Professional Documents
Culture Documents
The Impact of Health On Productivity
The Impact of Health On Productivity
I
mproving the living standards of popula- nomic and microeconomic relationship between
tions is a widespread societal objective. A health and productivity. This chapter reviews
cornerstone of living standards is the ability the theoretical underpinnings and empirical
of individuals to earn wages and profits in order evidence of this relationship. In particular, it
to purchase goods and services for consump- addresses the question: Would an improve-
tion. In turn, wages and profits reflect the value ment in the health status of working Canadians
of the goods and services produced in an econ- pay off in terms of higher aggregate produc-
omy and the productivity of the factor inputs tivity? The evidence presented comes from
used to produce them. Though living stan- many countries, both developed and develop-
dards, income and productivity are distinct con- ing, and spans a period of over 200 years. The
cepts, the three are very much related. The review focuses on implications for public pol-
correlation between labour productivity and icy and firm-level practices in developed
real wages both across countries and over time countries, particularly Canada.
is quite high, indicating the importance of pro- Figure 1 provides an organizing frame-
ductivity growth rates for the improvement of work for this review. It lists a number of
a country’s living standards (Harris 1999). strategies that the public and private sectors
Consequently, economists and historians have employed to promote the health of indi-
have focused much attention on better under- viduals and populations, as well as several
standing the determinants of productivity measures of health to assess the effectiveness
growth. There is increasing awareness that of these strategies. Some of the strategies are
human capital is a key factor. Traditionally, employed with the express intention of
human capital has been interpreted as edu- improving human capital and, in turn, pro-
cation and skills. Recently, however, increas- ductivity. Others have improving health as a
ing attention has been given to health as a specific goal. The more traditional strategies
form of human capital. of sanitation, nutrition and education are
Emile Tompa text 11/27/02 2:18 PM Page 182
Emile Tompa
FIGURE 1
Framework for the Review of Evidence on the Impact of Health on Productivity
Sanitation
Nutrition Health status Output per hour worked
Education Health and function Output per paid labour
Health promotion Sickness absence hour Output per capita
Healthy workplace Disability Output per worker
Occupation, health & Longevity Output per labour
safety force participant
Population health
public-sector interventions targeted at popu- absence) and career output (e.g., through decreased
182 lation health, but they nonetheless have impli- morbidity or increased longevity, resulting in a
cations for the health and productivity of the longer career). At the aggregate level, these indi-
labour force. Another traditional public-sec- vidual increases in output can translate into
tor intervention, occupational health and safe- increases in labour productivity (i.e., output per
ty, is targeted at the workplace and is focused hour worked, output per worker) and/or standard
on the reduction of accidents and chemical of living (i.e., GNP per capita) (e.g., by increas-
exposures and the resultant work-related ing the size of the active labour force relative to
injuries, illnesses and disabilities. The strate- the population).
gies of health promotion and healthy-work- This chapter proceeds as follows: The
place promotion are newer, firm-level next section, “Human Capital, Health and
initiatives developed since the 1970s from the Productivity,” reviews the theory on the
growing awareness that organizational-level demand for health and its relationship to the
interventions can be an effective means of pro- accumulation of human capital. “Historical
moting healthy lifestyles, reducing stress, Trends and Current Macroeconomic Evidence”
improving employee wellness, and reducing reviews the historical economic evidence con-
sickness-related absence and health-care costs cerning the relationship between health and
(Polanyi et al. 2000). Lastly, population health productivity growth as well as current macro-
is a new strategy based on the acknowledge- economic empirical work on measuring this
ment that the determinants of health are mul- relationship. “Occupational Health and Safety
tifactorial — biological, social and economic and the Cost of Work Disability” reviews the
— and that health policy needs to take a changing nature of work and the implications
broad, multisectoral approach (Frank 1995). for traditional approaches to occupational health
Figure 1 also lists several labour-produc- and safety regulation — its ability to influence
tivity and standard-of-living measures that can firm and worker behaviours and, through these,
be affected by improvements in the health of the health and productivity. “Health, Sickness
labour force. At the individual level, health can Absence and Firm-Level Practices” reviews the
directly increase general output (e.g., through microeconomic evidence concerning the rela-
enhanced physical energy and mental acuity), tionship between health and various produc-
yearly output (e.g., through reduced sickness tivity markers, with a focus on sickness
Emile Tompa text 11/27/02 2:18 PM Page 183
absence. Lastly, the key implications for poli- knowledge and health raises his or her produc-
cy and future research are summarized. tivity in both market and non-market activi-
ties. In the Grossman model, health capital
differs from other forms of human capital in its
HUMAN CAPITAL, HEALTH AND effect on these activities. Health capital deter-
PRODUCTIVITY mines the total amount of healthy time avail-
able for them, whereas knowledge capital
Grossman’s (1972, 2000) model for affects the productivity of the time spent on
health demand provides insights into the rela- them. This approach suggests that health cap-
tionship among health, human capital and ital provides a flow of healthy time that is uni-
consumption at the individual level, as well as form in quality, an “all or nothing” state. An
a framework for modelling human capital alternative formulation would be to have health
accumulation and its relationship to produc- capital bearing on both the quality and quanti- 183
tivity at the micro and macro levels. The main ty of healthy time. Like all capital, health depre-
contribution of this model is that it offers ciates over time and is assumed to do so at an
insights into modelling two key aspects of increasing rate with age. Consequently, invest-
human capital, health and education, and their ment is required to restore and/or maintain
relationship to labour supply, earnings and health stocks through household production
productivity. The model is based on Becker’s activities that include inputs such as exercise,
(1965) household-production concept, which nutrition and health care. The model does not
in turn is premised on the notion that utility expressly include spillover effects in the pro-
is obtained not directly from market goods and duction of health and education, which can be
services, but, rather, from final consumption an important contributor to the efficiency of
goods produced from market goods and serv- their production (e.g., the health of parents can
ices in conjunction with one’s own time. For affect child health outcomes, and some of the
example, leisure (a final consumption good) skills and knowledge acquired by a worker
may be produced with the purchase of movie through educational pursuits can be transmit-
tickets and one’s own healthy time. Some ted to colleagues).
household production processes provide utility There is significant interplay between dif-
directly (e.g., the production of leisure), where- ferent types of human capital, specifically
as others are inputs into other processes such between education and health. In the Grossman
as educational development or labour force par- model, higher levels of education are theorized
ticipation which provide utility indirectly. A to improve the efficiency of gross health invest-
fundamental aspect of the Grossman model is ment.1 The empirical literature substantiates the
that health or healthy time provides utility not existence of this relationship (Grossman and
only directly but also indirectly, since it is a Kaestner 1997). Whether it is causal — and in
critical input into many production processes, which direction — is not clear, though Grossman
as described above. Accordingly, health is both and Kaestner conclude that the evidence suggests
a final consumption good and a capital good. the existence of a pathway from education to
Human capital theory is premised on the health (i.e., individuals with higher education
notion that an increase in a person’s stock of are better at producing health). A third variable,
Emile Tompa text 11/27/02 2:18 PM Page 184
Emile Tompa
time preference, may also play an intermediate an increase in labour force participation —
role. Higher education may result in an indi- which may, in turn, result in increased per
vidual placing more value on the future (i.e., it capita income if these individuals are accom-
may lower an individual’s time preference, modated by the labour market.
which suggests that time preference is endoge-
nous). Alternatively, a lower discount rate may
encourage an individual to seek higher levels of HISTORICAL TRENDS AND
education, which, in turn, bears on the optimal CURRENT MACROECONOMIC
level of health capital investment.2 EVIDENCE
The Grossman model can be used to
identify an individual’s labour supply as a The empirical literature in economic his-
function of health. The principal implication tory provides substantive evidence concerning
184 of the model is that health is determined the productivity impact of increased life
endogenously (i.e., by the other variables in expectancy and reduced morbidity over the last
the model rather than by exogenous/external few centuries in Europe and the United States
factors). In principle, education is also deter- (e.g., Costa and Steckel 1995; Fogel 1991,
mined endogenously, but since most people 1994; Steckel 2001/2002). Fogel (1991) stress-
complete their education early in life it can be es the importance of long-run dynamics and
treated as an exogenous variable when model- presents evidence that improvements in health
ling labour supply, whereas health capital which began some 300 years ago in Europe
depreciates and requires ongoing investment and North America have not yet fully run
(Currie and Madrian 1999). As the Grossman their course. This work suggests that an
model suggests, health is also an important understanding of the key drivers of long-run
aspect of human capital, and an important dynamics may be of value to policy-making
input into market and non-market production in developed countries even today. To this
at the individual level. end, evidence from the historical economics
At the aggregate level, Bloom and literature is reviewed below, followed by
Canning (2000) identify four pathways by empirical evidence based on data from more
which health can affect productivity: a healthy recent periods that makes use of the growth-
labour force may be more productive because accounting framework. Similar measures of
workers have more physical and mental ener- health appear in both literatures.
gy and are absent from work less often; indi- Fogel (1991, 1994) presents historical
viduals with a longer life expectancy may trends in England, France and Sweden on two
choose to invest more in education and receive anthropomorphic measures associated with
greater returns from their investments; with nutrition, namely adult height and weight
longer life expectancy, individuals may be (also known as body mass index). Height and
motivated to save more for retirement, result- weight provide different information about
ing in a greater accumulation of physical cap- health. Adult height reflects the adequacy of
ital; and improvement in the survival and early-childhood nutrition, whereas adult weight
health of young children may provide incen- reflects the adequacy of adult nutrition. Using
tives for reduced fertility and may result in more recent evidence from Norway on the
Emile Tompa text 11/27/02 2:18 PM Page 185
relationship among height, weight and risk of about six hours of light work (1.09 hours of
mortality, Fogel estimates the proportion of heavy work). Essentially, those in the bottom
the decline in mortality in the three countries 20 percentile had such poor diets that they were
that can be associated with changes in these excluded from the labour force. As well, many
anthropomorphic measures since the 18th cen- of those in the top 40 percentile were below
tury. Research on the relationship among current North American standards of average
height, weight and chronic disease from more height and weight, and hence were likely sub-
recent US data provides further evidence on ject to premature chronic conditions and mor-
the detrimental health implications of below- tality. Subsequent improvements in nutrition
average height and weight (relative to current raised the energy levels of individuals in the
North American standards). Based on this bottom 20 percentile of consumption such
research, Fogel (1994) concludes that chronic that they were able to enter the labour force.
health conditions were significantly more These improvements in nutrition also sub- 185
prevalent throughout the life cycle prior to the stantially raised the capabilities of those
First World War. Consistent with Fogel’s already in the labour force. Fogel (1991, 1994)
work, Steckel (2001/2002) presents more estimates that health and nutritional improve-
recent evidence on anthropomorphic measures ments alone can explain some 30 percent of
as proxies of health, supporting the notion British growth in per capita income since
that health can influence productivity. He 1790. This value is similar to estimates of the
found that the simple correlation between productivity impacts of health found in cross-
average height and log of GDP per capita country studies using data from the last 50
ranges from 0.82 to 0.88. Furthermore, he years (World Health Organization 1999).
notes that the average height of Americans is Recent macroeconomic research on pro-
falling behind that of Northern Europeans, ductivity has emphasized the importance of
and that this trend may be reflective of grow- human capital. Like physical capital, human
ing income inequality in the United States. capital in the form of education and health is
Fogel (1994) presents evidence of the durable, lasting, and subject to accumulation
historical impact of population health on labour (Lucas 1988; Romer 1986). One approach to
force productivity drawn from estimates of incorporating human capital into the macro-
caloric intake in Britain and France in the 18th economic modelling of productivity is to aug-
and early 19th centuries. He estimates that the ment the neoclassical growth-accounting
daily caloric intake for individuals in the bot- equation developed by Solow (1956). Solow’s
tom 10 percentile of consumption in France approach to measuring multifactor productiv-
was so low that they did not have enough ener- ity growth is to associate it with the residual
gy for work, and that those in the next 10 per- amount of output growth not explained by
centile had energy for only three hours of light growth in the key inputs of labour and physi-
work (0.52 hours of heavy work). In England cal capital. This approach is founded upon sev-
the situation was somewhat better. Only indi- eral contentious assumptions. First, it assumes
viduals in the bottom 3 percentile of con- that technology (which is associated with the
sumption lacked enough energy for work, and residual) is exogenous, suggesting that labour-
those in the next 17 percentile had energy for productivity growth rates will be the same
Emile Tompa text 11/27/02 2:18 PM Page 186
Emile Tompa
across economies once they reach a steady state. reduced personal-care time required by family
This runs counter to the trends of sustained caregivers who are members of the labour force.
differences in growth rates observed across Human capital in the form of education
developed countries. Another assumption of has received much attention in cross-country
this approach is that there is perfect competi- empirical growth studies, and researchers have
tion such that market prices reflect social costs found considerable support for its importance
— that is, there are no information asymme- as a productivity driver (Harris 1999). Human
tries, appropriability problems, spillovers or capital in the form of health has received less
other externalities. Intuitively, these appear to attention, but the relatively few cross-country
be restrictive assumptions that are likely not studies that have included some measure of
met in the real world, suggesting that there is health have found that it does have a signifi-
a role for public policy. One can easily imag- cant and positive association with economic
186 ine spillover effects occurring from higher lev- growth.4 Many of the empirical growth stud-
els of human capital. They could have a ies that include health have focused on devel-
substantial impact not only on an individual’s oping countries (e.g., Bhargava et al. 2001;
own productivity, but also on the productivity Hicks 1979; Wheeler 1980), though some
of co-workers and on society as a whole. A have included a broader range of countries
number of measurement issues also arise with (Barro and Sala-i-Martin 1995; Bloom et al.
this paradigm, the most salient of which is 2001; Knowles and Owen 1995, 1997) and
how to deal with technological improvements some have focused specifically on OECD coun-
and quality changes embodied in both inputs tries (Knowles and Owen 1995, 1997; Rivera
and outputs.3 Quality improvements are rele- and Currais 1999a, 1999b).
vant for both physical and human capital. These studies often use rather crude
In response to the shortcomings of the measures of health, likely due to the lack of
Solow model, a new approach to growth account- data on more refined and comprehensive meas-
ing has evolved, one that attempts to model the ures that span both a reasonable time period
key determinants of growth as jointly endoge- and a number of countries.5 Most studies use
nous (Knowles and Owen 1997). One way to some measure of life expectancy or mortality
augment the Solow model is to include the accu- (life expectancy at birth, infant mortality rates,
mulation of human capital as well as physical adult survival rates), though two recent stud-
capital, while still treating technology as exoge- ies used per capita health-care expenditures
nously determined. This diminishes the impor- (Rivera and Currais 1999a, 1999b). Life
tance of exogenous technological growth. The expectancy has increased dramatically over the
endogenous growth literature attempts to cap- post-war period in many developed countries
ture two aspects of the impact of health on pro- (see Chart 1 for Canadian trends). Though
ductivity: its direct impact on the production mortality and life expectancy are important
process — for example, improvements in measures of health status, they may not cap-
health can increase productivity due to reduced ture the subtle changes in morbidity, health
incapacity, disability and days off sick; and its behaviours, health-related quality-of-life meas-
spillover impact — for example, an improve- ures, or other measures of health that are par-
ment in the health of seniors can result in ticularly salient to developed countries today.
Emile Tompa text 11/27/02 2:18 PM Page 187
Age
evidence of this fact. As is apparent in the
72
graph, the life-expectancy gradient is much
steeper for low levels of gross national income 70
per capita. Indeed, when Knowles and Owen 68
(1997) estimated their specification for 22
66
high-income countries, they found that the 187
1950
1955
1960
1965
1970
1975
1980
1985
1990
1995
health measure they used, life expectancy, was
no longer significant, likely due to the lack of Source: World Bank.
Emile Tompa
CHART 2
Gross National Income Per Capita (GNIpc) Versus Life Expectancy (LE) at Birth, 1997
90
80
70
LE
60
188 50
40
30
GNIpc 100 1,000 10,000 100,000
TABLE 1
Macroeconomic Growth Studies with Data from Developed Countries
that Include Measures of Health 189
Bhargava et log GDP log of adult 125 countries from varies by GDP
al.( 2001) growth rate per survival rate Pen World Tables +ve for low-income countries
capita 107 countries from -ve for high-income countries
World For the poorest countries, a 1%
Development change in adult survival rate is
Indicators associated with a 0.05%
(1965–90) increase in GDP growth rate.
Emile Tompa
CHART 3 CHART 4
Total Workers’ Compensation Indemnity Time Loss/Fatality and No Time Loss
Payments in Canada Compensation Claims Per Employee
60000 0.07
Expenditures in 1992$ (thousands)
0.05
40000
0.04
30000
0.03
20000
0.02
10000 0.01
191
0 0.00
1993
1985
1977
1997
1989
1992-93
1981
1984-85
1973
1976-77
1996-97
1988-89
1980-81
1972-73
Canada and other developed countries (see Chart modest general deterrence unless reinforced with
4 for Canadian trends). In Canada, injury claims penalties. Scholz and Gray (1997) found that reg-
decreased by 40 percent between 1990 and ulation facilitating cooperation, such as inspec-
1998, despite the fact that the labour force tions initiated by workers (regardless of penalty),
increased by 10 percent over this period can be more effective than coercive regulation
(Mustard et al. 2001). A number of explanations such as regular inspections, unless penalties are
have been put forward for this trend. Though imposed. The effectiveness of facilitative regula-
there is evidence to support some of these, it is tion is reinforced by Canadian evidence on the
not clear which factors are the predominant introduction of regulations requiring joint health
ones.8 More specifically, it is not clear what frac- and safety committees (Lewchuck et al. 1996).9
tion of the trend is attributable to the effective- The evidence on workers’ compensation experi-
ness of insurance and regulatory mechanisms. ence rating suggests that financial incentives can
There is a large body of empirical litera- be an effective means of improving occupational
ture on the effectiveness of insurance and regu- health and safety. The appeal of experience rat-
latory mechanisms using econometric techniques ing is that it ties the cost of work-related injuries
(reviews of this evidence are provided by closer to the firms experiencing them, while
Curington 1988; Hyatt and Thomason 1998, allowing firms the flexibility to find the most
Kralj 2000; and Smith 1992). Taken as a whole, efficient means of improving health and safety.
the empirical evidence on the impact of regula- One of the challenges of regulatory and
tion is mixed. The US evidence suggests that insurance mechanisms is that the nature of
standards can reduce certain types of injuries and work and labour market experience have
that a system of inspections provides, at best, changed profoundly since their introduction in
Emile Tompa text 11/27/02 2:18 PM Page 192
Emile Tompa
Canada, yet occupational health and safety reg- kets that determine the size and distribution of
ulation and workers’ compensation continue to productivity gains associated with health
focus on injuries and the impact of physical improvements. The framework focuses on a
and chemical exposures characteristic of man- particular proxy for health-related productivi-
ufacturing and resource-based industries. The ty gains, namely decreases in sickness absence,
shift away from manufacturing to services, the but the premise of the model can be general-
increasing use of technology, the introduction ized to all health-related productivity improve-
of new human-resource and management prac- ments.11 This framework is examined below,
tices, and the growth in demand for knowledge followed by a review of the microeconomic evi-
workers have all contributed to a dramatic dence concerning the relationship between
change in the nature of work-related injuries health and productivity, and employers’ efforts
and illnesses (Sullivan and Frank 2000). For to capture potential health-related productivity
192 example, mental-stress claims more than dou- gains through health-promotion initiatives.
bled in the United States between 1980 and Consistent with this literature, particular atten-
1987 (Gnam 2000).10 Furthermore, the growth tion is given to sickness absence as a proxy for
in non-standard work arrangements has not productivity. As background for this discussion,
only made it difficult for workers’ compensa- Canadian trends in sickness absence are provid-
tion boards to assign firm-level responsibility ed in Chart 5. Rates have been increasing for
for injuries and illnesses, but has also dramat- both men and women over the last few years.
ically changed the nature of labour market This may be due in part to the economic recov-
experiences. Other factors such as income ery during the period. Similar to work-related
inequality, job insecurity and unemployment accident claim rates, sickness absence rates have
have also been shown to have a bearing on the a cyclical component (i.e., work-related acci-
health of individuals and populations (Deaton dent claims tend to increase during periods of
2001; Platt et al. 1999). economic recovery due to factors such as the
increased pace of work). Nonetheless, the rate
for women has increased to levels above those
HEALTH, SICKNESS ABSENCE AND of the late 1980s.
FIRM-LEVEL PRACTICES Many empirical studies on the cost of
sickness absence and the cost-effectiveness of
At the microeconomic level, research into health-promotion initiatives assume that the
the impact of health on productivity focuses on dollar value of reducing sickness absence is
returns to employers and workers. Returns to simply the direct cost of wages paid to absent
the accumulation of health capital can be real- workers (assuming that they are paid for
ized by employers in the form of higher profits, absences). In the Pauly framework, wages are
by workers in the form of higher wages, or by actually the lower bound for losses. In many
both. Pauly et al. (forthcoming) provide a the- cases, total costs can be much higher due to
oretical framework (hereinafter the Pauly frame- indirect costs attributable to sickness absence.
work) to identify the principal characteristics of If the production process is team-based, or if
the production process, the market for the good a penalty is incurred for failure to achieve tar-
or service being produced, and the labour mar- get output, then the cost of sickness absence
Emile Tompa text 11/27/02 2:18 PM Page 193
1998
1991
1988
1995
1992
1989
1996
1993
1997
1990
1987
1999
Emile Tompa
and societal values of labour time may differ example, some people may believe that health
as well. If this difference is small, the gains status is predetermined and that changes in
from reducing sickness absence would be sim- health status are unavoidable, but an interven-
ilar to the full-employment case. In general, tion such as an influenza shot can significantly
the benefits from reduced sickness absence are reduce the incidence of colds and flu, and thus
greater than the wage rate. It should be noted may reduce sickness absence more readily than
that the impact of sickness absence on pro- an intervention designed to increase workers’
ductivity is greater when measured by output engagement in their work.
per worker than output per hour, since some Nonetheless, there is evidence to support
of the output not produced due to sickness the notion that health status has a strong influ-
absence is offset by the reduced number of ence on sickness absence. Most empirical stud-
hours worked. ies of sickness absence that have included
194 The Pauly framework underscores the measures of long-term health, chronic conditions
potential productivity gains of reducing sick- or health behaviours have focused on self-report-
ness absence through investments in health ed health status, smoking, illicit-substance use
capital. Implicit in this line of thinking is the and alcohol consumption. The Whitehall II
assumption that poor health is the principal study provides some of the most compelling evi-
reason for sickness absence.14 But the etiology dence for the impact of these factors on sickness
of sickness absence is quite complex; poor absence (Marmot et al. 1995; Marmot et al.
health is but one of many factors that have a 1993; North et al. 1993). Reported “average” or
bearing on sickness absence. Examples of other “worse” health over the 12 months preceding
factors include personality, job-related atti- the survey was associated with significantly
tudes and social context. Two recent literature higher levels of sickness absence compared to
reviews (Alexanderson 1998; Harrison and reported “good” health — a 60 percent
Martocchio 1998) identify a broad range of increase in short spells (seven days or less) and
causal factors investigated in the empirical lit- a twofold increase in long spells (more than
erature on sickness absence and find that few seven days). Significantly higher levels of sick-
studies have investigated the impact of health ness absence were also observed for individuals
on sickness absence.15 In particular, short-term reporting recurrent health problems, long-
health conditions are rarely studied. Nicholson standing illnesses or psychiatric symptoms.
and Martocchio (1995) describe this gap in the Mental health factors such as depression, anxi-
literature as a “black hole.” Alexanderson notes ety and emotional stress are a frequently report-
that even though most studies in her review ed cause of sickness absence, particularly
are from the fields of epidemiology and medi- among women (Stansfeld et al. 1995), and have
cine, most do not use a medical model or even been found to be significant predictors of
consider health status as an explanatory vari- absence and disability (Garrison and Eaton
able. This lack of attention to health is likely 1992; Kessler et al. 1999; Kouzis and Eaton
due to a focus on sources of variance that are 1994; Simon et al. 2001; Skodol et al. 1994).
perceived to be avoidable (and thus amenable Studies of health-related behaviours have found
to change), though the distinction between that smokers have higher rates of absence than
avoidable and unavoidable is blurry at best. For non-smokers (Bush and Wooden 1995; Leigh
Emile Tompa text 11/27/02 2:18 PM Page 195
1995; North et al. 1993; Parkes 1987) and growing cost of employee health care in the
that illicit-substance users have high rates of 1970s and 1980s, which was increasing at a rate
absence (Bass et al. 1996; Normand et al. of up to 30 percent per year (Conrad 1988). The
1990). The impact of alcohol consumption is focus on health-care costs in many US employer
more complex. Problem drinking (a level of initiatives is understandable given that health
alcohol consumption that results in social dys- insurance is generally provided by employers and
function) appears to be the point at which can make up a substantial component of the ben-
sickness absence is affected (Beaumont and efits provided to workers. In Canada, it is a less
Hyman 1987; Casswell et al. 1988), causing salient cost for employers, since health insurance
long-term absences in particular (Marmot et is primarily funded by the public sector, but is
al. 1993). nonetheless a factor to consider at the macro
Acute conditions related to respiratory level. Canada has also experienced substantial
and gastrointestinal conditions are the primary increases in health-care costs since the 1970s, and 195
reasons for short-term absences (Stansfeld et al. there is evidence that a growing proportion is
1995), yet, as noted, few studies have investi- paid by the private sector (Polanyi et al. 2000).
gated the nature and impact of short-term Studies of the impact of employer-spon-
health conditions on sickness absence. This is sored health-promotion initiatives generally use
surprising given that more immediate inroads quasi-experimental methods (participants self-
into reducing sickness absence might be made select into the program) or non-experimental
by addressing the factors that cause acute con- methods (no control group — e.g, before/after
ditions. For example, Nichol et al. (1995) found comparison) and cost-benefit or cost-effectiveness
a 43 percent drop in the rate of cold- and flu- analysis. The dearth of rigorous research, partic-
related sickness absences among adults receiv- ularly in the earlier literature in this field, has
ing an influenza vaccine instead of a placebo. been commented on in most reviews (Aldana
The evidence regarding the impact of 1998; Baun 1995; Fielding 1990; Heaney and
health status on sickness absence, though pre- Goetzel 1997; Messer and Stone 1995;
liminary, suggests that firm-level initiatives such O’Donnell 1997; Pelletier 1991, 1993, 1996,
as health promotion are one means by which 1999, 2001; Shaeffer et al. 1994; Shephard 1992;
employers can reduce absence and increase pro- Warner 1992; Warner et al. 1988).16 As a whole,
ductivity. Employers have undertaken a variety these studies show mixed evidence, with non-
of initiatives, ranging from targeted to multi- experimental studies generally demonstrating
component programs (fitness/exercise programs positive results and more rigorous studies
are the most widespread). A large empirical lit- demonstrating less positive results. On average,
erature on the impact of these initiatives on studies using experimental designs had positive
health and productivity has accumulated over the results approximately 25 percent of the time,
past 30 years. Most of the empirical studies have quasi-experimental designs 50 percent of the
considered one or more of five outcomes: risk time and non-experimental designs 100 percent
reduction or behavioural change, health/medical- of the time (Heaney and Goetzel 1997;
care costs, sickness absence, turnover and other O’Donnell 1997). In terms of sickness absence,
proxies of productivity. One of the motivations the evidence on the impact of health promotion
for these initiatives in the United States was the is mixed (Baun 1995).17 If the results of a broad
Emile Tompa text 11/27/02 2:18 PM Page 196
Emile Tompa
range of studies reviewed by Shephard (1992) can (Marmot et al. 1991, 1993), which has health
be accepted as the true program effects, then the effects that can endure even after retirement
impact on sickness absence is positive but mod- (Wolfson et al. 1990); (3) if behavioural and
est; most of the studies found effect sizes in the lifestyle changes are to be enduring, the social
range of 0.5 to two days per year of improved and cultural context that engendered them
attendance. In Messer and Stone’s (1995) review must be modified as well; (4) health-promo-
of cost-benefit studies, those that included the tion initiatives reach only a limited number of
cost of reduced sickness absence as a benefit workers, since they tend to be implemented in
found that benefits exceeded costs in all cases white-collar settings; and (5) unless other job-
(benefit-cost ratios ranged from 1.07 to 3.90); for related and organizational factors are also
some of the studies, the positive effects may also addressed, workers may perceive such initia-
have been driven by reduced health-care costs. tives as a self-serving effort by firms to reduce
196 The quality of research has improved their health-care and absenteeism costs.
substantially in the most recent generation of Indeed, it is becoming increasingly evi-
studies, providing a rationale for employers to dent that general health and functioning are
consider such programs as a means of reducing very much affected by work experiences. There
the health and economic costs of illness. The is a growing body of evidence showing that psy-
findings provide some support for the hypoth- chosocial workplace factors such as job control,
esis that health and fitness can have an impact psychological demands and social support have
on sickness absence and productivity, though an important bearing on workers’ health
the durability of these effects has not been (Shannon et al. 2001). This suggests that broad-
established. A realistic assumption is that a ly based organizational initiatives may be more
program will need to engage workers on an successful in improving workers’ mental and
ongoing basis if the changes are to be sus- physical health and productivity than focused
tained. Though most studies have found only health-promotion programs. Such broad-based
modest reductions in sickness absences as a initiatives can include a range of elements such
result of health-promotion programs, the direct as: redesigning worksites with ergonomic prin-
costs of these absences represent only a lower ciples in mind; redesigning work flows and
bound for the costs attributable to sickness communications channels to enhance commu-
absence, suggesting that productivity gains nication and social support; and providing flex-
may be higher. ible work hours, leave programs and daycare
Polanyi et al. (2000) are less optimistic facilities. Unfortunately, there are few experi-
about the potential for firm-level health-pro- mental studies in the literature investigating the
motion initiatives. They list five reasons why productivity impact of such initiatives.
such initiatives may be limited in their ability
to improve worker health: (1) this approach
does not address the sources of human motiva- SUMMARY AND CONCLUSIONS
tion and behaviour that bear on health, name-
ly the social and economic determinants of Economic well-being in developed coun-
health; (2) lifestyle is a less important factor tries is growing increasingly dependent on
bearing on health than socioeconomic status international markets and integrated trade. To
Emile Tompa text 11/27/02 2:18 PM Page 197
maintain a high standard of living in the glob- research focuses on life expectancy as a meas-
al marketplace, countries must remain com- ure of health, which can be an important pro-
petitive, which in turn depends on the ductivity driver for developing countries but
fostering of an innovative, productive labour is less salient for developed countries. In order
force. Researchers and policy-makers are to better serve the information needs of poli-
becoming increasingly aware that a country’s cy arenas in developed countries, future
ability to innovate and remain productive research in this area should identify more
depends on the characteristics and quality of refined measures of health to incorporate into
its human capital, key elements of which are their empirical analyses (e.g., measures of
education, skills and health. Most recently, the functional status and mental health).
links between population health and econom- Work-related injuries and illnesses are a
ic productivity have become a significant pol- major source of productivity losses for socie-
icy concern. With the aging of the labour force ty as a whole, suggesting that public policy 197
in Canada and many other developed countries, can play a crucial role in providing incentives
labour force health will become an even more for employers to dedicate more resources to
important issue in the near future. On the pos- occupational health and safety. Most devel-
itive side, there is evidence that older workers oped countries recognize the importance of
in developed countries are no less productive labour market institutions designed to pro-
than their younger counterparts. tect the health and safety of workers, and have
Fogel’s research in economic history established financial and regulatory mecha-
highlights the significance of population nisms for this purpose. Different policy levers
health for productivity growth. He provides are available under the umbrella of regulation
compelling evidence for the important role of and insurance, some of which are more effec-
nutrition, particularly in early childhood, on tive than others. The evidence suggests that
health and functioning throughout the life occupational health and safety regulation has
cycle. Fogel points out that these historic had only a modest impact, though prelimi-
trends still have a bearing on the health of pop- nary evidence shows that regulation focusing
ulations in developed countries today. This on facilitation rather than coercion may be
research provides valuable insights into the crit- more promising. The evidence for experience-
ical role that policy can play in supporting pop- rating suggests that it is effective, partly
ulation health and ultimately productivity. For because financial incentives allow employers
example, the enduring effect of childhood the flexibility to identify the most efficient
experiences is one of the themes in this work, means by which to improve workplace health
suggesting that financial support for low- and safety. Both insurance and regulation
income families, parental leave policies and tend to create incentives that focus on acute
child-care policies can help to ensure healthy injuries and the health impacts of physical
child and adult outcomes. The evidence from and chemical exposures characteristic of the
macroeconomic studies on health and produc- industrial sector. Consequently, many of the
tivity corroborates the evidence from econom- important factors affecting labour force and
ic history. Health is indeed an important population health in developed countries
driver of productivity even today. This area of remain outside the purview of traditional
Emile Tompa text 11/27/02 2:18 PM Page 198
Emile Tompa
insurance and regulatory domains, due to the ductivity. Identifying priorities in order to
changing nature of work and labour market achieve the greatest gains from the resources
experiences in these countries. invested is a difficult task at best. Certainly,
Sickness absence is an easily measured increasing health-care spending will not neces-
proxy for productivity that is often investigated sarily result in higher levels of population health,
in the organizational practices literature. Yet, as evidenced by the differences in per capita
surprisingly, the empirical literature on the caus- health-care expenditures and health profiles
es of sickness absence has given little attention across OECD countries. The multifaceted nature
to the role of health status as an explanatory vari- of the factors that influence health suggests that
able. The work that has been undertaken in this policies in a number of areas traditionally con-
area suggests that chronic and acute physical and sidered outside the purview of health policy may
mental conditions, as well as health-related be important avenues by which the public sector
198 behaviours, explain a significant portion of sick- can have an impact on population health. Key
ness absence. This evidence provides support for areas are labour market policy, education poli-
health-promotion initiatives as a means of reduc- cy, and child-care and parental leave policy.
ing sickness absence and increasing productivi- Furthermore, though improving population
ty. Empirical studies evaluating workplace health is an important societal objective, there
health-promotion programs find that they are are many other objectives competing for scarce
effective in reducing absence and health-care public resources. Achieving an optimal balance
costs, though these reductions are modest and when addressing societal objectives requires a
the durability of program effects is not known. sound understanding of the policy options, their
Nevertheless, the direct costs of absences are impact on the various objectives, and the costs
only a lower bound for productivity losses attrib- associated with each.
utable to them; the productivity gains to be real-
ized from decreasing sickness absences may be
substantially higher. Future research in this area NOTES
should focus on developing the tools to quanti-
1 Muurinen (1982) provides an alternative formulation
fy the various indirect costs associated with sick- in which education is theorized to lower the rate of
ness absence, as well as improving the depreciation of health stock.
methodological rigour of intervention studies. 2 A lower discount rate may also encourage investment
Furthermore, there is evidence to suggest that in health directly.
such initiatives should consider a broader set of 3 There are other measurement issues with regard to
capturing the full impact of health on welfare. In
organizational factors related to work and the particular, utility is derived directly from health, as
workplace that bear on the health and produc- well as indirectly through its role in market and non-
tivity of workers instead of narrowly focusing on market production activities, whereas only market
production activities are captured in standard output
behaviour and lifestyle. measures. The direct value of health may be captured
The fact that firm-level initiatives appear to some degree through market goods and services
purchased to improve health. The indirect value of
to have had a limited impact on productivity health from non-market activities may also be
suggests that the public sector has a role to play captured to some degree in this way. At the core of
in improving the health of the labour force and this measurement issue is the fact that standard
measures of output do not capture non-market
population as a whole and, in turn, overall pro- resources and activities. Costa and Steckel (1995)
Emile Tompa text 11/27/02 2:18 PM Page 199
compare alternative methodologies of measuring 10 Mental-stress claims still represent only a small
welfare changes arising from changes in the health of fraction of workers’ compensation claims in North
populations as an illustrative exercise to assess the America, though this may be due in part to
consistency of rankings. legislation passed in the early 1990s in many
4 Significance does not necessarily imply causality. jurisdictions limiting compensability for such claims
In fact, the direction of effects might run from (Gnam 2000).
productivity growth to health, rather than the 11 Decreases in sickness absence are frequently used as a
reverse. All of the studies address the reverse proxy for health-related productivity gains attributable to
causality issue by testing for potential employer-sponsored initiatives, primarily because sickness
endogeneity. absence is readily observable and measurable, and data
5 Currie and Madrian (1999) classify health measures regularly collected by human-resource departments. Two
typically available in the data sets of developed other proxies considered in the literature are disability and
countries into eight categories: self-reported health turnover. It should be noted that health improvements
status, presence of functional limitations on the can increase not only the amount of time available for
ability to work, presence of functional limitations on work, but also the quality of time spent at work. More
other activities, presence of chronic/acute health comprehensive empirical analyses attempt to assess the
conditions, health-care utilization, clinical entire range of direct and indirect benefits and costs 199
assessments of health, nutritional status and associated with a health improvement.
mortality. 12 In the case of a salaried worker who is required to
6 GDP per capita growth is highly correlated with make up lost production time, the benefits are in the
GDP per worker growth, with the latter driving the form of more convenient hours. Note that the value
former. None of the studies corrected for the fact that of lost leisure time is not captured by traditional
average hours of work can vary substantially from output measures but is nonetheless an important
country to country. aspect of social welfare.
7 Leigh et al. (2001) estimate that the total cost of 13 Firms may attempt to insure against losses by hiring
health care and lost productivity due to occupational extra workers as backups to cover for absences. This
injuries and illnesses in California is on par with the imperfect remedy will be relatively less costly for
costs of all cancers combined, and comparable to the larger firms than for smaller ones.
costs of heart disease and stroke. 14 One of the shortcomings of the model is its narrow
8 These explanations support one of two propositions: focus on sickness absence, rather than the impact of
that the trend reflects real decreases in work worker health on productivity in general. However,
disability, or that the trend is a reporting the basic tenets of the model can be generalized to
phenomenon. Following are some of the this broader perspective.
explanations offered: de-industrialization has 15 Harrison and Martocchio (1998) review over 500
resulted in a greater proportion of the labour force empirical studies on sickness absence culled from a
being employed in the service sector, which is variety of disciplines conducted from 1977 to 1996.
inherently safer; older capital is being replaced with 16 Frequently cited methodological shortcomings are:
new capital, which embodies ergonomic and failure to use a control group or randomization of
technological improvements that are inherently individuals between program and control groups, failure
safer; employers are doing a better job of instituting to adjust for confounders, lack of standardized measures
safety measures in response to insurance and used for exposure and outcome, short time period of
regulatory incentives or because of an increased studies, small sample sizes, use of self-reported measures,
awareness of the value of health capital; employers failure to consider all indirect costs and benefits associated
are adopting more aggressive claims-management with an initiative, bias introduced by evaluations being
practices; and workers’ compensation boards have performed by program advocates, and failure to consider
tightened their eligibility requirements. the worksite as the unit of analysis when initiatives are
9 This study also found that firms that are reluctant to implemented at a selection of worksites.
form joint health and safety committees show poorer 17 Baun (1995) reviews studies that assess a range of health-
health and safety performance, suggesting that the promotion initiatives for their impact in terms of
climate of the internal-responsibility system is an reducing sickness absence. He concludes that the
important element in its success. The authors evidence for the effectiveness of smoking-cessation
suggest that, if the internal-responsibility system is programs, health-risk assessment programs and exercise
to be effective, special measures may be required to programs is mixed. Only stress-reduction programs
educate employers and workers in such cases. appear to have a powerful effect on sickness absence.
Emile Tompa text 11/27/02 2:18 PM Page 200
Emile Tompa
Institute for Work and Health. 2002. Institute for Work Mustard, C.A., D. Cole, H. Shannon, J. Pole, T. Sullivan,
and Health Website: www.iwh.on.ca R. Allingham, and S. Sinclair. 2001. “Does the
Kessler, R.C., C. Barber, H.G. Birnbaum, R. Frank, P.E. Decline in Workers’ Compensation Claims
Greenberg, R.M. Rose, G. Simon, and P.S. Wang. 1990–1999 in Ontario Correspond to a Decline in
1999. “Depression in the Workplace: Effects on Short- Workplace Injuries?” Working Paper #168.
Term Disability.” Health Affairs 18(5):163-171. Toronto: Institute for Work & Health.
Knowles, S., and P.D. Owen. 1995. “Health Capital and Cross- Muurinen, J.M. 1982. “Demand for Health: A
Country Variation in Income Per Capita in the Mankiw- Generalised Grossman Model.” Journal of Health
Romer-Weil Model.” Economics Letters 48:99-106. Economics 1:5-28.
—— 1997. “Education and Health in an Effective-Labour Nichol, K.L, A. Lind, K.L. Margolis, M. Murdoch, R.
Empirical Growth Model.” Economic Record McFadden, M. Hauge, S. Magran, and M. Drake.
73(223):314-328. 1995. “The Effectiveness of Vaccination Against
Influenza in Healthy Working Adults.” New England
Kouzis, A.C, and W.W. Eaton. 1994. “Emotional Journal of Medicine 335(4):889-893.
Disability Days: Prevalence and Predictors.”
American Journal of Public Health 84(8):1304-1307. Nicholson, N., and J.J. Martocchio. 1995. “The
Management of Absence: What Do We Know?
Kralj, B. 2000. “Occupational Health and Safety: What Can We Do?” In Handbook of Human Resources 201
Effectiveness of Economic and Regulatory Management, ed. G.R. Ferris, S.E. Rosen and D.T.
Mechanisms.” In Workers’ Compensation: Foundations Barnum. Oxford: Blackwell.
for Reform, ed. M. Gunderson and D. Hyatt. Toronto:
University of Toronto Press. Normand, J., S.D. Salyards, and J.J. Mahoney. 1990. “An
Evaluation of Preemployment Drug Testing.” Journal
Leigh, P.J. 1995. “Smoking, Self-Selection and Sickness of Applied Psychology 75(6):629-639.
Absence.” Quarterly Review of Economics and Finance
35(4):365-386. North, F., S.L. Syme, A. Feeney, J. Head, M.J. Shipley,
and M.G. Marmot. 1993. “Explaining
Leigh, P.J., J.E. Cone, and R. Harrison. 2001. “Costs of Socioeconomic Differences in Sickness Absence: The
Occupational Injuries and Illnesses in California.” Whitehall II Study.” British Medical Journal
Preventive Medicine 32(5):393-406. 306:361-366.
Lewchuck, W., A.L. Robb, and V. Walters. 1996. “The O’Donnell, M. 1997. “Health Impact of Workplace
Effectiveness of Bill 70 and Joint Health and Safety Health Promotion Programs and Methodological
Committees in Reducing Injuries in the Workplace: The Quality of the Research Literature.” The Art of Health
Case of Ontario.” Canadian Public Policy 22(3):225-243. Promotion 1(3):1-7.
Lucas, R.E. Jr. 1988. “On the Mechanics of Economic Parkes, K.R. 1987. “Relative Weight, Smoking, and
Development.” Journal of Monetary Economics 22:3-42. Mental Health as Predictors of Sickness and Absence
Marmot, M.G., A. Feeney, M.J. Shipley, F. North, and from Work.” Journal of Applied Psychology 72:275-286.
S.L. Syme. 1995. “Sickness Absence as a Measure of Pauly, M.V., S. Nicholson, J. Xu, D. Polsky, P.M.
Health Status and Functioning: From the UK Danzon, J.F. Murray, and M.L. Berger. Forthcoming.
Whitehall II Study.” Journal of Epidemiology and “A General Model of the Impact of Sickness Absence
Community Health 49(2):124-130. on Employees.” Health Economics.
Marmot, M.G., F. North, A. Feeney, and J. Head. 1993. Pelletier, K. 1991. “A Review and Analysis of the Health
“Alcohol Consumption and Sickness Absence: From and Cost-Effective Outcome Studies of
the Whitehall II Study.” Addiction 88:369-382. Comprehensive Health Promotion and Disease
Marmot, M.G., G.D. Smith, S. Stansfield, C. Patel, F. North, Prevention Programs.” American Journal of Health
J. Head, I. White, E Brunner, and A. Feeney. 1991. Promotion 5(4):311-313.
“Health Inequalities Among British Civil Servants: The —— 1993. “A Review and Analysis of the Health and
Whitehall II Study.” The Lancet 337:1387-1398. Cost-Effectiveness Studies of Comprehensive Health
Mérette, M. 2002. “The Bright Side: A Positive View on Promotion and Disease Prevention Programs at the
the Economics of Aging.” Choices 8(1):1-28. Worksite: 1991-1993 Update.” American Journal of
Montreal: Institute for Research on Public Policy. Health Promotion 8(1):50-62.
Messer, J., and W. Stone. 1995. “Worksite Fitness and —— 1996. “A Review and Analysis of the Health and
Health Promotion Benefit-Cost Analysis: A Tutorial Cost-Effectiveness Studies of Comprehensive Health
Review of Literature, and Assessment of the State of Promotion and Disease Prevention Programs at the
the Art.” AWHP’s Worksite Health 2:34-43. Worksite: 1991-1993 Update.” American Journal of
Health Promotion 10(5):380-388.
Emile Tompa text 11/27/02 2:18 PM Page 202
Emile Tompa
—— 1999. “A Review and Analysis of the Clinical and Simon, G., C. Barber, H.G. Birnbaum, R. Frank, P.E.
Cost-Effectiveness Studies of Comprehensive Health Greenberg, R.M. Rose, P.S. Wang, and R.C. Kessler.
Promotion and Disease Management Programs at the 2001. “Depression and Work Productivity: The
Worksite: 1995-1998 Update (IV).” American Journal Comparative Costs of Treatment Versus
of Health Promotion 13(6):333-345. Nontreatment.” Journal of Occupational and
—— 2001. “A Review and Analysis of the Clinical and Environmental Medicine 43(1):2-9.
Cost-Effectiveness Studies of Comprehensive Health Skodol, A.E., S. Schwartz, B.P. Dohrenwend, I. Levav, and
Promotion and Disease Management Programs at the P.E. Shrout. 1994. “Minor Depression in a Cohort of
Worksite: 1998-2000 Update.” American Journal of Young Adults in Israel.” Archives of General Psychiatry
Health Promotion 16(2): 107-116. 51:542-551.
Platt, S., S. Pavis, and A. Gazala. 1999. Changing Labour Smith, R.S. 1992. “Have OSHA and Workers’
Market Conditions and Health: A Systematic Literature Compensation Made the Workplace Safer?” In
Review (1993-98). Dublin: European Foundation for the Research Frontiers in Industrial Relations and Human
Improvement of Living and Working Conditions. Resources, ed. D. Lewin, O.S. Mitchell, and P.D.
Polanyi, M.F.D., J.W. Frank, H.S. Shannon, T.J. Sullivan, Sherer. Madison, WI: Industrial Relations Research
and J.N. Lavis. 2000. “Promoting the Determinants of Association.
202 Good Health in the Workplace.” In Settings for Health Solow, R. 1956. “A Contribution to the Theory of
Promotion: Linking Theory and Practice, ed. B.D. Poland, Economic Growth.” Quarterly Journal of Economics
L.W. Green and I. Rootman. Thousand Oaks, CA: Sage. 70(1):65-94.
Rivera, and L. Currais. 1999a. “Economic Growth and Stansfeld, S.A., A. Feeney, J. Head, R. Canner, F. North,
Health: Direct Impact or Reverse Causation?” and M.G. Marmot. 1995. “Sickness Absence for
Applied Economics Letters 6:761-764. Psychiatric Illness: The Whitehall II Study.” Social
—— 1999b. “Income Variation and Health Expenditure: Science and Medicine 40(2):189-197.
Evidence for OECD Countries.” Review of Development Steckel R.H. 2001/2002. “Health Indicators of the
Economics 3(3):258-267. Standard of Living.” Indicators: The Journal of Social
Romer, P.M. 1986. “Increasing Returns and Long-Run Health 1(1):139-159.
Growth.” Journal of Political Economy 94(5): 1002-1037. Sullivan, T., and J. Frank. 2000. “Restating Disability or
Schaeffer, M.A., A.M. Snelling, M.O. Stevenson, and R.C. Disabling the State: Four Challenges.” In Injury and
Karch. 1994. “Worksite Health Promotion Evaluation.” the New World of Work, ed. T. Sullivan. Vancouver:
In Economic Impact of Worksite Promotion, ed. J.P. Opatz. UBC Press.
Champaign, IL: Human Kinetics Publishers. Warner, K.E. 1992. “Effects of Workplace Health
Scholz, J.T., and W.B. Gray. 1990. “OSHA Enforcement Promotion Not Demonstrated.” American Journal of
and Workplace Injuries: A Behavioral Approach to Public Health 82(1):126.
Risk Assessment.” Journal of Risk and Uncertainty Warner, K.E., T.M. Wickizer, R.A. Wolfe, J.E.
3(3):283-305. Schildroth, and M.H. Samuelson. 1988. “Economic
—— 1997. “Can Government Facilitate Cooperation? An Implications of Workplace Health Promotion
Informational Model of OSHA.” American Journal of Programs: Review of the Literature.” Journal of
Political Science 41(3):693-717. Occupational Medicine 30(2):106-112.
Shannon, H., C.A. Mustard, and G. Hepburn. 2001. Wheeler, D. 1980. “Basic Needs Fulfillment and
“Opportunities to Enhance Measurement of Economic Growth: A Simultaneous Model.” Journal
Workplace Health in Workplace and Employee of Development Economics 7:435-451.
Survey.” Discussion Paper. Toronto: Institute for Wolfson, M., G. Rowe, J.F. Gentleman, and M. Tomiak.
Work & Health. 1990. “Career Earnings and Death: A Longitudinal
Shephard, R.J. 1992. “A Critical Analysis of Work-Site Analysis of Older Canadian Men.” Journal of
Fitness Programs and Their Postulated Economic Gerontology 48(4):S167-S179.
Benefits.” Medicine and Science in Sports and Exercise World Health Organization. 1999. “WHO on Health and
24(3):354-370. Economic Productivity.” Population and Development
Review 25(2):396-402.