AdobeStock_138398039.jpeg
VoxEU Column Labour Markets Welfare state and social Europe

Technology, labour market institutions, and early retirement

Across OECD countries, promoting longer working lives is an important policy objective for mitigating fiscal pressures from population ageing. This column uses data from Finland to examine how technological change and access to early retirement pathways reinforce each other in pushing older workers out of employment. It finds that the probability of leaving employment is higher for individuals in occupations with higher automation risks and increases faster for individuals closer to the eligible age for early retirement pathways.Reforms that tighten access to such pathways substantially extend the working lives of older workers exposed to high automation risks, but have little effect on old workers exposed to low automation risks.

Across OECD countries, promoting longer working lives is an important policy objective for mitigating fiscal pressures from population ageing, notably increasing pension and healthcare expenditures. One barrier to increasing employment of older workers is workers engaged in codifiable, routine tasks are prone to being displaced by computers and robots (Gentile et al. 2020), a trend that may have been accelerated by the COVID-19 pandemic (Baldwin 2020, Chernoff and Warman 2021). Older workers are particularly exposed to this risk because, with shorter remaining working lives, they have weaker incentives to acquire new skills that would allow them to switch to tasks that are less likely to be automated. They may instead choose to retire early when facing rapid technological change (Ahituv and Zeira 2011, Hægeland et al. 2007). Another barrier is that a number of OECD countries have in place institutions that encourage early retirement, such as exceptional entitlements or looser criteria for unemployment and disability benefits reserved for older workers. These two factors reinforce each other in pushing older workers out of employment. That is, older workers highly exposed to technological change are more likely to exit the labour market when they gain access to early retirement pathways. Alternatively, those having access to early retirement pathways are more likely to use them when they are more exposed to technological change. 

In a recent paper (Yashiro et al. 2021), we explore such a complementarity for Finland. The country ranks as the top performer in adoption of digital technologies in socio-economic activities according to the European Commission's Digital Economic and Society Index (DESI). It also displays a considerably lower employment rate for older individuals than its Nordic peers (OECD 2020). The latter owes importantly to early retirement through the so-called unemployment tunnel, which is the combination of the entitlement to regular unemployment benefit up to 500 working days, and the extension of unemployment benefit until the retirement age, reserved for the unemployed aged 61 or more at the time their regular unemployment benefit expires. The inflow into unemployment increases significantly about two years before the eligible age for the extension of unemployment benefit, which is the earliest timing individuals can access the unemployment tunnel. Several reforms raised the eligibility age for this extension in the past, delaying the timing for entering the unemployment tunnel and thereby lengthening working lives each time (Kyyrä and Pesola 2020).1 The government of Finland recently decided to abolish the extension in 2025 for individuals born in 1965 or after.  

We combine Finland’s rich employee-employer database (FOLK) and three different OECD datasets capturing exposure to digital technologies at the occupation level: the risk of automation (Nedelkoska and Quintini 2018), the intensity of routine tasks (Marcolin et al. 2016), and the intensity in use of ICT skills (Grundke et al. 2017), all constructed from individual-level data of the OECD Survey on Adult Skills (PIAAC). Figure 1 compares the inflow into unemployment between individuals in occupations with higher-than-average automation risks and those with lower automation risks during the period 2007-2017, when the eligibility age for the extension of unemployment benefit varied between 59 to 61 (and thus the timing for entering the unemployment tunnel ranged between 57 and 59). The unemployment risk is clearly higher for individuals in occupations with higher automation risks. Moreover, the unemployment risk increases faster for these individuals in their late-50s, as implied by the steeper slope of schedule. The difference in the unemployment risk between individuals with high and low automation risks is largest at age 59, when all individuals can access the unemployment tunnel.  

Figure 1 Unemployment inflow by risk of automation, 2007-2017

 

Note: This figure compares the inflow into unemployment of individuals aged 50 and over between occupations with high automation risks and those with low automation risks, during the period 2007-2017. The red line corresponds to individuals in occupations with high automation risks while the blue line corresponds to those in occupations with lower automation risks.
Source: Authors’ computations.

We then estimate the probability of an individual aged 50 or over exiting employment after two consecutive years of employment as the function of exposure to digital technologies, a dummy variable indicating the individual’s access to the unemployment tunnel, and their interaction.2  

We find that: 

  • An individual aged 50 or above in occupations exposed to one standard deviation higher than the average risk of automation faces a 1.1 percentage point higher probability of exiting employment every year, if he or she does not have access to the unemployment tunnel. 
  • This probability is 2.2 percentage points higher if the individual has access to the tunnel. 
  • Gaining access to the unemployment tunnel increases the exit probability of an individual exposed to an average level of automation risks by 1.8 percentage points. 
  • The overall impact of higher automation risks and the unemployment tunnel therefore amounts to 4 percentage points, which implies an 80% increase in the probability of exiting employment for individuals aged 57-58. 
  • We obtain similar results when using other indicators to capture the exposure to digital technologies, such as intensity in routine tasks or ICT skills. 

Using the estimated coefficients, we simulate the impact of reforms that tighten access to the unemployment tunnel. Figure 2 illustrates that such reforms substantially extend the working lives of older workers exposed to high automation risks but have little effect on old workers exposed to low automation risks.

Figure 2 The probability of continuous employment under different reform scenarios

 

Note: This figure plots the average probability of remaining employed from age 50 onwards for two groups of older workers, one subject to higher than average automation risks (red lines) and another subject to lower than average risks (blue lines). Three reform scenarios are considered: (i) backtracking: the unemployment tunnel (UT) is made available earlier, at the age 57, as it was during 2012-2014; (ii) it is made available at age 59 as it is now, and (iii) extended unemployment benefit is abolished. See Yashiro et al. (2021) for detailed information.
Source: Authors’ computations.

These findings underscore the importance of reforms that tighten access to institutionalised pathways to early retirement in ensuring the inclusion of older workers in the future of work. While previous policy discussion often emphasised boosting lifelong learning opportunities, older workers will only have weak incentives to take up such opportunities if these early retirement pathways are left open. The recent decision by the Finnish government to abolish the extension of unemployment benefit is likely to encourage older workers more exposed to technological change to work longer and participate in upskilling opportunities. Employers will also have stronger incentives to provide more training to older employees, who can be expected to work until their mid-60s. There is, however, a need for targeted measures to increase the employability of older workers most affected by the reform through highly tailored training programmes (OECD 2020). It is also important to encourage older workers to participate in these programmes through better identification of their training needs, for instance through mid-career review or career guidance services, and certifying skills acquired through training (OECD 2019). 

Policymakers should also ensure that measures to protect jobs and businesses under the COVID-19 crisis are not used as a fast track to early retirement. In Finland, the extensive use of the temporary layoff scheme prevented a large increase in permanent job losses (OECD 2020). However, some layoffs may have targeted older workers more exposed to technological change with access to early retirement pathways. In our paper, we present preliminary evidence based on high-frequency data on unemployment inflows (including both temporary and permanent layoffs) that older workers exposed to higher than average automation risks experienced disproportionally larger increases in unemployment risks during the economic contraction, particularly at the ages when they can access early retirement pathways.      

Authors’ note: The views expressed are those of the authors and do not necessarily reflect those of the OECD or of the governments of its member countries. 

Reference

Ahituv, A and J Zeira (2011), “Technical progress and early retirement”, Economic Journal 121: 171–193.

Baldwin, R (2020), “Covid, hysteresis, and the future of work”, VoxEU.org, 29 May

Chernoff, A and C Warman (2021) “Down and out: Pandemic-induced automation and labour market disparities of COVID-19”, VoxEU.org, 2 February. 

Gentile, E, S Miroudot, G De Vries and K M Wacker (2020) “Robots replace routine tasks performed by workers”, VoxEU.org, 8 October.

Grundke, R, S Jameti, M Kalamova, F Keslairi and M Squicciarini (2017), “Skills and global value chains: a characterisation”, OECD Science, Technology and Industry Working Papers, 2017/05.

Hægeland,T, D Rønningen and K Salvanes (2007), “Adapt or withdraw? Evidence on technological changes and early retirement using matched worker-firm data”, NHH Dept. of Economics Discussion Papers, No. 22/07.   

Kyyrä, T and H Pesola (2020), “Long-term effects of extended unemployment benefits for old workers”, Labour Economics 62: 101777.

Marcolin, L, S Miroudot and M Squicciarini (2016), “The routine content of occupations: new cross-country measures based on PIAAC”, OECD Science, Technology and Industry Working Papers, 2016/02.

Nedelkoska, L and G Quintini (2018), “Automation, skills use and training”, OECD Social, Employment and Migration Working Papers, 202.

OECD (2020), OECD Economic Surveys: Finland 2020, OECD Publishing, Paris.

OECD (2019), Working Better with Age, Ageing and Employment Policies, OECD Publishing, Paris.

Yashiro N, T Kyyrä, H Hwang and J Tuomala (2021), “Technology, labour market institutions and early retirement: evidence from Finland”, OECD Economics Department Working Papers, 1659.

Endnotes

1 Other early retirement pathways in Finland include more lenient eligibility criteria for disability benefits (including non-medical factors) applied to applicants aged 60 or more, which results in the inflow into disability benefits soaring at age 60. The flexible retirement scheme introduced in 2005 allowing individuals to retire between 63 and 68 also triggered large retirement before the statutory retirement age of 65, even though the financial incentive for the early retirement was weak.

2 We control for the individual’s age, gender, educational attainment, marital status, area of residence, and year fixed effects. The coefficient on the access to the unemployment tunnel is identified by exploiting the past reforms that raised the eligibility age for the extension of unemployment benefit, which introduces exogenous variation in the age at which individuals can access the unemployment tunnel.

1,680 Reads