This post is taken from a longer study on the state of inequality in Germany commissioned by Forum New Economy. The full study by Charlotte Bartels (DIW) and Carsten Schröder (DIW) is titled “Income, consumption and wealth inequality in Germany: Three concepts, three stories?”. See also – the overview on methodological choices in research on inequality.
1. The distribution of income
Typically, world capital incomes grew more rapidly than labor incomes in many countries. The labor income share in national income fell as the functional income distribution shifted (Karabarbounis and Neimann, 2014). In Germany, capital income grew by a factor of 12.5 since the 1950s, while labor income grew by a factor of 7, according to the figures of the Federal Statistical Office. During the years of the Wirtschaftswunder, both capital and labor income grew at similar rates. Since the 1980s, capital income increased at much faster rate than labor income.
The increased importance of capital income usually leads to higher income inequality, because capital incomes are more concentrated than labor incomes. Capital, i.e. financial assets, real estate, and firms, is only held by a small share of the population, whereas the majority of the population depends on labor income to make a living. In Germany, the top decile of the income distribution received about 30 percent of national income in the 1960s. This income share quite steadily increased to about 40 percent today. At the same time, the income share of the bottom 50% decreased from more than 30 percent in the 1960s to less than 20 percent today (Bartels, 2019). However, increasing income inequality is not only linked to the shift in the functional income distribution between labor and capital, but also to increasing labor income inequality.
Since the 1980s, the distribution of labor income itself became increasingly unequal in Germany. Using IABS data, which is constructed from a 2 percent sample of social security records, Dustmann et al. (2009) show that wages have become increasingly unequal since the 1980s, when the difference between top and medium earners started to increase. In the 1990s, wage inequality at the bottom of the wage distribution also started to rise. Grabka and Schroeder (2018) assess trends in the inequality of hourly wages as well as monthly and annual earnings of dependent employees using SOEP data since the 1990s. In the following, we elaborate on their results on wage and earnings inequality measured by the 90:10 ratio, which is resistant to outliers and indicates the income of an earner at the 90th percentile (richer than 90 percent of earners) relative to one at the 10th percentile.
The 90:10 ratio of real hourly wages grew from 3.2 in 1994 to 3.9 in 2005. Real hourly wages of the ten percent poorest wage earners even decreased by more than 25 percent between 1998 and 2006. However, overall, this ratio remained quite stable over time. According to Dustmann et al. (2009) and Biewen and Seckler (2017), de-unionization plays a major role in exacerbating wage inequality, especially at the lower end of the distribution. Other papers highlight firm-specific wage differences (Antonczyk et al. 2010, Ohlert 2016) as well as personal characteristics or demographic changes (Glitz and Wissmann, 2017). Since the introduction of a minimum wage in 2015, inequality of hourly wages declined, and the 90:10 ratio declined to 3.5 in 2018 (Fedorets et al., 2020).
In contrast to the moderate increase of wage inequality, monthly and annual earnings inequality rose more sharply, as shown by Grabka and Schroeder (2018). Monthly earnings of the poorest ten percent of earners even halved between 1992 and 2005, whilst the richest ten percent improved their monthly and annual earnings by 20%. Subsequently, the 90:10 ratio, grew from four in 1992 to ten in 2006, oscillating near eleven today. The 90:10 ratio of annual earnings increased from eight in 1998, peaking at fifteen in 2010.
How can earnings inequality increase far more than wage inequality? Earnings are the product of worked hours and hourly wage. Indeed, from 1992 to 2012, the 10th decile increased the number of hours worked by ten percent, whilst the 1st decile works 25% hours less on average. Over the same period, the number of mini-jobs, which limit monthly income at 450€, grew from three million to seven and a half million. Additionally, Biewen et al. (2019) find negative selection of low-wage households into part time jobs, leading to an increase of earnings inequality. Nevertheless, Biewen and Plötze (2019) could only explain parts of the increase in the 90:10 earnings ratio (10% for men, 55% for women) with work-hour changes and part-time jobs.
These increased earnings inequalities accumulate over a lifetime such that lifetime earnings inequality has also increased. The inequality of lifetime earnings among cohorts born in the 1960s and 1970s is almost twice as high as for cohorts born in the 1930s (Bönke et al., 2015).
Finally, government taxes and transfers did not mitigate increased market income inequalities in Germany. Even though real equivalent disposable income of persons in private households in Germany rose by 15 percent on average between 1991 and 2015, the lower end of the income distribution did not benefit from the growth in real income. The Gini index of disposable household income remained rather stable between 1991 and 1999, then rose from 0.25 in 1999 to 0.29 in 2005. Unlike inequality in market income, inequality in disposable household income regressed only slightly between 2005 and 2009. After 2009, inequality in disposable household increased moderately until 2013 and stagnated since then (Goebel and Grabka, 2018; Goebel and Grabka, 2020).
2. The distribution of expenditures and consumption
Empirical studies on the distribution of household consumption are scarce. Most likely this is because of high data requirements: Ideally, information should be available for all kinds of consumption– from food to durable goods. Many surveys simply do not meet this requirement. Second, expenditures do not equate with consumption if households use certain commodities over a longer period. For such durable commodities, expenditures made at a particular point in time do not reflect the household’s current consumption of the same good. Instead, like a depreciation of durables in the books of a company, expenditures must be converted into a consumption flow over several periods (Krueger and Perri, 2006).
Although many studies focus on the distribution of income, consumption-based measures may be advantageous for conceptual reasons. As explained above, according to the permanent income hypothesis, current income has a permanent and a transitory component, with people basing consumption decisions on their permanent income. Thus transitory changes should cause current income to vary more than consumption, but the higher variation does not necessarily reflect equally large living standard variations (see Meyer and Sullivan, 2003, and references cited therein). According to the life-cycle hypothesis, current income rises throughout working ages and then declines around retirement. If people try to smooth marginal utility, then they will save less when they are young and old, but more in the middle of their lifetime. Combined, the two hypotheses imply that, relative to current income, savings rates should exhibit an inverse U-shaped pattern with respect to age. At the same time, however, households may align savings with interest rates. When different birth cohorts face different interest rate histories (at different ages) this may map into differences of income and consumption profiles.
These patterns of income and consumption have implications for the measurement of inequality. First, consumption smoothing implies that income inequality should be higher than consumption inequality, a conjecture that finds empirical support in several studies including Krueger and Perri (2006), Pendakur (1998), Attanassio et al. (2006), Johnson et al. (2005). Second, consumption smoothing in combination with cumulative effects of (bad) luck should result in rising consumption inequality over the lifecycle. Indeed, Deaton and Paxson (1994) and Storesletten et al. (2004) provide affirmative empirical evidence. Third, it is not ruled out that income and consumption distributions exhibit different intertemporal patterns, as documented, for example, in the empirical study of Meyer and Sullivan (2007).
For Germany, a few case studies estimate inequalities from consumption or consumption expenditures. All the studies we are aware of (e.g., Noll et al., 2007, Zaidi und de Vos, 2001, Fuchs-Schündeln et al., 2010, Hufe et al., 2018) rely on the Income and Expenditure Survey for Germany (Einkommens- und Verbrauchsstichprobe), provided by the German Statistical Office. All the cited studies focus on consumption expenditure and do not derive consumption flows.
For example, Hufe et al. (2018) study the 1993 to 2013 period, finding rising inequalities for all investigated concepts of material well-being: equivalent disposable income, market income, and consumption expenditures. Over this period, the Gini coefficient rises from about 0.27 to 0.29 for disposable income, from 0.37 to 0.39 for market income, and from 0.38 to 0.41 for consumption expenditures. At first glance, high consumption inequalities are at odds with the above argumentation, i.e., that consumption smoothing should lead to lower consumption inequalities. Notice, however, that the authors explore consumption expenditures. For expenditures, higher inequalities are not surprising because purchases of many durable goods involve high expenditures.
3. The distribution of wealth
Estimations of the distribution of wealth based on survey data tend to underestimate wealth inequality, as the wealthiest people and households are hardly ever included in their samples. This is due to two reasons: First, it is statistically unlikely to draw a sample that includes multi-millionaires or even billionaires. Second, these people often do not respond to surveys. Further, as the wealth tax was suspended in 1995, there is no administrative data that can be used to gain insights on the wealth distribution, thus survey data is crucial. We will review the literature with respect to the Gini coefficient, top wealth shares as well as the share of persons and households without significant wealth. In accordance with most research, we do not count the net present value of future pension claims as wealth.
Grabka and Halbmeier (2019), analyzing SOEP data, find that, in 2017, the wealthiest 10% of people own 56% of the total wealth (excluding motor vehicles), whilst the less-owning half of people only owns 1.3% of total wealth. This distribution corresponds to a Gini of 0.78 and has been relatively stable since 2004. The share of people with zero or negative wealth is 29% and increasing slightly since 2002 (27.6%).
Using the Income- and Expenditure Survey (EVS), Fuchs-Schündeln et al. (2010) find that the Gini of wealth on household level was 0.63 in 1978, rising to 0.68 in 1997. Over this period, the share of households with zero or negative wealth increased from 6.5% to 10.5%.
Another survey on the household-level is the Household Finance and Consumption Survey (HFCS) conducted by the national central banks, which allows international comparisons. According to the HFCS, the Gini of wealth in Germany was around 0.76 in both 2010 and 2014, with the wealthiest 10% of households owning 59.8% of total wealth. The HFCS additionally provides three other inequality measures, the Atkinson index, the Theil index, and the Pietra index. For both survey waves and all inequality measures, Germany exhibits the highest- or second-highest wealth inequality of all EU members.
Albers et al. (2020) combine wealth tax data, surveys, national accounts and rich lists to study the distribution of wealth in Germany from 1895 to 2018. They show that the wealth share of the top 1% has fallen by half, from close to 50% in 1895 to less than 25% today. Between 1993 and 2018, wealth of the top 10% and of the middle class (50-90%) has approximately doubled in real terms, while wealth of the bottom half has stagnated and their share in total wealth has nearly halved. Their estimated Gini coefficient based on a combination of EVS, national accounts and rich lists increases from 0.69 in 1993 to 0.74 in 2018.
4. The Case of Germany
This focus on Germany is because, like in many other countries, inequality is back on the academic agenda and in public debates. As in many other countries, empirical analyses document that in Germany, following a long period of stability, there are high and growing inequalities in income and wealth (e.g., OECD 2008, Bönke et al., 2019, Grabka and Schröder, 2018 and 2019). That inequalities are rising, however, is not uncontested in the literature. This is because inequality has many dimensions. For example, while the literature focuses on the development of inequalities within national borders, others broaden the perspective and study inequalities in transnational spatial or political entities (Milanovic, 2012 and 2019). Thus the measurement of inequalities has expanded, no longer just measuring how inequality evolves within the borders of a country, but also how differences in living standards evolve between countries. Of course, also the time horizon matters: Do we assess the development of inequality around the financial crisis (Bönke and Schröder, 2014), around German reunification (Fuchs-Schündeln et al., 2010), or since the 19th century (Bartels, 2019)?
Of course, also the choice of the focal variable matters. Much of the current literature focuses on income inequalities, as income (from labor), is essential for social and economic participation in a market economy. Another important material component is wealth. Wealth inequalities are much higher than income inequalities (Fuchs-Schündeln et al., 2010). However, due to data restrictions, wealth inequalities receive less attention in the literature. For conceptual reasons, inequalities in consumption may be more informative than inequalities in income or wealth: According to the permanent income hypothesis, current income has permanent and transitory components, with people basing consumption decisions on their permanent income. Thus transitory changes should cause current income to vary more than consumption, but the higher variation does not necessarily reflect equally large living standard variations (see Meyer and Sullivan, 2003, and references cited therein). Therefore, we use EVS data that allow us to study to what extent wealth and income inequalities translate into consumption inequality in Germany.
About the authors: Charlotte Bartels is a post-doctoral researcher at the German Institute for Economic Research (DIW), Carsten Schröder is Vice Director SOEP and Division Head Applied Panel Analysis in the German Socio-Economic Panel study
Read more on – the overview on methodological choices in research on inequality.
 Equivalent incomes (consumption) adjust for differences in needs. Hufe et al. (2018) make use of the OECD modified equivalence scale.