Long-run evolution of income inequality in the Nordic countries

Abstract

ABSTRACTThis paper surveys Nordic historic studies on the distribution of income to highlight similarities and differences between Denmark, Finland, Norway, and Sweden in the evolution of income concentration and income inequality over more than 140 years. Our descriptive analysis allows for a decomposition where we identify the contribution of the income share of the richest 1 per cent and the distribution of income among the other 99 per cent to overall inequality as measured by the Gini coefficient. The results show that the evolution of income concentration and inequality can be characterised by episodes rather than by secular cycles, which means that the evolution can neither be summarised by Kuznets’ inverse U nor by a U. The evidence on the role played by the share of the top 1 per cent for overall income inequality shows to be mixed and to vary across time and countries.KEYWORDS: InequalityincomesScandinaviaJEL: CODES: D31N33N34 AcknowledgementsWe would like to thank Jesper Roine, Petri Roikonen, Jakob Søgaard and Daniel Waldenström for sharing the data that has been used to produce the figures presented in this paper. We would also like to thank the editor Paul Sharp and three anonymous referees at SEHR for useful comments and suggestions.Disclosure statementNo potential conflict of interest was reported by the author(s).Notes1 Ólafsson and Kristjánsson (Citation2017) provide estimates of the share of the top 1 per cent for two years between the world wars (1927 and 1936) and for the period 1992–2017, while estimates of overall inequality measured by the Gini coefficient are only available for the recent three decades.2 Flodström was a crucial actor in the creation of modern economic statistics in Sweden. See Hellroth (Citation2011) for a study which discusses the role of Flodström.3 For the city of Stockholm 1870–1970 Bengtsson and Molinder (Citation2022) have been able to calculate both individual and household level inequality measures. However, they show to yield remarkably similar results.4 The top income approach of course has its clear proponents, most notably Piketty. See Piketty’s (Citation2014, pp. 246–268) discussion of top income shares versus overall measures of inequality like the Gini coefficient.5 Note that the Gini coefficient can be interpreted as the ratio between the average pairwise income differences in the population and twice the mean income, which means that the Gini coefficient becomes equal to 0 if and only if all population units have equal income. The other extreme is attained if and only if one unit receives the total income. In this case the Gini coefficient takes the value 1. When the Gini coefficient is equal to 0.5 then the average income difference is equal to the mean income.6 Nolan et al. (Citation2019, p. 1290) concur, arguing that «increases in inequality have often occurred in discrete ‘episodes’ rather than in a consistent fashion over time».7 To further explore the role of the financial sector and its crises in the income distribution, we would need data which allows for a decomposition of income by sector. Bengtsson and Molinder (Citation2021) use individual tax returns data to study top incomes in Stockholm in 1909, 1915, 1927, 1935 and 1950. The financial sector is shown to account for a considerable share of top incomes. Unfortunately, income by sector is not available before the financial crisis of 1907.8 Note that the significant fall in the income share of the top 1 per cent from 2005 to 2006 in Norway was solely due to the increase of the tax rate on dividends from 0 to 28 per cent in 2006, which gave owners of unlisted companies strong incentives to reduce the dividend pay-out and save most of the profit in holding companies. Thus, since retained business income is not included in personal tax records the conventional income statistics will underestimate top income shares from 2006 onwards. By accounting for retained business income Aaberge, Modalsli and Vestad (Citation2020) demonstrated that the share of the market income attributable to the top 1 per cent increased from 8–9 to 24 per cent in 2006–2007 and more than doubled after the financial crises in 2008. The effects of the reform have been discussed in further detail by Aaberge et al. (Citation2021) and Alstadsæter et al. (Citation2016).9 To reach the lowest levels, which were attained during the 1980s in Norway and early 1990s in Denmark, would have required a reduction of approximately 35 per cent of the Gini coefficient. It can be shown that this reduction in the Gini coefficient will correspond to introducing a 35 per cent proportional tax on income and then redistributing the derived tax revenue as equal sized amounts to the individuals (Aaberge, Citation1997). As an illustration assume that the mean income was 10,000 Euros. Then a person with 2000 Euros in income would have gained 2800 Euros, whereas a person with 50,000 Euros would have lost 14,000 Euros. After this tax-transfer intervention the income of the poorest would have increased from 2000 to 4800 Euros, while the income of the richest would have decreased from 50,000 to 36,000 Euros.10 For Sweden, estimates of overall measures of inequality are not available further back then in 1975.Additional informationFundingRolf Aaberge received support from the Research Council of Norway [grant numbers 261985; 341250]. Erik Bengtsson received support from the Swedish Research Council project grant 2018-01853, and Riksbankens Jubileumsfond grant M19-0231:1.