Chapter 4. The distribution of asset holdings

This chapter analyses patterns of asset holdings in selected OECD countries across both income and wealth distributions. The analysis draws on income and asset-holdings microdata for 18 countries from the Eurosystem Household Finance and Consumption Survey (HFCS). Patterns of asset holdings are found to vary substantially across both income and wealth distributions, with significant implications for the distributional effects of the taxation of household savings.

4.1. Introduction

A key consideration regarding the taxation of household savings is its impact on income and wealth inequality. This chapter provides an evidential base for examining the distributional effects of the taxation of household savings by analysing patterns of asset holdings in selected OECD countries across both income and wealth distributions. Detailed discussion of the tax policy implications of these patterns of asset holdings is left until Chapter 6.

The analysis draws on income and asset-holdings microdata for 18 countries from the second wave of the Eurosystem Household Finance and Consumption Survey (HFCS).1 The HFCS is a survey carried out by the European Central Bank and national central banks in the European Union to gather micro-level information on households’ assets and liabilities (European Central Bank, 2016). Further details on the HFCS are provided in Box 4.1.

Box 4.1. The Eurosystem Household Finance and Consumption Survey

Data source

This chapter draws on microdata from the second wave of the Eurosystem Household Finance and Consumption Survey (HFCS). The HFCS provides household-level data on income, asset holdings and consumption (together with various demographic variables) for 20 European Union countries following a harmonised approach to facilitate cross-country comparability of results. It provides a sufficiently detailed breakdown of household wealth across different assets to enable a detailed analysis of patterns of household wealth.

The HFCS includes 18 OECD member countries, which are the focus of the analysis in this chapter: Austria, Belgium, Estonia, Finland, France, Germany, Greece, Hungary, Ireland, Italy, Latvia, Luxembourg, the Netherlands, Poland, Portugal, the Slovak Republic, Slovenia and Spain.

Asset holding data generally relates to the time of the underlying household interviews which occurred between March 2013 and January 2015. Income data typically relates to the year preceding the interview (most commonly 2013, with interviews having occurred in 2014). The main exception is Spain for which interviews occurred between October 2011 and April 2012, and income data relates to 2010.1

In total, the HFCS covers more than 84 000 households. As it is household data, the HFCS does not focus on the asset distribution at the individual level. However, the amount of wealth that a household can accumulate can depend heavily on whether a household is a single-earner household or a household with several earners. Sample sizes range between countries (from 1 202 households in Latvia to 12 035 households in France). In the majority of countries, the surveys are stratified to oversample wealthier households.

To account for missing variables, Ireland, Italy and Finland adopt single imputation. All other countries adopt multiple imputation. Multiple imputation is used to handle missing data in a way that increases the accuracy of standard errors as compared to single imputation by taking account of the uncertainty regarding the imputed values. Multiple imputation works by imputing five versions of any missing data point in the dataset (following the same imputation procedure each time). This means that the final dataset effectively contains five complete datasets (one with each of the five imputed values, but with the exact same values for all other non-missing data). Final parameter estimates are the mean of the estimates from the underlying analysis undertaken separately on the five datasets.

In this study, results are presented across both gross income and net wealth deciles. Neither are equivalised. Within both the income and net wealth distributions, results are presented for a variety of financial and non-financial assets. These are outlined in Table 4.2.

Table 4.2. Main Asset Categories in the Household Finance and Consumption Survey
Real Assets	Financial Assets
Value of household’s main residence	Deposits in current and chequing accounts; deposits in saving accounts
Value of other real estate property	Bonds
Value of household’s vehicles	Publicly traded shares
Value of household’s other valuables	Mutual funds and managed accounts
Value of self-employed businesses	Occupational and private pensions
	Other assets (including value of money owed to households and value of non self-employment private business)
Note: Further detail is provided in ECB (2016).
Source: ECB (2016).

Data limitations

It is important to bear in mind the limitations of the results outlined in this chapter. First, only 18 European countries are covered as compared to the 40 countries considered in the effective tax rate (ETR) analysis contained elsewhere in this study. The results for these countries may not be representative of all OECD and G20 countries.

The results may differ from the true distribution of income and wealth inequality due to the problems with estimating income inequality with survey data. This kind of data is usually biased downwards at the top of the income distribution. One reason for this bias is that household survey samples may not contain households from the top 1%, 0.1% or even 0.01% of the income distribution. Research using other data sources such as tax return data suggests that income and wealth are concentrated in these very high components of the income distribution, and so estimates of income and wealth inequality may be biased downwards due to the omission of these high-wealth and high-income households. These data issues mean that the results should be interpreted with caution.

A further reason why survey data estimates of asset holdings across income and net wealth distributions should be interpreted with caution concerns the fact that the results in these data only account for income and wealth that is declared by households to statistical agencies. These data may be biased due to poor recollection on the part of survey respondents, lack of effort in reporting accurately, or unwillingness to report accurate figures due to fear of non-confidentiality.

Individuals may also underreport income and wealth that they conceal from public authorities. There is a large and growing literature documenting tax evasion efforts by high income and high wealth households, some of which suggests that a larger percentage of total income and wealth go unreported at the top of the income and net wealth distribution than at the bottom (Zucman, 2013, 2014). This suggests that when concealed wealth and income is taken into account, income and wealth inequality is even greater than the estimates in this study may suggest. This may be particularly the case with respect to financial assets and the income from them, as these kinds of assets are particularly mobile and so may be easily moved offshore in an attempt to evade tax. This topic is discussed further in Chapter 5. Nonetheless, these issues suggest that the results concerning income and wealth inequality should be interpreted as a lower bound of the true levels.

In the HFCS, occupational pension wealth is only presented where there is a value to an account (i.e. as in defined-contribution pension schemes). In defined-benefit pension schemes, no data is provided in the HFCS. To address this, this study takes the following approach. The value of defined-benefit occupational pensions is set, where they are present, to be equal to the mean account value of defined-contribution accounts in the corresponding income or net wealth decile for each country. Given the generosity of defined-benefit schemes relative to defined-contribution schemes, this approach means that the estimate of occupational pension wealth in this study is likely to be an underestimate and may be interpreted as a lower bound occupational pension wealth figure.

← 1. Further detail is provided in ECB (2016).

The main findings of the chapter are as follows:

Both income and net wealth inequality are substantial, although net wealth inequality is larger than income inequality. Most households at the bottom of the net wealth distribution have negative net wealth.
The income and net wealth distributions are correlated, but not perfectly so. There are many households with limited net wealth and high incomes, and vice versa.
Some asset holding patterns are the same across both the income and net wealth distributions.
- Those with high levels of both income and net wealth hold a larger share of their assets in the form of second residences, shares, bonds, self-employed businesses, and mutual and other managed funds.
- Those with low levels of both income and net wealth hold a larger share of their assets in the form of bank deposits, vehicles and valuables.
Some patterns are not the same across the income and net wealth distributions, particularly with respect to housing and housing debt.
- Those with higher levels of income hold a lower share of their assets in the form of their main residence. However, the share of the main residence in gross assets is high in the middle of the net wealth distribution, but not at the bottom or top. At the top, more wealth is held in financial assets, second residences and self-employed businesses. At the bottom, more wealth is held in bank deposits and vehicles.
- The share of pensions in total assets is rising with income, though it is roughly constant with respect to net wealth.
- Absolute debt levels are rising proportionately with income, but are roughly constant with net wealth. Most household debt is housing debt, and most housing debt is debt with regard to the main residence.
Net wealth rises with age up to roughly the age of retirement, and broadly declines thereafter, though net wealth remains substantial amongst the elderly.
For those households with self-employment wealth, this wealth is a substantial share of their total wealth.

These findings are summarised in Table 4.1.

Table 4.1. Summary of stylised facts on asset mixes

The results present challenges for tax policymakers, and for those assessing the distributional impacts of capital taxes. Some taxes, such as those on bank deposits, bond interest income, dividend income, and capital gains on second residences, would seem to have straightforward distributional consequences. However, other policies such as property taxes, capital gains on main residences, and mortgage interest deductibility, would seem to have differing distributional consequences across the income and net wealth distributions. This is particularly challenging given the fact that the main residence makes up a large share of household assets in most of the countries discussed. Chapter 6 discusses the policy implications of these results, drawing also on the results of other chapters in this study.

The chapter presents patterns of asset holdings across both income deciles and net wealth deciles. When asset mixes are considered, the results are presented as a share of gross wealth rather than net wealth. This is because in many instances negative net wealth levels would complicate the exposition of asset mixes if asset mixes were displayed as a share of net wealth. The difference between net wealth and gross wealth (i.e. household debt) also forms a key aspect of the distributional burden of savings taxation. The distribution of household debt is considered at the end of the chapter.

The chapter proceeds as follows. Section 4.2 discusses income inequality and the mix of assets across the income distribution. Section 4.3 provides a similar discussion but focusses on wealth inequality instead of income inequality. Section 4.4 discusses the importance of the lifecycle in patterns of net wealth and asset holdings. Section 4.5 discusses differences between income and net wealth, and discusses the correlation between these two distributions. It also highlights the importance of housing in the net wealth distribution and of housing debt in particular.

4.2. Income inequality and the mix of assets across the income distribution

Income inequality in the Household Finance and Consumption Survey

Income inequality in OECD countries is at its highest level for the past half century (OECD, In It Together: Why Less Inequality Benefits All, 2015). Across the OECD, the average income of the richest 10% of the population is about nine times that of the poorest 10%, up from seven times 25 years ago, although there are wide differences across OECD countries. (OECD, 2015) (OECD, 2015). Figure 4.1 shows evidence of inequality across the sample of countries in the HFCS data. Households in the top income decile earn an average of 28% of total household income in the HFCS, while households in the bottom 90% earn 3.1% of total income on average.

Income inequality varies across countries. This variation can be due to differences in market incomes (i.e. incomes before taxes and transfers) and differences in disposable income (i.e. incomes after accounting for the tax and transfer system in reducing inequality. OECD evidence suggests that the effectiveness of taxes and transfers has diminished in recent years.2 There are many facets to the distribution of income in OECD economies.

The HFCS data shows that income inequality is particularly marked in Latvia where households in the top income decile earn 36.4% of total household income, and is relatively modest in Poland where households in the top income decile earn 21.3% of total household income.

Variation in inequality stems in part from the pre-tax and post-tax labour income distribution. One way in which these levels of income inequality are impacted by tax policy is through labour taxes. Research both within and outside the OECD suggests that labour income inequality is rising (OECD, 2015; Alvaredo, Chancel et al., 2018). Labour taxes, at the same time have become less progressive over the last several decades (OECD, 2018).

Disparities in capital income and assets are another driver of income inequality. Economic research has also highlighted the extent to which increasing income inequality is also driven by capital ownership (Saez, 2017). All households own assets of different forms. These can be real assets such as cars, houses, or other valuables such as jewellery and art, or financial assets such as pensions, shares, bonds, and bank deposits. Understanding the distribution of these assets is a key issue in understanding how capital taxes impact inequality. Where capital taxes are highest on income from the kinds of assets that are held mostly by those with higher incomes, the tax system is more likely to reduce income inequality than if capital taxes are highest on income from the kinds of assets that are held mostly by those with lower incomes.

Real and financial assets

Overall, households hold a greater share of total gross wealth in the form of real assets than financial assets, although this is less pronounced for higher-income households. This can be seen in Figure 4.2, which shows the distribution of overall household gross wealth between real and financial assets across countries and income deciles. On average, households in the HFCS hold 86.6% of total gross wealth in the form of real assets compared to 13.4% as financial assets.3 Germany has the largest share of financial assets as a share of total assets (38.1% of total assets), while Latvia has the largest share of real assets (98% of total assets). Households in the top income decile hold a higher share of their gross wealth in financial assets than the average household, holding 18.1% of total gross wealth in the form of financial assets on average and 81.9% in the form of real assets on average. Financial asset holdings broadly form a larger share of total assets among those with higher levels of income and net wealth, although this is not the case for all countries.

Real asset holdings across the income distribution

Real assets form the majority of asset holdings for almost every income decile in almost every country covered by the HFCS. This means that the taxation of real assets is of crucial importance when considering the overall taxation of household savings. Figure 4.3 shows the distribution of real assets across income deciles, expressed as a share of total gross real assets. The relationship between household debt and net wealth is considered in Section 4.5.

Main residences are by far the largest single asset category for most households, particularly for lower income deciles. Figure 4.3 shows that main residences make up a smaller share of net wealth for higher income deciles than for lower income deciles (with the exception of the Netherlands). In Austria and Luxembourg for example, households in the lowest income deciles hold 79.6% and 82.5% of their gross real wealth in the form of their main residence, while households in the highest income deciles hold 31.2% and 36.4% respectively. However there are exceptions. For example, in the Netherlands the distribution of gross housing wealth across the income distribution does not vary as substantially. In the Netherlands, 82.9 % of gross real wealth held by households in the bottom income decile is held in the form of housing, while the equivalent figure is 63.0% for the top income decile. However, while the percentage of total gross wealth held in the form of the main residence is lower, the absolute amount of wealth held in this form is still likely to be higher for higher income deciles due to their higher incomes overall.

The share of gross real wealth held in the form of a self-employed business is increasing with income (see Box 4.2). Assets in this category rise from an average of 4.4% for households in the bottom income decile (with a highest value of 17% in France and 13.2% in Latvia) to an average value of 5.5% for households in the fifth decile and 19.4% for households in top decile (with highest values of 45.3% in Austria and 13.2% in Estonia). This highlights the importance of the tax treatment of the self-employed.

Box 4.2. Household assets and the self-employed

The distribution of real and financial assets is complicated for households where at least one individual is self-employed. For these individuals, a substantial amount of both net and gross wealth is held inside their business, regardless of whether the business is incorporated or not. Such assets may include financial assets such as business bank deposits, or other assets such as vehicles and equipment.

On average across all households, relatively little wealth is held inside self-employed businesses. However, for those who are self-employed, the amount held inside their businesses is substantial. Figure 4.5 shows this in more detail, dividing the HFCS sample into households where the head of household is self-employed and the overall sample. The blue bars show the average holdings of self-employment wealth across the entire population. The grey bars show the same data but for households with a self-employed head of household. In these households self-employment wealth is significantly higher than the population average.

On average, the households where the head of household is self-employed hold 33.9% of their total gross wealth inside their self-employed business, compared to 10.1% on average in the total population.

The taxation of the self-employed raises challenging issues from a tax policy perspective. Self-employment income is often taxed in a different way to other kinds of income such as wage income. Some self-employment income may not be immediately taxed under the personal income tax at all, but rather under the corporate income tax if businesses are incorporated.

Note: Data are for 2013-4 (see Box 4.1).

Source: Authors’ calculations based on Household Finance and Consumption Survey (2016).

In many countries the share of total gross real wealth held in the form of a second residence also rises with income. Gross real wealth held in the form of an additional residence rises from an average of 15.8% for households in the bottom income decile (with highest values of 38.4% in Portugal and 29.8% in Finland) to an average value of 17.7% for households in the fifth decile and 27.4% for households in the top decile (with highest values of 50.1% in Luxembourg and 41.1% in Ireland).

Financial asset holdings across the income distribution

The mix of financial assets varies substantially across countries. Figure 4.4 shows the breakdown of financial asset holdings – defined as a percentage of gross financial wealth – across income deciles for the countries in the HFCS sample. In this way it presents a more detailed breakdown of the financial assets presented in Figure 4.3.

Bank deposits (including deposit and savings or chequing accounts) are the most common form of financial asset for most households in the HFCS survey.4 This asset class makes up 60% of gross financial wealth when averaged across all income deciles. They make up a particularly large share of total assets for those at the bottom of the income distribution.

Bank deposits make up a smaller share of the asset mix for those at the top of the income distribution. They comprise 40.6% of gross financial asset holdings on average for households in the top income decile, and 63.4% of gross financial asset holdings on average for households in the lowest income decile. This could mean that high levels of taxation of interest on bank deposits could have negative distributional consequences for those with low levels of income, although it should be noted that the absolute amounts of bank deposits in these very low income deciles are very small.

4.3. Wealth inequality and the mix of assets across the wealth distribution

Wealth inequality in the Household Finance and Consumption Survey

While the distribution of incomes in society is of clear concern to policymakers, wealth inequality is another distributional dimension that is of particular interest to policymakers in the context of capital taxation. Results from the HFCS suggest that net wealth inequality is larger relative to income inequality. Figure 4.6 shows the distribution of net wealth by net wealth decile, expressed as a share of total household net wealth, for the 18 OECD countries in the HFCS. On average, households in the top decile hold 45.3% of total national household wealth, while the bottom 90% of households hold 54.7% on average. Wealth inequality is particularly marked in Latvia where households in the top decile hold 61.3% of total household wealth, and in Germany where households in the top decile hold 56.2%. Figure 4.7 ranks the value of the top deciles from Figure 4.6. Based on the HFCS data, the share of wealth held by the top decile is lowest in Poland and the Slovak Republic.

Those at the bottom of the net wealth distribution usually have negative net wealth. Households in the bottom decile have an average of -0.7% of total net wealth. In every country, the net wealth of the bottom decile in the country is negative, although in many countries the net wealth value is close to zero. Net wealth values in the lowest net wealth deciles are particularly low in the Netherlands and in Latvia, where the negative amount of net wealth is -3.1% and -2.3% of the total amount of national household wealth respectively. In the Netherlands, the second net wealth decile also has negative net wealth.

Real asset holdings across the net wealth distribution

Asset mixes across deciles vary more with respect to net wealth than with respect to income. As was shown in the discussion of assets across the income distribution, the main residence makes up a substantial share of assets. Figure 4.8 shows that on average across the countries in the HFCS sample, households’ main residence makes up a substantial share of both gross wealth (71.1% averaged across all countries and deciles). In Figure 4.8, wealth held in the main residence dominates for middle net wealth deciles in most countries, but less so for high and low deciles. The average share of the main residence in gross wealth holdings of the 4th to 6th decile is 80.7%, compared to an average of 71.1% overall. This suggests that most households in the middle of the income distribution hold a significant part of their gross wealth in the form of their main residence. The importance of the main residence for middle deciles varies from 55.1% in Austria to 91.5% in Belgium.

The share of vehicles and valuables varies widely across net wealth deciles.5 It makes up a substantial share of the gross wealth of those with low levels of net wealth overall, and a low share for those with higher levels of net wealth. The average share of vehicles amongst the lowest three deciles is 18.2% and the average share of valuables is 3.7%. On average, wealth held in vehicles and valuables makes up a small share of gross wealth for top deciles in all countries, with the highest value amongst the top decile being in the Slovak Republic at 3.7%. The highest overall value is in Austria at 13.5%. For most countries, the amount of wealth held in vehicles and valuables by the top decile is below 3%.

Main and other residences make up a large share of the total gross real wealth of those in lower deciles of the net wealth distribution. Among the lowest three deciles, this share is 65.7% of gross real wealth on average, varying between a lowest share of 34.4% in France and a highest of 85.6% in the Netherlands. Debt holdings are discussed further in Section 4.5 below.

Second residences form an increasing share of the gross wealth of those with higher net wealth levels. This share rises from a share of 14.9% of the gross real wealth of those in the bottom income decile on average to a share of 33% of the gross real wealth of those in the top income decile. These shares are highest in Ireland (60.3%) and Luxembourg (55.6%), and lowest in Poland (14.5%) and Slovenia (17.4%). This suggests that the taxation of second homes may be progressive with respect to net wealth as well as income. Other asset classes, including vehicles, valuables and wealth in the form of self-employed businesses also rise with income.

Households in top net wealth deciles also have high shares of gross wealth held in the form of self-employed businesses, and low shares of vehicles and valuables. Wealth held in self-employed business is particularly high in Slovenia, Austria and Estonia, where shares are 46.1%, 34%, and 33.3% respectively.

Financial asset holdings across the net wealth distribution

The variation in financial assets across countries and deciles is also substantial. Figure 4.9 shows this variation across the net wealth deciles. Bank deposits make up a much smaller share of the asset mix for those at the top of the net wealth distribution. They account for 42.4% of gross financial asset holdings on average for those in the highest net wealth decile. The highest share of bank deposits in the top net wealth decile is in Greece at 83.6% of total financial wealth, while the lowest share is in France at 20.3%. Households in these higher net wealth deciles are more likely to hold financial assets in other forms relative to those in lower net wealth deciles. Bank deposits comprise 52.4% of gross financial asset holdings on average for households in the lowest net wealth decile. The highest share of bank deposits in the lowest net wealth decile is in Latvia at 89.7% of total financial wealth, while the lowest share is in Ireland at 15.5%.

Variation in bank deposits could be driven in part by financial sophistication (as discussed for example by Chetty et. al., 2013). Low income and low net wealth taxpayers could choose to save in bank deposits – which tend to have relatively low returns – due to lack of financial literacy and knowledge about savings opportunities with higher returns. Risk-averse households may also perceive bank deposits to be safer than other types of financial assets.

Holdings of shares and bonds are concentrated at the top of the net wealth distribution. Households in the top net wealth decile hold, on average, 7.4% of all financial assets in the form of shares, while households in the bottom net wealth decile hold only 1.3% on average. Households in the top net wealth decile hold 5.3% of all financial assets in the form of bonds, while households in the bottom net wealth decile hold 0.4%. The average across all net wealth deciles is 1.9%.

A key source of variation in net wealth over time relates to individuals’ and households’ saving for retirement. In many countries, individuals engage in substantial private pension provision, which is a source of variation in asset holdings across wealth levels and age levels, as reflected in the HFCS dataset. Pension saving is a widely-used form of saving across the entire income distribution, and usually forms the second-largest share of financial assets held in HFCS countries, after bank deposits.

Holdings of pension wealth decline across the net wealth distribution as a share of total holdings, although only modestly. In the top net wealth decile, households hold 16.9% of all financial assets in the form of pensions, while households in the bottom net wealth decile hold 31.8%. The average across all net wealth deciles is 22.8%.

The variation in private pension wealth across countries is substantial. As shown in Figure 4.9, there are relatively high shares of private pension wealth in Ireland (59% of total gross financial wealth on average), in Germany (38.7% on average), in the Netherlands (36.5% on average), and in Belgium (36.2% on average). By contrast, shares of pension wealth in total gross financial wealth are much more modest in some lower-income OECD countries such as Greece and Estonia, where even the highest net wealth decile holds no more than 1.3% and 8% respectively (averages across net wealth deciles are 2% and 8.6% respectively). However private pension wealth is also low in some countries that have higher levels of GDP per capita such as Finland and Austria, who have 8.6% and 21% respectively of total financial wealth held in the form of pensions. These results should be caveated by the fact that not all forms of pension saving are fully accounted for in the HFCS, and in particular, defined-benefit pension schemes are not fully reflected in these results.

Within those countries where pension wealth is substantial, there is some variation in the extent to which pension wealth varies with overall net wealth. In the Netherlands, even those with low levels of net wealth have a substantial portion of their financial wealth in the form of pensions. Households in the lowest net wealth decile hold 26.7% of total gross financial wealth in the form of pensions, while households in the top et wealth decile hold 23.5%. These households in low net wealth deciles may nonetheless have very low absolute amounts of pension wealth (as their total and financial wealth is low overall). The pattern in the Netherlands contrasts with France, for example. In France, 11.7% of total gross financial wealth is held among those in the lowest net wealth decile, 13.5% in the second net wealth decile, and 39% in the second-from-top net wealth decile, and 46.7% in the top net wealth decile.

There are similarities and contrasts between the results for net wealth deciles and the results for net income deciles. Pension wealth holdings are substantial across both income and net wealth deciles in Ireland, in Germany, in France, and in the Netherlands. Bond holdings are above average in Italy across income and net wealth deciles, as are equity holdings in Finland, and mutual fund holdings in Belgium and Germany. By contrast, bond holdings in Hungary and the Netherlands are concentrated amongst high net wealth deciles, while they are spread much more evenly across income deciles.

Tax factors could also drive various savings outcomes. Households may vary the asset composition of their savings depending on how the tax system treats different categories of assets (see Chapter 1 for a discussion of the literature on this subject). This may be particularly the case with respect to financial assets, which have comparatively low switching costs (compared to residences or other immovable property). The tax treatment of different asset classes is the subject of the next chapter.

4.4. Ageing and net wealth

The within-decile distribution of income and net wealth provides a snapshot in time, however, households’ incomes and net wealth vary over time. As individuals enter the workforce, their income and net wealth are both likely to begin to rise compared to childhood. As individuals retire, their incomes often fall substantially, and their net wealth may decline as they dissave over their retirement years (although this dissaving may not apply to households with high levels of wealth). These trends may exist in each country independently of other tax and non-tax factors, but may be masked if the income and net wealth distributions are examined without taking age into account.

On average, household net wealth is highest in the years just prior to retirement. Figure 4.10 shows average and median net wealth across the age distribution. Overall, net wealth is highest amongst those from 50-59 years old, with an average net wealth level of EUR 205 501, and lowest amongst those from 20-29 years old, with an average net wealth level of EUR 68 600. The top three net wealth age groups are ages 50-59 years old, 40-49 years old and 60-69 years old, where the latter two groups have average net wealth levels of EUR 158 082 and EUR 222 938 respectively.

Median wealth across age groups is lower than average wealth across age group, which stems from the relatively high concentration of wealth at the top of the net wealth distribution. However, the stylised patterns across ages are similar to the patterns with respect to average wealth levels across age groups. Overall, net wealth is highest amongst those from 60-69 years old, with an median net wealth level of EUR 222 128, and lowest amongst those from 20-29 years old, with an average net wealth level of EUR 47 727.

The extent to which net wealth differs across age distributions varies across countries. In Austria and Belgium, for example, net wealth levels varies substantially, peaking at an average of EUR 506 811 and EUR 492 371 amongst those from 50 to 59 and 60 to 69 respectively (though this peak is not as pronounced with respect to median wealth levels). This compares to the lowest 10-year age bracket (ages 20 to 29) who have an average net wealth of EUR 282 329. By contrast, the Slovak Republic has a distribution of net wealth that is relatively flat in absolute terms across various age groups, with the wealthiest age group – those from 50 to 59 – having an average net wealth of EUR 76 772 (and a median level of EUR 61 474) compared to the age group with the lowest average net wealth - those from 20 to 29, who have a net wealth of EUR 22 165 (and a median level of EUR 8 777). This may suggest low levels of aggregate household saving for retirement. These results should be considered with caution however, given that the estimates of pension wealth in the NCFS are not comprehensive (see Box 4.1).

4.5. Net wealth, debt and housing

Net wealth inequality and income inequality

A complicating factor for those assessing the impact of savings taxation is the fact that income and net wealth inequality present different policy challenges. Sections 4.2 and 4.3 of this chapter have highlighted differing patterns of income and net wealth inequality and the differing patterns of asset holding across different levels of income and net wealth. However those individuals and households with high incomes may not be same as those with high levels of net wealth, and vice versa. This is particularly the case because variation in net wealth across society results in substantial part from consumption smoothing over time.

Those who earn a high income are also more likely to be wealthy, and those who earn a low income are more likely to have low levels of net wealth. Figure 4.11 shows the share of each net wealth decile that is contained within each income decile. On average, the probability of a household from the highest income decile being in the highest net wealth decile is 43.2%, while the probability of a household from the lowest income decile being in the lowest net wealth decile is 22.2%.

However, the correlation between income and net wealth is not perfect. Indeed, the data suggest that there are some households in the highest net wealth deciles who earn among the lowest incomes in the sample. Conversely, some households in the lowest net wealth deciles who earn among the highest incomes in the sample.

The absence of perfect correlation between income and net wealth may present challenges for tax policymaking. Efforts to tax those with high levels of net wealth may encounter difficulties where those with high levels of net wealth may have low levels of income, and so may encounter liquidity problems when paying taxes levied. From an equity perspective, it is not obvious whether the most appropriate “ability-to-pay” criterion is income or net wealth, or whether progressive taxation should be progressive with respect to income, or net wealth, or both. These issues are discussed further in Chapter 6.

Net wealth vs gross wealth

The preceding analysis presented results as a percentage of gross wealth. As noted, this was to avoid interpretation difficulties. However, there are significant differences in asset mixes expressed as a percentage of net wealth and gross wealth, particularly regarding housing. This section examines this in more detail. The data in the HFCS suggests that many households that have significant gross assets may be indebted on many of those assets. This is particularly the case for households with low levels of net wealth. For example, many households may finance their principal residence with debt. This in turn means that while gross housing wealth is substantial, net housing wealth may be far lower or even negative.

Housing and housing debt are key drivers in the variation between gross and net wealth across income and net wealth deciles. The results presented in Sections 4.2, Section 4.3 and 4.4 have shown that housing is the predominant form of real asset holding across most income and net wealth deciles for most countries in the HFCS sample.

The probability of owning a home is positively correlated with both income and net wealth. Figure 4.12 and Figure 4.13 show patterns of gross housing wealth across income and net wealth deciles. Figure 4.12 shows that those with low levels of net wealth are far less likely to have positive gross housing wealth (i.e. to own a home). Figure 4.13 shows that those with low levels of income are somewhat less likely to have positive gross housing wealth. Given that gross housing wealth is positive for some households in low net wealth deciles, it follows that mortgage debt associated with this housing wealth is likely to be substantial. The next section discusses household debt in more detail.

Distribution of debt holdings

This section brings together several stylised facts on income, net wealth and debt in this chapter. Debt, and particularly mortgage debt, plays a crucial role in driving asset holdings and influencing the distributional effects of the taxation of household savings. Figure 4.12 and Figure 4.13 have shown that gross housing wealth increases with both income and net wealth. Figure 4.11 has highlighted substantial differences in the income and net wealth distribution, as there are many households with high incomes and low levels of net wealth, and also households with low incomes and high levels of net wealth.

Those with gross housing wealth who are in the lowest net wealth decile have low net wealth precisely because they have substantial housing debt. Gross housing wealth (i.e. home ownership) is spread more evenly across income deciles than across net wealth deciles. It follows that some part of the variation between income deciles and net wealth deciles shown in Figure 4.11 is a function of housing debt.

As a share of gross wealth, those in low net wealth deciles have much higher levels of debt relative to their wealth when compared to those in high net wealth deciles. Figure 4.14 shows that the average amount of household debt in the lowest net wealth decile is 141.8% of total gross wealth. For this decile, housing debt is 91.9% of gross wealth (75.2% on the main residence, 16.7% on other residences). Non-mortgage debt is 49.8% of gross wealth.

For higher net wealth deciles, the share of non-housing debt falls substantially. Mortgage debt on the main residence does fall as a share of net wealth at higher net wealth levels, but not by as much as non-mortgage debt. Mortgage debt is 25.4%, 9.6%, and 2.5% of gross wealth in the 4th, 7th, and top net wealth deciles respectively. High levels of debt in part explain the negative levels of net wealth seen in Figure 4.6.

As a share of gross wealth, debt levels are far less correlated with income than they are with net wealth. Figure 4.14 also shows that while households above the lowest income deciles do have higher debt as a share of gross wealth, these shares rise only modestly with income. Debt levels are is 8.1%, 9.9%, and 13.6% of gross wealth in the 1st, 4th, and 7th income deciles respectively. Moreover, debt levels in the top income decile are lower than those of the 9th decile (at 12.6%, compared to 14.4%).

The results in the left panel of Figure 4.14 mask substantial variation within income deciles that can be seen on the right panel. Figure 4.14 shows that those with low levels of net wealth may have low levels of net wealth not necessarily because they have low levels of gross wealth, but because they have high levels of debt.

The absolute level of debt is broadly constant across net wealth deciles, but rises with income, on average, across countries. This variation is explored in Figure 4.15. Figure 4.15 shows absolute debt levels (i.e. in euros) as opposed to as a percentage of gross wealth. These figures show that those with higher income levels have higher debt levels on average, but that these debt levels rise roughly proportionately with income. By contrast, absolute levels of gross debt are roughly constant across net wealth deciles. This means that debt falls substantially as a share of gross wealth across net wealth deciles.

These data provide important context for tax policymakers assessing savings tax policies from a distributional perspective. Debt levels rise with income in absolute terms, so deductions for interest (such as mortgage interest) provide a greater absolute benefit to those with higher incomes. However these same deductions provide higher proportional benefits to those with low levels of net wealth. This is because housing forms a greater part of their gross wealth, and because housing debt is also more substantial relative to their gross wealth.

References

Alvaredo, F. et al. (eds.) (2018), The World Inequality Report, Harvard University Press.

Atkinson, A. and J. Stiglitz (1976), “The design of the tax structure: direct versus indirect taxation”, Journal of Public Economics, Vol. 6, pp. 55-75.

European Central Bank (2016), The Household Finance and Consumption Survey: Cross-country metadata information, Wave 2.

Brys, B., Perret, S., Thomas, A., and O’Reilly, P. (2016), “Tax Design for Inclusive Economic Growth”, OECD Tax Policy Working Papers, No. 26, OECD Publishing, Paris, https://doi.org/10.1787/5jlv74ggk0g7-en .

Causa, O. and M. Hermansen (n.d.), “The effectiveness of tax and transfer systems in reducing income inequality across OECD countries”, OECD Economics Department Working Papers (forthcoming).

Chamley, C. (1986), “Optimal taxation of capital income in general equilibrium with infinite lives.”, Econometrica, Vol. 54, pp. 607-622.

Golosov, M., Kocherlakota, N. and Tsyvinski, A. (2003), “Optimal Indirect and Capital Taxation”, Review of Economic Studies, Vol. 70, No. 3, pp. 569-587.

Judd, K. (1985), “Redistributive taxation in a simple perfect foresight model”, Journal of Public Economics, Vol. 29, pp. 59-83.

OECD (2018) Gearing Tax Policy to Make Inclusive Growth Happen, OECD, Paris (forthcoming).

OECD (2015), In It Together: Why Less Inequality Benefits All, OECD Publishing.

Saez, E. (2013), “Optimal progressive capital income taxes in the infinite horizon model”, Journal of Public Economics, Vol. 97, pp. 61-74.

Saez, E. (2017), “Income and wealth inequality: Evidence and policy implications”, Contemporary Economic Policy, Vol. 35, No.1, pp. 7-25.

Stantcheva, S. (2017), “Optimal Taxation and Human Capital Policies over the Life Cycle”, Journal of Political Economy, Vol. 125, No.6, pp. 1931-1990.

Zucman, G. (2014), “Taxing across Borders: Tracking Personal Wealth and Corporate Profits”, Journal of Economic Perspectives, Vol. 28, No.4, pp. 121-148.

Zucman, G. (2013), The missing wealth of nations: Are Europe and the U.S. net debtors or net creditors? Quarterly Journal of Economics, Vol. 128, No. 3, pp. 1321-1364.

Notes

← 1. The same analysis was replicated with data from the first wave of the HFCS, producing broadly consistent trends and results as those presented in this chapter for the second wave.

← 2. HFCS data contains information only about market income inequality. A discussion of the impact of the tax and transfer system is beyond the scope of this report. Recent OECD research has, however, considered these issues (Causa and Hermansen, n.d.).

← 3. Throughout, the results presented here are for the 18 OECD countries in the HFCS. When the text refers to HCFS average, it refers to 18-country averages, not 20-country averages.

← 4. This can include current accounts and savings accounts.

← 5. Valuables in this case refer to jewellery, works of art, antiques, etc.