual attends undergraduate, graduate, or professional schools. One potential limitation of the “school enrollment” variable in ACS is that the Post-9/11 GI Bill benefits may be applied to a wide range of educational and training programs, including but not limited to “school enrollment.” To the extent that some programs such as on-the-job training and apprenticeship programs may occur outside of schools, the college enrollment variable may underestimate the full effect of the Post-9/11 GI Bill. In addition, to the extent that veterans with service-connected disabilities may be more or less likely than those without such disabilities to participate in the educational and training activities not captured by the college enrollment variable, the difference in their college enrollment rates could over- or underestimate the overall effect of the Post-9/11 GI Bill. Table 1 indicates that on average, veterans have a higher college enrollment rate than non-veterans. Specifically, the college enrollment rate for non-veterans is 12%, whereas it is 20% for post-9/11 veterans without service-connected disabilities and almost 25% for veterans with service-connected disabilities. The large differences between veterans and non-veterans is partly due to the wide range of age groups included in this study.


Table 1. Descriptive Statistics for Post-9/11 Veterans and Non-Veterans, ACS 2008–2016


 

Non-veterans

 

Vet. w/o Disability

 

Vet. w/ Disability

 

Mean

Std. Dev.

 

Mean

Std. Dev.

 

Mean

Std. Dev.

College enrollment

0.120

0.325

 

0.200

0.400

 

0.246

0.431

Age

39.472

12.063

 

33.704

9.192

 

38.601

9.977

Male

0.458

0.498

 

0.828

0.377

 

0.834

0.372

White

0.748

0.434

 

0.701

0.458

 

0.701

0.458

Black

0.127

0.332

 

0.154

0.361

 

0.169

0.374

Hispanic

0.086

0.280

 

0.098

0.297

 

0.088

0.284

Asian & Pacific Islander

0.015

0.123

 

0.013

0.113

 

0.009

0.096

Native

0.007

0.085

 

0.008

0.087

 

0.007

0.084

Other race/ethnicity

0.017

0.131

 

0.027

0.161

 

0.026

0.160

Less than 1 year college

0.083

0.276

 

0.119

0.323

 

0.102

0.302

1 year or above college

0.207

0.405

 

0.274

0.446

 

0.276

0.447

Associate’s

0.097

0.296

 

0.117

0.322

 

0.155

0.362

Bachelor’s

0.217

0.412

 

0.165

0.372

 

0.195

0.396

Master’s

0.075

0.263

 

0.063

0.242

 

0.118

0.322

First professional

0.019

0.135

 

0.017

0.128

 

0.013

0.111

Doctoral

0.009

0.097

 

0.006

0.080

 

0.007

0.080

         

N

10,378,685

 

123,204

 

59,616

 


Figure 1 shows quite different age-enrollment profiles between veterans and non-veterans. As expected, college enrollment rates for non-veterans peak in their early 20s, fall sharply immediately thereafter, and continue to decline but not as rapidly after the initial drop. While college enrollment rates for veterans are lower than for non-veterans in their early 20s due to military service, the decline in college enrollment rates is moderate over time, resulting in higher enrollment rates for veterans than non-veterans beyond their early 20s. It is also interesting to observe that college enrollment rates are higher among veterans with service-connected disabilities than among those without such disabilities across age groups.


Figure 1. College enrollment rates by age, veteran status, and disability status

[39_23012.htm_g/00002.jpg]


For the remaining variables included in Table 1, post-9/11 veterans on average are younger than non-veterans in the sample because of the war era in which these veterans served. Veterans with service-connected disabilities on average are older than those without such disabilities, suggesting that those who have served longer in the military are more likely to have service-connected disabilities. When those who served in both the Gulf War and post-9/11 periods (i.e., those who served for a long period of time) are removed from the sample, the age distribution of veterans with service-connected disabilities is almost identical to that of veterans without such disabilities. For other demographic characteristics, men are over-represented (83%) in the veteran sample versus the non-veteran sample (46%). Minorities, especially Blacks, have a slightly higher representation in the veteran sample. Veterans and non-veterans also have different levels of educational attainment: veterans are more likely to have attended some college, but are less likely to have obtained bachelor’s degrees or above. However, this could be an artifact of distinct age distributions, because veterans as a group are younger than non-veterans in the analytic sample.


The ACS has two sets of disability measures. The first set, referred to in this study as ACS-defined disability, attempts to capture whether an individual has difficulties in at least one of six areas: hearing, vision, cognitive ability, ambulatory ability, self-care, and independent living. Data on individual disabilities have been collected since the ACS was fully implemented in 2005. This set of disability measures is available for both veterans and non-veterans. In the pooled ACS samples for 2005 through 2016, a moderately larger proportion of post-9/11 veterans between 20 and 60 years of age (11.60%) reported having ACS-defined disabilities than their non-veteran counterparts (9.37%).


The second set of disability measures related to service-connected disability pertain only to those with military service experience. Among the pooled ACS samples since 2008, 31.66% of post-9/11 veterans between ages of 20 and 60 reported having service-connected disabilities. Service-connected disabilities are different than functional disabilities as defined by ACS, which is illustrated by the significant difference between the proportion of post-9/11 veterans who reported having ACS-defined disabilities and the proportion who reported having service-connected disabilities. Nonetheless, one would reasonably expect a correlation between these two disability measures, i.e., veterans with service-connected disabilities are more likely to report ACS-defined disabilities than those without service-connected disabilities. Table 2 illustrates this relationship. Among the pooled sample of veterans since 2008, only 6.58% of those without service-connected disabilities reported ACS-defined disabilities, while a significantly higher proportion, 23.44% of those veterans with service-connected disabilities reported having ACS-defined disabilities. Still, the majority (77%) of veterans with service-connected disabilities did not report having ACS-defined disabilities.


Table 2. Proportion of Veterans Who Reported Disability and Service-connected Disability


  

ACS Defined Disability

 

Total

  

No

 

Yes

   

Service-connected Disability

No

115,095

(93.42%)

 

8,109

(6.58%)

 

123,204

Yes

45,641

(76.56%)

 

13,975

(23.44%)

 

59,616

          



Service-connected Disability Rating

0 percent

2,663

(89.45%)

 

314

(10.55%)

 

2,977

10–20 percent

13,873

(87.44%)

 

1,992

(12.56%)

 

15,865

30–40 percent

10,756

(83.36%)

 

2,147

(16.64%)

 

12,903

50–60 percent

7,550

(77.28%)

 

2,220

(22.72%)

 

9,770

70 percent & above

8,427

(56.44%)

 

6,504

(43.56%)

 

14,931

Not reported

2,372

(74.83%)

 

798

(25.17%)

 

3,170


For those veterans with service-connected disabilities, ACS provides information on disability ratings, which range from 0 to 100% in 10% increments. One would expect that a higher disability rating would be associated with an increased probability of reporting an ACS-defined disability, as illustrated in the lower panel of Table 2. On average, about 23% of veterans with service-connected disabilities reported having ACS-defined disabilities; in this group, 11%, 13%, 17%, 23%, and 44% reported disability ratings of 0%, 10%–20%, 30%–40%, 50%–60%, and 70% and above, respectively.


METHODS


The primary goal of this study is to examine whether the Post-9/11 GI Bill has improved college participation for post-9/11 veterans, especially those with service-connected disabilities. Employing a treatment-control research design, I used the triple-difference technique to identify the program effect based on the timing of program implementation. This technique was especially appropriate for this study because the goal is to evaluate the program effect between subgroups within the treatment group. The treatment group includes veterans who served in the post-9/11 era, some of whom had also served in the Gulf War era. I did not include active duty service members in the treatment group because they receive the monthly housing allowance in the absence of the Post-9/11 GI Bill. As a result, the enrollment effect of the expansion of GI benefit is minimal and insignificant (Zhang, 2018). A natural comparison group may include individuals who never served in military; however, this comparison might overlook important differences between veterans and non-veterans because serving in the military is part of the treatment (e.g., serving the military may alter individuals’ educational expectations that may in turn affect their college participation decisions). Consequently, I used a second comparison group that includes veterans from the Gulf War era—those who served between 1990 and 2001; however, because those who served before 9/11 were at least 27 years old in 2009, I limited the age range for both the Post-9/11 and Gulf War veteran samples between 27 and 60 years old for this set of analyses.


Within both the treatment and comparison groups, differencing between those with and without disabilities would effectively eliminate potential differences between veterans and non-veterans. However, the unavailability of data on service-connected disabilities before 2008 and non-equivalent definitions for ACS-defined disability and service-connected disability created challenges when estimating the effect of the Post-9/11 GI Bill within the natural experiment framework. Ideally, data for the service-connected disability variable would be available for all individuals in all years between 2005 and 2016. However, this is impossible because: (a) by definition, the service-connected disability data are only available for veterans; and (b) data for this variable have only been collected since 2008. I adopted two approaches to overcome these constraints. First, I used data for ACS-defined disabilities for both veteran and non-veteran groups, recognizing that the moderate correlation between ACS-defined disability and service-connected disability would lead to significantly attenuated effects due to measurement error. Second, I used service-connected disability data for veterans and ACS-defined disability data for non-veterans while restricting the pre-policy period to 2008 due to data unavailability for prior years.


Because ACS data are cross-sectional surveys conducted on a rolling basis, pre- and post-policy periods are clearly defined. Except for analyses to evaluate effects over time, I excluded data from 2009 because they span the pre- and post-policy periods. Formally, I used the following ordinary least squares regression:


 [39_23012.htm_g/00004.jpg]

[39_23012.htm_g/00006.jpg]        (1)


where yit is college enrollment for individual i in year t; Veti [39_23012.htm_g/00008.jpg] and [39_23012.htm_g/00010.jpg]. While [39_23012.htm_g/00012.jpg] estimates the effect of the Post-9/11 GI Bill for veterans without service-connected disabilities, [39_23012.htm_g/00014.jpg] yields the additional effect of the Post-9/11 GI Bill for veterans with service-connected disabilities versus veterans without such disabilities. I performed sub-group analyses based on age, sex, race/ethnicity, and disability ratings to provide in-depth and nuanced results.


This standard framework can be readily extended beyond simply comparing pre- and post-policy outcomes. Instead of having the[39_23012.htm_g/00016.jpg] dummy variable in the model, a full set of year fixed effects, each representing a different time period, can be included in the model, yielding the following model:


  [39_23012.htm_g/00018.jpg]

  [39_23012.htm_g/00020.jpg]     (2)


Estimates based on difference-in-differences and triple-difference methods require a “common trends” assumption, i.e., in absence of policy intervention, the treatment and comparison groups would follow parallel paths. I used strategies proposed by Mora and Reggio (2015) to test the “common trends” assumption. In addition, estimates in Equations 1 and 2 may suffer from incorrect statistical inference due to serial correlation (Bertrand, Duflo, & Mullainathan, 2004). In the study, given the large difference in college enrollment rates by age, it was reasonable to control for serial correlation within age cohorts. I used the reghdfe command in Stata to absorb a large number of fixed effects, and estimated clustered standard errors (Correia, 2016), which were noticeably larger when clustered by age. Other clusters (e.g., year or state) had little impact on standard errors.


RESULTS


I present our results in the following order. The section “College enrollment trends” illustrates time trends of college enrollment by veteran and disability status. The section “Effects based on ACS-defined disability" presents the overall impact of the Post-9/11 GI Bill on college participation among veterans with ACS-defined disabilities. Results are reported separately for men and women. The section “Effects based on service-connected disability” continues the analysis by using service-connected disabilities. A variety of potential threats to internal validity are discussed and sensitivity analyses are conducted to check the robustness of main findings. These additional materials are included in Appendix A. The section “Heterogeneous effects among veterans with service-connected disabilities” further examines potentially heterogeneous impacts across groups with different characteristics (i.e., age, race/ethnicity, educational attainment, and disability rating).

College enrollment trends


I began by plotting average enrollment rates for non-veterans and post-9/11 veterans by disability status from 2005 to 2016. Figure 2 is based on ACS-defined disability data for both veteran and non-veteran samples, and Figure 3 is based on service-connected disability data for the veteran sample. The first graph in Figure 2 shows that the college enrollment rate has increased for all four groups (i.e., veteran status by disability status) since 2009, likely due to the Great Recession. However, comparing the two non-veteran groups (non-veterans without disabilities represented by the dotted line with circles, and non-veterans with disabilities represented by the dotted line with triangles) suggests similar time trends before and after the adoption of the Post-9/11 GI Bill in 2009.10 The same comparison between veterans with disabilities and those without disabilities indicates a rather significant increase in college enrollment rates for veterans with disabilities after the implementation of the Post-9/11 GI Bill. In terms of sex differences, the increase in college enrollment rates among veterans with disabilities appears to be driven primarily by men. Among women, there is a great deal of variation in college enrollment rates over time probably due to the relatively small sample of female veterans.


Figure 2. Time trends of college enrollment rates by veteran and ACS-defined disability


 [39_23012.htm_g/00022.jpg]


 [39_23012.htm_g/00024.jpg]


[39_23012.htm_g/00026.jpg]


Figure 3 is based on service-connected disability data for the veteran sample and ACS-defined disability data for the non-veteran sample, revealing time trends starting from 2008 for the veteran sample. Differences in time trends between veterans with service-connected disabilities and those without service-connected disabilities are more pronounced than in Figure 2. While there are no noticeable differences between non-veterans with and without disabilities before and after 2009, the difference between veterans with and without disabilities has widened significantly since the adoption of the Post-9/11 GI Bill, and this is true for both male and female veterans. These observations foreshadow my regression results presented in the following two sections.


Figure 3. Trends of college enrollment by veteran and service-connected Disability


 [39_23012.htm_g/00028.jpg]


 [39_23012.htm_g/00030.jpg]


[39_23012.htm_g/00032.jpg]



EFFECTS BASED ON ACS-DEFINED DISABILITY


The first three columns in Table 3 presents results from a series of triple-difference regression models estimating the overall effect of the Post-9/11 GI Bill on college enrollment. In this set of analyses, ACS-defined disability data are used for both veterans and non-veterans. Two specifications are estimated here. In the first specification (the upper panel) I estimate the average effect of the Post-9/11 GI Bill, while in the second specification (the lower panel) I estimate the effects over time. I only report [39_23012.htm_g/00034.jpg] and [39_23012.htm_g/00036.jpg] in Equations 1 and 2 in the table in order not to overload this paper. Recall that [39_23012.htm_g/00038.jpg] represents the additional effect of the Post-9/11 GI Bill for veterans with disabilities when compared with veterans without disabilities, while [39_23012.htm_g/00040.jpg] represents the effect for veterans without disabilities. Using non-veterans as the comparison group and controlling for a variety of fixed effects and interaction effects, the first column indicates that the Post-9/11 GI Bill increased college enrollment rates of veterans without disabilities by 2.57 percentage points in the overall sample, with larger effects for men (2.78 percentage points) than for women. Most importantly, the increase among veterans with disabilities is significantly higher than among those without disabilities. The overall sample shows an additional increase of 2.13 percentage points. Similar effects are observed for both the male and female veteran samples, although the estimate is not statistically significant for the female sample due to relatively small number of female veterans. These are quite large increases when translated into relative change. Considering the average college enrollment rate of 21% for veterans without disabilities before the adoption of the Post-9/11 GI Bill, an increase of 2.57 percentage points represents a boost of approximately 12% induced by the Post-9/11 GI Bill. For veterans with disabilities, the average college enrollment was 19%; an increase of 4.70 percentage points (i.e., 0.0213 + 0.0257) thus represents a boost of nearly 25%.


Table 3. Effects of Post-9/11 GI Bill on College Enrollment for Veterans with ACS-defined Disabilities


 

Baseline

 

With unemployment and interaction terms

 

All

Men

Women

 

All

Men

Women

Veteran*After 2009*Disability

0.0213*

0.0222*

0.0254

 

0.0207

0.0197

0.0362

 

(0.0100)

(0.0106)

(0.0212)

 

(0.0114)

(0.0114)

(0.0238)

Veteran*After

0.0257***

0.0278***

0.0083

 

0.0198***

0.0214***

-0.0005

 

(0.0037)

(0.0035)

(0.0074)

 

(0.0037)

(0.0033)

(0.0075)

Unemployment

    

0.0006**

0.0009**

0.0003

     

(0.0002)

(0.0003)

(0.0002)

Unemployment*Veteran

    

0.0027***

0.0029***

0.0039*

     

(0.0007)

(0.0007)

(0.0018)

Unemployment*Disability

    

-0.0004

-0.0004

-0.0005

     

(0.0003)

(0.0003)

(0.0003)

Unemployment*Disability*Vet.

    

0.0007

0.0017

-0.0052

     

(0.0022)

(0.0024)

(0.0049)

R2

0.215

0.221

0.216

 

0.215

0.221

0.216

N

12748359

5778574

6969785

 

12748359

5778574

6969785

        

Veteran*Year 2009*Disability

0.0108

0.0154

-0.0106

 

0.0115

0.0122

0.0114

 

(0.0158)

(0.0202)

(0.0428)

 

(0.0176)

(0.0210)

(0.0406)

Veteran*Year 2010*Disability

0.0187

0.0182

0.0295

 

0.0199

0.0146

0.0573

 

(0.0125)

(0.0141)

(0.0421)

 

(0.0210)

(0.0240)

(0.0526)

Veteran*Year 2011*Disability

0.0274*

0.0400**

-0.0227

 

0.0285

0.0373

0.0009

 

(0.0126)

(0.0144)

(0.0314)

 

(0.0176)

(0.0197)

(0.0394)

Veteran*Year 2012*Disability

0.0333*

0.0325

0.0438

 

0.0342

0.0304

0.0614

 

(0.0164)

(0.0171)

(0.0328)

 

(0.0217)

(0.0225)

(0.0368)

Veteran*Year 2013*Disability

0.0030

0.0004

0.0168

 

0.0037

-0.0012

0.0306

 

(0.0138)

(0.0166)

(0.0289)

 

(0.0158)

(0.0173)

(0.0324)

Veteran*Year 2014*Disability

0.0227

0.0266

0.0163

 

0.0233

0.0259

0.0236

 

(0.0133)

(0.0139)

(0.0436)

 

(0.0139)

(0.0145)

(0.0432)

Veteran*Year 2015*Disability

0.0282

0.0243

0.0619

 

0.0285

0.0242

0.0649*

 

(0.0149)

(0.0137)

(0.0309)

 

(0.0150)

(0.0137)

(0.0306)

Veteran*Year 2016*Disability

0.0181

0.0185

0.0252

 

0.0182

0.0185

0.0261

 

(0.0141)

(0.0142)

(0.0314)

 

(0.0141)

(0.0142)

(0.0310)

Veteran*Year 2009

0.0056

0.0097*

-0.0111

 

-0.0120

-0.0056

-0.0383*

 

(0.0037)

(0.0044)

(0.0093)

 

(0.0083)

(0.0090)

(0.0154)

Veteran*Year 2010

0.0195**

0.0242***

-0.0050

 

-0.0023

0.0053

-0.0388

 

(0.0059)

(0.0058)

(0.0108)

 

(0.0106)

(0.0113)

(0.0204)

Veteran*Year 2011

0.0331***

0.0316***

0.0382**

 

0.0148

0.0158

0.0096

 

(0.0060)

(0.0060)

(0.0139)

 

(0.0097)

(0.0097)

(0.0193)

Veteran*Year 2012

0.0330***

0.0381***

0.0053

 

0.0192*

0.0262**

-0.0164

 

(0.0044)

(0.0045)

(0.0105)

 

(0.0073)

(0.0077)

(0.0161)

Veteran*Year 2013

0.0329***

0.0365***

0.0103

 

0.0218**

0.0269***

-0.0071

 

(0.0035)

(0.0046)

(0.0101)

 

(0.0064)

(0.0070)

(0.0130)

Veteran*Year 2014

0.0292***

0.0285***

0.0191

 

0.0235***

0.0236***

0.0101

 

(0.0041)

(0.0043)

(0.0099)

 

(0.0045)

(0.0046)

(0.0109)

Veteran*Year 2015

0.0149***

0.0166***

-0.0042

 

0.0125**

0.0145***

-0.0081

 

(0.0041)

(0.0035)

(0.0114)

 

(0.0042)

(0.0036)

(0.0116)

Veteran*Year 2016

0.0187**

0.0203***

-0.0050

 

0.0181**

0.0199***

-0.0059

 

(0.0059)

(0.0055)

(0.0118)

 

(0.0059)

(0.0055)

(0.0117)

Unemployment

    

0.0007***

0.0009***

0.0005*

     

(0.0002)

(0.0002)

(0.0002)

Unemployment*Veteran

    

0.0046**

0.0040*

0.0071

     

(0.0017)

(0.0017)

(0.0036)

Unemployment*Disability

    

0.0005

0.0005

0.0003

     

(0.0004)

(0.0005)

(0.0005)

Unemployment*Disability*Vet.

    

-0.0002

0.0008

-0.0059

     

(0.0035)

(0.0039)

(0.0070)

R2

0.216

0.222

0.216

 

0.216

0.222

0.216

N

13900840

6293923

7606917

 

13900840

6293923

7606917

Note: Variables included but not reported: veteran status, disability, year fixed effects, disability*after (or disability*year fixed effects in the lower panel), veteran*disability, age fixed effects, race/ethnicity fixed effects, state fixed effects, veteran*age dummies, veteran*race dummies, veteran*state dummies. Year 2009 was excluded from the upper panel. All models were weighted by ACS person weights. Standard errors (in parentheses) were clustered by age. * p < 0.05, ** p < 0.01, *** p < 0.001.


Estimating these effects over time reveals considerable variations. The lower panel shows a significant boost in enrollment rates for veterans with disabilities and those without disabilities immediately after the adoption of the Post-9/11 GI Bill. The effect, however, quickly plateaued in 2011 and 2012, and then decreased gradually in subsequent years. For example, for veterans without disabilities, enrollment effects were over 3 percentage points between 2011 and 2013, but declined to less than 2 percentage points in recent years. A comparison between veterans with disabilities and veterans without disabilities reveals an advantage of around 3 percentage points in 2011 and 2012 that became smaller after these initial years.


Several potential threats to internal validity are considered here. First, the assumption of parallel enrollment patterns between veterans and non-veterans is necessary because the latter group is used as the counterfactual group in the analysis. I estimated a variation of the time effect model by using 2005 as the base year to see whether there were noticeable differences in enrollment patterns between veterans and non-veterans prior to the adoption of the Post-9/11 GI Bill in 2009. Regression results (not reported here but available upon request) largely confirmed the assumption of parallel enrollment patterns, with all estimates of [39_23012.htm_g/00042.jpg] and [39_23012.htm_g/00044.jpg] statistically insignificant before 2009 when compared with 2005. In other words, while there were differences in college enrollment rates between veterans and non-veterans and between veterans with and without disabilities, their time trends appeared to be similar before the adoption of the Post-9/11 GI Bill.


Second, although the Post-9/11 GI Bill went into effect in August 2009, it became law in June 2008, making it possible that some veterans who would have attended college in 2008 in the absence of the Post-9/11 GI Bill had chosen to postpone their college enrollment, which could result in an over-estimation of the program effect when year 2008 is used as the only pre-policy period, which unfortunately is the only viable approach when service-connected disability is used in the next section. To check for this possibility, I replicated the first three columns in Table 3 but only included 2008 as the pre-policy period to see whether using 2008 as the pre-policy period would significantly increase the enrollment effects. Results (full regression results are available upon request) indicated very similar effects ([39_23012.htm_g/00046.jpg] = 0.0266; [39_23012.htm_g/00048.jpg] = 0.0265) for men when compared with using 2005–2008 as the pre-policy period (see Table 3, [39_23012.htm_g/00050.jpg] = 0.0222; [39_23012.htm_g/00052.jpg] =0.0278); however, the effects were noticeably larger for women ([39_23012.htm_g/00054.jpg]= 0.077; [39_23012.htm_g/00056.jpg] = 0.010), which is consistent with the low enrollment rate for women veterans with ACS-defined disability in 2008 (see Figure 2, the third graph). Because male veterans account for 83% of all veterans in the analytical sample, limiting the pre-policy period to 2008 is not likely to cause a large bias in the estimated effect.


Finally, the estimated effects may be susceptible to confounding factors, with the leading confounding factor in this particular case being the Great Recession that officially started in December 2007. While year fixed effects control for overall shift of enrollment patterns due to Great Recession, they do not account for potentially different effects of the Great Recession for veterans and non-veterans. For example, because financial aid is tighter in recession years (Clelan & Kofoed, 2017), access to Post-9/11 GI Bill education benefits may yield an extra boost to veterans. To test this possibility, I include unemployment rates by state and year, and all of its interaction terms with veteran and disability. Results are reported in the last three columns in Table 3. Not surprisingly, while there is an overall positive relationship between unemployment rate and enrollment rate, the relationship is much stronger for veterans. As a result, when the interaction between unemployment rate and veteran status is added to the model, the overall enrollment effect for veterans as a group decreases. However, the unemployment rate does not seem to have differential effects on veterans with disabilities. As a result, when unemployment rate and interaction terms are added to the model, the enrollment effect of Post-9/11 GI Bill decreases slightly for veterans without disabilities and remains essentially the same for veterans with disabilities. Although it can be argued that an amplified enrollment effect due to the Great Recession is part of the Post-9/11 BI Bill effect, it is in principle important to isolate the effect of the Great Recession from that of the Post-9/11 GI Bill. Consequently, I included the unemployment rate and its related interaction terms in all subsequent data analyses.   


EFFECTS BASED ON SERVICE-CONNECTED DISABILITY


Because a large proportion of veterans with service-connected disabilities did not report ACS-defined disabilities, the results in Table 3 could be potentially biased. To be specific, the estimated [39_23012.htm_g/00058.jpg] could be biased downward while [39_23012.htm_g/00060.jpg] could be biased upward. In this section, I use data for service-connected disabilities; however, the pre-policy period has to be restricted to 2008 due to a lack of data on service-connected disability prior to that date. The first three columns in Table 4 compare post-9/11 veterans with non-veterans. Results in the first column indicate positive but statistically insignificant effects of the Post-9/11 GI Bill on college enrollment rates for veterans without disabilities. However, the increase among veterans with service-connected disabilities is much higher than among those without service-connected disabilities. For example, in the overall sample, while the enrollment effect for veterans without service-connected disabilities is less than 1 percentage point (and statistically insignificant), the additional effect on veterans with service-connected disabilities is 5.01 percentage points. This large and significant increase in enrollment rates is observed in samples of both male and female veterans. Estimates of these effects over time reveal consistently large and significant effects for veterans with service-connected disabilities, with larger effects immediately after the policy adoption in 2009 and smaller effects in more recent years.


Table 4. Effects of Post-9/11 GI Bill on College Enrollment for Veterans with Service-connected Disabilities

 

Comparison: Non-veterans

 

Comparison: Gulf-war veterans

 

All

Men

Women

 

All

Men

Women

Veteran*After 2009*Disability

0.0501***

0.0490***

0.0610*

 

0.0376**

0.0358**

0.0378

 

(0.0102)

(0.0097)

(0.0267)

 

(0.0111)

(0.0104)

(0.0422)

Veteran*After 2009

0.0042

0.0054

-0.0153

 

0.0003

0.0029

-0.0044

 

(0.0061)

(0.0064)

(0.0114)

 

(0.0072)

(0.0060)

(0.0203)

R2

0.218

0.223

0.221

 

0.070

0.067

0.079

N

9401544

4324253

5077291

 

289462

244449

45013

        

Veteran*Year 2009*Disability

0.0343*

0.0345*

0.0325

 

0.0235

0.0232

0.0183

 

(0.0153)

(0.0160)

(0.0324)

 

(0.0216)

(0.0210)

(0.0579)

Veteran*Year 2010*Disability

0.0450**

0.0387*

0.0879*

 

0.0454*

0.0333

0.0901

 

(0.0156)

(0.0164)

(0.0365)

 

(0.0182)

(0.0191)

(0.0497)

Veteran*Year 2011*Disability

0.0711***

0.0719***

0.0805

 

0.0620*

0.0516*

0.1038

 

(0.0165)

(0.0179)

(0.0420)

 

(0.0227)

(0.0236)

(0.0583)

Veteran*Year 2012*Disability

0.0575***

0.0512***

0.1004**

 

0.0487**

0.0414**

0.0805

 

(0.0113)

(0.0125)

(0.0288)

 

(0.0139)

(0.0144)

(0.0538)

Veteran*Year 2013*Disability

0.0531***

0.0526***

0.0587*

 

0.0356*

0.0287

0.0633

 

(0.0122)

(0.0129)

(0.0259)

 

(0.0149)

(0.0181)

(0.0347)

Veteran*Year 2014*Disability

0.0523***

0.0543***

0.0537

 

0.0249

0.0283*

0.0093

 

(0.0121)

(0.0115)

(0.0277)

 

(0.0163)

(0.0133)

(0.0533)

Veteran*Year 2015*Disability

0.0485***

0.0461***

0.0631

 

0.0438**

0.0420**

0.0409

 

(0.0122)

(0.0125)

(0.0351)

 

(0.0127)

(0.0129)

(0.0484)

Veteran*Year 2016*Disability

0.0486***

0.0469***

0.0643*

 

0.0448**

0.0401*

0.0610

 

(0.0126)

(0.0118)

(0.0314)

 

(0.0145)

(0.0147)

(0.0436)

Veteran*Year 2009

-0.0098

-0.0049

-0.0330

 

-0.0063

-0.0007

-0.0402

 

(0.0097)

(0.0108)

(0.0208)

 

(0.0138)

(0.0136)

(0.0403)

Veteran*Year 2010

-0.0011

0.0056

-0.0408

 

-0.0065

0.0021

-0.0534

 

(0.0127)

(0.0128)

(0.0279)

 

(0.0165)

(0.0151)

(0.0415)

Veteran*Year 2011

0.0071

0.0069

0.0014

 

0.0100

0.0077

0.0172

 

(0.0123)

(0.0123)

(0.0240)

 

(0.0152)

(0.0151)

(0.0349)

Veteran*Year 2012

0.0123

0.0192*

-0.0283

 

0.0143

0.0204

-0.0207

 

(0.0081)

(0.0087)

(0.0203)

 

(0.0101)

(0.0101)

(0.0329)

Veteran*Year 2013

0.0100

0.0131

-0.0140

 

0.0055

0.0126

-0.0278

 

(0.0078)

(0.0085)

(0.0181)

 

(0.0097)

(0.0083)

(0.0306)

Veteran*Year 2014

0.0098

0.0085

0.0009

 

0.0100

0.0107

0.0068

 

(0.0061)

(0.0068)

(0.0135)

 

(0.0068)

(0.0056)

(0.0233)

Veteran*Year 2015

-0.0004

0.0016

-0.0224

 

-0.0031

-0.0010

-0.0021

 

(0.0065)

(0.0068)

(0.0135)

 

(0.0078)

(0.0068)

(0.0200)

Veteran*Year 2016

-0.0002

0.0017

-0.0247

 

-0.0097

-0.0066

-0.0156

 

(0.0090)

(0.0089)

(0.0167)

 

(0.0097)

(0.0086)

(0.0244)

R2

0.219

0.224

0.221

 

0.068

0.065

0.077

N

10553396

4839069

5714327

 

321956

272023

49933

Note: Variables included but not reported in all models: veteran status, disability, year fixed effects, disability*after (or disability*year fixed effects in the lower panel), veteran*disability, age fixed effects, race/ethnicity fixed effects, state fixed effects, veteran*age dummies, veteran*race dummies, veteran*state dummies, unemployment rate, unemployment*veteran, unemployment*disability, unemployment*veteran*disability. Year 2009 was excluded from the first panel. All models were weighted by ACS person weights. Standard errors (in parentheses) were clustered by age. * p < 0.05, ** p < 0.01, *** p < 0.001.


For veterans without service-connected disabilities, the effects are small and only significant in peak years. This small enrollment effect may be due to a couple of reasons. First, because the Post-9/11 GI Bill is an expansion of the MGIB which already provided quite generous educational benefits, the marginal effect of providing additional financial incentives is small. Second, not all veterans qualify for the full benefits provided by the Post-9/11 GI Bill. Because education benefits are tiered based on the length of time served on active duty, a large proportion of veterans only qualify for partial benefits. In addition, when veteran students attended college on a part-time schedule, their educational benefits would be adjusted accordingly. For example, a 9-credit course load qualifies 80% of full benefits. For these reasons, the average educational benefits received by post-9/11 veterans is far less than the full amount. In fiscal year 2016, the average post-9/11 educational benefit was $14,661, while the average MGIB educational benefit was $7,718 (U.S. Department of Veterans Affairs, 2018).


Because serving in the military itself could change veterans’ education expectations, it is plausible that being veterans, not necessarily receiving educational benefits, may make veterans more likely to attend colleges during economic downturns. To check for this possibility, I used veterans from the Gulf War era as the comparison group and re-estimated the triple-difference equation. In these comparisons, however, the age range was restricted to veterans between 27 and 60 years old because those who served before 9/11 were at least 27 years old in 2009. Results are reported in the last three columns in Table 4. The estimates using Gulf War veterans as the comparison group are somewhat smaller than those using non-veterans as the comparison group, although effects are still quite large and significant. The decrease in the average effect from 5.01 to 3.76 percentage points could be partly due to the change in age restrictions. To investigate this possibility, I re-estimated the first three columns of Table 4, but this time restricted the age range to between 27 and 60. Results (not reported here) indicate an average effect of 4.42 percentage points, suggesting that age range could be partially responsible for the decrease in the estimated effect between the first and last three columns in Table 4. Detailed analyses by age groups are presented in the next section.


Could this large effect for veterans with service-connected disability be driven by other disability related policies? I consider several possibilities including changes in disability status, ratings, and compensations. Detailed analyses are reported in Appendix A, and collectively they suggest that the large effect for veterans with service-connected disability cannot be explained by these salient factors.


HETEROGENEOUS EFFECTS AMOUNG VETERANS WITH SERVICE-CONNECTED DISABILITIES


Effects by Age


The overall effects of the Post-9/11 GI Bill on veterans’ college enrollment might have disguised variations across veterans in different age groups. I estimated the effects by age (i.e., a total of 41 regressions, one for each year of age) and results are reported in Appendix B. The estimated effect of the Post-9/11 GI Bill on college enrollment rates for veterans without service-connected disabilities oscillate around zero across age groups, suggesting a rather small average effect. However, the additional effects for those with service-connected disabilities are positive across most age groups, although most estimates are not statistically significant due to small sample sizes. Estimated effects are slightly larger for younger groups (i.e., 20s and early 30s) than for older groups, which is consistent with a smaller effect when the age range is restricted to between 27 and 60 as in the last three columns of Table 4. It is important to note that because college enrollment rates decrease with age, a larger percentage increase in enrollment does not necessarily translate into a larger relative change for younger veterans. For example, because the average enrollment rate for the 20 to 35 year olds is approximately 27%, an increase of 7.3 percentage points is equivalent to a 27% increase in the enrollment rate. In contrast, for the 40- to 50-year-olds, the enrollment effect is 5.3 percentage points, reflecting an increase of more than 40% relative to the average enrollment rate of 12.5% for this group of veterans.


Effects by Race/Ethnicity


Separate models were estimated for White, Black, Hispanic, Asian, and other races that include American Indian, Alaska native, and multiple races. Results are reported in Appendix C. These results suggest some but not large variations across racial groups. For example, the estimated effects for White and Black veterans (4.98 and 4.63 percentage points respectively) are very similar to the overall effect in Table 4 (5.01 percentage points); however, the effect is larger for the Hispanic group (8.01 percentage points). Similar patterns emerge when comparing estimated effects over time among different racial groups.


Effects by Educational Attainment


Because education benefits provided by the Post-9/11 GI Bill can be used for a variety of degree and training programs, it is important to examine how veterans with different levels of educational attainment react to financial incentives. Separate regressions for individuals with different levels of educational attainment are reported in Appendix D. Results indicate that except for the group with less than one year of college education, the Post-9/11 GI Bill has very small effects for those without service-connected disabilities; however, the additional enrollment effect for veterans with service-connected disabilities is large and significant for almost all levels of educational attainment. The large estimate for those with graduate degrees suggests that many veterans have taken advantage of educational benefits to pursue advanced degrees.


Effects by Disability Ratings


How do these effects vary by disability ratings? I adopted two approaches to answer this question. In the first set of analyses, I examined the conditional effects of the Post-9/11 GI Bill for veterans with different disability ratings. Results are presented in Appendix E. Scanning across columns in this table reveals quite consistent and large effects of the Post-9/11 GI Bill for veterans with different disability ratings. It appears that college enrollment rates for veterans with low disability ratings (i.e., 10% to 40%) increased more than for veterans with high ratings (i.e., over 50%), although the pattern is not unequivocal.


The other approach was to examine the interaction between service-connected disability and ACS-defined disability. Since a positive—albeit modest—correlation exists between these two measures of disability, one might expect a larger enrollment effect for veterans with service-connected disabilities but not ACS-defined disabilities than for those with both service-connected disabilities and ACS-defined disabilities. I performed separate regression analyses for these two groups; results are reported in Appendix F. This set of analyses yielded less ambiguous results than those reported in Appendix E: College enrollment rates improved more for veterans with service-connected disabilities but not ACS-defined disabilities, than for those with both service-connected-disabilities and ACS-defined disabilities.


DISCUSSION AND CONCLUSION


The results of this study provide strong evidence for the significant enrollment growth among veterans with service-connected disabilities after the adoption of the Post-9/11 GI Bill. While the Post-9/11 GI Bill has increased college enrollment for veterans without service-connected disabilities by less than 1 percentage point, the increase is much larger—over 5 percentage points—for veterans with service-connected disabilities. Over time, estimates show consistent, large and significant effects for veterans with service-connected disabilities, with larger effects immediately after the policy adoption in 2009 and smaller effects in more recent years. The large overall enrollment effect is robust to different disability measures, model specifications, and comparison groups used.


The large and significant enrollment effect for veterans with service-connected disabilities is reassuring. Since the average enrollment rate for veterans with service-connected disabilities was approximately 20% before the adoption of the Post-9/11 GI Bill, an increase of 5 percentage points is equivalent to a 25 percent increase. Because of lack of detailed data on actual educational benefits received by individuals in ACS, the results presented in this study cannot tell whether this large effect for veterans with service-connected disabilities is due to favorable provisions in the Post-9/11 GI Bill for them or due to lower opportunity costs. Maybe they work in tandem to produce this large enrollment boost. In the future, researchers may want to identify appropriate sources for data on detailed educational benefits to examine the mechanisms behind the effect.  


While the evidence for large and significant overall enrollment effects for veterans with service-connected disabilities is strong, the evidence for potentially heterogeneous effects within veterans with service-connected disabilities is less conspicuous. For example, the enrollment growth is similar for both male and female veterans with service-connected disabilities. All age groups have benefited substantially from the Post-9/11 GI Bill. Younger cohorts appear to have a slight edge in absolute increase of enrollment rates; however, older cohorts have an advantage in relative increase because of their low enrollment rates. Among all race/ethnicity groups, veterans with Hispanic background seem to be most responsive to the educational benefits provided by the Post-9/11 GI Bill. Enrollment effects for veterans with service-connected disabilities are consistent across different levels of educational attainment, suggesting that many veterans have taken advantage of educational benefits to pursue college and advanced degrees. Finally, results suggest a slightly larger enrollment effect for veterans with lower disability ratings than for those with higher disability ratings. College enrollment rates have increased more among veterans with service-connected disabilities but not ACS-defined disabilities than among those with both service-connected disabilities and ACS-defined disabilities.


The lack of significant variations in the program effects among veterans with service-connected disabilities is, in fact, welcoming news. It is encouraging that all veterans with service-connected disabilities—regardless of their sex, age, race/ethnicity, educational background, and disability ratings—were able to take advantage of the Post-9/11 GI Bill and return to college campus. The growth in the number of veterans on college campuses, especially those with service-connected disabilities, requires those affiliated with higher education institutions (e.g., faculty, staff, and administrators) to understand and meet the needs of this unique student population. On the one hand, research on veterans’ college experiences has revealed that, beyond access, veterans face challenges associated with service-related injuries and disabilities, as well as their status as non-traditional age students (Ford & Vignare, 2014; Steele, Salcedo, & Coley, 2010; Vacchi & Berger, 2014). Many colleges and universities have taken major steps in recent years to develop programs and services to serve veteran and military students (McBain, Kim, Cook, & Snead, 2012), which are likely to help veteran students succeed. On the other hand, it is not helpful to exaggerate the difficulties faced by veteran students on college campuses by drawing upon stereotypes (Vacchi, 2012). Embracing these veteran students while providing professional services will help them maximize their potentials in college and successfully transition to civilian life.


Notes


1. Under the Montgomery GI Bill, active duty members have one chance to buy into the MGIB ($100 per month for 12 months) and receive monthly education benefits after they complete a minimum service obligation.


2. A recent study by Elnitsky, Blevins, Findlow, Alverino, and Wiese (2018) examines medical challenges faced by veteran students and the effects of campus service on veteran students’ well-being.


3. Other benefits include a one-time relocation allowance, the “Yellow Ribbon” program that helps pay for expensive private colleges, the option to transfer benefits to family members after 10 years of service, extending the benefit eligibility period to 15 years (from 10 years under MGIB), and eliminating the $1,200 enrollment fee required by the MGIB (or refunding the MGIB enrollment fee upon conversion to the Post-9/11 GI Bill). The most recent expansion in the Forever GI Bill has eliminated the 15-year expiration period.


4. Since the Forever GI Bill was passed in 2017, the monthly housing allowance is now calculated based on where the student attends the majority of classes.


5. Author’s own calculation based on CPS Veteran Supplement data between 2009 and 2016.


6. It is important to note that the unemployment rate for veterans with service-connected disabilities (4.8%) is similar to that for veterans without service-connected disabilities (4.7%). The gap in employment-population ratio between veterans with and without service-connected disabilities is mainly due to the fact that a larger proportion of veterans with service-connected disabilities are not in the labor force.


7. Depending on the context, the definition of disability could vary. For example, to be eligible for Social Security disability benefits, a person must have a severe medical condition that is expected to last at least one year or result in death, and that prevents working at a “substantial gainful activity” level.


8. Service-connected disabilities are evaluated according to U.S. Department of Veterans Affairs’ Schedule for Rating Disabilities in Title 38, U.S. Code of Federal Regulations. Zero percent is a valid disability rating and is different from having no service-connected disabilities. A detailed schedule of rating disability is available at https://www.benefits.va.gov/warms/bookc.asp. To give a couple of examples, recurrent tinnitus is rated at 10%, while severe duodenal ulcer is rated 60%.


9. Relaxing the place of birth restriction does not change the results reported in this paper. The sample size of veterans over 60 years old is very small, and their college enrollment rates are very low.


10. Additional statistical tests (Mora & Reggio, 2015) between individuals with and without disabilities separately for veteran and non-veteran groups could not reject common pre-treatment trends.


Acknowledgements


This project is funded by Spencer Foundation (Grant #201800080); however, all views expressed in this article are those of the author. The author would like to thank Adam Fullerton at University of Nebraska Lincoln for his insights into the intricacies of the Post-9/11 GI Bills. The author would also like to thank several anonymous reviewers for their valuable comments and suggestions to improve the quality of the paper.  


References


Adams, K. S., & Proctor, B. E. (2010). Adaptation to college for students with and without disabilities:

Group differences and predictors. Journal of Postsecondary Education and Disability22(3), 166–184.


Angrist, J. D. (1993). The effect of veterans benefits on education and earnings. ILR Review46(4), 637–

652.


Angrist, J., Autor, D., Hudson, S., & Pallais, A. (2016). Updated results from a randomized evaluation of

post-secondary aid. Available at https://www8.gsb.columbia.edu/faculty-research


Barnard-Brak, L., Davis, T., Tate, A., & Sulak, T. (2009). Attitudes as a predictor of college students

requesting accommodations. Journal of Vocational Rehabilitation31(3), 189–198.


Barr, A. (2015). From the battlefield to the schoolyard: The short-term impact of the Post-9/11 GI

Bill. Journal of Human Resources50(3), 580–613.


Barr, A. (2019). Fighting for education: Veterans and Financial Aid. Journal of Labor Economics, 37(2), 509–544.


Bertrand, M., Duflo, E., & Mullainathan, S. (2004). How much should we trust differences-in-differences

estimates? The Quarterly Journal of Economics119(1), 249–275.


Bound, J., & Turner, S. (2002). Going to war and going to college: Did World War II and the GI Bill

increase educational attainment for returning veterans? Journal of Labor Economics20(4), 784–815.


Bureau of Labor Statistics (2017). Employment Situation of Veterans. Available at

https://www.bls.gov/news.release/vet.nr0.htm


Castleman, B. L., & Long, B. T. (2016). Looking beyond enrollment: The causal effect of need-based

grants on college access, persistence, and graduation. Journal of Labor Economics34(4), 1023–1073.


Clelan, E., & Kofoed, M. S. (2017). The effect of the business cycle on freshman financial aid.

Contemporary Economic Policy, 35(2), 253–268.


Correia, S. (2016). REGHDFE: Stata module to perform linear or instrumental-variable regression

absorbing any number of high-dimensional fixed effects. Statistical Software Components, August 6.


Deming, D., & Dynarski, S. (2010). College aid. In P. Levine & D. Zimmerman (Eds.), Targeting

investments in children: Fighting poverty when resources are limited (pp. 283–302). Chicago, IL:

University of Chicago Press.


Dynarski, S. (2000). Hope for whom? Financial aid for the middle class and its impact on college

attendance. National Tax Journal, 53(3), 629–661.


Elnitsky, C. A., Blevins, C., Findlow, J. W., Alverio, T., & Wiese, D. (2018). Student veterans reintegrating

from the military to the university with traumatic injuries: how does service use relate to health

status?. Archives of Physical Medicine and Rehabilitation99(2), S58–S64.


Ford, K., & Vignare, K. (2014). The evolving military learner population: A review of the literature. Online

Learning19(1), 7–30.


Goldrick-Rab, S., Kelchen, R., Harris, D. N., & Benson, J. (2016). Reducing income inequality in

educational attainment: Experimental evidence on the impact of financial aid on college

completion. American Journal of Sociology121(6), 1762–1817.


Hossler, D., Braxton, J., & Coopersmith, G. (1989). Understanding student college choice. In John Smart

(Eds.) Higher education: Handbook of theory and research (Vol. 5, pp. 231–288). New York: Agathon

Press.


Long, B. T. (2004). The impact of federal tax credits for higher education expenses. In C. M. Hoxby (Ed.),

College choices: The economics of where to go, when to go, and how to pay for it (pp. 101–168). Chicago,

IL: University of Chicago Press.


Madaus, J. W., & Miller, W. K. (2009). Veterans with disabilities in postsecondary education. Journal of

Postsecondary Education and Disability22(1), 10–17.


Mamiseishvili, K., & Koch, L. C. (2012). Students with disabilities at 2-year institutions in the United States:

Factors related to success. Community College Review40(4), 320–339.


McBain, L., Kim, Y. M., Cook, B. J., & Snead, K. M. (2012). From soldier to student II: Assessing campus

programs for veterans and service members. Washington, DC: American Council on Education.


Mora, R., & Reggio, I. (2015). didq: A command for treatment-effect estimation under alternative

assumptions. Stata Journal, 15(3), 796–808.


Murray, C., Lombardi, A., Bender, F., & Gerdes, H. (2013). Social support: Main and moderating effects

on the relation between financial stress and adjustment among college students with disabilities. Social

Psychology of Education16(2), 277–295.


Newman, L., Wagner, M., Knokey, A. M., Marder, C., Nagle, K., Shaver, D., & Wei, X. (2011). The Post-

High School Outcomes of Young Adults with Disabilities up to 8 Years after High School: A Report from

the National Longitudinal Transition Study-2. (NCSER 2011-3005). Menlo Park, CA: SRI International


Orr, A. C., & Hammig, S. B. (2009). Inclusive postsecondary strategies for teaching students with learning

disabilities: A review of the literature. Learning Disability Quarterly32(3), 181–196.


Perna, L. W. (2006). Studying college access and choice: A proposed conceptual model. In John Smart

(Ed.), Higher education: Handbook of theory and research (Vol. 21, pp. 99–157). The Netherlands:

Springer.


Perna, L. W. (2010). Toward a more complete understanding of the role of financial aid in promoting

college enrollment: The importance of context. In John Smart (Ed.), Higher education: Handbook of

theory and research (Vol. 21, pp. 129–180). The Netherlands: Springer.


Polloway, E. A., Patton, J. R., & Serna, L. (2008). Strategies for teaching learners with special needs.

Upper Saddle River, NJ: Prentice Hall.  


Simon, C. J., Negrusa, S., & Warner, J. T. (2010). Educational benefits and military service: An analysis

of enlistment, reenlistment, and veterans’ benefit usage 1991–2005. Economic Inquiry48(4), 1008–1031.


Sjoquist, D. L., & Winters, J. V. (2012). Building the stock of college-educated labor revisited. Journal of

Human Resources47(1), 270-–285.


Steele, J. L., Salcedo, N., & Coley, J. (2010). Service members in school: Military veterans' experiences

using the Post-9/11 GI Bill and pursuing postsecondary education. Santa Monica, CA: Rand Corporation.


U.S. Department of Veterans Affairs (2011). Veterans benefits administration annual benefits report

Fiscal Year 2010. Retrieved from http://benefits.va.gov/


U.S. Department of Veterans Affairs (2012). Veterans benefits administration annual benefits report

Fiscal Year 2011. Retrieved from http://benefits.va.gov/


U.S. Department of Veterans Affairs (2018). Veterans benefits administration annual benefits report

Fiscal Year 2016. Retrieved from http://benefits.va.gov/


Vacchi, D. T. (2012). Considering student veterans on the twentyfirstcentury college campus. About

Campus17(2), 15–21.


Vacchi, D. T., & Berger, J. B. (2014). Student veterans in higher education. In John Smart (Ed.), Higher

education: Handbook of theory and research (pp. 93–151). The Netherlands: Springer.


Zhang, L. (2018). Veterans going to college: Evaluating the impact of the Post-9/11 GI Bill on college enrollment. Educational Evaluation and Policy Analysis, 40(1), 82–102.



APPENDIX A


Additional analyses were conducted to examine whether the large enrollment effect for veterans with service-connected disability could be due to other factors including disability status, ratings, and compensations. First, it might be possible that veterans who wanted to attend college were more likely to file service-connected disability claims, although this is unlikely because service-connected disability ratings are required to receive a range of compensation including disability compensation, adapted housing grants, and service-disabled veterans’ insurance. Nonetheless, to explore this possibility, I plotted in Appendix Figure A the proportion of veterans with service-connected disabilities over time (the solid line with the left y axis) against the estimated effect of the Post-9/11 GI Bill for veterans with service-connected disabilities (the dashed line with the right y axis). The proportion of veterans with service-connected disabilities hovered around 27% between 2008 and 2011, and then increased gradually over the years, up to about 38% in 2016. This uptick in recent year is partially due to a change in law in July 2010 that made it easier for veterans seeking compensation for Post-Traumatic Stress Disorder (PTSD) by reducing the evidence needed if the PTSD stressor is related to fear of hostile military or terrorist activity and is consistent with the places, types, and circumstances of the veteran’s service. Following this change, there was an increase in the number of new disability compensation recipients who claimed PTSD in 2011 (U.S. Department of Veterans Affairs, 2012). If veterans who wanted to attend college were more likely to file for and subsequently receive disability ratings, one would expect an immediate increase in the proportion of veterans with service-connected disabilities after 2009. It is noteworthy that according to Veteran’s Affairs, it takes 118 days on average to process a service-connected disability claim. Most importantly, one would see a positive relationship between the proportion of veterans with service-connected disabilities and the estimated effect of the Post-9/11 GI Bill. In contrast, Appendix Figure 1 shows an increase in the estimated effect between 2009 and 2011, and a decrease in more recent years when the proportion of veterans with service-connected disabilities increased. The time trend in the estimated effect is consistent with the retrospective nature of the bill. In other words, the Post-9/11 GI Bill may have encouraged a large number of veterans who did not attend college under MGIB to change their college participation decisions under the Post-9/11 GI Bill, resulting in a temporary enrollment burst immediately after the bill’s adoption.


Second, the composition of disability ratings among veterans with service-connected disabilities have changed over time. In particular, there has been a gradual increase in disability ratings in recent years. Appendix Table 1 assembles data from Annual Benefits Reports published by the Department of Veterans Affairs. Recall that disability ratings range from 0 to 100% in 10% increments. Column 1 indicates that the weighted average disability ratings of all new disability compensation recipients has increased from 32% in early 2000 to 46% in recent years. If veterans with higher disability ratings are more likely to attend college, then an increase in disability ratings over time would likely improve college enrollment for veterans with service-connected disabilities as a group. In the ACS sample, the average enrollment rates are 24.4%, 23.7%, 25.6%, 25.6%, and 24.2% for disability ratings of 0%, 10%–20%, 30%–40%, 50%–60%, and 70% & above, respectively. The slightly higher college enrollment rates for veterans with high disability ratings cannot explain the large increase in college enrollment for veterans with service-connected disabilities as a group. Indeed, when the disability ratings were included in the regression model, the estimated [39_23012.htm_g/00062.jpg] only decreased slightly from 5.01 to 4.65.  


Finally, a less obvious but nonetheless important possibility is the increase of disability compensation over time, which could create an income effect assuming that college education is a “normal good.” While a thorough review of policies related to disability compensation does not reveal significant changes (except for the change related to PTSD in 2010) in recent years, the average annual disability compensation (Appendix Table 1, Column 2) has increased from about $8,000 in early 2000 to over $11,000 in recent years (numbers in 2016 constant dollars). Examining the compensation by disability ratings over time, however, suggests that the increase in average compensation largely reflects the increase in disability ratings over time but not a real increase in disability compensation by levels of disability ratings. For example, the last column indicates that the average compensation for veterans with a 50% disability rating has not changed since early 2000. While it is impossible to test directly whether an increased disability compensation due to an increase in disability ratings would lead to higher college enrollment rates, the small difference in college enrollment rates across different disability ratings suggests that the impact would be minimal, if it exists at all.


Appendix Figure 1. Proportion of Veterans with Service-connected Disability and Enrollment Effect over Time


 [39_23012.htm_g/00064.jpg]

Appendix Table 1. Disability Ratings and Disability Compensation of New Disability Compensation Recipients by Year


Year

Average disability ratings

Average compensation

Compensation for ratings of 50%

2001

31.7

8,036

10,920

2002

32.2

8,305

11,082

2003

32.8

8,567

11,002

2004

33.2

8,691

10,899

2005

33.6

8,784

10,883

2006

33.4

8,618

10,858

2007

35.0

8,129

10,733

2008

35.6

8,027

10,430

2009

35.9

8,732

11,097

2010

35.0

8,253

11,265

2011

37.0

8,741

10,413

2012

41.6

10,028

10,465

2013

43.6

10,692

10,648

2014

44.8

11,102

10,608

2015

45.8

11,675

10,776

2016

45.8

11,661

10,729

Data Source: Annual Benefits Report (U.S. Department of Veterans Affairs, various years). Disability compensations are in 2016 constant dollars.



APPENDIX B


EFFECTS OF POST-9/11 GI BILL ON COLLEGE ENROLLMENT ACROSS AGE GROUPS


(a) Effects on veterans without service-connected disabilities

[39_23012.htm_g/00066.jpg]

(b) Additional effects on veterans with service-connected disabilities

[39_23012.htm_g/00068.jpg]


APPENDIX C


EFFECTS OF POST-9/11 GI BILL ON COLLEGE ENROLLMENT BY RACE/ETHNICITY


 

White

Black

Hispanic

Asian

Other

Veteran*After 2009*Disability

0.0498***

0.0463

0.0801*

0.0435

0.0201

 

(0.0098)

(0.0287)

(0.0316)

(0.1012)

(0.0639)

Veteran*After 2009

0.0030

0.0139

-0.0206

-0.0281

0.0603

 

(0.0071)

(0.0136)

(0.0229)

(0.0441)

(0.0347)

R2

0.236

0.134

0.182

0.375

0.180

N

7342265

967081

698584

140405

253209

      

Veteran*Year 2009*Disability

0.0378*

0.0292

0.0715

-0.1541

0.0279

 

(0.0163)

(0.0522)

(0.0572)

(0.2037)

(0.0874)

Veteran*Year 2010*Disability

0.0457*

0.0336

0.1825**

-0.1345

-0.0900

 

(0.0171)

(0.0506)

(0.0661)

(0.1940)

(0.0910)

Veteran*Year 2011*Disability

0.0656***

0.1076*

0.1774**

-0.0828

-0.0997

 

(0.0155)

(0.0456)

(0.0539)

(0.1765)

(0.1018)

Veteran*Year 2012*Disability

0.0534***

0.0527

0.1610**

-0.2085

0.0496

 

(0.0127)

(0.0392)

(0.0468)

(0.1497)

(0.0765)

Veteran*Year 2013*Disability

0.0462**

0.0762

0.0967*

0.0697

0.0157

 

(0.0130)

(0.0390)

(0.0368)

(0.1338)

(0.0715)

Veteran*Year 2014*Disability

0.0519***

0.0461

0.0814

0.0875

0.0360

 

(0.0121)

(0.0382)

(0.0441)

(0.1023)

(0.0862)

Veteran*Year 2015*Disability

0.0558***

0.0068

0.0749

0.0572

0.0545

 

(0.0120)

(0.0340)

(0.0389)

(0.1231)

(0.0642)

Veteran*Year 2016*Disability

0.0458***

0.0663*

0.0689

0.0049

-0.0171

 

(0.0123)

(0.0326)

(0.0371)

(0.1127)

(0.0730)

Veteran*Year 2009

-0.0051

-0.0280

-0.0063

-0.0168

-0.0244

 

(0.0114)

(0.0332)

(0.0448)

(0.0763)

(0.0688)

Veteran*Year 2010

0.0047

-0.0154

-0.0352

0.0435

0.0165

 

(0.0151)

(0.0336)

(0.0464)

(0.0976)

(0.0764)

Veteran*Year 2011

0.0069

-0.0005

0.0073

-0.0455

0.0561

 

(0.0135)

(0.0312)

(0.0416)

(0.0845)

(0.0783)

Veteran*Year 2012

0.0160

0.0160

-0.0249

0.0256

0.0264

 

(0.0105)

(0.0252)

(0.0368)

(0.0891)

(0.0626)

Veteran*Year 2013

0.0094

0.0210

-0.0127

-0.0448

0.0597

 

(0.0090)

(0.0185)

(0.0289)

(0.0672)

(0.0533)

Veteran*Year 2014

0.0115

0.0143

-0.0135

-0.0798

0.0524

 

(0.0071)

(0.0209)

(0.0298)

(0.0418)

(0.0439)

Veteran*Year 2015

-0.0033

0.0175

-0.0146

-0.0367

0.0161

 

(0.0074)

(0.0181)

(0.0249)

(0.0574)

(0.0375)

Veteran*Year 2016

-0.0020

0.0006

-0.0218

0.0111

0.0869*

 

(0.0097)

(0.0174)

(0.0264)

(0.0503)

(0.0404)

R2

0.236

0.134

0.182

0.375

0.177

N

8270766

1078188

770590

154678

279174

Note: Same as in Table 4.



APPENDIX D


EFFECTS OF POST-9/11 GI BILL ON COLLEGE ENROLLMENT BY LEVELS OF EDUCATIONAL ATTAINMENT


 

<1 year

college

>=1 year, no degree

Associate’s degree

Bachelor’s degree

Advanced degree

Veteran*After 2009*Disability

0.0409

0.0831**

0.0058

0.0566**

0.0658**

 

(0.0329)

(0.0276)

(0.0283)

(0.0201)

(0.0205)

Veteran*After 2009

0.0458*

-0.0012

-0.0048

-0.0097

0.0144

 

(0.0196)

(0.0117)

(0.0219)

(0.0119)

(0.0157)

R2

0.293

0.418

0.222

0.139

0.079

N

831313

1800885

941374

2067647

1049569

      

Veteran*Year 2009*Disability

0.0143

0.0788*

0.0052

0.0410

0.0401

 

(0.0492)

(0.0367)

(0.0390)

(0.0228)

(0.0425)

Veteran*Year 2010*Disability

0.0529

0.0960

-0.0109

0.0365

0.0360

 

(0.0418)

(0.0509)

(0.0435)

(0.0344)

(0.0372)

Veteran*Year 2011*Disability

-0.0190

0.1380***

0.0356

0.0605

0.0942**

 

(0.0458)

(0.0366)

(0.0366)

(0.0384)

(0.0331)

Veteran*Year 2012*Disability

0.0485

0.1087**

-0.0073

0.0440

0.0912**

 

(0.0428)

(0.0328)

(0.0363)

(0.0281)

(0.0301)

Veteran*Year 2013*Disability

0.0215

0.0746*

0.0095

0.0787**

0.0822**

 

(0.0353)

(0.0336)

(0.0411)

(0.0247)

(0.0296)

Veteran*Year 2014*Disability

0.0548

0.0982*

-0.0122

0.0488

0.0569

 

(0.0328)

(0.0363)

(0.0365)

(0.0293)

(0.0285)

Veteran*Year 2015*Disability

0.0587

0.0731*

0.0023

0.0586*

0.0681**

 

(0.0381)

(0.0307)

(0.0343)

(0.0262)

(0.0227)

Veteran*Year 2016*Disability

0.0346

0.0794**

0.0273

0.0444*

0.0517*

 

(0.0390)

(0.0274)

(0.0324)

(0.0206)

(0.0231)

Veteran*Year 2009

0.0473

-0.0420

-0.0227

-0.0439

0.0107

 

(0.0331)

(0.0219)

(0.0364)

(0.0237)

(0.0223)

Veteran*Year 2010

0.0547

-0.0423

0.0072

-0.0402

0.0146

 

(0.0321)

(0.0281)

(0.0397)

(0.0297)

(0.0252)

Veteran*Year 2011

0.0813**

-0.0331

0.0097

-0.0202

0.0289

 

(0.0277)

(0.0260)

(0.0367)

(0.0283)

(0.0225)

Veteran*Year 2012

0.0515

-0.0003

0.0209

-0.0137

0.0010

 

(0.0299)

(0.0219)

(0.0339)

(0.0167)

(0.0197)

Veteran*Year 2013

0.0506*

0.0043

-0.0067

-0.0106

0.0117

 

(0.0235)

(0.0181)

(0.0263)

(0.0173)

(0.0188)

Veteran*Year 2014

0.0412

0.0028

0.0167

-0.0064

0.0337

 

(0.0219)

(0.0173)

(0.0202)

(0.0151)

(0.0183)

Veteran*Year 2015

0.0356

-0.0027

0.0025

-0.0220

0.0006

 

(0.0217)

(0.0139)

(0.0245)

(0.0142)

(0.0161)

Veteran*Year 2016

0.0509*

-0.0139

-0.0290

-0.0046

0.0225

 

(0.0236)

(0.0147)

(0.0271)

(0.0124)

(0.0196)

R2

0.291

0.420

0.222

0.139

0.079

N

935003

2022566

1052700

2320010

1175978

Note: Same as in Table 4.


APPENDIX E


EFFECTS OF POST-9/11 GI BILL ON COLLEGE ENROLLMENT BY DISABILITY RATINGS


 

0 percent

10–20 percent

30–40 percent

50–60 percent

70 percent & above

Veteran*After 2009*Disability

0.0319

0.0428*

0.0480**

0.0213

0.0522*

 

(0.0295)

(0.0171)

(0.0148)

(0.0238)

(0.0239)

Veteran*After 2009

0.0016

0.0027

0.0022

0.0025

0.0021

 

(0.0059)

(0.0059)

(0.0059)

(0.0060)

(0.0059)

R2

0.219

0.219

0.219

0.219

0.219

N

9349092

9360834

9358239

9355470

9360612

      

Veteran*Year 2009*Disability

0.0129

0.0378

0.0697**

-0.0200

0.0135

 

(0.0599)

(0.0224)

(0.0235)

(0.0382)

(0.0327)

Veteran*Year 2010*Disability

0.0061

0.0468*

0.0680*

-0.0006

0.0281

 

(0.0488)

(0.0230)

(0.0311)

(0.0345)

(0.0337)

Veteran*Year 2011*Disability

0.0778

0.0718*

0.0931***

0.0461

0.0223

 

(0.0388)

(0.0308)

(0.0242)

(0.0308)

(0.0303)

Veteran*Year 2012*Disability

0.0979*

0.0361*

0.0939***

-0.0005

0.0519

 

(0.0482)

(0.0159)

(0.0206)

(0.0320)

(0.0306)

Veteran*Year 2013*Disability

0.0987**

0.0531*

0.0841***

0.0082

0.0172

 

(0.0323)

(0.0210)

(0.0202)

(0.0291)

(0.0271)

Veteran*Year 2014*Disability

0.0476

0.0569*

0.0402*

0.0380

0.0343

 

(0.0320)

(0.0229)

(0.0196)

(0.0309)

(0.0266)

Veteran*Year 2015*Disability

0.0009

0.0380*

0.0513*

0.0190

0.0583*

 

(0.0405)

(0.0187)

(0.0194)

(0.0258)

(0.0263)

Veteran*Year 2016*Disability

0.0004

0.0379

0.0254

0.0215

0.0613*

 

(0.0415)

(0.0205)

(0.0156)

(0.0261)

(0.0251)

Veteran*Year 2009

-0.0034

-0.0036

-0.0011

-0.0058

-0.0113

 

(0.0087)

(0.0091)

(0.0087)

(0.0088)

(0.0092)

Veteran*Year 2010

0.0064

0.0065

0.0093

0.0036

-0.0033

 

(0.0115)

(0.0117)

(0.0122)

(0.0113)

(0.0117)

Veteran*Year 2011

0.0127

0.0131

0.0153

0.0105

0.0047

 

(0.0110)

(0.0116)

(0.0117)

(0.0107)

(0.0111)

Veteran*Year 2012

0.0164*

0.0167*

0.0183*

0.0147

0.0103

 

(0.0074)

(0.0078)

(0.0078)

(0.0074)

(0.0076)

Veteran*Year 2013

0.0121

0.0128

0.0140

0.0112

0.0076

 

(0.0071)

(0.0073)

(0.0073)

(0.0068)

(0.0074)

Veteran*Year 2014

0.0096

0.0103

0.0110

0.0095

0.0075

 

(0.0059)

(0.0059)

(0.0060)

(0.0059)

(0.0060)

Veteran*Year 2015

-0.0024

-0.0015

-0.0013

-0.0021

-0.0028

 

(0.0061)

(0.0062)

(0.0061)

(0.0061)

(0.0062)

Veteran*Year 2016

-0.0034

-0.0023

-0.0025

-0.0026

-0.0029

 

(0.0087)

(0.0087)

(0.0087)

(0.0087)

(0.0087)

R2

0.219

0.219

0.219

0.219

0.219

N

10496757

10509645

10506683

10503550

10508711

Note: Same as in Table 4.


APPENDIX F


EFFECTS OF POST-9/11 GI BILL ON COLLEGE ENROLLMENT BY ACS-DEFINED DISABILITY AND SERVICE-CONNECTED DISABILITY


 

ACS-defined Disability: No

 

ACS-defined Disability: Yes

 

All

Men

Women

 

All

Men

Women

Veteran*After 2009*Disability

0.0563***

0.0550***

0.0629*

 

0.0332

0.0319

0.0600

 

(0.0112)

(0.0109)

(0.0267)

 

(0.0215)

(0.0236)

(0.0465)

Veteran*After 2009

0.0040

0.0053

-0.0155

 

0.0070

0.0083

-0.0064

 

(0.0061)

(0.0065)

(0.0114)

 

(0.0062)

(0.0067)

(0.0110)

R2

0.219

0.223

0.221

 

0.219

0.224

0.221

N

9388431

4312983

5075448

 

9359500

4289419

5070081

        

Veteran*Year 2009*Disability

0.0375*

0.0374*

0.0353

 

0.0363

0.0388

0.0317

 

(0.0182)

(0.0185)

(0.0355)

 

(0.0292)

(0.0309)

(0.0703)

Veteran*Year 2010*Disability

0.0513**

0.0474*

0.0804*

 

0.0382

0.0251

0.1313

 

(0.0170)

(0.0186)

(0.0353)

 

(0.0293)

(0.0335)

(0.0672)

Veteran*Year 2011*Disability

0.0826***

0.0803***

0.1022*

 

0.0445

0.0584*

-0.0125

 

(0.0187)

(0.0205)

(0.0426)

 

(0.0262)

(0.0279)

(0.0667)

Veteran*Year 2012*Disability

0.0627***

0.0545**

0.1101***

 

0.0480*

0.0480*

0.0638

 

(0.0131)

(0.0161)

(0.0285)

 

(0.0197)

(0.0216)

(0.0706)

Veteran*Year 2013*Disability

0.0670***

0.0664***

0.0696*

 

0.0151

0.0159

0.0242

 

(0.0137)

(0.0147)

(0.0287)

 

(0.0261)

(0.0292)

(0.0477)

Veteran*Year 2014*Disability

0.0610***

0.0630***

0.0583

 

0.0280

0.0290

0.0467

 

(0.0138)

(0.0134)

(0.0304)

 

(0.0222)

(0.0244)

(0.0591)

Veteran*Year 2015*Disability

0.0481**

0.0455**

0.0580

 

0.0458

0.0407

0.0868

 

(0.0139)

(0.0135)

(0.0370)

 

(0.0241)

(0.0262)

(0.0531)

Veteran*Year 2016*Disability

0.0556***

0.0544***

0.0645*

 

0.0251

0.0201

0.0711

 

(0.0129)

(0.0123)

(0.0302)

 

(0.0251)

(0.0257)

(0.0540)

Veteran*Year 2009

-0.0088

-0.0056

-0.0235

 

-0.0042

-0.0013

-0.0126

 

(0.0094)

(0.0101)

(0.0218)

 

(0.0069)

(0.0072)

(0.0174)

Veteran*Year 2010

0.0001

0.0047

-0.0291

 

0.0055

0.0095

-0.0162

 

(0.0121)

(0.0116)

(0.0278)

 

(0.0090)

(0.0093)

(0.0136)

Veteran*Year 2011

0.0081

0.0061

0.0115

 

0.0120

0.0094

0.0215

 

(0.0121)

(0.0115)

(0.0245)

 

(0.0087)

(0.0095)

(0.0182)

Veteran*Year 2012

0.0131

0.0187*

-0.0205

 

0.0157*

0.0208**

-0.0132

 

(0.0077)

(0.0079)

(0.0207)

 

(0.0062)

(0.0073)

(0.0107)

Veteran*Year 2013

0.0106

0.0127

-0.0083

 

0.0118

0.0134

-0.0032

 

(0.0075)

(0.0079)

(0.0185)

 

(0.0062)

(0.0072)

(0.0145)

Veteran*Year 2014

0.0101

0.0082

0.0034

 

0.0095

0.0074

0.0059

 

(0.0061)

(0.0066)

(0.0137)

 

(0.0059)

(0.0064)

(0.0149)

Veteran*Year 2015

-0.0004

0.0016

-0.0218

 

-0.0024

-0.0006

-0.0213

 

(0.0064)

(0.0068)

(0.0134)

 

(0.0061)

(0.0064)

(0.0137)

Veteran*Year 2016

-0.0002

0.0017

-0.0257

 

-0.0033

-0.0016

-0.0264

 

(0.0089)

(0.0088)

(0.0166)

 

(0.0087)

(0.0086)

(0.0166)

R2

0.219

0.224

0.221

 

0.219

0.225

0.221

N

10539421

4827067

5712354

 

10507755

4801298

5706457

Note: Same as in Table 4.






Cite This Article as: Teachers College Record Volume 122 Number 3, 2020, p. 1-44
https://www.tcrecord.org ID Number: 23012, Date Accessed: 10/21/2021 7:31:38 PM

Purchase Reprint Rights for this article or review
 
Article Tools
Related Articles

Related Discussion
 
Post a Comment | Read All

About the Author
  • Liang Zhang
    NYU Steinhardt
    E-mail Author
    LIANG ZHANG is a professor of higher education at NYU Steinhardt School of Culture, Education, and Human Development. His research focuses on higher education economics, finance, and public policy, particularly on the role of governments and institutions in affecting institutional performances and student outcomes.
 
Member Center
In Print
This Month's Issue

Submit
EMAIL

Twitter

RSS