Jan Einhoff



PhD Candidate
DYNAMICS Research Training Group
Humboldt University & Hertie School

E-mail
Twitter
LinkedIn
GitHub

Thank you for visiting my website.

I am a PhD Candidate at the DYNAMICS Research Training Group, a joint programme of Humboldt University and the Hertie School in Berlin, funded by the German Research Foundation (DFG). I was a visiting researcher at the Finnish Centre for Pensions in 2023 and at the University of Wisconsin-Madison and the Max Planck Institute for Demographic Research in 2024. Later this year, I will be visiting Nuffield College in Oxford.

My research interests include life course dynamics, social stratification, comparative social policy, as well as novel quantitative methods for causal inference. For my dissertation, I investigate the class stratification of work-to-retirement transitions in the context of population ageing and extending working life policies in Europe. I use a wide range of quantitative methods, including sequence and decomposition analyses as well as targeted machine learning approaches. I also like to work with the potential outcomes framework and graphical causal models.

Besides my PhD research, I work as a Consultant at the OECD's Directorate for Science, Technology and Innovation, where I contribute to reviews of national innovation policy and a project exploring natural language processing tools and large language models for innovation policy analysis. In 2024, I was selected as a Young Thinker at the Centre for European Policy Studies (CEPS) to discuss policy priorities for the European elections.

You can find my CV here.

Here are some of my current research projects:

1. Cohorts' working life expectancies and working years lost in 21 European countries
[R&R] [Replication materials]
Details

Abstract: Across Europe, the extension of working lives has been a central policy goal for more than two decades. Working life expectancy (WLE) and working years lost (WYL) are well-suited demographic indicators for assessing countries’ progress towards achieving this goal. This article first reviews all available estimates of WLE and WYL for European countries. It then uses the largest available micro-level data – the European Labour Force Surveys (n > 10 million) – to estimate and project WLEs and WYL for cohorts of men and women aged 55–64 and 65–74 in 21 European countries. The results show that WLEs have generally increased, most rapidly in Central and Eastern Europe and in Western Europe. Northern European countries reach the highest levels of WLEs. However, country and gender differences remain large, especially when WLEs are adjusted for working hours. Correlational analyses suggest that working years have been gained primarily from successive cohorts losing fewer working years to retirement. The remaining WYL to retirement, to inactivity among women, and to unemployment in Southern Europe will be the main barriers to a further extension of working lives over the coming years.

2. Does owning your home make you retire early? A comparative analysis of Germany and the UK using targeted maximum likelihood estimation
[Submitted] [Replication materials]
Details

Abstract: Despite its key role in generating and consolidating social inequalities, little is known about the role of housing at critical life course transitions, particularly in later life. This article studies the effects of home ownership on the risk of entering retirement between age 51 and 65 in two distinct country contexts. After delineating the causal estimand and identification conditions, the analysis uses the doubly-robust, non-parametric Targeted Maximum Likelihood Estimation method with panel data from the German SOEP (n = 12,234) and the UK BHPS/UKHLS (n = 11,544). In line with theoretical expectations, home ownership is found to raise individuals’ retirement risks compared to renting by up to 21.3pp. in the UK and up to 7.4pp. in Germany. However, these effects are largest or only present for outright owners and close to statutory age thresholds. These findings highlight housing as a dimension of social stratification in work-to-retirement transitions and show how the institutional context moderates this relationship.

3. Occupational segregation and gender gaps in later life labour market participation: New evidence from causal decomposition analyses for Germany
[Work in progress]
Details
Abstract: Gender disparities in later-life labour market participation are large and persistent in most European countries. Compared to older men, older women are less likely to be employed, especially in full-time jobs, and more often inactive or retired. While theory and prior empirical work suggest that much of these disparities is due to men’s and women’s unequal positions in the occupational hierarchy, the precise extent to which segregation contributes to the observed disparities is unknown. To quantify this effect, this article uses a novel non-parametric and doubly-robust causal decomposition approach based on data from the German Socio-economic Panel (n=69,345). First, I document considerable disparities in several labour market outcomes between men and women aged 55 to 64 (in total, full- and part-time employment; unemployment; inactivity; and retirement). Surprisingly and contrary to expectations, I then find that a more gender-equal occupational structure would not substantially reduce these disparities, except for a 23% decline in the gender gap in part-time employment. The effects are slightly smaller in East than in West Germany and weaken over the three decades from 1991 to 2020. These findings qualify the role of occupational segregation in sustaining gender disparities in late working lives and illustrate the benefits of novel decomposition methods for social stratification research.

4. Towards flexibilisation? Age-period-cohort analyses of retirement entry sequences in Europe
[Work in progress]
Details
Abstract: In this article, I propose a novel combination of sequence and age-period-cohort analysis to study retirement entries in Europe. Based on data from the SHARE, sequence analysis is first used to extract and cluster sub-sequences that end in full retirement, which are then used in an age-period-interaction framework to disentangle the contributions of age, periods and cohorts to the changing prevalence of each cluster. I highlight the benefits of the approach over standard sequence analysis, including a clearer separation of timing and sequencing, larger and more homogenous clusters, and the option to consider time-varying covariates. Preliminary findings indicate a trend towards more diversity in retirement entries. More generally, the approach presented in the article has many applications in demographic research where age, period and cohort effects may bring about joint but distinct changes in life course transitions.