• Static Analysis: Activity Status
• Dynamic Analysis: Job Tenure
• Discussion

The preceding three sections have analyzed data from the National Longitudinal Survey-Youth. The analysis exploited the continuous employment histories in the NLS-Y to characterize the school-to-work transition in terms of labor market and school enrollment status at a point in time, the number of jobs held, and the age at entering 1-, 2-, and 3-year jobs. No other survey offers this rich level of detail with respect to employment histories.

The drawback of the NLS-Y data is that they refer only to a specific cohort of youth: the NLS-Y sample was drawn from individuals 14 to 21 in late 1978. As we noted in Section 2, our requirement that we observe the initial time of school leaving within the period covered by the continuous employment histories caused us to drop the older respondents. For example, our dropout sample is composed of respondents who were 14 to 17 years old at the time of the first interview; our high school graduate and some college sample includes those aged 14 to 19; our college sample is essentially all of the sampled individuals.

Thus, our results refer primarily to the youth cohort born in the two years before and the two years after 1960. There is considerable interest (and concern) about whether, compared with earlier cohorts, these cohorts have had a harder time making the transition from school to work. Similarly, there are concerns that the transition has become even more difficult since the NLS-Y cohort entered the labor market in the early 1980s. In this section, we try to place the experiences of the NLS-Y cohort of men in the context of earlier and later birth cohorts.

Unfortunately, data comparable to the NLS-Y do not exist for earlier or later cohorts. Instead, we used less comprehensive data from two supplements to the Current Population Survey (CPS). The CPS is a monthly survey of approximately 55,000 households conducted by the Bureau of the Census. The primary purpose of the survey is to provide the official monthly unemployment statistics and other labor market data. For each sampled individual aged 16 and older, the CPS collects basic demographic information (e.g., gender, race/ethnicity, age, educational attainment in years) and detailed information on labor market status for a specific reference week (e.g., labor force status, hours worked, job search efforts). Consistent with the focus in the CPS on measuring labor market status, individuals who do any work are coded as workers, regardless of whether they are enrolled in school.

In addition to the core labor force information, supplemental questions are included on both a regular and an irregular basis. A supplement to the October survey, available annually since 1968, includes detailed information on contemporaneous school attendance for respondents. Combining these supplemental data with the basic CPS information on labor market status, we constructed a time series of static activity status measures similar to those used in Section 3 of this report for young men in various age groups. In addition, in 1973, 1981, 1983, 1987, and 1991, a supplement to the January survey collected information on tenure on the current job. We exploited these data to evaluate changes over time in job tenure for various age groups.

Neither of these CPS data sources permit analyses strictly comparable with those we have conducted using the NLS-Y, our preferred data source. First, the schooling groups we defined using the CPS refer to current educational attainment, not attainment at the time of school leaving (as in our school-leaving groups using the NLS-Y). For example, NLS-Y respondents who were still in college as of the last point at which we observed them (and who have been in school continuously) would not be included in the analyses reported in the preceding sections. In the CPS analysis reported here, they would be included in the some college group. Second, the January CPS measures tenure on the current job. We noted in Section 4 that tenure on the current job provides an overly negative perspective on the transition to stable employment. Nevertheless, examining the time-series patterns in these static and dynamic measures provides useful insights into the likely school-to-work experiences of youth cohorts that entered the labor market between the late 1960s and the early 1990s.

Static Analysis: Activity Status

Since 1968, as part of the October survey, the CPS has included a battery of questions about current school attendance. Combining responses to these questions with responses to the basic labor market information, we constructed static measures of major activity status, assigning individuals hierarchically to the following four mutually exclusive categories: working full time, in school, working part time, and neither working nor in school. These four categories and the hierarchy follow the methodology used for the NLS-Y in Section 3. For example, individuals in school and working full-time (35 or more hours per week) are classified as working full time. Individuals in school and working part time (less than 35 hours per week) are classified as being in school.

Figures 5.1, 5.2, and 5.3 present time trends in these static measures separately for men aged 19 to 21, 23 to 25, and 27 to 29, respectively. Each figure contains four plots showing the percentage of individuals in each of four activity statuses, by year. In each plot (except that for the youngest age group, in which there are few young men who have graduated from college), we show the percentages separately for four education groups: high school dropouts, high school graduates, those with some college, and college graduates.

Across the four educational groups, the patterns reported in the figures show the expected relationships, once we account for the fact that many men in the younger age groups are still in school. In the youngest group (Figure 5.1), for example, most of the some college group are still in school. Thus, the some college group has the highest school attendance rates and the lowest rates in the other three categories. Full-time work, in contrast, is more prevalent among high school graduates compared with high school dropouts. High school dropouts, in turn, are slightly more likely (in most years) to work part time than are high school graduates. Finally, high school dropouts are much more likely than high school graduates to be neither in school nor working.

In the intermediate age group (Figure 5.2), after those progressing continuously through school have completed college, the relative patterns of the schooling groups are even easier to interpret. Those with more education are less likely to be in the residual category (neither working nor in school) and slightly less likely to be working part time. The fact that many of the some college and college graduate groups are still in school (about 20 percent) offsets the strong differentials in the fraction in the residual category. Consequently, there is no clear ordering across the years among schooling groups in the fraction engaged in full-time work.

Finally, turning to the late twenties (Figure 5.3), we can see that school attendance has essentially ended (under 10 percent even in the some college and college graduate groups). Likewise, a very small fraction (less than 10 percent) is classified as working part time, even for high school dropouts for whom part-time work is most common. In most years, well over 90 percent of men aged 27 to 29 are working full time. It is difficult to distinguish the rates for college graduates, some college, and high school graduates. High school dropouts are the exception, because they are less likely to be working full time compared with the other three schooling groups. The pattern is similar, although less pronounced, for the neither working nor in school outcome. Again, high school dropouts are the outlier, with much higher rates in each year compared with more educated men. For the other groups, there is a nearly stable ordering; the fraction in the nonactivity state declines as years of schooling increase.

Figures 5.1 through 5.3 also reveal several pronounced changes in the static measure of activity status that occur over this nearly 25-year period. Both absolutely and relative to their peers in higher schooling groups, the situation of high school dropouts has clearly deteriorated. For each of the age groups examined, the fraction engaged in full-time work has fallen approximately 20 to 25 percentage points between 1968 and 1991. For the 23 to 25 year olds, there has actually been a crossover. Through about 1980, dropouts were more likely to be working full time than were those with some college or college graduates. But by the end of the period, they were less likely to be working full time than college graduates and about as likely to be in this category as those with some college.

Almost all the decline in the fraction of high school dropouts engaged in full-time employment has been matched by a corresponding increase in the fraction not engaged in either work or school. There has been little change, in contrast, in the fraction in school or working part time. Among the other education groups there is a small increase (at most a few percentage points) in the fraction neither working nor in school, matched by a decline in the fraction working full time. These modest increases do not come close to the more than 15-percentage-point gain evident for high school dropouts.

Most of the change in activity status for dropouts and the other three schooling groups is concentrated in the period before 1980. While there is considerable cyclical fluctuation thereafter, the trend is less pronounced through the 1980s. For this reason, the school-to-work transition patterns we investigated for the NLS-Y cohort who entered the labor market in the early 1980s are likely to reflect the experiences of later cohorts who entered the labor market throughout the next decade. Conversely, the CPS data also suggest, particularly for high school dropouts, that cohorts of youth who entered the labor market in the late 1960s or the 1970s may have had a better school-to-work transition experience than the NLS-Y cohort.

Moreover, these results are broadly consistent with other trends in the labor market experiences of youths and young adults since 1968. Other authors working with cross-sectional data on employment patterns have found similar patterns, for dropouts, of absolute and relative decreases in full-time employment and increases in not working or not being in school (e.g., Juhn, 1992). The employment patterns evident for this period have been accompanied, as well, by significant changes in the wage structure. Wages for high school dropouts declined both absolutely and relative to those for more educated workers (Klerman and Karoly, 1994).

Dynamic Analysis: Job Tenure

We turn now to more dynamic characterizations of the labor market transitions of young men during the 1970s and the 1980s. Our analysis in Section 4 of this report exploits the longitudinal employment histories of the NLS-Y cohort to identify the age at which young men first entered into jobs that would eventually last 1, 2, and 3 years. Such longitudinal data do not exist for other cohorts. However, in some years (1973, 1981, 1983, 1987, and 1991) the January CPS has included questions concerning tenure in the current job.[24]

At the end of Section 4, we argue that current job tenure is not the ideal statistic for evaluating the extent of the difficulty in the transition to stable employment. Because current job tenure is always less than completed job tenure and because short jobs may be a beneficial aspect of a career involving longer jobs, current job tenure gives too pessimistic a view of the level of difficulty in the transition to stable employment. Nevertheless, we must rely on this measure to evaluate the likely changes over time in the transition to stable employment. If, as seems reasonable to expect, all three job-tenure measures discussed in Section 4 move together through time, the CPS data allow us to assess variation across youth cohorts in the transition to stable employment.

Figures 5.4, 5.5, and 5.6 present the results from the January CPS. Following the pattern of the earlier figures, results are reported separately for men in three age groups: 19 to 21, 23 to 25, and 27 to 29, respectively. Within each figure, we present four plots showing the percentage of men with 1, 2, 3, and 4 years of tenure on the current job. These percentages are among all men in the age and schooling group, regardless of whether the individual is working (nonworking men are considered to have zero tenure). As above, results are shown separately for the four schooling groups, defined by completed years of schooling as of the interview date. (Figure 5.4 is the exception; college graduates were omitted from it because samples are very small in the 19 to 21 year age group.)

Turning to the figures, consider first the basic patterns by age and schooling groups. Consistent with the results reported for the NLS-Y, at the youngest ages (Figure 5.4), high school graduates are the most likely to have been in a job at least 1 year (about 40 percent). By comparison, the some college and high school dropout groups have similar rates (about 25 to 30 percent). The differential between the three schooling groups narrows as the measure of job tenure lengthens.

For the middle age group (Figure 5.5), the fraction of those in the some college group with a given level of tenure is typically below the fraction for high school graduates but above the rate for high school dropouts. College graduates lag behind those with less schooling because they typically have entered the labor market at older ages. Even so, college graduates do better even than high school dropouts by the early 1980s for shorter tenures (1 or 2 years).

By the oldest age group (Figure 5.6), the percentage of young men with 1 year of job tenure--high school graduates, some college, and college graduates--looks similar (about 70 percent). Again, the slow start and then convergence of some college and college graduates is probably due to their later entrance into the labor market. Similar orderings are apparent for longer-tenure definitions. For each tenure measure, high school dropouts at these older ages are less likely to have a given level of tenure compared with the other three groups, especially by the early 1980s.

Beyond the age and education group differences, the January CPS data provide time-series information that allows us to judge the representativeness of the
NLS-Y cohort that entered the labor market in the early 1980s. Overall, the plots depict relative stability through time in the fraction with a given level of job tenure, although some modest changes did occur over the period covered by the CPS data. For each age group, the fraction with shorter job tenures (1 or 2 years) rose slightly or remained unchanged between 1973 and 1981. During the same period, there was a tendency for the fraction with longer job tenures to decline. The early 1980s, a period marked by back-to-back recessions, led to declines in short-term job tenure (1 year), particularly for high school dropouts in each age group. At the same time, the fraction with longer job tenure (2 to 4 years) increased in the early 1980s, most notably for older college graduates. The changes among 1983, 1987, and 1991 are relatively modest by comparison with the earlier shifts.

With the exception of high school dropouts, at each age the job-tenure rates are similar between 1973 and 1991. Dropouts, by comparison, stand out with an overall decline in the fraction with each measure of job tenure between 1973 and 1991. In almost every figure, a smaller fraction of dropouts are in a job that has lasted 1, 2, 3, or 4 years compared with each of the other schooling groups, a pattern that did not exist in 1973, particularly for those 19 to 21 (Figure 5.3) and 23 to 25 (Figure 5.4). Thus, again, dropouts experienced declines in these measures of job stability both absolutely and relative to more educated young men.

We conclude, as we did from the October CPS data, that results for static and dynamic labor market experiences based on the NLS-Y cohort are likely to closely reflect the experience of both earlier and later cohorts for men who entered the labor market with at least a high school degree. In contrast, high school dropouts who entered the labor market before the NLS-Y cohort probably experienced less difficulty in the transition from school to work; the reverse would be true for more recent labor market entrants.

Discussion

In this section, we have used supplemental data from the October and January CPS to explore the robustness of the school-to-work analysis based on the NLS-Y data. For the three highest schooling groups, the picture that emerges is one of relative stability in the early labor market experience. The fraction of high school graduates, those with some college, and college graduates engaged in work or school between the ages of 19 and 29 changed little during the 1970s and 1980s. Likewise, job-tenure distributions for young men in the mid-1980s (the period covered by the NLS-Y data) look similar to those for the early 1970s and for the early 1990s. Thus, we conclude that our NLS-Y-based characterization of the transition to stable employment is likely to reflect the experiences of earlier and later cohorts of U.S. youth.

The lowest schooling group, high school dropouts, is the exception to this general picture of stability through time. Compared with earlier cohorts, young dropouts today are less likely to be working full time and more likely to be neither working nor in school. At the same time, the job-tenure distribution for these less educated youth appears to have worsened through the 1970s and 1980s; the largest effect is at older ages and longer-tenure points. These results imply that our NLS-Y-based characterization of the transition to stable employment for high school dropouts is probably too pessimistic for the early 1970s and too optimistic for the early 1990s.

[24] These data have been used before to analyze various aspects of job tenure: e.g., Hall (1982), Ureta (1992).