Risk level assessment and occupational health insurance expenditure: A gender imbalance.
It is tested whether occupational risk explains differences in reimbursements from occupational-injury insurance schemes in relation to socioeconomic differences in all municipalities in Stockholm county, Sweden. An occupational risk level is formed, which considered the proportions of workers in various industrial sectors and the probability of a worker being injured in each. A regression analysis is performed, treating socioeconomic condition and risk level as predictors of reimbursement. After controlling for variation in socioeconomic factors, occupational-risk level explains the pattern of payments to men but not to women. From a gender perspective, it can be concluded that women, as a group, are not compensated for their occupational risks to the same extent as men. [C] 2002 Elsevier Science Inc. All rights reserved.
Keywords: Equity; Gender; Occupational injuries and diseases; Risk; Social insurance
Postinjury compensation is paid according to specific social-insurance agreements (between employers' associations and trade unions) that cover most workers in Sweden (Statistics Sweden, 1995). Reimbursements from national occupational-health insurance represent cost of illness, and the burden this imposes on society. At an individual level, reimbursement is linked to each recipient's income and occupational-health conditions. A person with higher income gets more payments every time (s)he becomes injured. But at the societal level during a one-year period, the scenario is different. The deprived areas, where the number of low-income persons is higher (Table 1), are supposed to receive more payments in total than affluent areas with higher number of persons with high-income. The reason is that the persons with low income get injured more often. Thus, both at a point of time and a one-year period, incidences of injury is higher in a deprived area than in an affluent area.
The aim of this study is, therefore, to analyze the effect of socioeconomic conditions and occupational risk involved on reimbursements from occupational-injury insurance to workers in different industrial sectors in the municipalities of Stockholm in Sweden. The hypotheses intend to test for both male and female workers if there is a socioeconomic difference in reimbursement payments and if occupational risk explains differences in reimbursements from the occupational-injury insurance scheme. Finally, we will test if occupational injury risk explains the reimbursements to male and female workers for given socioeconomic condition.
2. Materials and method
The study is designed as an ecological analysis of socioeconomic differences in reimbursements among the 25 municipalities in Stockholm county. The analysis is divided into two parts. The first part comprises a descriptive statistics of reimbursements from occupational injury insurance on the basis of the socio-economic condition of the municipalities. The second part consists of regression analysis in order to observe the reflection of socioeconomic condition and occupational risk level on reimbursements.
All 777,456 registered residents in gainful work (383,800 men and 393,656 women) in 1993 are encompassed by the study with regard to reimbursements. Reimbursements (sickness compensation and disability annuity) of occupational injury insurance from Sweden's social-insurance registry for 1993 are employed.
Sickness compensation is paid to persons whose work capacity is reduced by at least 1/4 or more due to injury or disease. Compensation for the first 14 days is paid by the employer, and is thereafter covered by the occupational injury insurance. Sickness compensation encompasses persons with minimum income of 896 US$ during a year. No reimbursement is provided for persons on income exceeding 7.5 times "base amount," which is set annually by the Swedish Government (base amount for 1993 = 38,565 US$).
Persons who, as a consequence of work injury, had their personal income permanently reduced by at least one-fifth has a right to compensation in the form of annuity. Generally, the level of the annuity is calculated in relation to the insured person's sickness-compensation-based income.
2.1. Socio-economic variables
Six variables are selected in order to classify the 25 municipalities in Stockholm county into three groups (deprived, intermediate and affluent). The variables, used elsewhere, are average income from gainful work, dwelling space, education, immigrant density, proportion of welfare recipients and proportion of unemployed (Bobak et al., 1998; Friis et al., 1998; Janlert, 1991; Kawachi et al., 1999; Smith et al., 1998).
2.2. Specific socioeconomic municipality-groups
The 25 municipalities of the Stockholm county are allocated to three groups, in accordance with the socio-economic variables through K-means cluster analysis (SPSS, 1999, version 9.0). Five, seventeen and three municipalities have been allocated to deprived, intermediate and affluent municipality-groups respectively. Table 1 shows that affluent municipalities are on average possessing higher proportion of all variables that enrich socioeconomic condition of an area and vice versa.
The second part of the paper comprises three regression models having reimbursements from occupational injury insurance as the dependent variable, while socioeconomic condition and occupational injury risk in the municipalities as independent variables both separately and together respectively. The small number of observations (25 municipalities) and correlation among six socio-economic variables create a problem to determine the contribution of variables on reimbursements. We, therefore, have created two socioeconomic deprivation factors (SEF1 and SEF2) through a factor analysis (SPSS, version 9.0).
2.3. Socioeconomic deprivation factors
Two socioeconomic deprivation factors are created from six variables through factor analysis (SPSS, version 9.0). The first factor, SEF1, represents a combination of three variables -- average income, dwelling space and educational level. We suggest that SEF1 has a negative relationship with reimbursement. The second factor, SEF2, encompasses high values on immigration density, proportion unemployed and proportion of welfare recipients. This factor is hypothesized to be positively related to reimbursements from the occupational-health insurance system.
2.4. Occupational risk-level (ORL) assessment
All occupational injuries registered in Sweden in 1993 (19,464 accidents and 57,118 diseases) that resulted in a sickness of more than seven days for insurance purposes are used to assess the risk level of nine industrial sectors (Statistics Sweden, 1995). Frequency of occupational injuries, that is incidence per worker during 1993, is used to measure the probability of being injured in a specific industrial sector. Distribution of workers between industries is used to assess the risk level of each municipality. The following formula summarizes the definition of occupational-risk level (ORL) of a municipality:
[Risk.sub.mi] = [euro] [P.sub.wj] * [P.sub.j] (j = 1,2, .., 10; i = 1,2, .., 25)
[P.sub.wj] = proportion of workers in industry j, and [P.sub.j] = probability of a worker to be injured if (s)he works in industry j.
Occupational risk levels of each municipality -for male, female and all workers are assessed. Nine industries and one nonspecified worker group (j = 10), following the Nordic Occupational Classification System (SNI, 1992), are included in the study.
2.5. The regression models
The regression models will have the following structure,
Log (Re) = [alpha] + [[beta].sub.1] * SEF1 + [[beta].sub.2] * SEF1 + [[beta].sub.3] * ORL + [epsilon]
where, Re implies the reimbursements per 10 000 persons, a is a constant, [[beta].sub.1] [[beta].sub.2] and [[beta].sub.3] are coefficients of socioeconomic deprivation factors, that is, SEF1, SEF2 and occupational risk level respectively. [epsilon] is the error term. The short regression models are also analyzed by excluding independent variables from the model above.
3.1. Part I
Socioeconomic difference in reimbursement payments -- descriptive statistics
During 1993 a total of USD 41,907,823 is paid out as sickness compensation, with male workers receiving 67%. USD 55,037,829 is paid in disability annuity, and 65% of this went to male workers. Table 2 shows reimbursements by payment category -- as total payments and per 10,000 workers in paid employment -- divided into the three municipality groups. The variation between municipalities in the intermediate and deprived groups is considerable, with ranges from 0.55 to 1.55 and from 0.96 to 1.80, respectively, compared with Stockholm County as a whole. All municipalities in the affluent group receive lower reimbursements than the county's average. All but one of the deprived municipalities shows higher reimbursement per 10,000 workers than the average for Stockholm County.
Payments to male workers, per 10,000 are much higher than those to female workers. Men received more than double the amount of reimbursement than women from the occupational sickness fund. The scenario is not much different with regard to the disability-annuity fund; in all affluent municipalities, reimbursement is much lower than the average for the County as a whole. By contrast, in the deprived-municipality group, all municipalities except Nynashanm showed higher reimbursement than the average for Stockholm County. Nynashamn received 4% less than the average.
3.2. Part II
In the regression analyses, the socioeconomic variables associated with low and high shares in reimbursements are compared (model 1, Table 3). Significant differences are observed for all workers. There is a significant difference in reimbursement related to both socioeconomic factors (SEF1 and SEF2)--for both men and women. As expected, the first factor (SEF1) is found to be negatively related to reimbursement, and the second (SEF2) positively related. Further, a simple regression analysis demonstrates that reimbursement from occupational-injury insurance is associated with risk level in the municipalities. Both of these outcomes are expected. The model (model 2, Table 3) provides a much stronger prediction for male than for female workers ([R.sup.2] = 62.4 for men and [R.sup.2] = 27.4 for women).
Male workers are at greater risk than female workers according to the risk-level assessment. The nationwide occupational risk levels (ORLs) for male and female workers are 0.208 and 0.181 respectively. The most hazardous municipalities in Stockholm are Norrtalje for men and Sodertalje for women. The safest municipality for male workers is Danderyd; its counterpart for female workers is Taby. The assessed ORL for every municipality is included in the regression models.
Socio-economic difference in ORL is observed. When both risk level and socioeconomic variations are taken into account at the same time, by means of a multiple regression model (model 3, Table 3), a difference between men and women is found with regard to reimbursements. After controlling for variation in socioeconomic condition, occupational-risk level explains the pattern of reimbursement payments to men. However, the occupational-risk level does not significantly explain the distribution of reimbursements to women. Nevertheless, differential reimbursement is explained by variation in socioeconomic conditions (model 1, Table 3).
Another multiple regression analysis, taking reimbursements as the dependent variable and income from gainful work, occupational risk level (ORL) and gender as the predictors has been performed. We have found the predictors, that is, income from gainful work (t-value 2.993; p-value = 0.004), ORL (t-value = 2.140; p-value = 0.038) and gender (t-value = 2.620; p-value = 0.012) to be significant.
In this study, socioeconomic condition and occupational risk levels of the municipalities are estimated independently. The socioeconomic condition is assessed on the basis of all inhabitants, while occupational risk levels solely involve the working population. A socioeconomic difference in occupational risk is found in the analysis. Consequently, reimbursements from occupational-injury insurance also explain the socioeconomic difference. Occupational-injury risk also explains differences in reimbursements, and also gender differences. The adjusted correlation coefficient for the male model is 62.4, while that for the female model is only 27.4. One of the most interesting questions posed is whether occupational-risk level is still powerful enough to explain the distribution of reimbursements to male workers after controlling for socioeconomic variation. Reimbursements to male workers are explained well by risk in the workplace, but not those to female workers.
The two main findings are that there is socioeconomic variation in reimbursements and that there is an inconsistency between payments made to men and women. Concentration of workers in hazardous jobs contributes to the accumulation of occupational risk as well as to social deprivation in some geographic areas. Restructuring of occupational composition in municipalities, so that there is greater evenness across occupations would lead to greater equality in reimbursements by municipality. However, it is not only occupational structure that determines the pattern of reimbursements. Socioeconomic factors, which may be influenced by governmental intervention, are also important. Socioeconomic inequality may also have some influence on occupational hazards in the workplace. Psychosocial stress from relative deprivation, disrupted social cohesion, insufficient investment in social capital and underinvestment in human resources have all been suggested as factors involved in ill-health at population level (Chiang, 19 99).
Women are represented in almost equal numbers as men on the labor market in Stockholm. However, women work fewer hours (Inregia, 1995; Statistics Sweden, 1994). Women suffer more from occupational diseases, whereas men have more occupational injuries (Statistics Sweden, 1995). Difference in applicability of regulation to judge the injury cases, i.e., between diseases and accidents, may explain the difference in reimbursements between women and men. Job segregation also contributes to the differential between the genders. Amount of reimbursement is directly related to level of income and prevalence of occupational hazards. Since women work fewer hours, their income level is much lower than that of men. In Stockholm county the income of men is 23% higher than that of women (Inregia, 1996). The proportion of women is higher in lower income groups, and decreases as income rises. This income effect can result in lower reimbursements to women. However, this is reflected in the analysis by controlling for socioecon omic conditions. One reason for the gender differential with regard to reimbursements is that occupational-health insurance is established on the basis of full-time employment. Accordingly, part-time workers, who are more likely to be female, will receive less compensation because of their lower income. This means that the level of risk for a full-time worker is better reflected in his/her reimbursements than that of a part-time worker. It might be argued that the probability of injury is higher for a full-time worker than for a part-time worker. When female workers are exposed to equal hazards as male at work, women, due to lower income are not fully compensated according to their occupational risk.
(*.) Corresponding author. Tel.: + 1-46-8-5177-79-89;
fax: + 1-46-8-30-73-51.
E-mail address: email@example.com (J. Khan).
Bobak, M., Pikhart, H., Hertzman, C., Rose, R., Marmot, M., 1998. Socioeconomic factors, perceived control and self-reported health in Russia. A cross-sectional survey. Social Science and Medicine 47, 269-279.
Chiang, T.L., 1999. Economic transition and changing relation between income inequality and mortality in Taiwan: regression analysis. British Medical Journal 319, 1162-1165.
Friis, R., Yngve, A., Person, V., 1998. Review of social epidemiological research on migrants' health: findings, methodological cautions, and theoretical perspectives. Scandinavian Journal of Social Medicine 26, 173-80.
Inregia, 1995. Statistik om Stockholms Lan: Forvarvsarbetande Befolkning 1993 efter Naringsgren och Kon, (Statistic about Stockholm County: Gainful Working Population 1993 according to Industrial Classification and Sex, Report number 9).
Inregia, 1996. Arsstatistik for Stockholms Lan och Landsting 1994-1996 (Statistical Yearbook of Stockholm County, 1994-1996).
Janlert, U., 1991. Work Deprivation and Health: Consequences of Job Loss and Unemployment. PhD thesis. Karolinska Institutet, Department of Social Medicine. Sundbyberg, Stockholm.
Kawachi, I., Kennedy, B.P., 1999. The relationship of economic inequality to mortality: does the choice of indicator matter? In: Kawachi, I. et al. (Eds.), The Society and Population Health Reader. The New Press: New York.
Smith, G.D., Hart, C., Watt, G., Hole, D., Hawhorne, V., 1998. Individual social class, area-based deprivation, cardiovascular diseases risk factors, and mortality: The Renfrew and Paisley study. Journal of Epidemiology & Community Health, 52, 399-405.
SNI 92, 1994. Standard for svensh naringsgrens-indelning 1992: innehalbbeskrivningar and nychlar (Swedish standard for classification of occupational industries: contents and keys), SCB, Stockholm.
Statistic Sweden, 1994. Arbetskraftsundersokningen, Arsmedeltal 1993 (Labor Force Investigation, yearly average 1993), SCB. Stockholm.
Statistics Sweden, 1995. Occupational Diseases and Occupational Accidents 1993. National Board of Occupational Health. SCB, Stockholm.
Table 1 Socio-economic variables distributed across three groups of municipalities (deprived, intermediate and affluent) in Stockholm county, mean and two standard deviations. 1993 Socio-economic Deprived municipalities variables (n (*) = 79,422) mean SD Average income (a) 21.48 1.08 Dwelling space (b) 1.41 0.04 Education (c) 12.04 1.46 Immigrants (d) 18.94 10.69 Welfare recipient (e) 10.00 2.74 Unemployed (f) 6.68 0.80 Socio-economic Intermediate municipalits variables (n (*) = 828,559) mean SD Average income (a) 23.96 0.97 Dwelling space (b) 1.40 0.07 Education (c) 14.74 0.98 Immigrants (d) 16.24 4.09 Welfare recipient (e) 8.82 1.67 Unemployed (f) 5.50 0.83 Socio-economic Affluent municipalities variables (n (*) = 158,095) mean SD Average income (a) 29.74 2.03 Dwelling space (b) 1.57 0.04 Education (c) 17.83 0.31 Immigrants (d) 12.63 0.58 Welfare recipient (e) 3.67 0.58 Unemployed (f) 3.88 0.68 (*)Refers to number of persons eligible for sickness compensation in each municipality. (a)Average income from gainful work in USD 1000 per person and year. (b)Number of rooms per person, (c)Percentage share of inhabitants with at least 3-year high school education. (d)(e)Percentage share in total population. (f)Percentage share of registered job seekers in total population. Table 2 Reimbursements (USD) of sickness compensation and disability annuity from occupational injury insurance, by total, per 10,000 inhabitants with gainful work and ratios between observed (O) and expected (E) reimbusements, 1993 Municipalities Sickness compensation Men Women Total per 10,000 Total per 10,000 Deprived Botkyrka 1,093,546 711,342 806,282 527,568 Norrtalje 1,055,629 945,396 437,411 425,828 Nynashamn 194,906 358,217 98,147 195,941 Sundbyberg 475,973 675,331 146,327 202,137 Sodertalje 1,583,513 845,126 819,976 464,208 Intermediate Ekero 240,655 465,034 102,561 199,807 Haninge 2,158,955 1,391,976 812,428 531,833 Huddinge 2,611,237 1,454,323 751,609 411,773 Jarfalla 1,036,962 725,402 397,203 275,281 Nacka 827,556 531,336 524,992 320,743 Salem 174,469 554,221 88,580 275,608 Sigtuna 553,243 688,028 151,312 191,631 Sollentuna 699,977 551,510 536,967 415,416 Solna 320,911 277,821 250,207 200,840 Stockholm 10,692,277 723,468 5,888,743 371,651 Tyreso 870,284 993,248 272,949 305,210 Upplands Bro 192,825 396,107 103,476 219,835 Upplands Vasby 845,365 952,524 371,189 413,581 Vallentuna 256,768 441,409 114,989 203,844 Vaxholm 81,644 452,572 90,981 515,765 Varmdo 527,729 814,396 137,706 229,242 Osteraker 886,266 1,089,047 516,214 671,979 Affluent Danderyd 127,372 203,015 32,065 49,490 Lidingo 240,335 279,981 166,594 185,620 Taby 389,206 264,784 151,312 103,532 Total 28,137,603 733,132 13,770,20 349,803 Municipalities Disability annuity Men Women Total per 10,000 Total per 10,000 Deprived Botkyrka 1,375,297 894,618 760,839 497,834 Norrtalje 2,133,659 1,910,854 814,409 792,844 Nynashamn 698,393 1,283,574 262,613 524,283 Sundbyberg 787,815 1,117,786 565,792 781,589 Sodertalje 3,402,976 1,816,180 2,367,878 1,340,511 Intermediate Ekero 407,196 786,853 181,530 353,653 Haninge 2,121,864 1,368,062 851,396 557,342 Huddinge 2,681,387 1,493,393 910,249 498,685 Jarfalla 1,460,714 1,021,836 800,906 555,067 Nacka 943,132 605,542 601,964 367,769 Salem 296,854 942,992 173,476 539,752 Sigtuna 746,664 928,571 380,522 481,917 Sollentuna 1,362,368 1,073,407 602,822 466,364 Solna 652,108 564,546 422,755 339,344 Stockholm 10,934,452 739,854 6,811,528 429,890 Tyreso 1,058,242 1,207,763 342,280 382,735 Upplands Bro 378,298 777,111 178,258 378,709 Upplands Vasby 1,008,741 1,136,609 466,811 520,123 Vallentuna 566,317 973,556 212,267 376,294 Vaxholm 121,335 672,586 67,013 379,895 Varmdo 839,955 1,296,227 336,036 559,407 Osteraker 590,151 725,179 308,574 401,684 Affluent Danderyd 163,926 261,279 145,290 224,247 Lidingo 353,641 411,977 358,517 399,461 Taby 646,307 439,694 382,312 261,589 Total 35,731,792 931,000 19,306,037 490,429 Municipalities O/E Deprived Botkyrka 1,06 Norrtalje 1,66 Nynashamn 0,96 Sundbyberg 1,11 Sodertalje 1,80 Intermediate Ekero 0,73 Haninge 1,55 Huddinge 1,54 Jarfalla 1,03 Nacka 0,73 Salem 0,92 Sigtuna 0,92 Sollentuna 1,00 Solna 0,55 Stockholm 0,90 Tyreso 1,15 Upplands Bro 0,71 Upplands Vasby 1,21 Vallentuna 0,81 Vaxholm 0,81 Varmdo 1,18 Osteraker 1,17 Affluent Danderyd 0,29 Lidingo 0,51 Taby 0,43 Total 1,00 Table 3 Correlation coefficients and outcomes of regression analyses (a) of socioeconomic differences in payments (b) of sickness compensation and disability annuity from occupational injury insurance, distributed across all municipalities in Stockholm county, by gender, 1993 Male Female Explanatory [R.sup.2] = 57.5 t-value p-value [R.sup.2] = 55.3 variables Beta Beta Model 1 SEF1 (c,f) -0.740 -5.395 0.000 -0.679 SEF2 (d,f) 0.503 3.607 0.001 0.503 Female Total Explanatory t-value p-value [R.sup.2] = 61.2 t-value p-value variables Beta Model 1 SEF1 (c,f) -4.815 0.000 -0.750 -5.707 0.000 SEF2 (d,f) 3.670 0.001 0.533 4.051 0.001 Male Female Explanatory [R.sup.2] = 62.4 t-value p-value [R.sup.2] = 27.4 variables Beta Beta Model 2 ORL (e) 0.800 6.391 0.000 0.552 Female Total Explanatory t-value p-value [R.sup.2] = 55.3 t-value p-value variables Beta Model 2 ORL (e) 3.172 0.004 0.756 5.536 0.000 Male Female Explanatory [R.sup.2] = 65.7 t-value p-value [R.sup.2] = 55.0 variables Beta Beta Model 3 SEF1 (c,f) -0.312 -1.465 0.158 -0.605 SEF2 (d,f) 0.300 2.016 0.057 0.493 ORL (e) 0.518 2.472 0.022 0.157 Female Total Explanatory t-value p-value [R.sup.2] = 64.1 t-value p-value variables Beta Model 3 SEF1 (c,f) -3.719 0.001 -0.478 -2.322 0.030 SEF2 (d,f) 2.997 0.007 0.391 2.566 0.018 ORL (e) 0.923 0.366 0.341 1.674 0.109 (a)Standardized regression coefficients and explained variation. (b)Payments age-standardized per 10,000 insured. (c)Comprising mean income, dwelling space and educational level. (d)Comprising proportion of immigrants, welfare recipients and unemployed in all municipalities. (e)Occupational risk level (ORL) calculated on the basis of occupational injury incidences in the Swedish industries and occupational structure of the municipalities. (f)99% of variation in the original socio-economic variables jointly explained by socioeconomic deprivation factors, SEF1 and SEF2.
|Printer friendly Cite/link Email Feedback|
|Author:||Khan, Jahangir; Jansson, Bjarne|
|Publication:||The Journal of Socio-Economics|
|Article Type:||Statistical Data Included|
|Date:||Nov 1, 2001|
|Previous Article:||Market management of the essential economy: An essay.|
|Next Article:||The dynamics of the mutualist movement in Portugal: A time series analysis.|