Intergenerational income mobility in Gothenburg, Sweden, 1925–1958, before the rise of the welfare state

Abstract

This paper investigates intergenerational mobility at the household level by using tax data for the city of Gothenburg, Sweden for two periods, 1925–1947 and 1936–1958, before the rise of the welfare state. Young households (selected persons under 18) are followed and income mobility (defined as changes in household disposable equivalent income) is followed across generations. In addition, socioeconomic mobility (defined as changes in the socioeconomic status of household head) is followed across generations. These two approaches of measuring mobility will tell us to what extent sons and daughters follow in their fathers' footsteps before the rise of the welfare state in Sweden. The results indicate significant intergenerational income mobility for both periods, while the period from 1925–1947 seems to be more mobile. In addition, socioeconomic mobility increased during the last period, 1936–1958. Hence, even before the rise of the welfare state, Sweden had high intergenerational income mobility.

Keywords::

• income mobility
• intergenerational
• household disposable income
• panel data
• Sweden

Acknowledgement

I wish to thank Björn Gustafsson Torun Österberg, two anonymous referees, and the participants in a seminar at the Department of Economic History, University of Gothenburg, for comments on a previous version.

Notes

^1. The final year of the study is not presented in the article but 2004 is the last year for collecting data on income from capital.

^2. Household disposable income is for one year, with information coming from tax files.

^3. The female labour force participation rate in Sweden today is approximately 76%.

^4. Author's own calculations from SCB (Swedish Statistics) http://www.scb.se/Pages/ProductTables____25795.aspx Sveriges folkmängd (i ett års klasser) 1860–2009. (Population in Sweden, in one age cohorts, 1890–2009.)

^5. In the ecclesiastic law from 1686, it was established that the sacristan should teach all children in the parish to read. The first education act came in 1842.

^6. Elementary school was extended from six to nine years followed by a reformed upper secondary school.

^7. For a more thorough description of the database, see Gustafsson and Johansson (2003) who also report Gini coefficients computed from the data.

^8. In 1925, the number of households in the original subsample was 1641; by 1936, there were 2197 households; in 1947, 2614 households; and finally by 1958, 3363 households. The increase in the number of households follows the growth of the population of Gothenburg.

^9. We do not know for sure whether the father is the head of the household during the first year of observation, but most probably this is the case. For a discussion of the definition of a father and the consequences for intergenerational mobility of living together with the biological father or not, see Björklund and Chadwick (2003).

^10. In addition, Long and Ferrie (2007, 2013) have a 22-year interval.

^11. See Österberg (2000).

^12. Economic growth was substantial in Sweden during the time of the study (see Dahmén, 1950; Lundberg, 1953).

^13. Daughter's value is significant at 0.15.

^14. The confidence interval for sons (0.0253–0.3485) and for daughters (0.1717–0.4740) is to some extent overlapping.

^15. As the number of observations in each panel is too few to make a division in gender reliable, the panels are merged after the quartile boundaries for each year have been established.

^16. The data originally had three different classifications. The richest classification consisted of between 959 and 1716 different occupational groups and the less detailed one consisted of 13 socioeconomic groups. Here the less detailed classification is used and merged into the categories unskilledworker, skilledworker, white collar, higher employees, self-employed and information missing.

^17. It is of course possible that all young men and women have not yet found partners and formed new households. Socioeconomic status could be from the partner, father or adult brother/relative with whom the selected person formed a household.

^18. The groups are presented in note 16, above.

^19. The immobility ratio is calculated from the total percentage of the matrix. Note that the matrices are not symmetrical (as they are in Section 5) as the number of unskilled households decreases and that of white-collar households increases over time.

^20. The occupational groups are regrouped into four, retaining unskilledworker, skilledworker and information missing and creating a new group named white collar, which is made up of the former groups white collar, higher employees and self-employed. These three former groups consisted of a small number of individuals. By merging these into a new larger group, the risk is reduced that one individual's change of socioeconomic position results in a large ratio of mobility.

^21. Socioeconomic immobility is here calculated as a line percentage.

Additional information

Funding

This work was supported by the Swedish Research Council for Health, Working Life and Welfare (FORTE) [grant number 2005-0180]