Abstract:
Savings are, apart from inheritances and transfers, the corner stone for the accumulation of wealth. Against the background of rising economic inequality in industrialized countries and the ongoing assessment of its root causes, analyses of the distribution of savings along the income and wealth distribution are of high interest for the question on whether mutual stimulation between income flows and wealth stocks contributes to rising inequality. We analyze the extent of the concentration of household savings in Germany by estimating saving amounts, saving rates and shares in aggregate savings for different classes of household income and household wealth in Germany. Our calculations are based on the Sample Survey of Household Income and Expenditure (in German: Einkommens- und Verbrauchsstichprobe - EVS), a large sample containing more than 40,000 households in Germany. We show that the concentration of savings in Germany is substantial, as in 2013 the top income decile's share in aggregate savings amounts to about 60 percent, whereas the lower half of the income distribution actually does not save at all. Conditional on the distribution of wealth the concentration of savings is somewhat less pronounced, but still apparent. Over the years 2003 till 2013 we find an increase of the concentration of household savings across the income and wealth distribution. Finally, based on a set of assumptions, we look beyond the top income threshold underlying the EVS dataset (18,000 euros of monthly net household income) in order to estimate bias - corrected saving rates for the top income groups which are considerably higher than those that can be calculated with our data set alone. Using these corrected saving rates as input parameters for a macro simulation of the distribution of household incomes and savings we find that the aggregate saving rate increases by two to three percentage points compared to the estimate based on EVS data alone. Also, the top decile and percentile groups' shares in aggregate savings are substantially higher compared to the estimates solely based on EVS data.