Retirement Reforms: Occupational Strain and Health

• DOI: http://doi.org/10.1111/irel.12242
• Author: Kantha Dayaram,Alistair McGuire
• Date: 01 July 2019
• Published date: 01 July 2019

Retirement Reforms: Occupational Strain and

Health*

KANTHA DAYARAM and ALISTAIR McGUIRE

A concurrent increase in the demand for state age pensions and health care has

led to reforms in delaying retirement. We employ thirteen waves of longitudinal

data to examine the mental and physical health effects of Australian men and

women at “early”and “traditional”retirement. We use before and after propensity

score matching (PSM) estimates between treatment and control groups of retired

and not retired individuals aged 60 and 65 years. The results indicate a negative

health effect according to occupational strain for both genders but a positive men-

tal health effect for retirees with access to self-funded retirement.

Introduction

Retirement reforms, population aging, and increased life expectancy com-

bined with low fertility rates in developed economies are critical issues that

impact workforce planning and labor dynamics. Nowhere is the aging popula-

tion more acutely observed than in the assessment of individual retirement

decisions. Such decisions are integrally bound to legislative agreements relat-

ing to the labor force, and the retirement income expectations that individuals

have. Retirement decisions are also affected by other individual characteristics;

however, Roberts, Rice, and Jones (2010) found health to be a key determi-

nant of early retirement. Rohwedder and Willis (2010) noted that the choice of

early retirement had negatively correlated with an individual’s mental health as

cognitive ability declined. Conversely, Coe and Lindeboom’s study (2008)

found that retirement timing had no impact on the physical and psychological

health outcomes of men. However, they acknowledged that early retirement

might have a positive, although temporary, impact on the self-reported health

of individuals. Calvo, Sarkisian, and Tamborini (2013) similarly found that

*The authors’affiliations are, respectively, Curtin University, Western Australia, Australia. E-mail:

k.dayaram@curtin.edu.au; and London School of Economics, London, UK. E-mail: A.J.McGuire@lse.ac.uk.

INDUSTRIAL RELATIONS, DOI: 10.1111/irel.12242. Vol. 58, No. 3 (July 2019). ©2019 The Regents of the

Universit y of Calif ornia Published by Wiley Periodicals, Inc., 350 Main Street, Malden, MA 02148, USA,

and 9600 Garsington Road, Oxford, OX4 2DQ, UK.

522

employees who retire early experience a number of poor health outcomes relat-

ing to self-reported physical and emotional health, whereas those that choose

to continue employment beyond the expected retirement age are generally not

associated with subjective health penalties. Although these studies have some

consistencies in their findings, there are also differing estimates on the health

effects with retirement subject to demographics, context, and measurement

bias. Notably, self-reported health measures carry endogeneity and measure-

ment error concerns; however, Roberts, Rice, and Jones (2010) maintained that

studies that employed subjective and objective measures yielded similar results

when predicting retirement behaviors. They further argued that because behav-

iors are being examined, an individual’s subjective assessment of their health

is more relevant rather than their objective health status. Disney, Emmerson,

and Wakefield (2006) made a similar claim but suggested using proxies for

health shocks that affect retirement behavior, as a way to address reporting

bias. We follow similar approaches with self-reports and examine health uti-

lization rates as proxy measures, as well as using the matching of retired and

nonretired individuals to overcome bias and a test of “double robustness”

(Athey et al. 2017).

While most of these studies examined the association of retirement on phys-

ical and or mental health there has been less investigation of the effects of

retirement decisions by gender, at different age categories and health utilization

rates that underlie the relationship between retirement and health. In this study,

we investigate the effects of retirement on health for males and females in two

subpopulations, at age 60 (considered early retirement) and at age 65 (consid-

ered traditional retirement) using Australian data. Although Australia has no

mandatory retirement age, new policy reforms define age eligibility criteria for

access to publicly provided pension income at age 67 and concessional tax

treatment at age 60 for access to employer-based superannuation funds. As the

majority of retirement decisions relate to the age criteria for these retirement

income access policies, most Australians retire at either 60 (early retirees) or

65 given that until July 2017 this was the age at which the public pension pro-

vision became available. We are specifically interested in whether retirement

differentially affects health in “early”retirees and “traditional”retirees.

We initially investigate the health effects on both genders and subpopula-

tions, using ordinary least square (OLS) regressions, subsequently testing for

measurement bias from self-assessed health and for potential endogeneity

using a propensity score matching (PSM) approach. Through matching we

reduce the covariate imbalance in the data that then reduces the selection bias.

This is a novel “double-robustness”method that employs a hybrid approach to

examine the mental and physical health characteristics of the treatment group

comprising retirees at ages 60 and 65 and a control group comprising

Retirement and Health Effects / 523