Wealth Inequality & House Prices: A study on the effect of the evolution of house prices on wealth inequality in the Netherlands between 2003 and 2018

(1)

11-6-2019

Wealth Inequality

& House Prices

A study on the effect of the evolution of house prices on wealth inequality in the Netherlands between 2003 and 2018

Annoes van der Zande, s1700464 Leiden University

Faculty Governance and Global Affairs Master of Public Administration

Specialization Economics & Governance Supervisor: E. Suari-Andreu

(2)

Abstract

This research looks at the association between house prices and wealth inequality. The main question is: “What is the association between the evolution of the house prices and the evo-lution of wealth inequality in the years 2003-2018?” The expectation is that there is a nega-tive association between house prices and wealth inequality: a decrease in house prices leads to more inequality and the other way around. This because the middle class has a relative higher share of housing assets compared to the higher class. Therefore, they are more affected by changes in house prices. With use of DHS-data, the households’ net worth is calculated. The Gini-coefficient and the mean-median ratio indicate how the net worth is distributed across the sample. Especially in the years of the financial crisis, when the house prices dropped, a clear association is visible between the decrease in house prices and an increase in inequality. To specify the results, the sample is divided in three groups: the bottom 50, middle 40 and the top 10. A decomposition of the household portfolio for each of these groups shows that especially the bottom 50 and the middle 40 group have a large share of housing wealth. The top 10 has relatively more financial wealth. This can explain the association between house prices and wealth inequality.

(3)

1. Introduction

The evolution of homeownership and house prices have been discussed often in the past dec-ades. In many OECD countries homeownership rates have increased significantly. In the Neth-erlands, the owner-occupied market now makes up for almost 60% of the housing market (Rijksoverheid, 2015). This increase can partially be explained by changes in household char-acteristics, but a significant part of the increase in homeownership rates remains unexplained. This leaves room for the possibility that public policy can also partially explain the increase in homeownership rates (Andrews & Sánchez, 2011). Homeownership was broadly supported by governments across Europe. For example, the Netherlands had several measures in place which made the financing of a house more attractive. This led to many homeowners refinancing their mortgage or taking out a second one. During the crisis, the bubble burst and house prices dropped. Resulting in the fact that in 2019 the Netherlands still has the largest mortgage debt of all European countries (Figure 1.1).

Figure 1.1 Homeownership rates in EU-countries (Source: Eurostat)

Along with this, the house prices also increased over the past decades, especially in the period leading up to the Great Financial Crisis. As Figure 1.2 shows, the Netherlands still has one of the highest increases in house prices.

0 10 20 30 40 50 60 70 80 90 100 Ro m an ia Cro ati a Li th u an ia Hu n g ary P o lan d Bu lg aria Esto n ia Latv ia M alt a Sp a in S lo v ak ia P o rtu g al Lu x em b o u rg Be lg iu m Italy F in lan d Cy p ru s Ire lan d Ne th erlan d s EU S we d en UK F ra n c e De n m a rk Au stri a Ge rm an y

Homeownership rates

(5)

4

Figure 1.2 Change of house prices relative to the year before in EU-countries (Source: Eurostat)

Figure 1.2 shows that the change in Dutch house prices is far above the European average. When it comes to homeownership rates, the Netherlands is more average. As can be seen from Figure 1.1, mostly the Eastern-European countries have the highest homeownership rates. Another remarkable trend of the past years is that after the drop of the house prices during the crisis, access to homeownership became more difficult. This is both a consequence of the fi-nancial crisis and the measures taken by the government. The consequences of all these trends can be serious. As Arundell & Doling (2017) mention: housing careers are central to one’s life course. When a housing career is uncertain or unstable, this may have significant impact on the wealth and wellbeing of the future generation.

What is often discussed in combination with housing, is wealth inequality. Wealth inequality is an important aspect of the broad picture of socio-economic inequalities. According to the OECD (2015) the Netherlands springs out when it comes to wealth inequality. As Van Bavel (2014) notices: the Netherlands is not the equal country it perceives to be. Within OECD coun-tries, wealth inequality is often twice as high as income inequality. When wealth inequality is measured as the share of wealth hold by the 10% at the top of the distribution, the Netherlands has one of the highest rates of the OECD countries. In 2016 more than 68% of the net wealth is hold by the top 10 percent of the distribution.

-1% 0% 1% 2% 3% 4% 5% 6% 7% 8% 9% 10% 11% Po rtu g al Ire lan d Latv ia Hu n g ary Ne th erlan d s S lo v ak ia Li th u an ia Lu x em b o u rg S p a in Bu lg aria P o lan d Cro ati a Esto n ia M alt a Ro m an ia Ge rm an y Au stri a De n m a rk EU UK Be lg iu m F ra n c e Cy p ru s F in lan d Italy S we d en

(6)

5 Looking closer at the households’ wealth can provide important information about how differ-ent types of households respond to financial shocks or other economic developmdiffer-ents (OECD, 2015). Especially in combination with housing this can be interesting, since housing can be an important part of the households’ wealth. Wealth allows people to smoothen consumption over time and can protect them from unexpected changes (OECD, 2011). While homeownership is seen as an important source of the wealth of a nation and the wealth of an individual, young people these days seem to be excluded from accessing the housing ladder. Overall, a declining rate of entry into the housing market can be seen amongst the under-thirties in OECD countries (McKee, 2012). Partially, this is the case because after the Global Credit Crunch (GCC) the criteria for mortgage lending have been tightened up. For example, nowadays a sizable down payment is necessary in most cases. This disadvantages the younger people, who have had less opportunity to save. According to Allegré and Timbeau (2015) the current combination of the rising house prices and the more difficult access to homeownership leads to a reinforcement of the already existing wealth inequality.

This research looks closer at all these trends of the past decades. It will therefore zoom in on the changes in the house prices and the changes in wealth inequality in the Netherlands. The main question in this research is: “What is the association between the evolution of the house prices and the evolution of wealth inequality in the years 2003-2018?” In this way the effects of the increasing house prices and the government support for homeownership from before the crisis can be measured. But also, the drop of the house prices and the recovery of the housing market that took place during the crisis and afterwards. In this way, the new conditions on the housing market in the Netherlands can be investigated. By doing this, it is possible to add knowledge to the scientific research from before the financial crisis. Previously, homeowner-ship was strongly encouraged by the government. Nowadays, the encouragement is less. It is of societal relevance to assess the effects of this overall policy change on homeownership crit-ically. Besides that, in these days it is more difficult for the younger generation to have access to homeownership. This is partially due to the increasing house prices (CBS, 2019b). Further-more, together with income inequality, wealth inequality is one of the essential components of individual well-being according to the OECD (2011). Therefore, it is of societal relevance to investigate the effects of these changes.

(7)

6 This research uses DHS data, which consists of information on a panel of more than 1500 Dutch households in the specified period. The data provides information on the house prices of the households, but also on other aspects of the households’ portfolio. The evolution of wealth inequality is measured using Gini-coefficients and the mean-median ratio. To see if the changes in wealth inequality are really due to the changes in house prices, the housing assets and finan-cial assets are excluded from the portfolio. Furthermore, a decomposition of the households’ portfolio is used to explain the trends in wealth inequality. For this decomposition, the popula-tion is divided in three different groups: the bottom 50, the middle 40 and the top 10 percent of the distribution.

First, in chapter 2, the institutional context of the Netherlands will be discussed. This chapter provides information on how government supported homeownership and what has changed in this support over the years. Next, the theoretical framework reviews the most important theories about the relation between homeownership, house prices and wealth inequality. Based on this, the expectation is that there will be a negative association between house prices and wealth inequality. An increase in house prices will lead to less wealth inequality while a decrease of the house prices is expected to lead to more inequality.

The research design provides details about the method of data collection and the method of analysis. The results from the analysis indicate that especially for the years 2010-2013 a clear association between the evolution of the house prices and the evolution of wealth inequality is visible. The drop in the house prices led to more inequality. Both the Gini-coefficient as well as the mean-median ratio showed this. The results are further specified by taking out the housing assets and the financial assets of the analysis. This again showed the importance of housing wealth as part of total wealth. Finally, a decomposition of the wealth portfolio by different percentile groups showed that compared to the lower- and middle-class, the upper-class relies relatively less on housing wealth and more on financial wealth. The research concludes with a summary of the results and a short discussion.

(8)

7

2. The Dutch Housing Market

In many countries, governments see homeownership as the preferred tenure, because of the positive externalities it brings. For example, research shows that homeowners are perceived to be better citizens because they generally invest more in their neighborhood. (Elsinga et al., 2009). Because of the perceived benefits of homeownership, it is supported by the government in many countries. However, Tomann (1996) questions whether a higher rate of homeowner-ship in the total housing market leads to a better distribution of wealth. He takes the example of Germany, that has a strong and successful renting sector. Nonetheless, many governments intervene in the housing market. These interventions can be introducing grants, subsidizing loans and offering tax relief amongst others (Elsinga et al., 2009). Another reason for the gov-ernment to intervene in the housing market is because lower-income households may have problems accessing the housing market. These groups can be seen as a risk to investors and will therefore have trouble getting a loan. Furthermore, an increase in house prices can lead to more support from the government to encourage homeownership, especially among middle- and low-income groups (Elsinga et al., 2009).

This research will focus on the period 2003-2018 in the Netherlands. During this period, the government had several policies in place to regulate the housing market. These regulations can be favorable for tenants, like the regulations about social rent, but they can also favor home-ownership. In the Netherlands the government has different (fiscal) measures in place to support homeownership. This chapter discusses the measures that were used during the period 2003-2018

2.1 Tax measures

In the Netherlands, there is a tax system in place that is favorable for homeowners and for the financing of homes. Rijksoverheid (2015) lists three measures that are beneficial for (future) homeowners. First of all, homeowners pay a real estate tax levied by local municipalities, but generally they are exempt from capital taxes. Secondly, when someone wants to purchase a house, government provides a reduction in the transaction taxes that come with it. Thirdly, government has set measures considering imputed rent and mortgage deductibility.

Especially the imputed rent and mortgage deductibility is special. As Rouwendal (2007) men-tions, the Netherlands is one of the few countries in the world were the interest on mortgage loans can be fully deducted from taxable income. This is a very beneficial measure for home-owners. What homeowners pay as net interest is the difference between the gross interest

(9)

8 payment and what they save on tax payment because of the deductibility. Especially since in the Netherlands the highest tax rate is 52% this leads to substantially lower net mortgage interest payments than gross payments (Rouwendal, 2007). Furthermore, it becomes attractive for homeowners to take out an interest-only mortgage. By doing this this they are able to maximize the benefits they can get from interest payment tax deductibility.

The imputed rent is calculated at a modest percentage of the value of the property (Rijksover-heid, 2015). With modest is meant that generally, it is set at a lower price than the market price. This means that the real estate tax the homeowners have to pay is actually lower than it could have been. The combination of these two measures led to the fact that in general homeowner-ship can have the benefit of paying a lower amount of income tax (Rouwendal, 2007).

One of the side effects of the mortgage interest tax-deductibility system is that in combination with the inelastic supply of houses, it can drive up the house prices (Besseling et al., 2008). Priemus (2010) estimates that because of the mortgage interest relief in combination with re-strictions on the supply side, the prices of owner-occupied houses in the Netherlands are 20% of higher value than they would have been without the measure.

2.2 Mortgage guarantee scheme

Another instrument used by the government to promote homeownership is the mortgage guar-antee scheme. This scheme is designed to protect homeowners with mortgages against (unex-pected) losses and makes it possible to soften the traditional requirements people have to meet before they become eligible for borrowing. With the help of the mortgage guarantee scheme, people who would otherwise be excluded from the housing market, can get finance (Elsinga et al., 2009). In the Netherlands, there is the Mortgage Guarantee Fund (NHG). Before this fund, each municipality was able to set its own conditions for borrowing, but after the housing-market crash in the 1980’s, the government decided to launch the NHG-guarantee scheme for borrow-ers. This scheme made it possible for lenders to get a full repayment of the original amount of money plus interest in case the borrower should default due to unexpected circumstances. Fur-ther, the NHG-guarantee made it possible to borrow without any down payment (Scanlon & Elsinga, 2014). This too, lead to an increase of borrowing.

(10)

9 2.3 Changes in the system

In 2008 the credit crunch also hit the Netherlands. Some banks went bankrupt and because of this banking crisis, getting a loan became more difficult. In the years before the crisis, the num-ber of mortgages and the mortgage debt increased every year. Furthermore, more and more critique was expressed considering the mortgage deductibility system. Now, the government was forced to rethink their housing policies. This resulted in the regulation of the mortgage market and the recent changes in the mortgage interest tax-deductibility system (Ronald & Dol, 2011).

In 2007 the Ministry of Housing, Spatial Planning and the Environment (VROM) reported that the gap between the renting population and the owner occupation should become smaller. This gap existed because of the increasing house prices which makes it difficult for renters to enter the housing market. One of their advisements was to convert mortgage interest relief, imputed rent and the rent allowance into a more tenure-neutral form of support, instead of favoring homeownership, as was the case before. They advised to phase out mortgage tax relief (VROM, 2007). However, the cabinet of Balkenende IV did not support such a reform. Therefore, only after the credit crunch, this report was reconsidered. In the meantime, the criticisms against subsidizing housing in general increased. The subsidizing would encourage over-consumption and leads to more borrowing than the borrows actually were financially capable of.

In 2013 it was decided that only people with an annuity or with a linear mortgage would be eligible for mortgage tax deductibility. This measure should prevent that people took out an interest-only mortgage. Furthermore, it was decided that maximum the mortgage deductibility rate is being phased out. This happens in gradual steps. The goal is that in 2023 the maximum rate is 37,05% (Rijksoverheid, 2019).

(11)

10

3. Theoretical Framework

The relation between housing and inequality has been researched in many ways. These re-searches contain both information on the relationship between homeownership and inequality and on the relationship between house prices and inequality. There are two main perspectives when it comes to homeownership and inequality. The first perspective assumes that housing circumstances depend on the individuals’ social-economic status. Therefore, inequality in these circumstances lead to inequality in homeownership (Kurz & Blossfeld, 2004). The second per-spective considers the relation from the other way: homeownership has an impact on inequality, specifically wealth inequality (Mulder 2004, Arundel & Doling 2016).

3.1 The evolution of homeownership and house prices

As Figure 3.1 shows, over the last decades not only the house prices increased, also the overall homeownership rates increased. Whereas the house prices declined with the financial crisis, the homeownership rates stayed more or less the same.

Figure 3.1 Trends in homeownership and house prices in the Netherlands (Source: CBS)

In the period up until 2007, multiple states in Europe tried to make home-ownership attractive, because of the perceived benefits it comes with (Andrews & Sánchez, 2011). This glorification of homeownership is also called “the ideology of homeownership”. In most countries it is as-sumed that homeownership generates different social benefits than for renters. Previous re-search supported this hypothesis. For example, homeowners are more involved in their neigh-borhood, they are better caretakers and they are socially more active (André & Dewilde, 2016).

46 48 50 52 54 56 58 60 62 €-€50.000 €100.000 €150.000 €200.000 €250.000 €300.000 €350.000 1 9 9 8 1 9 9 9 2 0 0 0 2 0 0 1 2 0 0 2 2 0 0 3 2 0 0 4 2 0 0 5 2 0 0 6 2 0 0 7 2 0 0 8 2 0 0 9 2 0 1 0 2 0 1 1 2 0 1 2 2 0 1 3 2 0 1 4 2 0 1 5 2 0 1 6 2 0 1 7 2 0 1 8 H o m eo w nersh ip ra tes H o u se p rices in e u ro 's

Trends in homeownership and house prices

(12)

11 These assumed positive outcomes, give policymakers an incentive to encourage homeowner-ship. Furthermore, homeownership is encouraged through public policy in OECD countries because it is an easy way to stimulate the economy and therefore generating economic growth (Andrews & Sánchez, 2011). As was shown in the institutional context chapter, the Dutch gov-ernment also encouraged homeownership.

The homeownership rates have increased over the recent decades, together with the house prices. This trend continued until the financial crisis. In some countries the increase in house prices led to a house bubble. For example, in America, the house price bubble partially caused the financial crisis. The increases in homeownership rates can be partially be explained by changes in the characteristics of households and the ageing of the population according to An-drews and Sánchez (2011). Further, statistical research suggests that there is a big role for the relaxation of down-payment constraints on mortgage loans in the growth of homeownership rates. But, when the tax-relief on the mortgage debt is relatively generous, the impact is smaller. With the Global Credit Crunch, private renting has revived again, while homeownership access declined in Europe (Arundell & Doling, 2017). Arundell & Doling (2017) argue that these changes in tenure are not only because of the financial crisis, but a result of more fundamental changes, especially in labor markets. This is why it is not likely that the situation of mass home-ownership will return. This can have several consequences.

3.2 Advantages of homeownership

Homeowners are different from renters in the sense that homeownership comes with a set of property rights. Especially the right to sell the property is important, since this can lead to capital gain. Homeownership is therefore a form of asset-based welfare, also called ‘housing wealth’. Furthermore, there are certain benefits that come together with homeownership. For example, it is shown that homeownership is important in life-cycle investment strategies and for asset-based welfare (Arundell & Doling, 2017).

3.3 The evolution of wealth inequality

While a lot of research focuses on income inequality, wealth inequality can also be an important aspect in evaluating a countries overall inequality. Wealth inequality is about the distribution of wealth amongst households and is thereby an important part of the households’ economic well-being (OECD 2018). Usually, wealth inequality is much more unequally distributed than income. Figure 3.2 gives the wealth inequality in Europe measured by the share of wealth held by the top 10 of the distribution.

(13)

12

Figure 3.2 Wealth inequality in Europe (Source: piketty.pse.fr/capital21c)

As can be seen from this graph: the top 10 percent of the distribution owns more than half of the total wealth. This indicates a much larger inequality than the average Gini-coefficient, in Europe which is around 0,30 (OECD, 2017).

While the Gini-coefficient of the Netherlands is a bit lower than the OECD average, when it comes to wealth inequality, they are far above average. As Van Bavel (2014) mentions, from an international perspective, wealth inequality in the Netherlands is high. A partial explanation is that the private assets are really unequally distributed amongst households. These private assets consist of housing assets and financial assets. Van Bavel (2014) showed that the top 10 percent of the Dutch households owns more than half of the total assets. Also, according to the OECD (2018) the top 10 of the distribution in the Netherlands owned 68% of total wealth, while in other countries the share was around 50%.

When focusing on the evolution of the inequality of wealth, Van Bavel (2014) notices that the inequality of assets decreased until the 1980’s. More recent statistics show that the wealth ine-quality remained rather high over the past decades (OECD, 2018). Changes in wealth now take place mostly at the very top and bottom of the distribution. This is also what Van Bavel (2014) found. He mentions four main causes of this inequality. One of the causes of the increasing inequality in assets is the growth in house prices and holdings. According to Van Bavel (2014) this is not only a trend over the past years, but a trend over the past 30 years. Because of this

0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100% 1810 1830 1850 1870 1890 1910 1930 1950 1970 1990 2010 Sh a re o f to ta l w ea lt h

Wealth inequality in Europe

(14)

13 increase in prices, the richest were able to get richer and profited most, while the poorest of society were not able to profit at all.

Piketty (2014) elaborates on the structure of inequality. According to Piketty (2014) income inequality comes forth out of three elements: inequality in income from labor, inequality in income from ownership of capital and inequality that come forth out of the interaction between the first two elements. Important to keep in mind is that the inequality of income from capital can be larger than the inequality of capital itself. Piketty (2014) cites the problem of the top 10-percent of the capital income distribution owning always more than 50% of all wealth (pp. 305). Generally, the bottom 50% of the wealth distribution owns almost nothing (always less than 10% and generally less than 5%) (Piketty 2014, pp.306). According to Piketty (2014) this so-called hyper-concentration of wealth is a result of a high rate of return on capital in combination with low overall growth. When the growth in a country is low, but the return on capital high, this benefits the people with capital acquired in the past the most. Therefore, initial inequalities are accumulated. As can be seen from Figure 3.2, Europe had a really high wealth inequality before World War I: the top decile owned close to 90% of the total wealth and there was almost no middle class. In the years that followed, there was a sharp drop. Now the top decile owns around 65% of total wealth and the middle class owns a larger share of the total wealth. The large gap between the return on capital and the growth rate explains why wealth concentration was so high before the WOI (Piketty & Saez, 2014).

3.4 Homeownership and inequality

Over the last years, there is a lot of evidence of rising inequalities across OECD countries. Not only income inequality, but also wealth inequality. One of the key components of wealth is housing wealth (Arundel, 2017). The combination of strong economical and labor conditions and supportive policies for homeownership, have led to increasing homeownership rates across Europe (Conley & Gifford, 2006; Kurz & Blossfeld, 2004). Besides that, there was also large political support for homeownership and different policies making homeownership more at-tractive (Arundel, 2017). Since homeownership after the financial crisis became more difficult to access, this has led to diminished opportunities for the accumulation of housing wealth among many households. Furthermore, prices have increased while at the same time the finan-cial risk also has increased. This can make the difference between people on the housing market even bigger. Kurz & Blossfeld (2004) and Bernardi & Poggio (2004) also showed that home-ownership can have an effect on inequality by playing a role in one’s life changes and class differences.

(15)

14 Bernardi & Poggio (2004) also researched the relationship between homeownership and social inequality in Italy and found that there are class differences in the chances of becoming a home-owner. This suggests that the relationship can be reversed: social inequality has also an influ-ence on homeownership. Norris and Winston (2012) investigated the features of homeowner-ship systems in EU countries with inequalities in outcomes. They have identified different themes on the relationship between homeownership and inequality. First, wider structural ine-qualities (such as various socio-demographic characteristics) are of influence on the access to the tenure. Secondly, the unequal social distribution of homeownership has an effect of its own on winder social inequalities. In accordance with previous research, they found that the poten-tial gains of homeownership can reinforce already existing inequalities. Inequality in access to homeownership affects inequalities in outcomes, but inequalities in outcomes also affect the access to homeownership.

Also, Kaas, Kocharkov and Preugschat (2015) find a strong negative correlation between wealth inequality and homeownership rates across countries. They used the Household Finance and Consumption Survey, published by the European Central Bank, to investigate this. The survey provides household level data for 15 European countries and includes about 62000 households. The focus lied on the nine largest countries in the survey. Wealth inequality was measured by the Gini coefficient. They found negative correlations between the homeowner-ship rates and the Gini coefficient. Furthermore, they used two decomposition methods to show this correlation. In the first decomposition method, they decomposed the Gini coefficient by homeownership status. For this they measured the relative contribution of subgroups to the overall Gini coefficient. The subgroups consisted of owners, renter, between and residual. Both the owner subgroup and the between-group turned out to be important, but only the between group accounts for the negative relationship between the Gini coefficient and the homeowner-ship rate. Between-group inequality is lower in countries with a smaller share of renters. The second decomposition method takes other potential explanatory variables into account using a regression of the recentered influence function (RIF) of the Gini coefficient on observables (Kaas et al., 2015).

(16)

15 Kaas et al. (2015) also looked at the importance of housing wealth for the average household’s portfolio and its impact on inequality. For this, they divided wealth in different components: net own housing wealth, net financial wealth, net real wealth and business wealth. The shares of net own housing wealth are extremely high, even in countries with low ownership rates. This shows that housing wealth is an important contributor to overall wealth inequality. Furthermore, they find that the housing component contributes on average 42% to the overall Gini coefficient (Kaas et al., 2015).

3.5 Wealth effects of housing

Housing wealth is less liquid than financial wealth. Active saving in financial assets is highly responsive to real capital gains and losses on financial assets but active saving may be less responsive to changes in housing wealth (Disney et al., 2010). It is believed that house price changes influence household consumption through wealth effects. This is because an increase in the housing wealth of households reduces the need for homeowners to save for the future and therefore it allows them to spend more than they would have otherwise spend. However, some argue that housing wealth cannot be treated as another asset in the household balance, because housing is both an investment good as well as a consumption good (Buiter, 2008).

Wolff (2013) focused mainly on the effects of housing wealth for the middle class in the period 2007-2010 in the United States. This period has one of the sharpest declines in stock and in house prices. The research looked at different specific issues. First it looked if the inequality of household wealth increased over time. Second it looked at the time-trends in home ownership and home equity and finally at time trends in average wealth and home ownership for different age groups. The main findings of this research are that the median wealth decreased over the years 2007-2010. Furthermore, the inequality of net worth rose sharply in this period, while in the previous two decades it hardly had changed. The decrease in median net worth and the rise in its inequality can be explained by the high leverage of middle-class families and the high share of homes in their portfolio.

The main concept of wealth used in this study is net worth (marketable wealth) which is the current value of all assets minus the current value of all debts. Wolff (2013) compared the trends in net worth and non-home wealth and home equity. Furthermore, he looked at wealth inequal-ity before and during the crisis. For this the Gini coefficient was used and the share of wealth held by the top percentile. In the years of the crisis, the wealth inequality increased. Further-more, non-home wealth turned out to be even more concentrated than net worth: the richest 1

(17)

16 percent owned 42% of total household non-home wealth in 2010 (this was only 35% for net worth) and the top 20% owned 95% (compared to 89% net worth). According to Wolff (2013) there are two factors that account for the changes in wealth distribution. The first is the change in income inequality and the second is the change in the ratio of stock prices to housing prices. Wolff (2002) showed that wealth inequality is positively related to the ratio of stock prices to housing prices. During the crisis the median wealth dropped. According to Wolff (2013) this was caused by the high debt to net worth ratio of middle-class households. Because the upper-class had a lower debt to net worth ratio, they were less affected. This explains why the overall inequality increased during the crisis.

Because the house prices went up in the years before the financial crisis, people, and especially people from the middle-class, were able to borrow against the new (higher) prices of their house or to refinance the mortgages of their house for the increased value. This led to higher debts in the middle-class. Eventually when the housing market collapsed, many households were “un-derwater”, meaning that the value of their mortgage debt was higher than the value of their home (Wolff, 2013). Wolff (2013) found that in general 30 percent of total assets of a house-hold consists of housing. While for the middle class this is two third of the assets. Therefore, it is only logical that an economic downturn that affects the housing market (like the financial crisis) will hurt the wealth of the middle class the most. What explains the position of the middle class during the financial crisis is their high degree of leverage and the high concentration of assets in housing (Wolff, 2013). Furthermore, the middle-class was relatively hit harder in their net worth from the decline in house prices than the upper 20 percent was from the stock market. The increased debt and leverage that the middle class acquired during the rise of the house prices made them more vulnerable during the recession.

(18)

17 3.6 Hypothesis

Considering the question “What is the association between house prices and wealth inequal-ity?” the theoretical chapter showed that there is a relationship between both homeownership and house prices and wealth inequality. The relationship between homeownership and wealth inequality can go both ways. As the chapter on housing policies in the Netherlands and the section about the evolution of homeownership in the Netherlands showed, homeownership in-creased because of the housing policies. This affected particularly the middle- and lower-classes as they were now able to buy a house, whereas they otherwise would not have been able to. Because of this, their housing wealth increased.

The expectation is that the middle class has a larger share of housing wealth in their household portfolio compared to the upper class. According the ECB (2013), for homeowners the domi-nant components of net worth are housing assets and the debts that are related to this. Financial assets and other liabilities have only limited impact on their net worth. Because of this, they are expected to gain or lose more wealth when house prices fluctuate. Therefore, for the researched period, the expectation is that the collapse in house prices reduced the wealth of the middle class. The rationale behind this is that the middle class is expected to be more dependent on their housing wealth than for example the higher classes. For the higher class, financial wealth is expected to be relatively of more importance. Furthermore, when house prices drop after purchase, this can lead to negative house equity and therefore more inequality (OECD, 2018). Concluding: the expectation is that house prices are negatively associated to the wealth inequal-ity.

(19)

18

4. Research Design

4.1 Case selection

This research looks at the period 2003-2018 in order to see how the evolution of the house prices affected the wealth inequality in the Netherlands. As Figure 3.1 already showed, house prices increased until the crisis and dropped afterwards. Therefore, this time period will be able to show both the effects of an increase in the prices as well as the effects of a decrease in the prices. As became clear from the theoretical framework, the house prices and the value of stocks changed significantly during the crisis.

The unit of analysis in this period are the households in the Netherlands. How much the house-holds are affected by the changes in the house prices, is investigated by using micro-level data from the Netherlands.

4.2 Data collection

This research uses data from the Dutch National Bank Household Survey. The DNB Household Survey collects information on a panel consisting of more than 1500 households in the Nether-lands since 1993. The data are collected through the CentERpanel, that collects data every week on different topics. The DHS data consists of different questionnaires, of which different dataset are constructed. For this research the following are used: (1) household information (HHI), (2) accommodation data (HSE) and (3) the aggregated wealth data (AGW). The data on household information and on accommodation are measured at the household level, therefore only the responses from the household heads will be taken into account in the sample. The dataset of aggregated wealth is at the individual level. Therefore, for the relevant variables the sum of the observations per household is taken to make the datasets more comparable. Only information on the household heads will be used in this research.

4.3 Operationalization of variables 4.3.1 The evolution of the house prices

In the DHS-dataset different measures of the value of houses and therefore the house prices are provided. Households are asked several questions concerning the current value of their house. Two of these measures can be valuable for this research. The first is the question that asks the respondents how much they think they will get for their house if they sold it today. This measures the evolution of the house prices is called the self-reported house price. The second measure is the WOZ-value of the house, which is the value determined by the municipality.

(20)

19 The advantage of using self-reported house prices is that it is easy to collect. Furthermore, it is most likely that the household bases its (consumption)decisions on what they think their house is currently worth (Engelhardt, 1996). Finally, the household has the most recent information about the status of the house (e.g. renovations). A drawback is that people can estimate the price of their house too high or too low. A too high estimation of their house prices can lead to a higher net worth. People overestimate themselves and think they are richer than they actually are. This can lead to a seemingly more inequal distribution than in reality, since the differences between people with and without a house become even bigger. On the other hand, if people overestimate themselves the differences between them and the top of the distribution become smaller, so the wealth inequality seems lower than it actually is.

Goodman and Ittner (1992) find that homeowners systematically overestimate the value of their house by an average of six percent compared to what the sales price would be. They point out that this can especially become a problem for microlevel studies in which the self-reported price is used as an independent variable. Because of the overestimation, it can lead to inconsistent parameter estimates. One of their solutions is to redefine the estimated wealth effect as the effect of perceived wealth since the consumer beliefs that he has the amount of wealth corre-sponding to the estimated house price (Goodman and Ittner 1992, p. 355). This because most owners do not sell and therefore it is not likely that the owners themselves notice that their house is worth less than they thought. Goodman and Ittner (1992) conclude that even though the self-reported house prices is slightly overvalued, for most research this is not a problem. This research focuses on the trends over time. Since the self-reported house price is asked yearly again to the respondents, the results might be biased upwards, but the research is still able to observe trends over time.

Another way to estimate the value of the house is by using the WOZ-value. WOZ represents “Waardering Onroerende Zaken” in Dutch, which is the valuation of real estate. The WOZ-values are determined by local municipalities for calculating property tax. They are reassessed every year based on the sales prices of comparable properties. An advantage of WOZ-values is that they are able to provide a more accurate assessment of the heterogenous effect of house prices (Zhang, 2019). However, in the DHS-dataset the WOZ-value is asked to a reference date, but this differs per period. From 2003 – 2006 the WOZ-value is asked with as reference data 1999, while after 2007 it is asked for the year before. This makes the data slightly biased. An-other drawback compared to the self-reported value, is that the WOZ-value does not take into account the actual condition of the house and other possible circumstances. Since the DHS-data

(21)

20 provides both measurements of the house prices, they will both be used in the analysis since both measurements have their advantages.

4.3.2 The evolution of wealth inequality

According to Cowell & Van Kerm (2015) the standard concept for wealth is current net worth, that is the difference between assets and debts. The current net worth indicates the amount of wealth of the household, which is an important aspect to assess how the household is doing economically. People can consume more than only their income because of their wealth or save income to add to their wealth. Further, wealth allows people to smoothen consumption or pro-tects them against unexpected changes in income (OECD, 2018). The expectation is that for the middle class, housing wealth is a larger part of their portfolio. They are thus more affected by the drop in the house prices than for example the upper classes. It is therefore necessary to construct the remaining of the housing portfolio. This construction is comparable with the household portfolio in the research from Kaas et al. (2015). The household portfolio consists of non-financial assets, financial assets, liabilities from mortgages and other liabilities. These var-iables are constructed with the use of the aggregated data on assets and liabilities from the DHS-data. More detailed information about the construction of these variables can be found in Ap-pendix A.

Table 4.1 Construction of the household portfolio

Assets Liabilities Non-financial assets: - House - Others Financial assets Mortgages Other debts

Net worth = total assets – total liabilities

As can be seen in Table 4.1, the assets and the liabilities are divided in different subcategories. Assets is subdivided into non-financial and financial assets. In Appendix I the specific parts can be found, ranked according to liquidity. The non-financial assets are subdivided in non-finan-cial assets from the households’ house and in non-finannon-finan-cial assets from other assets. These can be for example cars or boats. This extra subdivision is made so that the housing wealth can be excluded from the portfolio. For this same reason, the liabilities are subdivided in mortgages and other debts. When housing wealth is excluded, mortgages are not taken into account. With the use of the households’ portfolio, the households’ net worth can be calculated by subtracting the total liabilities from the total assets. This gives information about how the household is doing economically/financially. The net worth will be used to calculate the different measures of inequality.

(22)

21

Gini-coefficients

One of the most common measures of inequality is the Gini-coefficient. For example, the OECD (2008) states that the Gini is the best measure of income inequality. The Gini-coefficient measures the wealth inequality between households in a number between zero and one. Zero means that there is perfect equality and one means that one household has the full amount of income and the other households have nothing (CBS, 2015). Thus, the Gini measures the ine-quality in a given distribution. It can also be used for wealth ineine-quality. Although there are some specific issues with the Gini and measuring wealth inequality. These will be discussed later.

Figure 4.1 Calculation of the Gini-coefficient (based on DHS data in the year 2003)

Figure 4.1 shows how the coefficient can be calculated. It is common to calculate the Gini-index for a country based on the cumulative share of income, but it is also possible to calculate the Gini for other measures. The linear line indicates how wealth should be distributed in an equal society. In that case, 50% of the population would possess 50% of the wealth. The curved line is the Lorenz-curve. This curve indicates the actual distribution of the wealth. The more convex the Lorenz curve is, the greater the concentration of wealth at the top of the distribution (Cowell & Van Kerm, 2015). You can see for example that 90% of the population in this case only owns 56% of total wealth. This means that the upper 10% owns the remaining wealth. The Gini-coefficient can be calculated by dividing area A (the orange area) by area A + B.

0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100% 0% 5% 1 0 % 1 5 % 2 0 % 2 5 % 3 0 % 3 5 % 4 0 % 4 5 % 5 0 % 5 5 % 6 0 % 6 5 % 7 0 % 7 5 % 8 0 % 8 5 % 9 0 % 9 5 % 1 0 0 % Cum ula tiv e sha re o f Net w o rt h

Cumulative share of households

GINI = A / (A+B)

A B

(23)

22 Figure 4.2 has been constructed to show the distribution of net worth in the research sample. In this graph, net worth is based on the self-reported house prices.

Figure 4.2 Lorenz curve based on net worth in 2003

This Figure shows how net worth was distributed across the sample for the year 2003. As can be seen, the actual distribution is far from equal, meaning that net worth is very unequally dis-tributed in this sample. What furthermore springs out, is that a part of the net worth is negative. This can be seen from the shape of the slope of the Lorenz curve. Only where the slope of the Lorenz curve moves upwards, net worth is positive again. This leads to some difficulties when calculating the Gini-coefficients. Normally the Gini indicates the inequality with a number be-tween zero and one, but when net worth is negative a part of the Lorenz curve lies below the horizontal axis. In this case, the Gini-coefficient can become larger than one.

One of the most common practices is to eliminate the observations with a negative value, or to substitute them with zero (OECD, 2015). However, there are some drawbacks to this method. First of all, by excluding the negative values, you risk excluding a large part of the observations from the sample. Further, by excluding the negative values, it is more difficult to compare the different distributions because there the negative value is taken into account (De Battisti et al., 2019). Figure 4.3 shows the distribution for the year 2003 if all the negative values are substi-tuted with zero.

-20% -10% 0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100% 0% 5% ₁0% 1 5 % 2 0 % 2 5 % 3 0 % 3 5 % 4 0 % 4 5 % 5 0 % 5 5 % 6 0 % 6 5 % 7 0 % 7 5 % 8 0 % 8 5 % 9 0 % 9 5 % 1 0 0 % Cum ula tiv e sha re o f Net w o rt h

Lorenz curve net worth (y=2003)

(24)

23

Figure 4.3 Lorenz curve for year 2003 with only positive values

In this research almost 20% of the households has negative net worth. De Battisti et al. (2019) recommend the use of different ways to calculate the Gini when dealing with negative values. They make a distinction between when 1% of total observations has negative value, when 1-5% has negative value or when the share of negative values is bigger than that. Furthermore, De Battisti et al. (2019) note that when dealing with negative values, the Gini coefficient is no longer a measure of concentration. It has to be interpreted as a relative measure of variability with respect to the mean value. Since the Gini is such a common inequality measure, it will be calculated even though a substantive part of the sample has negative net worth. Further, Cowell & Van Kerm (2015) indicate that as long as the mean is positive, wealth shares and the Lorenz curve are well defined. Since this is the case in the sample, the Gini-coefficient will be used. In addition, other measures of inequality will be used as comparison to provide more accurate results.

To see how the house prices affect the inequality, the Gini is also calculated with the housing aspect left out of the distribution. In this case, the net worth is calculated by subtracting liabili-ties (from other debts) from total assets (without housing). This makes the number of negative values of net worth increase. As can be seen in Figure 4.4 the inequality increases: the orange line is even further from the equality line than the blue line.

Lorenz-curve Networth y=2003

(25)

24

Figure 4.4 Lorenz curve with only positive values, based on net worth with and without housing

The problem of negative values occurs throughout all calculations. For net worth 20% is nega-tive, for net worth without housing 37% is negative and for net worth without financial assets 31% of the values is negative.

Mean-median ratio

The previous paragraph showed that using a Gini-coefficient can be problematic when measur-ing wealth inequality, therefore the ratio between the mean and median net worth is also used. Since the mean value is influenced by outliers at the top of the distribution, the ratio between the mean and the median also indicates how much inequality there is in the sample. The median value indicates the ‘normal’ household. When the mean is higher than the median, this already indicates a right skewed distribution, which is one of the features of wealth inequality (Cowell & Van Kerm, 2015). Furthermore, wealth inequality often knows a large concentration of wealth with a spike around zero. This because a large part of the population has no financial wealth or debt.

Another way of measuring inequality is by focusing on the share of wealth held by the top of the distribution. However, since survey data has trouble with capturing the very top of the dis-tribution, this might not provide an accurate overview (Vermeulen, 2016). What can be done is look into more detail how the net wealth is distributed among society using percentile groups. This can be done for the lower class, the middle class and the upper class. Based on this, the share of wealth held by the different classes can be calculated. OECD (2018) reported that in

Lorenz-curve Networth y=2003

(26)

25 2015 the average ratio between mean and median net wealth was 2,6. The Netherlands had an extremely high ratio of almost 8. In contrast: half of the OECD countries had ration’s below 2 (OECD, 2018).

4.4 Method of analysis

In order to see how the evolution of the house prices has affected the wealth inequality, first the household portfolio including housing is researched. Based on this, each households’ net worth is calculated. With the use of the Gini-coefficient, the mean-median ratio and the percentile groups, the distribution of net worth across the sample can be measured. By doing this, both the distribution of housing wealth and the distribution of inequality over time can be calculated. Since wealth is usually very concentrated at the top, focusing on the share of wealth held by the top of the distribution can make more sense than measuring wealth inequality in a broader sense (Vermeulen, 2016). Therefore, the share of wealth held by the top 10 percent will also be taken into account.

Furthermore, to see if the housing aspect is the only factor affecting wealth inequality, the hous-ing assets can be taken out of the household portfolio. Table 4.2 gives an indication of how the portfolio looks like without the housing aspect. Based on this, net worth is calculated again. This makes it possible to compare the situation including housing and the situation without housing.

Table 4.2 Construction of the household portfolio without housing assets

Assets Liabilities

Non-financial assets: others Financial assets

Other debts Net worth = total assets – total liabilities

The financial assets were hit as well in the years of the financial crisis. This can also influence the households’ portfolio and the distribution. For this reason, the financial assets will also be taken out of the analysis. As OECD (2018) mentions in their report: financial assets are one of the main factors influencing wealth inequality, next to the level of homeownership.

Table 4.3 Construction of the household portfolio without financial assets

Assets Liabilities Non-financial assets: - house - others Mortgages Other debts Net worth = total assets – total liabilities

(27)

26 4.5 Validity & reliability

According to Cowell & Van Kerm (2015) there are two main problems specific to measuring wealth inequality, which make it difficult to use some traditional measures of relative inequality (like the Gini-coefficient). The first problem is that net worth can take a negative value. This research compensates for this by taking other measures of inequality into account as well. The second problem is that one of the characteristics of wealth inequality is that it has a very skewed distribution with a large tail. This can result in extreme data which has to be taken into account when interpreting the results.

There are some drawbacks to the use of survey data. First of all, survey questions can lead to respondents answering not accurately or honest enough. People have the tendency to make themselves look better than they actually are, this can lead to biased data. Furthermore, respond-ents can misinterpret the questions, leading to biased answers. Finally, there can be data errors when some respondents choose not to answer a question, while others do. Problems that are evident with the use of survey data are non-response and under-reporting issues.

As Vermeulen (2016) mentions, measuring the upper tail of the wealth distribution is difficult using survey data. This is important, since wealth is usually concentrated at the very top. With his research, Vermeulen (2016) proves in his research that survey data almost always underes-timates the upper tail of the wealth distribution. The main reason for this is “differential unit non-response”, which basically means that the richer households are less likely to participate in surveys. A solution to this problem can be to replace the upper tail of the wealth distribution by a Pareto distribution (Vermeulen 2016; Cowell & Van Kerm 2015). In this research the top of the wealth is represented by the upper decile of the distribution. With the interpretation of the results, it must be taken into account that it is likely that this group holds a larger percentage of the wealth than represented.

(28)

27

5. Results

5.1 The evolution of the house prices

In Figure 5.1 the evolution of the house prices is graphically represented for years 2007–2018. The first 5 years of the WOZ values are extremely high and therefore likely to be biased. A graph of the WOZ-values for these years can be found in Appendix B, in Figure 1B.

Figure 5.1 The evolution of the house prices

As can be seen in Figure 5.1, the self-reported house prices are higher than the WOZ-value, which is in accordance with the literature. Both measurements of house price follow more or less the same pattern. However, since the WOZ-value is very biased during the first five years of the researched period, the analysis will only use the self-reported house price. For the years 2007 – 2018 the analysis is redone with the WOZ-value as house prices. This analysis can be found in Appendix B.

As can be seen in Table 5.1, the house prices increased during the years under investigation. The current house price in 2018 is higher than the price was in 2003. The growth rate is relative to the previous year. As can be seen in Table 5.1, the house prices increased the most in the years 2007-2009. In 2010 only a small growth can be observed. After 2010 the house prices drop. The year 2018 is the first year since 2010 that the house prices actually grow relative to the year before.

€ -€ 50.000,00 € 100.000,00 € 150.000,00 € 200.000,00 € 250.000,00 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016 2017 2018

The evolution of house prices

(29)

28 Table 5.1 The evolution of the house prices in detail: index numbers and growth rates. Green indicates a decrease, red an increase

Self-reported house price Index 2003-100 Growth rate 2003 € 160.845,80 100,00 2004 € 164.279,30 102,14 1,02 2005 € 170.409,20 105,95 1,04 2006 € 173.394,00 107,80 1,02 2007 € 186.628,20 116,03 1,08 2008 € 197.162,80 122,58 1,06 2009 € 209.188,70 130,06 1,06 2010 € 213.411,40 132,68 1,02 2011 € 211.254,00 131,34 0,99 2012 € 199.261,70 123,88 0,94 2013 € 188.318,50 117,08 0,95 2014 € 180.507,20 112,22 0,96 2015 € 181.321,90 112,73 1,01 2016 € 179.208,60 111,42 0,99 2017 € 170.079,10 105,74 0,95 2018 € 171.526,80 106,64 1,01

5.2 The evolution of the household portfolio

Figures 5.2 and 5.3 show respectively the asset and the liability side of the household portfolio.

Figure 5.2 The evolution of the household portfolio: trends in assets

The shape of the total assets line follows roughly the same pattern as the housing assets line. It is also interesting to look at the financial assets since these are also expected to affect the wealth distribution. For example, the steep drop between 2012 and 2013 of the total assets can be traced

€-€50.000 €100.000 €150.000 €200.000 €250.000 €300.000 €350.000 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016 2017 2018

Trends in assets

financial assets non-financial assets: other non-fiancial assets: housing assets total non-financial assets total assets

(30)

29 back to a drop in the mean of the financial assets. What can be seen is that the value of financial assets increases up until 2011, then they drop. This is at the same time that the average house prices decrease. This indicates also that the financial assets follow the boom and bust from the crisis. The years of the crisis are both visible in the housing assets as well as in the financial assets. The other non-financial assets remain rather stable over time.

Figure 5.3 The evolution of the household portfolio: trends in liabilities

As can be seen in Figure 5.3, mortgages are the most important liabilities of the households. The average mean of mortgages increases during the years until 2015. Since the mortgages are such an important part of the households’ liabilities, the total liabilities follow the same trend. The households’ other debts form a really small part of the households’ total debts and remain stable over time.

What can be already seen from the figures above is that the net worth of the households will follow the boom and bust cycle in the housing market. This makes sense, because both house prices as well as financial assets followed roughly the same pattern. Figures 5.2 and Figure 5.3 therefore already showed in this paragraph already how important housing is for the house-holds’ net worth.

€-€10.000 €20.000 €30.000 €40.000 €50.000 €60.000 €70.000 €80.000 €90.000 €100.000 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016 2017 2018

Trends in liabilities

(31)

30 5.3 The evolution of wealth inequality

In this paragraph the households’ total wealth is first compared to the total wealth without hous-ing assets. Next, the households’ wealth is compared to total wealth without financial assets. Figure 5.4 shows the trends in the average net worth of the households.

Figure 5.4 Trends in net worth: with and without housing assets

Net worth with housing assets excluded

When housing is excluded, the trend in net worth is much more stable than including housing. This makes sense, because the house prices were influenced by the boom leading up to the crisis and crashed afterwards. The house prices decreased from 2010 on and so did the average net worth. When looking at net worth without housing assets, the net worth doesn’t drop. This again confirms that housing influences the net worth of the households a lot.

Net worth with financial assets excluded

Since housing is not the only asset-based wealth that can be of importance, also the financial assets are taken out of the net worth. Especially because in the years of the financial crisis, the financial market also suffered. The analysis from chapter 5.2 is redone, but this time using the net worth without the financial assets.

€-€20.000 €40.000 €60.000 €80.000 €100.000 €120.000 €140.000 €160.000 €180.000 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016 2017 2018

Trends in networth

(32)

31

Figure 5.5 Trends in net worth with and without financial assets

When the net worth is calculated without financial assets, it follows the same pattern as net worth including it, as can be seen in Figure 5.5. This is contrary to net worth without housing assets. The results confirm that the financial assets do not have as much influence on the average net worth as housing assets do.

Mean and sum net worth

As can be seen in Figure 5.6 the mean net worth doesn’t follow the exact same pattern as the sum of net worth. For example, the sum of net worth increased between 2009 and 2010 but the average decreased. This can already indicate an increase in inequality. Another interesting pe-riod is between 2014 and 2017. The sum of net worth increased and stayed at the new level. However, this increase cannot be noticed in the mean, there is even a decrease between 2016 and 2017. €-€20.000 €40.000 €60.000 €80.000 €100.000 €120.000 €140.000 €160.000 €180.000 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016 2017 2018

Trends in networth

(33)

32

Figure 5.6 Trends in net worth: mean and sum

When the housing assets are removed from the household portfolio, the average and the sum of net worth follow the same trend. Therefore, it seems to be that the housing asset leads to a deviating trend between the average and sum of the net worth among households.

Figure 5.7 Trends in net worth: mean and sum without housing assets

Figure 5.8 shows the same, this time with financial assets excluded. What especially springs out is that now the trend between the mean and the sum of net worth becomes even less visible.

€-€50 €100 €150 €200 €250 €300 €-€30.000 €60.000 €90.000 €120.000 €150.000 €180.000 2 0 0 3 2 0 0 4 2 0 0 5 2 0 0 6 2 0 0 7 2 0 0 8 2 0 0 9 2 0 1 0 2 0 1 1 2 0 1 2 2 0 1 3 2 0 1 4 2 0 1 5 2 0 1 6 2 0 1 7 2 0 1 8 S u m o f n et w o rt h in m il io n s M ea n net w o rt h

Trends in net worth

sum mean €-€5 €10 €15 €20 €25 €30 €35 €40 €-€3.000 €6.000 €9.000 €12.000 €15.000 €18.000 €21.000 €24.000 2 0 0 3 2 0 0 4 2 0 0 5 2 0 0 6 2 0 0 7 2 0 0 8 2 0 0 9 2 0 1 0 2 0 1 1 2 0 1 2 2 0 1 3 2 0 1 4 2 0 1 5 2 0 1 6 2 0 1 7 2 0 1 8 Su m o f net w o rt h in m ilio ns M ea n net w o rt h

Trends in net worth without housing assets

(34)

33

Figure 5.8 Trends in net worth: mean and sum without financial assets

Share of net worth held by the top of the distribution

The share of net worth held by the top of the distribution is one of the indicators of inequality (OECD 2015, Piketty 2014). The larger the share held by the top 10 percent, the larger the inequality. Figure 5.9 represents the share of total net worth held by the top 10 percent.

Figure 5.9 Trends in net worth: share of wealth held by the top 10

€-€30 €60 €90 €120 €150 €180 €210 €-€20.000 €40.000 €60.000 €80.000 €100.000 €120.000 €140.000 2 0 0 3 2 0 0 4 2 0 0 5 2 0 0 6 2 0 0 7 2 0 0 8 2 0 0 9 2 0 1 0 2 0 1 1 2 0 1 2 2 0 1 3 2 0 1 4 2 0 1 5 2 0 1 6 2 0 1 7 2 0 1 8 S u m o f n et w o rt h in m il io n s M ea n net w o rt h

Trends in net worth without financial assets

sum mean 40% 45% 50% 55% 60% 2 0 0 3 2 0 0 4 2 0 0 5 2 0 0 6 2 0 0 7 2 0 0 8 2 0 0 9 2 0 1 0 2 0 1 1 2 0 1 2 2 0 1 3 2 0 1 4 2 0 1 5 2 0 1 6 2 0 1 7 2 0 1 8

(35)

34 What is most striking about Figure 5.9 is that between 2007 and 2009 the share of wealth of the top 10 decreased sharply. When looking back at Figure 5.1 these are also the years that the house prices strongly increased. After 2009 up until 2013 the share of wealth held by the top 10 percent increased from 42% to 53%. Again, looking back at Figure 5.1, these are the years that the house prices strongly decreased. In 2016 the share of wealth held by the top 10 was the highest. Afterwards, the share declined to 53%, which is still higher than in the years before the crisis. The results of Figure 5.9 therefore also indicate an increase in inequality during the years that the house prices dropped.

5.3.1 Gini-coefficients

For the assets side of the household portfolio, Gini-coefficients are calculated. These are graph-ically represented in Figure 5.10. The categories ‘non-financial assets: other’ and ‘financial assets’ are most unequally distributed. Other non-financial assets include assets like boats, cars and second houses. It makes sense that these are most unequally distributed. Also, financial assets are likely to be unequally distributed. The housing assets are most equally distributed across the sample: they have the lowest Gini coefficients. This means that among the sample probably more people have access to housing wealth than to financial wealth. It is therefore likely that the financial wealth is more concentrated at the top of the distribution. This will be research in more detail in paragraph 5.4.

Figure 5.10 Gini-coefficients of different aspects of the household portfolio 0,45 0,50 0,55 0,60 0,65 0,70 0,75 0,80 0,85 0,90 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016 2017 2018

Gini-coefficients: household portfolio

total assets total assets (housing excluded) total non-financial assets non-financial assets: other non-financial asset: housing financial assets

(36)

35 Besides general trends in the net worth, the distribution of total net worth across the sample is calculated using the Gini-coefficient. The results are given in Figure 5.11. As can be seen from the Figure, net worth is more equal distributed when the housing assets are included. Again, the housing boom is clearly visible in the distribution of net worth.

Figure 5.11: Gini-coefficients of net worth and net worth without housing assets

Without housing assets

The Gini-coefficients of the net worth without housing assets seem to follow a more whimsical pattern. Table 5.2 is constructed to look at this more in depth. The first two columns show the absolute Gini-coefficient for net worth and net worth without housing. The next two columns show the changes in the Gini-coefficient in index numbers with 2003 as base value. Finally, the last two columns show the growth rate of the Gini-coefficient. The growth rate shows the change relative to the year before. The green areas represent a decrease in inequality, the red areas an increase. 0,60 0,65 0,70 0,75 0,80 0,85 0,90 2 0 0 3 2 0 0 4 2 0 0 5 2 0 0 6 2 0 0 7 2 0 0 8 2 0 0 9 2 0 1 0 2 0 1 1 2 0 1 2 2 0 1 3 2 0 1 4 2 0 1 5 2 0 1 6 2 0 1 7 2 0 1 8

Wealth Inequality & House Prices: A study on the effect of the evolution of house prices on wealth inequality in the Netherlands between 2003 and 2018

11-6-2019