How long is a South African working life?

Charles Simkins says high unemployment is having a serious effect

How long is a South African working life?

7 February 2017


The length of a working life is of interest because it, along with the average level of wages, determines aggregate lifetime earnings. It also determines the number of payments into contributory private and public social security systems, such as pensions and medical aid. The age at which a person works for the first time, the age at which the person works for the last time, and the extent of non-employment in between should also be considered against the pensionable age and unemployment insurance payments.

Using a simulation technique[1], it is possible to estimate aspects of working lives between the ages of 15 and 64. The Statistics South Africa panel data set constructed from the third and fourth quarters of the 2013 Quarterly Labour Force Survey are the starting point for the simulation. To use demographic terminology, the estimates are ‘period’ estimates rather than cohort estimates. They represent the experience of a hypothetical cohort successively exposed to the conditions in the panel data, rather than that of an actual cohort which would experience conditions at future points in time. An allowance has to be made for mortality and the life table for 2010-2015 from the 2015 United Nations World Population Prospects (2015 revision) is used for the purpose.  


Table 1 sets the estimates of the average age at which people enter employment for the first time, the average age at which they work for the last time, and the average age at which they retire. To qualify as retiree, a person must live for at least one quarter after working for the last time, in contrast to people who die when they are working. It also presents estimates of the number of years of employment, and the number of years of non-employment.

Table 1









Average age at entry into employment



Average age at exit of employment



Working life



Number of years of full-time employment[2]



Number of years of part-time employment



Number of years of employment



Number of years of non-employment



Average age at retirement



Average years lived from 15 to death or 65



Number of years of employment: survivors to 65



One measure of working life is the span between entry into employment and exit from employment for the last time.  On this measure, men have a working life of 31 years and women 28 years.  Another measure excludes the time not employed, in which case men have a working life of 21 years, and women 16 years.   It is the second measure which is relevant for the ability to contribute to social security.  

For instance, pensions are often calculated to give a pension of the number of years of contributions divided by 60 multiplied by average or final earnings.  Even if pensions were fully preserved, the replacement rate (annual pension divided by average or final wage) would be 45% for men and 34% for women.   And the rates would be lower if the part-time employed did not contribute.  Equally, men would be covered for 52% of their lives between 15 and 64 by employment based medical aid and women for 39%.  

These estimates assume that the self-employed as well as employees can participate in contributory pensions and medical aid, and that contributions would be made by part-time workers.  Note also that the average retirement age for men is 2.6 years short of the age at which people qualify for a government pension.  For women, it 5.8 years short.

Table 2 sets out the distribution of working lives (periods of employment only) by their length.

Table 2






Less than 10 years



10-15 years



15-20 years



20-25 years



25-29 years



30-34 years



35 years









40% of men are in employment for less than twenty years and 66% of women.  


The average period for which South Africans are employed is short, not because of unduly late entry into employment or unduly early retirement from it, but because the time between entry and exit spent out of employment between the two is high – 33% in the case of men and 43% for women.

This is mainly a consequence of high unemployment (both official and discouraged workers), but also because of periods of economic inactivity, more extensive among women than among men. Mortality also plays a considerable role.  Just 44% of men aged 15 reach their 65th birthday, and 53% of women.

High non-employment imposes a high level of dependency among adults of working age and it renders reliance on contributory social security inadequate.

Charles Simkins, Head of Research, HSF


[1] Simulation works by tracking the progress from state to state from one time period to the next (in this case, the interval is a quarter of a year) of a large number of individuals (in this case, 2500 men and 2500 women). The initial state and the progress from state to state is determined in each case by comparing a number randomly drawn from a uniform distribution between zero and one and comparing it to the relevant probability. There are some simplifying assumptions in the analysis, mainly the assumption that the state in one period depends only on the state in the single period before it, that survival rates are constant across five year age intervals, and that everyone who lives to age 65 retires on their 65th birthday if they have not already done so. So the results are approximate only and for a hypothetical cohort, but they are accurate enough to support the broad conclusions drawn. 

[2] Note that for a person to be enumerated as employed, he must have done at least an hour of work in the preceding week. This means that those wholly on leave in the preceding week are not counted as employed, even though they remain in post.