How large is the homeownership bias in personal income taxation?

Transcription

1 1 How large is the homeownership bias in personal income taxation? Francesco Figari 1,2, Gerlinde Verbist 3 and Francesca Zantomio 4 Institute for Social and Economic Research, University of Essex; 2 Department of Economics, University of Insubria; 3 Centre for Social Policy Herman Deleeck, University of Antwerp; 4 Department of Economics, Ca Foscari University of Venice Abstract Most countries income tax systems entail a favourable treatment of homeownership, compared to rental-occupied housing. Although a removal of this homeownership bias is generally advocated on efficiency grounds, lack of updated comparative evidence on its size, and on its distributional effect, is hampering the current tax policy debate. In this paper we present empirical measures of the homeownership bias inherent in the income tax systems of eight European countries Austria, Belgium, Finland, France, Germany, Italy, Spain and the United Kingdom. We then investigate its distributional pattern and assess to what extent the existing homeownership bias is mitigated by existing property taxation on the main residence. Results show that a sizeable bias in favour of homeowners is embedded in current income tax systems, with existing property taxation generally representing only a partial correction to this bias. Such evidence appears valuable in informing the policy debate on the search for new sources of tax revenues, and in particular for those less detrimental to growth and equity. Keywords Housing taxation; imputed rent; homeownership bias; income distribution; inequality; microsimulation JEL-codes D31, H23, I31, I32 Acknoweledgment: The authors gratefully acknowledge the support, guidance and comments received by Salvador Barrios and European Commission staff throughout the project Microsimulation modelling tool for the analysis of housing tax reforms in the EU (Tender JRC/SVQ/2013/J.1/0008/NC) commissioned by the Joint Research Centre IPTS, Knowledge for Growth Unit (Sevilla), European Commission. We use EUROMOD version G1.0 and we acknowledge the contribution of all past and current members of the EUROMOD consortium and its Director Holly Sutherland. The process of extending and updating EUROMOD is financially supported by the Directorate General for Employment, Social Affairs and Inclusion of the European Commission [Progress grant no. VS/2011/044]. The authors alone are responsible for the analysis and interpretation of the data reported here. We use microdata from the 20 EU Statistics on Incomes and Living Conditions (EU-SILC) made available by Eurostat under contract EU-SILC/2011/ (for Austria, Belgium, Finland, France and Germany), the national 20 SILC data made available by respective national statistical offices (for Italy and Spain) and the Family Resources Survey (FRS), made available by the UK Department of Work and Pensions (DWP) through the Data Archive. Material from the FRS is Crown Copyright and is used by permission. Data providers do not bear any responsibility for the analysis or interpretation of the data reported here. 1

2 1 Introduction The financial and economic crisis has revived interest towards this subject of residential housing taxation, which although discussed since a long time in economics (Aaron, 1970; Rosen, 1979; Poterba, 1992; Turnovsky and Okuyama, 1994), has recently moved into the spotlight in the public policy debate on both sides of the Atlantic (IMF, 2009; Glaeser, 20; Ceriani et al., 2011). After the crisis onset, governments have faced the need to consolidate their public finances, envisaging additional ways of increasing tax revenues. This typically raises challenges on both efficiency grounds (as taxation distorts economic incentives and thus hinders growth), and on equity grounds (as distributional concerns are raised with respect to the protection of the most vulnerable groups of population, already severely hit by the crisis). The crisis has drawn attention on the need to devise policy interventions capable of fostering economic recovery and growth, making it imperative to identify tax bases less detrimental to growth. In this respect, recurrent residential housing is regarded as a potential avenue for raising tax revenues, less harmful to economic efficiency and equity (European Commission, 2012; 2013; Lloyd, 2009; Mirrlees et al., 2011; Pellegrino et al., 2012) than other forms of taxation. In the Public Finance discourse, consensus has been formed on the opportunity of including in the personal income tax base the figurative return homeowners derive from their residence in terms of avoided cost of paying for the residential services they enjoy typically referred to as imputed rent (IR) (Marsh, 1943; Goode, 1960; Musgrave, 1967; Vickrey, 1993) 1. On equity grounds, considering the case of two individuals, identical in all respects (cash income, housing quality, etc.) except for one being an owner occupier and the other being a renter, although endowed with the same cash income, the owner occupier enjoys a higher command over resources, that is a larger potential consumption set, as he does not have to spend part of his budget on buying residential services. On efficiency grounds, owned housing represents a form of capital investment, with its return (i.e. IR) to be taxed as any other form of investment, in order to avoid creating distortions leading to inefficient investment decisions. However, the prevalent policy practice in most western countries departs from these theoretical recommendations, and rather provides for a favourable tax treatment of homeownership, resulting in a so-called homeownership bias. Most notably, in the context of Personal Income Taxation (PIT), IR and capital gains related to owner occupied housing are exempt 2. Besides, many countries provide income tax reliefs for mortgages incurred to purchase owner occupied housing, although more recently several countries are moving towards phasing out of such measures. A vast body of economic literature has underlined how similar tax provisions entail several drawbacks: on efficiency grounds, this lack of neutrality in taxation of different forms of investment (residential housing vs others) lead to the crowding-out of more productive alternative investments, which would be taxed more heavily 1 According to the comprehesive income taxation view, a fair tax base should include all the consumption possibilities that would leave unaltered the initial stock of capital, therefore also non cash incomes, such as imputed rent. 2 In the few countries where IR is subject to income tax, the corresponding notional rents are systematically lower than private market rents (Andrews et al., 2011) 2

3 heavily (Turnovsky and Okuyama, 1994). Also, the favourable tax treatment of residential housing has appeared to result in increased housing prices without necessarily expanding housing opportunity, but rather fostering inflation and volatility (Catte et al., 2004). A link has been established between the favourable tax treatment of residential housing with the excessive households indebtedness and the housing bubble that triggered the crisis in the US, later spread to Europe. On equity grounds, while violating horizontal and vertical equity principles (i.e. taxing the same those endowed with the same command over resources, and taxing more those endowed with higher command over resources), housing tax provisions, and mortgage interest reliefs in particular, have been shown to bear regressive distributional effects (Matsaganis and Flevotomou, 2007a; Andrews and Caldera Sánchez, 2011; Kneller et al., 1999; Johansson et al., 2008; Arnold et al., 2011). The reasons for such debatable practices are partly to be sought in the widespread belief that homeownership generates positive externalities (civic engagement, house maintenance, children s outcome, etc.) (Di Pasquale and Glaeser, 1999), although the literature has not reached consensus on the point, as several authors trace the link to a selection mechanism, with e.g. more civically engaged individuals more likely to become homeowners, pointing rather at homeownership hampering mobility and therefore employment (Bover et al., 1989; Cameron and Muellbauer, 1998; Boeri and Terrell, 2002). More bluntly, housing taxation is highly unpopular: the tax is highly visible as it is typically paid directly by the taxpayer, and existing tax measure often result from political attempts to maximise consensus (Chetty et al., 2009; Wood, 1990; Arnold et al., 2011). While the current policy debate is now considering scenarios of housing taxation reform, a common argument brought against the removal of the homeownership bias regards the lack of taxation on imputed rent as compensated by existing recurrent taxation on immovable property, although whether they should be regarded as substitutes, or rather as complements, remains debatable 3. Besides, when considering the option of removing the homeownership bias, one concern is that income inequality might be adversely affected, for example, in countries where older people live on lower cash incomes than the rest of population. Clearly, lack of systematic evidence on the grounds for such claims is hampering effective tax policy making. The overarching contribution of this paper is to provide novel evidence on the size and distributional effect of the homeownership bias inherent in current European personal income taxation systems, and to confront it with that of existing property taxes levied on the main residence. We consider eight European countries Austria, Belgium, Germany, Finland, France, Italy, Spain and the United Kingdom, who exhibit important variation in terms of current income tax treatment of homeowners (mainly relating to mortgage interest tax reliefs), as well as recurrent property taxation. 3 The ability to pay principle would call again for levying higher taxes from those, like homeowners, who, other things equal, have higher stocks of capital, which is known to be more unequally distributed than income. Residential housing taxation has also occupied a prominent role in the Theory of Fiscal Federalism with respect to the funding of local public goods, as embodying all the desirable features a local tax should have (coherence with the benefit principle of paying for the local services received; immovability of the tax base, hindering tax exportation, tax avoidance and tax evasion; distribution throughout the country). 3

4 In order to measure and assess the distributional effects of existing housing tax policies, we simulate counterfactual scenarios using a fiscal microsimulation approach (Bourguignon and Spadaro, 2006; Figari, Paulus and Sutherland, 2014). Microsimulation allows to estimate households and individuals disposable incomes resulting from current and alternative tax policy systems, considering not only housing taxation rules, but the overall set of tax and benefit rules. To carry out the microsimulation, we use EUROMOD, the multi-country European wide tax-benefit microsimulation model. As part of this contribution, we extend EUROMOD to include as far as the underlying micro data allow - property taxes in order to have a more complete overview of homeownership taxation. The underlying data reflect the population and housing market characteristics as observed after the onset of the financial and economic crisis that profoundly changed housing markets in Europe. The rest of the paper is organised as follows. Current housing taxation policies in the eight countries we consider are reviewed in the following section 2. Section 3 describes the methodology and provides details of its implementation (with further detail provided in an Appendix). Results are presented in Section 4, and preliminary conclusions drawn in Section. 2 Housing taxation policies in eight European countries In this section, we illustrate the current 4 residential housing tax policies in the eight European countries we consider: Austria, Belgium, Finland, France, Germany, Italy, Spain, United Kingdom. The choice of countries is motivated by the need to represent heterogeneous settings (Anglo-Saxon, central European, Mediterranean, Nordic ). We focus on the treatment of the main residence and cover homeownership bias provisions on the one hand (IR exemption and mortgage interest reliefs under PIT), and recurrent residential housing taxation 6 7 provisions on the other hand 8. A summary of each country policy in reported in Table 1. It is striking that in no country imputed rent is subject to PIT. This creates a notable lack of horizontal equity between homeowners and renters endowed with the same pre-tax potential 4 As of 2012 Eastern European countries are characterised by very limited private rental markets, and as such were not suited to this analysis as no imputed rent estimates could be derived. 6 We use here the term Property tax to refer to all other forms of recurrent housing taxation, although in some cases (e.g. see France and the UK), rather than a property tax, a residential services tax is levied from both owners and renters, and as such cannot be regarded as an actual property tax. 7 We also look at a few other aspects describing the overall treatment of housing under the national tax system such as whether a rent relief is allowed under PIT for renters of the main residence, the tax treatment of income from other properties under PIT, the presence of taxes on housing transactions, and taxation of capital gains on the primary residence. 8 Sources include European Commission Reports, the OECD, the European Taxation Database, the International Bureau of Fiscal Documentation Handbooks, EUROMOD Country Reports, and a variety of national sources, such as relevant Ministries websites. 4

5 consumption set, with renters ending up paying higher PIT, other things equal 9. Only in Italy and Spain (since 2009) we find a partial correction of the inequity, as in both countries renters are allowed a partial deduction of the rent paid from the PIT tax base. However, these reliefs are strictly limited in amount and concern only lower income households. Table 1: Overview of housing taxation: main characteristics tax treatment of own residence Austria Belgium Finland France Germany Italy Personal Income Tax Property taxation Transfer MI relief IR Type taxes tax, Limited amount allowed for low incomes, among most generous in EU, although being limited abolished in 20 (tax credit maintained for second properties) Real Estate Tax on standard value of property Real Estate Tax (regional and municipal rate on Revenues (as % of GDP) Latest values assessment Tax on capital gains primary residence Cadastral Income) Real Estate Tax (municipal tax) 0.64 Real Estate Tax (municipal rate on Cadastral Income) Real estate tax on standard value of property Real Estate Tax (municipal tax) (if realised within first years from acquisition) Spain United Kingdom Mortgage tax credit removed since 2012 (still compensation for house acquired before 2006). (after lengthy phasing-out), maintained on other taxed properties. Real Estate Tax and Net Wealth Tax (200 in Wales) (but full rollover relief available) As such, these measure cannot be regarded as a full correction for the inequity with respect to homeowners, which enjoy a full deduction of IR irrespective of their overall pre-tax income, 9 According to previous work (Figari et al., 2012), the difference in average tax rate between otherwise identical (same pre-tax potential consumption) owners and renters spanned from 1 to 3 percentage points in a set of six European countries. The source of the reported figures is the Taxes in Europe database and refers to the latest available figure (2011/2012), except in the case of Italy, where it has been calculated from official national tax revenues data.

6 so even if reaching the top marginal tax rates. Therefore, in all countries IR exemption creates an homeownership bias that is likely to lead to overinvestment in housing. As to mortgage interest tax reliefs under PIT, there appears to have been a tendency to reduce them. MI reliefs have been entirely abolished in Spain (2012) and France (20), reduced in Finland (since 2012; previously the country provided one of the most generous reliefs) and reformed in Belgium (since 200). Italy is the only 11 country, among those providing for a relevant MI relief (in Italy the maximum corresponds to more than 700 euro per year returned to the taxpayer), which has made no attempt at reducing it in recent years, despite its wellestablished regressive nature. Homeowners reliefs such as the MI one, represent an additional component adding to the homeownership bias, on top of the IR exemption, so their entire removal would anyway not suffice to cancel the homeownership bias in PIT taxation. A political argument against IR taxation builds on the existence of additional tax burden for homeowners stemming from recurrent residential housing taxation other than under PIT. In facts, in all the countries we consider recurrent housing taxation is provided for in the form of real estate taxation, that is an actual property tax. The UK is the only country where no property taxation of homeownership is present, as recurrent housing taxation hits both homeowners and renters in the form of a residential service tax. Recurrent housing taxation typically plays a role, although limited, in local taxation, with tax bases defined at the central level, and local governments having some autonomy in setting tax rates (often within centrally established lower and upper boundaries) and obtaining (at least some portion of) tax revenues. Although it is debatable that real estate taxation provides a substitute for IR taxation under PIT, thus correcting the homeownership bias, it appears that the reviewed countries choose to tax owner occupied housing only though the first type of taxation (except, as noted, the UK, where the homeownership bias is reinforced by the absence of property taxation 12 ). This choice has indeed advantages (for example, given PIT tax evasion widespread in some countries) but might not suffice to correct the homeownership bias; also, on equity grounds, real estate taxes are typically of proportional, rather than progressive, nature. Recurrent housing taxation varies a lot with respect to the amount of revenues raised. To assess the weight of housing taxation in each country, it is important to consider the integration of different housing tax measures, and also of housing tax provision with other components of the tax benefit system. For example, Austria and Germany display very light housing taxation, but no (or very limited, in the case of Austria) mortgage interest relief is granted to homeowners. The UK has removed homeowners reliefs, but a relevant homeownership bias still stems from both IR exemption and the lack of a property tax, as the Council Tax hits both homeowners and renters. Countries where housing taxation is heavier (UK, France), are also those who appear more developed in the use of tax exemptions and reductions (France) and benefits (the Council Tax Benefit in the UK) to lift the housing tax burden for the income poorest households. 11 The Austrian one is extremely limited in amount and coverage. 12 In Italy, potential correction of the homeownership bias is being compromised by the reform prospects of exempting the main residence from real estate taxation, in contrast with other European countries practices. 6

7 The assessment of housing values appears quite out of date in all countries, with some exceptions, most notably Finland, that, reflecting a feature of other Nordic Countries, provides for regular values re-assessments. In a few countries, interesting safeguard mechanisms are in place, to protect individuals who might be assets rich but income poor. These are the countries where a Wealth Tax (with partial exemptions for main residence) is also in place. It happens in France, where since 2012 the total tax burden cannot exceed the 0% of taxable income, and in Spain, where the sum of the Personal Income Tax and the Wealth Tax cannot exceed 60% of Taxable Income. Such mechanisms represent a promising avenue for enlarging the role of housing taxation while safeguarding the most vulnerable in societies. In what follows, our focus will be on considering, as potential compensation to the homeownership bias, taxation of homeowners, rather than on taxation based on residence (that would hit both homeowners and renters), which was observed in a few of the reviewed countries (UK, France, Italy). Table 2 provides a summary of the principles inspiring the two forms of recurrent housing taxation: residence taxation does not (and is not meant to) correct for the homeownership bias, neither in terms of equity (equal treatment of samecomprehensive-income owners and tenants) nor in terms of efficiency (neutrality w.r.t other capital income tax treatment) as it is rather inspired by the benefit principle according to which the amount paid in taxation should reflect the benefit received from public provisions of local goods. Table 2: Comparison of property and residence service taxation PROPERTY TAX To be paid by owners only Ability to contribute principle (equity in contribution, w.r.t to others) - could represent both Comprehensive income taxation and Wealth tax Aims at achieving neutrality w.r.t other forms of investment and correcting the homeownership bias Might be proportional or progressive, according to how other capital income is taxed RESIDENCE SERVICE TAX To be paid by resident owners and tenants Benefit principle (equity in exchange) No concern for neutrality w.r.t taxation of other forms of investment, no correction of the homeownership bias To be designed as progressive only if higher income individuals are regarded as those benefitting more from local services provided 7

8 3 Methodology and simulations 3.1 A measure of net imputed rent From a theoretical viewpoint, following Haig (1921) and Simons (1938), an appropriate comprehensive income tax base should reflect both monetary and non-monetary potential consumption opportunities. Imputed rent is one of the non-monetary income components that enhances homeowners potential consumption because they benefit from housing services they would otherwise need to pay for, thus depleting cash resources. Imputed rent can, therefore, be regarded as a component of the tax base; treating it differently from other types of income in defining the tax base is part of the homeownership bias. In order to estimate the extent of this bias an estimation of the rental value of own residence (imputed rent) is needed.. Various approaches can be used to calculate imputed rent on the basis of micro data. A general description of these can be found in Frick and Grabka (2003), Frick et al. (2007) and Frick et al. (20). They propose three methods: 1) the opportunity cost approach; 2) the capital market approach, and 3) the self-assessment approach. We use here the opportunity cost approach (also known as the rental equivalence method) to estimate the opportunity cost of housing in a non-subsidized rental market (see Frick and Grabka, 2003; Frick et al., 2007). We use a hedonic regression method in two steps. First, a regression is run on the population of private market tenants with rent as the dependent variable, using a wide range of variables concerning characteristics of the dwelling, the neighbourhood, occupancy etc. as covariates. Second, the resulting coefficients are applied to the owner-occupiers sample in order to get an estimate of imputed rent. As selection may occur in the owner status, this approach can be extended by including a correction for selection bias (e.g. through a Heckman procedure). We decide to provide our own estimates of imputed rent, as EUROSTAT documentation (the internal EUROSTAT document Countries experience: Imputed Rent (HY030G) ) indicates that it is not clear for all countries how the estimations are done. Juntto and Reijo (20) and Törmälehto and Sauli (20) point to some important issues of comparability for EU-SILC In their analysis of the distributive impact of imputed rent in EU-SILC, Törmälehto and Sauli (2013) indicate that there is still lack of transparency for more recent waves of SILC. Moreover, it is not clear what is meant by net imputed rent. The variable HY030N (net imputed rent) is not empty and different from HY030G only in France; consequently in all other countries this variable is either empty or equal to the gross amount. In Germany and Spain, the net variable is the only one that is filled in, while gross is empty. As Törmälehto and Sauli (2013) indicate, it is not clear what is actually meant by net : it can be net of actual costs borne by the occupant, or it can be net after taxes in cases where imputed rent would be taxed. As imputed rent is not taxed in the countries we study, the first issue is applicable; however, it is not clear which actual costs borne by the occupant are deducted. Due to lack of transparency, conclusive evidence concerning comparability of the imputed rents data in SILC cannot be reached neither with the available set of variables nor with the available information given in the national quality reports (Törmälehto and Sauli, 2013, 9). Hence, we will produce own estimates of both gross and net imputed rent for the eight countries 8

9 considered in this study, adopting a consistent methodology. Until now no studies have compared the imputed rent estimates available in EU-SILC with other sources (such as the estimates that will be presented in this paper); by providing estimates with a transparent and coherent methodology across countries, we hope to contribute to a better understanding of the procedure to estimate imputed rent. Table 3: Share of owners and private market tenants, number of households, EUROMOD input datasets, 20 Number of owners Number of private market renters Share of private market renters in total number of households AT 3,3 1, % BE 4,19 1, % DE 6,6, % ES 11,273 1, % FI 8,461 1, % FR 7,093 2, % IT 14,264 2, % UK 16,63 2, % Source: own calculations on EU-SILC (and FRS for UK). Number of cases refer to households, unweighted. As a threshold for using the opportunity cost approach, a minimum of % private market tenants in the population is recommended by EUROSTAT. Most of the countries we study here are above this threshold (more than 20% in AT, BE, DE; more than % in FR, IT and UK), while Spain is with 8% below it, and Finland is just around this threshold (see Table above). In order to derive a net imputed rent measure we deduct interest payments for owners with mortgage. The estimates are done on the EUROMOD input datasets of. This is the EU-SILC 20 version 1 (of 01/03/2012) for Austria, Belgium, Germany, Finland, France and Italy. EUROMOD input data for Italy is based on the National SILC, but as the personal and household identifiers are the same as in the UDB EU-SILC, we use the latter to estimate the imputed rent. Also for Spain EUROMOD input data are based on the National SILC. Because of differences in the identifiers, this dataset is the used to estimate imputed rent (instead of the UDB EU SILC). For the UK the FRS (Family Resources Survey 2009/) is used. For our eight countries we apply a hedonic regression estimation of the logarithm of rent (i.e. variable HH060, excluding all costs) actually paid by tenants on the private housing market (so excluding social housing and any reduced rent payments). In the case of UK, the estimation sample has been restricted to private tenants holding assured short-hold letting agreements (about % of private tenants), deemed most representative of the private rental market prices we want to capture. As covariates we use the ones that are traditionally used in the literature for estimating IR, and that refer to to type and size of the dwelling, quality of the dwelling and the neighbourhood, occupancy in years, geographical location, household income, etc. A Heckman selection correction is applied to correct for potential selection into the owner status. Selection may arise from substantial differences in terms of housing quality and other dwelling characteristics between private market tenants (the donors for the estimation) and 9

10 homeowners (the recipients ). We include in the Heckman regression a number of variables that may play a role in the selection. These are indicators of spatial segregation, or other factors that may cause a difference in e.g. quality of the house. Our estimations show that there is indeed selection bias in all countries (except Belgium and Spain). Next, the resulting coefficients are applied to the owner-occupiers sample in order to estimate their imputed rent. In order to maintain the proper variance in the resulting estimates of imputed rent, we add an error term to the predicted value of imputed rent (Frick et al., 20). This is done by adding an error component, randomly chosen from a distribution characterised by zero mean and variance set based on the difference between the standard deviation of the actual rent variable and the standard deviation of the predicted imputed rent variable (for tenants). The monthly amounts are included in the EUROMOD input file. In order to arrive at net imputed rent we deduct interest payments for owners with mortgage (due to lack of information, other costs cannot be deducted). For some households this results in negative imputed rents (see Table 4). For the simulations, negative net imputed rents can be put to zero, or can be included as a deduction from taxable income. Table 4: Number of cases with positive & negative imputed rent estimates, homeowners only Gross IR Positive (>0) Negative (<0) Net IR(no Net IR (with Net IR (no error error error correction) correction) correction) Net IR (with error correction) AT 3,3 3,20 3, BE 4,19 3,980 3, DE 6,6,9, ES 11,273,84, FI 8,461 8,419 8, FR 7,093 6,7 6, IT 14,264 13,74 13,712 2 UK 16,63 16,368 16, Source: own calculations on the basis of EU-SILC/SILC/FRS. Given that our methodology is transparent and coherent across countries, we use the estimates provided here for the simulations on taxing imputed rent. Moreover, we think that the variable with the error correction term might be the most appropriate, given that the variance is increased, thus allowing for more variation in the estimates. Simulations using the SILC variable or the variable without error correction can be used for sensitivity checks. 3.2 EUROMOD: a multi-country tax benefit model In order to assess the measure the homeownership bias and the impact of alternative tax policy options we simulate counterfactual scenarios by using a fiscal microsimulation approach which allows us to estimate household incomes under different tax options holding everything else constant and therefore avoiding endogeneity problems (Bourguignon and Spadaro, 2006). Baseline and counterfactual scenarios are implemented by using EUROMOD, the multi-country European wide tax-benefit microsimulation model. EUROMOD is a static model that provides measures of direct taxes, social insurance contributions, cash benefits as well as market incomes in a comparable way across countries.

11 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 taxbenefit policies and in understanding how tax-benefit policy reforms may affect income distribution, work incentives and government budgets in the short term. The policy reform simulations are performed on the EUROMOD input data (based on EU- SILC 20 and the Family Resource Survey 2009/ for the UK). In case of France and Spain, additional variables with information on current property tax values have been added from the national SILC data released by the respective National Statistical Institutes. Table : Overview of Policies coverage in EUROMOD MITR IR tax Property tax Rent tax Taxation income relief other properties Tax Tax Tax Tax Tax EM EM EM EM code code code code code EM Austria Limited No No - Yes No No - Yes S* Belgium Yes S No - Yes S No - Yes S Finland Yes S No - Yes I No - Yes S France No - No - Yes I No - Yes S Germany No - No - Yes I No - Yes S Italy Yes S No - Yes S Yes S* Yes S Spain Yes S No - Yes I Yes S Yes S United Kingdom No - No - Yes I No - No - Note: S = simulated standardly in EUROMOD; * Simulation refined for this research; I = included (i.e. information from input data). The baseline tax-benefit systems simulated refer to Appropriate price and income indices are used to update monetary variables from income reference period (i.e. 2009) to policy system year (i.e. 2012). The simulations of these policy systems have been crosschecked with administrative statistics and tested through a number of other applications (e.g. Bargain, 2007; Dolls et al., 2012; Bargain et al., 2014). In Spain, the regional dimension of housing tax policies is covered up to the point allowed by the input data: the regional "renting tax credit" is simulated. With respect to "Transfer taxes" and "Tax on capital gains primary residence, these are out of the scope of EUROMOD simulations because the underlying data do not include data neither on transfers of property nor on realised capital gains. 3.3 Measuring the homeownership bias As summarised in Table 6 below, the current treatment of housing in personal income taxation is somewhat disconnected from economic principles, which would recommend the inclusion of imputed rent (IR) in the personal income tax base, and the absence of Mortgage 11

12 Interest Tax Reliefs (MITR) on both equity (i.e. the comprehensive income taxation view) and efficiency grounds (i.e. avoiding the homeownership bias in investment decisions and the debt bias in financing decisions). In all of the reviewed countries though, IR is exempt from PIT. Also, some countries provide for a relief in mortgage interest, typically in the form of a tax credit (except for Belgium, where a deduction is allowed), although there seems to be a general trend of reduction in these provisions (again with the exception of Belgium). The fact that in all countries some form of recurrent housing taxation is found could be regarded as a correction to the homeownership bias, if property taxation were considered as a substitute to IR taxation under PIT. A different perspective though could regard the two forms of taxation as complements, rather than substitutes, with recurrent property taxation representing a form of wealth tax. Table 6: Overview housing tax policies for 8 EU countries PIT IR tax PIT MI relief Housing taxation Latest values assessment Austria, (very limited) Real Estate Tax (Property) 1973 Belgium (reformed since 200) Real Estate Tax (on Cadastral 197 Income, de facto Property tax) Finland, among most generous in EU, Real Estate Tax (Property) 2011 although being limited France abolished in 20 Both Property Tax and 1970 Residence Tax Germany Real Estate tax (Property) 1964 Italy Real Estate Tax (Property municipal tax), moving towards Residence Service tax for main residence Spain From 2012 onwards mortgage tax credit is removed, although compensation remains for those who acquired before January 20th, United Kingdom Real Estate Tax (Property) and Net Wealth Tax (phased out since 1974) Hybrid Council Tax (Residence Tax) (200 in Wales) We now asses the homeownership bias in personal income taxation by its different components and to what extent this bias is mitigated by the property tax. For this purpose we need to have a benchmark of a non-biased system. We consider two options for the definition of a non-biased system. A tax system is not biased in favour of homeowners when 1) imputed rent is tax exempt, there are no provisions for mortgage interest tax relief, there is no property tax and rent paid is fully deductible; or 2) imputed rent is taxed, there are no provisions for mortgage interest tax relief, there is no property tax and any existing rent rebates are removed. 12

13 4 Estimates of the homeownership bias in personal income taxation Figure 1 presents a measure ofthe homeownership bias according to the second definition, and to what extent this bias is mitigated by property taxes. The blue bars represent the homeownership bias and is measured by simulating the inclusion of imputed rent in taxable income and the abolition of the mortgage interest tax relief. These blue bars show that the current system implies a favourable treatment of owners vs renters, and thus an implicit redistribution of resources towards owners on average and along the income distribution. On average the bias represents more than % of household disposable income in Belgium, Finland, Italy and the United Kingdom; it is very small in France. If an unbiased PIT were in place the poorest homeowners would have to pay a relatively larger extra burden in Finland, Italy and UK; this implies that the homeownership bias is progressive, i.e. it favours the poorest homeowner tax payers. Removing the bias in this way would have a progressive (i.e. pro-poor) effect in Austria, Belgium, France, Germany and Spain. The fact that patterns are so different across countries result from differences in housing markets on the one hand and differences in the current personal income system on the other. Removing the rent rebate is only relevant for Italy and Spain, and the impact is very small (i.e. a small increase in the tax burden in both countries). 13

14 Tax burden as % of HH DPI Tax burden as % of HH DPI Tax burden as % of HH DPI Tax burden as % of HH DPI Tax burden as % of HH DPI Tax burden as % of HH DPI Tax burden as % of HH DPI Tax burden as % of HH DPI Figure 1: Homeownership bias in eight EU countries Fig Austria Belgium France ure Total Decile groups based on eq. DPI Total Decile groups based on eq. DPI Total Decile groups based on eq. DPI Finland Germany Italy Total Decile groups based on eq. DPI Total Decile groups based on eq. DPI Total Decile groups based on eq. DPI Spain UK Unbiased PIT, owners Total Decile groups based on eq. DPI Total Decile groups based on eq. DPI Property tax, owners Unbiased tax system, renters Unbiased tax system, owners vs renters 14

15 The property tax (green bars) mitigates to some extent the homeownership bias, but with the exception of France (and to a lesser extent Spain), this is rather (or very) limited. In addition, the distributional pattern of current property taxes tends to be quite regressive in all countries, and in particular in the countries where its weight is the highest (FR and ES). The blue line is the net tax burden change to be paid by owners to get a horizontal neutral system with respect to renters. If an unbiased tax system were in place (and considering the property tax as a compensation for the PIT bias), then the poorest homeowners would have to pay an extra burden that is relatively large in Finland and the UK. Homeowners would gain (i.e. paying less taxes) in France and also Spain the first two quintile groups due to the relatively large property tax. In the other countries, the richest homeowners would have to pay a relatively larger extra burden. In Italy the pattern is quite flat. Next steps in the empirical analysis include: - The decomposition of the homeownership bias over impute rent exemption and mortgage interest tax relief - Measurement of the homeownership bias according to the alternative definition (i.e. imputed rent is tax exempt, there are no provisions for mortgage interest tax relief, there is no property tax and rent paid is fully deductible) Conclusion Most countries income tax systems entail a favourable treatment of homeownership, compared to rental-occupied housing. Although a removal of this homeownership bias is generally advocated on efficiency grounds, lack of updated comparative evidence on its size, also in relation to existing forms of recurrent housing taxation levied from homeowners, and on its distributional effect, is hampering the current tax policy debate. In this paper we have considered eight European countries Austria, Belgium, Finland, France, Germany, Italy, Spain and the United Kingdom, observed after the onset of the recent crisis. Building on upto-date estimates of imputed rent, whose exemption from personal income taxation represents the major contributor to the existing homeownership bias, and using the multi-country tax benefit model EUROMOD (applied to data drawn from the Survey of Income and Living Conditions and UK Family Resources Survey), we have presented an empirical measure of the homeownership bias inherent in these countries income tax systems. We have investigated its distributional pattern (i.e. first round distributional effect) and assessed to what extent the existing homeownership bias is mitigated by existing property taxation on the main residence. Results show that a sizeable bias in favour of homeowners is embedded in current income tax systems, with existing property taxation generally representing only a partial correction to this bias. Such evidence appears valuable in informing the policy debate on the search for new sources of tax revenues, and in particular for those less detrimental to growth and equity.

16 References: Avram S., Figari F., Leventi C., Levy H., Navicke J., Matsaganis M., Militaru E., Paulus A., Rastrigina O. and Sutherland H., 2013, The distributional effects of fiscal consolidation in nine EU countries, EUROMOD Working Paper EM2/13. Bourguignon F. and Spadaro A., Microsimulation as a tool for evaluating redistribution policies, Journal of Economic Inequality 4(1): Figari, F., A. Paulus, H. Sutherland, P. Tsakloglou, G. Verbist, and F. Zantomio, 2012, Taxing home ownership: distributional effects of including net imputed rent in taxable income. EUROMOD Working Paper EM4/12 Figari, F., A. Paulus, H. Sutherland, 2014, Microsimulation and Policy Analysis in Handbook of Income Distribution Volume 2, edited by A. B. Atkinson and F. Bourguignon, Elsevier, forthcoming. ISER Working Paper 23/2014 Frick, J.R., Grabka, M.M., Imputed Rent and Income Inequality: A Decomposition Analysis for the UK, West Germany and the USA. Review of Income and Wealth, 49(4), Frick, J.R., Goebel, J., Grabka, M.M., Assessing the distributional impact of imputed rent and non-cash employee income in micro data: Case study based on EU-SILC (2004) and SOEP (2002). SOEP Paper 2, DIW, Berlin. Frick, J.R., Grabka, M.M., Smeeding, T., Tsakloglou, P., 20. Distributional effects of imputed rents in five European countries. Journal of Housing Economics, 19(3), Haig, R.M., The Concept of Income Economic and Legal Aspects, in: R.M. Haig (ed.) The Federal Income Tax. Columbia University Press, New York. Juntto, A. & M. Reijo, 20. The comparability of imputed rent in EU-SILC 2007 differences in variable definitions and methods concerning institutional housing, EUROSTAT Methodologies and Working Papers, Luxembourg: Publications Office of the European Union. Matsaganis, M, Flevotomou M., 2007, The impact of mortgage interest tax relief in the Netherlands, Sweden, Finland, Italy and Greece, EUROMOD Working Paper 2/07. OECD, 20, Annex A. Revenue Forgone Estimates of Main Tax Expenditures in OECD Countries, in Choosing a Broad Base Low Rate Approach to Taxation, OECD Publishing. Saunders, P., Smeeding, T., Coder, J., Jenkins, S., Fritzell, J., Hagenaars, A., Hauser, R., Wolfson M., Non-cash Income, Living Standards, Inequality and Poverty: Evidence from the Luxembourg Income Study, Discussion papers 3, Social Policy Research Centre (SPRC), University of new South Wales. Simons, H., Personal Income Taxation: the Definition of Income as a Problem of Fiscal Policy. University of Chicago Press, Chicago. Sutherland, H. and F. Figari, EUROMOD: the European Union tax-benefit microsimulation model. International Journal of Microsimulation 6(1), Törmälehto, V-M & Sauli H. (2013). The distributional impact of imputed rent in EU-SILC Eurostat Methodologies and Working Papers 2013 edition. Luxembourg. 16