SOEP Conferences

4th International German Socio-Economic Panel User Conference (SOEP2000)

SOEP Quicklinks:  





SOEPdata Documents


Poster Session Abstracts


1. Sascha O. Becker, European University Institute
Explaining School Enrolment Behavior in Germany and Italy

2. Steffen Brenner/ Joachim Schwalbach, Humboldt Universität zu Berlin
Firm Size and Management Quality

3. Hilke Brockmann, MPI for Demographic Research and Thomas Klein, Uni Heidelberg
Changing family formations, occupations and mortality

4. Michael Brookes, Middlesex University Business School
Gender Earnings Mobility: A Comparison Between Germany and the UK

5. Veronika V. Eberharter, University of Innsbruck
Convergence in Poverty Intensity and Poverty Mobility ? Empirical Evidence from Germany and the United States in the 90ies

6. Marcel Erlinghagen, Institut Arbeit und Technik (IAT), Gelsenkirchen
Unemployment and volunteer work in longitudinal perspective

7. Joachim R. Frick and Markus Grabka, German Institute for Economic Research, Berlin
Income Distribution and the Impact of Imputed Rent: Alternative Concepts of Measurement and the Sensitivity of Income Inequality

8. Karsten Hank, Max-Planck-Institute for Demographic Research, Rostock
Regional Opportunity Structures and Fertility in Western Germany

9. Asuncion Soro-Bonmati, European University Institute, San Domenico di Fiesole
Transition Patterns into Adulthood: a comparison of labour market transitions and transitions to independent living of youth in Germany, Italy and Spain

10. Daniela Vuri, European University Institute, San Domenico di Fiesole
Fertility and Divorce

nach oben

Explaining School Enrolment Behavior in Germany and Italy

Sascha O. Becker, European University Institute

Italy's university dropout rate is the highest of all OECD countries. As a consequence Italy also has one of the lowest rates of people holding a tertiary degree. In contrast, Germany's dropout rate is much lower and overall educational attainment higher than in Italy. In this paper, I address the factors explaining school enrollment and dropout behavior in both countries. The analysis for Italy, based on two Italian micro data sets, the Italian Survey of Household Income and Wealth (SHIW) and a special data set surveying high school leavers, shows that variables characterizing local labor market conditions exert a strong effect on enrollment and dropout behavior.

Using the German Socioeconomic Panel (GSOEP), I show that in Germany the decision to enroll in education seems to be independent of local economic factors.
For both countries, variables characterizing family background, play an important role.


nach oben

Firm Size and Management Quality

Steffen Brenner and Joachim Schwalbach, HU-Berlin

Economic theory predicts that management quality is related to firms size. Therefore, one would expected that the most talented managers should work for the largest firms. Indeed, the empirical studies found that managers' pay as a substitute for management quality is strongly tied to firm size. Exploring a subsample with micro data from the German Socio-Economic Panel for the period 1984 to 1998 we find a strong relationship between pay and size as well. Furthermore, using a structural approach we show that this relationship can partially be explained by education serving as a proxy of ability.


nach oben

Changing family formations, occupations and mortality

Hilke Brockmann and Thomas Klein

In all advanced countries life expectancy has significantly improved. Life is still prolonging at a considerable pace. Demographers have shown in numerous studies that wealth, education, occupational and marital status affect adult mortality. The findings are robust over time and across countries. However, the causal mechanism behind these findings are yet unclear. How does being single or having a lower occupational status translate into a higher probability to die early? The purpose of this poster-presentation (and following paper) is to bring demographic and sociological perspectives together. It compares the effect of different marital and occupational settings on mortality, based on GSOEP data.


nach oben

Gender Earnings Mobility: A Comparison Between Germany and the UK

Michael Brookes, Middlesex University Business School

It is becoming reasonably well established that when making cross-country comparisons the German mean female wage is closer to the mean male wage than it is for their British counterparts. (See Brookes, Hinks and Watson 1999, Black, Trainor and Smith 1999, as examples). The purpose of this paper is to explore the role that is played in this by earnings mobility, i.e. is the gender wage gap narrower in Germany because German women are more likely to move up the earnings distribution than British women, and less likely to move down.

Using the PACO Dataset, derived from the SOEP in Germany and the BHPS in the UK, earnings mobility is analysed using 3 different techniques. Firstly absolute mobility is compared across the 2 countries using the method pioneered by Fields and OK (1996), this allows for any absolute earnings mobility to be decomposed into the portion due to earnings growth and the portion due to earnings transfers. Secondly relative mobility is addressed. This is done firstly by ranking each worker within the overall earnings distribution in an initial year and then again in a subsequent year. This allows for the production of separate transition matrices for men and women which reveal the probabilities of being in any decile group in the subsequent year having been in a particular decile group in the initial year. From each matrix a Bartholomew index can be calculated, (Bartholomew 1973), which is used to make direct comparisons of the relative positions of British and German women, the index is also modified to take account of the direction and the distance of the mobility. Finally Ordered Probit models are estimated for both countries to establish which labour market characteristics are the most likely to increase (decrease) the probability of upward (downward) mobility. This analysis has so far been carried out for the years 1991-93, as additional years from the PACO Dataset have recently become available, it will be extended to include the years 1994-96 in the near future.

Given the fact that the gender wage differentials are narrower in Germany, it was expected that, all other things being equal, German women would be in a more favourable position in terms of earnings mobility than women in Britain, i.e. more likely to move up the distribution and less likely to move down. However, for the period 1991-93, this did not prove to be true, in all cases British women outperformed their German counterparts.

Unfortunately it is difficult to make any firm conclusions based upon this one result, since it is impossible to ignore the impact of economic growth. The more favourable position of UK women in this period may simply reflect the 2 economies being at different stages of their respective business cycles. However this should not be the case once the additional years have been included. Assuming that these initial results hold after further analysis it has interesting implications for the 2 economies. Firstly, for Germany, why are their greater opportunities for earnings mobility for British women? Secondly for Britain, if female employees are better placed in terms of earnings mobility, why is their gender wage gap still wider? It can only be because of the treatment of new and frictional labour market participants, in this initial case women who became employed after 1991 and women who were employed in 1991 but were no longer employed by 1993.


nach oben

Convergence in Poverty Intensity and Poverty Mobility?

Empirical Evidence from Germany and the United States in the 90ies 

Veronika V. Eberharter, University of Innsbruck

Cross country comparisons point out remarkable differences between Germany and the United States concerning labor market participation, income inequality and poverty in the 80´ies. In the 90´ies Germany experienced structural changes concerning female labor market participation and household composition. The distributional effects of these changes lead to an increasing polarization of individual and household incomes and increasing poverty affliction of certain groups of the population, e.g. women or families with children. The diverging economic indicators in the 90ies let expect a turning point in gendered poverty affliction in the United States and an increasing poverty intensity in Germany, expressed by poverty head count ratio and poverty aversion. From the longitudinal viewpoint this would mean a higher immobility of the poor respectively a higher mobility into poverty and a higher relative risk to stay poor. This antipodal development would lead to a convergence of poverty indicators in the two countries. We test these hypotheses using the international version of the socio-economic panel GSOEP-PSID (1980-1997) and choose the periods 1986-1991 and 1992-1997. Observation units are persons aged 20 to 50 years at the beginning of the respective observation periods with positive real (1991=100) equivalent (OECD Equivalence Scale) post government income. We use the conventional OECD poverty concept, defining poverty line with 0.5 median disposible household income.


nach oben

Unemployment and volunteer work in longitudinal perspective

An analysis of the West German subsample from the German Socio Economic Panel (GSOEP) for the years 1992 and 1996

Marcel Erlinghagen

Starting point of the analysis is the discussion about possibilities to defuse the crisis on the German labour market by supporting volunteer work. For that reason, the effects of unemployment on the probability to volunteer are of special interest. For this purpose, logistic regressions are estimated for the years 1992 and 1996, using longitudinal data from the West German subsample of the German Socio-Economic-Panel (GSOEP).

There is no evidence for an increasing propensity to take up or maintain volunteer work among the unemployed. In contrast, it is shown that the chance to volunteer especially increases with a higher educational degree or if the person lives in ‘secure' family circumstances. On the ‘volunteer market' similar qualifications are in demand, which also support a successful participation in the regular labour market.

Therefore, the hope that an assumed individually higher willingness to volunteer among the unemployed may contribute to cope with the general labour market crisis turns out to be misleading. Especially low-educated persons, being a problem group on the labour market, do not regard volunteering as an adequate activity for themselves.


nach oben

Income Distribution and the Impact of Imputed Rent:

Alternative Concepts of Measurement and the Sensitivity of Income Inequality

Joachim Frick and Markus Grabka

As early as 1968 the United Nations argued in favour of capturing imputed rent from owner-occupied housing in national account statistics; in 1977 the UN issued guide-lines recommending the inclusion of this income type along with other property income. Since then, there is increasing international literature as well as empirical research on the distribution of personal income showing the significant impact of imputed rent (as a most relevant income component of owner-occupiers) on income inequality. The overall magnitude of this effect depends on

1. the population share of owner-occupiers, which is also relevant for social policy decisions given the need for (additional) private old age provision. This also shows the need to consider individual preferences which drive the decision to invest in real estate rather than in financial assets.
2. the respective income advantage captured in the value of imputed rent. This issue is related to shortages on the housing market.

In addition to dis-aggregated information from national accounts micro-data from household surveys can offer at least two ways of estimating imputed rental values: a first subjective way is based on a self-assessment by owner-occupiers by asking for a "best guess" rent they would have to pay for their apartment if they were renters. A second and clearly more objective way is based on regression models on the amount of the rent paid by actual renters, simultaneously controlling for a variety of factors like housing size, construction year, quality of endowment, etc. Adapting Oaxaca`s (1973) decomposition analysis we apply these regression coefficients to the population of otherwise comparable owner-occupiers and their respective apartments or houses which results in an estimate of the true gross value at market prices. Deducting owner-specific costs for taxation, maintenance costs, and interest on mortgages yields a net value which can be interpreted as an income advantage of owner-occupiers.

Micro data from the German Socio-Economic Panel Study (GSOEP) is used to calculate both alternatives. In our empirical analyses we depict the absolute and relative relevance of imputed rent as given by both measures in contrast to a reference model where imputed rent is set to zero. In a second step we look at the respective impact on the income distribution when applying some robust inequality indicators (Gini coefficient and Mean Logarithmic Deviation). Given the still existing differences in the housing market we run these analyses separately for West- and East Germany as of 1998. In order to control for time effects we also look at West German results in the mid 1980s.


nach oben

Regional Opportunity Structures and Fertility in West Germany

Karsten Hank, Max-Planck-Institute for Demographic Research (Rostock)

In this paper it is argued that fertility decisions are not only determined byindividual characteristics, but also by the given opportunity structures of theindividual's environment. In this context, regional factors such as the degree ofurbanization, the provision of day care, and local labor market characteristics areconsidered to be of particular importance. Individual level data from the German Socio-Economic Panel (GSOEP) are linked with aggregate data from the ‘Statistik regional' database. First birth probabilities of women living in western Germany in the mid-1990s are then estimated, using a sample selected probit model. The results indicate some influence of regional characteristics on a woman's propensity to have a first birth in the observation period. They also suggest, however, that the analysis of contextual effects should not be reduced to ‘materialized' opportunity structures, but that broader social contexts (including social interaction processes) need to be taken into account.


nach oben

Transition Patterns into Adulthood

A comparison of labour market transitions and transitions to independent living of youth in Germany, Italy and Spain

Asuncion Soro-Bonmati, European University Institute

Recent surveys on European youth (EC (1997), OECD (1998)) point out that finishing education, getting a first job and leaving the parental home takes longer in Northern countries, such as Germany, than in Southern countries, such as Italy or Spain. This paper uses panel data from the German Socio-Economic Panel (GSOEP), the Bank of Italy Survey on Household Income and Wealth (SHIW) and the Spanish waves of the European Community Household Panel (ECHP) to test whether these differences are significant and if so, whether explanatory factors can be identified.

The aim of the paper is to examine whether young people in Germany have ''smoother'' transition patterns into adulthood, i.e. easier labour market entrance and easier transition to independent living, than young people in Italy and Spain. The analysis tries to shed light on whether differences in the educational system, the labour market structure and the role of the family in the three countries under study help to explain the cross-country differences in youth labour market entrance and the process of leaving home.

The first part of the paper shows that there are significant cross-country differences in both the labour market transition probabilities and the hazard rates of leaving home even after controlling for demographic variables such as age, gender and level of education. The main results can be summarized as follows. First, Italian students have significantly lower probability of moving into employment than German students. Second, the probability of exiting unemployment is significantly lower in Italy than in Germany. Third, the probability of moving out of employment is significantly higher in Spain than in Italy. Finally, Germans are significantly more likely to leave the parental home than Spanish and Italians.

The second part is devoted to the analysis of the factors that may explain some of the cross-country differences observed. The following hypotheses are investigated. First, the German dual system is successful as a bridge into the labour market while the Italian vocational system does not work, so that those graduating from the vocational tracks in Italy encounter problems to find jobs. Second, first job seekers in Germany do not seem to face difficulties in finding a first employment, while the main problem for those unemployed in Italy is finding a first job. Third, having part-time or short-term contracts leads to employment instability in Spain but not in Italy. Finally, the family plays a very active role among those facing bad labour market prospects in Italy and Spain but not in Germany, so that staying at the parental home is very common among Spanish and Italians looking for a first job and for those employed with short-term or badly paid jobs.


nach oben

Fertility and Divorce

Daniela Vuri, European University Institute

I am currently working on the first paper of my PhD dissertation, titled ''Fertility and divorce''. The starting point of the analysis is the empirical result, found in many papers, that couples with children tend to divorce less. Is it possible to infer from this finding that the presence of children discourages marital dissolution, that is, there is a causal effect between them? Or does it exist instead a problem of omitted variables bias, due to the presence of unobservable factors correlated to both fertility and probability of divorce? The goal of this paper is to give an answer to these questions. I do so by setting up a theoretical framework that links fertility and divorce decisions and by estimating this model using 2-stage instrumental variable techniques. The five waves 7-12 (1990-1995) of the German Socio-Economic Panel are used to build the sample on which estimation results are based.

In more detail, after a brief summary of the literature on fertility and marital instability, I formulate a dynamic decision model of fertility and divorce. In the context of a model of marital-specific investment, it is shown that marriage continuation probability is increasing in the number of children, and that the number of children is increasing in an unobservable measure of the quality of the marriage, which in turn influences the perceived marriage duration. Then, I indicate how to move from the model to the data. In particular, by making some assumptions on the functional forms of the expressions defined in the model, it is possible to write two equations empirically estimable: the first one describes the effect of the number of children on the probability of divorce, the second one represents the effect of the couple's subjective probability of divorce on its fertility. In order to obtain consistent estimates of the effects of children on marital behaviour, two possible strategies are available: (1) "single equation'' estimation, which consists in focusing on divorce equation and accounts for endogenous fertility by using IV methods; (2) "simultaneous equations'' estimation techniques, which consist in estimating the two equations simultaneously by full maximum likelihood estimation.

In this paper, I focus on the first estimation procedure (IV methods). In order to apply instrumental variables techniques, it is necessary to find a valid instrument for the divorce equation, that is, a variable affecting fertility but not influencing directly the decision to divorce. I argue that the exogenous source of variation in fertility choices can be ''the sex composition of previous children'', which exploits the widely observed phenomenon of parental preferences for ''balanced'' families in terms of the sex composition of their children. Some simple tabulations, documenting the relationship between the fraction of couples who have a third child and the sex of the first two children, confirm that couples with two children of the same sex are much more likely to have a third child than couples who have one boy and one girl.

Then the estimation results of the effect of fertility on divorce from both ordinary least squares regression and two-stage least squares instrumental variables regression are reported and interpreted.

Two important conclusions emerge from the analysis. First, the standard approach with no instrumenting fertility (OLS estimates) leads to a negative and significant estimate of the impact of exogenous changes of fertility on marital dissolution. Second, IV estimates that exploit the fertility consequences of sibling sex composition on marital instability contradict the OLS results. For instance, the implied instrumental variables estimate of the effect of fertility on divorce (0.010) is substantially above and of different sign with respect to the probability of divorce estimate by a conventional ordinary least squares procedure (-0.015). Nevertheless, the standard errors of the IV estimates are relatively large and one cannot reject the hypothesis that differences between the IV and OLS estimates are due to sampling errors. Therefore, more efficient estimation methods have to be identified in order to improve the statistical significance of the results. The research can continue in three directions:

1. ''single equation'' estimation methods by providing a better specification of the probability model (probit) or using propensity scores matching methods;
2. ''simultaneous equations'' estimation methods by applying full maximum likelihood estimation;
3. fully exploitation of the panel structure of the data in order to relate changes in childbearing to changes in marital instability. It requires the use of duration models that allow for time-varying variables. A reasonable starting point would be to model the potential simultaneity of marital dissolution and recent births, taking past births as weakly exogenous.

In the next months my aim is to implement the procedures just mentioned in order to evaluate the robustness of the results


nach oben