While college graduates, including those with advanced degrees, came close to maintaining their earnings after an initial decline in the early 1970's, those with less education continued to lose ground. The quantified slopes of the lines in this figure (not shown here) indicate that the rate of decline is negatively associated with education level. Those with the least education and the lowest earnings, the dropouts, are losing ground most quickly.[46] Dropouts would gain substantially if they had the earnings of high school graduates, and the rate at which they are losing ground would decrease somewhat. Nevertheless, they would still be experiencing a long-run decline in real earnings.
To assess the economic performance of GED recipients, we review studies using cross-sectional and longitudinal approaches. A discussion of the strengths and weaknesses of these analytic strategies is found in appendix C.
Simple comparisons of labor force participation showed that GEDs were more likely than dropouts but less likely than high school graduates to participate (see the "Difference" column in table B-6). However, in controlled analyses the differences among the 3 groups were slight or nil in all but one study (Passmore 1987). Credentials seem to have little effect on labor market participation.
In simple estimates of employment rates, GED males were a little more likely than dropouts to be employed (full-time or part-time). GED females and adults in general (both sexes) were considerably more likely to be employed. High school graduates had higher employment rates than GEDs, whether male or female. One study (Reder 1994) found that GED males and adults of both sexes had higher rates of full-time employment than other dropouts.
In controlled comparisons, GED males were 5 percent more likely than dropouts to be employed, and GED adults were about 6 percent more likely to be employed full-time. Male high school graduates and adult high school graduates in general had higher overall employment rates than their GED counterparts. There were no controlled comparisons of employment among females.
Simple comparisons of unemployment rates showed that GED recipients were more likely than high school graduates to be unemployed. Female GEDs were less likely than dropouts to be unemployed (in 2 of 3 studies), though the evidence regarding GED males and dropouts was mixed. The one controlled study of unemployment (Sum 1996) showed that among adults of both sexes, the GED rate is the same as that of dropouts and higher than that of high school graduates.
In general, the studies suggest that while GED attainment has little effect on labor-force participation or unemployment, it seems to increase full-time employment somewhat.
The results of controlled analyses were similar. The wages of GED males were still 6 to 12 percent higher than those of dropouts; the wages of GED females were up to 13 percent higher; and the wages of GED adults in general were 5 to 11 percent higher. In every study but one, GEDs earned more than other dropouts but less than high school graduates.
In controlled studies, then, GEDs tended to have higher wages than dropouts and lower wages than graduates. However, the introduction of certain controls in some analyses reduced the wage differences between GEDs and dropouts considerably. In one study, for example, introducing controls for level of education, literacy, and work experience reduced the difference between GED and dropout wages from 19 percent to 10 percent (Reder 1994). This suggests that much of the apparent wage effect of GED status may actually be an effect of other pivotal characteristics of GED recipients. Shortly we will examine the extent to which, and ways in which, key variables such as these affect the wages of GED recipients.
Simple and controlled comparisons of time worked (e.g., hours per week) and work experience (total weeks) tell a fairly consistent story (table B-6). GED males work less than or about the same amount as dropouts. GED females work more than dropouts. High school graduates, in turn, work more than GEDs, regardless of sex. Simple comparisons of job tenure for the three groups reveal a similar pattern--male GEDs have less tenure than dropouts, females have more, and high school graduates of both sexes have more than GEDs.[48]
Analysts disagree about whether the greater number of hours worked by GED females, as compared to dropouts, is due to the credential. Cao et al. (1993) and Cameron (1994) concluded that the differences between GEDs and other dropouts were not statistically significant. Cameron suggested that "all differences in the annual labor supply and accumulated work experience are governed by differences in fertility and marital decisions."
On the other hand, Maloney (1993) found that for women there were labor supply effects of obtaining a GED (or a high school diploma). He concluded from two controlled analyses that "the average dropout would see her actual hours of work rise by 17.6 and 17.3 percent with a high school diploma and 8.0 and 6.4 percent with a GED degree."[49]
In sum, GED males tend to work less than or about the same amount as dropouts and experience more job turnover. GED females tend to work more than dropouts and have less turnover, but there is disagreement about whether their increased working time is a result of GED attainment. GEDs in general tend to work less than high school graduates and have more turnover.
Controlled comparisons of annual earnings are found in three studies-Sum (1996), Cave and Bos (1994), and Tyler et al. (1997).
Sum's analysis of NALS data controlled for literacy, educational attainment, school enrollment status, marital status, and race-ethnicity, among other things. He found that GED males with no further education earned 13 percent more than dropouts, and GED females, 21 percent more. The magnitude of these differences, even after controlling for the higher literacy skills of GED recipients, is substantial. However, the GEDs still earned less than high school graduates.
Cave and Bos (1994) presented evidence from an experiment designed to evaluate the effect of JOBSTART, an education and training program for high school dropouts, all of whom were reading below the eighth-grade level. One component of the program was many hours of basic education, along with encouragement and assistance in getting a GED credential. Among the experimental JOBSTART group, 36 percent obtained the GED, compared to 21 percent in the control group. Combining the GED recipients in both groups and comparing them to nonrecipients in both groups, the authors found that the GEDs earned $12,766 compared to $9,405 for the nonrecipients during the third and fourth years of the program. This is a substantial difference, but the groups were not balanced with respect to many individual characteristics. To alleviate the possible bias, the authors used membership in the experimental group as an instrumental variable to predict receipt of the GED. Although the resulting estimates were not statistically distinguishable, the magnitudes of the estimates were similar to those of the simple comparison between GED recipients and nonrecipients.
Using data from the GED Testing Service and the Social Security Administration, Tyler et al. (1997) conducted a natural experiment to assess the effect of GED attainment on annual earnings. They compared the earnings of dropouts who passed the GED in states having relatively low passing standards with the earnings of other dropouts who took the test and received the same scores but did not pass because they lived in states with relatively high standards. This approach enabled the researchers to compare outcomes for dropouts having the same motivation to take the test and the same cognitive skills, some of whom received GED credentials (the treatment group) and some who did not (the controls), by virtue of the states they happened to live in.
Five years after taking the tests in 1990, white GED recipients earned 10-19 percent more than controls (in 3 experiments), but nonwhite GEDs gained no earnings benefit.[50] GED earnings gains increased year by year but were relatively low or negative in the first 2 years after attainment of the credential. The authors think that GED certification affects earnings by increasing opportunities for additional education and training, stimulating job search, and signalling skills to others. The relatively low returns to the GED in the first 2 years may be opportunity costs of acquiring more education and training.
In general, this research shows that GED recipients earn more than comparable dropouts and indicates or strongly suggests that a major factor in the increased earnings is the additional education and training that the GED enables recipients to acquire.
In Cameron and Heckman's (1993) analysis, GEDs had a mean 10.40 years of schooling, compared to 9.46 years for dropouts, and of course, 12 years for high school graduates. Excluding those with any higher education, the authors found that the wage differences between male diploma graduates, GEDs with 11 years of schooling, and dropouts with 11 years of schooling were much smaller than those between all male diploma graduates, GEDs, and dropouts.[51] When the authors added years of schooling to a model that included race and ethnicity, the wage coefficients of the GED and the high school diploma fell to near zero, except for high school graduates at age 28. Across all groups, the coefficients were jointly insignificant, suggesting that there was no additional "sheepskin" or certification effect of either credential on wages.
Murnane et al. (1995) reported findings similar to those of Cameron and Heckman. Their sample of male GEDs had a mean 9.12 years of schooling and dropouts had 8.79 years. Their analysis found 4 percent higher wages for each additional year of schooling completed, controlling for race, ethnicity, and mother's education. Once years of schooling were taken into account, GED certification had no additional effect on wages.
Cameron (1994) reported that among females at age 25, GEDs had a mean 10.3 years of schooling, compared to 9.5 years for dropouts. He found that earnings rose with years of schooling, as Cameron and Heckman did for males. Unlike Cameron and Heckman, however, he found that for females at ages 25, 28, and 30, high school graduates did earn more than GED recipients and other dropouts with 11 years of schooling. Also, GED recipients with 11 years of schooling earned significantly more than similar dropouts at ages 25 and 28. The findings for females at age 30 were in the same direction, but not significant.
Cao et al. (1993) reported that for women aged 22-29, each year of additional schooling completed was associated with $0.38 higher hourly wages. After adjusting for years of schooling, the difference in hourly earnings between diploma graduates and GED graduates fell from $0.33 per hour to an insignificant amount in the opposite direction. That is, the wage difference between GEDs and high school graduates was attributable to their respective years of schooling. However, controlling for years of schooling did not eliminate the wage difference between GEDs and other dropouts. Rather, the difference fell from $1.50 to $0.47 per hour. Other attributes of the two groups, or their situations, must explain the remaining difference.
In summary, among males of similar race, ethnicity, and maternal education, all differences in hourly earnings between the three groups can be accounted for by differences in years of secondary schooling completed. Neither the high school diploma nor the GED have a "sheepskin effect." Among females, some, but not all, of the wage differences can be accounted for by years of schooling, leaving open the possibility that there may be some GED credential effect.
Cameron and Heckman (1993) compared the wages of GED males with those of dropouts and high school graduates and then added indicators of postsecondary completion--2 years with a GED, 2 years with a diploma, and neither. Including these variables reduced the estimated difference in hourly wages between the groups. For 25-year-old males, the difference between dropouts and GEDs declined a little, from -6.7 to -5.8 percent; the difference between GEDs and high school graduates dropped from +16.2 percent to +8.8 percent. (The negative signs denote lower earnings than GEDs, the positive signs, higher earnings.) An analysis of the wages of 28-year-old males produced similar results.
Cameron (1994) estimated the hourly wages of females at age 25, first including and then excluding those with any formal schooling beyond high school (and those actually in school). The exclusion reduced the difference between GED wages and those of dropouts from -11.1 percent to -9.7 percent, and it reduced the difference between GED wages and those of high school graduates from +19.1 percent to +9.1 percent. Results were similar for 28-year-olds. Adding other control variables in a regression equation changed these results very little.
Maloney (1993) found differences similar to Cameron's among the three groups of women he examined, once those with postsecondary education were excluded from the analysis.
Variation in postsecondary attainment helps explain the wage differences between GED recipients and diploma graduates. To a lesser extent, variation in postsecondary attainment helps explain the difference between GED recipients and other dropouts. Nevertheless, the GED is still associated with increased wages vis-a-vis dropouts after the effect of postsecondary education is taken into account.
Cameron and Heckman (1993) estimated wage models for males, controlling on standard background variables and then adding AFQT scores. The difference in hourly wages between dropouts and GED recipients changed from -5.8 percent for 25-year-olds and -6.0 percent for 28-year-olds to +3.9 percent and +2.0 percent, respectively (dropouts earning more than GEDs). The differences between GEDs and high school graduates were less affected by introducing AFQT as a control: they fell from +8.8 and +11.9 percent to +7.3 and +10.5 percent. With the relatively small samples of GED recipients in the NLSY, neither the wage differences between groups nor the changes in differences passed conventional tests of statistical significance.
One finds the same pattern for females in Cameron's 1994 study. After estimating wage equations, controlling for standard background variables, he added AFQT as a control. Once again, the addition of AFQT to the equation reduced the wage differences between GEDs and dropouts to near zero, or to a positive advantage for dropouts, while reducing the differences between GEDs and high school graduates only a little. For example, among females at age 28, the difference between GEDs and dropouts fell from -7.8 percent to +1.9 percent, while the difference between GEDs and high school graduates dropped from +16.2 percent to +13.4 percent with the introduction of the AFQT variable.
The results of Garet, Jing, and Kutner's (1996) analysis are largely consistent with these findings, though there are differences. The authors estimated log hourly wage rates for GEDs and high school graduates at age 28, controlling for demographics, work experience, and local unemployment rates. Both groups were compared to high school dropouts. Adding an ASVAB composite score to the equation caused the GED coefficient for males to fall from a statistically significant .10 to a non-significant .046. For male high school graduates, the diploma coefficient dropped from .18 to .101 (both significant) with the introduction of the ASVAB composite. The coefficients for female high school graduates also dropped, but the change cannot be attributed entirely to the ASVAB.[52]
Maloney (1993) treated select ASVAB scores as a measure of cognitive achievement reflecting human capital accumulation. He attempted to estimate the direct effects of GED certification and of a high school diploma on the wages of females age 23-31, and to estimate their indirect effects on wages through the ASVAB.[53] Dropouts were the comparison group. Like Cameron and Heckman, Maloney found no significant direct effect of GED certification on wages after the ASVAB variable was included in the regression. The introduction of the ASVAB variable also reduced the direct effect of high school graduation, sometimes to nonsignificant levels.
In the Cao et al. (1993) analysis of females, the introduction of AFQT scores made the differences between the wages of high school graduates on the one hand and GEDs and dropouts on the other nonsignificant. The coefficient for the difference between GEDs and high school graduates (-.22) was smaller than the coefficient for the difference between dropouts and graduates (-.47).[54] Reder (1994) estimated the effects of the GED on wages directly and indirectly through a measure of cognitive skills, in this case, National Adult Literacy Survey scores in Oregon. Comparing GEDs to other dropouts, and controlling on a number of variables, he found a .107 GED effect on wages. Of this total, .047 or 44 percent was an indirect effect through literacy, and the remainder was a direct effect of the GED.[55]
As expected, the introduction of cognitive ability as a control greatly reduces or eliminates any wage differences between GEDs and dropouts but has a smaller effect on differences between GEDs and high school graduates. Controlling on ability statistically removes the GEDs' primary advantage over other dropouts--their greater cognitive skills, the basis on which they were selected as GEDs. On the other hand, because the cognitive skills of GEDs are fairly close to those of high school graduates, the difference in their wages is less affected by the introduction of ability as a control.
These analyses of the effects of secondary schooling, higher education, and cognitive ability suggest that accumulated human capital, reflected here in level of education and aptitude scores, accounts for most of the wage differences among high school graduates, GEDs, and dropouts.
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
In a multivariate analysis with controls for race, work experience, and unemployment, Garet, Jing, and Kutner (1996) estimated the indirect effects of postsecondary education on log wages through higher education. Dropouts were the omitted category. The authors found that completing some college (up to two years) was associated with a 6 percent wage increase for male GED recipients and a 7 percent increase for male high school graduates. Female high school graduates benefitted from some college, with a 9 percent increase in wages, but female GEDs did not. The authors further found that completing two years of college was associated with wage gains of 10 percent and 8 percent for male GEDs and high school graduates, respectively, and 27 percent and 19 percent for their female counterparts.
Cameron and Heckman (1993) and Cameron (1994) investigated these indirect effects, following a two-step procedure. The first step, which did not control for tenure and experience, yielded total effects of the GED and the high school diploma on hourly wages. The second, which did control for tenure and experience, yielded direct (or partial) effects. The difference between the total effect and the partial effect was the indirect effect of the GED or the high school diploma on wages through tenure and experience.[56]
As table 17 shows, the authors found that for 25-year-old males, including tenure and experience in the equation increased the effect of the GED from 0.060 to 0.071, rather than reducing it. This occurred because (as we saw in table B-6) the male GED recipients in these studies had less tenure and experience than other dropouts, and this deficit tended to lower their wages. Once tenure and experience were controlled, the wages of GEDs vis-a-vis dropouts increased. In fact, it was the dropouts who experienced a positive and indirect effect on wages through tenure and experience.
For similarly aged females, however, Cameron found that including tenure and experience in the equation decreased the effect of receiving the GED, as expected, from 0.102 to 0.064. For these females, the author estimated that between 37 and 74 percent of the total effect of the GED on wages was indirect through greater tenure and experience.
In sum, both male and female GEDs receive indirect wage benefits from their certification through additional education. Female GED recipients also benefit indirectly through additional tenure and experience. Male GEDs, however, do not increase their wages through tenure and experience, because they have less than dropouts in these studies.
![[Table of Contents]](/icons/up.gif){kind=icon}
[Labor Market Performance of GED Recipients (part 2 of 2)]