Paper Presented at the Economic History Society Annual Conference, Wolverhampton, Telford Campus
Paper presented at the Economic History Society Annual Conference, Wolverhampton, Telford Campus. 27–29 March 2015
Wealth inequality in Sweden 1750–1900[1]
Erik Bengtsson, Anna Missiaia, Mats Olsson and Patrick Svensson
Department of Economic History, Lund University
Preliminary results, please do not quote!
1. Introduction
The last fifteen years or so have seen an important upswing in the research interest in early modern wealth inequality. Previous economic inequality research has mainly built on data from the mid-19th century on and hypothesized patterns in the data (a 19th century increase in inequality) that could not be proven or disproven with the data used. As early as the mid-1990s (van Zanden 1995), the interest in earlier economic inequality, and especially wealth inequality, has grown. Wealth inequality is, in an agrarian economy, a better measure of inequality than the inequality of incomes. This is because a large part of the economic activity is not mediated by the market and the income of most economic activity cannot be measured. Instead, the ownership of wealth, and especially land, is the most important indicator of economic inequality in the preindustrial society (cf. Alfani 2010: 514).
This paper provides new estimates of wealth distribution in Sweden from 1750 to 1900. The previous literature on Swedish wealth inequality before 1900 comprises only of a problematic point estimate of 1800 (Soltow 1985) and some years in the 1870s and 1890s (Roine and Waldenström 2009); our data thus makes it possible to test many hypotheses and ideas about Swedish wealth inequality in the early modern and early industrial periods. One of these is Piketty’s (2014) claim that Sweden before the 1910s was not at all more equal than other capitalist countries. This claim, given that Sweden was a late industrializer, contrasts with the common statement that more economically advanced countries were more unequal (e.g. van Zanden 1995). Moreover, this study aims at extending Lindert’s (1986) analysis of rural vs. urban inequality and inequality by social groups to Sweden. Finally, we are testing the widespread view in Swedish economic history of a relatively equal peasant-dominated rural society.
2. Literature and hypotheses
The international literature on early modern wealth inequality
Historical wealth inequality has raised great interest in recent years. For the United States, Lindert and Williamson (1980) provide a synthesis of the studies made up to the late 1970s. Their main result is that wealth inequality went through four phases from the mid-17th century to the 1950. First, a period without much change from the mid-17th century to 1770. The second phase is from the 1770s to the 1860s and sees a rapid increase of inequality.[2] Phase three is from 1870 to 1910 with high and stable inequality. The fourth phase is one of decreasing inequality from the First World War to 1950. Lindert and Williamson claim that this is completely consistent with the Kuznets curve: first increasing inequality with industrialization and economic modernization, then decreasing inequality.
Lindert (1986) estimates wealth inequality for England and Wales for 1670, 1700, 1740, 1810, 1857 and 1875. He finds that inequality in ownership was high in the Victorian age, as social commentators pointed out in that period, but had been so also for the two hundred preceding years, although its composition had changed substantially. At the beginning of the period, land was the most important form of wealth, and was very unequally held. When land gradually became less important, this had an equalizing effect. But this effect was counteracted by a growing inequality in the ownership of non-land capital. The compositional change at the centre of Lindert’s analysis highlights how studies of wealth inequality should not reduce its topic to be only top shares or Ginis, but also investigate the composition of inequality and how it changes over time. Lindert also shows an interest in social groups – nobility, merchants, yeomen, peasants and so on – that inspires. However he also presents estimates of top wealth shares. The share of the top decile is, according to Lindert, rather trendless from 1670 to 1875, varying between 81 and 86 per cent. The top percentile share is more volatile, first decreasing from 49 per cent in 1670 to 39 in 1700 and 44 in 1740, then increasing dramatically to 55 per cent in 1810 and 61 per cent in 1875 (Lindert 1986: table 4). However, this increase seems to have been the result of polarization among the rich, as the top 5 per cent share actually did not bulge.
Van Zanden (1995) in a very influential paper gave new impetus to the literature and brought it, so to say, to the European continent. Van Zanden compiled the scattered findings from a large number of local studies of wealth inequality in late mediaeval and early modern Western Europe and claimed that the literature needed a uniting analytical framework. The framework he proposed was the Kuznets Curve, now extended from its earlier 19th–20th century shape to a “Super Kuznets Curve” for the early modern period, where the growth period of the 16th century starts increasing inequality which goes on until the 19th century, and the decrease starting in the early 20th century. For example Augsburg, for which there is unusually good data, in the 16th century shows the expected combination of economic growth and increasing inequality that one would expect in the early phase of a Kuznets Curve. Van Zanden also provides new evidence that the Super Kuznets Curve fits the case of Holland.
The later European literature has mainly continued to build on van Zanden’s 1995 paper. Alfani (2010), studying Ivrea, a city in northern Italy, also finds very high preindustrial urban inequality. Reis et al (2011) have data for Portugal in 1550, 1700 and 1770 and do not find increasing inequality from 1550 to 1770 but explain this with Portugal’s relative economic stagnation – i.e., the country is the opposite of the typical Kuznets Curve country. Ergene et al (2013) however find the opposite for 18th century Ottoman Empire: economic stagnation and declining inequality. Fernandez and Santiago-Caballero (2013) investigate the countryside around Madrid from 1500 to 1840 and find increasing inequality. Cankabal and Filiztekin (2013) for four cities in Anatolia from 1500 to 1840 find the expected positive correlation between economic development and inequality. Turner (2010) studies wealth inequality in Northern Ireland from 1858 to 2001 and find no change in inequality before the 1890s when inequality starts decreasing. The decrease after the 1890s is not surprising but the typical right hand side of the Kuznets Curve (albeit starting earlier than most), but it is more surprising that inequality did not increase during the industrialization years from 1858 to the 1890s. Turner says that one possible explanation is that Northern Ireland is “a more agrarian and landlord-dominated economy than England and Wales, and was less industrialized at this stage in its economic development” (p. 642).
Thus, the current literature on early modern wealth inequality in Europe is very much focused, almost single-mindedly so, on the (Super) Kuznets Curve and the relationship between economic development and inequality. This is certainly one connection that we will explore in our analysis, but as we will see from the Swedish literature, there are also more nuanced and fine-grained hypotheses to test.
The Swedish literature
There are two earlier direct studies of Swedish wealth inequality before 1900. Soltow (1985) uses the 1805 census where parish priests were obliged to report the socioeconomic status of their parishioners. Those who had more than 500 riksdaler above their yearly expenses were to be classified as rich, those who had less as rather rich, those who did not get by without hardship were classified as poor, and those who depended on gifts and alms to survive were classified as destitute. Soltow uses the social distribution into the four groups and combines this with wealth data from the 1800 wealth tax, which lists each owner of property and wealth that year. Soltow has a sample of 6309 persons and estimates a gini coefficient in 1800 of 0.73.[3]
Roine and Waldenström (2009) use estate tax data for 1873 to 1877 and 1906 to 1908 to extend the previous wealth distribution series which Spånt (1979) created for the period 1920 to the 1970s.[4] According to Roine and Waldenström’s estimates, the top decile held about 85 to 90 percent of total wealth in the 1870s and this did not change to 1906–08. (In this series there is only real change, with a steep decrease of inequality, after 1930.) The top percentile holds between 53 and 60 percent of total wealth in 1873–77 and 58 to 61 percent in 1906–08; again the picture is one of high and stable inequality. These two studies provide the data also for the reporting on Swedish wealth inequality in Piketty’s (2014: 344f) recent book. Piketty’s conclusion from these data in a comparative perspective is that “Sweden was not the structurally egalitarian country that we sometimes imagine” (p. 344); wealth is not equalized until after World War I.
Of course there is also a wider literature which is relevant to the topic of Swedish wealth inequality, even though it does not present direct estimates of wealth inequality. Soltow (1989) uses the same census classification in four social groups – rich, moderately rich, poor and destitute – as in his 1985 wealth inequality paper from the censuses from 1805 to 1855 to analyze the proportion of households in the different groups. He finds that both among rural and urban households, the share of the poor or destitute decreased and the share of rich or moderately riche increased. He interprets this as decreased inequality (p. 51) He estimates the Gini coefficient of land ownership to be 0.70 in 1805, 0.67 in 1845, 0.64 in 1879 and 0.58 in 1921, and the top decile share in same years as 60, 58, 54 and 48 per cent respectively, i.e. a continuous decrease of inequality. According to Soltow, “general inequality decreased at least a little; inequality above the median income decreased; destitution decreased. The outstanding aspect of the initial stages of the industrial revolution in Sweden is the fact that poverty did not increase. In fact, it decreased.” (p. 62) However, as we have noted his data only concerns the ownership of land, and of course during industrialization land decreased in importance as a source of wealth, as produced capital became more important. As we will see, we will have reason to revise Soltow’s picture of decreased inequality.
There are several other studies of specifically rural social differentiation during our period. Winberg (1975: 17) in a seminal study stresses that from 1750 to 1850 population growth is rapid in Sweden, and especially rapid is the growth of proletarians. The share of proletarian and semi-proletarian groups among rural households increases from one fifth in 1751 to half in 1850. Proletarianization is not necessarily the same thing as pauperization, but it is still reasonable to expect some increased wealth inequality to this social differentiation. Morell (1980) compiles the findings of several local studies of six parishes in east central Sweden from the 1770s to the mid-19th century and reports the main conclusion as one very compatible with Winberg’s: increased social differentiation in the peasant class. Söderberg (1978) counts the share who were poor – defined as those who by the standards of their own time did not have more than “vad yttersta nödtorften krävde” (to cover the absolutely lowest necessities) – in southern Sweden in 1821, 1851 and 1871. He finds that the share who was poor typically was around one fifth of the population. This share was rather stable but decreased in some southern regions from 1821 to 1871 (p. 23).[5] Söderberg (1978: 182) summarizes Lundsjö’s (1975) study of poverty from 1825 to 1860 as no country-wide trend but increases in poverty in eastern middle Sweden and a decrease in west Sweden (but from a higher level). Isacsson (1979) studies one parish in a protoindustrial area of central Sweden (Dalarna) from 1680 to 1860 and finds increasing social differentiation in the early 19th century; “the process toward a more homogenous peasant class during the 18th century was turned into its opposite at the beginning of the 19th century”. This is similar to Martinius’ (1977, 1982) argument that major peasant farmers started to distinguish themselves from ordinary peasants around 1850, although for Isacsson this process starts already around 1800 while Martinius stresses 1830 as a break point. Olausson (2004) studies social differentiation in western Wermland from the 17th century to the mid-19th century. He finds obvious differentiation already in the 17th century, increasing in the 18th century with growth of proletarian and semi-proletarian groups; he notes that his studied area was particularly dominated by nobles (p. 151). However in the second half of the 18th century he does find some specific expansion of peasant farmers which would indicate a more equal distribution; this is in the 1800 to 1825 period turned to its opposite, with instead a growth of on the one hand proletarian groups, on the other hand of the large estates.
On the contrary, we have very few studies of urban social differentiation. Lindberg (2007) studies income distribution in Stockholm from 1730 to 1810, but only has data for burghers, which of course represented only a small part of the population. Among the burghers he finds increasing income inequality and makes the more general claim that the 18th century was marked by a mercantilist “policy of enrichment of the rich and impoverishment of the poor”. Söderberg (1987) analyzes the distribution of property in Stockholm in 1715, 1799 and 1845. He finds an ownership Gini coefficient of 0.78 in 1715, 0.7 in 1799 and 0.68 in 1845 (p. 68). The share of the nobility or persons of standing falls, and the share of merchants and craftsmen increases. Bengtsson (2015) studies salary–wage differentials in Stockholm from 1833 to 1905. He finds that the difference between a white-collar worker or a professional on the one hand and a blue-collar worker on the other hand increases rapidly in the mid-19th century, to historically high levels in the 1880s and 1890s. In another study he also finds that the wages of urban labourers as well as manufacturing and agricultural workers increase less than GDP per capita does ca. 1830 to 1875, implying a bottom-led increase in inequality, where proletarians’ living standards do not increase at a pace with overall living standards (Bengtsson 2014). This would imply growing wealth inequality, given that, as Williamson (1986) phrases it, wealth distribution is the tail wagged by the income distribution dog. On the other hand, wages are found to increase slightly faster than GDP/capita in the 1880s and 1890s.