“Research on causes of sick-listing is done within many disciplines and it would be close to impossible to combine knowledge and theory from all the different sciences; medical, psychological, sociological etc. to one overall theory of causes for sick-listing” (SBU, 2003, p. 74).
Hoping for an „overall theory‟ of sick-listing or sickness withdrawal is likely to lead to despair. Sickness withdrawal is likely to have many underlying causes; medical, cultural, economical. One of the underlying causes of sickness withdrawal is health, let‟s begin there.
In the public health literature, the concept and importance of control seems to be more and more recognized as a major explanatory factor of health differences. Marmot‟s (2004) book: „The status syndrome: how social standing affects our health and longevity‟ showed that a social gradient in health is a widespread phenomenon. Statistics show (apparently pretty much whenever and wherever you look), that “health follows a social gradient: the higher the social position, the better the health… It runs from the top to the bottom of society, with less good standards of health at every level down the social hierarchy.” The social health gradient was first observed by Marmot through his famous „Whitehall studies‟ of British civil servants. A seemingly homogenous group: “It does not contain the richest or the poorest in the country; no one has a private jet and no one is unemployed or unthinkable for employment. All have great job security.” Still “the fact that civil servants in the second grade from the top have worse health than those at the top shows that we are dealing not only with the effects of absolute deprivation. Rather position in the hierarchy is important. This suggests some concept of relative rather than absolute deprivation… a psychosocial concept.”
In „The status syndrome‟ Marmot develop arguments for the importance of control and social participation: “the degree of control you have over your life and the opportunities you have to fully participate in society will strongly predict your health, your life quality and your longevity.” But Marmot cannot be blamed for focusing too much on too few causes of health. In a later (2006) book, he writes that “the social gradient has shown us how sensitive health is to social and economic factors, and so enabled us to identify the determinants of health among the population as a whole” (quotes from Marmot 2004, p. 58-59, 246 & 2006, p. 2, 6). Later in this paper (p. 13) a most helpful model of the social determinants of health is presented.
A brief overview of theories used in different scientific fields to understand health or sickness withdrawal will be given here9, followed by an attempt to establish a framework for thinking about regional variances in SIU.
A model that has been of central importance in the field of psychology
Sickness absence in this field has often been analyzed together with other forms of “withdrawal behaviors” such as quitting work (“turnover”) or late arrival (“tardiness”). These withdrawal behaviors have often been ranked, with tardiness as the mildest form and turnover the most extreme. These different forms of withdrawal have been viewed as consequences of low job satisfaction. A large empirical literature has tried to establish the connection between job satisfaction and withdrawal – without much success. (SBU 2003 p75)
A model used in the field of sociology
A theory best placed in this field of research is that of a „culture of absence‟, which has been defined, by Chadwick-Jones et al (1982) as “…beliefs and practices influencing the totality of absences – their frequency and duration – as they currently occur within an employee group or organization.” The culture of absence theory links actual absence to common perceptions regarding absence within different groups; such as: what are the legitimate grounds for absence, how long should a “reasonable” absence spell be, and so on and so forth. Operationalization10 and measurement of absence cultures are difficult, and the empirical literature on the topic is therefore limited (SBU 2003 p. 77-8)
“Decades of research in sociology and psychology have demonstrated that a sense of control of life in general is a robust predictor of physical and mental well-being” (Toivanen 2007)
Unlike psychological and economical theories (which will be referenced shortly) stress theories have more of an interdisciplinary character, with important contributions from the medical, sociological, and psychological fields. These theories tend not to focus on absence but on sickness generally. A large share of the research has been on medical problems linked to stress, such as cardiovascular diseases and the more psychological symptoms of anxiety and depression. To varying degrees these theories propose different causal chains leading to sickness absence, it is for example suggested that stress affects the employees motivation to work, or that sickness absence is used as a method of stress control. (SBU 2003, p. 80)
Concerning the work situation it is “without a doubt” Karaseks “demand-control” theory that has garnered the most attention and generated the most empirical research. According to this theory it is particularly jobs that combine high demands with low control that have negative health consequences, these jobs are referred to as “high-strain”. (SBU 2003, p.
The critical argument that is made clear in Marmot (2004) is that once people pass a certain minimum level of material well-being (as in Sweden) – another kind of well being is more central: “Autonomy…and the possibilities we have for full social engagement and participation is of decisive importance for health… It is inequalities within these areas that have great importance for creating the social health gradient. Behind the status syndrome is the degree of control and participation.”
Another example of a stress theory is that of “Person Environment Fit” where stress is defined as a lack of fit, or poor interplay, between the person and the environment. A third example is the “Effort-Reward Imbalance” theory which is rather self-explanatory. (SBU 2003, p. 81
Medical explanatory models
Medical evaluations are fundamental in the sickness insurance process. A doctor decides both if a person „really‟ is sick, and whether it means a reduction in work capacity for the individual, which is a criterion to be eligible for sickness insurance. Sick-listing can be viewed as a certificate that the patient has a reduced work capacity, it can also be seen as an ordination or prescription, where sickness absence is ordered or prescribed on the assumption that continued work may worsen or prolong the sickness spell. In some cases the doctor may order sick-listing primarily to protect the patient‟s surroundings, as would be the case if the patient had an infectious disease. (SBU 2003, p. 82)
“During the last 40 years, especially during the periods when sickness absence has increased, doctor‟s sickness certification practice have been discussed and even mentioned as causing the increase in sickness absence…It is often asked if doctors have the right, or sufficient education for this task.” (SBU p. 23) An open question is how much the view on the very concept of sickness varies between doctors and within the public, over time and within groups. These kinds of questions are dealt with by for example the historical and philosophical sciences (and will not be further considered in this thesis). (SBU 2003, p. 83
Economical and rational choice theories
Worker absence due to illness is not something economists have been terribly interested in, causes of absence from work in general is more studied, but then theory is often formed only to explain the part of absence that is unrelated to sickness, that is, that part of all absence that may be considered a result from rational choices made by healthy employees from a utility maximization perspective (SBU, 2003). In a stylized way, one could say that economists have tended to ask: why do people go to work at all? While for example psychologists have asked: why don‟t people go to work?
Economical theories on sickness absence usually take as their starting point the assumption of persons being rational actors, seeking to maximize personal welfare or “utility”. Personal utility is seen as a function of consumption (demanding time at work) and leisure (time off work). In the simplest version of this model zero absence compensation is assumed. Utility is then maximized for a certain distribution between consumption and leisure; deciding the amount of „sickness‟ absence11. It follows from the model that „sickness‟ absence will increase with increased contractual hours of work. An increased wage will have two offsetting effects, a person‟s leisure (time off work) will be more „expensive‟ (greater alternative costs through the “income effect”), but with a higher wage a person will also be able to work less for the same amount of consumption as when the wage was lower (the “substitution effect”).
This simple model has then been expanded in several ways, crucially by assuming sickness insurance compensation. “A central theme in economical research is the negative effects of [insurance compensation].” (SBU 2003, p. 78) Two such effects are „adverse selection‟ and „moral hazard‟.
„Adverse selection‟ refers to the case where the insured and the insurers have asymmetric information (i.e. access to different information). If insurance is voluntary and the asymmetric information problem is severe, predominantly high-risk individuals will be interested in insurance, normal and low risk individuals would not want to be insured since they would have to pay unfairly high premiums due to the „gaming of the system‟ performed by the high risk individuals.
If all workers are forced to be insured, and the asymmetric information problem is not too severe, as is the case with the Swedish sickness insurance system (the asymmetric information problem is reduced in the Swedish system due to the medical criteria‟s for sick-listing mentioned above). In the Swedish „single payer‟ system then, the risk is spread over all insured, something economic theory12 (and evidence) indicate is more efficient for the society, and probably beneficial even for most low-risk individuals. If the system was voluntary (private insurance only) the low risk group might opt out of insurance altogether or be forced to pay unfair premiums (assuming asymmetric information).
„Moral hazard‟ means that insured (here: workers) change their behavior according to the insurance system, so, if the sickness insurance system is more generous there will be more absence and vice versa. In a system with 100% absence compensation (as for example in Norway13) the simple model would predict that all workers were absent all the time since there would be no opportunity cost of „leisure‟ (not working). Since this prediction is proven false, other non-economical or longer term costs of absence have been built in to more advanced models. Health can also be introduced into these models, partly by assuming absence to be a function of actual sickness, partly by assuming that the value of leisure increases with actual sickness. (SBU 2003, p. 78-9)
Moral hazard is likely to be a relevant phenomena concerning regional SIU and will be touched upon below. Asymmetric information however is not likely to be relevant in explaining the regional variance in SIU. As Lindbeck et al (2009) write:
Since the Swedish sick-pay insurance system is mandatory and uniform, the rules and the availability are well-known to all individuals in the country. (Even immigrants are informed about the details of the social insurance system when settling down in Sweden.) … [T]he acquisition and interpretation of information about the Swedish sick-pay insurance system is a trivial task. This means that group effects can hardly be interpreted as the result of the dissemination of information about the availability of sickness benefits. This strengthens our interpretation of group effects as the results of social norms, rather than of information. (p.4
Labor mobility and health
A connection between labor mobility and sickness-withdrawal with stress as the „medium‟ is implied by existing theory. In summary: 1) Higher labor mobility implies better job-matching for individuals. 2) According to the “Person Environment Fit” stress theory this would reduce stress. 3) Stress in turn is identified as the cause of several illnesses in the medical literature.
Job transition theory (what happens to workers who quit) previously regarded turnover from a negative stress perspective; turnover itself was seen as a potentially disruptive and stressful event. More recently however, transition theory regards turnover from a positive, “proactive growth perspective”. According to this perspective, workers who turnovers strive to find work that fits their personality and allows active learning and growth. From this perspective then, turnover reduces rather than produces strain. “The empirical research seems to support this last „positive‟ perspective” (de Croon et al 2004).
The idea that higher labor mobility implies better job-matching was recently echoed in a Swedish report. First there are a number of studies connecting agglomeration with increased productivity; Anderson & Thulin (2008) refers to at least three Swedish14, and a number of foreign studies supporting this connection. They reference Yankow and Wheeler (both 2006) whom investigate the causes of this „urban productivity premium‟ and in their conclusions they both point out that better job-matching in “thick and dense markets” is likely to be an important explanation to the „urban productivity premium‟ (see Andersson & Thulin 2008 p. 26-9 for references)
Why would a simple job mismatch have health effects? Brunner and Marmot (2006), two medical scientists write the following on the topic of stress:
We are now beginning to recognize that people‟s social and psychological circumstances can seriously damage their health in the long term…The power of psychosocial factors to affect health makes biological sense. The human body has evolved to respond automatically to emergencies. This stress response activates a cascade of stress hormones which affect the cardiovascular and immune systems…if the biological stress response is activated too often and for too long, there may be multiple health costs. These include depression, increased susceptibility to infection, diabetes, high blood pressure, and accumulation of cholesterol in blood vessel walls, with the attendant risks of heart attack and stroke. (p. 28) To summarize:
- If a worker is poorly fitted, or mismatched, that is: if he does not like his place of work, he will experience stress. (Person Environment theory)
- Long term stress is recognized to be the cause of multiple health costs. (Brunner & Marmot 2006)
- In regions with higher labor mobility rates there is evidence of better job-matching, (see Andersson & Thulin 2008 for references) therefore regions with higher labor mobility rates can be expected to have less stressed i.e. healthier workers and therefore „consume‟ less sickness insurance.
Empirical support for a connection between labor mobility and health
Rothstein & Boräng (2005) theorized that “lack of mobility on the labor market is a cause of the increase long-term sick listed and early retirees” (p. 187). They argued that low labor mobility in Sweden compared to Denmark could be a cause for the higher rates of sickness withdrawal in Sweden compared to Denmark. They refer to a number of studies to establish the difference in labor mobility (p. 197-8):
Denmark has the highest labor mobility in Europe (Denmark‟s Statistics 2004). The share of newly employed (less than one year‟s employment) is 23 percent in Denmark, in comparison with barely 16 percent in Sweden (von Otter 2003)…In Sweden the share with ten years or longer tenures was 46.7 percent in 2000 while it was 31.1 percent in Denmark (Auer and Cazes 2003)…Labor mobility within Sweden‟s public sector is lower than the EU-average while publicly employed are overrepresented among sick-listed. In the Danish public sector labor mobility is in line with the national average and the Danish public sector does not have nearly as strong overrepresentation of sick-listed as the Swedish public sector (Auer & Cazes 2003; Swedish Ministry of Finance 2002).
Rothstein & Boräng (2005) also refer to at least nine Swedish studies, and a number of foreign which they (convincingly) argue support the idea of a connection between low labor mobility and sickness absence. They conclude that: “it appears to be important from a health perspective to have a job that one likes and there appears to be many people whom for health reasons want to change jobs but does not. An increase in voluntary workplace changes could therefore have a positive effect on workers health.” (p. 191)
Rothstein & Boräng (2005) suggested that the low labor mobility generally in Sweden15 (in the country as a whole), may be a cause of high SIU. This thesis will investigate if regional differences within Sweden show signs of affecting SIU. If such signs are found it will strengthen R&B‟s theory, and conversely if they are not found that would at least indicate that the effects of regional variation in labor mobility on SIU is not strong, something they themselves may well expect, paraphrasing: “We want to emphasize that we imagine this idea of low labor mobility to explain only a small part of the high SIU in Sweden” (p. 187-8). They point to the difficulties in testing their theory: “Ideally it would be done through a massive panel database (i.e. where a large group of individuals are followed over a long time). Lacking these data we have tried to substantiate our theory by comparing Sweden and Denmark.” (p. 193). This thesis is partly an attempt to „test‟ this theory with regional data. Since the regional variance in labor mobility is expected to have only a small effect on SIU, therefore we need to control for confounding variables, other factors that affect SIU. What are those factors?
Understanding the regional variance in SIU
There are only three earlier regressions studies on regional variation in SIU in Sweden published by the Swedish Social Insurance Agency. Of these studies Dutrieux & Sjöholm‟s (2003) paper is the most helpful. The other two studies by Olsson (2004 & 2006) are very short on theory. All three studies are explorative in character, and many regional variables assumed to be connected with SIU are tested for significance. The 2003 study lays some theoretical grounds for thinking about the subject, summarized below.
Theoretically, two categories of factors affecting regional SIU rates can be distinguished
- Factors affecting the populations‟ demand of SIU. For example: demographics (e.g. age of population), health situation, social and economic „well-being‟ labor market situation, attitudes.
- Factors affecting the will/possibilities of the instances granting sickness compensation to actually grant it. For example: density of doctors, attitudes and knowledge of doctors, and resources and routines (culture) at social insurance offices
This thesis will analyze variables exclusively from the former category, i.e. variables believed to affect the need (or demand) for sick listing. The reason is simply that variables from the second category were unavailable.
The regional populations demand for sickness insurance, i.e. their SIU, may depend on at least three kinds of explanations.
- SIU should depend on health.
- SIU can be seen as the result of a cost-benefit analysis the individual makes between the utility (and possibility) of working versus not working.
The need for SIU may depend on the working environment for employees.
Concerning point i) SIU should depend on health. Brunner & Marmot (2008) present a conceptual framework (figure 4, below) where “factors operating beyond the level of the individual, as well as individual characteristics are recognized. Thus, Social structure (top-left of the diagram) influences well-being and health (bottom right).” The authors describe their model as follows:
“The influences of social structure operate via three main pathways. Material circumstances are related to health directly and via the social and work environment. [The social and work environment] in turn shape psychological factors and health-related behaviors.” (p. 8
2 Theoretical framework
2.1 Understanding the regional variance in SIU
2.2 Earlier research on regional variation in SIU in Sweden
3 Dependant variables
7 Conclusions .
GET THE COMPLETE PROJECT