Paper for the Sixth Meeting of the Society for the Study of Economic Inequality (ECINEQ), July 13-15, 2015, Luxembourg

Transcription

1 Paper for the Sixth Meeting of the Society for the Study of Economic Inequality (ECINEQ), July 13-15, 2015, Luxembourg Using the Household Finance and Consumption Survey (HFCS) for a joint assessment of income and wealth taxes: Prospects, limitations and suggestions for policy simulations <Draft paper, please do not quote> Francesco Figari 1, Sarah Kuypers 2, Gerlinde Verbist 2 1 University of Insubria and ISER University of Essex 2 Centre for Social Policy, University of Antwerp Abstract We explore the prospects for using the Eurosystem Household Finance and Consumption Survey (HFCS) dataset as an underlying micro-database for policy simulation across euro zone countries. In particular, we consider the issues to be addressed and the advantages arising from building a database from the HFCS for the EU tax-benefit model, EUROMOD. EUROMOD is currently running mostly on EU-SILC data, but is built in a way that maximises its flexibility and possibility to simulate tax-benefit policies on different databases. This will allow expanding the policy domains currently covered in EUROMOD with dimensions like wealth taxation, which recently gained much prominence, in the academic as well as the public debate. As the HFCS only contains gross income amounts which are not suitable for redistributive analysis, the purpose of this paper is to derive net incomes by simulating the gross-to-net transition with EUROMOD taking into account all important details of the social security and personal income system. In order to identify the issues and illustrate their importance a trial database for Belgium is constructed. We conclude that, although transforming the HFCS into a database for EUROMOD would require a significant amount of effort, this is surely to be worthwhile because of the interesting possibilities to extend the policy scope of EUROMOD and to consider jointly the redistributive effect of income and wealth taxes. Moreover, the derivation of disposable income allows one to consider the joint distribution of income, wealth and consumption, which can be used to analyse issues relating to inequality and poverty. Key words: EUROMOD, HFCS, simulations, gross-to-net incomes, wealth taxation JEL Classification: C15, H24, I3 1

2 Using the Household Finance and Consumption Survey (HFCS) for a joint assessment income and wealth taxes: Prospects, limitations and suggestions for policy simulations Francesco Figari 1, Sarah Kuypers 2, Gerlinde Verbist 2 1 University of Insubria and ISER University of Essex 2 Centre for Social Policy, University of Antwerp 1 Introduction The increasing accumulation of private wealth in Europe appears as one of the most striking evolutions in the distributional literature over the last 40 years. The aggregate private wealthnational income ratios have nowadays returned to levels observed in the 19th century, ranging from 400% to 700 %. Such levels are determined by different economic factors, such as the long-run asset price recovery effect, high saving rates and low economic growth rates, at least partially sustained by pro-capital policies (Piketty and Saez, 2013). Focusing on household resources the ratios between private wealth and disposable income are even higher; the rise of freely provided public services and in-kind transfers such as health and education that occurred since the 1970s is an important factor in explaining why disposable income has declined relative to national income (Piketty and Zucman, 2014). High wealth-income ratios are not necessarily bad but they raise challenging issues about capital taxation (Piketty, 2011, 2014) and the overall structure of inequality (Davies, 2009). First, while in the last 50 years the contribution of wealth taxes to government revenues has diminished, there are strong arguments nowadays for broadening the existing tax bases to include wealth and income from wealth both on horizontal and vertical equity grounds. As long as wealth is more unequally distributed than income, wealth taxes are attractive in distributional terms. Moreover, as far as economic efficiency is concerned, wealth taxes minimise economic distortions by taxing fixed factors (Hills, 2013). In a recent contribution, Piketty (2013) proposes to go beyond the national boundaries and suggests the introduction of a comprehensive wealth tax at the European level, based on the market value of the net personal worth. Second, given that wealth is in general very concentrated (with a Gini coefficient ranging between 0.5 and 0.8 over time and across countries) and correlated with income, the inequality of wealth is likely to exacerbate overall inequality. Taxing wealth is a way to reduce this inequality. Hence, it is important to assess the role of the different wealth components across countries, in order to set appropriate tax-free allowances and concentrate the tax burden on the wealthy part of the population, given the increasing role of housing assets in the household s portfolio along the entire income distribution (Figari, 2013). In such a context the need for more comprehensive and integrated data on individual well-being is widely recognised, as highlighted in the Report by the Commission on the Measurement of Economic Performance and Social Progress (Stiglitz, Sen and Fitoussi, 2009). In order to identify better 2

3 measures of economic performance in a complex economy and thus going Beyond GDP, Stiglitz, Sen and Fitoussi recommend to consider income, consumption and wealth and to give more prominence to their joint distribution. New household surveys as those developed as part of the Luxembourg Wealth Study (Jäntti et al. 2013) and the Eurosystem Household Finance and Consumption Network (HFCN, 2013a) represent a milestone in this ongoing process to better measure individual well-being. Nevertheless, empirical research faces important and severe limitations that limit potential future attainments mainly due to data availability. This paper aims at contributing to the recent developments in this area by exploring the prospects for using the Eurosystem Household Finance and Consumption Survey (HFCS) dataset as an underlying database for a tax-benefit microsimulation model. In particular, we consider the issues to be addressed and the advantages arising from building a database from the HFCS for the EU-wide tax-benefit model, EUROMOD. Although it is currently running mostly on EU-SILC data, EUROMOD is built in a way that maximises its flexibility and possibility to simulate tax-benefit policies on different databases (Sutherland and Figari, 2013). The main advantages of incorporating the HFCS data in EUROMOD are twofold. First, it allows us to expand the policy domains currently covered in EUROMOD with dimensions like wealth taxation, which recently gained much prominence in the academic as well as the public debate. In addition to budgetary and distributional analysis of current wealth taxes, the model based on HFCS data would allow for an integrated assessment of taxable capacity taking into account direct taxes on income and wealth and tackling challenging issues such as those faced by asset rich/income poor households (Hills, 2013). Moreover, it would enable to estimate the impact of reforms in wealth taxation in interaction with other tax-benefit policies. Second, as the HFCS contains only gross income amounts which are not suitable for redistributive analysis, we derive net incomes by simulating the gross-to-net transition with EUROMOD taking into account all important details of the social security and personal income system. For the first time, this allows us to consider the joint distribution of disposable income, wealth and consumption based on information coming from the same survey, potentially comparable across countries and time. In order to identify the issues and illustrate their importance a trial database for Belgium is constructed. We conclude that, although transforming the HFCS into a database for EUROMOD would require a significant amount of effort, this is surely to be worthwhile because of the interesting possibilities to extend the policy scope of EUROMOD and also to consider jointly the redistributive effects of income and wealth taxes. Moreover, the derivation of disposable income allows one to consider the joint distribution of income, wealth and consumption. In the next section we briefly describe the advantages and limitations of tax-benefit models. In section 3 we discuss what the HFCS data can contribute to policy simulation in EUROMOD. In section 4 we discuss the assumptions and transformations needed to construct a EUROMOD database on the basis of the HFCS data, where Belgium is used as a case study. Section 5 then studies the results of the derivation of net incomes for the HFCS data and validates them against the EU-SILC and where possible external sources. The last section concludes. 3

4 2 Purpose of a tax-benefit model on the HFCS data The main advantage of a tax-benefit model is that it allows one to focus quite accurately on the objectives of social and economic policy, on the tools employed, and on the structural change experienced by those to whom the measures apply. Unlike a macroeconomic model, a microsimulation model allows one to simulate individual decision units. These decision units are in the case of the HFCS data households and the individuals that live in them. Fiscal rules are incorporated into the model as accurately as possible, so that the impact on the individual characteristics of a decision unit becomes apparent; the impact of social security and taxation may, after all, vary considerably for different units. The various decision units may also be aggregated according to different characteristics (e.g. age, social and professional category). As such, the model allows one to test the redistributive potential of different tax-benefit systems, while taking due account of social and demographic variables. Another important advantage of this method is that it allows one to study a set of policy measures from two distinct perspectives. On the one hand, one can focus on the cumulative effect of the various measures, and therefore also on the impact of the entire set of transfer-oriented measures. On the other hand, a microsimulation model offers the possibility of dissecting complex measures (e.g. step-by-step tax calculation for a household), so that the impact of each step may be considered separately. As described in Figari et al. (2015) different types of analysis are facilitated by using a microsimulation approach, among else: - impact of tax-benefit policy changes (e.g. reforms regarding wealth and income taxation) on income-based indicators and related statistics (e.g. poverty and inequality indicators); - impact of demographic factors on disposable income through the effects of tax-benefit policies (e.g. public support to families contingent on presence of children, see Figari et al., 2011); - impact of policy changes over time (e.g. profiles of gainers and losers of a policy indexation and policy reforms); - impact of policy changes on social indicators capturing work incentives (e.g. effective marginal tax rates, participation tax rates) or social inclusion (e.g. multiple deprivation). Of course, simulation models also have inherent limitations. These models use empirical data that are either obtained by means of surveys or from administrative sources. As such, the accuracy of the results depends on the quality of the data (e.g. adequate information about the relevant socioeconomic characteristics, a sufficiently large sample). Another limitation is the cost involved in constructing and maintaining such a model: developing a tax-benefit model requires time and money. Therefore, one will need to make certain considerations in terms of policy areas covered, incorporation of demographic and macro-economic processes or behavioural reactions. In order to exploit the cross-country dimension of the HFCS data, it is quite natural to build a database from the HFCS for EUROMOD, the EU-wide tax-benefit model, rather than for separate national tax-benefit models. Moreover, EUROMOD is built in a way that maximises its flexibility and possibility to simulate tax-benefit policies on different databases. 4

5 EUROMOD simulates cash benefit entitlements and direct tax and social insurance contribution liabilities on the basis of the tax-benefit rules in place and information available in the underlying datasets. Instruments which are not simulated (mainly contributory pensions), as well as market income are taken directly from the data (Sutherland and Figari, 2013). As such, EUROMOD is of value in terms of assessing the first order effects of tax-benefit policies and in understanding how taxbenefit policy reforms may affect income distribution, work incentives and government budgets in the short term. Currently EUROMOD runs on the EU-SILC data, which has only limited information on wealth and income from wealth. Incorporating the HFCS-data will allow expanding the policy domains currently covered in EUROMOD with dimensions like wealth taxation. This will enable simulations relating to issues like a tax shift from income to wealth (a currently hotly debated topic in e.g. Belgium). It will help to understand and measure the redistributive role of these policies, in relation to the other taxbenefit rules. With subsequent waves of the HFCS coming available, the microsimulation model will also enable to investigate changes over time and to determine to what extent these are due to changes in the underlying population or to changes in the policies. The second purpose of running EUROMOD on HFCS data is to derive a proper measure of disposable income, as the HFCS contains only gross income amounts which are not suitable for redistributive analysis. For the first time, this allows us to consider the joint distribution of disposable income, wealth and consumption based on information coming from the same survey, potentially comparable across countries and time. 3 HFCS and its perceived advantages over EU-SILC The Eurosystem Household Finance and Consumption Survey (HFCS) is a new dataset covering detailed household wealth, gross income and consumption information (Eurosystem Household Finance and Consumption Network [HFCN from now onwards], 2013a). It is the result of a joint effort of all National Banks of the Euro zone, three National Statistical Institutes 1 and the European Central Bank (ECB). The first wave was made available to researchers in April 2013 and contains information on more than households in 15 Euro area member states which were surveyed mostly in 2010 and Ireland and Estonia are not included, but joined in the second wave (fieldwork period is 2014). Moreover, Latvia, who joined the Euro zone on the 1st of January 2014 has also carried out the survey for the second wave. An important shortcoming of the direct research use of the HFCS data is that it only covers gross income amounts which make them for instance unsuitable for the analysis of issues of inequality and redistribution. Nevertheless, the income components that are covered in the HFCS are largely the same as those surveyed in EU-SILC. More specifically, the HFCS gross income concept includes the following components: employee income, self-employment income, rental income from real estate 1 Of France, Finland and Portugal 2 Exceptions are France (2009/2010), Greece (2009) and Spain (2008/2009) 5

6 property, income from financial investments, income from pensions (public, occupational & private), regular social transfers, regular private transfers, income from private business and income from other sources (HFCN, 2013b, p.108). The major differences with the income concept of EU-SILC are presented in Table 1. First, it is clear that in the category of employee income the HFCS only asks respondents about cash and near cash income, while EU-SILC also captures non-cash income. Secondly, pensions from mandatory employer-based schemes are included in public pensions in EU- SILC, while they are covered under private pensions in the HFCS (HFCN, 2013a, p.100). Finally, income received by people under 16 is covered in EU-SILC, but not in the HFCS. In contrast, the HFCS covers income from other types of sources (such as capital gains or losses from the sale of assets, prize winnings, insurance settlements, severance payments, lump sum payments upon retirement), while EU-SILC does not. However, considering the joint patterns of income and wealth inequality in Belgium, Kuypers et al. (2015) show that despite these methodological similarities, a non-negligible difference in gross income distributional outcomes exists between the HFCS and EU-SILC, mainly at the top of the distribution, which is arguably the consequence of the oversampling strategy implemented in the HFCS (see below for more details). Such differences suggest that is not enough to look at median incomes (HFCN, 2013, p. 100) to provide a reliable comparison between different surveys. Table 1: Comparison of gross income components HFCS and EU-SILC HFCS EU-SILC Employee income (Cash & near cash income) Employee cash or near cash income Non-cash employee income Self-employment income Cash benefits or losses from self-employment Rental income from real estate property Income from rental of a property or land Income from financial investments Income from private business other than selfemployment Interest, dividends, profit from capital investments in unincorporated business Public pensions (old-age pension, survivor pension, disability pension) Occupational & private pensions Unemployment benefits Other social transfers (family/children related allowances, housing allowances, education allowances, minimum subsistence, other social benefits) Old-age benefits, Survivor benefits, Disability benefits Pensions from individual private plans Unemployment benefits Family/children related allowances Housing allowances Education-related allowances Sickness benefits Social exclusion not elsewhere classified Regular private transfers Regular inter-household cash transfer Income received by people aged under 16 Income from other sources Source: HFCN (2013) & European Commission The HFCS dataset contains some very interesting features. First, the very wealthy are oversampled such that a better coverage of the top of the income and wealth distributions is obtained. This is necessary because there exist large sampling and non-sampling errors as a consequence of the large skewness of the wealth distribution. In particular the wealthiest households are less likely to respond 6

7 and more likely to underreport, especially in the case of financial assets (Davies et al., 2011). Moreover, it also makes the rather small sample more representative. Hence, in contrast to EU-SILC which should represent the entire income distribution and is used to identify poor households, the HFCS focusses on the top of the distribution (HFCN, 2013a, p.98-99). Since taxes typically have a larger impact on the top of the distribution the implementation of the HFCS in EUROMOD should lead to more accurate outcomes on the distributional and budgetary effects of taxation. The HFCN (2013b, p.21) indicates that this oversampling strategy in some countries comes at the expense of coverage at the bottom of the distribution, but it is not clear to what extent this is the case in practice. As a consequence, the benefit side of the redistributive system may still be better covered by EU-SILC. A second interesting feature of the HFCS data is that a multiple imputation technique was used to deal with selective item non-response (in the form of five different imputations). In other words, crucial income and wealth information does not need to be imputed by researchers in the process of building the database. This imputation is not standardly performed in EU-SILC, implying that the researcher has to make decisions. Moreover, five different imputations will clearly lead to more accurate outcomes than a single imputation. The number of covariates used for the imputation, however, largely differs between countries as well as by income or asset type 3. Moreover, the concrete variables that are used for these imputations are not documented. Therefore, the quality of imputations for individual countries may be hard to evaluate (Tiefensee & Grabka, 2014). The largest added value from using the HFCS data as an underlying database for EUROMOD is that it covers much more detailed information on wealth issues. This will allow the expansion of policy domains currently covered in EUROMOD with taxation of wealth and income from wealth. In the trial database created for this paper, however, this was not yet implemented. We only constructed the database in the same manner as the one based on EU-SILC and as it is currently needed for its inclusion into EUROMOD to get a distribution of disposable income and to measure the redistributive effect of the tax-benefit system. In order to evaluate the suitability of the HFCS as a EUROMOD database we construct a trial database and validate the main outcomes of running EUROMOD on the HFCS, comparing them with those obtained using EU-SILC as input database. The HFCS data potentially supplies micro data on 15 euro area member states. However, the quality and reliability of the HFCS data is not clear yet for all countries. For Belgium an extensive validation of the HFCS data against external data sources such as EU-SILC and SHARE indicates that the HFCS is sufficiently reliable for the study of income and wealth in Belgium (Kuypers et al., 2015). Practical issues in the creation of this database are discussed in the following part. 3 For example, the imputation of missing values of employee income is based on 224 covariates in Spain, while the Netherlands use only 5 variables (HFCN, 2013a, p.51). 7

8 4 The data requirements for EUROMOD and the HFCS: a case study for Belgium Figari et al. (2007) list a set of basic data requirements that a database must fulfil in order to be incorporated in a sensible way in EUROMOD. These are: - The database used must be a recent, representative sample of households, large enough to support the analysis of small groups and with weights to apply to population level and correct for non-response; - The database must contain information on primary gross incomes by source and at the individual level, with the reference period being relevant to the assessment periods for taxes and benefits. When benefits cannot be simulated, information on the amount of these benefits, gross of taxes, is required for each recipient; - The database must contain information about individual characteristics and withinhousehold family relationships; - It must contain information on housing costs and other expenditures that may affect tax liabilities or benefit entitlements; - Specific other information on characteristics affecting tax liabilities or benefit entitlements (examples include weekly hours of work, disability status, civil servant status, private pension contributions) is also necessary; - The same reference period(s) should apply to personal characteristics (e.g. employment status) and income information (e.g. earnings) corresponding to it. In principle this implies the recording of status variables for each period within the year; - There should be no missing information from individual records or for individuals within households. Where imputations have been necessary, detailed information about how they were done is necessary. In general, most of these requirements are met for the HFCS data (see also previous section). We now provide more details on how the HFCS scores on these requirements for Belgium, which is presented here as a pilot exercise. We make use of the UDB 1.1 data version of the HFCS (February 2015 release) on which we construct a new dataset containing the mean estimate over the five imputations for each case where such a multiple imputation was done. We highlight issues of sample size, reference period, imputation of missing information, the disaggregation of certain variables into more detailed information, etc. Sample The UDB data for Belgium include information on 5,506 individuals living in 2,327 households. They were surveyed between April and October 2010, so that the reference income period is The oversampling of the wealthy was implemented in Belgium based on the NUTS 1 region and the average income by neighbourhood of residence, which results in an effective oversampling rate of the top 10% equal to 47 per cent (HFCN, 2013c). As mentioned before, missing information on crucial variables is multiply imputed, so that in principle the full sample can used for the construction of the EUROMOD input database. However, following common EUROMOD conventions, in the creation of the EUROMOD input database children that were born after the end of the income reference period are deleted from the sample. In the HFCS we only know the age of the individual at the time of the 8

9 interview, not the year in which they were born. We assume all individuals aged 0 years to be born after the income reference period. Since most Belgian interviews were done in the second half of 2010 this assumption is relatively acceptable. In case of Belgium it concerns 18 children that are still in their first year of life. Hence, the final sample covers 5488 individuals. Some descriptive statistics used for the grossing up to the level of the full population (10.8 million people) are presented in Table 2. A comparison with those for EU-SILC immediately shows that the HFCS sample is much smaller and therefore its statistical reliability may be lower. Table 2: Descriptive statistics of sample and weights Observations Mean weight Median weight Min weight Max weight HFCS 5,488 1, , ,205.7 EU-SILC 14, ,523.1 Source: own calculations based on HFCS Reference period The HFCS questionnaire asks individuals to declare incomes received in 2009, but all aspects relating to assets and debt holdings as well as demographic and economic characteristics refer to the time of the interview. We have to make the assumption that these aspects have not changed compared to the income reference period. For example, since we only know the age of individuals at the time of interview and not the birth year we cannot just subtract 1 year for age because we do not know whether the person has already celebrated its birthday when the interview takes place. Moreover, we do not know whether an individual has perhaps experienced a change in its labour market status, marital status, etc. We deem it reasonable to assume that the largest share of individuals has not experienced a change in their main demographic and economic characteristics, or that such a change has no large impact on the outcomes. In sum, the practice is basically the same as the one used when deriving an EU-SILC based EUROMOD input database Adjustments of variables With the exception of certain variables, EUROMOD input variables on labour market information, incomes, benefits, etc. need to be covered at the individual level. As in EU-SILC a number of these components are surveyed at the household level in the HFCS. In order to divide these between individuals we followed the same process that was developed for the EU-SILC based input database. The components for which this applies are: - Rental income from real estate property - Income from financial investments - Income from regular social transfers - Income from other sources Important to note is that the EUROMOD variable INCOME: other (yot) in the EU-SILC refers to income received by individuals younger than 16 years, while it refers to income received from other sources in the HFCS. Variables that could not be created using the HFCS as an underlying dataset and which are used in EUROMOD - Belgium are firm size (lfs) for the calculation of employer contributions and Belgian cadastral income of the own residence (khooo) for the calculation of personal income taxes. 9

10 Disaggregation of social transfers In the original HFCS dataset all incomes from regular social transfers (except pensions and unemployment benefits) are covered under one aggregated variable (HG0110), while EUROMOD requires all types of benefits to be covered separately. As this variable includes income sources received both at the individual (e.g. educational allowances) and the household level (e.g. housing allowances, family benefits, ) and are not mutually exclusive their breakdown into separate components is not straightforward. Child benefits and social assistance for Belgium can be accurately simulated in EUROMOD and are considered to be the most important, and also the most widespread among households in the case of child benefits. Therefore, we opted to create a EUROMOD variable BENEFIT: Other (bot) which was set equal to the aggregate reported variable and to simulate the child benefit and social assistance in EUROMOD 4, after which these two values are subtracted from the aggregate variable. As a result of this process we have three output variables: one containing the simulated child benefits, one containing the simulated social assistance benefits and one including all other types of benefits covered in HG0110. Where the simulated benefits turn out to be larger than the reported amounts, we decided first to use the simulated benefits and assume no other benefits when the difference between simulated and declared benefits is smaller than 150 euros per month. Second, those households where no social benefits are declared and there are indications of a reconstituted family, child benefits are set to zero. This is because in Belgium mothers typically receive child benefits so that if the father and child are part of a new household in the dataset no child benefits are received by them. Finally, for the remainder of households the difference between declared and simulated social benefits is relatively large, with a large share of households with dependent children not declaring any social benefits (about 85% of the remaining households) 5. Therefore, we assume for all remaining households that they have forgotten to declare child benefits. We simply make use of the simulated child benefits, while the declared benefit amount, if any, is considered to refer to other types of social benefits. In case of income received from public pensions all types of benefits are also included in one aggregated variable (PG0310). In order to obtain separate EUROMOD input variables for old age benefits, survivor benefits and disability benefits the aggregated variable was imputed. As we assume these three types of benefits to be mutually exclusive, we imputed old age benefits as all pension income received from the age of 65 onwards and survivor benefits as those pensions received under the age of 65 by widowed persons. Finally, disability benefits are all those pensions received by someone who is permanently disabled according to one of its declared labour status. Imputation of main residence mortgages The HFCS dataset covers very detailed information on mortgages held for the main residence, among others the monthly payment that is made. However, EUROMOD requires a specification of the part that is paid in interest and the capital part. First, it should be noted that wealth information in the HFCS refers to the time of the survey, hence We assume that mortgage payments are the same in 2009 as in Mortgages that were taken or refinanced in 2010 are not included. For households that refinanced their loan in 2010 we lose the information on the fact that they did have a mortgage in However, we do not know the specificities such as the old interest rate of this 4 Similarly to the EU-SILC based simulations, the amounts of social assistance are adjusted for non-take-up of benefits with a random non take-up correction. 5 Overall, approximately 46.7% of households with dependent children do not report any social benefits. 10

11 loan. This only involves 8 households. Second, we assume all households to have made a payment during 12 months. This could be a problem if the mortgage was taken or expired in the income reference period. Furthermore, we used the following formula to split the mortgage repayment into an interest and a capital part: interest part = repayment [1 (1 + i) (k n 1) ] Where i refers to the interest rate, n to the duration of the mortgage and k to the time of the mortgage period that already passed. Subtracting this interest part from the repayment amount gives the capital part. We have detailed information on duration and interest rate for the first two mortgages only. For the third mortgage onwards we only have information on the monthly repayment. We opt to apply the parameters from the second mortgage on the payment for the other mortgages. In the Belgian sample 16 households have a third mortgage for their main residence. Missing regional information Unfortunately the HFCS UDB data do not include information on the region households live in. In the construction of the trial database we arbitrarily assumed all household to live in Flanders as this region has the largest population share and this will have the smallest impact on the EUROMOD outcomes. However, the EUROMOD module simulating the Flemish contribution to care insurance was not switched on, because all households would then be eligible to pay this specific contribution 6. In other cases the effect on outcomes is probably smaller as they are applicable in all regions but only differ in level of rates. The lack of NUTS1 information, however, will turn out to be a large problem in the future, because the sixth state reform involves a substantial transfer of tax-benefit competences from the federal to the regional level. 5 Simulating net incomes using the Belgian HFCS data In this part we discuss the outcomes from the EUROMOD process based on the Belgian HFCS data. Moreover, in order to validate these results we compare them to those obtained by the EU-SILC database and where possible to other available sources. We start by analysing how well the HFCS and EU-SILC data represent the Belgian population. Table 3 provides an overview of some basic socio-demographic indicators. Overall, the HFCS and EU-SILC data appear to represent the Belgian population in a similar way. While the age and gender distribution are highly similar, the results for highest education achieved and tenure status are found to diverge slightly more. Relating to the latter, both surveys cover about the same home-ownership rate, but they differ somewhat in their subdivision on mortgage holdings. Comparing the outcomes for both surveys with those of external sources, however, indicates that both samples are not completely representative for the Belgian population. While the HFCS more closely represents the distribution of highest education achieved than EU-SILC, it appears to underestimate the share of self-employment. However, in the HFCS respondents can declare several labour statuses. In our 6 As this contribution amounts to maximum 50 euro per year, the overall effect of this omission is probably negligible. 11

12 definition of labour market status we used only the one that was reported first. It is possible that an individual declares self-employment as the second, third, etc. labour status at the time of interview (2010), but has worked in self-employment throughout the main part of the income reference period (2009) (see discussion of reference period above). Table 3: Comparison of socio-demographic characteristics between HFCS, EU-SILC and EUROSTAT HFCS 2009 EU-SILC 2009 External 2009 Age Gender Female Male Highest education achieved (*) Not completed primary Primary Lower secondary Upper secondary Post-secondary Tertiary N/A (*) N/A 26.1 Labour market status Pre-school Employer or self-employed Employee Pensioner Unemployed Student Inactive Sick or disabled Other Family worker (**) (**) Marital status Single Married Separated Divorced Widowed N/A N/A Tenure status Owned on mortgage Owned outright Rented Reduced rented Social rented Free user N/A N/A N/A (***) 28.8 (***) Notes: outcomes are estimated for the EUROMOD sample, not the survey sample; (*) external data only on persons aged 15 years and over, first figure refers to joint category of not completed primary and primary education; (**) includes all children that are entitled to child benefits; (***) figures for 2011, only breakdown into owned versus rented house is 12

13 available, rest category (2.2%) refers among others to collective residential accommodations which are typically not part of a survey sample Source: own calculations and external sources: age, gender and marital status: EUROSTAT based on CENSUS, education attained: FOD Economics, Department Statistics, labour market status: Data warehouse labour market and social protection of the Crossroads Bank for Social Security, tenure status: CENSUS 2011 Now we move on to the analysis of the distributional outcomes. All figures in this part are computed for individuals based on their household disposable income equivalised by the OECD modified scale 7 and expressed in annual terms, unless mentioned otherwise. Table 4 shows the outcomes of EUROMOD based on the HFCS in terms of income inequality. We show outcomes for disposable as well as original income (including pensions), the difference between them being the inclusion of taxes, social insurance contributions and benefits. The HFCS median is found to be very similar to the EU-SILC median (see also HFCN (2013, p. 100) for a crosscountry evidence), while mean estimates appear to be somewhat higher based on the HFCS, especially for original income. Nevertheless, the comparison of inequality indices requires more attention and further investigations as they show a rather large discrepancy between the outcomes of the HFCS and EU-SILC. As we will discuss below, this will likely be the consequence of the oversampling strategy applied in the HFCS. The Kakwani measure for progressivity is shown in Table 5. As expected due to the oversampling strategy, all components of the tax-benefit system are found to be more progressive in the HFCS compared to the EU-SILC database. Table 4: Comparison of income inequality indicators between HFCS and EU-SILC EUROMOD 2009 based on HFCS EUROMOD 2009 based on EU-SILC EU-SILC incomes 2009 Disposable income Mean 21,995 20,036 21,201 Median 18,977 18,919 19,469 Gini coefficient Income quintile ratio (S80/S20) Original income Mean 29,435 25,247 25,635 Median 22,263 22,638 22,917 Gini coefficient Income quintile ratio (S80/S20) Source: own calculations Table 5: Comparison of progressivity HFCS and EU-SILC Kakwani index EUROMOD 2009 based EUROMOD 2009 based on HFCS on EU-SILC Taxes Social insurance contributions Social benefits Source: own calculations 7 The OECD equivalence scale is constructed by giving the first adult a weight 1, any additional individuals aged 14 years or over 0.5, while individuals younger than 14 count for

14 In Table 6 we show how the impact of the tax-benefit system is distributed across deciles. We find that the difference in inequality is mainly driven by divergence at the top and the bottom of the income distribution. While the average equivalised disposable income in the 10 th decile is equal to 58,545 based on the HFCS, it is only 37,580 for EU-SILC. The difference in average disposable income in the bottom decile is approximately 33% higher in EU-SILC than in the HFCS. Moreover, differences are mainly found with regard to taxes and social insurance contributions, which are typically based on the income level, while outcomes for the benefits that are received are much more similar as eligibility is often based on non-monetary aspects such as the presence of children in order to qualify for child benefits for instance. This again indicates that the difference in outcomes between the two surveys can mainly be attributed to the HFCS oversampling strategy. Table 6: Comparison between HFCS and EU-SILC of averages of different components by decile of equivalised disposable income Disposable Social insurance Original income Benefits Taxes income contributions Decile EUROMOD 2009 based on HFCS 1 6,177 2,657 3, ,215 8,466 3, ,725 12,884 2, ,869 16,683 2,737 1,863 1, ,021 20,463 2,516 2,933 2, ,958 23,844 2,817 4,062 2, ,497 30,071 2,063 6,138 3, ,110 35,073 2,090 7,859 4, ,043 44,248 1,315 11,244 5, , ,382 2,183 34,731 9,289 Total 21,995 29,435 2,559 6,961 3,038 EUROMOD 2009 based on EU-SILC 1 8,235 4,191 4, ,089 9,304 3, ,210 13,134 3,287 1,149 1, ,181 16,704 3,225 2,057 1, ,090 20,184 3,117 3,054 2, ,887 24,459 2,662 4,368 2, ,960 28,485 2,418 5,601 3, ,408 33,583 2,446 7,493 4, ,744 40,812 2,084 10,217 4, ,580 61,668 2,298 19,335 7,051 Total 20,036 25,247 2,994 5,373 2,832 Source: own calculations Table 7 shows a comparison of poverty rates for several poverty thresholds. At each poverty line the share of poor individuals is higher in the HFCS compared to SILC, although the gap decreases slightly at higher poverty thresholds. However, the outcomes are closer to the poverty rates based on reported EU-SILC disposable incomes. It is well-known that disposable income simulated in EUROMOD differs to an important extent from the reported disposable incomes in the UDB EU-SILC data (Hufkens et al., 2014). At the official threshold of 60% of median equivalised disposable income the poverty rate is equal to 15.3% for the HFCS EUROMOD database, 11.7% for the EU-SILC EUROMOD database and 14.6% for the UDB EU-SILC. As was shown in Table 4 the HFCS and EU-SILC 14

15 medians are similar. Hence, the difference in poverty rates can hardly be attributed to a difference in poverty thresholds. However, Table 6 showed that HFCS figures of net disposable incomes are lower at the bottom of the distribution and higher at the top of the distribution compared to the EU-SILC outcomes. This directly impacts the number of individuals below 60% of median income. If we look at the distribution of poverty across age categories in Table 8, it appears that the overrepresentation of poor individuals in the HFCS compared to EU-SILC is not evenly distributed by age. While the number of poor individuals is relatively similar for the two surveys in the age categories of 0-15 and 65+, the poverty rates in the other three age categories are much larger, especially for individuals aged between 16 and 29. Table 7: Comparison of poverty rates at different poverty lines between HFCS and EU-SILC Percentage of individuals below: EUROMOD 2009 based on HFCS EUROMOD 2009 based on EU-SILC EUROSTAT (EU- SILC) 40% of median equivalised 5.4% 2.3% 4.1% disposable income 50% of median equivalised 9.5% 5.5% 7.9% disposable income 60% of median equivalised 15.3% 11.7% 14.6% disposable income 70% of median equivalised disposable income 22.6% 19.8% 23.8% Source: Hufkens et al. (2014) and own calculations Table 8: Comparison of poverty rates by age group between HFCS and EU-SILC Age group: EUROMOD 2009 based on HFCS EUROMOD 2009 based on EU-SILC EU-SILC incomes % 14.8% 18.5% % 11.1% 14.3% % 9.4% 11.4% % 9.5% 11.6% % 15.3% 19.5% Note: poverty line is set at 60% of median equivalised income Source: Hufkens et al. (2014) and own calculations In short, this preliminary validation exercise of a EUROMOD database constructed on the HFCS indicates that outcomes based on simulated disposable incomes are reasonable. This is in line with Kuypers et al. (2015), which include a similar validation exercise for gross incomes comparing HFCS Belgium with the EU-SILC and the SHARE database. There are, however, some remarkable differences which warrant further investigation. The largest discrepancies are found with regard to the level of inequality, which is found to be largely driven by divergences at the top of the distribution, which in turn is assumed to be the consequence of the HFCS oversampling strategy. Kennickell (2008) and Bover (2008) argue that on top of its correction for nonresponse oversampling of the wealthy also provides more precise estimates of wealth in general and of narrowly held assets as standard errors are much smaller. Since the income and wealth distributions are highly correlated, especially at the top (e.g. Alvaredo et al., 2013), oversampling will also result in more accurate estimates of the top of the income distribution as well as of income sources that are typically received by a select group. Therefore, we expect the HFCS to capture the level of inequality more closely to reality than EU-SILC. Vermeulen (2014), however, shows that despite the oversampling strategy wealth shares of the top 5 15

16 and 1% are still underestimated. It is not clear whether this is also the case for the income distribution. Some particular aspects should be borne in mind in the use of the HFCS-EUROMOD database and the interpretation of its outcomes. First, the HFCS sample is considerably smaller than the EU-SILC sample. Therefore one should be careful in interpreting results for small subgroups. Second, an analysis of some socio-demographic characteristics indicated that the sample is not fully representative for the Belgian population. Most importantly the HFCS might slightly underestimate the share of self-employment as main labour status. The largest limitation of the HFCS, however, is the fact that the income reference period and the reference time of other aspects does not coincide. Moreover, the reference period also differs between separate countries, which will complicate crosscountry analyses. 6 Conclusion This paper explores the feasibility of considering the HFCS data as an underlying database for the European tax-benefit model EUROMOD. We created a trial database for Belgium and validated some aggregate results by comparing outcomes to those obtained when EU-SILC is used as underlying database as well as to external databases. These first results indicate that it is feasible to use the HFCS database as EUROMOD input data, despite some of the outcomes need further investigation. The main differences exist with regard to the level of inequality, which is found to be largely driven by divergences at the top of the distribution, which in turn is assumed to be the consequence of the HFCS oversampling strategy. As our discussion above indicated, the oversampling of wealthy households might result in more accurate estimates of income and wealth at the top. Another conclusion from our research is that a comparison of results between EU-SILC and the HFCS cannot be based just on medians alone. It is important to look at the distribution, as our outcomes show that there are some discrepancies at especially the bottom and the top of the distribution. The reasons for these discrepancies should be investigated in more depth. Hence, our preliminary conclusion is that, although transforming the HFCS into a database for EUROMOD would require a significant amount of effort and the simulation results require a detailed scrutiny to assess their reliability against external statistics and results based different input data, this is surely to be worthwhile because of the interesting possibilities to extend the policy scope of EUROMOD and also to consider the joint distribution of disposable income, wealth and consumption. In a future extension of this paper a second trial database for Italy will be constructed. Since the HFCS data for Italy originate from the conversion of an existing national survey (i.e. Survey on Household Income and Wealth (SHIW)) the strengths and weaknesses of these data are well known. Moreover, much more variables are available for Italy, such as imputed rent and net incomes for instance, which will largely contribute to the validation of using the HFCS as an underlying database for tax-benefit microsimulation in EUROMOD. 16

17 7 References Alvaredo, F., Atkinson, A. B., Piketty, T., & Saez, E. (2013). The top 1 percent in international and historical perspective. Journal of Economic Perspectives, 27(3), Bover, O. (2008). Oversampling of the wealthy in the Spanish Survey of Household Finances (EFF). Irving Fisher Committee Bulletin, 28, pp Davies, J. B. (2009). Wealth and economic inequality. In W. Salverda, B. Nolan, & T. M. Smeeding, The Oxford Handbook of economic inequality (pp ). Oxford: Oxford University Press. Davies, J. B., Sandström, S., Shorrocks, A., & Wolff, E. N. (2011). The level and distribution of global household wealth. The Economic Journal, 121(551), Eurosystem Household Finance and Consumption Network. (2013a). The Eurosystem Household Finance and Consumption Survey - Methodological report for the first wave. ECB Statistics Paper No1, 112p. Eurosystem Household Finance and Consumption Network. (2013b). The Eurosystem Household Finance and Consumption Survey - Results from the first wave. ECB Statistics Paper No2, 112p. Figari, F. (2013). Should we make the richest pay to meet fiscal adjustment needs? - Discussion. The role of tax policy in times of fiscal consolidation (pp ). European Economy, Economic Papers 502. Figari, F., Levy, H., & Sutherland, H. (2013). Using the EU-SILC for policy simulation: Prospects, some limitations and some suggestions. Comparative EU Statistics on Income and Living Conditions: Issues and challenges (pp ). Eurostat Methodologies and Working Papers, European Communities. Figari, F., Paulus, A., & Sutherland, H. (2015). Microsimulation and policy analysis. In A. B. Atkinson, & F. Bourguignon, Handbook of Income Distribution Volume 2B. Amsterdam: Elsevier-North Holland. Hills, J. (2013). Safeguarding social equity during fiscal consolidation: which tax bases to use? The role of tax policy in times of fiscal consolidation (pp ). European Economy, Economic Papers 502. Hufkens, T., Spiritus, K., & Vanhille, J. (2014). EUROMOD Country Report Belgium Jäntti, M., Sierminska, E., & Van Kerm, P. (2013). The joint distribution of income and wealth. In J. C. Gornick, & M. Jäntti, Income inequality. Economic disparities and the middle class in affluent countries (pp ). Stanford: Stanford University Press. Kennickell, A. B. (2008). The role of oversampling of the wealthy in the Survey of Consumer Finances. Irving Fisher Committee Bulletin, 28, pp