A rich picture of life from longitudinal data

Dr Loretta Platts

As we live longer, we need a better understanding of what an ageing population means for society, government and ourselves.

A picture of life in the UK and elsewhere can give us that, and help to bring about policies which support more people living and working – sometimes for decades – beyond retirement age.

A project to understand later life

The three-year WHERL (wellbeing, health, retirement and the lifecourse) project was established to do just that. We examined matters such as:

These were just a few of the strands of a major project funded by the ESRC and MRC, and involving the Institute of Gerontology and the Institute of Psychiatry at King’s College London, University College London, the Universities of Manchester and Toronto, and the Pensions Policy Institute. Our most recent paper was published in Ageing & Society in March.

Longitudinal data

The bedrock of our work was longitudinal data – such as Understanding Society and the English Longitudinal Study of Ageing (ELSA). These resources allowed us to examine the lives of men and women born in Britain between 1920 and 1949, and to build up a detailed picture of their experiences in the labour market and their family lives. These life histories are enormously helpful in understanding people’s working patterns in later life, and what they do in retirement.

To build these life histories we use optical matching analysis (OMA) – a technique which originated in molecular biology to study DNA sequences. In the social sciences, we use each ‘strand’ of a person’s life as seen in the data, and compare it with the strands of other people’s lives to observe the differences between them. Then we group similar lives together.

An aside: Limitations to optimal matching analysis

We were looking at people over a large number of years – from ages 16 to 54 – and over that time, no one life is going to match their group exactly. OMA allows us to place people in a group where their experiences most closely resemble the reference sequence.

So, someone who has many spells of unemployment – but which are all brief – might be counted as employed full-time throughout the period because their experience of the labour market most closely matches someone whose employment wasn’t interrupted. And if someone experienced marital disruption (such as divorce or being widowed), but remarries quickly, they might be classified as long-term married, because across the whole time period, they have more years married than not.

Patterns of work

But OMA does allow us to observe patterns in data that would otherwise be invisible because individual life histories are so diverse. The most striking we found have been marked gender differences in employment history. Almost all the men either worked full-time from 16-54 or worked full-time and left employment early (at around 49) – 96% of them fitted into one of these two trajectories.

By contrast, women’s lives fitted into seven different trajectories:

  • Only 20% worked full time throughout their careers – and they were less likely to be self-employed than their male counterparts.
  • Just under 10% were in full-time work and left employment early.
  • Almost a third of women showed what’s called ‘weak attachment’ to the labour market – about 10% did not work, and 20% were family carers.
  • Just over a third interrupted their careers to care for a family, and returned full-time (16%) or part-time (18%).
  • A smaller group (8%) moved from full-time work to part-time work, and stayed predominantly working part-time until they were 54.

Consequences of child rearing for pension income

One of our main findings was that leaving paid work to care for children substantially reduces retirement income. Among people aged 50 or older, the gendered division of labour means it is overwhelmingly women who have had more responsibility for family care and thus have had lower employment rates than men. Women who interrupted their careers for 10 years to raise children and then returned to work full-time were projected to have substantially lower income in retirement than any of the most common male life courses, including men who retired at only 49.

Even for men and women who followed a similar life course, working full-time throughout, the gender pay gap generated higher pensions for men. A median-earning man (50th percentile) can expect a similar pension to a high-earning women (70th percentile), when both have worked full-time throughout their adult life. We also found that the relative impact of motherhood may be greater for higher-paid women, since a higher proportion of their retirement income comes from private pensions.

Women could attempt to increase their pension income by returning to paid work sooner after having children. However, our modelling showed that returning to work part-time made little difference to pension income, since earnings from such jobs were too low. It made a greater difference to pensions for women to retire later, if there are jobs available and they are well enough to work into their late sixties and early seventies. 

In short, our work with pensions modelling showed how many aspects of the pensions system are designed around a certain male life course. Typical life courses for women deviate from this model, and this, as well as their lower earnings, is generating substantially lower pension income for women.

Informing government policy

Using Understanding Society – both wave-to-wave data, and data from life history interviews – was what made our research possible, and especially this detailed picture of life in Britain. This was the first step in our broader project to learn more about the labour market and family experiences of older workers, and how their experiences influence their working patterns leading up to and past state pension age – and the consequences of working later in life for people’s health and financial position.

That kind of understanding can help to inform government policy as the British population continues to be made up more and more of older members. For example, if policies that aim to extend working lives beyond state pension age don’t take into account differences in labour market histories, health and sociodemographic characteristics, they could exacerbate existing inequalities. And policies that aim to redress inequalities throughout the life course might be more effective in encouraging and enabling people to work beyond state pension age than policies focusing on the retirement transition, when it might be too late.

Now based in Sweden, I’m still working in this fascinating field – and currently working on some research which compares unretirement in Britain, Germany and Russia. I hope to return to this blog with news on that when it is published.

Author

Dr Loretta Platts

Loretta Platts is a researcher in public health and gerontology at the Stress Research Institute, Stockholm University