The Impact of Health on Productivity: Empirical Evidence ... · omy and the productivity of the...

181

INTRODUCTION

Improving the living standards of popula-tions is a widespread societal objective. Acornerstone of living standards is the ability

of individuals to earn wages and profits in orderto purchase goods and services for consump-tion. In turn, wages and profits reflect the valueof the goods and services produced in an econ-omy and the productivity of the factor inputsused to produce them. Though living stan-dards, income and productivity are distinct con-cepts, the three are very much related. Thecorrelation between labour productivity andreal wages both across countries and over timeis quite high, indicating the importance of pro-ductivity growth rates for the improvement ofa country’s living standards (Harris 1999).Consequently, economists and historianshave focused much attention on better under-standing the determinants of productivitygrowth. There is increasing awareness thathuman capital is a key factor. Traditionally,human capital has been interpreted as edu-cation and skills. Recently, however, increas-ing attention has been given to health as aform of human capital.

Over the last few years a growing bodyof literature has developed on the macroeco-nomic and microeconomic relationship betweenhealth and productivity. This chapter reviewsthe theoretical underpinnings and empiricalevidence of this relationship. In particular, itaddresses the question: Would an improve-ment in the health status of working Canadianspay off in terms of higher aggregate produc-tivity? The evidence presented comes frommany countries, both developed and develop-ing, and spans a period of over 200 years. Thereview focuses on implications for public pol-icy and firm-level practices in developedcountries, particularly Canada.

Figure 1 provides an organizing frame-work for this review. It lists a number ofstrategies that the public and private sectorshave employed to promote the health of indi-viduals and populations, as well as severalmeasures of health to assess the effectivenessof these strategies. Some of the strategies areemployed with the express intention ofimproving human capital and, in turn, pro-ductivity. Others have improving health as aspecific goal. The more traditional strategiesof sanitation, nutrition and education are

TH

E R

EV

IEW

OF

EC

ON

OM

IC P

ER

FO

RM

AN

CE

AN

D S

OC

IAL P

RO

GR

ES

S | 2

00

2

The Impact of Health onProductivity: Empirical

Evidence and PolicyImplications

Emile Tompa

Emile Tompa text 11/27/02 2:18 PM 81

public-sector interventions targeted at popu-lation health, but they nonetheless have impli-cations for the health and productivity of thelabour force. Another traditional public-sec-tor intervention, occupational health and safe-ty, is targeted at the workplace and is focusedon the reduction of accidents and chemicalexposures and the resultant work-relatedinjuries, illnesses and disabilities. The strate-gies of health promotion and healthy-work-place promotion are newer, firm-levelinitiatives developed since the 1970s from thegrowing awareness that organizational-levelinterventions can be an effective means of pro-moting healthy lifestyles, reducing stress,improving employee wellness, and reducingsickness-related absence and health-care costs(Polanyi et al. 2000). Lastly, population healthis a new strategy based on the acknowledge-ment that the determinants of health are mul-tifactorial — biological, social and economic— and that health policy needs to take abroad, multisectoral approach (Frank 1995).

Figure 1 also lists several labour-produc-tivity and standard-of-living measures that canbe affected by improvements in the health of thelabour force. At the individual level, health candirectly increase general output (e.g., throughenhanced physical energy and mental acuity),yearly output (e.g., through reduced sickness

absence) and career output (e.g., through decreasedmorbidity or increased longevity, resulting in alonger career). At the aggregate level, these indi-vidual increases in output can translate intoincreases in labour productivity (i.e., output perhour worked, output per worker) and/or standardof living (i.e., GNP per capita) (e.g., by increas-ing the size of the active labour force relative tothe population).

This chapter proceeds as follows: Thenext section, “Human Capital, Health andProductivity,” reviews the theory on thedemand for health and its relationship to theaccumulation of human capital. “HistoricalTrends and Current Macroeconomic Evidence”reviews the historical economic evidence con-cerning the relationship between health andproductivity growth as well as current macro-economic empirical work on measuring thisrelationship. “Occupational Health and Safetyand the Cost of Work Disability” reviews thechanging nature of work and the implicationsfor traditional approaches to occupational healthand safety regulation — its ability to influencefirm and worker behaviours and, through these,health and productivity. “Health, SicknessAbsence and Firm-Level Practices” reviews themicroeconomic evidence concerning the rela-tionship between health and various produc-tivity markers, with a focus on sickness

182

Emile Tompa

FIGURE 1

Framework for the Review of Evidence on the Impact of Health on Productivity

Health Measures

Health statusHealth and functionSickness absenceDisabilityLongevity

Labour ProductivityMeasures

Output per hour workedOutput per paid labour

hourOutput per workerOutput per labour

force participant

Standard of LivingMeasures

Output per capita

Strategies

SanitationNutritionEducationHealth promotionHealthy workplaceOccupation, health &

safetyPopulation health

Emile Tompa text 11/27/02 2:18 PM Page 182

absence. Lastly, the key implications for poli-cy and future research are summarized.

HUMAN CAPITAL, HEALTH ANDPRODUCTIVITY

Grossman’s (1972, 2000) model forhealth demand provides insights into the rela-tionship among health, human capital andconsumption at the individual level, as well asa framework for modelling human capitalaccumulation and its relationship to produc-tivity at the micro and macro levels. The maincontribution of this model is that it offersinsights into modelling two key aspects ofhuman capital, health and education, and theirrelationship to labour supply, earnings andproductivity. The model is based on Becker’s(1965) household-production concept, whichin turn is premised on the notion that utilityis obtained not directly from market goods andservices, but, rather, from final consumptiongoods produced from market goods and serv-ices in conjunction with one’s own time. Forexample, leisure (a final consumption good)may be produced with the purchase of movietickets and one’s own healthy time. Somehousehold production processes provide utilitydirectly (e.g., the production of leisure), where-as others are inputs into other processes suchas educational development or labour force par-ticipation which provide utility indirectly. Afundamental aspect of the Grossman model isthat health or healthy time provides utility notonly directly but also indirectly, since it is acritical input into many production processes,as described above. Accordingly, health is botha final consumption good and a capital good.

Human capital theory is premised on thenotion that an increase in a person’s stock of

knowledge and health raises his or her produc-tivity in both market and non-market activi-ties. In the Grossman model, health capitaldiffers from other forms of human capital in itseffect on these activities. Health capital deter-mines the total amount of healthy time avail-able for them, whereas knowledge capitalaffects the productivity of the time spent onthem. This approach suggests that health cap-ital provides a flow of healthy time that is uni-form in quality, an “all or nothing” state. Analternative formulation would be to have healthcapital bearing on both the quality and quanti-ty of healthy time. Like all capital, health depre-ciates over time and is assumed to do so at anincreasing rate with age. Consequently, invest-ment is required to restore and/or maintainhealth stocks through household productionactivities that include inputs such as exercise,nutrition and health care. The model does notexpressly include spillover effects in the pro-duction of health and education, which can bean important contributor to the efficiency oftheir production (e.g., the health of parents canaffect child health outcomes, and some of theskills and knowledge acquired by a workerthrough educational pursuits can be transmit-ted to colleagues).

There is significant interplay between dif-ferent types of human capital, specificallybetween education and health. In the Grossmanmodel, higher levels of education are theorizedto improve the efficiency of gross health invest-ment.1 The empirical literature substantiates theexistence of this relationship (Grossman andKaestner 1997). Whether it is causal — and inwhich direction — is not clear, though Grossmanand Kaestner conclude that the evidence suggeststhe existence of a pathway from education tohealth (i.e., individuals with higher educationare better at producing health). A third variable,

183

The Impact of Health on Productivity: Empirical Evidence and Policy Implications


time preference, may also play an intermediaterole. Higher education may result in an indi-vidual placing more value on the future (i.e., itmay lower an individual’s time preference,which suggests that time preference is endoge-nous). Alternatively, a lower discount rate mayencourage an individual to seek higher levels ofeducation, which, in turn, bears on the optimallevel of health capital investment.2

The Grossman model can be used toidentify an individual’s labour supply as afunction of health. The principal implicationof the model is that health is determinedendogenously (i.e., by the other variables inthe model rather than by exogenous/externalfactors). In principle, education is also deter-mined endogenously, but since most peoplecomplete their education early in life it can betreated as an exogenous variable when model-ling labour supply, whereas health capitaldepreciates and requires ongoing investment(Currie and Madrian 1999). As the Grossmanmodel suggests, health is also an importantaspect of human capital, and an importantinput into market and non-market productionat the individual level.

At the aggregate level, Bloom andCanning (2000) identify four pathways bywhich health can affect productivity: a healthylabour force may be more productive becauseworkers have more physical and mental ener-gy and are absent from work less often; indi-viduals with a longer life expectancy maychoose to invest more in education and receivegreater returns from their investments; withlonger life expectancy, individuals may bemotivated to save more for retirement, result-ing in a greater accumulation of physical cap-ital; and improvement in the survival andhealth of young children may provide incen-tives for reduced fertility and may result in

an increase in labour force participation —which may, in turn, result in increased percapita income if these individuals are accom-modated by the labour market.

HISTORICAL TRENDS ANDCURRENT MACROECONOMICEVIDENCE

The empirical literature in economic his-tory provides substantive evidence concerningthe productivity impact of increased lifeexpectancy and reduced morbidity over the lastfew centuries in Europe and the United States(e.g., Costa and Steckel 1995; Fogel 1991,1994; Steckel 2001/2002). Fogel (1991) stress-es the importance of long-run dynamics andpresents evidence that improvements in healthwhich began some 300 years ago in Europeand North America have not yet fully runtheir course. This work suggests that anunderstanding of the key drivers of long-rundynamics may be of value to policy-makingin developed countries even today. To thisend, evidence from the historical economicsliterature is reviewed below, followed byempirical evidence based on data from morerecent periods that makes use of the growth-accounting framework. Similar measures ofhealth appear in both literatures.

Fogel (1991, 1994) presents historicaltrends in England, France and Sweden on twoanthropomorphic measures associated withnutrition, namely adult height and weight(also known as body mass index). Height andweight provide different information abouthealth. Adult height reflects the adequacy ofearly-childhood nutrition, whereas adult weightreflects the adequacy of adult nutrition. Usingmore recent evidence from Norway on the

184

Emile Tompa


relationship among height, weight and risk ofmortality, Fogel estimates the proportion ofthe decline in mortality in the three countriesthat can be associated with changes in theseanthropomorphic measures since the 18th cen-tury. Research on the relationship amongheight, weight and chronic disease from morerecent US data provides further evidence onthe detrimental health implications of below-average height and weight (relative to currentNorth American standards). Based on thisresearch, Fogel (1994) concludes that chronichealth conditions were significantly moreprevalent throughout the life cycle prior to theFirst World War. Consistent with Fogel’swork, Steckel (2001/2002) presents morerecent evidence on anthropomorphic measuresas proxies of health, supporting the notionthat health can influence productivity. Hefound that the simple correlation betweenaverage height and log of GDP per capitaranges from 0.82 to 0.88. Furthermore, henotes that the average height of Americans isfalling behind that of Northern Europeans,and that this trend may be reflective of grow-ing income inequality in the United States.

Fogel (1994) presents evidence of thehistorical impact of population health on labourforce productivity drawn from estimates ofcaloric intake in Britain and France in the 18thand early 19th centuries. He estimates that thedaily caloric intake for individuals in the bot-tom 10 percentile of consumption in Francewas so low that they did not have enough ener-gy for work, and that those in the next 10 per-centile had energy for only three hours of lightwork (0.52 hours of heavy work). In Englandthe situation was somewhat better. Only indi-viduals in the bottom 3 percentile of con-sumption lacked enough energy for work, andthose in the next 17 percentile had energy for

about six hours of light work (1.09 hours ofheavy work). Essentially, those in the bottom20 percentile had such poor diets that they wereexcluded from the labour force. As well, manyof those in the top 40 percentile were belowcurrent North American standards of averageheight and weight, and hence were likely sub-ject to premature chronic conditions and mor-tality. Subsequent improvements in nutritionraised the energy levels of individuals in thebottom 20 percentile of consumption suchthat they were able to enter the labour force.These improvements in nutrition also sub-stantially raised the capabilities of thosealready in the labour force. Fogel (1991, 1994)estimates that health and nutritional improve-ments alone can explain some 30 percent ofBritish growth in per capita income since1790. This value is similar to estimates of theproductivity impacts of health found in cross-country studies using data from the last 50years (World Health Organization 1999).

Recent macroeconomic research on pro-ductivity has emphasized the importance ofhuman capital. Like physical capital, humancapital in the form of education and health isdurable, lasting, and subject to accumulation(Lucas 1988; Romer 1986). One approach toincorporating human capital into the macro-economic modelling of productivity is to aug-ment the neoclassical growth-accountingequation developed by Solow (1956). Solow’sapproach to measuring multifactor productiv-ity growth is to associate it with the residualamount of output growth not explained bygrowth in the key inputs of labour and physi-cal capital. This approach is founded upon sev-eral contentious assumptions. First, it assumesthat technology (which is associated with theresidual) is exogenous, suggesting that labour-productivity growth rates will be the same

185



across economies once they reach a steady state.This runs counter to the trends of sustaineddifferences in growth rates observed acrossdeveloped countries. Another assumption ofthis approach is that there is perfect competi-tion such that market prices reflect social costs— that is, there are no information asymme-tries, appropriability problems, spillovers orother externalities. Intuitively, these appear tobe restrictive assumptions that are likely notmet in the real world, suggesting that there isa role for public policy. One can easily imag-ine spillover effects occurring from higher lev-els of human capital. They could have asubstantial impact not only on an individual’sown productivity, but also on the productivityof co-workers and on society as a whole. Anumber of measurement issues also arise withthis paradigm, the most salient of which ishow to deal with technological improvementsand quality changes embodied in both inputsand outputs.3 Quality improvements are rele-vant for both physical and human capital.

In response to the shortcomings of theSolow model, a new approach to growth account-ing has evolved, one that attempts to model thekey determinants of growth as jointly endoge-nous (Knowles and Owen 1997). One way toaugment the Solow model is to include the accu-mulation of human capital as well as physicalcapital, while still treating technology as exoge-nously determined. This diminishes the impor-tance of exogenous technological growth. Theendogenous growth literature attempts to cap-ture two aspects of the impact of health on pro-ductivity: its direct impact on the productionprocess — for example, improvements inhealth can increase productivity due to reducedincapacity, disability and days off sick; and itsspillover impact — for example, an improve-ment in the health of seniors can result in

reduced personal-care time required by familycaregivers who are members of the labour force.

Human capital in the form of educationhas received much attention in cross-countryempirical growth studies, and researchers havefound considerable support for its importanceas a productivity driver (Harris 1999). Humancapital in the form of health has received lessattention, but the relatively few cross-countrystudies that have included some measure ofhealth have found that it does have a signifi-cant and positive association with economicgrowth.4 Many of the empirical growth stud-ies that include health have focused on devel-oping countries (e.g., Bhargava et al. 2001;Hicks 1979; Wheeler 1980), though somehave included a broader range of countries(Barro and Sala-i-Martin 1995; Bloom et al.2001; Knowles and Owen 1995, 1997) andsome have focused specifically on OECD coun-tries (Knowles and Owen 1995, 1997; Riveraand Currais 1999a, 1999b).

These studies often use rather crudemeasures of health, likely due to the lack ofdata on more refined and comprehensive meas-ures that span both a reasonable time periodand a number of countries.5 Most studies usesome measure of life expectancy or mortality(life expectancy at birth, infant mortality rates,adult survival rates), though two recent stud-ies used per capita health-care expenditures(Rivera and Currais 1999a, 1999b). Lifeexpectancy has increased dramatically over thepost-war period in many developed countries(see Chart 1 for Canadian trends). Thoughmortality and life expectancy are importantmeasures of health status, they may not cap-ture the subtle changes in morbidity, healthbehaviours, health-related quality-of-life meas-ures, or other measures of health that are par-ticularly salient to developed countries today.

186

Emile Tompa


Furthermore, it is likely that the relationshipbetween health and productivity found in thesestudies is driven by data from developingcountries in the sample. Developed countriessuch as Canada and the United States are vir-tually indistinguishable across the health meas-ures used (Harris 1999). Chart 2 providesevidence of this fact. As is apparent in thegraph, the life-expectancy gradient is muchsteeper for low levels of gross national incomeper capita. Indeed, when Knowles and Owen(1997) estimated their specification for 22high-income countries, they found that thehealth measure they used, life expectancy, wasno longer significant, likely due to the lack ofvariability for this measure in the sub-sample.Rivera and Currais (1999a, 1999b) attempt tomake a case for their use of health-care expen-ditures rather than life expectancy as a meas-ure of health, but the meaningfulness of thisproxy of health for developed countries is alsoquestionable. Variations in health-care expen-ditures in developed countries are not highlycorrelated with health measures such as lifeexpectancy and infant mortality (comparingJapan and the United States highlights thispoint), and it is not clear whether the margin-al dollar spent on medical care reflects mor-bidity improvements.

A fully specified growth model shouldinclude all key inputs into the productionprocess and all drivers of productivity, includ-ing measures of all forms of human capital; oth-erwise one cannot be sure whether a particularvariable directly affects growth or is simply aproxy for missing factors. In particular, humancapital should include a measure of educationand skills, as well as health. All but one studyreviewed included some measures of educationand/or skills such as average years of schooling,primary/secondary/university enrolment, adult

literacy or years of experience. Interestingly, theeducation and skills variables were not signifi-cant in most specifications in the studies, withthe exception of Rivera and Currais (1999a,1999b) and Barro and Sala-i-Martin (1995),suggesting that health may be a more impor-tant determinant of productivity, particularlyfor developing countries. These results mightbe driven by other factors, such as the nature ofthe proxy being used for education or measure-ment error in the data.

The magnitude of the coefficient forhealth is difficult to compare across studies,due to a number of differences in the specifi-cations. Some studies used different measuresof health, while others used comparable oneswith slight variations in their specification.Some studies used the growth of labour pro-ductivity as the dependent variable, while oth-ers used the growth of total factor productivity;those that used the former generally made useof one of two denominators, an estimate of thesize of the labour force or the total population.6

There were other specification differences aswell, making direct comparison impossible.

187


Ag

e

66

68

70

72

74

76

78

80

1995

1990

1985

1980

1975

1970

1965

1960

1955

1950

CHART 1

Canadian Life Expectancy at Birth

Source: World Bank.


Nonetheless, a comparison of the range ofresults provides a sense of the impact thathealth can have on productivity. To this end,Table 1 contains information on the elastici-ties and percentage effects of health from stud-ies that included developed countries and fromwhich data could be extracted. Studies byRivera and Currais (1999a, 1999b) andKnowles and Owen (1995, 1997) suggest thatbetween 21 and 47.5 percent of GDP growthper worker (working-age person) over the last25 to 30 years can be explained by improve-ments in the health of populations (defined ashealth-care expenditures and life expectancy)at the country level. As noted, this range issimilar to the value estimated by Fogel in hiseconomic history work. Bloom et al. (2001)also found a significant relationship betweenhealth and GDP growth. Each extra year of

life expectancy is estimated to increase a coun-try’s GDP by 4 percent.

Endogenous growth studies substantiatethe importance of health for productivitygrowth, particularly in developing countries.Health may be equally important for growthin developed countries, but different aspects ofhealth, such as morbidity, vitality, mentalhealth and mental acuity, are likely more crit-ical for these countries than increases in lifeexpectancy. With the shift from manufactur-ing to services and the increasing importanceof new technologies in developed countries, thehuman-capital needs of the labour force havechanged. Intuitively, one can foresee an increas-ing role for mental health and acuity forknowledge workers providing high-end serv-ices. There is evidence of health improvementson this front. Recent research in the field of

188

Emile Tompa

90

80

70

60

50

40

30100 1,000 10,000 100,000

LE

GNIpc

CHART 2

Gross National Income Per Capita (GNIpc) Versus Life Expectancy (LE) at Birth, 1997

Source: World Bank.


psychology has found that populations indeveloped countries experienced steady gainsin intellectual ability over the course of the20th century, and some work suggests thatthis is attributable to improvements in healthand nutrition (World Health Organization1999). Further decreases in morbidity couldalso be critical for future productivity gains,given the aging of the labour force in many

developed countries. Moreover, according toMérette (2002) there is no evidence that theperformance of older workers is systematicallylower than that of younger workers. This maybe due in part to improvements in the healthof older workers and in part to a decrease in thephysical demands of industries in many devel-oped countries (as a result of both the shifttowards service industries and technological

189


TABLE 1

Macroeconomic Growth Studies with Data from Developed Countries that Include Measures of Health

Study Productivity Health Countries and Elasticitiesmeasures measures time period

Rivera andCurrais(1999a)

Rivera andCurrais(1999b)

Bhargava etal.( 2001)

Knowles andOwen (1995)

Knowles andOwen (1997)

Bloom,Canning andSevilla (2001)

Barro andSala-i-Martin(1995)

log differenceof GDP perworker, 1960–90

log differenceof GDP perworker, 1960–90

log GDPgrowth rate percapita

log difference ofGDP perworking ageperson, 1960–85

log difference ofGDP perworking-ageperson, 1960–85

log GDPgrowth rate

GDP growthrate per capita

log percentageof GDP spenton health care

log percentageof GDP spenton health care

log of adultsurvival rate

log of (80 yearsless lifeexpectancy atbirth)

log of (80 yearsless lifeexpectancy atbirth)

log of lifeexpectancy atbirth

log of lifeexpectancy atbirth

24 OECDcountries(1960–90)

24 OECDcountries(1960–90)

125 countries fromPen World Tables107 countries fromWorldDevelopmentIndicators(1965–90)

84 countries62 developing22 high-income(1960–85)

77 countries55 developing(1960–85)

Information notprovided

134 countries(developing anddeveloped)(1965–85)

0.21–0.22

0.28–0.33

varies by GDP+ve for low-income countries-ve for high-income countriesFor the poorest countries, a 1%change in adult survival rate isassociated with a 0.05%increase in GDP growth rate.

0.3810.3820.03

0.4490.475

0.04Each extra year of lifeexpectancy leads to anincrease of 4% in GDP.

0.046-0.082An estimate of .064 meansthat a one standarddeviation increase in lifeexpectancy (13 years) raisesGDP growth rates per capitaby 1.4 % per year.


improvements in manufacturing). Nonetheless,there is a need for further research into theimpact of health on productivity in developedcountries using measures of health that aremore salient to these countries.

OCCUPATIONAL HEALTH ANDSAFETY AND THE COST OF WORKDISABILITY

If health capital complements firm-spe-cific human capital in that it increases thereturns to firm-specific skills and knowledge,one might expect that employers would bewilling to bear the cost of investing in thehealth of workers in order to reap the benefitsof productivity gains. But if health capital isgeneric rather than firm-specific, the fact thatworkers can take it with them from job to jobsuggests that firms might be unwilling to bearthese costs, even if health capital increasesworker productivity (Currie and Madrian1999). In reality, health capital likely hassome degree of complementarity and somegeneric aspects. If this is the case, firms mayvoluntarily invest in the health of workers butnot necessarily to a socially optimal level.Consistent with this notion, developed coun-tries have recognized the importance of labourmarket institutions designed to protect thehealth and safety of workers through financialand regulatory mechanisms. The main policylevers for providing such incentives are occu-pational health and safety regulation and expe-rience-rated workers’ compensation insurance.Occupational health and safety regulation cov-ers a broad range of procedural and equipmentstandards and is generally enforced through asystem of inspections and fines. Experience-rated workers’ compensation insurance provides

financial incentives for safety consciousness byvarying insurance premiums at the sectoral andfirm level, in an effort to tie the costs of injuryand illness as closely as possible to the employ-ers responsible for them, without undulypenalizing any one firm for costly and unpre-ventable accidents.

In Canada, the direct cost of work-relat-ed injuries and illnesses exceeded $5.7 billionin calendar year 2000 (Institute for Work andHealth, 2002). This estimate includes indem-nity payments, insurance administrationexpenses and medical services that are paid byemployers through workers’ compensation pre-miums (Chart 3 provides data on the growth ofindemnity payments over the 1972-96 period).These direct costs substantially underestimatethe true cost of productivity losses attributableto work-related injuries and illnesses. The indi-rect cost estimate for Canada is $12 billion.This includes costs incurred by employers toaccommodate injured workers who return towork, recruitment and training costs incurredfor replacing injured workers, earnings lost byworkers due to injury and the lost home pro-duction of workers. Even these direct and indi-rect costs likely underestimate the true socialcost. For instance, they do not include costsassociated with pain and suffering or home careprovided by family members, and the numberof claims is less than the true number of work-related injuries.7 Clearly, the financial burdenof work-related injuries and illnesses is sub-stantial, but we do not know what proportionof this burden is preventable, the expendituresnecessary to reduce the burden, or whetherinsurance and regulation are the most effectivemeans of reducing the burden.

Over the past 10 to 15 years the numberof work-related injury and illness claims hasdecreased substantially in many jurisdictions in

190

Emile Tompa


Canada and other developed countries (see Chart4 for Canadian trends). In Canada, injury claimsdecreased by 40 percent between 1990 and1998, despite the fact that the labour forceincreased by 10 percent over this period(Mustard et al. 2001). A number of explanationshave been put forward for this trend. Thoughthere is evidence to support some of these, it isnot clear which factors are the predominantones.8 More specifically, it is not clear what frac-tion of the trend is attributable to the effective-ness of insurance and regulatory mechanisms.

There is a large body of empirical litera-ture on the effectiveness of insurance and regu-latory mechanisms using econometric techniques(reviews of this evidence are provided byCurington 1988; Hyatt and Thomason 1998,Kralj 2000; and Smith 1992). Taken as a whole,the empirical evidence on the impact of regula-tion is mixed. The US evidence suggests thatstandards can reduce certain types of injuries andthat a system of inspections provides, at best,

modest general deterrence unless reinforced withpenalties. Scholz and Gray (1997) found that reg-ulation facilitating cooperation, such as inspec-tions initiated by workers (regardless of penalty),can be more effective than coercive regulationsuch as regular inspections, unless penalties areimposed. The effectiveness of facilitative regula-tion is reinforced by Canadian evidence on theintroduction of regulations requiring joint healthand safety committees (Lewchuck et al. 1996).9

The evidence on workers’ compensation experi-ence rating suggests that financial incentives canbe an effective means of improving occupationalhealth and safety. The appeal of experience rat-ing is that it ties the cost of work-related injuriescloser to the firms experiencing them, whileallowing firms the flexibility to find the mostefficient means of improving health and safety.

One of the challenges of regulatory andinsurance mechanisms is that the nature ofwork and labour market experience havechanged profoundly since their introduction in

191


Exp

end

iture

s in

199

2$ (t

hous

and

s)

0

10000

20000

30000

40000

50000

60000

1996

-97

1992

-93

1988

-89

1984

-85

1980

-81

1976

-77

1972

-73

CHART 3

Total Workers’ Compensation IndemnityPayments in Canada

Source: Human Resources and Development Canadaand Statistics Canada

Cla

ims

per

em

plo

yed

ind

ivid

ual

0.00

0.01

0.02

0.03

0.04

0.05

0.06

0.07

1997

1993

1989

1985

1981

1977

1973

time loss no time loss

CHART 4

Time Loss/Fatality and No Time LossCompensation Claims Per Employee

Source: Human Resources and Development Canadaand Statistics Canada.


Canada, yet occupational health and safety reg-ulation and workers’ compensation continue tofocus on injuries and the impact of physicaland chemical exposures characteristic of man-ufacturing and resource-based industries. Theshift away from manufacturing to services, theincreasing use of technology, the introductionof new human-resource and management prac-tices, and the growth in demand for knowledgeworkers have all contributed to a dramaticchange in the nature of work-related injuriesand illnesses (Sullivan and Frank 2000). Forexample, mental-stress claims more than dou-bled in the United States between 1980 and1987 (Gnam 2000).10 Furthermore, the growthin non-standard work arrangements has notonly made it difficult for workers’ compensa-tion boards to assign firm-level responsibilityfor injuries and illnesses, but has also dramat-ically changed the nature of labour marketexperiences. Other factors such as incomeinequality, job insecurity and unemploymenthave also been shown to have a bearing on thehealth of individuals and populations (Deaton2001; Platt et al. 1999).

HEALTH, SICKNESS ABSENCE ANDFIRM-LEVEL PRACTICES

At the microeconomic level, research intothe impact of health on productivity focuses onreturns to employers and workers. Returns tothe accumulation of health capital can be real-ized by employers in the form of higher profits,by workers in the form of higher wages, or byboth. Pauly et al. (forthcoming) provide a the-oretical framework (hereinafter the Pauly frame-work) to identify the principal characteristics ofthe production process, the market for the goodor service being produced, and the labour mar-

kets that determine the size and distribution ofproductivity gains associated with healthimprovements. The framework focuses on aparticular proxy for health-related productivi-ty gains, namely decreases in sickness absence,but the premise of the model can be general-ized to all health-related productivity improve-ments.11 This framework is examined below,followed by a review of the microeconomic evi-dence concerning the relationship betweenhealth and productivity, and employers’ effortsto capture potential health-related productivitygains through health-promotion initiatives.Consistent with this literature, particular atten-tion is given to sickness absence as a proxy forproductivity. As background for this discussion,Canadian trends in sickness absence are provid-ed in Chart 5. Rates have been increasing forboth men and women over the last few years.This may be due in part to the economic recov-ery during the period. Similar to work-relatedaccident claim rates, sickness absence rates havea cyclical component (i.e., work-related acci-dent claims tend to increase during periods ofeconomic recovery due to factors such as theincreased pace of work). Nonetheless, the ratefor women has increased to levels above thoseof the late 1980s.

Many empirical studies on the cost ofsickness absence and the cost-effectiveness ofhealth-promotion initiatives assume that thedollar value of reducing sickness absence issimply the direct cost of wages paid to absentworkers (assuming that they are paid forabsences). In the Pauly framework, wages areactually the lower bound for losses. In manycases, total costs can be much higher due toindirect costs attributable to sickness absence.If the production process is team-based, or ifa penalty is incurred for failure to achieve tar-get output, then the cost of sickness absence

192

Emile Tompa


can be much higher if a perfect substitute isnot available to replace an absent worker (i.e.,if there is firm-specific knowledge capital). Inthe case of team production, sickness absencecan also reduce the productivity of co-work-ers. Penalties associated with failure to achievetarget output can also be significant if pro-duction is time-sensitive (e.g., perishablegoods and travel/transportation services).

The benchmark case used to elaborate theframework is a single homogenous product anda simple production process in which wagesreflect the incremental value of production (inwhich case the cost of absences is the wage rate).If health improvements are observable andtransferable to a new employer (e.g., smokingcessation and weight loss), then a worker whosehealth improves will receive higher wage offersfrom competing employers. Consequently, inthe long run the benefits of health improve-ments are fully captured by the worker in theform of higher wages.12 There is a large litera-ture in labour economics investigating the

impact of health status on wages, income andlabour force participation. A thorough reviewof this literature is provided by Currie andMadrian (1999). In general, research supportsthe notion that health is associated with wagesand income, though the magnitude of itsimpact appears to be sensitive to the measureof health used in the particular study. One ofthe patterns emerging from Currie andMadrian’s review is that health has a greaterimpact on the number of hours worked than onthe wages received by workers. Even if workersdo capture the benefits of health improvementsin the long run, employers may still have anincentive to undertake health-promotion ini-tiatives due to competitive pressures and inorder to reap the short-run benefits, particular-ly if there are complementarities between healthcapital and firm-specific knowledge capital(Currie and Madrian 1999).

Three factors determine the degree towhich the wage rate underestimates the costof an absence: the extent to which the pro-duction process relies on team work, the sizeof the penalty incurred for failure to achievetarget output levels, and the cost of replacingan absent worker with an equally productiveone. With full employment, the wage rate isa good measure of lost output for cases inwhich there is a perfect substitute for a work-er at the same wage rate (assuming that, in fullemployment, the wage rate reflects a conver-gence of firm, worker and societal values oflabour time). The cost of absence exceeds thewage rate if the replacement worker is lessproductive or costs more and if the productionprocess relies on team work or a penalty isincurred for failure to achieve target outputlevels.13 If there is less than full employment,the prevailing wage rate may differ from theequilibrium-wage rate, so the firm, worker

193


Day

s

0

1

2

3

4

5

6

2000

1999

1998

1997

1996

1995

1994

1993

1992

1991

1990

1989

1988

1987

male female

CHART 5

Canadian Absence Rates for Illness andDisability, Full-Time Paid Workers

Source: Statistics Canada (data series L91404 & L9140).


and societal values of labour time may differas well. If this difference is small, the gainsfrom reducing sickness absence would be sim-ilar to the full-employment case. In general,the benefits from reduced sickness absence aregreater than the wage rate. It should be notedthat the impact of sickness absence on pro-ductivity is greater when measured by outputper worker than output per hour, since someof the output not produced due to sicknessabsence is offset by the reduced number ofhours worked.

The Pauly framework underscores thepotential productivity gains of reducing sick-ness absence through investments in healthcapital. Implicit in this line of thinking is theassumption that poor health is the principalreason for sickness absence.14 But the etiologyof sickness absence is quite complex; poorhealth is but one of many factors that have abearing on sickness absence. Examples of otherfactors include personality, job-related atti-tudes and social context. Two recent literaturereviews (Alexanderson 1998; Harrison andMartocchio 1998) identify a broad range ofcausal factors investigated in the empirical lit-erature on sickness absence and find that fewstudies have investigated the impact of healthon sickness absence.15 In particular, short-termhealth conditions are rarely studied. Nicholsonand Martocchio (1995) describe this gap in theliterature as a “black hole.” Alexanderson notesthat even though most studies in her revieware from the fields of epidemiology and medi-cine, most do not use a medical model or evenconsider health status as an explanatory vari-able. This lack of attention to health is likelydue to a focus on sources of variance that areperceived to be avoidable (and thus amenableto change), though the distinction betweenavoidable and unavoidable is blurry at best. For

example, some people may believe that healthstatus is predetermined and that changes inhealth status are unavoidable, but an interven-tion such as an influenza shot can significantlyreduce the incidence of colds and flu, and thusmay reduce sickness absence more readily thanan intervention designed to increase workers’engagement in their work.

Nonetheless, there is evidence to supportthe notion that health status has a strong influ-ence on sickness absence. Most empirical stud-ies of sickness absence that have includedmeasures of long-term health, chronic conditionsor health behaviours have focused on self-report-ed health status, smoking, illicit-substance useand alcohol consumption. The Whitehall IIstudy provides some of the most compelling evi-dence for the impact of these factors on sicknessabsence (Marmot et al. 1995; Marmot et al.1993; North et al. 1993). Reported “average” or“worse” health over the 12 months precedingthe survey was associated with significantlyhigher levels of sickness absence compared toreported “good” health — a 60 percentincrease in short spells (seven days or less) anda twofold increase in long spells (more thanseven days). Significantly higher levels of sick-ness absence were also observed for individualsreporting recurrent health problems, long-standing illnesses or psychiatric symptoms.Mental health factors such as depression, anxi-ety and emotional stress are a frequently report-ed cause of sickness absence, particularlyamong women (Stansfeld et al. 1995), and havebeen found to be significant predictors ofabsence and disability (Garrison and Eaton1992; Kessler et al. 1999; Kouzis and Eaton1994; Simon et al. 2001; Skodol et al. 1994).Studies of health-related behaviours have foundthat smokers have higher rates of absence thannon-smokers (Bush and Wooden 1995; Leigh

194

Emile Tompa


1995; North et al. 1993; Parkes 1987) andthat illicit-substance users have high rates ofabsence (Bass et al. 1996; Normand et al.1990). The impact of alcohol consumption ismore complex. Problem drinking (a level ofalcohol consumption that results in social dys-function) appears to be the point at whichsickness absence is affected (Beaumont andHyman 1987; Casswell et al. 1988), causinglong-term absences in particular (Marmot etal. 1993).

Acute conditions related to respiratoryand gastrointestinal conditions are the primaryreasons for short-term absences (Stansfeld et al.1995), yet, as noted, few studies have investi-gated the nature and impact of short-termhealth conditions on sickness absence. This issurprising given that more immediate inroadsinto reducing sickness absence might be madeby addressing the factors that cause acute con-ditions. For example, Nichol et al. (1995) founda 43 percent drop in the rate of cold- and flu-related sickness absences among adults receiv-ing an influenza vaccine instead of a placebo.

The evidence regarding the impact ofhealth status on sickness absence, though pre-liminary, suggests that firm-level initiatives suchas health promotion are one means by whichemployers can reduce absence and increase pro-ductivity. Employers have undertaken a varietyof initiatives, ranging from targeted to multi-component programs (fitness/exercise programsare the most widespread). A large empirical lit-erature on the impact of these initiatives onhealth and productivity has accumulated over thepast 30 years. Most of the empirical studies haveconsidered one or more of five outcomes: riskreduction or behavioural change, health/medical-care costs, sickness absence, turnover and otherproxies of productivity. One of the motivationsfor these initiatives in the United States was the

growing cost of employee health care in the1970s and 1980s, which was increasing at a rateof up to 30 percent per year (Conrad 1988). Thefocus on health-care costs in many US employerinitiatives is understandable given that healthinsurance is generally provided by employers andcan make up a substantial component of the ben-efits provided to workers. In Canada, it is a lesssalient cost for employers, since health insuranceis primarily funded by the public sector, but isnonetheless a factor to consider at the macrolevel. Canada has also experienced substantialincreases in health-care costs since the 1970s, andthere is evidence that a growing proportion ispaid by the private sector (Polanyi et al. 2000).

Studies of the impact of employer-spon-sored health-promotion initiatives generally usequasi-experimental methods (participants self-select into the program) or non-experimentalmethods (no control group — e.g, before/aftercomparison) and cost-benefit or cost-effectivenessanalysis. The dearth of rigorous research, partic-ularly in the earlier literature in this field, hasbeen commented on in most reviews (Aldana1998; Baun 1995; Fielding 1990; Heaney andGoetzel 1997; Messer and Stone 1995;O’Donnell 1997; Pelletier 1991, 1993, 1996,1999, 2001; Shaeffer et al. 1994; Shephard 1992;Warner 1992; Warner et al. 1988).16 As a whole,these studies show mixed evidence, with non-experimental studies generally demonstratingpositive results and more rigorous studiesdemonstrating less positive results. On average,studies using experimental designs had positiveresults approximately 25 percent of the time,quasi-experimental designs 50 percent of thetime and non-experimental designs 100 percentof the time (Heaney and Goetzel 1997;O’Donnell 1997). In terms of sickness absence,the evidence on the impact of health promotionis mixed (Baun 1995).17 If the results of a broad

195



range of studies reviewed by Shephard (1992) canbe accepted as the true program effects, then theimpact on sickness absence is positive but mod-est; most of the studies found effect sizes in therange of 0.5 to two days per year of improvedattendance. In Messer and Stone’s (1995) reviewof cost-benefit studies, those that included thecost of reduced sickness absence as a benefitfound that benefits exceeded costs in all cases(benefit-cost ratios ranged from 1.07 to 3.90); forsome of the studies, the positive effects may alsohave been driven by reduced health-care costs.

The quality of research has improvedsubstantially in the most recent generation ofstudies, providing a rationale for employers toconsider such programs as a means of reducingthe health and economic costs of illness. Thefindings provide some support for the hypoth-esis that health and fitness can have an impacton sickness absence and productivity, thoughthe durability of these effects has not beenestablished. A realistic assumption is that aprogram will need to engage workers on anongoing basis if the changes are to be sus-tained. Though most studies have found onlymodest reductions in sickness absences as aresult of health-promotion programs, the directcosts of these absences represent only a lowerbound for the costs attributable to sicknessabsence, suggesting that productivity gainsmay be higher.

Polanyi et al. (2000) are less optimisticabout the potential for firm-level health-pro-motion initiatives. They list five reasons whysuch initiatives may be limited in their abilityto improve worker health: (1) this approachdoes not address the sources of human motiva-tion and behaviour that bear on health, name-ly the social and economic determinants ofhealth; (2) lifestyle is a less important factorbearing on health than socioeconomic status

(Marmot et al. 1991, 1993), which has healtheffects that can endure even after retirement(Wolfson et al. 1990); (3) if behavioural andlifestyle changes are to be enduring, the socialand cultural context that engendered themmust be modified as well; (4) health-promo-tion initiatives reach only a limited number ofworkers, since they tend to be implemented inwhite-collar settings; and (5) unless other job-related and organizational factors are alsoaddressed, workers may perceive such initia-tives as a self-serving effort by firms to reducetheir health-care and absenteeism costs.

Indeed, it is becoming increasingly evi-dent that general health and functioning arevery much affected by work experiences. Thereis a growing body of evidence showing that psy-chosocial workplace factors such as job control,psychological demands and social support havean important bearing on workers’ health(Shannon et al. 2001). This suggests that broad-ly based organizational initiatives may be moresuccessful in improving workers’ mental andphysical health and productivity than focusedhealth-promotion programs. Such broad-basedinitiatives can include a range of elements suchas: redesigning worksites with ergonomic prin-ciples in mind; redesigning work flows andcommunications channels to enhance commu-nication and social support; and providing flex-ible work hours, leave programs and daycarefacilities. Unfortunately, there are few experi-mental studies in the literature investigating theproductivity impact of such initiatives.

SUMMARY AND CONCLUSIONS

Economic well-being in developed coun-tries is growing increasingly dependent oninternational markets and integrated trade. To

196

Emile Tompa


maintain a high standard of living in the glob-al marketplace, countries must remain com-petitive, which in turn depends on thefostering of an innovative, productive labourforce. Researchers and policy-makers arebecoming increasingly aware that a country’sability to innovate and remain productivedepends on the characteristics and quality ofits human capital, key elements of which areeducation, skills and health. Most recently, thelinks between population health and econom-ic productivity have become a significant pol-icy concern. With the aging of the labour forcein Canada and many other developed countries,labour force health will become an even moreimportant issue in the near future. On the pos-itive side, there is evidence that older workersin developed countries are no less productivethan their younger counterparts.

Fogel’s research in economic historyhighlights the significance of populationhealth for productivity growth. He providescompelling evidence for the important role ofnutrition, particularly in early childhood, onhealth and functioning throughout the lifecycle. Fogel points out that these historictrends still have a bearing on the health of pop-ulations in developed countries today. Thisresearch provides valuable insights into the crit-ical role that policy can play in supporting pop-ulation health and ultimately productivity. Forexample, the enduring effect of childhoodexperiences is one of the themes in this work,suggesting that financial support for low-income families, parental leave policies andchild-care policies can help to ensure healthychild and adult outcomes. The evidence frommacroeconomic studies on health and produc-tivity corroborates the evidence from econom-ic history. Health is indeed an importantdriver of productivity even today. This area of

research focuses on life expectancy as a meas-ure of health, which can be an important pro-ductivity driver for developing countries butis less salient for developed countries. In orderto better serve the information needs of poli-cy arenas in developed countries, futureresearch in this area should identify morerefined measures of health to incorporate intotheir empirical analyses (e.g., measures offunctional status and mental health).

Work-related injuries and illnesses are amajor source of productivity losses for socie-ty as a whole, suggesting that public policycan play a crucial role in providing incentivesfor employers to dedicate more resources tooccupational health and safety. Most devel-oped countries recognize the importance oflabour market institutions designed to pro-tect the health and safety of workers, and haveestablished financial and regulatory mecha-nisms for this purpose. Different policy leversare available under the umbrella of regulationand insurance, some of which are more effec-tive than others. The evidence suggests thatoccupational health and safety regulation hashad only a modest impact, though prelimi-nary evidence shows that regulation focusingon facilitation rather than coercion may bemore promising. The evidence for experience-rating suggests that it is effective, partlybecause financial incentives allow employersthe flexibility to identify the most efficientmeans by which to improve workplace healthand safety. Both insurance and regulationtend to create incentives that focus on acuteinjuries and the health impacts of physicaland chemical exposures characteristic of theindustrial sector. Consequently, many of theimportant factors affecting labour force andpopulation health in developed countriesremain outside the purview of traditional

197



insurance and regulatory domains, due to thechanging nature of work and labour marketexperiences in these countries.

Sickness absence is an easily measuredproxy for productivity that is often investigatedin the organizational practices literature. Yet,surprisingly, the empirical literature on the caus-es of sickness absence has given little attentionto the role of health status as an explanatory vari-able. The work that has been undertaken in thisarea suggests that chronic and acute physical andmental conditions, as well as health-relatedbehaviours, explain a significant portion of sick-ness absence. This evidence provides support forhealth-promotion initiatives as a means of reduc-ing sickness absence and increasing productivi-ty. Empirical studies evaluating workplacehealth-promotion programs find that they areeffective in reducing absence and health-carecosts, though these reductions are modest andthe durability of program effects is not known.Nevertheless, the direct costs of absences areonly a lower bound for productivity losses attrib-utable to them; the productivity gains to be real-ized from decreasing sickness absences may besubstantially higher. Future research in this areashould focus on developing the tools to quanti-fy the various indirect costs associated with sick-ness absence, as well as improving themethodological rigour of intervention studies.Furthermore, there is evidence to suggest thatsuch initiatives should consider a broader set oforganizational factors related to work and theworkplace that bear on the health and produc-tivity of workers instead of narrowly focusing onbehaviour and lifestyle.

The fact that firm-level initiatives appearto have had a limited impact on productivitysuggests that the public sector has a role to playin improving the health of the labour force andpopulation as a whole and, in turn, overall pro-

ductivity. Identifying priorities in order toachieve the greatest gains from the resourcesinvested is a difficult task at best. Certainly,increasing health-care spending will not neces-sarily result in higher levels of population health,as evidenced by the differences in per capitahealth-care expenditures and health profilesacross OECD countries. The multifaceted natureof the factors that influence health suggests thatpolicies in a number of areas traditionally con-sidered outside the purview of health policy maybe important avenues by which the public sectorcan have an impact on population health. Keyareas are labour market policy, education poli-cy, and child-care and parental leave policy.Furthermore, though improving populationhealth is an important societal objective, thereare many other objectives competing for scarcepublic resources. Achieving an optimal balancewhen addressing societal objectives requires asound understanding of the policy options, theirimpact on the various objectives, and the costsassociated with each.

NOTES

1 Muurinen (1982) provides an alternative formulationin which education is theorized to lower the rate ofdepreciation of health stock.

2 A lower discount rate may also encourage investmentin health directly.

3 There are other measurement issues with regard tocapturing the full impact of health on welfare. Inparticular, utility is derived directly from health, aswell as indirectly through its role in market and non-market production activities, whereas only marketproduction activities are captured in standard outputmeasures. The direct value of health may be capturedto some degree through market goods and servicespurchased to improve health. The indirect value ofhealth from non-market activities may also becaptured to some degree in this way. At the core ofthis measurement issue is the fact that standardmeasures of output do not capture non-marketresources and activities. Costa and Steckel (1995)

198

Emile Tompa


compare alternative methodologies of measuringwelfare changes arising from changes in the health ofpopulations as an illustrative exercise to assess theconsistency of rankings.

4 Significance does not necessarily imply causality.In fact, the direction of effects might run fromproductivity growth to health, rather than thereverse. All of the studies address the reversecausality issue by testing for potentialendogeneity.

5 Currie and Madrian (1999) classify health measurestypically available in the data sets of developedcountries into eight categories: self-reported healthstatus, presence of functional limitations on theability to work, presence of functional limitations onother activities, presence of chronic/acute healthconditions, health-care utilization, clinicalassessments of health, nutritional status andmortality.

6 GDP per capita growth is highly correlated withGDP per worker growth, with the latter driving theformer. None of the studies corrected for the fact thataverage hours of work can vary substantially fromcountry to country.

7 Leigh et al. (2001) estimate that the total cost ofhealth care and lost productivity due to occupationalinjuries and illnesses in California is on par with thecosts of all cancers combined, and comparable to thecosts of heart disease and stroke.

8 These explanations support one of two propositions:that the trend reflects real decreases in workdisability, or that the trend is a reportingphenomenon. Following are some of theexplanations offered: de-industrialization hasresulted in a greater proportion of the labour forcebeing employed in the service sector, which isinherently safer; older capital is being replaced withnew capital, which embodies ergonomic andtechnological improvements that are inherentlysafer; employers are doing a better job of institutingsafety measures in response to insurance andregulatory incentives or because of an increasedawareness of the value of health capital; employersare adopting more aggressive claims-managementpractices; and workers’ compensation boards havetightened their eligibility requirements.

9 This study also found that firms that are reluctant toform joint health and safety committees show poorerhealth and safety performance, suggesting that theclimate of the internal-responsibility system is animportant element in its success. The authorssuggest that, if the internal-responsibility system isto be effective, special measures may be required toeducate employers and workers in such cases.

10 Mental-stress claims still represent only a smallfraction of workers’ compensation claims in NorthAmerica, though this may be due in part tolegislation passed in the early 1990s in manyjurisdictions limiting compensability for such claims(Gnam 2000).

11 Decreases in sickness absence are frequently used as aproxy for health-related productivity gains attributable toemployer-sponsored initiatives, primarily because sicknessabsence is readily observable and measurable, and dataregularly collected by human-resource departments. Twoother proxies considered in the literature are disability andturnover. It should be noted that health improvementscan increase not only the amount of time available forwork, but also the quality of time spent at work. Morecomprehensive empirical analyses attempt to assess theentire range of direct and indirect benefits and costsassociated with a health improvement.

12 In the case of a salaried worker who is required tomake up lost production time, the benefits are in theform of more convenient hours. Note that the valueof lost leisure time is not captured by traditionaloutput measures but is nonetheless an importantaspect of social welfare.

13 Firms may attempt to insure against losses by hiringextra workers as backups to cover for absences. Thisimperfect remedy will be relatively less costly forlarger firms than for smaller ones.

14 One of the shortcomings of the model is its narrowfocus on sickness absence, rather than the impact ofworker health on productivity in general. However,the basic tenets of the model can be generalized tothis broader perspective.

15 Harrison and Martocchio (1998) review over 500empirical studies on sickness absence culled from avariety of disciplines conducted from 1977 to 1996.

16 Frequently cited methodological shortcomings are:failure to use a control group or randomization ofindividuals between program and control groups, failureto adjust for confounders, lack of standardized measuresused for exposure and outcome, short time period ofstudies, small sample sizes, use of self-reported measures,failure to consider all indirect costs and benefits associatedwith an initiative, bias introduced by evaluations beingperformed by program advocates, and failure to considerthe worksite as the unit of analysis when initiatives areimplemented at a selection of worksites.

17 Baun (1995) reviews studies that assess a range of health-promotion initiatives for their impact in terms ofreducing sickness absence. He concludes that theevidence for the effectiveness of smoking-cessationprograms, health-risk assessment programs and exerciseprograms is mixed. Only stress-reduction programsappear to have a powerful effect on sickness absence.

199



REFERENCES

Aldana, S.G. 1998. “Financial Impact of Worksite HealthPromotion and Methodological Quality of theEvidence.” The Art of Health Promotion 2(1):1-8.

Alexanderson, K. 1998. “Sickness Absence: A Review ofPerformed Studies with (sic) Focused on Levels ofExposures and Theories Utilized.” ScandinavianJournal of Social Medicine 26(4):241-249.

Barro, R., and X. Sala-i-Martin. 1995. “EmpiricalAnalysis of a Cross Section of Countries.” In EconomicGrowth. New York: McGraw-Hill.

Bass, A.R., R. Bharucha-Reid, K. Delaplane-Harris, and M.A.Schork. 1996. “Employee Drug Use, DemographicCharacteristics, Work Reactions, and Sickness Absence.”Journal of Occupational Health Psychology 1(1):92-99.

Baun, W.B. 1995. “The Impact of Health PromotionPrograms on Sickness Absence.” In Worksite HealthPromotion Economics, ed. R.L. Kaman. Champaign, IL:Human Kinetics Publishers.

Beaumont, P.B, and J. Hyman. 1987. “The Work PerformanceIndicators of Problem Drinking: Some British Evidence.”Journal of Occupational Behaviour 8(1):55-62.

Becker, G.S. 1965. “A Theory of the Allocation of Time.”Economic Journal 75(299):493-517.

Bhargava, A., D.T. Jamison, L.J. Lau, and C.J.L.Murray. 2001. “Modeling the Effects of Health onEconomic Growth.” Journal of Health Economics20(3):423-440.

Bloom, D.E., and D. Canning. 2000. “The Health andWealth of Nations.” Science’s Compass 287:1207-1209.

Bloom, D.E., D. Canning, and J. Sevilla. 2001. “TheEffect of Health on Economic Growth: Theory andEvidence.” Working Paper #8587. Cambridge, MA:National Bureau of Economic Research.

Bush, R., and M. Wooden. 1995. “Smoking and AbsenceFrom Work: Australian Evidence.” Social Science andMedicine 41(3):437-446.

Casswell, S., L. Gilmore, and T. Ashton. 1988. “EstimatingAlcohol-Related Sickness Absence in New Zealand.”British Journal of Addiction 83(6):677-682.

Conrad, P. 1988. “Worksite Health Promotion: The SocialContext.” Social Science and Medicine 25(5):485-489.

Costa, D.L., and R.H. Steckel. 1995. “Long-Term Trendsin Health, Welfare, and Economic Growth in theUnited States.” Historical Paper #76. Cambridge,MA: National Bureau of Economic Research.

Curington W.P. 1988. “Federal versus State Regulation: TheEarly Years of OSHA.” Social Science Quarterly 69:341-360.

Currie, J., and B.C. Madrian. 1999. Health, Health Insuranceand the Labor Market. New York: Elsevier Science.

Deaton, A. 2001. “Health, Inequality, and EconomicDevelopment.” Working Paper #8318. Washington,D.C.: National Bureau of Economic Research.

Fielding, J.E. 1990. “Worksite Health PromotionPrograms in the United States: Progress, Lessonsand Challenges.” Health Promotion International5(1):75-84.

Fogel, R.W. 1991. “The Conquest of High Mortality andHunger in Europe and America: Timing andMechanisms.” In Favorite of Fortunes: Technology, Growthand Economic Development Since the Industrial Revolution,ed. D. Lander, C. Higgonet and H. Rosovsky.Cambridge, MA: Harvard University Press.

—— 1994. “Economic Growth, Population Theory andPhysiology: The Bearing of Long-Term Processes onthe Making of Economic Policy.” Working Paper#4638. Washington, D.C.: National Bureau ofEconomic Research.

Frank, J.W. 1995. “The Determinants of Health: A NewSynthesis.” Current Issues in Public Health 1:233-240.

Garrison, R., and W. Eaton 1992. “Secretaries, Depressionand Sickness Absence.” Women and Health 18:53-76.

Gnam, W. 2000. “Psychiatric Disability and Workers’Compensation.” In Injury and the New World of Work,ed. T. Sullivan. Vancouver: UBC Press.

Grossman, M. 1972. “The Demand for Health: A Theoreticaland Empirical Investigation.” Occasional Paper #119,Cambridge, MA: National Bureau of Economic Research.

—— 2000. “The Human Capital Model.” In Handbook ofHealth Economics, ed. A.J. Culyer and J.P. Newhouse.Amsterdam: Elsevier.

Grossman, M., and R. Kaestner. 1997. “Effects ofEducation on Health.” In The Social Benefits ofEducation, ed. J.R. Behrman and N. Stacey. AnnArbor: University of Michigan Press.

Harris, R.G. 1999. “Determinants of CanadianProductivity Growth: Issues and Prospects.”Discussion Paper #8. Ottawa: Industry Canada.

Harrison, D.A., and J.J. Martocchio. 1998. “Time forAbsenteeism: A 20-Year Review of Origins, Offshoots,and Outcomes.” Journal of Management 24(3):305-350.

Heaney, C.A., and R.Z. Goetzel. 1997. “A Review ofHealth-Related Outcomes of Multi-ComponentWorksite Health Promotion Programs.” AmericanJournal of Health Promotion 11(4):290-308.

Hicks, N.L. 1979. “Growth vs. Basic Needs: Is There aTrade-off?” World Development 7:985-994.

Hyatt, D., and T. Thomason. 1998. Evidence on the Efficacyof Experience Rating in British Columbia: A Report to theRoyal Commission on Workers’ Compensation in BC.Vancouver: Royal Commission on Workers’Compensation in BC.

200

Emile Tompa


Institute for Work and Health. 2002. Institute for Workand Health Website: www.iwh.on.ca

Kessler, R.C., C. Barber, H.G. Birnbaum, R. Frank, P.E.Greenberg, R.M. Rose, G. Simon, and P.S. Wang.1999. “Depression in the Workplace: Effects on Short-Term Disability.” Health Affairs 18(5):163-171.

Knowles, S., and P.D. Owen. 1995. “Health Capital and Cross-Country Variation in Income Per Capita in the Mankiw-Romer-Weil Model.” Economics Letters 48:99-106.

—— 1997. “Education and Health in an Effective-LabourEmpirical Growth Model.” Economic Record73(223):314-328.

Kouzis, A.C, and W.W. Eaton. 1994. “EmotionalDisability Days: Prevalence and Predictors.”American Journal of Public Health 84(8):1304-1307.

Kralj, B. 2000. “Occupational Health and Safety:Effectiveness of Economic and RegulatoryMechanisms.” In Workers’ Compensation: Foundationsfor Reform, ed. M. Gunderson and D. Hyatt. Toronto:University of Toronto Press.

Leigh, P.J. 1995. “Smoking, Self-Selection and SicknessAbsence.” Quarterly Review of Economics and Finance35(4):365-386.

Leigh, P.J., J.E. Cone, and R. Harrison. 2001. “Costs ofOccupational Injuries and Illnesses in California.”Preventive Medicine 32(5):393-406.

Lewchuck, W., A.L. Robb, and V. Walters. 1996. “TheEffectiveness of Bill 70 and Joint Health and SafetyCommittees in Reducing Injuries in the Workplace: TheCase of Ontario.” Canadian Public Policy 22(3):225-243.

Lucas, R.E. Jr. 1988. “On the Mechanics of EconomicDevelopment.” Journal of Monetary Economics 22:3-42.

Marmot, M.G., A. Feeney, M.J. Shipley, F. North, andS.L. Syme. 1995. “Sickness Absence as a Measure ofHealth Status and Functioning: From the UKWhitehall II Study.” Journal of Epidemiology andCommunity Health 49(2):124-130.

Marmot, M.G., F. North, A. Feeney, and J. Head. 1993.“Alcohol Consumption and Sickness Absence: Fromthe Whitehall II Study.” Addiction 88:369-382.

Marmot, M.G., G.D. Smith, S. Stansfield, C. Patel, F. North,J. Head, I. White, E Brunner, and A. Feeney. 1991.“Health Inequalities Among British Civil Servants: TheWhitehall II Study.” The Lancet 337:1387-1398.

Mérette, M. 2002. “The Bright Side: A Positive View onthe Economics of Aging.” Choices 8(1):1-28.Montreal: Institute for Research on Public Policy.

Messer, J., and W. Stone. 1995. “Worksite Fitness andHealth Promotion Benefit-Cost Analysis: A TutorialReview of Literature, and Assessment of the State ofthe Art.” AWHP’s Worksite Health 2:34-43.

Mustard, C.A., D. Cole, H. Shannon, J. Pole, T. Sullivan,R. Allingham, and S. Sinclair. 2001. “Does theDecline in Workers’ Compensation Claims1990–1999 in Ontario Correspond to a Decline inWorkplace Injuries?” Working Paper #168.Toronto: Institute for Work & Health.

Muurinen, J.M. 1982. “Demand for Health: AGeneralised Grossman Model.” Journal of HealthEconomics 1:5-28.

Nichol, K.L, A. Lind, K.L. Margolis, M. Murdoch, R.McFadden, M. Hauge, S. Magran, and M. Drake.1995. “The Effectiveness of Vaccination AgainstInfluenza in Healthy Working Adults.” New EnglandJournal of Medicine 335(4):889-893.

Nicholson, N., and J.J. Martocchio. 1995. “TheManagement of Absence: What Do We Know?What Can We Do?” In Handbook of Human ResourcesManagement, ed. G.R. Ferris, S.E. Rosen and D.T.Barnum. Oxford: Blackwell.

Normand, J., S.D. Salyards, and J.J. Mahoney. 1990. “AnEvaluation of Preemployment Drug Testing.” Journalof Applied Psychology 75(6):629-639.

North, F., S.L. Syme, A. Feeney, J. Head, M.J. Shipley,and M.G. Marmot. 1993. “ExplainingSocioeconomic Differences in Sickness Absence: TheWhitehall II Study.” British Medical Journal306:361-366.

O’Donnell, M. 1997. “Health Impact of WorkplaceHealth Promotion Programs and MethodologicalQuality of the Research Literature.” The Art of HealthPromotion 1(3):1-7.

Parkes, K.R. 1987. “Relative Weight, Smoking, andMental Health as Predictors of Sickness and Absencefrom Work.” Journal of Applied Psychology 72:275-286.

Pauly, M.V., S. Nicholson, J. Xu, D. Polsky, P.M.Danzon, J.F. Murray, and M.L. Berger. Forthcoming.“A General Model of the Impact of Sickness Absenceon Employees.” Health Economics.

Pelletier, K. 1991. “A Review and Analysis of the Healthand Cost-Effective Outcome Studies ofComprehensive Health Promotion and DiseasePrevention Programs.” American Journal of HealthPromotion 5(4):311-313.

—— 1993. “A Review and Analysis of the Health andCost-Effectiveness Studies of Comprehensive HealthPromotion and Disease Prevention Programs at theWorksite: 1991-1993 Update.” American Journal ofHealth Promotion 8(1):50-62.

—— 1996. “A Review and Analysis of the Health andCost-Effectiveness Studies of Comprehensive HealthPromotion and Disease Prevention Programs at theWorksite: 1991-1993 Update.” American Journal ofHealth Promotion 10(5):380-388.

201



—— 1999. “A Review and Analysis of the Clinical andCost-Effectiveness Studies of Comprehensive HealthPromotion and Disease Management Programs at theWorksite: 1995-1998 Update (IV).” American Journalof Health Promotion 13(6):333-345.

—— 2001. “A Review and Analysis of the Clinical andCost-Effectiveness Studies of Comprehensive HealthPromotion and Disease Management Programs at theWorksite: 1998-2000 Update.” American Journal ofHealth Promotion 16(2): 107-116.

Platt, S., S. Pavis, and A. Gazala. 1999. Changing LabourMarket Conditions and Health: A Systematic LiteratureReview (1993-98). Dublin: European Foundation for theImprovement of Living and Working Conditions.

Polanyi, M.F.D., J.W. Frank, H.S. Shannon, T.J. Sullivan,and J.N. Lavis. 2000. “Promoting the Determinants ofGood Health in the Workplace.” In Settings for HealthPromotion: Linking Theory and Practice, ed. B.D. Poland,L.W. Green and I. Rootman. Thousand Oaks, CA: Sage.

Rivera, and L. Currais. 1999a. “Economic Growth andHealth: Direct Impact or Reverse Causation?”Applied Economics Letters 6:761-764.

—— 1999b. “Income Variation and Health Expenditure:Evidence for OECD Countries.” Review of DevelopmentEconomics 3(3):258-267.

Romer, P.M. 1986. “Increasing Returns and Long-RunGrowth.” Journal of Political Economy 94(5): 1002-1037.

Schaeffer, M.A., A.M. Snelling, M.O. Stevenson, and R.C.Karch. 1994. “Worksite Health Promotion Evaluation.”In Economic Impact of Worksite Promotion, ed. J.P. Opatz.Champaign, IL: Human Kinetics Publishers.

Scholz, J.T., and W.B. Gray. 1990. “OSHA Enforcementand Workplace Injuries: A Behavioral Approach toRisk Assessment.” Journal of Risk and Uncertainty3(3):283-305.

—— 1997. “Can Government Facilitate Cooperation? AnInformational Model of OSHA.” American Journal ofPolitical Science 41(3):693-717.

Shannon, H., C.A. Mustard, and G. Hepburn. 2001.“Opportunities to Enhance Measurement ofWorkplace Health in Workplace and EmployeeSurvey.” Discussion Paper. Toronto: Institute forWork & Health.

Shephard, R.J. 1992. “A Critical Analysis of Work-SiteFitness Programs and Their Postulated EconomicBenefits.” Medicine and Science in Sports and Exercise24(3):354-370.

Simon, G., C. Barber, H.G. Birnbaum, R. Frank, P.E.Greenberg, R.M. Rose, P.S. Wang, and R.C. Kessler.2001. “Depression and Work Productivity: TheComparative Costs of Treatment VersusNontreatment.” Journal of Occupational andEnvironmental Medicine 43(1):2-9.

Skodol, A.E., S. Schwartz, B.P. Dohrenwend, I. Levav, andP.E. Shrout. 1994. “Minor Depression in a Cohort ofYoung Adults in Israel.” Archives of General Psychiatry51:542-551.

Smith, R.S. 1992. “Have OSHA and Workers’Compensation Made the Workplace Safer?” InResearch Frontiers in Industrial Relations and HumanResources, ed. D. Lewin, O.S. Mitchell, and P.D.Sherer. Madison, WI: Industrial Relations ResearchAssociation.

Solow, R. 1956. “A Contribution to the Theory ofEconomic Growth.” Quarterly Journal of Economics70(1):65-94.

Stansfeld, S.A., A. Feeney, J. Head, R. Canner, F. North,and M.G. Marmot. 1995. “Sickness Absence forPsychiatric Illness: The Whitehall II Study.” SocialScience and Medicine 40(2):189-197.

Steckel R.H. 2001/2002. “Health Indicators of theStandard of Living.” Indicators: The Journal of SocialHealth 1(1):139-159.

Sullivan, T., and J. Frank. 2000. “Restating Disability orDisabling the State: Four Challenges.” In Injury andthe New World of Work, ed. T. Sullivan. Vancouver:UBC Press.

Warner, K.E. 1992. “Effects of Workplace HealthPromotion Not Demonstrated.” American Journal ofPublic Health 82(1):126.

Warner, K.E., T.M. Wickizer, R.A. Wolfe, J.E.Schildroth, and M.H. Samuelson. 1988. “EconomicImplications of Workplace Health PromotionPrograms: Review of the Literature.” Journal ofOccupational Medicine 30(2):106-112.

Wheeler, D. 1980. “Basic Needs Fulfillment andEconomic Growth: A Simultaneous Model.” Journalof Development Economics 7:435-451.

Wolfson, M., G. Rowe, J.F. Gentleman, and M. Tomiak.1990. “Career Earnings and Death: A LongitudinalAnalysis of Older Canadian Men.” Journal ofGerontology 48(4):S167-S179.

World Health Organization. 1999. “WHO on Health andEconomic Productivity.” Population and DevelopmentReview 25(2):396-402.

202

Emile Tompa


The Impact of Health on Productivity: Empirical Evidence ... · omy and the productivity of the...

Documents

Transcript of The Impact of Health on Productivity: Empirical Evidence ... · omy and the productivity of the...