Sickness absence in compulsory and voluntary health insurance: the case of Sweden at the turn of the twentieth century

ABSTRACT

At the turn of the twentieth century, Swedish health insurance was organised according to the Western European models of both voluntary, ‘fraternal’ principles and compulsory, ‘factory scheme’ principles. In this paper, we trace the characteristics of both organisational forms, and compare the sickness absence by considering the role of risk selection and mitigation across a large panel of voluntary and compulsory health insurance societies operating in Sweden between 1900 and 1910. We find that voluntary societies used a wide set of rules and practices in order to select and monitor members in order to keep down the number of sick cases. Compulsory societies applied shorter waiting periods and offered more medical treatment, leading to more frequent but shorter sickness absences.

KEYWORDS:

• Health insurance
• sickness absence
• developing economies
• voluntary insurance
• compulsory insurance

JEL-CODES:

• N00
• N23

Acknowledgements

The paper has benefited from comments and criticism from the editor and anonymous referees. We gratefully acknowledge the financial support received from Umeå University through the grant ‘Mobility and regional development’ and for support for the projects: Forte, ‘Sociala normer och sjukförsäkring’ and the Swedish Research Council ‘Efficiency and equality in private and public insurance’.

ORCID

Lars Fredrik Andersson http://orcid.org/0000-0002-1413-3707

Notes

1 Unhealthy people will face the same cost, but receive greater benefits than healthy people unless such differences in risk are priced correctly.

2 Sickness funds were most common in urban areas. 52.2% of all sickness fund members were situated in cities in 1909. The sickness funds established in the countryside were largely situated in municipalities or connected to factories or mills (Lindeberg, 1949).

3 In France, only miners were to be compulsorily insured (Murray, 2007).

4 Sickness insurance in the case of loss of income was the foundation of the business of Swedish occupational health societies, although they could also provide medical support, for wives and children too, burial insurance and a small pension in the case of invalidity or old age. In some compulsory societies, only a certain part of the workforce was permitted in the society, excluding less skilled workers.

5 Voluntary societies could use penalties in the cases when the rules of the society were infringed. The penalties were larger in compulsory societies, but the difference is not significant. The penalties were on average between 0.20 and 0.40 SEK per member and the average premium was on 10.27 SEK.

6 Riksarkivet, Stockholm (RA), Socialförsäkringskommittén. Sjukkassestadgar mm. 320147, vol. 28: 1915: ‘Stadgar för Tumpa bruksarbetares begravningskassa’.

7 RA, Socialförsäkringskommittén, Sjukkassestadgar mm. 320147, vol. 28: 1915: ‘Festskrift till sjukkasserörelsens i Finnspång 100-års jubileum’.

8 Besides the support from employers, both compulsory and voluntary societies often received donations from philanthropists, but also from the local parish. In some cases, societies viewed it to be the local parish’s obligation to support the society financially. The health insurance society for workers in Grycksbo received support from the local parish, but it also demanded support from a parish close by, since the society had members from that parish as well (Bergström, 1999).

9 The sample encompasses in total 847 societies across the country. To examine organisational form, we are, however, forced to restrict the sample to societies supplying policy charters. After imposing that restriction, the data covers 512 societies, of which 145 are compulsory. Most of the societies are in operation for 7 years of more, and only a few are active for only one or two years. The distribution of T is fairly similar for voluntary and compulsory societies and, for our sample of societies in comparison to the population of societies. A comparison between societies included in the sample and the population shows that the distribution of T is similar. For a few individual variables, the sample data differed from the population as a whole. When running t-test for all variables, we find that the sampled societies was significant younger (15.3 compared to 18.0) and larger (236 compared to 205). For the other variables, no significant differences were identified.

10 To compare benefits across time and space is expressed in real terms with the 1905 price level of Stockholm as reference year/place. The fixed price calculations are based on cost-of-living-standard indices constructed for each city between 1901 and 1910. The cost-of-living includes prices for food stuff, fuel and rent (Kommerskollegii, 1904–1912; Socialstyrelsen, 1919, 1933).

11 About half of the Swedish occupational health insurance societies based their underwriting on ex post premium payments.

12 RA, Socialförsäkringskommittén, Sjukkassestadgar mm. 320147, vol. 28: 1915: ‘Stadgar för Skromberga Arbetares Fortsättningskassa’; ‘Stadgar för Tumpa bruksarbetares begravningskassa’. See also Lindeberg (1949, p. 151).

13 RA, Socialförsäkringskommittén, Sjukkassestadgar mm. 320147, vol. 28: 1915: ‘Stadgar för Sjukkassan Eriksbergs bryggeris arbetareförbund’; ‘Stadgar för Hargs arbetarförenings sjuk- och begrafningshjälpkassa’.

14 RA, Socialförsäkringskommittén, Sjukkassestadgar mm. 320147, vol. 28: 1915:‘A. B. Borås klädningstygsfabriks arbetares sjuk- och begrafningshjälpskassa’. See also Lindeberg (1949, p. 152).

15 RA, Socialförsäkringskommittén, Sjukkassestadgar mm. 320147, vol. 28: 1915: ‘Förvaltningsberättelse över Jonsereds sjuk- och begravningskassas verksamhet 1917’; ‘Stadgar för Hargs arbetarförenings sjuk- och begrafningshjälpkassa’; ‘Stadgar för Sandviks sjukkassa’.

16 RA, Socialförsäkringskommittén, Sjukkassestadgar mm. 320147, vol. 28: 1915: ‘Arbetspersonalens sjuk- och begravningskassa vid AB Atlas 1887’; ‘Gäddvikens arbetares sjuk och begravningskassa’.

17 RA, Socialförsäkringskommittén, Sjukkassestadgar mm. 320147, vol. 28: 1915: ‘Stadgar för arbetarnes sjukkassa vid Fr. Kurzels Vigognespinneri i Malmö’.

18 Riksarkivet (RA), Riksförsäkringsanstalten, Registrerade sjukkassor, Första och andra sjukförsäkringsbyrån: ‘Statistiska redogörelser 1905’.

19 When running the model the share of old members as dependent, we identified a significant impact of growth of members (–0.15*), age of society (0.61***), and compulsory membership (–8.21***). The regression analysis based on average age had an Adj R-squared of 0.36, and the model on old members an adj. R-squared on 0.26). ***, **,* denotes significant at the 1%, 5% and 10% level respectively.

20 Riksarkivet (RA), Riksförsäkringsanstalten, Första och andra sjukförsäkringsbyrån, Statistiska redogörelser: ‘Allmänna sjuk-och begravningskassan 1916’.

21 For IV regressions one could consider the correlation between the endogenous regressor and the instrument. In our analysis, the correlation is around 0.5 and highly significant. To test for a weak instrument, one commonly used diagnostic is the F-stat for the significance of an instrument (in the first stage regression of endogenous regressor). A widely used rule of thumb is that an F-stat of less than 10 is a weak instrument. The F-stat we arrive at is above 10 across all of the specifications. A more formal test is proposed by Stock and Yogo (2005).They provide critical values for single endogenous repressors. Based on their critical values from the first regression output, we can reject the null hypothesis of a weak instrument based on our F-stat reported in the first stage regression.

22 RA, Riksförsäkringsanstalten, Registrerade sjukkassor, Första och andra sjukförsäkringsbyrån: ‘Statistiska redogörelser 1905’: ‘Af arbetsgivaren har direkt, utan kassans förmedling, bestridts läkarevård och medikamenter för delägarare af kassan till belopp af: … ’

23 Arbetareförsäkringskomiténs betänkande, 3, Statistiska undersökningar, 6, Sjuk- och begrafningskassor; Lindeberg (1949).

Additional information

Funding

This work was supported by the Forskningsrådet om Hälsa, Arbetsliv och Välfärd [grant number 2012-0812] and Vetenskapsrådet.