The consequences of a rising proportion of older people have long been seen as a possible macroeconomic problem with adverse implications for growth, real interest rates, public finances, and inflation (Baldwin and Teulings 2014, Bloom 2019, Goodhart and Prahan 2020). The sensitivity of economic growth to an ageing society is apparent in the fact that, for example, two-thirds of the increase in US employment between 1990 and 2020 consisted of workers aged between 50 and 74.
More recently, the US and the UK have both experienced rising inflation and tight labour markets, but employment rates have failed to reach pre-pandemic levels. According to the IMF (Pizzibelli and Shibata 2022), the main reason for this is older workers withdrawing from the labour force. Looking forward, the participation rates of older workers are likely to be a defining issue for economic success (OECD 2019).
Figure 1 US employment growth by age, 1990-2020 (millions)
Raising employment at older ages requires either greater willingness to work on the part of older workers and/or greater willingness on the part of firms to hire them. A range of measures can contribute to this objective: increasing incentives to work through social security reforms (Rogerson and Wallenius 2022), improving the health of older workers (Coile et al. 2017) and their education (Burtless 2013), greater use of automation (Acemoglu and Restrepo 2022), and measures tackling age discrimination (Neumark et al. 2019).
An especially important lever is to create ‘age-friendly’ jobs for which older workers have a comparative preference or advantage. For example, if available jobs become less strenuous for older workers or enable them to use their skills more effectively, then they will have greater willingness to work and may also face greater remuneration. Reflecting this current OECD policy is to “promote employability of workers throughout their working lives” by “creating a supportive age-friendly working environment” (OECD 2019).
Age-friendly jobs also have an additional advantage: the prospect of minimising the impact of rising older employment on the rest of the workforce. Sorting into these jobs by older people will lessen the downward pressure on the wages of younger workers through reducing direct competition.
An “age-friendly working environment” requires that older workers have distinct skills and preferences – a result that receives considerable support (e.g. Hudomiet et al. 2019, Jetsupphasuk and Maestas 2019, Ameriks et al. 2020). Older workers prefer occupations that provide flexible scheduling, reduced job stress, less demanding cognitive and physical work, less commuting time, and the opportunity to work from home. While these characteristics are desirable for all workers, our evidence, as well as survey results, suggest they are particularly attractive to older workers.
In recent research (Acemoglu et al. 2022), we build on these insights to construct an index of age-friendly jobs. Specifically, we use Natural Language Processing (NLP) methods to construct an occupational ‘Age-Friendly Index’ (AFI) by combining information on the preferences and aptitudes of older workers together with occupational characteristics.
Our starting point is detailed textual descriptions of 873 occupations listed in the O*NET database. These cover more than 244 job attributes broken down across abilities, interests, work values, work styles, skills, knowledge, work activities and work context, and the relative importance of a total of 16,804 occupational tasks.
We also have textual definitions of the nine occupational characteristics that Jetsupphasuk and Maestas (2019) find most appealing for older workers. Using NLP methods, we examine the degree of correlation between these definitions and descriptions, weighting the age-friendly characteristics by the absolute and relative preferences of older workers. The result is a ranking of occupations by their age-friendliness and a measure of how this has changed over time.
We perform two tests to confirm the validity of our index. First, we compare age-friendliness rankings from survey participants with those produced by our AFI: the two cohere strongly with a correlation coefficient of 0.83. Secondly, we use regression analysis and confirm our AFI predicts which occupations older workers sort into.
Figure 2 Shifts in employment share by decile of the Age-Friendly Index (AFI)
Figure 2 depicts the change in employment shares between AFI deciles (based on 2020 rankings) and shows a big shift into above-average age-friendly jobs. In total, these increased by 49 million – more than the overall increase in employment of 42 million. Around 90% of this was due to within-occupation effects, with around three-quarters of occupations becoming more age-friendly, and only 10% due to faster growth in the most age-friendly occupations.
These statistics point to the US occupational structure becoming more supportive of older workers. But closer examination reveals that older workers were not the main beneficiaries. Top quartile age-friendly jobs increased by 33 million between 1990 and 2020 but only 15 million went to workers aged over 50.
Table 1 shows that older workers are more likely to be in the most age-friendly jobs and in general experienced the largest increase in the likelihood of working in age-friendly occupations. But it also shows two surprising features. Many of the new age-friendly jobs have been taken up by young women and young graduates, and older male non-college graduates have not gained much. By 2020, the latter were no more likely to be in age-friendly jobs than younger male non-college graduates.
Table 1 Probability (%) in a top quartile age-friendly job
Note: Table shows proportion of different demographic groups employed in top quartile age-friendly occupations. Quartiles are defined by numerical values based on 2020 AFI.
This represents a puzzle. If comparative advantage is operating, older workers should accept lower wages and be disproportionately represented in age-friendly occupations. Table 1 suggests either these comparative advantage forces are weak or other factors are important.
One reason why comparative advantage may be weak is that the characteristics that make occupations age-friendly appeal strongly to other demographic groups. For instance, many of the features of our age-friendliness index overlap with those Goldin (2014) identifies as appealing to female workers and supporting gender equality.
Similarly, cognitive non-physical work with autonomy and flexible scheduling appeals to college graduates. In other words, age-friendly jobs may just be good jobs, especially for more educated and female workers. Even though in relative terms these occupational characteristics are preferred by older workers, the difference may be small. This dilutes the impact of age-friendly jobs in supporting older employment and in limiting the impact on younger workers.
Comparative advantage may also be stymied due to other labour market frictions and distortions. For instance, if there is limited mobility among older workers (due to accumulation of firm-specific human capital, age discrimination in hiring, or fixed costs of job transitions), then they may not benefit from rapid improvements in age-friendliness in fast-growing sectors.
Another possibility is firms and workers engage in rent-sharing so that firms prefer higher productivity workers regardless of comparative advantage. Under this scenario, if college graduates of all ages like age-friendly jobs, then they will be major beneficiaries. Support for this interpretation comes from the fact that the most age-friendly jobs tend to be those with the highest wages and have benefited from the highest wage growth. This form of rent-sharing would also explain why non-college older workers may be rationed out of age-friendly jobs and prevented from moving out of the more physically demanding jobs with difficult work conditions they currently occupy.
The challenge is that male non-college graduates are disproportionately employed in industries with low levels of age-friendliness – around one in three are working in manufacturing and construction. The situation is made worse by the fact that out of all older workers, this group have experienced the largest increase in labour force participation, have the lowest level of health and have witnessed the worst declines in health. That raises the question of whether more age-friendly employment for this group would help to maintain better health.
Our results show the US labour market in the last 30 years has been highly supportive of rising employment among older workers because of a substantial increase in age-friendly occupations. But a large proportion of this increase has been to the benefit of other groups and not just older workers.
These findings have two major implications. The first is that just promoting the creation of age-friendly jobs, may be insufficient for promoting employment and income growth for older workers and shielding younger employees from the impact.
The second is that age-friendly policies need to be designed with other labour market imperfections in mind. There are too many differences among older workers (especially graduates and non-graduates) and similarities between younger and older workers (again especially among graduates) for purely age-based policies alone to have major impact.
More targeted policies aimed particularly at male non-graduates and policies that tackle underlying market imperfections – such as the cost and difficulties of job transition among older workers – are promising areas to explore.
Acemoglu, D, N Mühlbach and A J Scott (2022), “The Rise of Age-Friendly Jobs”, NBER Working Paper 30463 and forthcoming in Journal of the Economics of Ageing.
Acemoglu, D and P Restrepo (2022), “Demographics and Automation”, Review of Economic Studies 89(1): 1-44.
Ameriks, J, J Briggs, A Caplin, N Lee, M Shapiro and C Tonetti (2018), “Older Americans would work longer if jobs were flexible”, VoxEU.org, 29 October.
Baldwin, R and C Teulings (2014), Secular Stagnation: Facts, Causes and Cures, CEPR Press.
Bloom, D E (2019), Live Long and Prosper? The Economics of Ageing Populations, CEPR Press.
Burtless, G (2013), “Can Educational Attainment explain the rise in Labor Force Participation at Older Ages?”, Centre for Retirement Research, Boston College.
Coile, C, K Milligan and D A Wise (2017), “Health Capacity to Work at Older Ages”, in Social Security Programs and Retirement around the World, University of Chicago Press.
Goldin, C (2014), “A Grand Gender Convergence: Its last chapter”, American Economic Review 104(4): 1091–1119.
Goodhart, C and M Pradhan (2021), “The Great Demographic Reversal: Ageing Societies, Waning Inequality and an Inflation Revival”, VoxEU.org, 17 February.
Hudomiet, P, M Hurd, A Parker and S Rohwedder (2019), “The Effects of Job Characteristics on Retirement”, NBER Working Paper.
Jetsupphasuk, M and N Maestas (2019), “What do older workers want?”, in D E Bloom (ed) Live Long and Prosper? The Economics of Ageing Populations, CEPR Press.
Neumark, D, I Burn and P Button (2019), “Is It Harder for Older Workers to Find Jobs? New and Improved Evidence from a Field Experiment”, Journal of Political Economy 127(2): 922–970.
OECD (2019), Working Better with Age, OECD Publishing.
Pizzibelli, C and I Shibata (2022), “Why Jobs are Plentiful While Workers are Scarce”, IMF.
Rogerson, R and J Wallenius (2022), “Shocks, Institutions and Secular Changes in Employment of Older Individuals”, in NBER Macroeconomics Annual 2022, University of Chicago Press.