As we have noted, the STW movement is powered in part by a desire to solve persistent problems in the youth labor market, especially for young people who do not hold bachelor's degrees. Murnane and Levy (1996), among others, argue that the mismatch between the skills high schools have traditionally imparted and those the market now rewards is evident in the increasing divergence between high school and college graduates' wages (Karoly, 1996; Levy & Murnane, 1992). To illustrate, Murnane and Levy (1996) point out that in 1979, the average 30-year-old man holding only a high school diploma earned $27,700 (in 1993 dollars). By 1983, this amount had declined to $23,000, and by 1993 to $20,000.
In addition to boosting wages by improving the skills and knowledge of high school graduates, STW initiatives are also intended to create new institutional arrangements that may remedy persistent unemployment and job instability among young adults. These may take a toll on youths' long-term career prospects. By reducing dropout rates, improving academic achievement, and imparting work-related skills, STW programs may allow young people a smoother transition to stable, learning-intensive employment. This sort of early labor market experience may then lead to long-term stability, further training, and higher wages.
A significant body of literature suggests that young people's typical STW transitions often include a wasteful period of floundering, during which high school graduates flip and flop from job to job without much sense of purpose or career direction (Hamilton, 1990; Osterman & Iannozzi, 1993). Compared to other industrialized countries, young Americans are less likely to be in jobs they have held for five years or more, even after reaching age 35 (Stern et al., 1995, p. 7). Instability in the youth labor market is seen as harmful to young people, especially minority and low-income. It is also seen as damaging the national economy, if the difficulty of getting and keeping qualified workers causes employers to leave the country.
From the viewpoint of economic efficiency, Klerman and Karoly (1995a) have identified three separate sets of issues in this argument:
First, when people are out of work involuntarily, they are not accumulating the knowledge and skill that are acquired through work experience. When jobs end, young people are at risk of finding themselves unemployed or out of the labor market. Apart from the benefit of increased leisure time, such non-employment is evidently unproductive. Furthermore, young people do not accumulate experience or more specific forms of human capital while not employed. Of course, not all job-leavers become unemployed. Parsons (1991), for instance, found more than half of job quitters in his study had already arranged for their next job before leaving their current one. A period of unemployment is more likely for individuals whose job loss is not a result of their own decisions. Second, moving from one employer to another may waste firm-specific knowledge and skill. When an employment relationship ends, the value of the knowledge that the particular employer and the employee have about each other is lost. One measure of that value is the higher wages that tend to be paid to employees who remain for a longer time with a particular firm. However, as Altonji and Shakotko (1987) and Topel (1991) have pointed out, longer tenure with a given employer may not actually cause higher wages, because job-leavers are unlikely to be a random sample of the population. In other words, it is precisely those employees who would not have experienced productivity and wage gains who are likely to leave a given job. This type of self-selection will statistically give rise to positive associations between tenure and earnings which may not be causal. After further analysis, however, Topel concluded that most of this association actually is causal. Jacobson, Lalonde, and Sullivan (1993) presented evidence that points to a similar conclusion.
Third, if employers expect young people to move around, they may be less likely to invest in additional training for them, creating a vicious cycle of low investment and high mobility. High job mobility in the youth labor market may weaken firms' incentives to invest in young workers, since the period of time over which they can profit from such investments becomes on average short and uncertain (see Becker 1964).
As opposed to these concerns, a different theory holds that instability may reflect productive investment in job shopping by which employers and employees eventually find "good matches" that raise their mutual productivity for an extended period thereafter (Heckman, 1994; Johnson, 1978; McCall, 1990). As Heckman (1994) put it, "Job shopping promotes wage growth. Turnover is another form of investment, not demonstrably less efficient than youth apprenticeships" (p. 105). The models motivating this view generally involve situations in which people and jobs differ in various respects, and therefore some people are more productive than others in certain jobs. No workers or firms have prior information on what the best matches are, however, and in order to obtain such information people must take different jobs and switch when they find others for which they are better suited (Flinn, 1986; Mincer & Jovanovic, 1981). High turnover may be positive in this perspective, not only because individuals make higher wages as they make better matches, but also because aggregate productivity is increased. The implication is that policies or programs aimed at reducing job shopping would be detrimental to young people and society in general.
Two recent studies shed new light on this dispute. One by Klerman and Karoly (1995a) found that the transition from school to work is smoothest for four-year college graduates and roughest for high school dropouts, compared to high school graduates or those with "some college." This result is more consistent with the view of instability as a negative experience that people try to avoid because if job shopping were a form of positive investment it is not clear why people with more schooling would get less of it. Furthermore, a second study performed by Gardecki and Neumark (1995) revealed that people who experienced more unstable employment in the first year or two after leaving school tended to continue having more unstable employment three or four years later. Again, this seems more consistent with the idea that unstable employment is a bad thing because if it led to better job matches it should not be expected to go on for so long.
The best data for analyzing unemployment and instability in the youth labor market comes from the National Longitudinal Survey of Youth (NLSY), which began in 1979 with a panel of almost 13,000 respondents. In a previous study by Veum and Weiss (1993), the NLSY revealed that between the ages of 18 and 27, the average high school graduate not continuing on to higher education held almost six different jobs and experienced more than four unemployment spells. Klerman and Karoly (1995a) have recently analyzed the NLSY data in greater detail. Their analysis shows that young people with more schooling experience more stable employment after they leave school.
Klerman and Karoly divide respondents into five school-leaving groups (SLGs),[2] according to the level of schooling they have completed at the time they had been out of school for more than three to five months:
Individuals may subsequently return to school, but this does not affect their SLG category (i.e., being in a given SLG is not necessarily the same as having the current schooling indicated by the SLG's label).
Static Analysis
Klerman and Karoly first consider a static picture of the labor market, analyzing the percentage of members from each SLG in 1990 who were engaging in one of four activities: (1) working full-time (35 or more hours per week), (2) attending school and not working full-time, (3) working part-time and not attending school, and (4) neither working nor attending school. Someone working full-time and attending school would be classified as only working full-time. Table 1 describes the distribution of men across these four activities by age group.
| Age | N | Working Full-Time | In School, Not Working Full-Time |
Working Part-Time, Not in School |
Not Working, Not in School |
High School Dropouts | |||||
| 17
|
437
|
51.7
|
16.7
|
9.4
|
22.2
|
| 18
|
820
|
48.0
|
20.6
|
10.0
|
21.3
|
| 19
|
1,031
|
55.6
|
13.2
|
9.5
|
21.7
|
| 20
|
1,070
|
62.7
|
9.1
|
7.0
|
21.3
|
| 21
|
1,069
|
66.7
|
6.6
|
6.1
|
20.6
|
| 22
|
1,056
|
70.6
|
5.7
|
6.4
|
17.3
|
| 23
|
1,039
|
71.3
|
5.1
|
5.3
|
18.4
|
| 24
|
1,014
|
73.1
|
3.3
|
7.0
|
16.6
|
| 25
|
992
|
77.9
|
2.2
|
5.8
|
14.2
|
| 26
|
898
|
77.5
|
1.7
|
5.3
|
15.5
|
| 27
|
661
|
79.6
|
1.1
|
4.9
|
14.4
|
| 28
|
412
|
75.2
|
3.0
|
3.3
|
18.6
|
| 29
|
193
|
79.9
|
1.1
|
2.6
|
16.4
|
High School Graduates | |||||
| 18
|
446
|
58.4
|
4.5
|
19.8
|
17.3
|
| 19
|
1,025
|
62.1
|
9.2
|
13.4
|
15.3
|
| 20
|
1,173
|
66.6
|
9.4
|
12.3
|
11.6
|
| 21
|
1,177
|
71.2
|
8.3
|
9.2
|
11.4
|
| 22
|
1,165
|
76.5
|
6.5
|
6.8
|
10.1
|
| 23
|
1,157
|
80.8
|
5.0
|
5.8
|
8.4
|
| 24
|
1,143
|
84.2
|
2.4
|
5.9
|
7.5
|
| 25
|
1,123
|
87.9
|
1.9
|
4.3
|
5.9
|
| 26
|
1,035
|
87.4
|
1.5
|
4.4
|
6.6
|
| 27
|
817
|
88.9
|
2.1
|
5.2
|
3.8
|
| 28
|
598
|
90.3
|
1.0
|
3.2
|
5.4
|
| 29
|
376
|
89.3
|
3.6
|
2.1
|
5.0
|
| 30
|
183
|
89.5
|
1.9
|
3.9
|
4.8
|
Some College | |||||
| 19
|
165
|
66.6
|
10.3
|
13.4
|
9.7
|
| 20
|
385
|
64.6
|
18.6
|
9.4
|
7.4
|
| 21
|
536
|
63.6
|
23.6
|
4.4
|
8.3
|
| 22
|
620
|
62.0
|
25.2
|
7.7
|
5.2
|
| 23
|
656
|
66.0
|
17.4
|
8.8
|
7.7
|
| 24
|
675
|
76.4
|
10.6
|
5.6
|
7.3
|
| 25
|
668
|
80.7
|
8.3
|
6.9
|
4.1
|
| 26
|
634
|
82.9
|
7.2
|
5.4
|
4.5
|
| 27
|
516
|
85.3
|
4.9
|
5.1
|
4.8
|
| 28
|
425
|
89.7
|
3.2
|
4.2
|
2.9
|
| 29
|
323
|
87.0
|
3.4
|
4.0
|
5.5
|
| 30
|
228
|
85.8
|
4.8
|
3.1
|
6.3
|
| 31
|
156
|
85.8
|
5.1
|
2.7
|
6.3
|
College Graduates | |||||
| 23
|
247
|
80.8
|
5.4
|
6.0
|
7.8
|
| 24
|
286
|
82.6
|
6.7
|
3.6
|
7.2
|
| 25
|
292
|
90.7
|
3.9
|
2.6
|
2.8
|
| 26
|
278
|
87.0
|
7.3
|
2.8
|
3.0
|
| 27
|
242
|
89.5
|
3.2
|
4.1
|
3.1
|
| 28
|
205
|
96.1
|
1.3
|
1.7
|
1.0
|
| 29
|
168
|
94.1
|
2.2
|
0.8
|
2.8
|
This table shows that the STW transition is smoother for men with more schooling. This can be seen by holding constant the number of years that have passed since school leaving. For instance, by age 25, or approximately five years after leaving school, 11% of those with some college education are either without work or working only part-time. Less than 3% of college graduates are in this situation at age 28, about five years after graduation. In contrast, 24% and 14% are the corresponding figures for high school dropouts and graduates at age 22 and 23, respectively. By this measure, high school graduates and those with some college are similar; both are doing substantially better than high school dropouts, but substantially less well than four-year college graduates.
Klerman and Karoly also subdivide the SLGs into three ethnic groups: black non-Hispanics, Hispanics, and white non-Hispanics. This last category includes Asians and other ethnic groups. Overall, comparisons between SLGs within ethnic groups show similar patterns of activity. Within given SLGs, blacks experience the highest rates of unemployment or partial employment and whites the least, with Hispanics generally in between. A relevant result is that black and Hispanic high school graduates experience employment rates which are close to those of high school dropouts presented in Table 1.
A similar analysis for women shows that the fraction of women neither working nor in school is higher than for men at any given age. Further, employment rates do not increase as quickly as for men, and there is less evidence of convergence among the three highest SLGs than among men.
To summarize, these "snapshots" of men's and women's early labor market careers show substantial unemployment or part-time employment among high school dropouts and black and Hispanic high school graduates. This situation improves by the late twenties but inactivity rates are still relatively high. The experience for high school graduates is substantially better, and similar to that of individuals with some college. College graduates experience the fastest and most complete transition to full-time employment.
Dynamic Analysis
The static analysis presented so far does not necessarily indicate a high degree of instability or churning in the youth labor market because the observed degree of inactivity could be due to a particular group of youths consistently being unemployed, underemployed, or out of the labor force--as opposed to large numbers of young people experiencing these conditions from time to time. Clarification of this issue requires a dynamic analysis showing how young adults move between various states of work and school, how many jobs they hold in the process, and how long these jobs last.
Previous research suggested a significant degree of churning (e.g., see Topel & Ward, 1992; Veum & Weiss, 1993). Klerman and Karoly consider two indicators. The first is the number of jobs that members of each SLG have held by a given age, and the second is the age at which members of each SLG attain various job tenure states. Table 2 presents the mean number of jobs held by male members of each SLG at a given age. To describe the amount of variance in youths' experience, it also presents the number of jobs held by youths at the median, the 25th percentile, and the 75th percentile of the jobs-held distribution. This gives an idea of how the "typical" individual's experience differs from that of his counterparts closer to the extremes of the jobs-held distribution.
| Age | High School Dropouts | High School Graduates | Some College | College Graduates | ||||||||||||
| Mean | 25th | 50th | 75th | Mean | 25th | 50th | 75th | Mean | 25th | 50th | 75th | Mean | 25th | 50th | 75th | |
| 17
|
0.5
|
0
|
0
|
1
|
0.0
|
0
|
0
|
0
|
0.0
|
0
|
0
|
0
|
0.0
|
0
|
0
|
0
|
| 18
|
1.4
|
0
|
1
|
2
|
0.4
|
0
|
0
|
1
|
0.0
|
0
|
0
|
0
|
0.0
|
0
|
0
|
0
|
| 19
|
2.5
|
1
|
2
|
4
|
1.5
|
1
|
1
|
2
|
0.3
|
0
|
0
|
0
|
0.0
|
0
|
0
|
0
|
| 20
|
3.5
|
2
|
3
|
5
|
2.5
|
1
|
2
|
3
|
1.0
|
0
|
1
|
2
|
0.0
|
0
|
0
|
0
|
| 21
|
4.3
|
2
|
4
|
6
|
3.3
|
2
|
3
|
4
|
1.8
|
0
|
2
|
3
|
0.0
|
0
|
0
|
0
|
| 22
|
5.2
|
3
|
5
|
7
|
4.0
|
2
|
4
|
5
|
2.7
|
1
|
2
|
4
|
0.4
|
0
|
0
|
1
|
| 23
|
5.9
|
3
|
5
|
8
|
4.7
|
2
|
4
|
6
|
3.6
|
2
|
3
|
5
|
1.3
|
1
|
1
|
2
|
| 24
|
6.7
|
4
|
6
|
9
|
5.3
|
3
|
5
|
7
|
4.4
|
2
|
4
|
6
|
2.0
|
1
|
2
|
3
|
| 25
|
7.3
|
4
|
7
|
10
|
5.9
|
3
|
5
|
8
|
5.0
|
3
|
5
|
7
|
2.5
|
1
|
2
|
3
|
| 26
|
8.0
|
5
|
7
|
10
|
6.3
|
3
|
6
|
9
|
5.6
|
3
|
5
|
8
|
2.9
|
1
|
2
|
4
|
| 27
|
8.6
|
5
|
8
|
10
|
6.7
|
4
|
6
|
9
|
6.2
|
3
|
5
|
8
|
3.3
|
2
|
3
|
4
|
| 28
|
8.6
|
5
|
8
|
10
|
7.2
|
4
|
6
|
10
|
6.7
|
4
|
6
|
9
|
3.6
|
2
|
3
|
5
|
| 29
|
8.8
|
6
|
9
|
10
|
7.6
|
4
|
7
|
10
|
7.0
|
4
|
6
|
9
|
3.8
|
2
|
3
|
5
|
| 30
|
-
|
-
|
-
|
-
|
7.8
|
4
|
7
|
10
|
7.3
|
4
|
7
|
10
|
-
|
-
|
-
|
-
|
| 31
|
-
|
-
|
-
|
-
|
-
|
-
|
-
|
-
|
7.7
|
4
|
7
|
10
|
-
|
-
|
-
|
-
|
In general, the mean numbers of jobs shown in the table is smaller than those found by Veum and Weiss (1993). This difference is possibly due to the fact that Veum and Weiss counted jobs starting at age 18, while Klerman and Karoly begin counting at the time a young person leaves school.[4] Nevertheless, the table confirms young men tend to hold a large number of jobs in years immediately following school leaving. For instance, a high school dropout at the median of the number of jobs-held distribution will have held six jobs by age 24 and nine jobs by age 29. There is also significant variance in experience within SLGs. At the 75th percentile, the number of jobs held is usually at least twice the number at the 25th percentile.
Comparing different SLGs five years after the typical age of school leaving (high school dropouts at age 22, high school graduates at age 23, individuals with some college at age 25, and college graduates at age 28) shows the mean numbers of jobs held are 5.2, 4.7, 5.0, and 3.6, respectively. Four-year college graduates experience less job-changing than the other three groups. Analyzing racial or ethnic groups separately shows a similar pattern within each group.
To explore the dynamic component of the STW transition further, Klerman and Karoly consider three concepts of job stability, labeled L, E, and C:
Table 3 presents the percentage of men, by SLG, with job tenure of 1, 2, and 3 years under each of these three measures (in this case, lower numbers indicate greater degrees of labor market instability).
| Age | N | 1
Year
|
2
Years
|
3
Years
|
||||||
| L
|
E
|
C
|
L
|
E
|
C
|
L
|
E
|
C
| ||
| 16
|
1,132
|
0.0
|
1.8
|
0.0
|
0.0
|
1.0
|
0.0
|
0.0
|
0.5
|
0.0
|
| 17
|
1,122
|
2.1
|
10.2
|
2.1
|
0.0
|
5.2
|
0.0
|
0.0
|
2.4
|
0.0
|
| 18
|
1,106
|
11.9
|
23.0
|
9.0
|
1.2
|
11.9
|
1.2
|
0.0
|
6.8
|
0.0
|
| 19
|
1,093
|
27.5
|
39.7
|
15.0
|
5.7
|
25.7
|
3.5
|
0.8
|
17.3
|
0.6
|
| 20
|
1,082
|
49.0
|
49.1
|
28.5
|
13.7
|
34.5
|
8.0
|
2.8
|
24.2
|
1.3
|
| 21
|
1,074
|
63.2
|
52.1
|
34.5
|
28.6
|
38.5
|
18.2
|
7.5
|
29.7
|
5.0
|
| 22
|
1,059
|
72.2
|
57.8
|
36.4
|
39.1
|
43.4
|
22.8
|
18.4
|
35.4
|
13.0
|
| 23
|
1,040
|
77.4
|
61.1
|
40.7
|
46.6
|
47.7
|
23.9
|
26.0
|
38.5
|
15.0
|
| 24
|
1,017
|
81.9
|
64.3
|
44.3
|
54.1
|
49.3
|
27.2
|
32.4
|
40.9
|
17.0
|
| 25
|
994
|
86.4
|
67.0
|
46.3
|
61.4
|
50.3
|
30.5
|
40.1
|
40.5
|
19.9
|
| 26
|
902
|
90.0
|
65.8
|
46.6
|
66.2
|
51.0
|
31.3
|
47.1
|
42.0
|
22.8
|
| 27
|
662
|
91.8
|
70.2
|
47.3
|
70.3
|
57.2
|
33.4
|
52.7
|
46.8
|
25.3
|
| 28
|
413
|
93.7
|
68.1
|
50.3
|
73.7
|
58.4
|
33.2
|
56.8
|
48.5
|
25.5
|
| 29
|
194
|
93.8
|
72.0
|
58.4
|
75.6
|
43.5
|
60.8
|
27.2
| ||
| 16
|
1,227
|
0.0
|
0.0
|
0.0
|
0.0
|
0.0
|
0.0
|
0.0
|
0.0
|
0.0
|
| 17
|
1,225
|
0.0
|
0.1
|
0.0
|
0.0
|
0.1
|
0.0
|
0.0
|
0.1
|
0.0
|
| 18
|
1,217
|
0.1
|
11.5
|
0.1
|
0.0
|
5.7
|
0.0
|
0.0
|
3.7
|
0.0
|
| 19
|
1,202
|
14.2
|
44.2
|
11.6
|
0.1
|
26.0
|
0.1
|
0.0
|
18.5
|
0.0
|
| 20
|
1,193
|
47.3
|
59.1
|
34.2
|
7.3
|
40.7
|
6.5
|
0.1
|
29.8
|
0.1
|
| 21
|
1,179
|
67.4
|
63.7
|
42.4
|
27.4
|
49.0
|
21.5
|
5.2
|
39.5
|
4.3
|
| 22
|
1,168
|
77.1
|
66.1
|
45.3
|
41.8
|
54.8
|
28.0
|
19.7
|
46.0
|
15.5
|
| 23
|
1,160
|
84.9
|
75.2
|
52.2
|
53.4
|
60.4
|
34.1
|
30.5
|
50.0
|
21.4
|
| 24
|
1,143
|
90.5
|
74.5
|
54.7
|
63.7
|
62.7
|
37.5
|
41.3
|
53.0
|
25.9
|
| 25
|
1,125
|
94.3
|
78.7
|
58.1
|
70.9
|
67.1
|
41.4
|
49.9
|
58.6
|
30.7
|
| 26
|
1,035
|
95.9
|
81.0
|
61.3
|
77.4
|
70.8
|
45.4
|
57.0
|
63.5
|
33.8
|
| 27
|
817
|
98.2
|
86.4
|
66.8
|
83.4
|
76.0
|
51.1
|
63.6
|
67.3
|
39.1
|
| 28
|
598
|
98.6
|
85.2
|
68.8
|
87.1
|
76.2
|
53.4
|
70.1
|
67.1
|
42.9
|
| 29
|
376
|
99.0
|
85.6
|
66.8
|
89.0
|
76.2
|
52.0
|
75.7
|
68.7
|
44.3
|
| 30
|
183
|
99.2
|
91.2
|
73.3
|
91.7
|
56.1
|
79.5
|
45.8
| ||
| 16
|
732
|
0.0
|
0.0
|
0.0
|
0.0
|
0.0
|
0.0
|
0.0
|
0.0
|
0.0
|
| 17
|
729
|
0.0
|
0.0
|
0.0
|
0.0
|
0.0
|
0.0
|
0.0
|
0.0
|
0.0
|
| 18
|
727
|
0.0
|
0.5
|
0.0
|
0.0
|
0.0
|
0.0
|
0.0
|
0.0
|
0.0
|
| 19
|
724
|
0.6
|
7.4
|
0.4
|
0.0
|
4.2
|
0.0
|
0.0
|
3.5
|
0.0
|
| 20
|
721
|
8.4
|
23.0
|
6.3
|
0.0
|
15.0
|
0.0
|
0.0
|
9.7
|
0.0
|
| 21
|
707
|
24.8
|
35.5
|
17.5
|
4.5
|
25.4
|
4.1
|
0.0
|
17.0
|
0.0
|
| 22
|
700
|
41.0
|
48.1
|
25.6
|
15.6
|
34.3
|
11.8
|
3.6
|
27.2
|
3.4
|
| 23
|
693
|
57.3
|
57.5
|
37.4
|
26.8
|
41.4
|
17.7
|
10.2
|
33.7
|
8.3
|
| 24
|
687
|
69.1
|
67.6
|
42.0
|
38.6
|
51.4
|
25.2
|
17.9
|
42.7
|
12.9
|
| 25
|
672
|
81.2
|
76.4
|
50.3
|
48.0
|
60.1
|
29.9
|
28.8
|
50.8
|
20.0
|
| 26
|
637
|
89.8
|
81.3
|
59.0
|
61.0
|
64.6
|
36.5
|
36.3
|
54.8
|
21.3
|
| 27
|
517
|
94.3
|
81.4
|
59.6
|
71.1
|
69.6
|
43.7
|
49.4
|
59.2
|
28.9
|
| 28
|
425
|
96.7
|
86.0
|
64.7
|
75.2
|
73.1
|
45.8
|
58.8
|
63.5
|
34.0
|
| 29
|
323
|
98.9
|
84.5
|
66.8
|
81.6
|
75.0
|
48.8
|
63.5
|
67.8
|
36.6
|
| 30
|
228
|
98.9
|
80.8
|
64.4
|
86.9
|
72.3
|
52.2
|
70.2
|
66.1
|
38.3
|
| 31
|
156
|
99.1
|
69.5
|
87.5
|
53.6
|
76.8
|
45.2
| |||
| 16
|
309
|
0.0
|
0.0
|
0.0
|
0.0
|
0.0
|
0.0
|
0.0
|
0.0
|
0.0
|
| 17
|
307
|
0.0
|
0.0
|
0.0
|
0.0
|
0.0
|
0.0
|
0.0
|
0.0
|
0.0
|
| 18
|
307
|
0.0
|
0.0
|
0.0
|
0.0
|
0.0
|
0.0
|
0.0
|
0.0
|
0.0
|
| 19
|
306
|
0.0
|
0.0
|
0.0
|
0.0
|
0.0
|
0.0
|
0.0
|
0.0
|
0.0
|
| 20
|
304
|
0.0
|
0.0
|
0.0
|
0.0
|
0.0
|
0.0
|
0.0
|
0.0
|
0.0
|
| 21
|
303
|
0.0
|
0.2
|
0.0
|
0.0
|
0.0
|
0.0
|
0.0
|
0.0
|
0.0
|
| 22
|
301
|
0.2
|
18.9
|
0.0
|
0.0
|
14.2
|
0.0
|
0.0
|
12.4
|
0.0
|
| 23
|
297
|
20.8
|
56.5
|
18.3
|
0.0
|
38.2
|
0.0
|
0.0
|
33.0
|
0.0
|
| 24
|
295
|
61.1
|
73.2
|
47.6
|
16.0
|
59.2
|
15.8
|
0.0
|
48.8
|
0.0
|
| 25
|
294
|
78.8
|
85.3
|
58.9
|
40.9
|
71.3
|
35.9
|
13.6
|
60.4
|
12.6
|
| 26
|
279
|
90.4
|
85.2
|
68.9
|
60.7
|
75.3
|
48.6
|
35.4
|
66.7
|
31.5
|
| 27
|
242
|
94.9
|
89.7
|
70.5
|
73.4
|
79.8
|
55.4
|
50.5
|
73.0
|
43.1
|
| 28
|
205
|
97.4
|
84.3
|
68.9
|
81.3
|
77.2
|
54.0
|
60.6
|
72.2
|
43.5
|
| 29
|
169
|
98.2
|
88.8
|
68.2
|
84.1
|
81.1
|
59.0
|
68.0
|
79.5
|
49.1
|
By all three measures, the table shows that four-year college graduates do substantially better than high school graduates or individuals with some college, who in turn do substantially better than high school dropouts. For example, making a comparison about five years after each SLG has left school, 39.1% of high school dropouts at age 22 have held at least one job that lasted two years or more. Among high school graduates at age 23, the percentage is a considerably higher 53.4%. Among the "some college" group at age 25, the figure is 48.0%. And for the four-year college graduates at age 28 it is much higher at 81.3%.
To summarize, both static and dynamic measures of employment stability show that college graduates do considerably better than those with only some college,[5] whose experience is similar to that of high school graduates. High school dropouts have the most difficult STW transition, doing substantially worse than high school graduates.
These findings have two direct implications for designing STW systems and programs:
If employment instability among recent school leavers were only a temporary phenomenon, it would not be a great concern for policy. A recent analysis of the NLSY data by Gardecki and Neumark (1995), however, reveals a strong correlation between employment experience in the first year or two after leaving school and in the subsequent three or four years. Although this correlation does not prove that early instability causes later instability, it does indicate that early instability fails to produce more stable job matches, at least within three or four years.
The relation between early employment and subsequent experience can be studied using the probability tree presented in Table 4.
Months Worked in Each Year, Any Job, First Five Post-Schooling Years
| Year 1 | Year 2 | Year 3 | Year 4 | Year 5 | |
| Months >9 .93 | |||||
| Months >9 .89 | |||||
| Months <9 .07 | |||||
| Months >9 .91 | |||||
| Months >9 .65 | |||||
| Months <9 .11 | |||||
| Months <9 .35 | |||||
| Months >9 .84 | |||||
| Months >9 .92 | |||||
| Months >9 .59 | |||||
| Months <9 .08 | |||||
| Months <9 .09 | |||||
| Months >9 .48 | |||||
| Months <9 .41 | |||||
| Months <9 .52 | |||||
| Months >9 .55 | |||||
| Months >9 .84 | |||||
| Months >9 .80 | |||||
| Months <9 .16 | |||||
| Months >9 .63 | |||||
| Months >9 .37 | |||||
| Months <9 .20 | |||||
| Months <9 .63 | |||||
| Months <9 .16 | |||||
| Months >9 .69 | |||||
| Months >9 .37 | |||||
| Months <9 .31 | |||||
| Months <9 .37 | |||||
| Months >9 .32 | |||||
| Months <9 .63 | |||||
| Months <9 .68 | |||||
| Months >9 .89 | |||||
| Months >9 .88 | |||||
| Months <9 .11 | |||||
| Months >9 .82 | |||||
| Months >9 .46 | |||||
| Months <9 .12 | |||||
| Months <9 .54 | |||||
| Months >9 .52 | |||||
| Months >9 .90 | |||||
| Months >9 .36 | |||||
| Months <9 .10 | |||||
| Months <9 .18 | |||||
| Months >9 .27 | |||||
| Months <9 .64 | |||||
| Months <9 .73 | |||||
| Months <9 .45 | |||||
| Months >9 .82 | |||||
| Months >9 .75 | |||||
| Months <9 .18 | |||||
| Months >9 .33 | |||||
| Months >9 .32 | |||||
| Months <9 .25 | |||||
| Months <9 .68 | |||||
| Months <9 .48 | |||||
| Months >9 .70 | |||||
| Months >9 .26 | |||||
| Months <9 .30 | |||||
| Months <9 .67 | |||||
| Months >9 .22 | |||||
| Months <9 .74 | |||||
| Months <9 .78 | |||||
The first column presents a breakdown of individuals' work experience in their first year after high school. It shows that 55% of students had more than nine months work experience that year, while the remaining 45% had less. The Year 2 column indicates that 84% of the first group, but only 52% of the second, were able to secure employment for more than nine months in the subsequent year. This suggests that initial bad experiences can have significant negative effects on subsequent ones.
The strong correlation between experience in the first two years and in subsequent years is evident in Table 5, which presents the probability of reporting zero, one, two, three, or four subsequent years of working for a period greater than nine months after various experiences in the first year or two.
| Probabilities
of Full-Year Work in Additional . . .
|
|||||
| No
Years
|
One
Year
|
Two
Years
|
Three
Years
|
Four
Years
| |
| Full-year
work in first year |
0.03 | 0.05 | 0.08 | 0.21 | 0.63 |
| No
full-year work in first year |
0.19 | 0.15 | 0.14 | 0.19 | 0.33 |
| Full-year
work in first and second year |
0.02 | 0.06 | 0.18 | 0.75 | |
| No
full-year work in first year; full-year work in second year |
0.08 | 0.09 | 0.19 | 0.64 | |
| No
full-year work in first or second year |
0.39 | 0.22 | 0.19 | 0.20 | |
To illustrate interpretation of this table, one can compare the first and second rows in which individuals are classified by whether or not they worked more than nine months the first year. The first column shows that those who worked less than nine months in the first year had a significantly higher probability (0.19 vs. 0.03) of not working more than nine months in any of the subsequent years, and a much lower probability (0.33 vs. 0.63) of working more than nine months in all four subsequent years. The next three rows show even larger differences after two initial years of not working nine months or more. This data suggests that, although adverse initial experiences do not doom young people to bad experiences forever after, the correlation with experience a few years later is strong.
Finding a job is one thing; keeping it is another. Gardecki and Neumark also analyze year-to-year changes in whether young people worked four or more quarters for the same employer. They construct the probability diagram in Table 6.
| Year 1 | Year 2 | Year 3 | Year 4 | Year 5 | |
| Tenure >4 .95 | |||||
| Tenure >4 .92 | |||||
| Tenure <4 .05 | |||||
| Tenure >4 .90 | |||||
| Tenure >4 .56 | |||||
| Tenure <4 .08 | |||||
| Tenure <4 .44 | |||||
| Tenure >4 .84 | |||||
| Tenure >4 .88 | |||||
| Tenure >4 .49 | |||||
| Tenure <4 .12 | |||||
| Tenure <4 .10 | |||||
| Tenure >4 .30 | |||||
| Tenure <4 .51 | |||||
| Tenure <4 .70 | |||||
| Tenure >4 .47 | |||||
| Tenure >4 .92 | |||||
| Tenure >4 .96 | |||||
| Tenure <4 .08 | |||||
| Tenure >4 .53 | |||||
| Tenure >4 .54 | |||||
| Tenure <4 .04 | |||||
| Tenure <4 .46 | |||||
| Tenure <4 .16 | |||||
| Tenure >4 .89 | |||||
| Tenure >4 .46 | |||||
| Tenure <4 .11 | |||||
| Tenure <4 .47 | |||||
| Tenure >4 .44 | |||||
| Tenure <4 .54 | |||||
| Tenure <4 .56 | |||||
| Tenure >4 .92 | |||||
| Tenure >4 .93 | |||||
| Tenure <4 .08 | |||||
| Tenure >4 .90 | |||||
| Tenure >4 .36 | |||||
| Tenure <4 .11 | |||||
| Tenure <4 .64 | |||||
| Tenure <4 .41 | |||||
| Tenure >4 .78 | |||||
| Tenure >4 .42 | |||||
| Tenure <4 .22 | |||||
| Tenure <4 .10 | |||||
| Tenure >4 .42 | |||||
| Tenure <4 .58 | |||||
| Tenure <4 .58 | |||||
| Tenure <4 .53 | |||||
| Tenure >4 .80 | |||||
| Tenure >4 .85 | |||||
| Tenure <4 .20 | |||||
| Tenure >4 .36 | |||||
| Tenure >4 .17 | |||||
| Tenure <4 .15 | |||||
| Tenure <4 .83 | |||||
| Tenure <4 .59 | |||||
| Tenure >4 .86 | |||||
| Tenure >4 .31 | |||||
| Tenure <4 .14 | |||||
| Tenure <4 .64 | |||||
| Tenure >4 .22 | |||||
| Tenure <4 .69 | |||||
| Tenure <4 .78 | |||||
The overall results are quite similar to those found for experience in the sense that there appears to be a strong relationship between job tenure in one year and the next. Those who do not have a long-tenure job in their first post-schooling year are much less likely to have long-tenure jobs in each of the following four years.
As before, once an individual switches to working continuously for an employer, the difference between that worker and a worker with a longer history of continuous work with an employer is much smaller. Nevertheless, as with experience, the differences never disappear, and the strong correlation between performance in successive years translates into large subsequent differences. Table 7, which is analogous to GN4, presents the corresponding information.
| Probabilities of Additional . . . | |||||
| No Years | One Year | Two Years | Three Years | Four Years | |
| Tenure
>4 in first year |
0.02 | 0.06 | 0.09 | 0.17 | 0.66 |
| Tenure
<4 in first year |
0.20 | 0.12 | 0.18 | 0.19 | 0.30 |
| Tenure
>4 in first year and second year |
0.04 | 0.06 | 0.12 | 0.79 | |
| Tenure
<4 in first year; >4 in second year |
0.03 | 0.09 | 0.13 | 0.74 | |
| Tenure
<4 in first year or second year |
0.34 | 0.17 | 0.24 | 0.24 | |
Once again considering the first two rows, individuals who do not have a long-tenure job in their first year out of school have a higher probability of never having long tenure in any of the following four years (0.20 vs. 0.02), and a smaller probability (0.35 vs. 0.66) of long tenure in all four subsequent years.
Multivariate analysis that controls for observed characteristics of individuals does not qualitatively change the results from those reached by observation of the probability trees, either for experience or tenure. Gardecki and Neumark find that individuals can recover from adverse labor market experiences, but that even when the analysis controls for ability and other traits, previous experience matters.
In sum, the Gardecki and Neumark study shows a strong correlation between getting or keeping a job in the first two years out of school, and getting or keeping a job in the three or four years after that. Coming on top of previous research that showed early work experience correlated with subsequent wages (D'Amico & Maxwell, 1990; Ellwood, 1992; Lynch, 1989; Meyer & Wise, 1992), these results suggest that stabilizing employment for young people right out of high school may help them later on. Smoothing the initial STW transition may indeed help young people get onto a higher trajectory of employment and earnings. Along with the reconfirmation by the Klerman and Karoly (1995a) study that finishing high school--and, beyond that, a bachelor's degree--is associated with more successful transition to work, these findings support policies and practices aimed at encouraging students to complete high school, enabling them to go on toward a bachelor's degree, and facilitating the initial transition from school to stable employment.
[3] The authors do not include this last group in any of the tables reproduced below because the samples are too small.
[4] Freeman and Medoff (1982) provide an overview of why rates of youth labor force activity differ across surveys.
[5] The "some college" group in Klerman and Karoly's analysis is very diverse. It is possible that certain subgroups, in particular holders of associate degrees, may in fact do better than the others, but an NLSY sample is too small for testing this.
[6] Previous research indicates that the high school diploma is not merely a signal for other characteristics, but does reflect learning that results in higher wages and lower unemployment (Stern, Paik, Catterall, & Nakata, 1989).