Project Information

The impact of health changes on the development of older workers' earned income

Project objective

The overall objective of the study is to examine the effects of changing health status on the development of individually earned income in the late phase of the working life. After the legal possibilities for early retirement were gradually abolished as a response to demographic change since the mid-1990s, health restrictions proved to be the decisive reason for an early exit of older employees from the labour market. However, at least in Germany the policy response to the imminent demographic change was not limited to restricting an early transition to regular retirement, but also entailed significant aggravations regarding access to a disability pension. For this reason, Germany is now one of the European countries in which it is particularly difficult to obtain a pension due to reduced earning capacity. Instead of receiving a disability pension, persons with impaired health now have to reduce their weekly working time, change their job or end employment without any compensatory payment. This development is problematic for at least three reasons. Firstly, the individual old-age provision of the affected group of people is threatened because it cannot be built up normally and/or may have to be claimed early. Secondly, this risk results in subjectively perceived threats for younger employees to a successful working life.

Research questions

The focus of the project lies on the effects of a worsening health status on individual employment participation and individual earnings in the late employment phase. In concrete terms, we are analysing to what extent a worsening of the individual's health status leads to changes in the earned income (research question 1). The longitudinal nature of the data also makes it possible to analyse whether the observable effects on earned income are of a long-term or even permanent nature, or whether earned income returns to its original level (or above) after a certain regeneration period (research question 2). An improvement of the individual health status is therefore explicitly considered. In addition to the relationship between health status and earned income dynamics, we will also investigate which employment-related adjustments of the health-impaired employees lead to the observable changes in individual earned income (research question 3). In this regard, adjustments related to the respective employment relationship, such as changes in weekly working hours, the reduction of paid overtime and changes in the concrete field of activity, are primarily considered.

It is unlikely that a change in health status will have the same effect on earned income and employment status for all individuals surveyed. Instead, we assume that various factors within and outside individuals have a significant influence on the nature of employment-related changes. Accordingly, the study additionally examines the extent to which other individual, occupational and work-related characteristics moderate the effects of worsening health on individual earnings and employment status (research question 4). Furthermore, we consider the household level by examining to which extent potential income losses of the health-impaired partner are compensated at the household level (research question 5). It is conceivable, for example, that a partner might re-enter or enter working life for the first time, or that he or she might increase the weekly working hours.

Data and Methods

To answer the research questions raised, we use the SHARE-RV and SOEP data. Due to their panel structure, both data sets allow a differentiated analysis of the research questions. A clear advantage of panel data sets is that we can observe the same individuals over time and therefore can analyse not only differences between individuals, but also changes in the same individual. Thus, both the problem of unobserved heterogeneity and the problem of selectivity are significantly lower than is the case when comparing different individuals over time.

We use sequence data analysis to classify income profiles over time. By using sequence data analysis, it is possible to investigate the link between individual statuses and the chronological sequence of statuses. For the classification of health restrictions, we use the Latent Class Analysis (LCA). Similar to cluster analysis, LCA is a method for the classification of observations and has been increasingly used in sociological research in recent years. A decisive difference compared to cluster analysis is that within the framework of LCA the distribution assumptions regarding the classification features within each latent class are determined empirically and not theoretically. This means that measures are available for identifying the number of latent classes that are statistically more reliable than those in cluster analysis.

The panel structure of the data makes it possible to use fixed-effects regression (FE regression) to investigate the relationship between health restrictions and earned income. The great advantage of FE regression compared to other methods for the analysis of panel data is that FE regression focuses exclusively on intraindividual relationships for the construction of estimators. Thus, the exogeneity assumption underlying regression models can be formulated less strictly, because there is no violation of the assumption even if relevant constant personal or household characteristics remain unobserved. Correspondingly, fixed-effect estimators are more suitable for empirically substantiating statements about causal relationships than OLS or random effects regression estimators.


Prof. Dr. Martin Brussig, Tom Heilmann, Dr. Andreas Jansen: Individuelle Einkommensverläufe unter besonderer Berücksichtigung gesundheitlicher Veränderungen in der späten Erwerbsphase. Forschungsnetzwerk Alterssicherung. FNA-Fachgespräch, virtuell, 05.10.2021

Project data

Term of the project:
01.10.2018 - 30.09.2021

Reseach department:
Employment – Integration – Mobility

Project management:
Prof. Dr. Martin Brussig,Dr. Andreas Jansen

Project team:
Susanne Enssen, Tom Heilmann

Forschungsnetzwerk Alterssicherung (FNA) / Forschungsportal der deutschen Rentenversicherung