Printer Friendly

Intersegment and racial differences in nonemployment mobility.

This paper is an exploration of intersegment and racial differences in job-leaving experience, which links ideas found in mobility studies of the labor market segmentation school and the literature which describes unemployment as a dynamic process. Findings suggest that there are significant differences in nonemployment mobility patterns across the segment structure, lending support to researchers who view the labor market as segmented. In addition, after controlling for segmentation, racial differences in nonemployment mobility are revealed, suggesting that race, too, is an important theoretical category that cannot be ignored.

1. Introduction

Previous work on mobility in the labor market segmentation framework has focused attention primarily on mobility between labor segments (e.g. Rosenberg, 1980 and Rumberger and Carnoy, 1980). Transitions of workers from the various labor segments to a nonemployment state, however, are not incorporated into these analyses. Labor market segmentation literature does suggest that secondary segment workers may be significantly more unstable in their jobs than workers in the primary segments (e.g. Harrison, 1972). Instability may be manifested as a higher probability of transition to nonemployment for workers in secondary segment occupations in comparison to their primary segment counterparts.

More recently, research in dynamic explanations of unemployment has also been concerned with job-leaving behavior. Coleman (1989), for example, established that the primary determinant of the unemployment rate is the frequency of occurrence of unemployment spells (transitions from employment to unemployment). Duration of these spells, on the other hand, is found to be a less important determinant of the overall unemployment rate (for other research in this area see also Clark and Summers 1979, 1982, Marston, 1976 and Moser, 1986). Coleman's findings imply that it may be more useful to direct research into causes of unemployment that focus on job-leaving behavior. In this literature, however, there has been little attempt to explicitly incorporate notions of labor market segmentation.

This paper links the idea found in the mobility models of the labor market segmentation literature with those found in dynamic explanations of unemployment in an attempt to address three sets of labor market issues. First, if labor market segmentation theorists are correct in their claim that distinct and qualitatively different segments exist in the labor market, significant differences in nonemployment mobility patterns across the segment structure may be expected.(1) The observation of differential patterns of nonemployment mobility, then, would tend to support the agenda of researchers who view the labor market as segmented.

Second, if transitions to nonemployment play a significant role in the determination of the unemployment rate, then it is important for labor market segmentation researchers to gain a clearer insight into how individual characteristics, labor market conditions and labor market structure enter into the determination of nonemployment mobility. It may also be useful for analysts in the dynamic unemployment tradition to recognize the potential contribution of labor market segmentation ideas to their research. Third, racial differences in labor market experience are a common theme in labor market literature. In this research, equations are estimated selecting for race in an attempt to determine the extent to which race as well as segmentation is an important characteristic in the analysis of nonemployment mobility. Observed racial differences in nonemployment mobility controlling for segmentation may also suggest that the tri-partite segment structure is not sensitive to racial differences among segments, and that race as well as segment location is an important theoretical category.

The remainder of this paper is organized as follows: labor market structure will be discussed in Section 2, followed by Section 3, which contains a description of the econometric model. Section 4 contains a discussion of the empirical results, while Section 5 concludes the analysis.

2. Labor Market Structure

Labor market segmentation literature suggests that the labor market is composed of three labor segments, the independent primary, the subordinate primary and the secondary (Reich, Gordon and Edwards, 1973 and Piore, 1975). Occupations in the two primary segments are generally considered "good" jobs, offering relatively high pay, stability, fringe benefits and opportunity for advancement. Secondary segment jobs are usually defined as "bad" jobs based on the absence of the above characteristics. The independent primary is comprised of professional, managerial and craft positions. These jobs are usually well-paying, often offer good fringe benefit packages and opportunities for upward mobility. Administrative rules are often not codified and rigid; rather, they are characteristically embedded in an internalized code of conduct.

The other primary segment, the subordinate primary, is also composed of jobs, which provide good pay and benefits, are relatively stable, but which are characterized by specific tasks linked to relatively narrowly defined job classifications. Opportunities for upward mobility typically depend on tenure in a particular organization. Administrative rules in this segment are usually institutionalized and impersonal, and are accompanied by more or less well-defined grievance procedures.

The secondary segment is comprised of jobs with poor pay and working conditions. Often full-time secondary segment employment is not sufficient to lift a worker over the officially defined poverty line. Occupations in this segment typically are not thought to be part of mobility chain through which a worker has access to promotion opportunities. Work rules in this segment are often unwritten and arbitrary with little or no institutionalized recourse for employee grievances.

3. Economic Specification

Mobility is described as a stochastic process in which workers move from the three labor segments to a nonemployment state. The transition probability, [P.sub.ij] is the conditional probability of making a transition to state j at time t, given location in state i at time t - 1. The model describes transition probabilities from each of the three segments (i = 1,2,3) to a nonemployment state (j = 4). The three employment states are defined based on the labor market structure defined above, where state 1 is "employment in the independent primary," state 2 is "employment in the subordinate primary," state 3 is "employment in the secondary." The transition probability [P.sub.ij] can, therefore, be written [P.sub.i4].

Over the relevant interval [t - 1, t], workers in the three employment segments (i = 1,2,3) face two alternatives described by the variable m: transition to a nonemployment state occurs when m = 1, otherwise m = 0. Assuming the transition probabilities are described by a logistic functional form, and dropping subscript 4 for notational convenience in equations (1) and (2), we have

[P.sub.i](m = 1) = 1/1 + exp( - [x.sub.i'beta])

(i = 1,2,3), (1) where [x.sub.i] is a vector of explanatory variables and [beta] is a vector of unknown parameters. The log-odds of the transition probabilities are, therefore, expresses as a linear function of the explanatory variables by

1n [P.sub.i(m = 1)/1 - [P.sub.i](m = 1)] = [x.sub.i'beta]

(i = 1,2,3). (2)

A description of the independent variables along with expected signs is found in Table 1.


4. Empirical Results

Data used to estimate the model presented above are obtained from the Panel Survey of Income Dynamics (PSID) for the years 1981 - 1985. The respondents in the sample are male heads of household between the ages of 25-54. Selecting only "prime-age" males is an attempt to minimize the effects of both younger, and hence less stable, workers and older workers nearing retirement age. Further, only respondents that were household heads as of 1981 or before are included. This is necessary to avoid double counting those households where split-offs occurred. A split-off occurs when a household splits into two or more separate households.

A respondent included in the sample has an opportunity to enter on four occasions, once for each transition period. If, for example, a respondent is 22 years old in 1981 when he splits off from his original household, he would be excluded from the sample until the time he turns 25. At that time he would be included. This hypothetical individual would only contribute one year of observations to the sample.

The proposed mobility model requires that occupations are allocated into the segment structure as discussed above. Gordon (1986) proposes a methodology in which information on three-digit census occupation and industry codes is used to place occupations into segments. Two important conditions underlie the Gordon method: first, segment categories are determined based on characteristics of jobs rather than workers who hold the jobs, and second, labor market outcomes such as wages and turnover rates are excluded as much as possible from segment definitions (see Waddoups, 1989 for a more detailed summary of the Gordon procedure). Table 2 contains results of this allocation procedure. As expected, a higher concentration of white workers is found in the independent primary segment, while a larger proportion of nonwhite workers is found in both the secondary segment and nonemployment state. As is evident from Table 3, whites make up a majority of the sample at 68.1 percent, with African Americans and Hispanics comprising 28.5 and 2.8 percent, respectively.
Proportions of Respondents in the Various
 White Nonwhite
Independent Primary 55.6 23.1
Subordinate Primary 23.1 24.8
Secondary 14.8 33.5
Nonemployed 6.6 17.0
Source: Calculated from PSID data on male heads of
households ages 25-54. All proportions are averaged
over the 1981-85 period.
Racial Composition of Sample
Race Percent
Caucasian 68.1
African American 28.5
Hispanic American 2.8
Other .6
Source: Panel Study of Income Dynamics (1981-85).

Parameter estimates generated by maximum likelihood estimation for white, nonwhite and a combined white/nonwhite group of respondents are presented in Tables 4,5, and 6 below. Estimation results indicate that the parameter estimates on the RACE variable are positive and statistically significant at the .01 level for transitions from the two primary segments to nonemployment ([P.sub.14 and P.sub.24]). These results suggest that being nonwhites is positively correlated to nonemployment risk in the two primary segments. Nonwhites seem either to have access to less stable jobs within each of the primary segments possibly indicating racial segregation in the job market (Carter, 1982 and Doeringer and Piore, 1971), or to experience racial discrimination in lay-off and firing procedures.


The UNEMP variable, which is a proxy for the overall level of economic activity, is positively correlated with the probability of transition from the secondary segment to nonemployment ([P.sub.34]) for combined white/ nonwhite and white groups of workers. For nonwhites on the other hand, the statistical significance is found on the estimates on the UNEMP variable in the equation describing the probability of transition from both the independent and subordinate primary segments ([P.sub.14 and [P.sub.24]). These results suggest that nonemployment risk due to labor market conditions is allocated most heavily to white in the secondary segment and to nonwhites in the two primary segments. The results are also consistent with a racially segregated labor market, in which nonwhites are more often allocated to those occupations within the two primary segments that are more prone to nonemployment risk than the jobs of their white counterparts. In terms of nonemployment mobility and labor market conditions, primary segment nonwhites appear to have similar labor market experience to secondary segment whites. Nonwhite secondary segment workers on the other hand appear to be relatively unaffected by general labor market conditions.

Human capital variables included in the model are educational attainment (EDUC) experience (EXP) and job-tenure (JOBTEN) These variables may allow workers access to those occupations within the various segments that are less prone to nonemployment risk. Thus, they would exert negative influence on nonemployment transition probability. Results indicate that levels of formal education have a strong negative correlation with probabilities of transition from the independent primary to nonemployment ([P.sub.14]) for both racial categories, while the parameter estimates are uniformly insignificant predictors of transitions from the secondary to nonemployment ([P.sub.34]). As labor market segmentation theories suggest, formal education is a more important predictor of labor market outcomes for independent primary workers than for secondary workers. For white subordinate primary workers. the probability of transition to nonemployment is significantly affected by the level of education, but for their nonwhite counterparts, this effect is not present. Again, it appears as if nonwhites in subordinate primary employment, experience the labor market similarly to whites in the secondary segment.

In the equations modeling white transitions from the independent primary ([P.sub.14)] and nonwhite transitions from the subordinate primary ([P.sub.24]), the parameter estimates on the EXP variable are statistically significant and negative. This is in contrast to equations modeling secondary segment worker behavior, where labor market experience is not correlated with the probability of transition to nonemployment. These results suggest that increased labor market experience does not lead to a more secure position in the primary segment for independent primary nonwhites, subordinate primary whites or secondary segment workers in either racial group. A possible explanation for results found for primary segment workers may be that, on average, workers within the two segments, but in different racial groups, are holding different kinds of jobs. For example, it is likely that independent primary nonwhites hold relatively more craft positions, rather than managerial or professional positions. In craft positions such factors as age and skill obsolescence may be a more important factor increasing nonemployment risk. The positive sign on the [EXP.sup.2] variable indicates that the experience effect increases the probability of transition to nonemployment at an increasing rate. For the affected segments and racial groups, experience appears to be an increasingly important deterrent to nonemployment risk as its magnitude grows.

The JOBTEN variable is found to be negatively correlated with the probability of transition to nonemployment. For all labor segments and both racial groups this variable is negatively related to the transition to nonemployment and the estimates are statistically significant at the .01 level. These results are as expected for workers in the primary segments. For secondary segment workers, however, the findings are somewhat surprising. Conventional labor market segmentation theory suggests that secondary segment workers may not be prevented from nonemployment mobility simply based on job tenure. This effect is largely due to a lack of institutional structures, such as seniority and pension benefits, designed to decrease turnover. These results do, however, concur with Rosenberg (1980) and Rumberger and Carnoy (1980), who suggest that secondary work may not be as transitory as once thought. Indeed, while, the mean level of job tenure for secondary workers is less than mean levels for the other segments, the average length of job tenure for secondary segment work is still nearly 6.5 years (tenure in a specific job situation in the independent and subordinate primary segments are 6.7 and 7.9 years, respectively (Waddoups, 1989)), which contradicts the type of severe instability that is often presented in some labor market segmentation accounts.

Estimates on other explanatory variables may also be of interest. The parameter estimate on marital status is statistically significant and positive for white primary segment workers. For nonwhites, this effect is confined only to the independent primary segment. Results on the CHILD variable suggest that the existence of children in the household is positively correlated with transitions to nonemployment for primary segment workers. For secondary segment workers, on the other hand, CHILD is negatively correlated with nonemployment transition probabilities. A possible explanation for the difference in sign may be the relative lack of a social safety net for secondary segment workers. For example, secondary segment employees may be less likely than their primary segment counterparts to meet eligibility requirements for unemployment benefits. In this case transition to nonemployment may be more costly for those workers with such household responsibilities as children to support. The parameter estimate on REGION is statistically insignificant for equations modelling mobility to nonemployment for white workers; however, for nonwhites residence in the south exerts a negative impact on nonemployment risk for those in the independent primary and the secondary segments. Finally, it appears that union membership is negatively related to nonemployment mobility for all secondary segment workers, suggesting unions may especially help to insulate these workers from unemployment risk. Union membership also tends to insulate nonwhite independent primary segment workers from transitions to nonemployment.

5. Conclusions

This paper joins ideas found in the labor market segmentation literature with those found in dynamic explanations of unemployment to examine intersegment and racial differences in patterns of nonemployment mobility. Two major conclusions may be draw from this research. First, the tri-partite segment structure does seem to be picking up significant differences in mobility patterns from the labor segments to nonemployment. This result lends support to research agendas which view the labor market as segmented. Findings, for example, indicate that variables representing the level of economic activity, educational attainment, labor market experience, children in the household and race are all found to differ significantly in their explanatory power across the segment structure.

Second, the analysis suggests that race as well as segment location is an important theoretical category. Nonwhite subordinate primary segment workers, for example, seem to experience nonemployment mobility in much the same way as white secondary segment workers. This result suggests that segment location does not entirely account for differential labor market experience, but that race too is an important factor which must be taken into account.

To summarize, the primary purpose of this study is to show that segment location and race are both important categories to include in the study of job-leaving behavior. This research focuses attention primarily on determinants of transitions to nonemployment. Another interesting area for further investigation would consist of an analysis of intersegment and racial differences in nonemployment duration.


(1.) The nonemployment category is an aggregation of unemployment and nonparticipation states. This aggregation is necessary because the number of transitions to nonparticipation for prime-age males is so small among some labor segments that in some cases maximum likelihood estimation cannot be conducted. Earlier research shows that white workers in all labor segments are slightly less likely to move to both unemployment and nonparticipation states than their nonwhite counterparts. Additionally, secondary segment workers of both racial groups are somewhat more prone to nonemployment risk than those in the two primary segments (Waddoups, 1989).


Carter, Michael J. "Competition and Segmentation in Internal Labor Markets," Journal of Economic Issues, 1982, 16,4 1063-77. Clark, Kim B. and Lawrence H. Summers. "Labor Market Dynamics and Unemployment: A Reconsideration," Brookings Papers on Economic Activity, 1, 1979, 13-72. Clark, Kim B. and Lawrence H. Summers. "Unemployment Insurance and Labor Market Transitions," in Martin Neil Bailey ed. Workers, Jobs and Inflation, Washington, D.C.: The Brookings Institution, 1982, 279-323. Coleman, Thomas S. "Unemployment Behavior: Evidence from the CPS Work Experience Survey," The Journal of Human Resources, 1989, 24, 1-38. Doeringer, Peter and Michael Piore. Internal Labor Markets and Manpower Analysis, Lexington, MA: Lexington Books, 1971. Gordon, David M. "Procedure for Allocating Jobs into Labor Segments," working paper, New School for Social Research, May, 1986. Harrison, Bennett, "Public Employment and the Theory of the Dual Economy," in Sheppard, Harold L., Harrison, Bennett and Spring, William J. eds. The Political Economy of Public Service Employment, Lexington, MA: Lexington Books, 1972, pp. 41-75. Marston, Stephen T., "Employment Instability and High Unemployment Rates," Brookings Papers on Economic Activity, 1,1976, 169-210. Piore, Michael, "Notes for a Theory of Labor Market Stratification," in Richard C. Edwards, Michael Reich and David M. Edwards eds. Labor Market Segmentation, Lexington, MA: D.C. Heath, 1975, 125-50. Reich, Michael, David M. Gordon and Richard C. Edwards. "A Theory of Labor Market Segmentation," American Economic Review, 63,3, (May, 1973), 359-365. Rosenberg, Sam, "Male Occupational Standing and the Dual Labor Market," Industrial Relations, 19, 1, 1980, 35-49. Rumberger, Russell W. and Martin Carnoy, "Segmentation in the U.S. Labor Market: Its Effects on the Mobility and Earnings of Whites and Blacks," Cambridge Journal of Economics, 4,2, 1980, 117-32. Waddoups, Jeffrey, A Labor Market Segmentation Approach to Labor Force Flows, unpublished doctoral dissertation, University of Utah, 1989.
COPYRIGHT 1992 Omicron Delta Epsilon
No portion of this article can be reproduced without the express written permission from the copyright holder.
Copyright 1992 Gale, Cengage Learning. All rights reserved.

Article Details
Printer friendly Cite/link Email Feedback
Author:Waddoups, Jeffrey; Assane, Djeto
Publication:American Economist
Date:Mar 22, 1992
Previous Article:The publication incidence of replications and critical commentary in economics.
Next Article:The efficient tariff: systematically balancing security and welfare concerns.

Related Articles
Job mobility and wage growth: evidence from the NLSY79.
Self-employment, entrepreneurship, and the NLSY79: researchers have used the rich data from the 1979 cohort of the National Longitudinal Survey of...
Are male veterans at greater risk for nonemployment than nonveterans?

Terms of use | Copyright © 2016 Farlex, Inc. | Feedback | For webmasters