Does Machine Learning Offer Added Value Vis-à-Vis Traditional Statistics? An Exploratory Study on Retirement Decisions Using Data from the Survey of Health, Ageing, and Retirement in Europe (SHARE)

Do machine learning algorithms perform better than statistical survival analysis when predicting retirement decisions? This exploratory article addresses the question by constructing a pseudo-panel with retirement data from the Survey of Health, Ageing, and Retirement in Europe (SHARE). The analysis consists of two methodological steps prompted by the nature of the data. First, a discrete Cox survival model of transitions to retirement with time-dependent covariates is compared to a Cox model without time-dependent covariates and a survival random forest. Second, the best performing model (Cox with time-dependent covariates) is compared to random forests adapted to time-dependent covariates by means of simulations. The results from the analysis do not clearly favor a single method; whereas machine learning algorithms have a stronger predictive power, the variables they use in their predictions do not necessarily display causal relationships with the outcome variable. Therefore, the two methods should be seen as complements rather than substitutes. In addition, simulations shed a new light on the role of some variables—such as education and health—in retirement decisions. This amounts to both substantive and methodological contributions to the literature on the modeling of retirement.

Using HRS-ONET Linked Data to Study Subjective and Objective Mismatch Between Work Demands and Capacity at Older Ages

Mismatch between demands of work and workers’ ability to meet those demands may play an important role in retirement decisions. This presentation extends earlier work using Health and Retirement Study data linked to O*NET to develop measures of discrepancy between individual’s own reports of physical and mental abilities and 1) their perceptions of the physical and mental demands of their jobs and 2) O*NET ratings of the physical and mental demands of their jobs. In particular, we utilize newly available linked information using 2010 Census codes and 2019 O*NET ratings that reflect more current jobs. We then examine the impact of each type of mismatch (subjective and objective) on retirement timing. Overall, we find a stronger connection between subjective mismatch relative to objective mismatch. We discuss implications of this finding in terms of the value of the O*NET linkage and potential interventions aimed at extending working lives for positive aging.

The Role of Occupational Switching on Retirement of American Midlife Workers

An increasing number of midlife Americans are financially unprepared for retirement. This is a problem because of the increasing life expectancy that prolongs the need for financial resources. One way to resolve this problem is to postpone full retirement by having bridge jobs that provide more time to work and accumulate retirement savings. While having a bridge job means numerous labor market behavior such as working for different employers and reducing work hours and intensity, there is a limited number of studies focused specifically on how switching occupations can contribute to retirement decisions with a longer time frame. This study investigated the association between occupational switching and retirement patterns of American midlife workers aged between 50 to 59 years using the Health and Retirement Study longitudinal data from 2004 to 2016, Occupational Information Network data, and American Community Survey from 2003 to 2016. The changes in occupational demandingness before and after switching occupations were reflected by using mover design event study regression with fixed effects. In general, occupational switching is associated with later retirement until two to three years after switching occupations for both genders, yet this association varies by the directions of the change in occupational demandingness.

Introduction to special issue on institutional influences on retirement, health and well-being

Countries make differing policy choices. They can serve as a scientific laboratory for drawing lessons on the policy paths to follow or to avoid and the consequences of those institutional choices on individuals at older ages. In this special issue we bring together six articles that evaluate the influence of institutions on retirement decisions, health and well-being of older adults using common data that have emerged with the international network of health and retirement studies to study key life outcomes such as health, work, and lifecycle transitions at older ages.

The Effect of Pension Subsidies on the Retirement Timing of Older Women

I estimate the effect of additional pension benefits on women’s retirement decisions by examining a German pension subsidy program. The subsidies have a kinked relationship with the recipients’ past pension contributions, creating a sharply different slope of benefits for similar women on either side of the kink point. I find that a 100 euro increase in the monthly benefit induces female recipients to claim their pensions six months earlier. Recipients also adjust their labor supply by using unemployment insurance as a stepping stone to retirement and by reducing time spent in marginal employment. A back-of-the-envelope calculation suggests that the ratio of behavioral to mechanical costs for this subsidy program is 0.25, which is smaller than that of many other income support programs.

Factors Influencing Retirement Decisions among Blue-Collar Workers in a Global Manufacturing Company—Implications for Age Management from A System Perspective

The maintenance of older workers and determining the appropriate age for retirement are growing issues related to the fact that fewer people, still active in working life, have to provide for more non-working people due to increased life expectancy. As a result, retirement age has started to rise in many countries, and employers need to find ways to maintain an older and healthy work force, not least to avoid the loss of important experience. The aim of the current study was to increase the knowledge of factors influencing the retirement decisions among blue-collar workers in different national settings. A survey and semi-structured interviews were conducted with a sample of 100 blue-collar workers in Sweden, the Netherlands, and France, aged 55 years and older, within a global manufacturing company. Based on the results, implications for companies’ age management strategies were discussed from a system perspective. Factors contributing to both retirement and to a prolonged work life were found on individual, organisational, and societal levels. This indicates the importance of a system perspective when planning for age management interventions.

Migration, work, and retirement: the case of Mexican-origin populations

Mexico and the United States both face rapid population aging as well as older populations with high poverty rates. Among the most vulnerable populations of retirement age in either nation are Mexican immigrants to the United States. This work uses data from the U.S. Health and Retirement Study and the Mexican Health and Aging Study to assess retirement decisions among persons born in Mexico and working in either nation as well as such decisions by non-Hispanic Whites in the United States. Social security system incentives matter for the retirement of Mexican immigrants in the U.S. but not for return-migrants in Mexico.