VoxEU Column Labour Markets Productivity and Innovation

Age, wage, and productivity

Ageing populations are a concern for many developed countries, with increasing dependence on the working population expected. Despite this, there is relatively little research on how productivity changes with age. This column argues that while older people do not run as fast, there is no evidence of a mental productivity decline and little evidence of an increasing pay-productivity gap. The negative effects of ageing on productivity should not be exaggerated.

Over the coming decades, European countries will experience a steep increase in the share of elderly people and a steep decline in the share of people of prime working-age. The number of workers retiring each year will increase and eventually exceed the number of new labour market entrants. The ratio of older inactive persons per worker could rise to almost one older inactive person for every worker by 2050 (OECD 2006).

Population ageing occurs because birth rates are low and people are living longer. Since 1960 life expectancy at age 65 has increased from 13 to 17 years for men and from 15 to 20 years for women. Not only is the labour force ageing, the length of working lives has been declining because workers are retiring earlier than they used to.

With an ageing labour force, the labour market position of older workers is a matter of policy concern. Currently, in many countries older workers are not very likely to lose their job but once they have lost their job they need a long time to find a new one. This situation is often attributed to the gap between wage and productivity, i.e. older workers having a wage that is higher than their productivity. At their current employer, older workers are protected by employment protection legislation including seniority rules. But once older workers become unemployed, employers are reluctant to hire them.

Surprisingly little is known about this relationship. Most employers – and probably most employees – seem to believe in a rule of thumb that average labour productivity declines after some age between 40 and 50. This assumption is so common that few attempts have been made to gather supporting evidence: “why bother to prove the obvious?” (Johnson 1993).

It is not easy to establish the relationship between age and productivity. Productivity is difficult to measure at the level of the individual since it is usually a group phenomenon. Since a group of workers usually consists of workers of different ages the relationship between age and productivity is not straightforward.

Physical productivity

Only on rare occasions is it possible to establish the productivity of individuals. A well-known example is physical productivity in sports contests. In recent research I have analysed the results from an amateur 10km run in the Netherlands (Van Ours 2010). The data refer to the period 1998-2008. The upper part of Figure 1 gives an overview of the observations showing a tendency for the speed to go down with age but at any given age there is a huge variation in average speed. The lower part of Figure 1 presents the average speed by age group showing the average speed goes down from more than 15 kilometres per hour (km/h) for runners younger than 25 to about 13 km/h for participants aged 40. After 40, the average speed hardly drops. Taking into account differences in running ability, the average drop in running speed is 0.6% per year for men and 0.4% per year for women. So, physical productivity declines with ageing – but not a lot.

Figure 1. Running 10 km, 1998-2008

a. All data

b. Average speed per age

Cognitive abilities

Productivity may change over the life cycle because cognitive abilities change with age. To get some idea about this relationship I have studied how publishing in economics journals by members of the Department of Economics of the Tilburg School of Economics is related to their age (Van Ours 2010). To establish a publication score, impact factors of journals are used. The top part of Figure 2 gives a graphical representation of the available information. As shown there is a lot of variation in publications. There are many years for which individual economists have no publication at all. But there are also several observations of individuals who had a publication value of more than 20 within one year. The bottom part of Figure 2 shows average publication scores by year. Apart from publications being a bit lower below age 35 there is no obvious age pattern in these annual publication scores. From an analysis in which time-invariant individual characteristics are taken into account it appears that productivity in publishing increases with age up to age 50 and stays constant after that.

Figure 2. Publishing in economics journals, 1977-2008

a. All data

b. Average publication score by age

Firm level relationship between age, wage and productivity

Recent studies on the relationship between age, wage, and productivity using matched worker-firm panel data are inconclusive about whether or not there is a pay-productivity gap for older workers. To study the age related pay-productivity gap, Lenny Stoeldraijer and I have used matched worker-firm data from Dutch manufacturing firms over the period 2000-2005 measuring productivity as value added per worker (van Ours and Stoeldraijer 2010). The results from a pooled time series – cross-section analysis are shown in Figure 3a. This graph indicates that from age 40 onwards productivity goes down while wage costs do not.

But this analysis is incomplete.

  • First, it may be that there are time-invariant differences in productivity between firms with a young workforce and firms with an older workforce that are unrelated to the age structure of these firms.
  • Second, it may be that changes in age composition are not exogenous to changes in productivity.

It could be that there is a negative productivity shock, which induces firm to fire young workers, causing the average age of the workforce to increase. The negative productivity shock might then seem to be due to the increase in average age of the workforce whereas in fact there is an exogenous explanation for this correlation.

As shown in Figure 3b, after accounting for time-invariant differences between firms and accounting for potential endogeneity the pattern of the squeezed productivity-pay gap disappears. In fact at higher ages productivity seems to increase more than wage costs. Nevertheless, the patterns in Figure 3b are estimates with a lot of imprecision so the main conclusion from this graph is that the age profiles for productivity and wage costs are not so different.

Figure 3. Age profiles of productivity and wage costs (in 1000 Euros per worker) – Manufacturing

a. Pooled time series – cross section

b. After accounting for time-invariant differences between firms and potential Endogeneity of the age structure

So what?

To the extent that running performance represents physical productivity Figure 1 presents evidence of a productivity decline after age 40. To the extent that publishing in economics journals represents mental productivity Figure 2 shows that there is no evidence of a productivity decline, even after age 50. Figure 3 shows that when measured at the firm level there is little evidence of an increasing pay-productivity gap at higher ages of the workforce. These empirical findings are limited to the extent that they are based on Dutch data focusing on single dimensions of productivity. Running is used as an example of physical fitness, publishing as an example of mental ability. Both samples used in the analysis concern small groups that are most likely not representative for the Dutch labour force.

Despite the limitations of the empirical analysis some conclusions can be drawn. My main conclusion is that the potential negative effects of ageing on productivity should not be underestimated; they should not be exaggerated either.

Moreover, there is no need to worry too much about age-related productivity declines or an age related pay-productivity gap – pay and productivity seem to go hand-in-hand as workers grow older. Maybe increasing firm-specific knowledge and experience is responsible for this.

Nevertheless, the labour market position of older workers will remain an area of policy concern. It remains the case that once older workers become unemployed they lose the firm-specific human capital and older unemployed are less likely to find new work.


Johnson, Paul (1993), “Ageing and European economic demography”, in: Johnson, Paul and Klaus F Zimmermann (eds), Labour markets in an ageing Europe, Cambridge University Press, Cambridge.

OECD (2006), “Live longer, work longer”, Paris.

Van Ours, Jan C (2010), “Will you still need me when I’m 64?”, De Economist, forthcoming.

Van Ours, Jan C and L. Stoeldraijer (2010), “Age, wage and productivity”, CEPR Discussion paper 7713.

6,509 Reads