Abstract
According to standard measures of income inequality, the Nordic countries rank among the most equal economies in the world. This paper studies whether and how this picture changes when the focus is on inequality of income composition, meaning the heterogeneity in individuals’ factor income shares. We show that, for all countries, a shift in capital incomes toward the top since the early 1990s causes rising heterogeneity in individuals’ factor income shares. To explain this result, we highlight the role of dual taxation systems. For Denmark in 2009–2013, Finland (1990–2007), and Norway (1991–2005), rising capital shares contributed to changes in personal income inequality, while for Sweden our results lead to disregard the capital share as a determinant of increasing income inequality.
1 Introduction
The link between the functional and personal distributions of income has again become a field of interest in recent years. The debate was initiated by Piketty and Zucman (2014) and Piketty (2014), predicting that higher capital shares of income in advanced economies (and, correspondingly, falling labor shares, Karabarbounis and Neiman (2014)) would inevitably lead to higher personal income inequality in the decades to come. The study by Bengtsson and Waldenström (2018) broadens the debate by empirically analyzing the relationship between factor shares and income inequality (proxied by top income shares) in the longer run. One of their take-away messages is that the link between the functional and personal distribution of income varies substantially over time and across countries, being contingent on different institutional contexts and production technologies.
In this research, we intend to contribute to the debate on the link between the functional and personal income distributions, by providing evidence on this relationship for the Nordic countries in recent decades. Ranaldi and Milanovic (2021) find that the evidence from the Nordic countries constitutes a puzzle: While the Nordics are still the lands of equality with respect to the more conventional Gini coefficient, they exhibit high inequality in income composition. This implies (a) that ownership of factor income is highly polarized across the total income distribution, with the bottom receiving mostly labor income and the top mostly capital income; and (b) that rising capital share might lead to sustained increases in personal income inequality in the years to come. By focusing on a longer time span and country-specific details, we show that income composition inequality has not always been high in the Nordics, and that current high levels of heterogeneity in factor income shares in Ranaldi and Milanovic (2021) are associated to recent taxation reforms.
Methodologically, we adopt a recent inequality dimension introduced by Ranaldi (2021), labeled income composition inequality. How is this new dimension useful? The prediction of rising inequality in response to rising capital shares made by Piketty (2014) and Piketty and Zucman (2014) rests on a set of necessary conditions regarding the transmission mechanism from the functional to the personal income distribution. First, capital incomes ought to be concentrated in fewer hands than in the case of labor income. Second, recipients of capital income ought to be, to a large extent, also the richest in the income distribution. These two necessary (and sufficient) conditions are analytically pinned down in Milanovic (2017). The methodological innovation in Ranaldi (2021) is to reduce these two conditions to the estimation of a single, aggregate, summary statistic. The degree of inequality in income composition, i.e., the heterogeneity in individuals’ income factor shares, is measured by the Income Composition Inequality (ICI, hereafter) index.
Ranaldi (2021) shows that a positive and high level of the ICI index would, alone, suffice to support the prediction of increasing inequality due to rising capital shares made by Piketty (2014) and Piketty and Zucman (2014).[1] Rather than focusing on a large set of countries as in Bengtsson and Waldenström (2018), we study the countries that rank as the most egalitarian with respect to the pre-tax market income distribution, namely, the Nordic countries.[2] Does a low level of pre-tax market income inequality prevent a rising capital share of income from increasing the level of inequality?
Our starting point is the empirical evidence identifying the increase in personal income inequality in the Nordic countries since the early 1990s, as documented in Aaberge et al. (2018a). In other words, income inequality will act as the dependent variable in our conceptual framework. The countries under analysis are Denmark, Finland, Norway and Sweden. The level of personal income inequality (Gini coefficient) in the Nordic countries is plotted in Figure 1.
Personal income inequality in the Nordics.
This figure shows yearly time series for the Gini coefficient (1980–2013, pre-tax national income, total population, unit = individuals, age = adults, equal-split series) for Denmark (blue), Finland (red), Norway (yellow), and Sweden (purple).
Source: World Inequality Database WID.World (2020).
Let us focus on the countries and periods in which income inequality increased most significantly. We observe that the Gini coefficient for Norway increased from 0.251 to 0.384 in the period 1989–2005. For Sweden, the year 1990 marks the lowest Gini, 0.27, which then increases up to 0.31 in 2005. In Finland, the lowest Gini is in 1993 at 0.31 points, climbing all the way to 0.368 in 2011. In Denmark, the Gini is stable at approximately 0.32 up to 2009, followed by a jump to a level of 0.37 in 2013. Overall, regardless of country differences, in the post-1990 era, the Nordic economies have been subject to prolonged periods of increases in the level of personal income inequality.
The main determinant of personal income inequality under scrutiny in this study is the net capital share of income in the Nordic countries for these years, similar to the model specifications proposed by Bengtsson and Waldenström (2018).[3] Figure 2 shows that the Nordic economies differ more from one another in recent decades with respect to capital share dynamics than with respect to personal income inequality. In brief, while the capital share of income is shown to increase for Norway (yellow) and Finland (red) from approximately the beginning of the 1990s to 2005, the series for Denmark (blue) and Sweden (purple) appear to be rather stable throughout the period. How much of the variation in the series of Gini coefficients (Figure 1) can be explained by the evidence in Figure 2?
Capital share of income in the Nordics.
This figure shows the capital share of income (1975–2013, defined as value added net of capital depreciation minus employee compensation over national income) for Denmark (blue), Finland (red), Norway (yellow), and Sweden (purple). Data were retrieved from the Bengtsson–Waldenström (BW) Historical Capital Shares Database in Bengtsson and Waldenström (2018), Version 2.0, March 2017.
The main contribution of this work is to provide evidence of the structural change in income composition taking place in Nordic countries since the early 1990s. For all the countries under analysis, we identify rising inequality in the composition of individual incomes, mostly due to a shift in capital incomes toward the top of the income distribution. A higher share of the wealthy having their incomes primarily composed of capital income also implies a higher fraction of the poor mostly relying on labor income and can be interpreted as a gradual move toward classical capitalism [4] in the Nordic countries in recent decades.
Furthermore, we analyze the extent to which the evidence of increasing inequality in income composition hinges on changes in the taxation of factor income. We highlight that the introduction of dual income taxation reforms in the early 1990s in the Nordics, which reduced the degree of progressivity in capital income taxation, lies behind the shift in capital incomes toward the top of the distribution and, hence, the strengthening of the responsiveness of the Gini coefficient to changes in the aggregate capital share.
Finally, we show that, for Sweden, changes in the capital share of income did not have a key impact on the level of personal income inequality. For Denmark, Finland, and Norway, the evidence is more mixed, indicating that rising capital shares might have contributed to increasing inequality. In Denmark, this applies to the period from 2009 to 2013, during which income composition inequality is high, indicating that the jump in the capital share of income in these years contributed to the rapid increase in income inequality. For Finland and Norway, this especially applies to the period from the early 1990s up to the outbreak of the financial crisis. A summary of these results will be provided in Table 9.
The structure of the paper is as follows. Section 2 presents the data and the methodology. Section 3 constitutes the core of the paper, reporting the main results for each country taken individually. Section 4 discusses the role of taxation and includes robustness checks, while Section 5 concludes the paper.
2 Data and Methodology
2.1 Data Sources
The micro data on income employed in this research are retrieved from the Luxembourg Income Study Database, LIS (2020). Cross-nationally harmonized LIS data allow a meaningful cross-country comparison of the estimates of income composition inequality. The LIS (2020) database divides current income (pre-tax market income) into factor and transfer income. In turn, factor income comprises labor income and capital income. Labor income is composed of wage income and self-employment income (including farm income). Capital income includes interest and dividends and rental income. Transfer income includes public and private pensions, public social benefits and private transfers. Capital gains received are available in the LIS (2020) data as extraordinary income sources. For all four countries under analysis, the source of the data included in the LIS (2020) database is register data from the national statistical institutes, with relatively high population coverage.
The series of Gini coefficients for the Nordic countries are, to ensure comparability, all retrieved from the World Inequality Database, WID.World (2020). The Gini coefficients are computed from pre-tax national income, and the unit of analysis is individual adults (equal-split series). These data are employed in Figure 1.
The series for the capital share of income (value added net of capital depreciation minus employee compensation) are obtained from the Bengtsson–Waldenström Historical Capital Shares Database (Bengtsson and Waldenström 2018). The capital share is computed as the sum of capital incomes (interest, profits, dividends, realized capital gains) divided by value added calculated at factor cost, net of capital depreciation. Regarding the income of the self-employed, Bengtsson and Waldenström (2018) utilize the proportional method, treating 65–70% of the income of the self-employed as labor income, with the residual being treated as capital income. These data are employed in Figure 2.[5]
2.2 Baseline Income Definition
In the core of the paper, we adopt a single baseline definition of income, based on the LIS (2020) factor income, hence excluding transfer income. The use of market factor income, rather than net disposable income, allows us to measure the pre-tax concentration of factor income across the income distribution. We define capital income (Π) as the sum of property income (Πpr), comprising rental income, interest, dividends and capital gains[6] and the capital component of net self-employment income (Πse). Formally, we write:
Labor income (W) includes wage income (W wa) and the labor component of net self-employment income (W se). Formally, we write:
The capital and labor components of net self-employment income are imputed following Glyn (2011).[7] The unit of analysis is the individual: to this end, we use household-level data adjusted by the LIS (2020) database sample weight (hpopwgt), and multiplied by the number of household members, given by the variable nhhmem in the LIS (2020) data variable list.
2.3 Income Composition Inequality
Ranaldi (2021) introduces and defines the properties of a novel inequality dimension: income composition inequality. Income composition inequality focuses on the heterogeneity in individuals’ factor income shares across the income distribution. The concept of inequality of income composition can be explained by first dividing total income Y into two income sources, e.g., total income given by capital π and labor w (although this applies to any other pair of sources), such that (w + π)/Y = Y/Y = 1. Individual i’s income share is given by y i = Y i /Y, and it can be written in terms of the share of the two factor incomes she receives:
where α i and β i are the relative shares of capital and labor for individual i. Individuals are indexed by their (total) income ranking. Ranaldi (2021) proposes a statistic to measure the degree of income composition inequality. The income composition inequality (ICI, hereafter) index is constructed by means of the concentration curves for each income source.[8]
The two other building blocks of the index are the zero and maximum concentration curves, reflecting the situation of minimal income composition inequality (implying that each individual owns the same share of both income sources along the income distribution) and maximal income composition inequality (with income sources separately owned at the top and at the bottom of the income distribution). Ranaldi (2021) then constructs the ICI index by taking the area given by the difference between the concentration curve for a given income source scaled down to the factor share and the zero concentration curve.[9] In more formal terms, define the area between the zero concentration curve and the concentration curve for source z (with z = π, w) by
The ICI index can assume any level in the range [−1, 1]. When income composition inequality is minimal, the area between the concentration curve for source z and the zero concentration curve tends to zero, implying ICI z = 0. In contrast, a state of the world with maximal income composition inequality can imply either ICI z = −1 or ICI z = 1, depending on which end of the income distribution owns income source z. A useful formulation of the index presented in Ranaldi (2021), highlighting the role of the areas below the concentration curves, is the following:
where π and w are the capital and labor shares of income, respectively, and
2.3.1 The Transmission Mechanism
Let us explain why the transmission mechanism between the functional and personal income distributions is better understood through the degree of income composition inequality. Whenever the degree of income composition inequality is high, the ICI index will display a level close to −1 or 1. In the case of an ICI index value close to 1, we have a state of the world in which the wealthiest individuals own relatively more capital income as part of their income than individuals at the bottom of the distribution. This means that a sudden increase in the capital share of income would imply a substantial increase in the income owned by the top of the income distribution, thereby resulting in higher personal income inequality. The opposite happens when the ICI index is close to −1, indicating that the poorest part of the population has higher shares of their income composed of capital income with respect to the wealthiest. In the latter case, a sudden increase in the capital share of income would imply a substantial reduction in the income owned by the top of the income distribution, thereby resulting in a lower level of personal income inequality.
In other words, a high degree of income composition inequality is associated with a strong transmission mechanism between the functional and the personal income distribution. In contrast, the transmission mechanism from the functional to the personal income distribution is weak under a low degree of income composition inequality. The above reasoning can be formalized following Ranaldi (2021). Deriving the Gini coefficient
Equation (6) identifies the connection between the ICI index and the derivative of the Gini index with respect to changes in the capital income share. We will refer to this derivative as the degree of responsiveness of the Gini index to increases in the aggregate capital income share.[11] Recall that the right-hand side of Eq. (5) finds the difference
3 Main Results
This section presents the main results of the paper: each of the four countries is analyzed within a standalone subsection. A summary of the results from this section is provided in Table 9.
3.1 Denmark
Denmark ranks among the countries with the lowest income inequality in the world (Alvaredo et al. 2018; Chancel 2021). However, recent empirical evidence has shown a tendency of widening disparities. Atkinson and Søgaard (2016) study the long-run evolution of top taxable income shares in Denmark, and find a tendency of rising inequality at the very top of the income distribution in recent years, up to 2010. Søgaard (2018) seeks to explain these changes in top income shares, identifying (i) the role of taxation in conjunction with (ii) the role of capital income. On the one hand, reductions in top marginal tax rates proposed in the Danish Tax Reform Act of 1993 (Lange, Pedersen, and Sørensen 1999) and implemented from 1994 onward lowered the progressivity of taxation of capital incomes (Rubolino and Waldenström 2020). Lower progressivity of capital income taxation might have led to higher concentration of capital incomes at the top of the distribution and hence higher income inequality. However, neither Søgaard (2018) nor Rubolino and Waldenström (2020) identify a causal effect from policy changes in taxation in the case of Denmark. On the other hand, Søgaard (2018) notes that the role of different types of capital income (a composition effect) has been decisive for the dynamics of the top 1%. He documents a shift away from interest income (due to reduced interest rates) and toward higher dividends (due to lower capital income taxation). Although Søgaard (2018) offers mostly correlation evidence, the plausible hypothesis is that this increased fraction of dividends in the net capital income received by households in Denmark is responsible for the increase in top income shares documented in the literature.
To shed light on the heterogeneity of income composition throughout the income distribution, let us start to analyze the concentration of factor incomes in Denmark over the period 1987–2016. Table 1 shows the concentration of capital income across the total income distribution.[13] The income distribution is divided into four main groups, the bottom 50%, the middle class 50–90%, and the top 10%, which is further divided into the bottom half and the top half of the decile.[14]
Capital income shares by total income group, Denmark 1987–2016.
| Total income group | 1987 | 1992 | 1995 | 2000 | 2004 | 2007 | 2010 | 2013 | 2016 |
|---|---|---|---|---|---|---|---|---|---|
| 0–50% | 21% | 19% | 15% | 13% | 12% | 11% | 10% | 9% | 8% |
| 50–90% | 28% | 26% | 26% | 20% | 22% | 19% | 18% | 18% | 17% |
| 90–95% | 8% | 7% | 9% | 8% | 8% | 8% | 8% | 8% | 7% |
| 95–100% | 43% | 47% | 50% | 59% | 58% | 62% | 64% | 66% | 68% |
| Capital share (BW) | 17% | 20% | 22% | 21% | 19% | 18% | 17% | 19% | 20% |
| Gini pre-tax national | 0.33 | 0.32 | 0.32 | 0.34 | 0.32 | 0.33 | 0.35 | 0.37 | na. |
| income (WID) | |||||||||
| Gini total factor | 0.44 | 0.47 | 0.47 | 0.46 | 0.47 | 0.47 | 0.50 | 0.51 | 0.51 |
| income (LIS) |
From Table 1 and considering the full period 1987–2016, we can infer that the bottom 90% of the income distribution in Denmark lost a substantial fraction of capital income, dropping from a total share of 49% in 1987 to 25% in 2016. In parallel, the fraction of capital incomes accruing to the top 10% increased radically by 24%. In other words, a dramatic shift in the concentration of capital incomes at the top of the income distribution has taken place throughout the period. This change in factor income concentration and composition of individual incomes is not captured by the aggregate statistics on the functional and personal distribution of income. On aggregate, the capital share of income increased slightly from 17% in 1987 to 22% in 1995, followed by a mild decrease in the following years. The Gini coefficient from WID.World (2020) was fairly stable at approximately 0.33 up to 2007 but exhibits a jump in more recent years to a level of 0.37 in 2013.[15] See Appendix D for a discussion on entropy measures. Before commenting on these dynamics, let us focus in Table 2 on the other source of factor income, namely, labor incomes.
Labor income shares by total income group, Denmark 1987–2016.
| Total income group | 1987 | 1992 | 1995 | 2000 | 2004 | 2007 | 2010 | 2013 | 2016 |
|---|---|---|---|---|---|---|---|---|---|
| 0–50% | 18% | 15% | 16% | 17% | 16% | 17% | 15% | 14% | 14% |
| 50–90% | 58% | 60% | 59% | 58% | 58% | 58% | 58% | 58% | 58% |
| 90–95% | 11% | 11% | 11% | 11% | 11% | 11% | 11% | 11% | 11% |
| 95–100% | 13% | 14% | 14% | 14% | 15% | 14% | 16% | 16% | 17% |
| Capital share (BW) | 17% | 20% | 22% | 21% | 19% | 18% | 17% | 19% | 20% |
| Gini pre-tax national | 0.33 | 0.32 | 0.32 | 0.34 | 0.32 | 0.33 | 0.35 | 0.37 | na. |
| income (WID) | |||||||||
| Gini total factor | 0.44 | 0.47 | 0.47 | 0.46 | 0.47 | 0.47 | 0.50 | 0.51 | 0.51 |
| income (LIS) |
Table 2 tells a different story, with labor incomes being relatively more stable across the income distribution throughout the period of analysis. The only notable change is the decrease in the share of labor incomes accruing to the bottom 50% of the distribution, from 18% in 1987 to 14% in 2016. In parallel, the top 5% received a 4% higher share of labor incomes in 2016 than in 1987. Overall, Tables 1 and 2 indicate that the top of the income distribution has increased its share of capital incomes, with the incomes of the very wealthy becoming more capital intensive, while the opposite is the case at the bottom of the distribution. This development was mainly driven by the shift in the concentration of capital incomes toward the top. As explained in Eq. (5), the difference between the areas under the concentration curves for labor and capital
Evolution of the areas of factor concentration curves for Denmark.
First, focus on the blue series of the area of the concentration curve for capital,
Income composition inequality index in Denmark, 1987–2016.
This figure plots the series of the ICI index for Denmark using micro data on gross income from the Luxembourg Income Study Database (Luxembourg Income Study (LIS) Database, 2020) for the following years: 1987, 1992, 1995, 2000, 2004, 2007, 2010, 2013, and 2016. Capital shares going to the top 5% of the total income distribution are also reported.
The ICI index departs from very low values in the late 1980s and early 1990s (between 0.05 and 0.15). The low values of the ICI index in these years indicate a low degree of income composition inequality across the income distribution. Recall that in these years, income inequality in Denmark was historically low, and capital incomes had not yet started to accrue to top incomes as they will in subsequent years.
In the early 1990s, the Danish tax reform (Lange, Pedersen, and Sørensen 1999) lowered top marginal tax rates on capital incomes, and in response, capital incomes shifted rapidly toward the top of the income distribution. This development is clearly visible in the period 1992–2000 during which the ICI index increases substantially (in line with the evidence from Atkinson and Søgaard (2016), Aaberge, et al. (2018a), Table 1 and Figure 3). Interestingly, this jump in the concentration of capital incomes is not reflected in the evolution of the Gini index for total income, which only slightly increases between 1992 and 2000.
From 2000 up to 2013, the ICI index continues to grow (between 0.4 and 0.55), with the exception of period 2013–2016. To summarize the main result for Denmark, the evidence from the ICI index series is unambiguous in the sense that a substantial increase in the inequality in income composition took place in the country throughout the period, mostly due to changes observed in the concentration of capital incomes.
Finally, we attempt to shed light on the relation between the aggregate capital share of income and the personal distribution of income by jointly analyzing the evolution of the net capital share, our estimates for the ICI index and the Gini coefficient in Figure 5.
Capital share, ICI index and Gini: Denmark, 1987–2016.
The left-hand side graph plots the series of the capital share of income (value added net of capital depreciation minus employee compensation) in the period 1987–2013. This series is obtained from the Bengtsson–Waldenström Historical Capital Shares Database in Bengtsson and Waldenström (2018). The graph in the middle again plots the series of the ICI index for Denmark for the period 1987–2016, as shown in Figure 4. The graph on the right-hand side plots the Gini coefficient for Denmark from 1987 to 2013 using data from the World Inequality Database WID.World (2020).
We start with the period from 1987 to 1995, a range of years with an increasing capital share. The ICI index shows low values, indicating a weak transmission mechanism, and consistently, the Gini coefficient appears to be quite stable. The subsequent period, from 1995 to 2009, is characterized by a volatile capital share and an equally volatile (but rather stationary) Gini coefficient. The absolute values for the ICI index are now higher in magnitude (0.25–0.4), suggesting that the functional distribution becomes more relevant for the determination of the personal income distribution. The period between 2009 and 2013 shows an increasing capital share (from 13% to approximately 19%), positive and rather high values of the ICI index (between 0.4 and 0.55), and sustained increases for the Gini coefficient. Due to the relatively high values of the ICI index, which are associated with higher responsiveness of the income Gini coefficient to the capital share, we claim that the functional distribution qualifies as one of the drivers of the jump in personal income inequality in Denmark since 2009. In summary, the working hypothesis that the functional distribution of income has been the main driver of changes in income distribution is largely rejected, with the exception of the period from 2009 to 2013.
3.2 Finland
The literature on the evolution of income inequality in Finland (Eriksson and Jäntti 1997; Jäntti et al. 2010) shows that after a period of overall inequality reduction from the 1960s up to the 1990s, this decreasing trend has reversed. Both in terms of the top shares of income and of the overall income distribution summarized by the Gini coefficient, inequality in Finland has been increasing since the 1990s. In particular, Jäntti et al. (2010) document a surge in the share of income accruing to the top 1% since the mid-1990s. According to the authors, these dynamics are mostly due to a change in the composition of incomes, corresponding to an increase in the fraction of capital income going to the top of the income distribution. A key determinant of this development is the 1993 dual income tax (DIT) reform, which shifted taxation on capital income in Finland from progressive to proportional.
As done for Denmark, let us start by focusing on factor income trends in Finland. Table 3 presents the concentration of capital income across the income distribution for Finland between 1987 and 2016. As before, the income distribution is divided into four groups, the bottom 50%, the middle class 50–90%, and the top 10%, which is further divided into the bottom half and the top half of the decile.
Capital income shares by total income group, Finland 1987–2016.
| Total income group | 1987 | 1991 | 1995 | 2000 | 2004 | 2007 | 2010 | 2013 | 2016 |
|---|---|---|---|---|---|---|---|---|---|
| 0–50% | 16% | 16% | 11% | 6% | 7% | 7% | 9% | 9% | 11% |
| 50–90% | 27% | 30% | 25% | 18% | 18% | 19% | 20% | 24% | 20% |
| 90–95% | 13% | 12% | 12% | 8% | 9% | 8% | 8% | 10% | 10% |
| 95–100% | 45% | 42% | 53% | 67% | 66% | 66% | 63% | 57% | 59% |
| Capital share (BW) | 14% | 5% | 24% | 28% | 28% | 29% | 22% | 19% | 20% |
| Gini pre-tax national | 0.32 | 0.31 | 0.34 | 0.35 | 0.34 | 0.35 | 0.36 | 0.36 | n.a. |
| income (WID) | |||||||||
| Gini total factor | 0.41 | 0.42 | 0.49 | 0.50 | 0.51 | 0.51 | 0.51 | 0.51 | 0.53 |
| income (LIS) |
Table 3 shows that capital incomes go steadily more toward the top 5% of the income distribution up to the year of the financial crisis (2007), with the bottom 95% losing a fraction of 19% of total capital incomes when comparing 2007 with 1987. On aggregate, the capital share of income rises from 14 to 29% in the years of the crisis, before it declines to 19% in 2013. In summary, a shift in the concentration of capital incomes at the very top (5%) of the income distribution took place up to 2007. We present in Table 4 the dynamics of the concentration of the other factor income, namely, labor.
Labor income shares by total income group, Finland 1987–2016.
| Total income group | 1987 | 1991 | 1995 | 2000 | 2004 | 2007 | 2010 | 2013 | 2016 |
|---|---|---|---|---|---|---|---|---|---|
| 0–50% | 21% | 20% | 15% | 16% | 16% | 16% | 14% | 13% | 13% |
| 50–90% | 55% | 56% | 58% | 58% | 58% | 58% | 58% | 58% | 58% |
| 90–95% | 10% | 10% | 12% | 11% | 11% | 12% | 12% | 12% | 12% |
| 95–100% | 14% | 14% | 15% | 15% | 15% | 15% | 15% | 16% | 17% |
| Capital share (BW) | 14% | 5% | 24% | 28% | 28% | 29% | 22% | 19% | 20% |
| Gini pre-tax national | 0.32 | 0.31 | 0.34 | 0.35 | 0.34 | 0.35 | 0.36 | 0.36 | n.a. |
| income (WID) | |||||||||
| Gini total factor | 0.41 | 0.42 | 0.49 | 0.50 | 0.51 | 0.51 | 0.51 | 0.51 | 0.53 |
| income (LIS) |
Table 4 shows that labor incomes accrue steadily more toward the top 50% of the income distribution. The share accruing to the bottom 50% of the distribution decreases from 21% in 1987 to 13% in 2016; hence, a fraction of 8% of capital incomes shifted to the top half of the distribution.
Overall, Tables 3 and 4 show that the very top of the income distribution substantially increased its share of capital incomes up to the year of the financial crisis, in addition to a slight increase in its share of labor incomes. Hence, in terms of income composition, the incomes of the very wealthy became more capital intensive, while the opposite occurred at the bottom of the distribution. The post-crisis period 2010–2016 instead presents levels of concentration similar to those of the pre-crisis period. As for Denmark in the previous section, before we proceed to estimate the ICI index itself, we present the series of the areas below the concentration curves for labor and capital, as shown in Figure 6.
Evolution of the areas of factor concentration curves for Finland.
First, focus on the (blue) series of the area of the concentration curve for capital,
Income composition inequality Index in Finland, 1987–2016.
This figure plots the ICI index estimates for Finland using micro data on factor income from LIS (2020) for the following years: 1987, 1991, 1995, 2000, 2004, 2007, 2010, 2013, and 2016. Capital shares going to the top 5% of the total income distribution are also reported.
Figure 7 shows that in the late 1980s and early 1990s, the ICI index for Finland was rather low, indicating a low degree of income composition inequality. Low values of the ICI index (regardless of the sign) imply that the composition of incomes is quite homogeneous across the income distribution. This also implies that the concentration of factor incomes across the income distribution is rather similar for both factor incomes. This picture radically changes from 1995 up to the years after the financial crisis of 2007–2010. The ICI index exhibits a surge, reaching approximately 0.6 points in 2000 and 2004. In those years, the highly positive values of the ICI index indicate that capital incomes are now concentrated more toward the top of the income distribution relative to labor incomes, with high-income earners deriving the majority of their income from capital and low-income earners deriving a large share of their income from labor. In summary, our estimates of the degree of income composition inequality in Finland accord with the evidence of Jäntti et al. (2010) on the surge in the share of income going to the top 1%. Jäntti et al. (2010) identified the surge in capital incomes for the top 1% as the cause of increased concentration of incomes at the very top of the distribution. Our ICI index estimates confirm that there has been a shift in the composition of incomes across the distribution, especially from 1995 to 2007, due to capital incomes moving from lower deciles of the income distribution to the very top.
In Figure 8, we compare the evolution of the net capital share of income, the ICI index and the Gini coefficient calculated on market income for the period 1980–2016. Figure 8 shows that Finland witnessed a significant increase in the capital share of income between the early 1990s and 2007. This coincides with progressively higher values of the ICI index, which indicate, ceteris paribus, a strengthening of the transmission mechanism between the functional and personal distributions of income. In the same period, the Gini coefficient also shows an increase from approximately 0.31 to approximately 0.35.[16] Although the ICI index starts from low values in the early 1990s, the evolution of the ICI index in this period indicates a relevant role of the functional distribution of income in the determination of personal income inequality. In the final period from 2008 to 2013, it is plausible that factors other than the functional distribution of income played a role in the dynamics of the Gini coefficient. In conclusion, the hypothesis implying that the surge in the capital share of income has been an important driver of increased inequality in Finland seems to apply mostly for the central period of our analysis, from the early 1990s up to the years of the financial crisis.
Capital share, ICI index and Gini: Finland, 1987–2016.
The left-hand side graph plots the capital share of income in the period 1987–2013, obtained from Bengtsson and Waldenström (2018). The graph in the middle plots the ICI index for Finland for the period 1987–2016. The graph on the right-hand side plots the Gini coefficient for Finland from 1987 to 2013, obtained from the WID.World (2020).
3.3 Norway
Aaberge and Atkinson (2010) use tax data to document that top income shares in Norway started to increase in the late 1980s and early 1990s. The sharp rise in income shares accruing to the wealthiest households is recorded during the 1990s, in response to the introduction of the dual income tax (DIT) system in 1992. The dual income tax reform of 1992 (Alstadsaeter 2007; Sørensen 1994; Thoresen 2004) shifted the tax rate on capital incomes to be proportional, while that on labor incomes remained progressive. As a result of this decrease in progressivity (Thoresen 2004), capital incomes became more unevenly distributed in the 1990s, which contributed to increased income inequality. Aaberge, Atkinson, and Modalsli (2020) extend the analysis in Aaberge and Atkinson (2010), demonstrating that after the sharp increase in top income shares in 2005 due to the upcoming tax dividend reform, they subsequently stabilized at the levels of the late 1990s.[17] They also present the evolution of the pre-tax Gini coefficient, which increased between 1989 and 2013. Specifically, after a significant increase between the late 1980s and 2000, the Gini exhibited turbulent behavior around the years of the previously mentioned tax reforms and a slight increase thereafter. A third dimension of inequality, which is investigated in the literature on Norway and is highly relevant for our analysis, is the association between labor and capital incomes at the top of the distribution (the wage-capital composition of top incomes). As shown by Aaberge, Atkinson, and Königs (2018) for the top half of the income distribution, the association between capital and labor incomes in Norway grew between 1995 and 2005. However, when we focus more narrowly on the top 1% of capital income earners, this association declined in the period 1995–2005. To obtain a more comprehensive view on the evolution of inequality in income composition across the distribution, we turn to the evolution of both capital and labor income shares for different groups. As done for previous countries, we focus on the bottom 50%, the middle 50–90%, the 90–95 percentiles and the top 5%.
Table 5 documents the concentration of capital income shares by income group over the period 1979–2013. We observe significant overall changes in the shares for all groups (except the 90–95 group) in the income distribution. From 1979 to 1991, we observe a shift in the concentration of capital incomes from the top 10% toward the bottom 90% (the share of capital income going to the top 10% decreased to 49% from a value of 62% in 1979). From 1991 to 2004, we observe a striking shift in the concentration of capital incomes. In this period, the top 5% increased its share by 37 percentage points, potentially in response to the reduced progressivity of the taxation of capital incomes. The process of amplification of the share of capital incomes going to the top 5% in conjunction with the reduction of the shares going to the bottom 90% continued until 2004. From 2004 to 2013, the shares of capital incomes accruing to the bottom 90% regained some percentage points, while the top 5% appears to have lost slightly in the aftermath of the financial crisis. However, in these years, the wealthiest 5% still held far higher shares of capital incomes relative to the end of the 1980s, while the lower and middle classes held a lower share during these years.
Capital income shares by total income group, Norway 1979–2013.
| Total income group | 1979 | 1986 | 1991 | 1995 | 2000 | 2004 | 2007 | 2010 | 2013 |
|---|---|---|---|---|---|---|---|---|---|
| 0–50% | 16% | 21% | 22% | 13% | 8% | 5% | 11% | 8% | 10% |
| 50–90% | 21% | 30% | 28% | 20% | 14% | 12% | 22% | 21% | 23% |
| 90–95% | 10% | 11% | 9% | 7% | 7% | 7% | 10% | 9% | 10% |
| 95–100% | 52% | 38% | 40% | 60% | 70% | 77% | 58% | 62% | 57% |
| Capital share (BW) | 19% | 19% | 24% | 23% | 32% | 32% | 34% | 30% | 31% |
| Gini pre-tax national | n.a. | 0.26 | 0.25 | 0.29 | 0.33 | 0.34 | 0.33 | 0.32 | 0.32 |
| income (WID) | |||||||||
| Gini total factor | 0.38 | 0.38 | 0.42 | 0.45 | 0.47 | 0.49 | 0.47 | 0.48 | 0.48 |
| income (LIS) |
In addition to the weighted data sample of capital incomes from LIS (2020) plotted in Table 5, for Norway, we also have access to register data on capital incomes for the whole population from Statistics Norway (Statistics Norway 2020). For each year from 1993 to 2015, Figure 9 plots the average of the capital incomes accruing to the top 5%, 10% and to the 50–90th income percentiles (rank variable is total income). Figure 9 substantially confirms the evidence in Table 5. First, in the years post-DIT reform, capital incomes for the top income percentiles increased, especially for the top 5%. Second, it is clear that 2000 and 2005 were years in which anticipations of upcoming dividend tax reforms determined higher capital incomes for the top 5%. Regarding the aggregate statistics on the functional and personal income inequality also summarized in Table 5, the capital share of income increases monotonically from 19% in 1986 to 34% in 2007, before it declines slightly in the last years considered. In other words, in addition to a structural shift in the shares of capital incomes accruing to the different income groups, we also observe an overall increase in the share of total income going to capital up to 2007. Regarding the Gini coefficient, both series in Table 5 show a sustained increase, mostly in the years subsequent to the tax reform of 1992. See Appendix D for a discussion on entropy measures.
Average capital incomes for different total income percentiles.
This figure plots register data on income retrieved from Statistics Norway, Statistics Norway (2020). Population is ranked according to total income, including capital and labor. We compute the average capital incomes (including capital gains) received by each total income percentile.
Similar to the other Nordic countries, in Norway, we witness relatively more stable shares of labor income. Table 6 shows changes in the shares of labor income that are of smaller magnitude with respect to the turbulent shares observed for capital incomes. Nevertheless, the share of labor income accruing to the bottom 50% decreased from 24% in 1979 to 17% in 2013, while the share of labor income going to the top 50% increased from 77% in 1979 to 83% in 2013. This shows that the bottom of the distribution is also losing ground with respect to labor incomes.
Labor income shares by total income group, Norway 1979–2013.
| Total income group | 1979 | 1986 | 1991 | 1995 | 2000 | 2004 | 2007 | 2010 | 2013 |
|---|---|---|---|---|---|---|---|---|---|
| 0–50% | 24% | 23% | 20% | 18% | 19% | 18% | 18% | 17% | 17% |
| 50–90% | 55% | 55% | 56% | 58% | 57% | 58% | 57% | 57% | 57% |
| 90–95% | 10% | 10% | 10% | 11% | 11% | 11% | 11% | 11% | 11% |
| 95–100% | 12% | 12% | 13% | 13% | 13% | 13% | 15% | 15% | 15% |
| Capital share (BW) | 19% | 19% | 24% | 23% | 32% | 32% | 34% | 30% | 31% |
| Gini pre-tax national | n.a. | 0.26 | 0.25 | 0.29 | 0.33 | 0.34 | 0.33 | 0.32 | 0.32 |
| income (WID) | |||||||||
| Gini total factor | 0.38 | 0.38 | 0.42 | 0.45 | 0.47 | 0.49 | 0.47 | 0.48 | 0.48 |
| income (LIS) |
Let us turn to use the evidence in Tables 5 and 6 as building blocks of the ICI index, namely, the series of the areas below the concentration curves for capital (blue) and labor (red), the difference of which at each point in time uniquely determines the sign and magnitude of the ICI index, as explained in Eq. (5). Figure 10 shows the series of the areas below the concentration curves for both factor incomes in Norway in the period 1979–2013.
Evolution of the areas of factor concentration curves for Norway.
Similar to the cases of Denmark and Finland, the (blue) series of the area below the concentration for capital shows the highest variation in Figure 10, declining substantially from the early 1990s to 2004. However, in contrast to the cases of Denmark and Finland, the series of
We now focus on our main contribution: the series of the ICI index for Norway in Figure 11. In the initial years 1979–1986–1991, the ICI index decreases from approximately 0.4 to approximately 0.2. This implies that the composition of incomes in the late 1970s and 1980s for the lower parts of the distribution of income gradually becomes more similar to that of the upper parts. The monotonic tendency toward higher levels for the ICI index in Norway starts in the 1990s, with a clear jump in the period 1991–1995. In our view, this tendency can be attributed to the shift in the concentration of capital incomes toward the top (as shown in Table 5 and Figure 10) after the DIT reform of 1992. The ICI index reaches a rather high value in 2004, slightly above 0.7.[18] In the period 2007–2013, the ICI index again decreased to lower values similar to those of the late 1990s.
Income composition inequality index in Norway, 1979–2013.
Authors’ computation based on LIS (2020) data. This figure shows the yearly ICI index for Norway using the following LIS waves: 1979, 1986, 1991, 1995, 2000, 2004, 2007, 2010, and 2013. Capital shares going to the top 5% of the total income distribution are also reported.
In summary, we estimate for Norway a clear tendency toward a higher degree of income composition inequality throughout the period from 1991 to 2004, potentially due to the tax reform of 1992 that introduced proportional taxation of capital incomes. This implies that throughout the 1990s, the incomes of the upper part of the income distribution became increasingly capital intensive with respect to the incomes of the poor. In the aftermath of the financial crisis (post-2007), Table 5 and Figure 10 show instead levels of concentration of capital incomes in Norway closer to those of the late 1990s. This is reflected in the lower values of the ICI index in these years, in the range 0.4–0.5. Finally, to detect whether there is a direct relationship between changes in the functional distribution of income and changes in personal inequality, we jointly investigate in Figure 12 the evolution of the capital share of income, the ICI index and the Gini coefficient for income. We divide the analysis into three subperiods. In the first subperiod 1979–1991, the ICI index displays a low and falling degree of income composition inequality, while the capital share of income appears to be rather volatile. In these years, personal income inequality is clearly decreasing. In other words, the estimation of the ICI index for these years leads us to exclude the functional distribution of income being a main driver of the series of the Gini in Figure 12. The most interesting subperiod is between 1991 and 2005. This period is characterized by a clearly increasing capital share of income (from 24% in 1991 to approximately 36% in 2005), significant positive levels of the ICI index (especially in the years from 1995 to 2004) and increasing Gini (from approximately 0.25 in 1991 to approximately 0.38 in 2005). In our view, the increase in the capital share in these years is partially responsible for the increase in income inequality. The final subperiod, between 2005 and 2013, instead witnessed a decreasing capital share and a stable or decreasing Gini (excluding 2005, which was affected by the dividend tax reform). The transmission mechanism between the functional and personal distributions seems to be weaker in these years (with respect to the values of the ICI index in 2000 and 2004).
Capital share, ICI index and Gini: Norway, 1979–2013.
The left-hand side graph plots the series of the capital share of income for the period 1979–2013 derived from the Bengtsson–Waldenström (BW) Historical Capital Shares Database in Bengtsson and Waldenström (2018). The graph in the middle plots the series of the ICI index for Norway for the period 1979–2013. The graph on the right-hand side plots the Gini coefficient for Norway from 1980 to 2013, source: World nequality Database (WID.World 2020).
In conclusion, for Norway, the hypothesis that changes in the capital share of income constitute a driver of increased personal income inequality seems to hold uniquely for the central subperiod of our analysis (from the early 1990s to 2005, similar to what observed for Finland). In these years, and especially from 1995 to 2005, due to capital incomes shifting dramatically from lower deciles of the income distribution to the very top, the high levels of the ICI index, which are associated with a high partial derivative of the Gini coefficient with respect to changes in the capital share, allow us to claim that the functional distribution has indeed been actively driving the increase in personal income inequality.
3.4 Sweden
Domeij and Flodén (2010) document inequality trends in Sweden using labor incomes for the period 1978–2004 and examine both pre-tax and disposable earnings. The main message is that a clear increase in the dispersion of pre-tax earnings occurred in Sweden in the early 1990s, a result that weakens when instead focusing on the inequality of disposable earnings. The authors explain this development through the gradual reformation of the centralized wage-setting system in the Swedish economy from the 1980s, which reduced the coverage of collective bargaining agreements and increased the use of industry-wide bargaining. García-Peñalosa and Orgiazzi (2013) as well claims that the increase in income inequality that started in 1980s was mainly due to greater earnings dispersion. Analyzing the long-run dynamics of the labor share in Sweden from the 1900s to the 2000s, Bengtsson (2014) shows that the share of labor income began to decline in the early 1980s. Hence, not only did inequality in pre-tax earnings rise, as shown in Domeij and Flodén (2010) but the share of total income accruing to labor also decreased since the 1980s, mostly to the advantage of capital earners in particular and to top income earners in general.
Focusing on the long-run concentration of top incomes in the Swedish economy from 1903 to 2004, Roine and Waldenström (2008) and Roine and Waldenström (2012) show that the portion of capital incomes generated by capital gains plays a decisive role in inequality estimates. When capital gains are included in the definition of capital income, the top 10% income share has increased substantially since the 1990s. Excluding capital gains instead leads to an increase in the share of income going to the top 10% that is more in line with other countries in continental Europe. As shown in Roine and Waldenström (2008) and Roine and Waldenström (2012), since the 1990s, a series of tax reforms that gradually decreased the marginal tax rate on capital (relative to the taxation of labor incomes) created incentives for capital earners to realize larger shares of their investments. In turn, increased capital gains boosted capital incomes at the top of the distribution. The magnitude of this jump is also partially explained by asset price increases in post-1980 deregulated financial markets. On the other hand, this implies that in the period before the 1990s, the concentration of capital incomes at the top of the income distribution in the available income data is largely underestimated. We will return to this point further below.
Overall, the above evidence points in the direction of a higher capital share of income, greater concentration of capital incomes at the very top, and, in turn, a higher Gini coefficient for Sweden starting in the 1980s.
As we have done for the countries considered previously, let us analyze how the approach of considering the inequality of income composition brings novel insights to this literature. Table 7 presents the concentration of capital income across the (total) income distribution for Sweden in between 1975 and 2005.[19]
Capital income shares by total income group, Sweden 1975–2005.
| Total income group | 1975 | 1981 | 1987 | 1992 | 1995 | 2000 | 2005 |
|---|---|---|---|---|---|---|---|
| 0–50% | 32% | 43% | 36% | 31% | 10% | 9% | 10% |
| 50–90% | 28% | 34% | 30% | 38% | 60% | 17% | 20% |
| 90–95% | 9% | 7% | 8% | 9% | 12% | 8% | 8% |
| 95–100% | 31% | 16% | 26% | 22% | 18% | 67% | 63% |
| Capital share (BW) | 12% | 11% | 14% | 13% | 18% | 13% | 13% |
| Gini pre-tax national | na. | 0.27 | 0.28 | 0.28 | 0.31 | 0.29 | 0.31 |
| income (WID) | |||||||
| Gini total income (LIS) | 0.42 | 0.43 | 0.46 | 0.49 | 0.53 | 0.53 | 0.51 |
Regarding capital incomes, the overall picture shows high volatility in the fraction of capital incomes accruing to most of the income groups. Excluding the initial and final years, the clearest trend is of an enormous shift of capital incomes from the bottom 50% to the top 5% between 1981 and 2000. The bottom 90% loses a fraction close to 50% of capital incomes in the period 1981–2005, while the opposite is the case for the top 10%. On aggregate, the capital share of income rose from 11 to 18% in the period 1981–1995 before it declined back to the levels of the 1980s. Regarding the Gini coefficients, both series from WID.World (2020) and LIS (2020) show a monotonic increase throughout the period, with much of it occurring in the period 1981–1995 for both measures. See Appendix D for a discussion on entropy measures. Before we proceed, we describe the dynamics of the concentration of labor incomes in Table 8.
Labor income shares by total income group, Sweden 1975–2005.
| Total income group | 1975 | 1981 | 1987 | 1992 | 1995 | 2000 | 2005 |
|---|---|---|---|---|---|---|---|
| 0–50% | 20% | 17% | 16% | 13% | 12% | 15% | 15% |
| 50–90% | 55% | 57% | 58% | 59% | 58% | 57% | 57% |
| 90–95% | 10% | 11% | 11% | 12% | 12% | 11% | 12% |
| 95–100% | 15% | 15% | 15% | 16% | 18% | 17% | 17% |
| Capital share (BW) | 12% | 11% | 14% | 13% | 18% | 13% | 13% |
| Gini pre-tax national | n.a. | 0.27 | 0.28 | 0.28 | 0.31 | 0.29 | 0.31 |
| income (WID) | |||||||
| Gini total income (LIS) | 0.42 | 0.43 | 0.46 | 0.49 | 0.53 | 0.53 | 0.51 |
Although with smaller magnitudes than in the case of capital incomes, Table 8 presents a trend of labor incomes shifting from the bottom 50% to the top 50% throughout the period of analysis. In summary, Tables 7 and 8 indicate that the very top of the income distribution has dramatically increased its share of capital incomes, in addition to the mild increase in its share of labor incomes. From the perspective of income composition, this implies that the incomes of the very wealthy became more capital intensive, while the opposite is the case at the bottom of the distribution. Most of this shift in concentration occurred in the two decades from 1981 to 2000.
Figure 13 plots the series of the areas below the concentration curves for capital (blue) and labor (red). Similar to the cases of Denmark, Finland and Norway, the (blue) series of the area below the concentration for capital presents dynamics with higher variation in Figure 13, decreasing steadily from the early 1980s to 2000. The series of
Evolution of the areas of factor concentration curves for Sweden.
In the following, we attempt to synthesize these dynamics into one a summary statistic, the ICI index for Sweden in the period 1975–2005. Figure 14 shows the results. The evidence shown in Figure 14 is somewhat striking. In contrast to the other Nordic economies previously analyzed in this paper, a significant fraction of the available series for the ICI index for Sweden is of negative sign. The ICI index for Sweden is below zero for the four observations available from 1975 to 1992, with a value of approximately −0.2 for the years 1981, 1987, and 1992. This situation appears to change gradually from 1995 onward, showing a rapid increase in the ICI index from approximately 0 in 1995 to approximately 0.5 in 2000–2005, reflecting the gradual increase in capital incomes accruing to the upper deciles of the income distribution shown in Table 7 and Figure 13. Focusing on the trend, the result emerging from Figure 14 is of increasing inequality in the income composition in Sweden throughout the period from 1981 to 2000. This result accords with the evidence obtained by Roine and Waldenström (2008, 2012 showing that capital incomes gradually became increasingly concentrated at the top of the income distribution. As explained above, Roine and Waldenström (2008, 2012 highlight the fact that only from the 1990s onward have Swedish capital owners started to realize larger fractions of their investments. Therefore, until the 1990s, the available data on capital incomes largely underestimate the concentration of capital incomes at the top of the income distribution. Before the 1990s, high marginal tax rates on capital and various loopholes in tax legislation are the reasons why capital incomes do not fully appear in the incomes of individuals ranked at the top of the income distribution. To analyze in greater detail the contribution behind the evidence in Figure 14, let us jointly observe in Figure 15 the series of the capital share of income, the ICI index and the Gini coefficient (WID.World 2020).
Income composition inequality index in Sweden, 1975–2005.
Authors’ computation based on LIS (2020) data. The figure shows the yearly ICI index for Sweden estimated using the following LIS waves: 1975, 1981, 1987, 1992, 1995, 2000, and 2005. Capital shares going to the top 5% of the total income distribution are also reported.
Capital share, ICI index and Gini in Sweden, 1975–2005.
The left-hand side graph plots the series of the capital share of income (value added net of capital depreciation minus employee compensation) in the period 1975–2005. This series is obtained from the Bengtsson–Waldenström (BW) Historical Capital Shares Database in Bengtsson and Waldenström (2018). The graph in the middle again plots the series of the ICI index for Sweden in the period 1975–2005, as shown in Figure 14. The graph on the right-hand side plots the Gini coefficient for Sweden from 1980 to 2005, source: World Inequality Database (WID.World 2020).
Let us divide the period 1975–2005 into two different subperiods. First, from 1975 to 1995, the series of the capital share increases significantly, with a jump from below 8% in 1977 up to approximately 18% in 1995. The ICI index shows negative values throughout the subperiod 1975–1992 before it becomes slightly positive in 2005. Hence, we would expect a rather limited (and, if any, negative) effect from the functional distribution of income on the Gini coefficient in those years. The evidence of a stable Gini coefficient of approximately 0.27–0.28 throughout the 1980s partially confirms a story of limited response of the level of personal income inequality to changes in factor shares of income. Second, the period from 1995 to 2005 witnesses again a rather volatile series of the capital share of income, jumping to 18% in the first half of the 1990s and subsequently declining in 2000 back to 12%. In these years, the ICI index jumps from approximately 0 in 1995 to 0.5 in 2000. This implies that the degree of income composition inequality is initially low but then increases to higher values in the years 2000–2005. Hence, the only period for which the dynamics of the capital share of income can be considered among the key determinants of the Gini coefficient is the period 2000–2005. In those years, however, volatility in both the capital share and the Gini coefficient lead us to exclude the possibility that the functional distribution has had a very significant effect on the level of personal income inequality. Overall, in contrast to the other Nordic countries, our results indicate that the functional income distribution cannot be identified as one of the determinants of increasing income inequality in Sweden.
4 Discussion
The four country-level analyses in Section 3 show how movements in capital incomes, which increasingly go to the top of the total factor income distribution, are responsible for the increase in income composition inequality in all the Nordic economies. The evidence of increased capital income shares at the top of the income distribution starting in the early 1990s is in line with findings in the literature (Aaberge and Atkinson 2010; Jäntti et al. 2010; Roine and Waldenström 2012; Søgaard 2018). However, the relation between the latter result and the increase in income composition inequality across the income distribution is a novel contribution. Furthermore, this case study has clearly shown how low income inequality may be accompanied by substantial heterogeneity in the composition of income. In Milanovic (2017)’s terminology, these findings would mean that the Nordics are moving toward a classical capitalism type of society, as ownership of factor incomes (capital and labor) is increasingly separated, creating two social groups that are non-overlapping in their income levels.
These main results are summarized in Table 9, in which the subperiods for which rising capital shares actually contributed to rising income inequality are highlighted (in gray). In each subperiod in which we observe a strong increase in the capital income share in conjunction with high income composition inequality, we also observe a strong increase in the Gini coefficient. The shaded subperiods are those characterized by relatively high responsiveness of the Gini coefficient to increases in the capital share of income. It emerges from our results that the use of the income composition inequality framework allows us to appreciate time-varying and country-specific relations in greater detail, as we also claim in Appendix A. Furthermore, the increase in income composition inequality and, relatedly, the stronger sensitivity of personal income inequality to changes in the capital share might represent a challenge to the Nordic egalitarian model in the future, as the thought experiment in Appendix B illustrates. A related aspect that emerged in the country sections is the tight relation of the ICI index to changes in taxation regimes. Hence, in Subsection 4.1, we will investigate in greater detail the role of taxation and, in particular, the introduction of dual income taxation. Finally, robustness checks are presented in Subsection 4.2.
4.1 Dual Income Tax Reforms and Progressivity
From the core of the paper, it emerges that the dual income tax (DIT) reforms in the Nordic countries in the early 1990s contributed to important shifts in the concentration of capital incomes across the income distribution and hence to rapid changes in the degree of income composition inequality. The aim of this subsection is therefore to investigate the role of taxation reforms in the dynamics of income composition inequality.
Sørensen (1994) describes in detail how the Nordic countries moved in the early 1990s away from the principles of global income taxation (GIT) (where global refers to the sum of a taxpayer’s income from all sources) toward a system of dual income taxation (DIT) in which taxation on capital and labor income is differentiated. The characteristic of the DIT system is that the lowest marginal tax rate for labor and transfer incomes (labeled personal [20] or earned income) is chosen as the proportional (flat) tax on capital incomes (encompassing all types of capital incomes). In other words, capital incomes are taxed at a flat rate considerably lower than the effective rate on personal incomes from labor and transfers. Overall, DIT reforms imply a lower progressivity of the tax system, potentially increasing the dispersion of after-tax incomes across the distribution, compared to the before-tax income distribution (Thoresen 2004).
However, claims on the causal distributional effects of the DIT systems ignore an important behavioral aspect regarding capital gains (as highlighted by Roine and Waldenström (2012) for Sweden) triggered by taxation reforms. Before the reforms, substantial shares of the incomes of shareholders were retained profits in firms; hence, capital gains were not realized, and the corresponding capital incomes did not appear at the top of the distribution. The DIT reforms lowered marginal tax rates on capital incomes (also broadening the tax base by eliminating tax favors for specific types of capital investments), triggering an extensive distribution of dividends and capital gains that contributed to the shift in the concentration of capital incomes toward the top of the distribution.
Readapted from Sørensen (1994), Table 10 presents the DIT reforms and the resulting marginal tax rates on personal and capital incomes in the Nordic countries. Table 10 clearly shows that the progressivity of the taxation of capital incomes has decreased in the Nordic countries as a consequence of the DIT reforms, especially with respect to taxation of labor incomes.
Summary of results by country.
| Income composition inequality | ||||||
|---|---|---|---|---|---|---|
| Country | Period | Variation (weak/strong) or sign (high/low) | ||||
| Cap. share | mu p | mu w | ICI index | Gini | ||
| Denmark | 1987–2016 | + (Weak) | − (Strong) | − (Weak) | + | + (Strong) |
| 1987–1995 | + (Strong) | − (Strong) | − (Weak) | + (Low) | − (Weak) | |
| 1995–2009 | − (Strong) | − (Strong) | − (Weak) | + (Low) | = | |
| 2009–2013 | + (Strong) | − (Weak) | − (Weak) | + (High) | + (Strong) | |
| Finland | 1987–2016 | + (Weak) | − (Strong) | − (Weak) | + | + (Strong) |
| 1990–2007 | + (Strong) | − (Strong) | − (Weak) | + (High) | + (Strong) | |
| 2008–2013 | − (Weak) | + (Weak) | − (Weak) | + (High) | + (Weak) | |
| Norway | 1979–2013 | + (Strong) | − (Strong) | − (Weak) | + | + (Weak) |
| 1979–1991 | + (Weak) | + (Weak) | + (Weak) | + (Low) | − (Strong) | |
| 1991–2005 | + (Strong) | − (Strong) | − (Strong) | + (High) | + (Strong) | |
| 2005–2013 | − (Weak) | + (Weak) | − (Weak) | + (High) | − (Weak) | |
| Sweden | 1975–2005 | + (Weak) | − (Strong) | − (Weak) | −/+ | + (Strong) |
| 1975–1995 | + (Strong) | − (Strong) | − (Weak) | − (Low) | + (Strong) | |
| 1995–2005 | − (Weak) | − (Strong) | + (Weak) | + (High) | − (Weak) | |
-
mu p is the area below the concentration curve for capital income; mu w is the area below the concentration curve for labor income. Heuristically, we define a high income composition inequality index as being above 0.4.
Marginal income tax rates in the Nordics, before/after DIT reforms.
| Country | (1) | (2) |
|---|---|---|
| Tax rates personal incomea | Tax rates capital income | |
| Denmark | ||
| Before 1993–1994 reform | 50–68 | 50–56 |
| After 1993–1994 reform | 38–58 | 38–44 |
| Finland | ||
| Before 1993 reform | 25–57 | 25–57 |
| After 1993 reform | 25–57 | 25 |
| Norway | ||
| Before 1992 reform | 26.5–50 | 26.5–40.5 |
| After 1992 reform | 28–41.7 | 28 |
| Sweden | ||
| Before 1991 reform | 36–72 | 36–72 |
| After 1991 reform | 31–51 | 30 |
-
aThis table is rearranged from Table 1 (page 59) in Sørensen (1994). Sørensen (1994) also includes the corporate income tax rates before and after the introduction of the DIT system, which have been excluded from the table above since they lie beyond the scope of this paper.
In this work, rather than being interested in disentangling the causal distributional effects of the DIT reforms (as done, for instance, in Thoresen (2004) for Norway), we are interested in how taxation policy changes may alter the degree of income composition inequality. Based on the evidence from the core sections of this paper, our informed guess is that the introduction of the DIT reforms led to a jump in the concentration of capital incomes toward the top, implying a lower area of the concentration curve for capital incomes and correspondingly higher values of the ICI index. Since a higher value of the ICI index is associated with a stronger relationship between the functional and personal distributions of income, we expect in the aftermath of the DIT reforms to observe a stronger responsiveness of the personal income distribution to changes in capital share of income.
Figure 16 plots, for each of the four countries under analysis, the series of the partial derivative of Eq. (6) obtained by substituting the series of the areas below the single-factor concentration curves,
Gini responsiveness and dual income taxation.
For each of the four countries under analysis, we plot here the series of the partial derivative of Eq. (6) obtained by substituting the series of
4.2 Robustness: The Inclusion of Transfer Income
We perform a robustness check by including transfers in the core definition of income given in Eqs. (1) and (2). Specifically, we include public pensions (both contributory and not), private pensions, other public social benefits and private transfers (cash transfers) as labor income in Eq. (2). This allows us to check whether our main results are robust when including components that may be considered as deferred income. In particular, including public contributory pensions also implies that elderly people are placed in the correct position in the income distribution instead of being considered poorer than they effectively are. Since transfer income affects more of the bottom part of the income distribution (especially public social benefits), our guess is that the new income definition entails a more labor-intensive (capital-intensive) income for the poor (wealthy) relative to the baseline definition of income. Other things being equal, a higher concentration of labor income toward the bottom of the distribution increases the area under the concentration curve for labor, shifting the ICI index upwards. The results are presented in Figure 17.
ICI index in the Nordics, including transfer income.
Authors’ computation based on LIS (2020) data. The figure shows the yearly ICI index for Denmark, Finland, Norway and Sweden estimated using all available LIS (2020) waves, with and without transfer income.
As expected, we find higher values for the ICI index for all the Nordic countries and throughout the period of analysis (the magnitude of the increase is approximately 0.1–0.2 points). This evidence confirms the hypothesis that the polarization in income composition between top earners with capital-intensive incomes and bottom earners with more labor-intensive incomes becomes stronger when transfer income is taken into account. Moreover, the trends in the index closely follow those for the baseline definition of income in all Nordic countries, such that our conclusions are robust to the inclusion of transfer income.
5 Concluding Remarks
In this research, we provide novel estimates on the degree of income composition inequality in Nordic countries in recent decades. In our view, the set of contributions of the paper can be summarized into two main strains.
First, this paper provides novel evidence of the structural change in income composition taking place in Nordic countries since the early 1990s. This finding is robust to country heterogeneity in trends and levels of the capital shares of income. We document rising inequality in the composition of individual incomes, mostly derived from a shift in capital incomes toward the top of the income distribution. We also provide a discussion of the potential determinants of these dynamics, identifying the reforms of capital income taxation that took place in the early 1990s (dual income taxation reforms) in the Nordic countries. We highlight that the lower progressivity of capital income taxation introduced by the DIT reforms lies behind the shift in capital incomes toward the top of the distribution that led to rising inequality in income composition and strengthens the responsiveness of personal income inequality to changes in the aggregate capital share.
Second, we use the estimates of the degree of income composition inequality to understand whether changes in the capital share of income can be considered among the determinants of the level of personal income inequality in the Nordic economies in recent decades. In summary, our descriptive analysis shows that for Sweden, rising capital shares of income did not have a significant impact on the level of personal income inequality. For Finland and Norway, the results show that a high degree of income composition inequality in the period from the 1990s up to approximately 2005 implies that changes in the capital share indeed contributed to increasing personal income inequality. For Denmark, the same applies but only for the period from 2009 to 2013.
We argue that if the trends of increased heterogeneity among individual factor shares uncovered in this work persist and if these are coupled with more significant increases in the aggregate capital share, the transmission from the functional to the personal distribution of income might represent an important factor potentially increasing personal income inequality in the Nordic economies in the future.
Funding source: Horizon 2020 Framework Programme
Award Identifier / Grant number: 730998
Acknowledgements
We thank R. Aaberge, E. Bengtsson, A. Cupak, M. Guerini, B. Milanovic, J. Modalsli, T. Munzi, M. Occelli, H. Omar, G. Pallante, A. Pietropaoli, M. Ranaldi, J. Roine, A. Roventini, E. Rubolino, J. E. Søgaard, the Editor in charge of our submission Till Requate, anonymous reviewers, seminar participants at the European Economic Association 2020 Virtual Conference, at the Luxembourg Income Study (LIS), at the BI Norwegian Business School, and at the Sant’Anna School for Advanced Studies for helpful comments and suggestions. The research leading to these results received funding from the European Union’s Horizon 2020 research and innovation programme under grant agreement No. 730 998, InGRID-2 – Integrating Research Infrastructure for European expertise on Inclusive Growth from data to policy. All data and codes to fully replicate the results of this paper are publicly available on OSF here. All errors are our own.
A Cross-Country Evidence
In this section of the Appendix, we shift the focus to cross-country analysis. We do not provide a fully-fledged econometric analysis of the long-run relationship between the capital share of income and income inequality in the Nordic countries, as already done in Bengtsson and Waldenström (2018). Rather, the following cross-country evidence will help to interpret the country-specific results provided in subsections 3.1–3.4. First, Table 11 provides some summary statistics.
Summary statistics.
| Variable | Observations | Mean | Std. Dev. | Min | Max |
|---|---|---|---|---|---|
| Gini of total income (WID.World) | 114 | 0.318 | 0.028 8 | 0.252 | 0.384 |
| Capital share of income | 128 | 19.78 | 7.002 | 5.250 | 37.66 |
| Top marginal tax rate (capital, %) | 134 | 42.29 | 16.87 | 25 | 72 |
| GDP per capita (current USD) | 131 | 34,978 | 20,294 | 4,697 | 102,913 |
| Market capitalization (% of GDP) | 98 | 48.04 | 42.27 | 2.676 | 258.4 |
| Government expenditure (% of GDP) | 131 | 23.04 | 2.612 | 17.90 | 27.93 |
-
Capital share of income (1975–2013) for Denmark, Finland, Norway, Sweden were retrieved from the Bengtsson–Waldenström (BW) Historical Capital Shares Database in Bengtsson and Waldenström (2018). Time series for the Gini coefficient (WID.World) (1980–2013) for Denmark, Finland, Norway, Sweden, were retrieved from: World Inequality Database WID.World (2020). Top marginal tax rates on capital incomes are taken from Table 10. GDP per capita, current USD; market capitalization of listed domestic companies, as a % of GDP; general government final consumption expenditure, as a % of GDP, are retrieved from the World Development Indicators of the World Bank.
Let us introduce the baseline linear specification. Dependent variable is the Gini coefficient (WID.World 2020) for country i = 1, …, 4 and time t; μ i comprises country fixed effects; t captures time fixed effects; CS it is the capital share in value added for country i = 1, …, 4 and time t; X it is a matrix of control variables (top marginal tax rate on capital incomes; GDP per capita, current USD; market capitalization of listed domestic companies, as a % of GDP; general government final consumption expenditure, as a % of GDP), and finally ɛ it is the random error term.
Table 12 presents the results of the baseline GLS specification, with fixed effects and controls introduced step-wise. The non-significant estimate in column (4) of Table 12 indicates, if anything, that the transmission mechanism measured by the ICI index has been weak on average for the countries analyzed.[23]
Capital share and income inequality in the Nordics.
| (1) | (2) | (3) | (4) | |
|---|---|---|---|---|
| Gini(WID) | Gini(WID) | Gini(WID) | Gini(WID) | |
| Capital share of income | 0.265*** | 0.158* | 0.179 | 0.114 |
| (0.100) | (0.091) | (0.146) | (0.088) | |
| Controls | No | No | Yes | Yes |
| Year F.E. | No | Yes | No | Yes |
| Country F.E. | No | Yes | No | Yes |
| Observations | 114 | 114 | 93 | 93 |
-
Robust standard errors in parentheses. * p < 0.10, ** p < 0.05, *** p < 0.01.
-
Column [1] reports estimates of β 1 without controls and fixed effects, which are introduced step-wise in columns [2, 3, 4].
Apart from standard endogeneity caveats that can be raised regarding the model specification in Eq. (7), we intend to raise a fundamental issue here. Estimating the shape of the relationship between the functional and the personal distribution of income, as done in Eq. (7) through the GLS coefficient β 1, is in our view a sub-optimal approach for two main reasons. First of all, the relationship we aim to estimate between functional and personal income inequality is inherently non-constant and country-specific, as shown in the core sections of this paper. Hence, approaching it from a cross-country perspective limits a deeper understanding of the underlying forces. Second and most importantly, the shape of the relationship between the functional and the personal distribution grasped in Eq. (7) through the coefficient β 1, is only one side of the story. The degree of responsiveness of the Gini coefficient to changes in capital share of income is crucially hinging on the strength of the transmission mechanism – the degree of income composition inequality. This point is illustrated by the two hypothetical cases below:
(High β 1 , low ICI index in absolute value). In this case, the degree of income composition inequality is low (i.e., the ICI index close to zero). A high value of the β 1 indicates that changes in the Gini coefficient strongly correlate with changes in the capital share of income. However, the low degree of income composition inequality allows us to disregard the functional distribution as a key determinant of the Gini coefficient.
(High β 1 , high ICI index in absolute value). The degree of income composition inequality is high (e.g., ICI index is close to one). A high value of the β 1 indicates that changes in the Gini coefficient strongly correlate with changes in the capital share of income. This time, the high degree of income composition inequality highlights the functional distribution as a key determinant of the Gini coefficient.
To convey graphically the argument above, Figure 18 plots the β
1 coefficient from Table 12 jointly with the derivatives in Eq. (6) by employing the series of the
Comparison between pooled derivatives and regression coefficient.
The horizontal green line is the β 1 from column 4 of Table 12, representing the average of the estimated elasticity of the Gini coefficient to the functional income distribution. The colored dots represent instead the elasticity in Eq. (6) and estimated by using LIS (2020) data for each country/year.
B A Simulation Exercise
In this section of the Appendix we introduce a simple simulation exercise, aimed at showing the relevance of the approach of income composition inequality for understanding the evolution of personal income inequality. To this end, we introduce a benchmark country in order to provide context to our thought experiment, namely the United States. The Nordic countries and the US are at the two extremes of the OECD countries for what concerns income inequality. How does that modify when we shift the focus to income composition inequality? A similar approach, although with focus on relative income mobility, was followed by Björklund and Jantti (1997).
For comparability reasons, we utilize the LIS (2020) data for the estimation of the ICI index for the US. For the last wave available, year 2016, the ICI index turns out to be equal to 0.26. It follows that, whereas the Nordic countries stand out in the comparison with the US as being relatively more egalitarian in terms of income inequality, this needs not be the case for what concerns inequality in income composition. In fact, the degree of income composition inequality is higher for Denmark, Finland and Norway in the latest years, being respectively equal to 0.50, 0.42 and 0.43.[24]
A lower income composition inequality in the US means that, paradoxically, a rising capital share of income might represent a larger threat for the dynamics of income inequality in the Nordic countries, compared with the US, all else held constant. This represents a further reason for which it might be relevant from a policy point of view to follow the development of income composition inequality in the Nordic countries.
In order to make our point clearer, we construct a deterministic forecasted path for the Gini coefficient conditional on initial values calibrated on data for the Nordic countries and for the United States, leaving aside Sweden due to lack of recent observations. We use data on average rates of return on capital from the database in Jordà et al. (2019). From the same database we compute growth rates for GDP in the different countries. Moreover, as in the previous sections, we consider net capital shares from Bengtsson and Waldenström (2018) and Gini coefficients on market incomes from WID.World (2020). For rates of return on capital, GDP growth rates and net capital shares, we average the available observations from 1994, year from which we observe a structural change in the series for the ICI index in the Nordic countries, to the most recent observations. For what concerns the initial value of the Gini coefficient (our variable of interest), we start from country specific empirical values for the latest observation available.
An important building block of our simulation exercise is given by the responsiveness of the Gini coefficient with respect to changes in capital income shares, estimated on LIS microdata through Eq. (6). In particular, for each country we consider the average of the estimates of the degree of responsiveness in the period 1994–2016. Starting from the empirical values of our variables of interest and from the country-specific average responsiveness rates, we simulate a simple system that evolves according to the following laws of motion.
For each country, GDP grows at a constant rate of growth g i , namely the empirical average calculated for the years 1994–2015, such that the evolution of GDP reads:
Capital in this system evolves as follows:
with r i assumed to be fixed over time, heterogeneous across countries, and equal to the empirical average rate of return on capital. Based on this information, at each time step we compute the capital share of income, as described in Eq. (10):
For the purpose of this thought experiment, the Gini coefficient of income is assumed to depend (i) on past Gini coefficients and (ii) on the degree of responsiveness of the Gini coefficient to the change in the capital share of income,
We simulate the above system of equations for T = 30 years, and show the resulting percentage growth in the Gini coefficient for each of the countries considered. We present the simulation results in Table 13.
Simulation results.
| Country | ϵ i | r i − g i | Gini growth (%) |
|---|---|---|---|
| Norway | 0.227 6 | 0.110 0 | 33 |
| Denmark | 0.133 4 | 0.087 3 | 0.64 |
| Finland | 0.258 4 | 0.127 7 | 62 |
| United States | 0.140 0 | 0.070 0 | −2.5 |
| Norway | β 1 = 0.114 | 0.110 0 | 16 |
| Denmark | β 1 = 0.114 | 0.087 3 | 0.55 |
| Finland | β 1 = 0.114 | 0.127 7 | 27 |
-
In this example T = 30. r, g and responsiveness are assumed to be fixed in time and equal to the empirical average observation from the year 1994 (after introduction of DIT) to the year 2013 for each country. β is the coefficient estimated in Section A and is equal to 0.114.
In this hypothetical scenario, assuming no other factor is affecting income inequality and hypothesizing the constancy over time of r i , g i and of the degree of responsiveness ϵ i , we find that inequality in Norway grows by 33%. Inequality in Denmark would only slightly increase (0.64%), while inequality in Finland would witness a major increase (62%). For United States inequality would instead slightly decrease. These simulated paths depend both on the difference between r i and g i ,[25] in line with Piketty (2014), and on the responsiveness of the Gini coefficient with respect to changes in the capital share.
In the final three simulations in Table 13, we assume that the responsiveness is equal to the β 1 coefficient found in the regression exercise in Section A and simulate the same forecasted path. Assuming this common degree of responsiveness would underestimate the resulting inequality in all Nordic countries considered here, confirming our statement that having country specific degrees of responsiveness, as those estimated through the income composition inequality approach, is more informative.
We do not claim that this simulation exercise can be used for forecasting future inequality levels in the Nordic countries, since, in reality, there are numerous confounding factors that affect the evolution of personal income inequality. Furthermore, we are not taking into account multiple sources of endogeneity in the evolution of the different variables. However, we believe that this stylized exercise with alternative scenarios helps to show how the degree of income composition inequality is a key statistic that should be taken into consideration for policy aimed at reducing personal income inequality.
C Growth Incidence Curves
As a complementary evidence for the evolution in capital and labor incomes across the total factor income distribution in LIS data we present growth incidence curves for each of the four countries under analysis. Figure 19 plots the factor income incidence curve for Denmark, Finland, Norway and Sweden respectively, capturing the percentage change in capital and labor income across the total income rank in between the first period available (1979–1987) and the last one (2005–2016).
Factor income incidence curves.
These figures plot country-by-country the change (%) in the factor income shares, for each decile of the total income distribution.
In line with the analysis in the main text, it is visible how the very top experienced surges in their capital incomes, while the rest of the distribution witnessed relatively minor changes in their labor and capital incomes in percentage terms.
D Decomposing Inequality: Varying Weights Across Income Groups
In this section, we add brief evidence on generalized entropy measures, to complement the analysis in the main text of the paper, which relies mostly on the Gini as the only aggregate measure of inequality. Table 14 presents, for each country and for each year, several entropy measures, attributing heterogeneous weights to different income groups. Assigning a higher weight to the bottom of the total income distribution results in large changes in the corresponding entropy measure, GE(−1), with overall increasing trends in all countries. This implies that reductions in income shares at the bottom are a key contributor to variations in inequality measures in the period. An overall increasing trend is confirmed also when considering the other entropy measures, with GE(1) (Theil index) and GE(2) assigning higher weights to upper parts of the distribution. The only exception is GE(0) for Denmark, which exhibits an overall flat trend.
Generalized entropy measures.
| Country | Year | GE(−1) | GE(0) | GE(1) | GE(2) |
|---|---|---|---|---|---|
| Denmark | 1992 | 937.920 94 | 1.133 58 | 0.394 43 | 0.404 06 |
| 1995 | 165 000 | 1.184 42 | 0.396 20 | 0.561 24 | |
| 2000 | 213 000 | 1.124 40 | 0.388 67 | 0.570 77 | |
| 2004 | 169 000 | 1.069 41 | 0.395 04 | 1.466 35 | |
| 2007 | 170 000 | 1.172 20 | 0.429 61 | 0.714 50 | |
| 2010 | 268 000 | 1.197 75 | 0.457 47 | 1.128 53 | |
| 2013 | 235 000 | 1.198 56 | 0.477 94 | 2.167 37 | |
| 2016 | 172 000 | 1.057 47 | 0.512 55 | 9.779 62 | |
| Finland | 1987 | 37.539 11 | 0.618 48 | 0.282 14 | 0.278 21 |
| 1991 | 26.140 33 | 0.587 95 | 0.285 86 | 0.282 28 | |
| 1995 | 28.134 90 | 0.673 44 | 0.341 00 | 0.382 58 | |
| 2000 | 42.676 95 | 0.705 07 | 0.424 08 | 1.150 51 | |
| 2004 | 94.159 84 | 0.912 96 | 0.452 66 | 0.960 30 | |
| 2007 | 137.448 32 | 0.968 49 | 0.475 84 | 1.176 14 | |
| 2010 | 180.167 77 | 1.063 38 | 0.455 90 | 0.660 50 | |
| 2013 | 228.362 79 | 1.132 37 | 0.459 64 | 0.725 23 | |
| 2016 | 228.776 54 | 1.164 23 | 0.469 57 | 0.621 82 | |
| Norway | 1979 | 3.043 39 | 0.358 64 | 0.209 92 | 0.291 37 |
| 1986 | 6.610 21 | 0.382 78 | 0.213 78 | 0.201 32 | |
| 1991 | 29.579 00 | 0.587 36 | 0.297 44 | 0.301 06 | |
| 1995 | 370.571 72 | 0.890 46 | 0.372 09 | 0.432 01 | |
| 2000 | 623.339 20 | 0.938 74 | 0.448 73 | 0.833 79 | |
| 2004 | 172 000 | 1.158 42 | 0.543 68 | 2.366 86 | |
| 2007 | 944.949 99 | 0.950 97 | 0.431 40 | 0.900 56 | |
| 2010 | 139 000 | 1.050 47 | 0.455 23 | 1.048 17 | |
| 2013 | 169 000 | 1.025 53 | 0.427 12 | 0.628 33 | |
| Sweden | 1975 | 6.164 99 | 0.470 23 | 0.264 00 | 0.271 64 |
| 1981 | 6.421 81 | 0.543 02 | 0.296 50 | 0.280 82 | |
| 1987 | 11.070 54 | 0.680 04 | 0.355 94 | 0.370 74 | |
| 1992 | 30.117 60 | 0.800 90 | 0.392 00 | 0.381 81 | |
| 1995 | 151.054 16 | 0.559 97 | 0.299 60 | 0.302 07 | |
| 2000 | 162.150 71 | 0.901 34 | 0.576 69 | 2.433 31 | |
| 2005 | 458.892 99 | 0.900 35 | 0.472 76 | 5.530 80 |
-
The table shows different general entropy measures of inequality, GE(a). The higher the a, the more sensitive becomes GE(a) to differences at the top of the income distribution. Vice versa, the lower (and negative) the a, the more sensitive becomes GE(a) to income differences at the bottom of the distribution. GE(1) corresponds to the Theil index.
References
Aaberge, R., and A. B. Atkinson. 2010. “Top Incomes in Norway.” In Top Incomes: A Global Perspective. Oxford: Oxford University Press.Search in Google Scholar
Aaberge, R., C. André, A. Boschini, L. Calmfors, K. Gunnarsson, M. Hermansen, and J. Søgaard. 2018a. Increasing Income Inequality in the Nordics. Copenhagen: Nordic Council of Ministers.Search in Google Scholar
Aaberge, R., A. B. Atkinson, and S. Königs. 2018b. “From Classes to Copulas: Wages, Capital, and Top Incomes.” The Journal of Economic Inequality 16 (2): 295–320. https://doi.org/10.1007/s10888-018-9386-x.Search in Google Scholar
Aaberge, R., A. B. Atkinson, and J. Modalsli. 2020. “Estimating Long-Run Income Inequality from Mixed Tabular Data: Empirical Evidence from Norway, 1875–2017.” Journal of Public Economics 187: 104196. https://doi.org/10.1016/j.jpubeco.2020.104196.Search in Google Scholar
Albarran, P., R. Carrasco, and M. Martinez-Granado. 2009. “Inequality for Wage Earners and Self-Employed: Evidence from Panel Data.” Oxford Bulletin of Economics & Statistics 71 (4): 491–518. https://doi.org/10.1111/j.1468-0084.2009.00549.x.Search in Google Scholar
Alstadsaeter, A. 2007. “The Achilles Heel of the Dual Income Tax: The Norwegian Case.” Finnish Economic Papers 20 (1): 5–22.Search in Google Scholar
Alvaredo, F., L. Chancel, T. Piketty, E. Saez, and G. Zucman. 2018. World Inequality Report 2018. Cambridge, MA: Harvard University Press.10.4159/9780674984769Search in Google Scholar
Atkinson, A. B. 2009. “Factor Shares: the Principal Problem of Political Economy?” Oxford Review of Economic Policy 25 (1): 3–16. https://doi.org/10.1093/oxrep/grp007.Search in Google Scholar
Atkinson, A. B., and J. E. Søgaard. 2016. “The Long-Run History of Income Inequality in Denmark.” The Scandinavian Journal of Economics 118 (2): 264–91. https://doi.org/10.1111/sjoe.12143.Search in Google Scholar
Barth, E., K. Moene, and F. Willumsen. 2015. “The Scandinavian Model: An Interpretation.” Journal of Public Economics 117: 60–72.10.1016/j.jpubeco.2014.04.001Search in Google Scholar
Bengtsson, E. 2014. “Labour’s Share in Twentieth-Century Sweden: A Reinterpretation.” Scandinavian Economic History Review 62 (3): 290–314. https://doi.org/10.1080/03585522.2014.932837.Search in Google Scholar
Björklund, A., and M. Jantti. 1997. “Intergenerational Income Mobility in Sweden Compared to the United States.” The American Economic Review 87 (5): 1009–818.Search in Google Scholar
Brewer, M., and L. Wren-Lewis. 2016. “Accounting for Changes in Income Inequality: Decomposition Analyses for the UK, 1978–2008.” Oxford Bulletin of Economics & Statistics 78 (3): 289–322. https://doi.org/10.1111/obes.12113.Search in Google Scholar
Chancel, L. 2021. “Ten Facts about Inequality in Advanced Economies.” In Combating Inequality: Rethinking Government’ s Role. Cambridge, MA: The MIT Press.10.7551/mitpress/13469.003.0003Search in Google Scholar
Domeij, D., and M. Flodén. 2010. “Inequality Trends in Sweden 1978-2004.” Review of Economic Dynamics 113 (1): 179–208. https://doi.org/10.1016/j.red.2009.10.005.Search in Google Scholar
Eriksson, T., and M. Jäntti. 1997. “The Distribution of Earnings in Finland 1971-1990.” European Economic Review 41 (9): 1763–79. https://doi.org/10.1016/S0014-2921(96)00044-X.Search in Google Scholar
Fiorio, C. V. 2011. “Understanding Italian Inequality Trends.” Oxford Bulletin of Economics & Statistics 73 (2): 255–75. https://doi.org/10.1111/j.1468-0084.2010.00606.x.Search in Google Scholar
Fochesato, M., and S. Bowles. 2015. “Nordic Exceptionalism? Social Democratic Egalitarianism in World-Historic Perspective.” Journal of Public Economics 127: 30–44. https://doi.org/10.1016/j.jpubeco.2014.03.001.Search in Google Scholar
García-Peñalosa, C., and E. Orgiazzi. 2013. “Factor Components of Inequality: A Cross-Country Study.” Review of Income and Wealth 59 (4): 689–727. https://doi.org/10.1111/roiw.12054.Search in Google Scholar
Glyn, A. 2011. “Functional Distribution and Inequality.” In The Oxford Handbook of Economic Inequality. Oxford: Oxford University Press.10.1093/oxfordhb/9780199606061.013.0005Search in Google Scholar
Iacono, R. 2018. “The Nordic Model of Economic Development and Welfare: Recent Developments and Future Prospects.” Intereconomics: Rev. Eur. Eco. Policy 53 (4): 185–90. https://doi.org/10.1007/s10272-018-0747-2.Search in Google Scholar
Iacono, R., and M. Ranaldi. 2022. “The Evolution of Income Composition Inequality in Italy, 1989–2016.” Review of Income and Wealth, https://doi.org/10.1111/roiw.12555.Search in Google Scholar
Jäntti, M., M. Riihelä, R. Sullström, and M. Tuomala. 2010. “Trends in Top Income Shares in Finland.” In Top Incomes: A Global Perspective. New York: Oxford University Press.Search in Google Scholar
Jaumotte, F., and C. Osorio Buitron. 2020. “Inequality: Traditional Drivers and the Role of Union Power.” Oxford Economic Papers 72 (1): 25–58. https://doi.org/10.1093/oep/gpz024.Search in Google Scholar
Jordà, Ò., K. Knoll, D. Kuvshinov, M. Schularick, and A. M. Taylor. 2019. “The Rate of Return on Everything, 1870–2015.” Quarterly Journal of Economics 134 (3): 1225–98. https://doi.org/10.1093/qje/qjz012.Search in Google Scholar
Kakwani, N. C. 1977a. “Applications of Lorenz Curves in Economic Analysis.” Econometrica 45 (3): 719–27. https://doi.org/10.2307/1911684.Search in Google Scholar
Kakwani, N. C. 1977b. “Measurement of Tax Progressivity: An International Comparison.” The Economic Journal 87 (345): 71–80. https://doi.org/10.2307/2231833.Search in Google Scholar
Karabarbounis, L., and B. Neiman. 2014. “The Global Decline of the Labor Share.” Quarterly Journal of Economics 2129 (1): 61–103. https://doi.org/10.1093/qje/qjt032.Search in Google Scholar
Lange, K., L. H. Pedersen, and P. B. Sørensen. 1999. “The Danish Tax Reform Act of 1993: Effects on the Macroeconomy and on Intergenerational Welfare.” In Macroeconomic Perspectives on the Danish Economy, pp. 134–64. UK: Palgrave Macmillan.10.1057/9780230501034_5Search in Google Scholar
LIS 2020. Luxembourg Income Study (LIS) Database. Luxembourg: LIS. Also available at http://www.lisdatacenter.org (multiple countries; January 2020 - April 2020).Search in Google Scholar
Milanovic, B. 2017. “Increasing Capital Income Share and its Effect on Personal Income Inequality.” In After Piketty: The Agenda for Economics and Inequality, edited by H. Boushey, B. de Long, and M. Steinbaum, 235–9. Cambridge, MA: Harvard University Press.10.4159/9780674978195-011Search in Google Scholar
Norway, S. 2020. Property Income. Also available at https://www.ssb.no/a/metadata/conceptvariable/vardok/10/en.Search in Google Scholar
Piketty, T. 2014. Capital in the 21st Century. Cambridge, MA: Harvard University Press.10.4159/9780674369542Search in Google Scholar
Piketty, T., and G. Zucman. 2014. “Capital is Back: Wealth-Income Ratios in Rich Countries 1700–2010.” Quarterly Journal of Economics 8129 (3): 1255–310. https://doi.org/10.1093/qje/qju018.Search in Google Scholar
Ranaldi, M. 2021. “Income Composition Inequality.” Review of Income and Wealth, https://doi.org/10.1111/roiw.12503.Search in Google Scholar
Ranaldi, M., and B. Milanovic. 2021. “Capitalist Systems and Income Inequality.” Journal of Comparative Economics, https://doi.org/10.1016/j.jce.2021.07.005.Search in Google Scholar
Rao, V. M. 1969. “Two Decompositions of Concentration Ratio.” Journal of the Royal Statistical Society: Series A 132 (3): 418–25. https://doi.org/10.2307/2344120.Search in Google Scholar
Roine, J., and D. Waldenström. 2008. “The Evolution of Top Incomes in an Egalitarian Society: Sweden, 1903-2004.” Journal of Public Economics 292 (1–2): 366–87. https://doi.org/10.1016/j.jpubeco.2007.06.012.Search in Google Scholar
Roine, J., and D. Waldenström. 2012. “On the Role of Capital Gains in Swedish Income Inequality.” Review of Income and Wealth 958 (3): 569–87. https://doi.org/10.1111/j.1475-4991.2011.00475.x.Search in Google Scholar
Roine, J., and D. Waldenström. 2015. “Long-run Trends in the Distribution of Income and Wealth.” In Handbook of Income Distribution, vol. 2, 469–592. Elsevier.10.1016/B978-0-444-59428-0.00008-4Search in Google Scholar
Roine, J., J. Vlachos, and D. Waldenström. 2009. “The Long-Run Determinants of Inequality: What Can We Learn from Top Income Data?” Journal of Public Economics 93 (7–8): 974–88. https://doi.org/10.1016/j.jpubeco.2009.04.003.Search in Google Scholar
Rubolino, E., and D. Waldenström. 2020. “Tax Progressivity and Top Incomes: Evidence from Tax Reforms.” The Journal of Economic Inequality 18: 261–89. https://doi.org/10.1007/s10888-020-09445-8.Search in Google Scholar
Shorrocks, A. F. 1982. “Inequality Decomposition by Factor Components.” Econometrica 50 (1): 193–211. https://doi.org/10.2307/1912537.Search in Google Scholar
Søgaard, J. E. 2018. “Top Incomes in Scandinavia: Recent Developments and the Role of Capital Income.” In Increasing Income Inequality in the Nordics, edited by R. Aaberge. Copenhagen: Nordic Council of Ministers.10.6027/a8f75d4b-enSearch in Google Scholar
Sørensen, P. B. 1994. “From the Global Income Tax to the Dual Income Tax: Recent Tax Reforms in the Nordic Countries.” International Tax and Public Finance 21 (1): 57–79. https://doi.org/10.1007/BF00874089.Search in Google Scholar
Thoresen, T. O. 2004. “Reduced Tax Progressivity in Norway in the Nineties: The Effect from Tax Changes.” In International Tax and Public Finance, vol. 811, pp. 487–506.10.1023/B:ITAX.0000033989.32569.5eSearch in Google Scholar
WID.World 2020. World Inequality Database. https://wid.world/data.Search in Google Scholar
© 2022 Roberto Iacono et al., published by De Gruyter, Berlin/Boston
This work is licensed under the Creative Commons Attribution 4.0 International License.
