Ageing Workforces, Ill‐health and Multi‐state Labour Market Transitions*
| DOI | http://doi.org/10.1111/obes.12379 |
| Published date | 01 February 2021 |
| Author | Xueyan Zhao,Mark N. Harris,Eugenio Zucchelli |
| Date | 01 February 2021 |
199
©2020 The Department of Economics, University of Oxford and JohnWiley & Sons Ltd.
OXFORD BULLETIN OF ECONOMICSAND STATISTICS, 83, 1 (2021) 0305–9049
doi: 10.1111/obes.12379
Ageing Workforces, Ill-health and Multi-state Labour
Market Transitions*
Mark N. Harris,†Xueyan Zhao‡ and Eugenio Zucchelli§,††
†Curtin University, Perth, Western Australia 6845, Australia
‡Monash University, Melbourne VIC 3800, Australia
§Lancaster University, Lancaster, LA1 4YG, UK (e-mail: e.zucchelli@lancaster.ac.uk.)
††Madrid Institute for Advanced Study (MIAS), Universidad Aut ´onoma de Madrid, Madrid,
28049, Spain (e-mail: eugenio.zucchelli@uam.es)
Abstract
We provide novel evidence on the effects of ill-health on the dynamics of labour state
transitions by considering retirement as mobility between full-time work, part-time work,
self-employment and inactivity. We employ a dynamic multi-state model which accounts
for state dependence and different types of unobservables. Our model allows for both
individual heterogeneity and labour-state gravity as well as correlations between labour
market states. We estimate this model on rich longitudinal data from the Household, Income
and Labour Dynamics in Australia Survey. We find that both ill-health and health shocks
greatly increase the probability of leaving full-time employmentand moving into inactivity.
Simulated dynamic trajectories suggest larger impacts of long-term health conditions than
those of a one-off health shock and some evidence of health-drivenretirement pathways via
part-time work and self-employment. Our findings also indicate that the effects of health
changes could be underestimated and the magnitude of true labour market state dependence
overestimated if individual effects or labour dynamic transitions are not accounted for in
the model.
JEL Classification numbers: C23, I10, J24, J2
*We thank ManuelArellano, Alan Duncan, Andrew Jones, Martin Karlsson, David Madden, Pedro Mira, Karen
Mumford, Aurora Ortiz-Nunez, Nigel Rice, Arne Risa-Hole, KarlTaylor, Ian Walker,Jeffrey Wooldridge, Maurizio
Zanardi and attendants to seminars at UniversidadAut ´onoma (Madrid), CEMFI, City (London), Duisburg-Essen and
York (UK) as well as participants to the 2017 International Health Economics Association (iHEA) WorldCong ress
and the 2017 joint European Economic Association/European Meeting of the Econometric Society (EEA/ESEM) for
valuable comments. We also thank the editor and two anonymous referees for their useful suggestions. This paper
uses unit record data from the Household, Income and Labour Dynamics in Australia (HILDA)Sur vey. The HILDA
Project was initiated and is funded by theAustralian Government Department of Families, Community Services and
Indigenous Affairs (FaCSIA)and is managed by the Melbourne Institute of Applied Economic and Social Research
(MIAESR). The findings and views reported in this paper, however, are those of the author and should not be
attributed to either FaHCSIA or the MIAESR. Eugenio Zucchelli gratefully acknowledgessupport from the Tom´asy
ValienteFellowship funded by the Madrid Institute for Advanced Study (MIAS), UniversidadAut ´onoma de Madrid,
and grants SI1/PJI/2019-00326 and H2019/HUM-5793 funded by the Comunidad de Madrid. Wealso acknowledge
support from the Australian Research Council (ARC).
200 Bulletin
I. Introduction
An ageing population poses a threat and a fundamental burden to the sustainability of
any social security system (Gruber and Wise, 2009; Bloom et al., 2010). This demo-
graphic change, combined with the generosity of pension systems and disability benefit
schemes in the majority of developed economies, also has profound consequences for
the labour markets (B¨orsch-Supan, 2003; D’Addio et al., 2010; ILO, 2016). According
to the United Nations (2017), the global population aged 60 years or over in 2017 more
than doubled since 1980 and is predicted to double again by 2050. In Australia, a coun-
try with one of the longest life expectancies in the world (OECD, 2017), the number
of working age people between 15 and 64 years for every person aged 65 or over has
fallen from 7.3 people in 1974–75 to 4.5 people in 2015. By 2054–55, this proportion
is projected to be nearly halved again to 2.7 people (Commonwealth of Australia, 2015).
Early exits from the labour market and the increased fragmentation of individuals’labour
market trajectories also highlight the need for re-examining the determinants of individ-
uals’ labour market choices, particularly in the latter part of the life cycle. Thus, identi-
fication of both determinants and trajectories of labour transitions at older ages would
allow governments and policy makers to formulate policies to avoid the loss of con-
tribution from a potentially active labour force. Importantly, although the literature has
established that ill-health is strongly associated with labour market decisions, including
retirement choices (e.g. Disney et al., 2006; Lindeboom and Kerkhofs, 2009; Lindeboom,
2012; Blundell et al., 2016b), multiple health-driven pathways into retirement need to
be considered to fully capture the complexity of the labour market transitions of older
workers.
The main objective of this paper is to explore the effects of ill-health and health shocks
on individuals’ labour market transitions of older workers by employing a highly flexi-
ble dynamic multi-state panel data model with several novel features. More specifically,
we consider retirement as a multi-state process and examine the effects of health and
health shocks on the mobility between full-time employment, part-time employment,self-
employment and inactivity, using a dynamic DOGIT (Gaudry and Dagenais, 1979) Ordered
Generalised Extreme Value (DOGEV) model (Fry and Harris, 2005). This model extends
the conventional multinomial logit (MNL) model to further allow for the error terms of
the utilities of some of the choices to be correlated, relaxing the undesirable independence
of irrelevant alternatives (IIA) restrictions, while still being computationally much sim-
pler than alternative models such as the multinomial probit (MNP). Our specification also
jointly accounts for labour market state dependence, individual-level unobserved hetero-
geneity, gravity to particular labour states due to choice heterogeneity, as well as potential
endogeneity of self-reported health. In this way,we can more precisely distinguish between
the effects of past employment experience,health and other key observable characteristics,
as well as further unobservable individual and choice-specific effects on employment be-
haviour.We estimate this model using a sample of older individuals drawnfrom the first 13
waves (2001–13) of the Household, Income and Labour Dynamics in Australia (HILDA)
Survey (Watson and Wooden, 2012). Given its wealth of both health and work-related
variables, HILDA is uniquely suited to study the relationship between health and labour
supply.
©2020 The Department of Economics, University of Oxford and JohnWiley & Sons Ltd
To continue reading
Request your trial