Health status and labour market outcome: Empirical evidence from Australia

• Date: 01 May 2019
• Published date: 01 May 2019
• DOI: http://doi.org/10.1111/1468-0106.12257

ORIGINAL MANUSCRIPT

Health status and labour market outcome:

Empirical evidence from Australia

Kannika Damrongplasit

1

| Cheng Hsiao

2,3

| Xueyan Zhao

4

1

Chulalongkorn University, Bangkok, Thailand

2

University of Southern California, Los Angeles,

CA, USA

3

Xiamen University, Xiamen, Fujian, China

4

Monash University, Melbourne, Victoria,

Australia

Correspondence

Kannika Damrongplasit, Faculty of Economics,

Chulalongkorn University, Phayathai Road,

Bangkok 10330, Thailand.

Email: kannika.d@chula.ac.th

Funding information

China Natural Science Foundation Grant, Grant/

Award number: #711311008 and #71631004;

Chulalongkorn Economics Research Center;

Australian Research Council, Grant/Award

number: DP0880086

This paper uses eight waves of Australia Household,

Income and Labour Dynamics data to study the issues of

state dependence and the short-run and long-run response

to health shocks on the labour market. We consider six

alternative panel data binary dependent variable models

with different ways of modelling labour market dynamics

and individual heterogeneity. We find that the key results

with regard to labour market dependence and the impacts

of health shocks are sensitive to model specification and

pooling of male and female samples with differences as

large as sixfold. Specification analysis is conducted and

favours the dynamic fixed effects logit model for separate

male and female samples. Methods for evaluating dynamic

response paths to a one-time health shock for binary out-

comes are also suggested and results are presented.

1|INTRODUCTION

The present paper estimates a dynamic fixed effects (FE) binary outcome variable model using a

rich panel data set, and demonstrates how important policy effects can be overestimated using a

dynamic random effect (RE) and other model specifications. The impacts of individuals’health on

their labour market decisions have increasingly been of interest to governments around the world.

Both population ageing and increasing prevalence of chronic conditions in the developed countries

have major economic implications for the labour markets. Chronic conditions such as diabetes, car-

diovascular diseases and mental health diseases continue to affect increasing proportions of the

populations (Zhang, Zhao, & Harris, 2009). This trend, combined with the trend of population age-

ing and increasing availability of disability welfare in many countries, not only places an increasing

burden on the health-care systems around the world, but also poses a significant challenge to the

labour markets due to early exits. Conversely, any medical research or public health strategies aimed

at reducing chronic diseases will also have a positive flow-on effect on the labour market. Therefore,

understanding the link between individuals’health conditions and their labour market outcomes is

crucial for government policy design and for a comprehensive calculation of the burden of diseases.

Received: 25 July 2017 Revised: 16 January 2018 Accepted: 25 January 2018

DOI: 10.1111/1468-0106.12257

Pac Econ Rev. 2019;24:269–292. wileyonlinelibrary.com/journal/paer © 2018 John Wiley & Sons Australia, Ltd 269

It is well established both in theory and in empirical studies that health plays an important role

in an individual’s labour participation decision. Becker (1964), Grossman (1972) and Currie and

Madrian (1999) regard health as a type of human capital endowment akin to education that is linked

to the labour market performance. Chirikos (1993) and Dwyer and Mitchell (1999) postulate that

poor health makes work more difficult and less fulfilling, thus increasing the utility of leisure rela-

tive to the utility derived from work. On the other hand, Dwyer and Mitchell (1999) and Cai and

Kalb (2006) argue that the income effect from lower earnings associated with poor health could

dominate the substitution effect between work and leisure.

However, quantifying the impact of a health shock or the onset of chronic conditions on an indi-

vidual’s labour market decision is a challenging task. There are many empirical studies measuring

the impact of health on labour participation decisions (Bazzoli, 1985; Bound, Schoenbaum, Stineb-

rickner, & Waidmann, 1999; Cai, 2010; Cai & Kalb, 2006; Disney, Emmerson, & Wakefield, 2006;

Dwyer & Mitchell, 1999; Garcia-Gomez, Jones, & Rice, 2010; Siddiqui, 1997; Zhang et al., 2009;

Zissimopoulos & Karoly, 2007). Most of these use cross-sectional data. Econometric methodologies

using cross-sectional data rely on the generalized law of large numbers to hold; that is, individual

outcomes are random draws from a population that has a constant mean conditional on some observ-

able factors. Furthermore, the estimated impact of a change in an observable factor is considered

instant and stays there forever. However, inertia in human behaviour and institutional and technolog-

ical rigidities have led many to believe that :“all interesting economic behaviour is inherently

dynamic, dynamic models are the only relevant models; what might superficially appear to be a

static model only conceals underlying dynamics, since any state variable presumed to influence pre-

sent behaviour is likely to depend in some way on past behaviour; and cross-sectional data may

effectively be precluded from studying the dynamics, but in which dynamics affects what is

observed”Nerlove (2002, p. 46).

There are already some studies using panel data to estimate dynamic labour participation models

(e.g. Knights, Harris, & Loundes, 2002; Oguzoglu, 2007, 2010; Zucchelli, Harris, & Zhao, 2012),

but they use RE dynamic discrete choice models to capture individual effects. Factors affecting indi-

vidual outcomes are numerous. Observable explanatory variables often only capture part of the fac-

tors affecting individual outcomes. If the observed outcome is also a function of individual-specific

effects that are persistent across time, then the observed dynamic response could be spurious in the

sense that it also captures the time-persistent individual-specific effect. Although panel data allow

the impacts of unobserved individual-specific effects to be controlled, the RE specification assumes

that the individual-specific effects are random draws from a common distribution and are uncorre-

lated with the included observable explanatory variables for the same individuals. If the individual

effects are, indeed, correlated with the included explanatory variables (e.g. Chamberlain, 1980),

which is most likely the case in empirical studies, the estimates of the coefficients of observable fac-

tors based on a RE specification could be misleading.

Labour market status is a clear example where strong state persistence is observed and an FE

model is more suitable to control for time-invariant individual effects. The labour participation deci-

sion could depend on an individual’s unobserved innate ability. For instance, a career-oriented indi-

vidual could still opt to work despite poor health, while a hedonic-oriented individual might prefer

to enjoy the leisure despite good health. These unobserved individual-specific effects are more likely

to be correlated with the observed confounding covariates, favouring an FE specification. Labour

market outcomes are also inherently dynamic and possess true state dependence where current out-

comes depend on past outcomes regardless of individual heterogeneity. Separation of the individual-

specific effects and true state dependence (Heckman, 1981a,b; Hsiao, 2014) is critical in providing

a correct assessment of the magnitude of health impacts on labour market outcomes. With

270 DAMRONGPLASIT ET AL.