MPIDR Working Paper
Eignet sich das Mikrozensus-Panel für familiensoziologische Fragestellungen? Untersuchung am Beispiel der Frage nach den ökonomischen Determinanten der Familiengründung
Kreyenfeld, M. R., Zühlke, S., Schmidtke, K.
Can we use the micro-census panel for studying family dynamics?
First applications on the economic determinants of family formation
MPIDR Working Paper WP-2009-024, 32 pages.
Rostock, Max Planck Institute for Demographic Research (October 2009)
The German microcensus is a rotating panel in which respondents are interviewed once per year for four years in a row. Recently, the microcensus panel 1996-1999 has been made available for scientific usage. This paper discusses the potential of the microcensus panel for inves-tigations in the field of fertility research. In the first part of the paper, we demonstrate how the data can be transformed into spell format so that event history models can be applied. Respondents become “under observation” when they are interviewed for the first time in 1996. The “risk set” are female respondents aged 16-38 who are childless at first interview. They contribute exposure time to our investigation until the last interview is conducted or have a first child. Based on this data set up, we have estimated a first birth model. Estimates are very much in line with previous findings on fertility behavior in Germany. But even though that our first investigations provide reasonable results, there are several shortcomings of the data, which we also discussed:
First, fertility histories are not surveyed in the microcensus. As others have done before, we have reconstructed fertility careers of female respondents based on the ages and numbers of the children who live in the same family unit. This procedure gives fairly good results for female respondents who are below age 39 at time of interview. Older respondents, and therefore fertility at higher ages, cannot be investigated with this data.
A second drawback is that there are no monthly employment histories available. However, information on the employment status at time of interview can be used. We assumed here that the employment status that we measure at time of interview is fixed for the period of twelve months after interview. This is a very strong assumption for some indicators, such as unemployment. However, for other employment indicators (such as whether the person is working in a public or private enterprise, whether the working contract is term-limited or not) this assumption should not be problematic.
A third downturn of the data is panel-attrition. The microcensus is a household-survey. When respondents leave the household, they are not surveyed any longer. This involves that panel-attrition is related to residential mobility. In a ‘sensitivity analysis’ we show that panel attrition does not seem to bias our investigations. The paper concludes with a critical discussion on the scope and limits of the microcensus panel for fertility investigations.