FINANCIAL CONSEQUENCES OF CAREER BREAKS A LATENT GROWTH MODEL ON REGISTER DATA Auteur(s): Frans Dorien, Dimitri Mortelmans en Caroline Masquillier
1.
Abstract
This study examines the financial consequences of career breaks. Using a sample of administrative data with information on individuals living in Flanders, the wage differentials after re-entry on the labour market are analysed. The data span the period from the second quarter of 1998 to the fourth quarter of 2006. A group of people with a career break and a control group of people without a career break were compared. In each group there were 90 414 individuals. For the analyses, additional selections were made after the initial sampling in order to control for sample selectivity. The study used multilevel models for change – also known as latent growth models – to analyse the data. The results suggest that there are significant initial wage differences after the break between people with a career break compared to people without a break in disadvantage of the first group. The wage differentials between men were greater than those between women. Moreover, a significant additional wage growth was found after the break for women but not for men. This could indicate that men are stigmatized more on the labour market for taking a career break than women. These results are explored in more depth by introducing the type of career break for men and women in the models. After controlling for important covariates (i.e. the number of children, age of the youngest child, age and income of 1998), there were significant differences according to type of career break (full-time (public/private), part-time (public/private), thematic leaves). Men and women with a full-time break had significant lower wages than men and women without a break respectively. A possible explanation is that in this type of break there is a total disconnection of the job, which can send a stronger signal to employers that these people are less committed than people who take part-time breaks or thematic leaves. Nevertheless, women with a part-time break, either public or private, also experience lower wages than women with thematic leaves and women without a break. This could be due to the fact that it is common for women to take such a thematic leave, and it therefore does not send a signal of lower job commitment. The wages of men with a thematic leave did not significantly differ from men without a break. An explanation could be that such leaves are becoming more and more socially acceptable and will consequently not lead to significant wage differentials. From these analyses, it is not clear what the exact consequences are of taking a career break. It is only possible to examine the initial wage differentials after the break at re-entry to the labour market and the subsequent differentials in wage growth from 2003 until 2006. There might also be selection-effects – like limited job ambition – that might underlie the wage differentials found in the study. Even though controlling for these selection effects was tried by including income before the break, it is not possible to fully grab all possible selection effects with this data.
Key words: Career breaks; latent growth model; financial consequences; longitudinal modelling; gender differences
2.
Study objectives
The career break scheme enjoys a great popularity and the number of people entering the system of career breaks still increases rapidly every year in Belgium. Over the years, the number of people doubled from 111 994 in 2001 to 223 319 in 2007, up to a record of 261 059 in 2010 (RVA, 2010). Despite this popularity, little is in fact known about the consequences of the Career Break Scheme. In international research, it has been shown that employees who interrupt their careers will be penalized on the labour market. Studies have shown the negative effects of taking a career break on career development in terms of fewer promotions and wage depreciation compared to individuals with continuous working patterns. However, it is still unclear in the literature how much wage loss can be expected. Since career interrupters might not always be aware of these negative effects before taking a career break, it is important to investigate the effects of career breaks. In this study, the administrative data of the datawarehouse Labour Market and Social Security was used to estimate the wage differentials after a career break. Persons taking a career break were compared to similar persons without a career break. The wage differentials after the break were looked at separately for men and women because they are expected to experience different effects.
1
Clarification of key concepts The work-life balance is a priority in a series of initiatives of the EU policy concerning childcare, work time and flexible arrangements that can have a direct effect on labour participation, family and life quality (Eurofound, 2004). Flexible arrangements like the career break scheme are considered useful instruments to achieve a better work-life balance (Hyman, 2004). Since 1985, the Belgian career break scheme allows employees to interrupt their careers temporarily while receiving a (limited) replacement income from the government. Initially, the underlying idea was to create more space on the labour market for the unemployed. The full-time career break allows employees to interrupt their careers and stop the work activities completely for a limited period of time. With a part-time break, it is possible to reduce the number of working hours temporarily with 50%, 33%, 25% or 20% (RVA, 2010). Throughout the years, the Career Break Scheme gradually shifted from an economic to a more individual-oriented measure by adding thematic leaves (i.e. parental leave, leave for medical assistance or leave for palliative care). In 2002, this policy shift was recognized and confirmed with the name change to ‘Time Credit’ in the private sector (Vanderweyden, 2002). The new Career Break Scheme is promoted as a way to increase the quality of life with a specific emphasis on the work-life balance.
3.
Methods and data
3.1 Data In order to analyse the wage differentials after taking a career break, longitudinal data are needed with an adequate representation of the number of people who took a career break in the past. Both requirements are met by using administrative data. Such data could be obtained from the Datawarehouse Labour Market and Social Security. The Datawarehouse manages and distributes longitudinal data from various social security institutions in Belgium. A sample spanning the period from the second quarter of 1998 to the fourth quarter of 2006 was obtained. Each year, 10% of the number of people of each type of career break was selected. A total of 90 414 individuals living in Flanders were selected at random. In order to compare the wages and wage growth. A ‘control group’ was designed where the same number of working people without a career break was randomly selected each year (total of 90 414). This ‘control group’ was also sampled to match the career interrupters on age and gender on a yearly basis. For the analyses, additional selections were made after the initial sampling. Only persons with one career break were taken in the analysis, as it is possible that these persons were selected more than once and had a break of more than one year. Taking the year 2003 as starting point in the analyses, a selection was made of people ending their career break in 2003. A final selection was made to include only people under the age of 50, because the career break motives of people younger than 50 and people of 50 years or older are of a different nature and relate to different social outcomes. It is also more probable that the older group career interrupters do not return to the labour market afterwards. It is therefore impossible to examine the wage differentials after the break for these people. 3.2 Methods The research questions are tested with multilevel models for change also known as latent growth models. These models provide a more profound way to examine these data compared to traditional regression analyses because they allow a two-level model, with the repeated measures on the first level and the individual persons on the second. The wage income is the dependent variable. The gross income measured on quarterly basis adjusted for inflation was used. The year 2003 (end of the break) is the starting point in the analyses. In this way, it is possible to take up income measures from before the break and several quarters after the break. The analyses examined longitudinal growth models for persons younger than 50 years. First, two simple models were fitted (i.e. the unconditional means model and the unconditional growth model). These unconditional models partition and quantify the outcome variation in two important ways: the first model describes the amount of variance on each level across people regardless of time, and the second model across both people and time by including time as a predictor in the model to examine the wage growth after the break at re-entry on the labour market. These models allow examining whether there is systematic variation in outcome that is worth exploring and showing where the variation is located (between or within people) (Singer & Willet, 2003). Both models also provide baselines for subsequent comparisons of following models where substantive predictors, like career break type and gender, are added on the second level to explain the variance between persons.
2
4.
Findings
Before the main findings of the analysis will be presented, a brief overview of the profile of the career break scheme in Flanders is given. Subsequently, five hypotheses will be outlined, distilled from the literature review, about the effects of career breaks in a longitudinal perspective. Lastly, the results from the analysis will be discussed. 4.1 The career break scheme in Flanders Taking a career break has become increasingly popular. The popularity appears to be double. On the one hand, there is a decrease in the number of full-time career breaks, while on the other hand there is an enormous increase in the number of part-time career breaks. It is primarily the rise in part-time breaks that explains the growth of career breaks over the past ten years. In 2010, almost 130 000 people took a part-time career break, either in the private or in the public sector. Little more than 85 400 of these people were aged 50 years or older. This stands in great contrast to the full-time options where less than 10 000 people took a full-time career break in 2010. The thematic leaves are also rather high, with almost 42 000 people taking a leave for a specific reason. Almost all of them belong to the middle age category. There are important differences between men and women in the use of career breaks. Taking a career break is still more common among women compared to men. 66% of the people taking a career break are female. This hegemony of women becomes even more apparent among the full-time breaks and thematic leaves (respectively 77% and 72%). However, this percentage is lower among the part-time breaks (63%), where the part-time options are gradually finding their way to men. This is indicative for the fact that the number of male interrupters went up considerably over the years. Among the part-time breaks, it is clear that men are mostly 50 years or older. This shows that the expansion is mainly situated among the older men in light of their retirement. The full-time interruptions remain at a rather constrained level, while the number of men taking a thematic leave also increases notably. This could be a sign that it is slowly becoming more acceptable for men to take a break for family reasons. It can be expected that there are different motivations between men and women to take a career break. Women generally interrupt their careers between ages 25 and 35, indicating that they are mostly motivated by familial reasons. Men on the other hand are usually 50 years or older. They generally use a break as an end of career measure. The distinct profile of persons taking a career break in Belgium is confirmed in the literature (Steegmans & Valgaeren, 2001). According to a study of Desmet, Glorieux and Vandeweyer (2007) the main motivation to take a break is because of children both for women and men under the age of 50. Part of the men with a full-time break nevertheless use it to try out another job or begin an own business (Devisscher, 2006). There has been an increasing popularity of career breaks suggesting that these flexible policy measures seem to address a certain need to interrupt careers temporarily. Moreover, it appears that taking a (part-time) break is becoming more acceptable and popular among men and women. The profile of career interrupters is nevertheless different from the ‘average’ Belgian employee, indicating that not every employee has the same chance of taking a career break. Higher income groups for example would be more able and thus more likely to take a career break (Devisscher, 2006). This suggests a potential selection of people that are more inclined to take an interruption. The possibility of self-selection needs to be taken into consideration in the analyses when examining wage differentials after a career break. 4.2 Hypotheses from the literature review From the literature, five hypotheses are derived regarding the effects of career breaks on subsequent wages. Moreover, each theory that lays the foundations for a particular hypothesis will be briefly outlined. Hypothesis 1: A career break leads to a significant wage loss during the break because the replacement income is much lower. Persons are not likely to receive the same wage after as before the break. Career breaks would not only cause the accumulation of human capital to stop during the break, it can also cause the existing human capital to deteriorate as the ‘Human Capital Theory’ states. In this theory the productivity of individuals is reflected by wages and job opportunities. This productivity increases as education, training and job experience increase (Becker, 1964). Therefore, it can be assumed that a career break interrupts the accumulation of work experience and can even cause the human capital to deteriorate during the break. As a consequence, it can lead to a lower productivity and reduce job opportunities. Consequently career interrupters will be confronted with relative lower wages after the break than before the break. After the career break, lower wages for career interrupters compared to continuous workers are expected. People with a career break would have lower initial wages after the break than people without a career break. Hypothesis 2: The wage loss immediately after the break is followed by a recovery period. During this period, the wage growth of people taking a career break would be faster compared to people without a break because a catch-up effect in work experience is expected. The recovery period after the break is seen by the ‘Human Capital Theory’ as a
3
phase where previously eroded human capital is restored (i.e. a catch up effect). Given such a process would be quicker and less costly than building up new human capital, this process and the wage growth will eventually slow down and continue to grow at the rate of the worker without a career break (Mincer & Ofek, 1982). On the long run, the wage losses would therefore be compensated. Hypothesis 3: It is hypothesized that longer career breaks would have a more negative effect on wages after the break than shorter career breaks. This can be explained by the ‘Human Capital Theory’ and ‘Signalization Theory’. The latter theory states that a temporary career break can be a signal to employers that these persons would be less committed to their job and prioritize their family over their work. This signal can play an important role in decisions regarding promotions where higher commitment is expected (Forrier & Sels, 2006). People who interrupt their career are in fact stigmatized for doing so (Beblo & Wolf, 2002). Therefore, it is possible that thematic leaves, that are usually short, lead to lower wage differentials between people with and people without a break, while part-time or full-time career breaks (in the public sector), that are generally used for a longer period of time, should experience lower wages compared to people without a break. There should be a clear distinction between types of career breaks, where the wage differential is lower or even non significant for people with a thematic leave compared to people without a career break, while people with other types of breaks should have higher wage differentials with people without a career break. Hypothesis 4: Contrary to the expected wage losses in hypothesis 1, it is also possible that a career break can have positive effects on wage levels. The ‘theory of New Home Economics’ (Becker, 1985; 1991) provides a possible explanation. This theory places productivity central and suggests that wage differentials between men and women with and without children are due to the lack of energy of women with children because of the amount of time and energy they have to put in their family and household. People with children are therefore less productive on their jobs which can have a negative impact on job opportunities and lead to lower wages. In light of this theory, it is possible that a career break can have a positive effect since it could allow employees to build up energy and increase productivity. Because of this fresh start after the break, it might be that the growth rates are higher for people with a career break compared to people without a career break, leading to higher wages on the long run for people with a career break. Hypothesis 5: Men with a career break are expected to be penalized more on the labour market than women with a career break. The wage differentials among men with and without a break would be higher than the wage differentials among women with and without a break. Since a career break is generally used by women, it is suggested that this is already reflected in the lower wages of women compared to men. In this respect, the ‘Statistical Discrimination Theory’ (Arrow, 1973 in: Román & Schippers, 2005) states that career breaks can encourage employers to discriminate certain groups. The employer would make a trade off based on previous experiences with former members of a certain group. Often a feature of an individual is ascribed to the whole group. Because mostly women take a career break, this is anticipated and taken into account for women in general. They are viewed as less productive in the long run causing them to have lower wages and job opportunities compared to men where a break is not expected (Schneer & Reitman, 1990). Men on the other hand are not expected to take a career break, and would more likely be stigmatized for doing so (‘Signalization Theory’). An interaction effect between sex and career break is expected to be significant where the initial wage differential after the break among men with or without a break is expected to be higher than the wage differential among women with or without a break. 4.3 Results It can be seen from the analyses that there are significant differences in initial wages in 2003 and changes in wage trajectories after 2003 between people with a career break compared to similar people without a career break (Hypothesis 1). The wage differentials are much higher between men with and without a break than between women with and without a break. Moreover, women with a break show a significant larger wage growth after the break than women without a break. The additional wage growth suggests that the wage differential would disappear in the long run (catch-up effect) (Hypothesis 2). For men, no significant differences in wage growth after the break are observed. Men seem to be ‘penalized’ more on the labour market in terms of wage than women for taking a career break (Hypothesis 5). This male penalty can be due to the fact that men are stigmatized more for taking a break. Since it is not common for men, this can send a signal of being less committed to their job, and hence lead to lower wages (‘Signalization Theory’). According to the ‘Statistical Discrimination Theory’, the lower wage loss found for women could also be due to the fact that women are penalized on the labour market with lower wages regardless of a career break because employers already anticipate a potential break. Examining different types of career breaks shows that women with a career break, regardless of the type of career break, have significant lower wages than women without a break after controlling for age, income of 1998 and childrelated covariates. Men with a career break have lower wages than men without a career break, except for men with thematic leaves and men with a part-time break in the private sector after controlling for age, income of 1998 and child-related covariates. Remarkable was the fact that the child-related variables did not have a significant effect on
4
the wages (differentials) of men which might suggest that men take career breaks for personal or work-related reasons. The wage differentials for men were a lot smaller or even non-significant for the breaks in the private sector. Only the breaks of 1 year could be examined. It could be a sign that for men the duration of the break is an important factor seeing wage differential for thematic leaves and for part-time breaks in the private sector were not significant. The wage differential found for full-time breaks in the private sector was considerably lower than the wage differential for full-time breaks in the public sector. Among men and women, those with full-time breaks, either public or private, seem to have the highest wage differential after the break compared to people without a break (Hypothesis 3). This is possibly due to the fact that they are totally disconnected from their job for a certain period of time. This can be explained by the loss of human capital that even deteriorates during the period. But it can also be explained in terms of the ‘Signalization Theory’ where a full-time break can send a stronger signal of lower job commitment than the other types of career breaks. Women with a part-time break, either public or private, also have lower wages compared to the continuous working women, but because they keep a link to their job, it is possible the wage differential is far less extensive than among full-time interrupters. Women with a thematic leave experience only a slightly lower wage than women without a break. Men did not have a lower wage after a thematic leave compared to men without a break. A possible explanation is that thematic leaves are popular and socially acceptable, and therefore does not send a signal of lesser work commitment leading to low or even non-significant wage differentials. Thematic leaves are also generally used less than one year. The smaller duration of this type of career break can also explain these small wage differentials.
5.
Conclusions and policy implications
As a general conclusion it can be said that there are significant initial wage differences after the break between people with a career break compared to people without a break in disadvantage of the first group. All the hypothesis were validated, except for the fourth which states that it is possible to find higher wages for people with a career break compared to people without a career break in the long run. Data with a longer span are needed in order to examine this hypothesis. The wage differentials between men with and without a career break were greater than those between women. Furthermore, a significant additional wage growth is found after the break for women but not for men. This finding could point to the fact that men are stigmatized more on the labour market for taking a career break than women. It could also indicate a selective entry into the career break scheme. Women and men with a full-time break had significant lower wages than women and men without a break. A possible explanation is that in this type of break there is a total disconnection of the job, which can send a stronger signal than part-time breaks or thematic leaves. Women with a part-time break, either public or private, nevertheless also experience lower wages than women with thematic leaves and women without a break. This could be due to the fact that it is common for women to take such a leave, and it therefore does not send a signal of lower job commitment. The wages of men with a thematic leave did not significantly differ from men without a break. This possibly indicates that such leaves are becoming more and more socially acceptable and will consequently not lead to significant wage differentials. From these analyses, it is only possible to examine the initial wage differentials after the break at re-entry to the labour market and the subsequent differentials in wage growths from 2003 until 2006. For men as well as for women, there was still some unexplained variance left on both levels, within persons (level 1) and between persons (level 2: initial status and rate of change). There are potential other predictors that can explain these wage differences on both levels that could not be examined or included in the models since the amount of available variables was limited. For example, there was no information on education (not available in administrative data), or other work-related variables that were only attained for people with a career break (full-time/part-time, working hours, quality of work). Because of these limitations, the wage differentials can only be explained partially. Because of the profile of people who are inclined to take a career break, the differences in wages can also be due to selection effects. It is probable that specific groups of employees are more likely to take a career break than others, such as persons with lower labour market commitment and less ambition to make promotions. It could be the case that these persons would have lower wages regardless of a career break. The administrative data used for these analyses, did only partially allow to control for possible selection-effects (e.g. there was no information on educational level, job ambition, psychological variables, ‌). The income level of 1998 was included, as a proxy for a possible lower attachment to the labour market before the break. Nevertheless, the wage differences found are significantly large, especially for men. The authors do not want to conclude anything on the labour market attachment of people taking a career break. The data do not allow doing so. Even though controlling for selectivity was tried, a mere control for income in 1998 does not fully prove this hypothesis. On the other hand, a wage penalty for men after a career break was found (again) (Frans, Mortelmans, 2009). The policy implications are important. Even though a gender gap is found in the analyses, when it concerns career interruptions the consequences show a surprising consistent gap for men who interrupt their career. These effects are
5
only valid for men below 50 because cases where the formula is used as an early exit strategy were not included. In that case, the social acceptance might be much higher (i.e. a cost reduction for the company). But early or mid-career men are clearly punished for leaving their job temporarily. Aside from these wage analyses, the authors want to stress the other potential effects of career breaks. As showed earlier on the Belgian Household Panel Data (PSBH), interrupting ones career can also have beneficial effects on one’s health. Stepping out the rat race might also have beneficial effects even though wages are affected in a negative way. These effects cannot be tested with these administrative data but should not be forgotten in the larger picture when evaluating the results from this study. On the other hand, even though other effects may be beneficial, the perceived negative signal of leaving ones job might also have consequences further in the career. The effect could only be tested for some years after the break. The effects on job mobility and long term effects of interruptions were not tested. Negative wage effects on the short term may as well have worse long term effects in terms of career success and subjective well-being in the job if the catch-up effect appears to be non-existent. Especially for weaker groups on the labour market, this should be taken into account when promoting long leaves from the labour market.
Full reference of study report(s) and or paper(s) and other key publications Frans, D., Mortelmans, D., & Masquillier, C. (2010). Financial consequences of career breaks. A latent growth model on register data, Steunpunt WSE, Antwerpen: Cello, Universiteit Antwerpen. Arrow, K.J. (1973). The Theory of discrimination. In: O. Ashenfelter, & A. Rees (Reds.), Discrimination in labor markets (pp. 333). Princeton: University Press. Beblo, M., & Wolf, E. (2002). Wage penalties for Career Interruptions: An Empirical Analysis for West Germany. ZEW, Discussion Papers, 02-45. Becker, G. (1964). Human Capital: A theoretical and Emperical Analysis, with Special Reference to Education. New York: Columbia University Press. Becker, G. (1985). Human Capital, Effort, and the Sexual Division of Labor. Journal of Labor Economics, 3(1), pp. 33-58. Desmet, B., Glorieux, I., & Vandeweyer, J. (2007). Wie zijn de loopbaanonderbrekers? Socio-demografische kenmerken, motivaties en arbeidshouding van loopbaanonderbrekers. Brussel: VUB. Devisscher, S. (2004). The career break (Time Credit) Scheme in Belgium and the Incentive Premiums by the Flemish Government (discussion paper), Brussel: Idea Consult. Eurofound (2004). Living to work - working to live: Tomorrow's work-life balance in Europe. European Foundation for the Improvement of Living and Working Conditions, 8. Forrier, A., & Sels, L. (2006). De prijs van een onderbroken loopbaan. Over.Werk tijdschrift van het Steunpunt WAV, ACCO, 4, pp. 73-83. Frans, D., & Mortelmans, D. (2009). Longitudinale gevolgen van loopbaanonderbrekingen. WSE-report. Hyman, J., Scholarios, D., & Baldry, C. (2005). Getting on or getting by? Employee flexibility and coping strategies for home and work. Work, employment and society, 19(4), pp. 705-725. Mincer, J., & Ofek, H. (1982). Interrupted work careers: depreciation and restoration of human capital. The journal of human resources, 17, pp. 3-24. RomĂĄn, A., & Schippers, J. (2005). De gevolgen van vrijwillige en onvrijwillige nonparticipatie voor loopbanen. Over.Werk tijdschrift van het steunpunt WAV, 4, pp. 131-135. RVA (2010). Statistieken en Studies: interactieve statistieken, [13/05/2010, www.rva.be ]. Schneer, J. A., & Reitman, F. (1990). Effects of Employment Gaps on the Careers of M.B.A.'s: More Damaging for Men than for Women?. The Academy of Management Journal, 33(2), pp. 391-406. Singer, J.D. ,& Willet, J.B. (2003). Applied Longitudinal Data Analysis: Modeling Change and Event Occurrence. New York: Oxford University Press. Steegmans, N., & Valgaeren, E. (2001). Mannen en vrouwen op de drempel van de 21ste eeuw. Een gebruikershandboek genderstatistieken. Diepenbeek: SEIN-LUC. Vanderweyden, K. (2002). 'Van loopbaanonderbreking naar tijdskrediet: Een verhaal van een wijzigend maatschappelijk discours', PSW-paper: 32.
6