Cross section data suggest that the relationship between age and hourly earnings is an inverted-U shape. Evidence from panel data does not necessarily confirm this finding suggesting that older workers may not experience a reduction in earnings at the end of their working life. In this paper we use panel data on males for Great Britain in order to examine why the two types of data provide conflicting conclusions. Concentrating on the over 50s, several hypotheses are examined: overlapping cohorts, job tenure, job-changing, labour supply behaviour and selectivity bias. Cohort and individual fixed effects partly explain the divergent conclusions. However, for fully, year-on-year employed individuals, there is no evidence of earnings decline at the end of working life. We find no role for selectivity due to retirement, although shorter working hours or partial retirement along with job-changing late in life do provide an explanation for why hourly earnings decline for certain older workers. We find no evidence that the process of ageing itself leads to lower earnings as suggested by the cross section profile.
