This was published under the 2010 to 2015 Conservative and Liberal Democrat coalition government
Evidence on individual’s circumstances at or near retirement, matched with information on their lifetime earnings.
Research published today by the Department for Work and Pensions (DWP) provides descriptive evidence from the English Longitudinal Study of Ageing (ELSA) on individual’s circumstances at or near retirement, matched with information on their lifetime earnings from National Insurance administrative records.
Linking this data provides evidence on how retirement incomes relate to lifetime earnings. It provides improved estimates on the level and distribution of retirement resources of those approaching retirement, including the characteristics of those who are at greater risk of not making appropriate retirement saving decisions.
The main findings from this research can be divided into two key themes:
Lifetime earnings and retirement income
- On average, those with low lifetime earnings had income in retirement that was higher relative to their lifetime earnings than individuals with higher lifetime earnings. This is influenced by the redistribution inherent in the state pension rules and other benefits.
- The comparison between the earnings profiles for single men and single women by age cohort highlights how these two groups are different. Single men are found to have higher earnings than single women in the same cohort but they have experienced much slower earnings growth across cohorts. Cohort effects appear to be very sizeable for single women, reflecting the increased participation of women in the labour market over time.
- Single women, many of whom were previously married, tend to have had relatively low levels of lifetime earnings. However, on average their retirement incomes are higher relative to their lifetime earnings than for single men and couples, reflecting that previously married women may have benefited from the pension rights of their former partners.
Estimates vs. administrative records
- The estimates of state pension wealth made before information from the NI records was available were found to be very similar up to around the thirtieth percentile, after which they start to diverge, with the distribution of previously estimated state pension rights being above that of true state pension rights.
- The estimates were found to be more accurate for men and never married women than they were for currently married women or women who were previously married and are now separated or widowed.
- Among both men and women the previous estimates were found to have been more accurate for younger groups than they were for those within five years of the SPA.
- Overall the linked subsample can be considered representative of the individuals in the full ELSA sample, in terms of its observable characteristics.
- However, older people and the self-employed are under-represented in the linked subsample so particular care would need to be taken with any analyses that looked specifically at these subgroups.
Notes to Editors
- DWP Research Working Paper No. 81 “Retirement outcomes and lifetime earnings: descriptive evidence from linked ELSA - NI data” is published on 10th August 2010.
- The research was conducted on behalf of DWP by the Institute for Fiscal Studies. The report authors are Antoine Bozio, Rowena Crawford, Carl Emmerson and Gemma Tetlow.
- The report is available free on the DWP website: http://www.dwp.gov.uk/asd/asd5
- The research is based on the analysis of a unique UK dataset, which combines information from household survey data on individuals’ circumstances at or near retirement with information on their lifetime earnings from administrative records, to look at how retirement incomes relate to lifetime earnings. The English Longitudinal Study of Ageing (ELSA) has interviewed a sample of individuals aged 50 and over every other year since 2002. Respondents to the survey are also asked for permission to link to their National Insurance (NI) records, which are held by HM Revenue and Customs. This research uses data from the first (2002-03) wave of the survey, for respondents who gave consent for their records to be linked.