AMERICA’S DROPOUT PROBLEM
Chapter 2
Table of Contents
Theoretical Functions of the GED 4
HUMAN CAPITAL HYPOTHESES 4
STUDYING FOR TEST INCREASES SKILL 4
Life Experience 6
Signaling Hypotheses 7
COGNITIVE SKILL SIGNAL 7
Multidimensional Signaling 8
Previous Research 12
DELAYED IMPACT 15
HETEROGENEOUS RETURNS 16
Low Skilled Dropouts 17
PRISONERS 19
IMMIGRANTS 21
OLDER COHORTS 22
LABOR MARKET OUTCOMES 23
DIRECT EFFECTS OF THE GED 25
THE CURRENT POPULATION SURVEY (CPS) 26
THE NATIONAL LONGITUDINAL SURVEY OF YOUTH (NLSY) 27
MEAN OUTCOMES 28
MALES 28
Females 29
Controlled Analysis 31
EMPLOYMENT AND LABOR SUPPLY 31
MALES 32
Females 37
Hourly Wages 40
MALES 40
Females 41
Annual Earnings 42
MALES 43
Females 43
Conclusion 44
REFERENCES 46
CHAPTER
2
GED Labor Market Outcomes
James J. Heckman and Paul LaFontaine
G
ED certification is now the major social program offered to dropouts looking to lift themselves out of poverty and improve their lives. Many major government programs such as job training, welfare to work, and prison educational programs feature GED acquisition as a primary goal. Students identified by school districts as “at risk” for dropping out are often pushed into “alternative” high schools or dropout prevention programs aimed at preparing them to take the GED test. The GED is a massive program due, in no small part, to official endorsement by government and local school districts. We have shown that fifty percent of all young high school dropouts now GED certify and that this accounts for 15-20% of all high school credentials issued in the U.S. each year. Therefore, the question of what benefits a GED certificate provides these young dropouts is of critical social importance.
Remarkably, despite the magnitude of the program, the economic and social consequences of GED certification were not seriously investigated until the early 1990s. Prior to that, the GED Testing Services’ (GEDTS) claim that a GED was equivalent to a conventional high school diploma went largely unchallenged and untested. GEDTS argued, and still does, that those who obtain the GED demonstrate that they are the academic equivalents of high school graduates and should be considered as such for “admissions, military, licensing, and employment purposes”.[1] Based on these claims, government agencies counted GEDs as if they had actually graduated from high school. In addition, most two and four year colleges accepted the GED to satisfy admission requirements.
Initial research into the outcomes of GEDs focused primarily on comparing them with high school graduates. We now know that the two degrees are by no means equivalent. Using a nationally representative sample of young adults, it was shown that traditional high school graduates outperform GEDs in measured test scores, hourly wages, employment, labor supply, annual earnings, the ability to hold down a job and completion of post-secondary education programs.[2] In short, GEDs are not as highly valued by the labor market as high school graduates and are not academically equivalent. No scholars dispute the nonequivalence of the two degrees and a large body of subsequent research supports this position.[3]
Current scholarship focuses on discovering what benefits, if any, GED acquisition offers to dropouts in search of better opportunities. This question, as well as the more general issue of the effectiveness of late intervention social programs in improving the outcomes of low skilled workers, is now of greater importance than ever. As the introduction demonstrated, America’s low skilled workers are losing ground every year in an increasingly technology oriented economy. Finding an effective means of raising the living standards of low skilled workers is now of critical sociologic and economic importance.
Theoretical Functions of the GED
GED acquisition could potentially affect labor market outcomes through many channels. One possibility is that the GED may directly increase the human capital of low skilled workers (i.e. the skills valued by employers) through time spent studying for the exam. Second, the GED may act as a signal to employers of the skills they desire and therefore of greater productivity. And finally, GED acquisition provides access to post-secondary schooling and training opportunities. This chapter tests the first two possibilities and leaves the indirect benefits of certification (e.g. post-secondary education) until the next chapter.
Human Capital Hypotheses
Studying for Test Increases Skill
One reason why GED acquisition may be valuable to those who obtain it is that while studying for the exam, either through participation in GED preparatory classes or through self-study, a dropout may significantly increase the skills desired by employers. A standard human capital model predicts that individuals invest in education when the return to the skills acquired outweighs the cost of acquiring them.[4] A large body of empirical research shows that schooling and classroom training have strong positive effects on earnings and employment, and that individuals do in fact respond to these incentives. However, GED certification is very different from a traditional schooling program that requires substantial investments in skills. Obtaining a GED requires no formal schooling and dropouts are able to take the exam with little to no preparation. The amount of time GED candidates spend preparing for the exam is a strong indicator of what the return may be. If the GED requires significant investment in terms of skill acquisition, then we would expect a substantial direct payoff to this investment. Alternatively, if the GED requires little to no investment, then the returns should be limited.
The GED does not appear to require much investment in new skills for the average applicant to pass. Two surveys conducted by the GED Testing Service in 1980 and 1989 asked examinees exiting GED testing centers to report how much time and money they spent preparing for the test. In 1980, GED examinees spent a median twenty hours and only $10 preparing for the test. By 1989, the median study time increased to thirty hours. Twenty percent did not prepare in any way. Seventy-five percent of examinees spent less than 100 hours in preparation, which is equal to about one month of high school. Those who pass the GED have completed 10 years of schooling on average. Essentially, examinees tried to make up two years of high school through one month of study, a tall order.
Despite the generally low levels of preparation, 70% of those taking the test pass on any given attempt. Test takers who fail one or more sections may retake sections of the exam until all sections are passed without penalty. Test preparation programs subsidized by federal and state governments tend to focus on test-taking strategies and items specific to the GED exam.[5] Test scores on the exam can be substantially improved through short-term coaching sessions that teach test taking skills rather than any real academic content.[6] This demonstrates that most GED examinees need little, if any, investment in new skills or knowledge to pass. Since the investment required to GED certification is low, we would expect the economic returns to be low as well—in common parlance, “no pain, no gain.”
Even though most applicants do not need substantial study to pass the GED exam, one line of research suggests that certification has a strong positive effect on wages for those who leave high school with very low skills.[7] Those who dropped out with higher academic skills were found to receive no benefits. Even though the median study time was 30 hours in 1990, 25% of GED recipients did study for 100 hours or more. For instance, low skilled immigrants may require substantial investments in order to pass the exam. Many foreign GED candidates invest hundreds of hours in basic skills training and language acquisition and this may result in substantial returns. We pay particular attention to these issues when estimating returns to GED certification by analyzing the returns for these groups separately.
Life Experience
Another commonly stated reason why the GED should be valuable is that it allows high school dropouts to certify lessons they have learned through “life experience”. The implication is that cognitive skills are acquired with age through everyday experiences. This is a throwback to the early days of GED testing when GED candidates bore the benefits of extensive military training and had potentially gained many skills outside the classroom. However, the vast majority of today’s GED candidates have little new life experience since they finished schooling, since they are typically under the age of 25. The modal age of GED certification is only 18 years old. Even for those dropouts with life experience, there is little to no documented evidence that one can raise cognitive skills through everyday activities such as reading newspapers or following the directions of a recipe. On the contrary, research shows that cognitive skills tend to decline with age and with time since and individual completes schooling.[8]
Signaling Hypotheses
Cognitive Skill Signal
The human capital model assumes that employers are able to easily observe differences in skill levels between prospective job candidates. Employers use a variety of mechanisms to differentiate potential employees including references, probationary periods and interviews. If this information is costly or difficult for the firm to obtain, then it may rely on other less costly observable measures of applicant quality such educational attainment or exam certification. Instead of leading to the acquisition of new skills, GED certification may simply signal the presence of individual characteristics already present prior to the acquisition of the GED. These skills may be difficult or costly for employers to observe without certification by the GED. The educational signaling hypothesis (i.e. education is valuable by signaling traits desired by employers) has been the subject of major theoretical and empirical treatments in the economic literature.[9] Schools may not improve the skills of students but may simply identify those with desirable characteristics. Under this view, the output of schools is information about the relative abilities of students. Applying the signaling argument to the present context, the GED would have value simply as a harbinger of traits desired by employers that not all high school dropouts possess.
At first glance, this argument has merit. We demonstrated in Chapter 2 that those dropouts who are able to pass the GED exam have higher cognitive skills than other high school dropouts. This observable signal that a worker possesses high ability is valued by both employers and workers since it increases productivity, solving an asymmetric information problem. Workers are paid on the value of their marginal product of labor and the GED could act as a sign of productivity, identifying workers that are more productive and therefore should receive a higher wage. A signal is effective only if it efficiently separates workers on the basis of skills valued by employers. However, if the skill level certified by a GED is observable through other channels before certification, then GED acquisition may not be associated with economic gain.
Multidimensional Signaling
Traditionally, the economic literature on signaling assumed that educational credentials provided a one dimensional measure of ability, commonly viewed as cognitive skill.[10] An alternative interpretation is that education is multi-dimensional and may signal a multitude of abilities that are valued by employers. A number of surveys find that employers in low skilled labor markets value docility, dependability and persistence over cognitive ability or independent thought.[11]
Although current education policy focuses primarily on raising measured achievement test scores, schools have long been considered important for habituating students to societal norms and proper modes of behavior. Many scholars argue that a primary purpose of schooling is to prepare students for the workplace by teaching them appropriate work habits, attitudes, and social skills. They find that the list of important traits schooling fosters includes obeying authority figures, punctuality, cooperation, ability to work in groups, orderliness and the ability to stay on task.[12] Economists and sociologists have produced a large body of evidence showing the role of these noncognitive skills in predicting wages, occupational choice and educational attainment.[13] They argue that the return to schooling primarily operates through learning how to adapt to working in highly structured and bureaucratic organizations, as is found in most companies. Education socializes students to the norms of society and those who most successfully internalize these norms reap the largest rewards in the labor market.
The lack of a socialization component in the process of attaining a GED implies that noncognitive skills are not affected or measured by GED acquisition. The GED, much like today’s educational policy, is concerned only with the cognitive skills of dropouts. It completely ignores important noncognitive behavioral traits. If these are the traits most valued in low skilled labor markets, then we would expect GEDs to have the same labor market outcomes as uncertified dropouts. The act of dropping out may signal that an individual has difficulty adapting to the expected norms of an institutional environment. Although taking and passing the GED exam signals higher cognitive skills, it does not demonstrate successful internalization of behavioral traits that many employers desire.
Using the idea of education as a multidimensional signal, a richer story emerges regarding the relative merits of GED certification and high school graduation. Suppose there are two categories of traits signaled by education: cognitive and noncognitive skills. Both types of skills are valued by employers because they are crucial to the production process, and an employer’s ideal worker possesses both types of skill. High school graduation would likely convey high levels of both skills, whereas dropping out would signal low endowments of one or both types of skill.
Evidence on who takes the GED, presented in Chapter 4, suggests that GED certifiers send the troubling message that they are intelligent but have lower noncognitive skills than both high school graduates and uncertified dropouts. We label this a “mixed signal” since the positive cognitive traits are countered by a negative signal on noncognitive skills.[14] In terms of labor market outcomes, this model would allow the GED to have no effect on earnings or employment despite higher cognitive skills. Since both cognitive and noncognitive skills enter the production process, passing the exam does not necessarily lead to higher productivity expectations. Under a multi-dimensional signaling model, we would expect high school graduates to have better wage and employment outcomes than GEDs even after adjusting for higher endowments of cognitive skill.
Further evidence that supports the multi-dimensional signaling story is provided by the performance of GED recipients in the military. Attrition from basic training among new enlistees is extremely costly for the Armed Forces. It is estimated that for each enlistee who does not complete basic training the military loses $29,000.[15] High school dropouts and GED-certified high school equivalents had basically the same attrition rates from the U.S. military over the period 1977-79. Both groups exited at twice the rate of high school graduates. In response to growing concerns over the significantly higher attrition rates among GED holders, the military in 1982 raised the minimum AFQT score requirement for GED recipients above the score required of traditional high school graduates. In a study of military attrition, a high school diploma was the single best predictor of whether an enlistee completed basic training. Holding degree status constant, AFQT scores had almost no predictive power. This suggests that those holding a high school diploma possess productive and valued characteristics that are not captured by simple cognitive test scores. Evidence of the importance of noncognitive skills in the military is summarized in Chapter 6.
Previous Research
The hypotheses stated above provide a basis for why we may suspect a causal link between GED acquisition and better labor market outcomes. Unfortunately, researchers immediately encounter a number of difficult statistical problems when trying to separate fact from fiction in order to empirically test this proposition, and research into the labor market outcomes of GED recipients often appears to give conflicting answers. These contradictory results are primarily due to two issues of first order empirical importance that arise when trying to evaluate the effects of social programs.
The first distinction that needs to be made is direct versus indirect effects of GED acquisition. Direct effects are those benefits that are conferred to program participants from the GED title itself, while indirect effects are those resulting from further schooling or training obtained after certification. Many studies claiming to find positive effects of GED certification on wages and other labor market outcomes do not distinguish between direct returns from certification versus the indirect effects that result from subsequent education. While there is no disagreement among researchers that attending college has substantial economic rewards, there is considerable debate on whether the GED title in and of itself is of value. The policy implications of these two separate questions are quite different. If the GED has direct value then it may be socially valuable to subsidize GED courses and promote GED certification through Job Corps and other government programs. If the only value of the GED is in obtaining access to and completing post-secondary schooling, then resources may be better allocated towards programs that directly promote post-secondary schooling for dropouts. We are careful to separate out these two types of effects in our analysis. This chapter dedicated solely to the study of direct benefits.
The second obstacle in determining potential GED effects is self-selection. As with the evaluation of many social programs, there has never been a controlled randomized experiment examining the benefits of GED certification. Instead, we are forced to evaluate data where people have self-selected non-randomly into treatment and control groups on the basis of some unobserved characteristics (e.g. intelligence, motivation, family income etc.). Generally, more able students complete more schooling or pass certification tests because it is easier for them to do so. In this case, any observed differences in earnings reflect differences in pre-existing ability and not the effects of more schooling or training.
Claims that increased schooling “causes” increased earnings, and improved employment prospects enhance socioeconomic performance, may simply reflect the correlation between high levels of achievement, enhanced schooling, and exam proficiency for the more able. Even after controlling for all observable characteristics, GEDs may have better outcomes due to some unknown characteristics or process that results in both GED certification and better employment prospects. In comparing the outcomes of GED recipients with those of uncertified dropouts, we need to separate out those factors that are attributable to the GED program from those that are due to the self-selection process.
The prediction of “Dewey Defeats Truman" in the 1948 presidential election by the Chicago Daily Tribune is one of the most famous cases of selection bias. Prior to the election an interview was conducted by telephone to determine who would win the election. Interviewees were polled by phone during the day and thus the sample was confined to only those households with a telephone and someone home during the day to answer it. This non-random sample of voters predicted that Dewey would win by a wide margin. Dewey received quite a shock on Election Day when he lost to Harry Truman after expecting to win in a landslide. Analogously, GED recipients are not a random sample of dropouts since they self-select into the program. Specifically, they self-select primarily on the basis of having higher cognitive ability (making the test easier to pass) and lower noncognitive ability. Since higher levels of cognitive ability are also associated with higher wages, what appears to be a positive impact of the GED may be a spurious result since it is generated by the selection process and not by the program itself.
Many articles in the press and proponents of the GED program (and even professional economists) point to more GED recipients in higher paying jobs with higher annual earnings, or more GEDs going to college, to show that the GED has a positive effect on dropout outcomes. They ignore the selection process and claim that these outcomes resulted from acquiring a GED and not from unobservable characteristics of those who chose to GED certify that would have caused them to have the same outcomes regardless of whether they GED certified. We are careful in our analysis to sort out differences based on selection versus direct effects caused by the GED program itself.
Following the publication of Cameron and Heckman’s (1993) original analysis of the outcomes of GED recipients, a number of studies have been published that attempt to assess whether the GED program offers any return to high school dropouts. A number of potentially valid criticisms of the original analysis have been voiced. The new research agrees that a GED is not the equivalent of a high school graduate and instead focuses on the dropout versus GED comparison. The latest findings of this literature can be grouped into two principle categories.
Delayed Impact
Cameron and Heckman’s work only examined the outcomes of GED recipients at ages 25 and 28 since these were the oldest available ages at the time their paper was written. Some advocates of the GED program argue that gains to GED certification may not manifest themselves until later ages. If GED certification opens up access to occupations that are closed to high school dropouts, then the wage trajectories of GED recipients may be steeper five to ten years after certification.
The evidence related to whether the GED leads to higher wage growth has been slightly mixed, but overall the latest evidence suggests that growth is not attained. An initial study by Murnane, Willett, and Boudett (1995)[16] using NLSY data (the same data used for the Cameron and Heckman study), found that the wages of GED recipients after certification increased at a rate 2.4% greater than uncertified dropouts. Similarly, annual hours worked increased at a 3.3% higher rate and annual earnings at a 5.6% higher rate. These results looked promising for GED proponents; however, a major problem prevents us from interpreting this evidence as a causal impact of the GED. The study did not control for unobserved abilities. Specifically, they did not control for the higher ability of GED recipients observed before they GED certify. They attributed the higher growth rates in GED wages and labor supply to the title, when it was likely due instead to higher ability dropouts selecting to take the GED.
The same authors later corrected this problem and re-estimated similar models using the same NLSY data.[17] Using the longitudinal structure of the NLSY to control for unobserved heterogeneity and analyze long range impacts, they find that both the direct and indirect effects of GED certification were quite low. Using methods that eliminate unobserved differences, the increase in hourly wage five years after completing the GED is only four percent and statistically insignificant for low skilled individuals and zero for those who left school with higher skills. The models of earnings tell a similar story. The authors estimate that the predicted difference between the annual earnings of GEDs five years after obtaining the title is only $288 and not statistically significant. Including the effects of further college and training to estimate the total effect does not help greatly. The estimates were 4.7% higher hourly wages and $488 dollars more in income per year. Both estimates are not statistically significant.
Heterogeneous Returns
The second issue explored by the latest GED research is whether there is a disparate impact of the GED program across different race groups or other subpopulations. Cameron and Heckman’s analysis constrained the estimated impact of GED certification to be the same across all groups. An important consideration is whether the GED has a small or zero impact on the outcomes of the majority of dropouts but a more substantial impact for certain groups of dropouts (e.g. immigrants or prisoners). Estimates from specifications that constrain the impact to be the same for all groups could be small and insignificant, while estimates that allow the impact to differ by group could uncover statistically significant impacts for subpopulations. We examine whether the GED has a greater impact for the following four groups of dropouts.
Low Skilled Dropouts
Some claim that the GED has a greater impact on the outcomes of those who left school with weak skills compared to those who left school with relatively high skills.[18] There are a number of potential mechanisms through which this sort of differential impact could arise. First, those with strong skills may quit school to pursue potential employment opportunities. For these people, the GED would hold little value since they already have strong employment histories and commitment to the labor market. Those who drop out with weak skills and poor work habits have few employment prospects, and as a result, obtaining a GED may send a strong signal that they have experienced a change in attitude and are now committed to the labor market. A second potential mechanism is that very low skilled dropouts might accumulate a substantial amount of human capital while studying to pass the GED. We saw that the median study time for the GED is quite small on average. For those who possess very weak academic skills, this investment may be much more substantial.
Tyler, Murnane, and Willett (2000)[19] utilize a quasi-experimental research design that exploits the fact that in 1990 states had differential GED passing standards. This interstate variation in GED passing standards allowed them to compare the earnings of dropouts who passed the GED in states having relatively low standards with the earnings of dropouts who took the test and received the same scores but did not pass because they lived in states with higher standards. This approach, sometimes called a “natural experiment”, allowed the researchers to compare the earnings of dropouts with similar motivation and cognitive skills, some of whom received their GED and some of whom did not by virtue of the states where they lived. They estimated the impact of the GED on earnings to be 10 to 19 percent for white GED holders, substantially larger than previous estimates in the literature.
They reconciled these new higher estimates with previous results by arguing that as a result of their research design, their estimates were driven by GED holders who barely passed the GED exams: GED holders with the lowest cognitive skill levels. In contrast, previous authors had adopted an implicit assumption that the impact of the GED on earnings was independent of skill level. However, Rubinstein (2006)[20] presents a critique that shows that these results are spurious due to a weakness in methodology. Essentially, the positive effects found by Tyler, Murnance and Willett, result from differences in the labor markets of low- and high-pass score states.[21]
Further confirmation of this error in methodology is provided by one of the authors’ later papers. Tyler and Lofstrom (2004)[22] once again use the natural experiment methodology to study the signaling value of a GED, but this time they restrict the sample to only those who GED certify in one state, Texas. In 1997, the state of Texas was forced to raise GED passing standards by the GED testing service. The authors compared Texas dropouts who acquired a GED before the passing standard was raised in 1997 to dropouts with the same test scores who failed the GED exams after the standard was raised. Under this design, the Rubinstein methodological critique is no longer applicable, since both GED recipients and non-recipients are compared in the same labor market. Not surprisingly, Tyler and Lofstrom found no evidence of a positive GED effect on earnings. This result is robust to a wide range of alternative specifications and across a wide range of sub-samples. They conclude by arguing that “our results are the most definitive to date regarding the value of the GED credential as a labor market signal for male dropouts.” We also find no evidence in this chapter that returns are higher for low skilled dropouts.
Prisoners
The impact of obtaining a GED for those serving time in prison may be quite different then for those who certify “on the outside”. Using data on criminal justice offenders who were incarcerated in Florida state prisons between 1994 and 2002, Tyler and Kling (2001)[23] estimate the effects of obtaining a GED credential while incarcerated on post-release quarterly earnings, probability of employment, and probability of recidivating. They find that minority offenders who obtain a GED while in prison have slightly higher earnings employment probabilities the first after three years after release than minority dropouts who do not obtain a GED while in prison. After three years the effect vanishes. They find no evidence that a GED improves the labor market outcomes of white offenders lacking a high school diploma. They also find that a GED does not improve the recidivism rate of any group.
Generally, their findings indicate that the value of a prison GED is quite low and that previous estimates in the literature overstated this value because they lacked sufficient controls for important observable differences between offenders who do and do not leave prison with a GED. The authors further state they are wary to make causal inferences from these weak effects since there is evidence that non-white offenders who obtain a prison GED had somewhat better labor market experiences before they entered prison than non-white offenders who do not obtain a GED.
We present our findings on the impact of a GED credential for prisoners in Chapter 7. Our findings are consistent with those presented in the Kling study. There is only a small advantage in earnings and employment in the three years immediately following release and this fades out after this time. We explain this anomaly by noting that this is the same amount of time that GEDs are on probation. It appears that those who GED certify are either behaving better while they are being monitored or receiving preferential treatment from their parole officers. Using data that tracks criminal offenders through the different stages of the criminal justice system, we find—like Tyler and Kling—that obtaining a GED while imprisoned has no impact on reducing recidivism rates.
Immigrants
The GED was developed for the U.S. population and may play a particularly important role for immigrants. The literature suggests that the returns to credentials acquired in a migrant’s destination country are generally higher than those obtained in their country of origin.[24] The GED may induce the foreign-born to acquire U.S. specific skills and/or signal to employers that an immigrant possesses greater motivation as well as higher academic and language skills than uncertified immigrants.
It is also possible that migrants schooled outside the U.S. spend significant amounts of time preparing for the exam (for example, learning English). Additionally, while school quality is relatively similar across the United States, it varies considerably from country to country, and a potential employer may be less familiar with the quality of education in specific countries. Finally, foreign born GED recipients may have considerably greater noncognitive skills than their native born counterparts. For all of these reasons, a foreign born GED could have greater value. It may signal to employers that a potential employee schooled abroad does have a certain level of U.S.-relevant skill and may fare better in the labor market than those who received a high school diploma outside the U.S.
Clarke and Jaeger (2006) use the CPS to estimate the differential return for foreign born and native GEDs. They find the relative returns to be larger for the foreign-born than for natives. Immigrants who possess a GED earn more than observationally equivalent migrants who received a traditional high school diploma outside of the U.S. We show in the next chapter that the Clarke and Jaeger study suffers from a number of important biases. The differences in cognitive skill distributions between foreign born dropouts and GEDs are even greater than we observed for natives. This tells us that the GED program is even more selective among this population. After controlling for test score differences between the two groups, we find no evidence that the GED provides a signal of greater productivity for foreign born dropouts.
Older Cohorts
The NLSY data used by Cameron and Heckman was for individuals born between 1957 and 1964. There is no reason to believe that the returns to GED certification have remained constant across birth and certification cohorts. Previous studies have not estimated the impact of the GED over a wide range of birth cohorts. All previous research on the effects of the GED used data sets that contained a narrow band of ages and years. There are many reasons why we might expect returns to vary.
In the next chapter, using the Panel Survey of Income Dynamics (PSID), we look at the effects of the GED on all ages over a substantial period of time (1968 to 2000). Results indicate that males who obtained an early GED do receive higher earnings relative to other high school dropouts of the same cohort. The CPS data shows a similar pattern with older GED workers receiving significantly higher returns than their younger GED counterparts.
Our finding that the GED return for older workers is higher than that for younger workers may be explained by several different hypotheses: (a) the return to unskilled young workers has declined more than the return to unskilled old workers; (b) older workers with GEDs come from better quality cohorts (the “GED has become corrupted” hypothesis); (c) effects of GED certification take time to appear; and (d) the ability bias between GED recipients and uncertified dropouts is diminishing. We find in the next chapter that the GED ability bias between GEDs and uncertified dropouts was greater for older cohorts and this explains the higher returns, uncorrected for ability, for older workers.
Labor Market Outcomes
Substantial anecdotal evidence and survey data suggest that having a GED diploma should give an advantage to those seeking employment. One obvious employment advantage of the GED may come from the public sector. Some federal, state and local agencies require that GED recipients be treated the same as high school graduates in terms of hiring decisions. Jobs often require and ask applicants if they have completed a high school diploma or GED. Today in many states, such as North Carolina and California, a GED is the legal equivalent of a high school diploma when applying for state and local government positions as well as for admission to and funding of public post-secondary education.
In the private sector, employer surveys across a number of decades and regions find that most employers accept a GED as a substitute for a traditional high school diploma. The GED testing service claims that 95% of U.S. employers consider a GED credential to be the same as high school graduation with respect to hiring, advancement and wages.[25] A number of independent studies find lower acceptance rates, but they generally corroborate the claim that a GED enjoys broad employer acceptance. For instance, a survey of Wisconsin employers found that nearly 80% accepted the GED certificate as an acceptable alternative to a high school diploma in hiring decisions.[26] However, some apprenticeship programs in the construction trades would only accept a GED if the candidate could also demonstrate at least a 10th grade reading level. Other studies from states across the nation and at different time periods suggest that broad employer acceptance of GED credentials is not unique to Wisconsin (Table 1). The mean acceptance rate across all surveys was 80%.
These studies of employer acceptance are limited in their usefulness and suffer from a number of potential biases. First, many of these studies have very low response rates, suggesting that non-response bias may be a serious issue. Second, the surveys do not determine if GED recipients are hired into the same positions, at the same wages, experience similar wage growth and promotions, or are able to keep their positions.
It is also not necessarily true that employer impressions are accurate. There is a widespread belief in our society that a GED is the equivalent of a high school diploma. The ACE has campaigned tirelessly to promote this view and it has had a marked effect on both public and private institutions. Employers, responding to a questionnaire may automatically respond that the two are equivalent and may even act on this belief. But later after learning the traits of an individual candidate, they might reassess their position. As we will see in Chapter 6, the Armed Forces initially accepted the GED as equivalent until they performed a systematic study and determined that equivalency was wasting resources through high attrition. In the civilian labor market, the same type of misconception may still persist due to the rhetoric of the ACE as opposed to the historic reality of GED performance on the job.
Direct Effects of the GED
We now explore the direct effect the GED has on labor market outcomes. In this chapter we present a detailed snapshot of the labor market outcomes of dropouts, GED recipients and high school graduates at age 30 using the NLSY data. This age was chosen because most education has been completed by this age and recent research suggests that significant benefits do not manifest themselves until at least five years after certification.[27] We exclude those who have been incarcerated[28] or who have completed one year of college or more by age 30. It should be noted that a far greater percentage of high school graduates completed some college than both GEDs and permanent dropouts; therefore the high school graduates examined in this section are essentially the worst of the overall high school graduate pool. Thus we are comparing the outcomes of uncertified dropouts and GEDs to low ability high school graduates. The full economic returns to both GED certification and high school graduation, including both direct and indirect effects, are examined in Chapter 5. Evidence on the returns to GED certification across the life cycle and for older cohorts is delayed until the next chapter.
Before we begin the analysis the reader should be acquainted with the major data sources used in this first section. The following is a brief summary of the two major datasets. Further exposition about each data source is provided in the technical appendix.
The Current Population Survey (CPS)
The Current Population Survey (CPS) is a monthly survey of approximately 55,000 American households collected by the Census bureau. Due to its large sample size, the long period over which it has been collected, and its perceived quality, the CPS has become the primary data source for understanding a host of important economic issues, including the U.S. earnings structure, racial wage gaps and returns to education. It is typically used to compute national high school graduation rates and other major social statistics. In 1998, the CPS Monthly and March surveys began distinguishing between GED recipients and traditional high school graduates. The large sample sizes for various racial and ethnic groups, as well as the wide range of available ages, appear to make the CPS ideal for studying the effects of GED acquisition on labor market outcomes.
Unfortunately, three potentially serious problems and limitations plague the CPS data.[29] First, the CPS contains no measure of ability. As we showed in Chapter 8, those who choose to obtain a GED have much higher ability than uncertified dropouts. Therefore, returns to a GED over an uncertified dropout may be due to unmeasured ability differences and not a result of obtaining the GED. Second, when the wage of a GED worker is missing, the CPS frequently assigns them the wage of a traditional high school graduate. Since high school graduates earn significantly higher wages than GEDs, the result is a strong upward bias in the return to a GED. Third, CPS data show that a large fraction of workers have no reported earnings because they are unemployed or out of the labor force.
Despite these problems we use the CPS data as a foundation from which to begin our analysis of the GED program. After the above problems are addressed and estimation is performed carefully, the returns to GED certification and other educational estimates using CPS data are similar to those obtained from other, cleaner, data sources such as the NLSY.
The National Longitudinal Survey of Youth (NLSY)
The principal data set used in this chapter is the National Longitudinal Survey of Youth (NLSY). The NLSY includes both a randomly chosen sample of 6,111 U.S. youth and a supplemental sample of 5,295 randomly chosen Black, Hispanic, non-Black and non-Hispanic economically disadvantaged youths. Both of these samples are drawn from the non-institutionalized civilian population. All youths were age 13-20 in 1978 and were interviewed annually beginning in 1979. Today we have over 20 years of data on the outcomes of each subject (they are now in their late 30’s to early 40’s).
The NLSY has a number of advantages over CPS data for analyzing the effects of GED attainment. First, it contains a much richer variety of measures on family background, schooling history, work history, welfare history, marital and fertility choices, and geographic location in each year. Second, the NLSY collected detailed information on school attendance, behavior and outcomes, including data indicating a possession of a GED degree by year and month. Third, the data set includes both direct and indirect measures of both cognitive and noncognitive skills. Finally, attrition and non-response in this data are very low compared to CPS data.
Mean Outcomes
Males
At age 30, male GED recipients who do not complete any college look remarkably similar to high school dropouts in terms of their labor market characteristics. Table 2(a) shows GED recipients enjoy a small premium in hourly wages of about $1 over dropouts, but they have lower levels of job tenure, less total work experience, work fewer hours per year, and have lower levels of employment. Statistical tests measuring whether the two groups differ significantly in all of these outcomes tell us that GED recipients and dropouts have the same outcomes in all of these dimensions. In contrast, both groups are significantly inferior to traditional high school graduates without college on these characteristics.
Due to the fewer hours GED recipients work per year compared to dropouts, the small premium in hourly wage they receive does not translate into significantly higher annual earnings. Despite higher wages and significantly higher levels of cognitive skills, GED recipients earn just $600 more per year than uncertified dropouts and this difference is not statistically significant.
The evidence supports the hypothesis that GED recipients are unable to hold jobs due to their poor noncognitive skills. GED recipients changed jobs a significantly higher number of times than both dropouts and high school graduates resulting in the lowest tenure among all three groups. While job turnover can be a positive sign of career advancement and result in higher earnings, in the case of GED recipients this does not appear to be the case. The low annual earnings of male GEDs tells us that they are not advancing in their careers but rather are unable to hold onto jobs after getting them.
Females
The value of the GED credential may be different for females than it is for males. Occupations available to unskilled female dropouts may differ from those available to young male dropouts. In addition, the decision to leave high school among females may be due less to a lack of motivation or academic ability and more to an interruption in their studies caused by fertility. Therefore, the returns to GED certification may be fundamentally different for males and females. As we noticed in the tables of family characteristics presented in the Chapter 2, female GED recipients and dropouts are more likely than high school graduates to have given birth in their high school years. As the demands of childcare diminish for these individuals, they may decide to obtain the GED in order to help support their families and qualify for post-secondary education and training. The female GED may send a different signal to employers than her male counterpart.
Given the different labor supply characteristics and reasons for taking the exam, we separately analyze the mean labor market outcomes of female GED recipients, dropouts, and high school graduates. Table 2(b) shows that female GED recipients still are paid about a dollar an hour more than other dropouts. In contrast to males, we find clear differences in labor force participation rates. By age 30, female GEDs have accumulated more total work experience, work more weeks and hours per year, and are more likely to hold a job at the time of survey. However, when we restrict the sample to those who worked at least one week in the past year, we find no difference, with the exception of total experience, between GEDs and dropouts in these characteristics. As we observed for males, despite having greater overall labor force attachment, GED recipients have the same tenure on the job as dropouts and a higher number of jobs ever worked. It again appears that GED recipients are unable or unwilling to hold onto a job after they find one. Evidence suggests that the GED may be helping some dropouts gain entry into the labor force, but when they get there they do not retain their new jobs.
It appears from these mean results that the primary effect of the GED is bringing some females into the labor force. However, as we saw for the full sample of males, after they decide to participate they perform the same as dropouts in all the measured dimensions. As bleak as these results appear to be, we should caution the reader that unadjusted comparisons of means can be misleading. This is likely to be true in evaluating the impact of the GED on labor market outcomes, particularly for women. GED recipients attain more secondary school before they drop out, have higher AFQT scores and come from better family backgrounds. Unadjusted comparisons of GED recipients and dropouts would show greater earnings for GED’s, but this could be attributable solely to preexisting differences and not the title itself. All of these factors produce a bias in favor of higher wages and earnings of GED recipients compared to high school dropouts who do not attain the GED. Below we tackle the question of whether this advantage over dropouts is due to the characteristics of those who obtain the GED (e.g. characteristics such as those measured by the AFQT) or due to the value of the GED in signaling desirable characteristics to employers.
Controlled Analysis
Using a wide range of statistical analyses and adjustments we find that the simple story told by the means of the previous section holds up to a more rigorous treatment.[30]
Employment and Labor Supply
This section presents evidence on the effect of a GED degree on the labor supply of GED holders. Two main findings emerge from comparing GED-holders to other high school dropouts’ employment before and after the treatment. First, we find no evidence that the GED degree affects the labor supply of males. This holds when labor supply is measured either by annual work hours or employment status. For women, the story is slightly different. Women who GED certify are more likely to be holding a job at a given time but they do not end up working more annual hours than uncertified dropouts (after controlling for differences in family structure and fertility decisions). This evidence points to GED women having higher rates of job turnover that result in shorter job spells. We also demonstrate that for those women searching for work, a GED does not increase their probability of finding employment.
Regression analysis is the basic statistical tool used in this chapter to explore the relationship between schooling and labor market outcomes[31]. The more complex factor model structure, introduced in Chapter 4, is also used to explore the relationship of cognitive and noncognitive skills to labor market outcomes. In this section the dependent variable is either the annual worked hours, or binary variables indicating the employment status of the individual. In all specifications of the labor supply equation we control for age[32], region of residence, local unemployment rate, year of survey, marital status and race. Two further specifications control for years of schooling and schooling adjusted AFQT scores.
Males
We find GED-certified males are the same as permanent high school dropouts in terms of their labor supply and likelihood of working. Figure 1 presents models estimating the log of annual hours worked for 30 year old male dropouts, GED holders and high school graduates. Even before controlling for their higher test scores, GED recipients actually work 5% fewer hours per year at age 30 than permanent dropouts. However, this difference is not significant in a statistical sense[33].
These results show a simple relationship between education and labor supply. Causal interpretations of the estimated effect of education on labor market outcomes are difficult to make because people self-select into educational categories based on unobservable factors. For example, GED recipients score higher on tests of cognitive ability and have greater years of completed schooling relative to permanent dropouts before they take the GED examination. These preexisting productive factors are valued highly by the labor market and were not caused by the individual obtaining a GED. The GED did not make the person smarter, but rather the individual chose to GED certify because he is smarter than the average high school dropout. In this case the GED would appear to have a positive impact, but this probably reflects self-selection into the GED pool. Likewise, high school graduates are thought to complete secondary schooling because they are smarter and more persistent. For these reasons, to obtain causal estimates of the effect of education on labor market outcomes we include measures of cognitive and noncognitive skills in our models whenever possible.
After controlling for schooling-corrected AFQT scores, the poor outcomes of GEDs relative to both dropouts and graduates become even more pronounced. For instance, GEDs work 7% fewer hours at age 30 than uncertified dropouts of the same ability level. High school graduates, on the other hand, work 6% more than dropouts and 13% more than GED recipients, and these differences are always significant.
Even after adjusting for cognitive differences, there is still a clear return to graduating high school. As we saw in the previous chapter high school graduates also have higher endowments of noncognitive skill. Schooling fosters these skills. We also saw that people self-select into schooling choices based on both cognitive and noncognitive skills. To determine the causal impact of the GED and high school credential on labor market outcomes we also need to control for noncognitive differences. One proxy that we introduce which imperfectly controls for both cognitive and noncognitive ability is completed schooling. Adding this variable allows us to not only partially control for noncognitive differences but also assess whether there is a return to high school graduation over dropping out beyond the educational content of an additional two years of school that a graduate obtains.
After controlling for the total number of years of secondary schooling completed in the final model, we find that we cannot reject the hypothesis that dropouts, GED recipients and high school graduates have the same annual labor supply characteristics. In other words, there are no labor supply benefits to high school graduation or GED certification beyond that accrued from the time spent in the classroom[34]. This brings up a theme that arises repeatedly throughout this book: succinctly, “hard work pays off”. In human capital terms, we find that it is the time invested that is important and not certification levels. There is no quick or easy way to acquire the skills or rewards attained through conventional classroom instruction and training.
Strikingly similar results are obtained for the probability of employment when comparing the three educational categories. This outcome measures whether or not an individual is employed at the time the NLSY survey was given.[35] Despite the generally high acceptance of a GED credential among employers, it does not appear that those who obtain the title have any advantage over dropouts who do not have the title (Figure 2). Even before controlling for test scores, we estimate that GEDs work less than permanent dropouts, although this is once again not a statistically significant difference. Controlling for schooling and test scores only strengthens this conclusion. High school graduates are once again found more likely to be employed until we adjust for completed schooling.
A deeper understanding of the poor employment outcomes of those who fail to complete high school is provided by our factor model (it estimates both the cognitive and noncognitive abilities of all schooling groups). As we saw in the previous chapter, male GED recipients have the lowest noncognitive skill endowments, followed by dropouts and then high school graduates. Interestingly, the ordering of employment outcomes follows the same pattern. This is not surprising given the evidence presented by Bowles and Gintis and many others that finds employers value noncognitive traits such as dependability and motivation over raw cognitive ability; especially in low skilled positions.
Bowles and Gintis also find that these noncognitive skills are fostered through participation in schooling and that a large part of the payoff to schooling is a reward for higher noncognitive skills. Our estimation of labor supply and employment by schooling level backs up this theory. One proxy for noncognitive skill, albeit far from perfect, is years of completed schooling. When this measure is entered into our employment and labor supply estimation we find no differences between dropouts, GEDs and high school graduates.
Figure 3 and Figure 4 (not yet available) demonstrate the veracity of the Bowles and Gintis argument that employers place a high value on noncognitive skills. Each of these figures gives a graphic picture of how cognitive and noncognitive skills affect the outcome under examination (in Figure 3, employment; in Figure 4, hours worked). The top panel displays the joint distribution by deciles[36] on the effects of cognitive and noncognitive skills on outcome. The bottom two panels display the marginal effects of moving across deciles of one factor while holding the other factor at its mean. For example, in the case of employment, those in the bottom decile of the noncognitive skill distribution and of mean intelligence have a 60% probability of being employed. In contrast, those in the top decile are employed with a probability of nearly 100%.
The noncognitive factor is the most powerful predictor of whether or not someone is employed and of how many hours they supply to the labor market. The inability of GEDs to maintain steady employment despite having the same cognitive ability as high school graduates is understood through this model: GEDs generally lack the noncognitive skills necessary to find and/or hold it.
We quickly note that this finding is in sharp contrast to the claims made by Herrnstein and Murray in their book The Bell Curve. Using a biased and imperfect proxy of cognitive ability and no measure of noncognitive skill, they find that cognitive ability dominates in predicting employment outcomes. We show that this is not even half of the story. Noncognitive measures are better predictors of employment and this is ignored in The Bell Curve analysis. For many of life’s outcomes, noncognitive skills are as great, and in many cases better, predictors of success than IQ.
The important lesson from this section is that male GED recipients are working the same hours and gaining the same annual work experience as uncertified high school dropouts. At least by age 30, GED acquisition has done little to change the work habits or job search capabilities of those dropouts who chose to certify. The potential advantage gained from GED certification through high employer acceptance does not appear in the data. There are three potential explanations for this. First, employer surveys could be inaccurate. Second, GED recipients may be unable to either find or keep jobs. Lastly, they may not be applying to jobs in the first place. In all probability it is a combination of all three. A story that is consistent with both the evidence on labor market outcomes and the survey data is that employers are willing to hire a GED recipient but they quickly learn about their relatively poor noncognitive skills and let them go (or GEDs leave the job before the employer has a chance to realize this).
Females
Unlike males, female GEDs have more work experience and greater labor force attachment than uncertified dropouts. Is this a result of obtaining the GED or rather other unobserved characteristics that caused these women to both obtain certification and work more? It is plausible that differences in annual labor supply and accumulated work experience are the result of differences in marital and fertility decisions rather than a result of obtaining the GED. In other words, women may choose to GED certify because they are already planning on increasing their labor supply and job search activities after childcare demands decrease; what appears to be a result of the GED could actually be due to these confounding factors. Since we saw that a GED has no effect on male employment and labor supply outcomes this hypothesis seems more credible than the alternative of employers rewarding female GED holders but not male.
Figure 1 presents estimates of the log of annual hours worked for 30 year old females. To control for the effects of child rearing and family structure on female labor supply, we add a number of variables to this model that we did not have in the male specification. These include: children in the household, spouse’s income, whether the woman is pregnant and whether there is a baby (a child less than a year old) or toddlers (ages1-4) in the household. Analogous to males, women who hold a GED credential do work slightly more than uncertified dropouts; however, this difference is again not significant. As with males, high school graduates consistently work more than both groups and this difference is significant. After controlling for test scores and years of schooling, no difference in labor supply is observed between all three groups. Schooling and acquired skills once again dominate in explaining who works and how much they work. The greater number of annual hours worked by GED recipients in the raw data is completely explained by accounting for differences in family structure and pre-existing conditions such as test scores and schooling.
Estimates of the probability of being employed are shown in Figure 2. GED recipients are still less likely to be employed than high school graduates, but now we find what appears to be strong and significant effects of GED certification on employment. At age 30, female GED recipients are 15% more likely to be employed than uncertified dropouts. Controlling for ability and schooling lowers the employment benefit to a 10% greater chance of working over uncertified dropouts, but this is still a statistically significant advantage.
The above results present an interesting puzzle. Why do female GEDs work as many hours per year as other high school dropouts yet have significantly higher probabilities of being employed at any given time? In order to resolve this question we first ask whether a GED provides any benefit to those women actually seeking employment. To do so, we restrict the previous analysis to those women who are either employed or unemployed (meaning that they are seeking employment) in Figure 2. For those women looking for work, the GED does not give an advantage in finding employment over other dropouts. This implies that any employment gains result from GED women deciding to look for work, which changes their status from “out of the labor force” to either “employed” or “unemployed”. While the GED does not help women find a job it does appear that those who certify are more likely to be seeking employment and less likely to be out of the labor force. However, this is not necessarily a result tied to obtaining a GED. Many women take the GED as part of a general recommitment to the labor force. It is difficult to separate out the true effects of obtaining the GED from the changes in behavior or circumstances that would have taken place even had these women not obtained the title. The same forces that led women to take the GED may also be motivating them to increase their job search and labor force participation. Ideally, we would have data that tracked the labor force participation of both women who chose to take the GED and did not pass versus those who succeeded. It seems reasonable—especially since the GED does not appear to help people find jobs—that once we restricted our sample to those who had decided to take the test, we would observe no difference in labor market activity between those who passed and those who failed. In any case, when all is said and done at the end of the year, female GED recipients do not work more total hours than similar uncertified dropouts. We will see that this is a result of shorter job durations and higher rates of job turnover among GEDs.
Noncognitive skill is even more predictive of annual labor supply and employment for females than for males. Only 20% of females in the bottom decile of the noncognitive distribution are employed at the time of survey, whereas 96% of those in the top decile are employed. This is more evidence that GED holders and dropouts struggle to find and keep steady employment because of deficiency in noncognitive skills. Nearly 50% of those who drop out of school are in the bottom two deciles of the noncognitive distribution. This, coupled with the dominance of noncognitive skills in determining who is employed, gives those who drop out little chance of success in the job market.
Hourly Wages
Does GED certification have any direct impact by age thirty on the hourly wages of those who obtain the credential? Hourly wage is an outcome of great interest because economic theory tells us that it measures how productive someone is in the workplace. More precisely, workers are paid at the marginal product of their labor. So in this section we examine whether obtaining a GED makes a worker more productive in the labor market than they otherwise would have been.
Males
Do employers pay a premium to those holding a GED credential? The answer given in Figure 3 appears to be no. Before adjusting for ability, the return to GED certification is a positive 7% higher hourly wage. High school graduates earn 16% more per hour than dropouts. After adjusting for ability we find that there is once again no direct benefit of certification on wages. Interestingly, there is still a 10% positive return to high school graduation. When including years of completed schooling in the model, GED recipients actually earn less than uncertified dropouts, but this is not statistically significant. The positive returns associated with graduating high school are again found to operate exclusively through the additional years of education. Returns are consistent and there is no premium (jump in pay) for graduating.
We find that cognitive ability is a better predictor of hourly wages for dropouts (Figure 6), whereas noncognitive ability is more predictive for GEDs (Figure 7). After controlling for both the noncognitive and cognitive factors in the wage equation at age 30, we find very similar results to those obtained using AFQT and years of schooling proxies. After controlling for both factors we find that GEDs earn a statistically insignificant 1.7% higher hourly wage (Table 3).
Females
Nearly the same hourly wage pattern as observed for males is evident for females. Figure 3 shows the initial GED dropout difference to be greater than we observed for males; but the dropout high school gap is also much wider. Before correcting for ability, GED recipients earn 12.8% more than uncertified dropouts. In contrast to males this difference is significant. High school graduates earn a 27% premium over dropouts. After adjusting for higher ability, a 5% gap remains between GEDs and dropouts; but this is not significant. High school graduates still earn a substantial premium over the other two groups, earning 19% more than uncertified dropouts. Females high school graduates continue to differ from their male counterparts when controlling for completed schooling and ability and still earn significantly more than both GED recipients and dropouts. GEDs and dropouts show no significant differences in wages after controlling for observed differences in schooling and intelligence. This is consistent with the male results.
The slopes of the cognitive and noncognitive curves in Figure 8 show that female GED recipients are paid slightly more for their noncognitive than cognitive abilities. However, both factors are important in determining wage rates. Female dropouts (Figure 9), have steeper wage profiles along the cognitive factor. This means that increasing a female dropout’s cognitive factor will bring about a greater increase in hourly wage than an identical increase in the noncognitive factor. Consistent with the story for males, noncognitive traits are more important for GEDs in achieving positive labor market outcomes. For dropouts, poor outcomes are more a result of cognitive deficiencies. After controlling for both types of skill, we see (Table 3) that GED certification has zero impact on the wages of thirty year olds who chose to GED certify.
Annual Earnings
Annual earnings provide a composite measure of labor market performance, by taking into account both labor supply and hourly wage rates. If zero effect is found, we can conclude that GED attainment is not affecting either the hourly wage rates or annual labor supply of GED holders.
Males
In short, male GEDs earn no more in a year than high school dropouts who never got the degree. In fact, it appears they earn even less (Figure 10). Despite significantly higher measured test scores, GED recipients earn only 2% more than dropouts. Meanwhile high school graduates, who have the same measured ability as GEDs, enjoy a 24% premium in wages. After controlling for AFQT, we find that GEDs actually earn 6% less than dropouts with similar cognitive ability. This is consistent with the finding that GEDs have lower noncognitive skills. As we’ve already seen, the GED title gives them no premium in terms of hourly wages and relatively poor noncognitive skills cause GEDs to work fewer hours at this wage rate than other dropouts. High school graduates earn 17% more annually over dropouts at the same ability level, and 23% more than GEDs. Since high school graduates have significantly higher noncognitive skills than both groups, we can interpret this as the noncognitive premium to graduating high school. Adjusting for completed schooling, largely a measure of noncognitive skill, lowers this figure to 6.5% (statistically insignificant).
Females
Female GEDs have noncognitive skills comparable to high school dropouts and so we would expect that their earnings would be the same as other dropouts of the same ability level, not worse (as we observed for males). In fact, this is what we see in Figure 10. Female GED recipients earn slightly more than dropouts with the same cognitive ability, while high school graduates earn 25% more. Adding schooling to the equation dampens the GED effect to only 3.4%. Again, the GED is not helping these people earn more than we would expect from an uncertified dropout at the same ability and schooling level. High school graduates, on the other hand, earn significantly more than dropouts and GEDs with similar backgrounds and skills.
Conclusion
This chapter evaluates the importance of educational attainment, cognitive skills, and noncognitive skills in determining labor market outcomes at age 30. We find that both cognitive skills and noncognitive skills (such as self-discipline and persistence) contribute substantially to labor market performance. We find evidence that labor markets price out both types of skills. Part of what appears to be a return to education reflects the monetary value of unobserved noncognitive skills that are correlated with educational attainment. High school graduates have higher endowments of both types of skills relative to GEDs and dropouts. As a result graduates have better labor market outcomes in all measured dimensions. High school graduates who do not attend college have higher: earnings; hourly wages; hours worked; probability of employment; accumulated work experience and tenure than both uncertified dropouts and GED recipients.
A second finding is that GEDs who do not attend college have the same labor market outcomes as observationally similar dropouts. Comparing the unadjusted mean wages of GED recipients and dropouts shows a small premium to GED certification. However, after controlling for measures of both cognitive and noncognitive ability, GED recipients have the same or worse labor market outcomes as uncertified dropouts.
The GED program screens dropouts only based on cognitive skills. Dropouts with higher cognitive skills relative to other dropouts choose to GED certify. However, these individuals lack basic noncognitive skills that negatively affect their outcomes. The lack of perseverance and social skills that led them to dropout in the first place also leads them to have poor economic outcomes despite higher levels of cognitive skill than uncertified dropouts. We find no improvement in GED outcomes after acquisition of the degree and conclude that the GED exerts no treatment effect for those who choose to certify.
References
Araujo, Aloisio, Daniel Gottlieb, and Humberto Moreira. 2004. “A Model of Mixed Signals with Applications to Countersignaling and the GED.” Working paper, Sociedade Brasileira de Econometria.
Becker, Gary. 1964. Human Capital: A Theoretical and Empirical Analysis with Special Reference to Education. New York: Columbia University Press.
Berg, Ivar. 1970. Education and Jobs: The Great Training Robbery. New York: Praeger.
Bernstein, B. 1977. Class, Codes, and Control: Towards a Theory of Educational Transmissions, 2nd ed. London: Routledge & Kegan Paul.
Boesel, David, Nabeel Alsalam, and Thomas Smith 1998. Educational and Labor Market Performance of GED Recipients. Washington, D.C.: U.S. Department of Education.
Bossert, Steven T. 1979. Tasks and Social Relationships in Classrooms: A Study of Instructional Organization and its Consequences. New York: Cambridge University Press.
Bossert, Steven T. 1988. “Cooperative Activities in the Classroom.” Review of Research in Education 15: 225-250.
Bourdieu, P. and J.C. Passerson. 1977. Reproduction in Education, Society and Culture. London: Sage.
Boesel, David, Nabeel Alsalam, and Thomas Smith. 1998. Educational and Labor Market Performance of GED Recipients. Washington, D.C.: U.S. Department of Education.
Bowles, Samuel and Herbert Gintis. 2001. “The Inheritance of Economic Status: Education, Class, and Genetics.” In International Encyclopedia of the Social and Behavioral Sciences: Genetics, Behavior, and Society, Vol. 6, Marcus Feldman and Paul Baltes, eds. New York: Oxford University Press and Elselvier: 4132-41.
Bowles, Samuel and Herbert Gintis. 2002. “The Inheritance of Inequality.” The Journal of Economic Perspectives 16, no. 3: 3-30.
Bowles, Samuel, Herbert Gintis, and Melissa Osborne. 2001. “The Determinants of Earnings: A Behavioral Approach.” Journal of Economic Literature 39, no. 4: 1137–76.
Bratsberg, Bernt and James F. Ragan Jr. 2002. “The Impact of Host-Country Schooling on Earnings: A Study of Male Immigrants in the United States.” The Journal of Human Resources 37, no. 1: 63-105.
Cameron, Stephen and James J. Heckman. 1993. “The Nonequivalence of High School Equivalents.” Journal of Labor Economics 11, no. 1, part 1: 1-47.
Cameron, Stephen. 1994. “Assessing High School Certification for Women who Drop Out.” University of Chicago. Presented at the annual meeting of the American Economics Association, Washington, DC, January 1995.
Clark, Melissa A. and David A. Jaeger. 2006. “Natives, the foreign-born and high school equivalents: new evidence on the returns to the GED.” Journal of Population Economics 19, no. 4: 769-793.
Collins, Randall. 1971. “Functional and Conflict Theories of Educational Stratification.” American Sociological Review 36: 1002-19.
Collins, Randall. 1979. The Credential Society: An Historical Sociology of Education and Stratification. New York: Academic Press.
Dreeben, Robert. 1967. On What is Learned in School. Reading, MA: Addison-Wesley.
Friedberg, Rachel M. 2000. “You Can't Take It with You? Immigrant Assimilation and the Portability of Human Capital.” Journal of Labor Economics 18, no. 2: 221-251.
Gintis, Herbert. 1971. “Education, Technology, and the Characteristics of Worker Productivity.” The American Economic Review 61, no. 2: 266-279.
Heckman, James J. and Paul A. LaFontaine. 2006. “Bias-Corrected Estimates to GED Returns.” Journal of Labor Economics 24, no. 3.
Jackson, PW. 1968. Life in Classrooms. New York: Holt, Rinehart, & Winston.
Jencks, Christopher, Marshall Smith, Henry Acland, and Mary Jo Bane, David Cohen, Herbert Gintis, Barbara Heynes, and Stephen Michelson. 1972. Inequality: A Reassessment of the Family and Schooling in America. New York: Basic Books.
Laurence, Janice H. 1987. Military Enlistment Policy and Education Credentials: Evaluation and Improvement (FR-PRD-87-33). Alexandria, VA: Human Resources Research Organization.
Malizio, Andrew and Douglas Whitney. 1982. “Examinee and High School Senior Performance on the GED Tests: A Detailed Analysis.” GED Testing Service. Research Studies no. 3. Washington, D.C.: American Council on Education.
Mally, Nell and Mary Charuhas. 1977. “A Survey: Perceptions of the College of Lake County's G.E.D. Program as Seen by Former Students and Area Personnel Managers.” Lake County, IL: College of Lake County.
Mueser, Peter. 1979. “The Effects of Noncognitive Traits.” In Who Gets Ahead? The Determinants of Economic Success in America by Christopher Jencks et al. New York: Basic Books
Murnane, Richard J., John B. Willett, and John H. Tyler. 2000. “Who Benefits from Obtaining a GED? Evidence from High School and Beyond.” The Review of Economics and Statistics 82, no. 1: 23-37.
Murnane, Richard J., John B. Willett, and Kathryn Parker Boudett. 1995. “Do High School Dropouts Benefit from Obtaining a GED?” Educational Evaluation and Policy Analysis 17, no. 2: 133-147.
Murnane, Richard J., John B. Willett, and Kathryn Parker Boudett. 1999. “Do Male Dropouts Benefit From Obtaining A GED, Postsecondary Education, And Training?” Evaluation Review 23: 475-503.
Pawasarat, John and Lois Quinn. 1986. “Research on the GED Credential and its Use in Wisconsin.” Employment and Training Institute. Milwaukee, Wis.: University of Wisconsin-Milwaukee.
Quinn, Lois. 1993. The Test that Became an Institution: A History of the GED. Milwaukee: Employment and Training Institute, University of Wisconsin-Milwaukee.
Rubinstein
Schoeni, Robert F. 1997. “New Evidence on the Economic Progress of Foreign-Born Men in the 1970s and 1980s.” The Journal of Human Resources 32, no. 4: 683-740.
Spence, Michael. “Job Market Signaling.” The Quarterly Journal of Economics 87, no. 3: 355-374.
Tyler, John H. 2001. “What Do We Know About the Economic Benefits of the GED? A Synthesis of the Evidence from Recent Research.” Unpublished, Brown University.
Tyler, John H. and Jeffrey R. Kling. 2003. “What is the Value of a ‘Prison GED’?” Unpublished, Brown University.
Tyler, John H. and Magnus Lofstrom. 2004. “Measuring the Returns to the GED: Using an Exogenous Change in GED Passing Standards as a Natural Experiment.” IZA Discussion Paper No. 1306.
Tyler, John H., Richard J. Murnane, and John B. Willett. 2000. “Estimating the Labor Market Signaling Value of the GED.” The Quarterly Journal of Economics 115, no. 2: 431-468.
Tyler, John H., Richard J. Murnane, and John B. Willett. 2003. “Who Benefits from a GED? Evidence for Females from High School and Beyond.” Economics of Education Review 22: 237–247.
-----------------------
[1] Malzio and Whitney 1982.
[2] The results for males are in Cameron and Heckman (1993); results for females are in Cameron 1994.
[3] See Boesel, Alsalam and Smith (1998) for a summary of the literature on this point.
[4] Becker 1964
[5] Quinn 1993
[6] Quinn 1993
[7] Murnane, Willett and Boudett 1999; Murnane, Willett and Tyler 2000; Tyler, Murnane and Willett 2000a and 2000b; See Tyler “What do we know” 2001 for a summary of these results.
[8] Wagner (1995)
[9] See for example, Berg. 1970; Spence, 1973; Collins, 1979
[10] See Spence 1979
[11] See the survey by Bowles, Gintis and Osborne.
[12] See Bossert 1979, 1988; Dreeben 1968; Jackson 1968
[13] See Bernstein 1977; Bourdieu and Passerson 1977; Bowles and Gintis 1976, 2001, 2002; Collins 1971, 1979; Gintis 1971; Jencks 1972; Mueser essay in Jencks 1979.
[14] This gives rise to important differences from traditional signaling models. See Moreira and Gottlieb 2002
[15] This estimate is in 2000 dollars. See Lawrence 1987.
[16] “Do High School Dropouts Benefit from Obtaining a GED?”
[17] Murnane, Willett, and Boudett (1999) “Do Male Dropouts Benefit from Obtaining a GED, Post-Secondary Education, and Training?”
[18] Murnane, 1996
[19] “Estimating the Labor Market Signaling Value of the GED”
[20] See technical appendix B
[21] The 10-19% return they find for GED recipients can be generated for high school dropouts living in low and high pass score states who do not GED certify.
[22] "Measuring the Returns to the GED: Using an Exogenous Change in GED Passing Standards as a Natural Experiment"
[23] “The Value of a Prison GED”
[24] See Friedberg (2000), Schoeni (1997) and Bratsberg and Ragan (1999).
[25] GED Official Web Site: Welcome to the GED
[26] Pawasarat and Quinn (1986)
[27] Interested readers should see Tyler (2001) for a summary of this evidence.
[28] Returns to GED certification for the incarcerated population are examined in Chapter 7.
[29] For a full discussion of these issues see Heckman and LaFontaine (2006)
[30] The interested reader should consult Appendix A for the technical details of these models.
[31] Specifically, OLS regressions and probit models. See the technical appendix for further details.
[32] Age is used instead of work experience since work experience is endogenous and a proxy for noncognitive skill endowments.
[33] We cannot reject the null hypothesis that a GED and dropout work the same hours annually at the 5% level. See the technical appendix for a detailed description of hypothesis testing.
[34] These are commonly referred to as “sheepskin” effects in the literature.
[35] Those who are employed but not currently at their jobs (i.e. people on vacation or sick leave) are counted as employed. Those who ever served time in jail, currently are in the military or enrolled in school are once again excluded from the analysis.
[36] A decile is one tenth of a distribution, i.e. the distribution is broken into ten bins of equal size.
................
................
In order to avoid copyright disputes, this page is only a partial summary.
To fulfill the demand for quickly locating and searching documents.
It is intelligent file search solution for home and business.
Related download
Related searches
- america s education system failing
- america s home grant program bank of america
- america s most influential people
- america s top facial plastic surgeons
- america s founding principles
- america s home grant program
- america s broken criminal justice system
- what are america s core values
- america s education system is failing
- problems with america s education system
- america s best management consulting firms
- america s top ten most wanted