Abstract
Using a large German linked employer–employee data set and methods of competing risks analysis, this paper investigates gender differences in job separation rates to employment and nonemployment. In line with descriptive evidence, we find lower job-to-job and higher job-to-nonemployment transition probabilities for women than men when controlling for individual and workplace characteristics and unobserved plant heterogeneity. These differences vanish once we allow these characteristics to affect separations differently by gender. When additionally controlling for wages, we find that both separation rates are considerably lower and also significantly less wage-elastic for women than for men, suggesting an interplay of gender differences in transition behaviour and the gender pay gap.
This is a preview of subscription content, log in via an institution to check access.
Similar content being viewed by others
Notes
For an empirical study investigating worker turnover utilising only workplace characteristics, see Anderson and Meyer (1994).
Using the same data set, Boockmann and Steffes (2010) investigate the determinants of job duration in Germany and find that exit rates are influenced both by individual- and firm-level characteristics (such as labour market institutions). They show that the effects of these characteristics differ between exits to employment and unemployment, but their analysis is confined to male employees. This leaves open whether women move in a similar or different way.
In contrast, Frederiksen et al. (2007) apply a discrete duration model accounting for unobserved firm heterogeneity by including fixed firm effects, but they do not distinguish between separations to employment and nonemployment and, although estimating separate equations for men and women, they do not discuss gender differences in the overall separation rates and wage elasticities.
Details about the IAB Establishment Panel are given by Kölling (2000).
For details about the different LIAB models and their versions, see Jacobebbinghaus (2008).
Note that our data set does not allow us to distinguish between voluntary quits and involuntary dismissals. A crude approximation, which is in line with empirical evidence from other German data sets (see, e.g., Burda and Mertens 2001), would be that job-to-job moves are predominantly voluntary quits, while most of the separations to nonemployment may be involuntary dismissals.
See, for example, Bartelheimer and Wieck (2005) for a transition matrix between employment and nonemployment based on the German Socio-Economic Panel, which allows stratification of the ‘unknown’ category into detailed categories.
Note that our results are virtually identical when just considering transitions to unemployment instead of pooling separations to unemployment and to ‘unknown’ states into separations to nonemployment.
Since there is no detailed information on the number of hours worked, we exclude employees working part-time (at any time in the observation period). We further exclude establishments with a workforce of less than five and more than 1,000 employees because works councils, which cannot be set up in establishments with a workforce of less than five employees and which exist in virtually all establishments with more than 1,000 employees, are found to be one important determinant of both separations to employment and to nonemployment (e.g., Boockmann and Steffes 2010; Hirsch et al. 2010b). Moreover, apprentices and a small number of employees experiencing recalls are excluded. In addition, we keep only individuals which were on 1st of January 2000 between 16 and 55 years old, where the upper bound should ensure that the transitions into nonemployment are not due to (early) retirement. Finally, notifications which start and end at the same day and benefit notifications which correspond to employment notifications at the same time are deleted.
As already said in the data section, we unfortunately do not have (reliable) information on workers’ marital status and number of children, which would allow us to investigate this conjecture in more detail. We would expect the gender difference mainly to show up for married women with children and a much weaker difference, if any at all, for childless singles.
The sectors distinguished are (1) agriculture, hunting, and forestry (including fishing), (2) mining, quarrying, electricity, gas, and water supply, (3) manufacturing, (4) trade and repair, (5) construction, (6) transport, storage, and communication, (7) financial intermediation, (8) business activities, (9) other activities, as well as (10) non-profit organisations and public administration.
Note that a plot of the Nelson–Aalen baseline hazard after the Cox regression points at an overall negative duration dependence (see Appendix Figure 1). This seems plausible: The longer a worker stays in a specific match, the better should have been that match on average, and therefore the lower should be his or her separation probability to employment.
It is tempting to interpret this result as reflecting gender segregation in the labour market—with less attractive working conditions and more voluntary quits in more female-dominated establishments and men being more likely to voluntarily move out of these establishments than women. However, note that this interpretation is far-fetched given that the identification of the establishment characteristics’ coefficients via stratified partial likelihood rests on within-establishment variation in these characteristics (see, e.g., Ridder and Tunalı 1999).
A plot of the Nelson–Aalen baseline hazard after the Cox regression points at negative duration dependence in the first ten years of tenure and virtually no duration dependence afterwards (see Appendix Figure 2). This seems plausible insofar as more tenured workers should be less likely to be dismissed than new entrants owing to their higher specific human capital and dismissal protection legislation, where this effect should add less and less to employees’ employment stability when workers’ tenure increases further.
This difference is not necessarily an indication of discrimination against foreigners. An alternative explanation would be that it reflects foreigners’ higher propensity of leaving the labour force due to backward migration.
Again, one might be tempted to tell a gender segregation story explaining the gender difference in the effect of the proportion of female workers in the establishment’s workforce on workers’ separation rate to nonemployment. But as explained in footnote 16, stratified partial likelihood estimation of the coefficients of the establishment characteristics relies on within-establishment variation, casting serious doubts on such an interpretation.
Against this background, the negative (but due to the low precision in estimation insignificant) coefficient of the female dummy in the fully interacted stratified Cox model for the separation rate to nonemployment may be due to possible correlation of other control variables with the omitted wage of the worker.
For discussion of both approaches, see Manning (2011, pp. 991–997).
Another potential source of endogeneity could be self-selection of high-ability (and thus high-wage) workers into jobs with long expected employment duration (e.g., Altonji and Williams 2005). This may be particularly relevant in the high-wage jobs not included in our sample but less so for the uncensored low- and medium-wage jobs we predominantly observe.
As a check of robustness, we also estimated the following models by including a dummy for wage censoring rather than excluding spells with censored wages, where the dummy’s coefficient captures the average wage effect for wage-censored observations. The results, which are available upon request, are affirmative to our findings from the models excluding censored spells presented below.
At this stage, one may object that part of the wage effect found may be demand-driven, rather than a supply-side response. However, since we control for both observed and unobserved determinants of establishment’s layoff behaviour, we think this to be less of a problem. Note further that our sample does not include plant closings because it consists of a balanced panel of establishments. Nonetheless, we redid our analysis excluding downsizing establishments (i.e., establishments with an employment reduction of at least 25 % during our period of observation) as a check of robustness. We found that our results did not change qualitatively and are thus robust to this exercise.
Again, we have to admit that ideally we would have checked whether gender differences in separation rate elasticities and levels are primarily present for married women with children at home as opposed to childless single women for whom domestic constraints should not play a role. Unfortunately, this is not possible due to the absence of such information in our data.
Note that our estimation results are virtually identical when additionally stratifying the Cox models by sex (so that women and men have different plant-specific baseline hazards). This specification, the results of which are available on request, is not reported here as it does not identify level differences in women’s and men’s transition rates.
References
Alda H, Bender S, Gartner H (2005) The linked employer–employee dataset created from the IAB establishment panel and the process-produced data of the IAB (LIAB). Schmollers Jahrbuch (J Appl Soc Sci Stud) 125(2):327–336
Altonji JG, Williams N (2005) Do wages rise with job seniority? A reassessment. Ind Labor Relat Rev 58(3):370–397
Anderson PM, Meyer BD (1994) The extent and consequences of job turnover, Brookings papers on economic activity. Microeconomics 1994:177–248
Bartelheimer P, Wieck M (2005) Arbeitslosigkeit und Unterbeschäftigung. In: Soziologisches Forschungsinstitut, Institut für Arbeitsmarkt- und Berufsforschung, Institut für Sozialwissenschaftliche Forschung, and Internationales Institut für empirische Sozialökonomie (eds) Berichterstattung zur sozioökonomischen Entwicklung in Deutschland—Arbeit und Lebensweisen. Erster Bericht. Wiesbaden (VS Verlag für Sozialwissenschaften), pp 271–302
Barth E, Dale-Olsen H (2009) Monopsonistic discrimination, worker turnover, and the gender wage gap. Labour Econ 16(5):589–597
Bender S, Haas A, Klose C (2000) The IAB employment subsample 1975–95. Schmollers Jahrbuch (J Appl Soc Sci Stud) 120(4):649–662
Berger MC, Black DA (1998) The duration of Medicaid spells: an analysis using flow and stock samples. Rev Econ Stat 80(4):667–675
Blau FD, Kahn LM (1981) Race and sex differences in quits by young workers. Ind Labor Relat Rev 34(4):563–577
Boockmann B, Steffes S (2010) Workers, firms, or institutions: what determines job duration for male employees in Germany? Ind Labor Relat Rev 64(1):109–127
Bowlus AJ (1997) A search interpretation of male–female wage differentials. J Labor Econ 15(4):625–657
Burda MC, Mertens A (2001) Estimating wage losses of displaced workers in Germany. Labour Econ 8(1):15–41
Burdett K, Mortensen DT (1998) Wage differentials, employer size, and unemployment. Int Econ Rev 39(2):257–273
Cameron AC, Trivedi PK (2005) Microeconometrics: methods and applications. Cambridge University Press, Cambridge
Cox DR (1972) Regression models and life-tables. J R Stat Soc, Series B (Methodological) 34(2):187–220
Cox DR (1975) Partial likelihood. Biometrika 62(2):269–276
Frederiksen A (2008) Gender differences in job separation rates and employment stability: new evidence from employer–employee data. Labour Econ 15(5):915–937
Frederiksen A, Honoré BE, Hu L (2007) Discrete time duration models with group-level heterogeneity. J Econ 141(2):1014–1043
Galizzi M (2001) Gender and labor attachment: do within-firms’ relative wages matter? Ind Relat 40(4):591–619
Hirsch B (2009) The gender pay gap under duopsony: Joan Robinson meets Harold Hotelling. Scot J Polit Econ 56(5):543–558
Hirsch B (2010) Monopsonistic labour markets and the gender pay gap: theory and empirical evidence. Springer, Berlin and Heidelberg
Hirsch B, Schank T, Schnabel C (2010a) Differences in labor supply to monopsonistic firms and the gender pay gap: an empirical analysis using linked employer–employee data from Germany. J Labor Econ 28(2):291–330
Hirsch B, Schank T, Schnabel C (2010b) Works councils and separations: voice, monopoly, and insurance effects. Ind Relat 49(4):566–592
Jacobebbinghaus P (2008) LIAB-Datenhandbuch Version 3.0, Forschungsdatenzentrum des Instituts für Arbeitsmarkt- und Berufsforschung, Nuremberg, Datenreport Nr. 3/2008.
Jenkins SP (2005) Survival analysis, unpublished manuscript, Institute for Social & Economic Research, University of Essex
Keith K, McWilliams A (1999) The returns to mobility and job search by gender. Ind Labor Relat Rev 52(3):460–477
Klein JP, Moeschberger ML (2003) Survival analysis: techniques for censored and truncated data, 2nd edn. Springer, New York
Kohaut S, Schnabel C (2003) Tarifverträge—nein danke!? Ausmaß und Einflussfaktoren der Tarifbindung west- und ostdeutscher Betriebe. Jahrbücher für Nationalökonomie und Statistik (J Econ Stat) 223(3):312–331
Kölling A (2000) The IAB-establishment panel. Schmollers Jahrbuch (J Appl Soc Sci Stud) 120(2):291–300
Light AL, Ureta M (1992) Panel estimates of male and female job turnover behavior: can female nonquitters be identified? J Labor Econ 10(2):156–181
Loprest PL (1992) Gender differences in wage growth and job mobility. Am Econ Rev (Papers and Proceedings) 82(2):526–532
Manning A (2003) Monopsony in motion: imperfect competition in labor markets. Princeton University Press, Princeton
Manning A (2011) Imperfect competition in the labor market. In: Ashenfelter OC, Card DC (eds) Handbook of labor economics, vol 4B. Elsevier, Amsterdam, pp 973–1041
OECD (2009) How do industry, firm and worker characteristics shape job and worker flows?. In: OECD 714 Employment Outlook 2009, Chapter 2. Paris, OECD, pp 117–163
Pencavel JH (1972) Wages, specific training, and labor turnover in U.S. manufacturing industries. Int Econ Rev 13(1):53–64
Postel-Vinay F, Robin JM (2002) Equilibrium wage dispersion with worker and employer heterogeneity. Econometrica 70(6):2295–2350
Ransom MR, Oaxaca RL (2010) New market power models and sex differences in pay. J Labor Econ 28(2):267–290
Ridder G, Tunalı İ (1999) Stratified partial likelihood estimation. J Econ 92(2):193–232
Robinson JV (1933) The economics of imperfect competition. Macmillan, London
Royalty AB (1998) Job-to-job and job-to-nonemployment turnover by gender and education level. J Labor Econ 16(2):392–443
Therneau TM, Grambsch PM (2000) Modeling survival data: extending the Cox model. Springer, New York
van den Berg GJ, Ridder G (1998) An empirical equilibrium search model of the labor market. Econometrica 66(5):1183–1221
Viscusi WK (1980) Sex differences in worker quitting. Rev Econ Stat 62(3):388–398
Author information
Authors and Affiliations
Corresponding author
Additional information
We would like to thank Jürgen Deinhard, Thorsten Schank, Rainer Winkelmann, and an anonymous referee for very helpful suggestions. We further appreciate the comments of participants of the 2011 annual conference of the Ausschuss für Bevölkerungsökonomie of the Verein für Socialpolitik and of the economics research seminar at the University of Augsburg.
Appendix
Appendix
Smoothed Nelson–Aalen baseline hazard after Cox regression of the workers’ instantaneous separation rate to employment
Smoothed Nelson–Aalen baseline hazard after Cox regression of the workers’ instantaneous separation rate to nonemployment
Rights and permissions
About this article
Cite this article
Hirsch, B., Schnabel, C. Women Move Differently: Job Separations and Gender. J Labor Res 33, 417–442 (2012). https://doi.org/10.1007/s12122-012-9141-1
Published:
Issue Date:
DOI: https://doi.org/10.1007/s12122-012-9141-1