Urban Studies, Vol. 42, No. 1, 127– 140, January 2005
Renter Illusion: Fact or Fiction? Jens Blom-Hansen [Paper first received, January 2004; in final form, July 2004]
Summary. The renter illusion hypothesis holds that renters underestimate their property tax burden and therefore support excessive levels of local expenditure. A number of studies have reached results that are consistent with the hypothesis. But so far empirical investigations have proceeded from the assumption that renter illusion is a fact and have investigated the implications of this fact. No-one has investigated renter illusion as such. In other areas of fiscal illusion, survey analyses of voter perceptions have yielded considerable insight into the validity of illusionary hypotheses. This paper follows this method and investigates a representative sample of 1000 Danish renters’ and home-owners’ perception of the local property tax system. The evidence raises doubts about the renter illusion hypothesis. The survey findings are corroborated by an analysis of property taxation in Danish municipalities.
Introduction The property tax is often hailed as one of the most visible taxes since it is paid directly by the taxpayer at regular intervals. But there is an important reservation: renters. In most systems, property tax bills go to property owners, not to occupants, so renters never see the exact amount of property tax assessed on their residence. This does not mean that renters avoid the burden of property taxes. Property taxes on rental units are often shifted onto renters. But it means that for renters the property tax is as invisible as, say, a VAT hidden in the price of commodities. This fact may cause renters to underestimate their true tax price and to support higher levels of local expenditure than home-owners who pay property taxes directly. This is a kind of fiscal illusion known as ‘renter illusion’ (Oates, 1988). Several studies find that jurisdictions with a large number of renters spend more on
local public services and have higher tax levels. A range of other studies finds that renters desire more public services than home-owners. The renter illusion hypothesis thus seems supported by empirical evidence. However, studies of renter illusion have so far focused on the implications of illusion— i.e. they proceed from the assumption that renter illusion is a fact. The validity of the assumption itself has not been investigated. In other areas of fiscal illusion, survey analyses of voter perceptions have yielded considerable insight into the validity of illusionary hypotheses (May, 1982; Sørensen, 1992; Winter and Mouritzen, 2001). This paper follows this approach and investigates a representative sample of Danish renters’ and home-owners’ perception of the local property tax system. The evidence suggests that the renter illusion hypothesis may be difficult to sustain.
Jens Blom-Hansen is in the Department of Political Science, University of Aarhus, Denmark. Fax: þ45 8613 9839. E-mail: [email protected]
Helpful comments from the following persons on earlier versions of the paper are gratefully acknowledged: Jørgen Grønnegaard Christensen, Mark Hallerberg, Rune Sørensen, Jørn Rattsø, Niels Jørgen Mau Pedersen, Hans Nikolaisen, Søren Serritzlew, Lotte Bøgh Andersen, Anne Binderkrantz and five anonymous referees of Urban Studies. All remaining flaws are the author’s responsibility. 0042-0980 Print=1360-063X Online=05=010127 –14 # 2005 The Editors of Urban Studies DOI: 10.1080=0042098042000309739
The validity of this survey result is explored in an analysis of property taxation in Danish municipalities. This analysis follows the traditional approach to renter illusion—that is, it investigates whether the fraction of renters in a jurisdiction has an upward effect on local budgets. This is found not to be the case. The analysis thus corroborates the survey results. The paper is structured as follows. The next section introduces in greater detail the phenomenon of renter illusion. The third section provides a brief introduction to local property taxation in Denmark which is the setting for the ensuing two empirical analyses. The fourth section investigates the actual extent of renter illusion by survey data of Danish renters’ and home-owners’ perceptions of the property tax system. The fifth section analyses whether the proportion of renters has an upward effect on property taxation in Danish municipalities. The paper ends with a brief conclusion. Renter Illusion The concept of fiscal illusion refers to a systematic misperception of the costs and benefits of public programmes. Fiscal illusion is thus more than fiscal ignorance, which may be understood as imperfect information without any particular bias. Fiscal ignorance is a necessary, but not sufficient, condition for the existence of fiscal illusion (Buchanan, 1967, ch. 10). Although fiscal illusion, in principle, may result in both overestimation and underestimation and occur in relation to both revenue systems and expenditure programmes, it is the systematic underestimation of tax burdens that has received the greatest attention in the literature. The danger of this kind of fiscal illuson is that, if the real costs of public programmes are systematically underestimated, then voters are induced to support an excessive size of public sector (Oates, 1988; Dollery and Worthington, 1996). In the local public finance literature, much attention has been paid to a specific type of fiscal illusion known as renter illusion. The
basis of this kind of fiscal illusion is that property taxes—an important revenue source in most local government systems—are not levied on all voters, but only on owners of local property. Renters do not receive property tax bills. This does not mean that renters avoid the burden of property taxes since property taxes on rental units are often shifted onto renters. But renters do not face the same visible tax price that confronts property owners. The true costs of local public goods may thus be concealed from renters who, consequently, may be led to support excessive levels of local expenditure. A considerable number of studies have reached results that are consistent with the renter illusion hypothesis. They are summarised in Table 1.1 The table shows that renter illusion has been investigated from two approaches. The first, and dominant, approach is to analyse the relationship between the number of renters in a jurisdiction and the spending or tax level in this jurisdiction. This is the approach followed by Barr and Davis (1966), Bergstrom and Goodman (1973), Lovell (1978), Gronberg (1980), Beck (1984), Brazer and McCarty (1987), Schwab and Zampelli (1987), Heyndels and Smolders (1994), Dollery and Worthington (1999) and Gemmell et al. (2002). Most of these studies find that, other things equal, jurisdictions with a relatively large fraction of renters tend to spend or tax more. The second approach is to analyse renter preferences for example, as revealed in local referenda. This is the approach followed by Hanushek (1975), Peterson (1975), Martinez-Vazquez (1983), Schokkaert (1987) and Moomau and Morton (1992). Most of these studies find that renters tend to be more in favour of increased spending than home-owners. All in all, although the empirical evidence is not unequivocal, it is consistent with the renter illusion hypothesis in general. However, the evidence may also be consistent with other hypotheses. Martinez-Vazquez (1983) argues that renter support for expanded budgets may be quite rational behaviour. First,
Table 1. Chronological survey of studies of renter illusion
Measure of renter illusion
64 county governments of Pennsylvania
Expenditure per capita
Percentage of voters owning property
Bergstrom and Goodman (1973)
826 municipalities located in 10 US states
Percentage of local property occupied by owner
Two referenda on expenditure increases in Cleveland Household voting for tax and expenditure proposals in US school districts 136 Connecticut towns
Probability of voting for expenditure increase
Percentage of homes owner-occupied
Desired level of school spending
Percentage of renters among the adult population
Proportion of homes owner-occupied
83 Chicago suburbs
Total municipal expenditure
Percentage of homes owner-occupied
Referendum on expenditure increases in St Louis 219 California municipalities
Percentage ‘yes’ votes
Percentage of households that are renters
General municipal expenditure
Percent of owneroccupied housing
Lower expenditure in counties with a relatively large fraction of owners Lower expenditure in municipalities with a relatively large fraction of owners Owners less inclined to vote for expenditure increases Renters support higher levels of school expenditure
Lower expenditure in towns with a relatively large fraction of owners Lower expenditure in suburbs with a relatively large fraction of owners Renters more inclined to vote for expenditure increases Lower expenditure in municipalities with a relatively large fraction of owners
Barr and Davis (1966)
Finding consistent with renter illusion?
Table 1. Continued
Measure of renter illusion
Finding consistent with renter illusion?
School and general expenditure
Percentage of owneroccupiers
Lower expenditure in districts with a relatively large fraction of owners
Demand for public spending
Dummy for homeownership
Schwab and Zampelli (1987)
73 US cities and counties
Percentage of homeownership
Moomau and Morton (1992)
Referendum to increase property taxation in New Orleans 302 Flemish municipalities
Probability of a ‘yes’ vote
Percentage of households that are renters
No difference between renters’ and owners’ expenditure demands Lower expenditure in cities and counties with a relatively large fraction of owners Renters more inclined to vote for property tax increases
Percentage of residences that are non-owner occupied
46 local governments in Tasmania, Australia 54 English and Welsh counties
Proportion of homes owned
Percentage of tenants
Heyndels and Smolders (1994)
Dollery and Worthington (1999) Gemmell et al. (2002)
No difference in expenditure levels among municipalities with large and small fractions of renters Lower expenditure in localities with a relatively large fraction of owners Lower expenditure in localities with a relatively large fraction of tenants
600 school districts in Connecticut, New Jersey and Virginia Survey data from a sample of 2404 inhabitants in Puurs, Belgium
Brazer and McCarty (1987)
since renters often consume less housing than home-owners of the same income level, their property tax share will often be less than that paid by their home-owner counterparts. Secondly, property taxes may be shifted onto renters, but this may take time or public rent control may hinder it. In sum, renters may face a lower tax price than home-owners. If renters correctly perceive this, support for expanded budgets is rational. Note here that the ‘rational renter’ hypothesis also rests on a particular type of renter perception of the local property tax system—namely, that renters understand the tax system as well as home-owners. But since the implication of the ‘rational renter’ hypothesis and renter illusion is the same—namely, upward pressure on local expenditure and taxation—the evidence presented in Table 1 cannot discriminate between them. This somewhat unsatisfactory state of affairs is due to the fact that the literature has so far provided only indirect tests of the renter illusion hypothesis. The focus has been on the implications of renters’ perceptions of the property tax system rather than on perceptions as such. Both of the abovementioned approaches to the renter illusion hypothesis as well as the competing rational renter hypothesis proceed from axiomatic assumptions of renters’ perceptions of property tax systems and then investigate the implications of these assumptions. A much more direct test of renter illusion is to investigate renters’ perceptions of the property tax system. Such an analysis would also be able to discriminate between the renter illusion hypothesis and the competing rational renter hypothesis. For the renter illusion argument to be true, renters must systematically underestimate their property tax burden— and do this to a larger degree than homeowners. For the rational renter argument to be true, there should be no particular difference between renters’ and owners’ perceptions of property taxation. In what follows, such a test is conducted on the basis of survey data of Danish citizens, but first a brief introduction to the Danish local property tax system is in order.
The Danish Property Tax System The local property tax system in Denmark is summarised in Table 2. It shows that there are three main types of property taxes. First, local governments must tax land within their jurisdiction. As specified in the Danish local property tax act, local governments may set the rate between 0.6 and 2.4 per cent of land values. Within each local government, all types of land are taxed at the same rate. Certain types of public property are exempt from the land tax, but local governments may decide to levy a special land tax in this case. Secondly, a building tax may be levied on commercial and industrial property. However, this tax is not used by many local governments (except big cities). A building tax may also be levied on public buildings exempt from land tax. Finally, local authorities must levy a tax on all local property (land and buildings). This is a tax with a fixed rate that leaves no local autonomy in the tax decision. The tax-base—land and building values—for all three property taxes is assessed by central government authorities. Property taxes is not tax-deductible. Table 2 further shows that property taxation is not the most important local tax source in Denmark. Property taxes only constitute 10 per cent of local tax revenue. As in the other Scandinavian countries, income tax is far the most important local tax. From a local perspective, the three property taxes are not equally interesting. The land tax (row 1 in Table 2) is the only property tax that both grants local authorities some autonomy and generates a sizeable amount of revenue. The analyses in the following two sections focus on just this tax. Although the land tax is a minor (but not unimportant) tax with minimum and maximum rates specified by the central government, there is much variation in the actual choice of tax rate. Compared with the far more important income tax—whose rate is decided at the full discretion of the local authority—there is more variation in local decisions on land taxation—see the
Table 2. The Danish local property tax system Revenue (2000) Billion DKK
Percentage of all local tax revenue
Land value (market price)
Land value (market price)
Building value (market price)
Building value (market price)
Property value (land and building at market price)
Rate 1. Local land tax General rule Special rule for public property owners
2. Local building tax Commercial and industrial buildings Public buildings 3. Local property tax
0.6–2.4 per cent (uniform rate within each jurisdiction) Central government owner: same rate as general rule Other public owners: maximum of 1 per cent Maximum of 1 per cent (uniform rate within each jurisdiction) Maximum of 0.5 per cent (uniform rate within each jurisdiction) 1 per cent (rate fixed by central government)
Total Source: Statistics Denmark (2001).
coefficient of variation (standard deviation relative to the mean) in Table 3. Measuring Renter Illusion by Survey Data This section provides an analysis of Danish renters’ and home-owners’ perception of the property tax system. Denmark is a promising
setting in which to explore the occurence and implications of renter behaviour. A third of all inhabitants are renters. Intermunicipal variation is high: renters as a percentage of inhabitants varies across the 275 municipalities from 10 to 68 per cent. Public rent control explicitly allows property taxes to be shifted onto renters.
Table 3. A comparison of the Danish local land and income tax, 2000
Local decision on tax rate Share of total local tax revenue (percentage) Minimum actual rate (percentage) Maximum actual rate (percentage) Mean Standard deviation Coefficient of variation Source: Statistics Denmark (2001).
Local land tax
Local income tax
Limited to 0.6–2.4 per cent 4.5
1.3 0.5 0.4
21.0 0.9 0.04
The analysis is based on interview data from a representative sample of Danish citizens. The sample comprises 1018 Danish citizens of more than 17 years of age. All interviews were conducted by telephone by a consultancy firm as part of an omnibus survey in the period from 5 – 7 August 2002. The sample was randomly chosen from registers of telephone numbers.2 A total of 947 persons responded to the questions used in the analysis (N in column 1 of Table 4),
which gives a response rate of 93 per cent. The exact wording of the interview question was: “Property tax is partly paid as land tax to your municipality. What is your local land tax rate, approximately?” The word “approximately” was added to invite respondents to offer an answer. The analysis thus focuses on the respondents’ estimates of the local land tax rate, which is the most important local property tax source in Denmark (see row 1 in Table 2). Clearly, knowledge of the
Table 4. Analysis of renters’ and home-owners’ perception of the land tax system
Renter Sex (male) Age Education Primary school Secondary school Vocational training Short tertiary education Medium tertiary education Long tertiary education Occupation Pensioner Manual worker Skilled worker White-collar worker Self-employed Miscellaneous Political affiliation Far left wing party Social democratic party Small bourgeois party Liberal party Other party Constant R 2 (adjusted) N
Hypothesis 3 Do renters underestimate the land tax rate? (estimation method: logistic)
Hypothesis 1 Are renters more prone to respond ‘don’t know’? (estimation method: logistic)
Hypothesis 2 Are renters’ estimate of the land tax rate less accurate? (estimation method: OLS)
0.5 21.1 20.0
3.1 21.7 20.4
20.1 0.5 20.0
(ref.) 20.4 20.1 0.1 21.0 21.5
(ref.) 23.4 23.7 27.1 22.4 24.1
(ref.) 0.0 22.2 20.8 20.4 21.4
(ref.) 21.0 20.5 20.3 21.4 20.5
(ref.) 25.5 26.9 24.4 27.3 27.8
(ref.) 20.6 0.2 0.1 20.1 7.4
(ref.) 0.2 20.2 20.0 0.7 4.3 — 947
(ref.) 22.1 0.8 21.8 22.8 32.2 0.26 111
(ref.) 20.3 20.7 21.9 21.5 2.6 — 111
Notes: p , 0.01; p , 0.05; p ,0.10. Entries in the table are unstandardised coefficients. Dependent variables: Hypothesis 1: Dichotomous (0 ¼ respondents give a positive answer; 1 ¼ respondents answer ‘don’t know’). Hypothesis 2: Numerical difference between respondents’ estimate of their land tax rate and the true rate. Hypothesis 3: Dichotomous (0 ¼ respondents overestimate or make correct estimates of their local land tax rate; 1 ¼ respondents underestimate their local land tax rate). Source: Interview data from a representative sample of Danish citizens (see text).
land tax rate does not measure the full extent of renter illusion or renter rationality, but it is nevertheless a central aspect of these concepts. The data show that 88 per cent of all respondents state that they do not know their local land tax rate. When interpreting this high degree of fiscal ignorance, it should be kept in mind that the Danish local land tax is a minor tax (see Table 2). For the present purpose, it is more interesting to focus on differences between renters and homeowners. Three types of differences may be hypothesised. They should all hold if the renter illusion hypothesis is true. – Hypothesis 1: When asked to state their local land tax rate, renters respond ‘don’t know’ to a larger degree than owners (renter ignorance). – Hypothesis 2: When answering the question, renters’ estimates of the local land tax rate are less correct than owners’ estimates (renter ignorance). – Hypothesis 3: Compared with home-owners, renters underestimate their local land tax rate (renter illusion). The first two hypotheses focus on renter ignorance—i.e. unbiased lack of information. As noted above, ignorance is a necessary, but not sufficent, condition for the existence of illusion. The first two hypotheses thus must hold if renter illusion is a fact, but do not directly demonstrate the existence of illusion. This is the object of the third hypothesis which directly focuses on illusion—i.e. systematic underestimation of the land tax rate. The three hypotheses are analysed in Table 4. The table confirms the first two hypotheses, but fails to confirm hypothesis 3. The first column analyses hypothesis 1. It shows which individual characteristics lead to a higher probability of answering ‘don’t know’ to the question of the size of the local land tax rate (positive values indicate increasing probabilities). Being a renter increases this probability. In other words, renters seem to have less knowledge of their local land tax rate. This result indicates that renters suffer from fiscal ignorance (but not necessarily
fiscal illusion). The control variables show that sex, age, education and occupation all influence the probability of answering ‘don’t know’.3 The second and third columns analyse hypotheses 2 and 3 respectively. They focus on the respondents who did not answer ‘don’t know’, but volunteered a guess of their local land tax rate (note the drop in N). Column 2 analyses whether renters generally misperceive their local land tax rate. The dependent variable in this model is the numerical distance between the respondents’ estimate of their local land tax rate and its true size—i.e. in this model there is no distinction between overestimation and underestimation. Positive values indicate increasing numerical differences. The table shows that renters are more inclined to make a wrong estimate than home-owners. This result also indicates that renters suffer from fiscal ignorance (but not necessarily fiscal illusion). The control variables show that estimates of the local land tax rate increase in precision with the respondents’ age and that occupation is also an important predictor of the quality of the respondents’ estimates.4 Column 3 investigates which individual characteristics lead to a higher probability of underestimating the local land tax rate (positive values indicate increasing probabilities). Being a renter does not increase this probability. The renter variable’s coefficient is statistically insignificant (significance: 0.92). In other words, there is no evidence of downward bias in renters’ misperceptions of their local land tax rate. This result indicates that, while renters may suffer from fiscal ignorance, they do not suffer from fiscal illusion.5 The control variables show that education and political affiliation influence the probability of underestimating the land tax rate.6 Table 4 is problematic evidence for both the renter illusion and the rational renter hypothesis. According to the renter illusion hypothesis, renters favour expanded budgets because they underestimate their property tax burden. Column 3 shows that there is no systematic underestimation of the local land tax rate among renters. According to the
rational renter hypothesis there should be no difference between renters’ and home-owners’ perceptions of the property tax system. Columns 1 and 2 show that there is. In sum, the results in Table 4 raise doubts about both hypotheses. If renters suffer from fiscal ignorance, but not from illusion, it seems doubtful that the fraction of renters should have an upward effect on property taxation. To investigate whether this is in fact the case—and thus to explore the validity of the findings from the survey data—the next section will analyse the relationship between renters and local property taxation in Denmark.
Do Renters Have an Upward Effect on Property Taxation? This section presents an analysis following what Oates (1988, p. 68) terms the general approach to empirical studies of fiscal illusion (corresponding to what I called the dominant approach among the studies summarised in Table 1). Studies using this approach start with a model of the form E ¼ aX þ bF þ u where, E is a measure of budgetary size; X is a vector of explanatory variables in a world without fiscal illusion; F is a vector of variables designed to measure fiscal illusion; and u is the disturbance term. The dependent variable in the following analysis is property taxation. Alternatively, the analysis could have focused on spending. But since renter illusion—like other types of fiscal illusion—should manifest itself on both the revenue and the expenditure side of the budget, both measures can be used (see Oates, 1988, p. 68). The most relevant operationalisation of property taxation in a Danish context is land taxation (see Table 2). Land taxation is measured in three different ways to capture as much of the phenomenon as possible and to check the robustness of the explanatory variables: land tax rates, land tax revenue and local tax structure. While the first two measures are straightforward,
the last needs explanation. If there is a positive relationship between the number of renters and property taxation, municipalities with a large number of renters should rely more on this tax instrument than other municipalities. Danish municipalities are financed by income taxes, property taxes and corporate income taxes, but are only allowed some autonomy in the area of income and land taxation (see Table 3). In the corporate income tax area, municipalities receive a fixed share of a centrally determined rate and base. Therefore, effects on the local tax structure must show themselves as variations in the land tax relative to the income tax. The explanatory variables in a world without renter illusion are the following six, which have all been demonstrated in the literature to have an impact on property taxation: the size of the tax-base, income per capita, expenditure needs, the number of farmers, party politics and tax export potential. In what follows, a more detailed argument for the choice of these explanatory factors is given. First, it is plausible that the size of the taxbase affects taxation, but the direction of the relationship is theoretically indeterminate. A large base requires a lower tax rate to generate a given revenue. But a large base may also make it attractive to maintain similar, or slightly higher, tax levels than comparable municipalities with smaller tax-bases. This strategy would have a relatively large income effect and may be cheap politically since voters may be expected to focus on rates. Prud’homme and Navarre (1992) find that the size of the base has a statistically significant effect on property tax rates, but in their (French) case the effect is positive on the tax rate on non-developed land and negative on the tax rate on developed land. Sørensen (2002) finds that, in Norway, the tax-base is negatively related to property taxation. The size of the tax-base is measured as land values in DKK per inhabitant in the following analysis. Secondly, several studies indicate that citizens’ income is important for property taxation. Income levels may affect the choice of
tax-base but, like land values, with a theoretically indeterminate direction. Prud’homme and Navarre (1992) and Sørensen (2002) find a negative relation, whereas Allers et al. (2001) and Chicoine and Walzer (1986) find a positive relation. Sjoquist (1981) and Nordstrand (1978) report inconclusive results. Local income is measured as per capita income in the following analysis. Thirdly, it is necessary to include expenditure needs to control for demographic, social and structural differences among the municipalities. Allers et al. (2001) and Borge and Rattsø (2002) include measures of expenditure needs for this reason. Further, since the number of renters is likely to be positively correlated with various kinds of social problems, it is imperative to disentangle the number of renters from the size of expenditure needs. In Denmark, local expenditure needs are equalised in order to make it possible to provide services based on a given local tax burden. However, equalisation is not complete. In the Greater Copenhagen region (48 municipalities), 85 per cent of intermunicipal differences in expenditure needs are equalised. In the rest of the country (225 municipalities), the equalisation level is 45 per cent. So, important differences remain even after equalisation. Expenditure needs are measured according to the definition used in the Danish intermunicipal equalisation system, but net of grants from the equalisation system. Fourthly, the number of farmers is included to capture the political influence of agriculture. Farmers traditionally prefer income taxes to property taxes since their incomes are relatively small compared with the size of their property. Since Danish municipalities can choose between these two tax sources, the influence of farmers must be controlled for. Nordstrand (1978) found that, at least in the 1970s, agriculture had a significant impact upon local property taxation in Denmark. The strength of agriculture in Danish local tax decisions received national attention in 1986 when the present minimum land tax rate of 0.6 per cent was introduced by the socialist majority in the Danish Parliament in a deliberate move to force rural municipalities
to introduce (or increase) land taxation and thus start taxing farmers. Farmers are measured by the number of farmhouses per inhabitant in the following analysis. Fifthly, the studies by Allers et al. (2001), Sørensen (2002), Borge and Rattsø (2002) and Nordstrand (1978) indicate that party politics is important. The ideological composition of the local council has an impact on property taxation independent of the composition of the local electorate: socialist local governments use property taxation more than bourgeois ones (but see Prud’homme and Navarre (1992) for an inconclusive finding). Party politics is measured as the political affiliation of the local mayor in the following analysis. Finally, the analysis controls for the municipalities’ land tax export potential. Several studies demonstrate that, if given the opportunity, local authorities export property taxes (Chicoine and Walzer, 1986; Thirsk, 1982; Sjoquist, 1981; Bowman, 1974; Hogan and Shelton, 1973). These studies generally argue that the property tax may be exported by taxing commercial and industrial property. In Denmark, property tax export is likely to be more prevalent in relation to another type of property, namely, holiday homes. These are summer cottages typically owned by nonresidents and used for recreation by the owner or rented out to tourists. In some Danish municipalities, summer cottages make up a considerable proportion of the local housing stock, especially in coastal municipalities with attractive beaches. On average, summer cottages constitute 10 per cent of the local housing stock, but this number varies from 0 to 74 per cent across municipalities. It has already been demonstrated by Nordstrand (1978) that ‘summer cottage municipalities’ are more prone to land taxation. The land taxation export potential is measured as the number of summer cottages per inhabitant in the following analysis. The variables and their operationalisation are summarised in Table 5. The results of the analysis are reported in Table 6. With three exceptions, the analysis includes all Danish municipalities.7 All data are from 2000.
Table 5. Variables used in regression analysis Variable Dependent variables Land tax rate Land tax revenue Tax structure Independent variables Renters Land values Per capita income Expenditure need (after equalisation) Farmers Party politics Summer cottages
Local land tax rate in thousandths of the land value Revenue from local land tax in DKK per inhabitant (log.) Local land tax rate relative to local income tax rate
Statistics Denmark (2002)
Number of renters as percentage of local population Land values in DKK per inhabitant Local income tax base in DKK per inhabitant Local expenditure need þ (distance to national average x equalisation level) in DKK per inhabitant Number of farmhouses per inhabitant Mayor (dummy variable: 1 ¼ socialist; 0 ¼ bourgeois) Number of summer cottages per inhabitant (log.)
Statistics Denmark (2002)
Ministry of Interior (2002) Statistics Denmark (2002)
Ministry of Interior (2002) Ministry of Interior (2002) Ministry of Interior (2002) Statistics Denmark (2002) ˚ rbog (2001) Kommunal A Statistics Denmark (2002)
Table 6. Analysis of land taxation, 2000
Renters Land values Per capita income Expenditure need (after equalisation) Farmers Party politics Summer cottages Constant R 2 (adjusted) N
Model 1 Dependant variable: land tax rate (estimation method: Tobit)a
Model 2 Dependant variable: land tax revenue (estimation method: OLS)
Model 3 Dependant variable: tax structure (estimation method: OLS)
20.1 (22.2) 0.00006 (4.6) 20.0001 (23.9) 0.003 (4.0) 255.6 (23.1) 3.2 (4.6) 0.7 (4.4) 245.9 — 272
20.004 (21.2) 0.00001 (13.6) 20.000007 (22.9) 0.0002 (3.2) 23.1 (22.5) 0.3 (5.3) 0.06 (5.3) 2.8 0.67 272
20.004 (22.0) 0.000003 (5.8) 20.000005 (24.4) 0.00009 (3.3) 22.5 (23.6) 0.1 (4.5) 0.03 (4.5) 21.1 0.39 272
Notes: p , 0.01; p , 0.05; p ,0.10. Entries in the table are unstandardised coefficients (T-values in parentheses). a Tobit estimation is used since the dependent variable is truncated. This is true in a very real sense. The present upper and lower limit of local land tax rates were deliberately introduced in 1986 by the Danish parliament to force low (high) tax municipalities to increase (decrease) taxation. The Tobit model allows for both upper and lower truncation (at the maximum and minimum tax rates).
Table 6 shows that the renter illusion hypothesis is not confirmed. The renter variable has a negative sign in all three models. This means that land taxation is reduced as the number of renters increases. This is exactly the opposite of what would be expected under both the renter illusion argument and the rational renter argument. But the table also shows that the effect of the renter variable is not stable across models. In model 2, the variable is not statistically significant, even at the 10 per cent level (although the sign is the same as in models 1 and 3). This casts some doubt on the strength of the relationship. The most fair interpretation of the analysis is that there seems to be a weak, but negative, relationship between the fraction of renters and land taxation. This result stands in contrast to the many US studies summarised in Table 1, but is quite comparable with the two European studies of renter illusion using Oates’ general approach—namely, the Belgian study by Heyndels and Smolders (1994) and the UK study by Gemmel et al. (2002). In both these studies, a negative, but weak, relationship is found between the number of renters and the size of local budgets. The control variables are all strong determinants of property taxation. The tax-base (land values) has a strong positive influence, whereas per capita income is negatively related to property taxation. High expenditure needs (after equalisation) lead to high taxation. Farmers have a dampening effect on property taxation. The party politics variable confirms that socialist local governments are prone to higher taxation than bourgeois ones. Finally, the summer cottage variable confirms that tax export potentials lead to higher taxation.
Conclusion This paper has investigated the renter illusion hypothesis according to which renters underestimate their property tax burden and therefore support excessive levels of local expenditure. In contrast to prior studies, this
paper provides a direct test of renter illusion using survey data to uncover renters’ and home-owners’ perception of the local property tax system. The analysis focuses on renters’ and home-owners’ knowledge of the local property tax rate in Denmark and analyses both fiscal ignorance (unbiased misperception) and fiscal illusion (biased misperception). The analysis demonstrates that renters suffer from ignorance, but not illusion. More tests of renters’ and home-owners’ perceptions of property taxes are needed to reach firm conclusions. Do renters, for instance, correctly perceive that property taxes are shifted onto them? But the analysis nevertheless raises doubts about the microfoundation of the renter illusion hypothesis. The analysis of survey data is followed up by an analysis of renter illusion from the ‘general approach’ of regressing the size of local budgets on a set of explanatory variables and a variable capturing renter illusion. This analysis shows that the number of renters has only a weak (but negative) impact upon property taxation, which is the opposite of what would be expected from the renter illusion hypothesis. This finding corroborates the analysis of the survey data. These results are consistent with a number of European studies of renter illusion, but not with the considerable number of US studies which have been made of this question (which a look at Table 1 will confirm). This raises a natural question: what lies beneath the often-found positive relationship between renter fractions and the size of local budgets? Heyndels and Smolders (1994, p. 335) suggest that this may be related to the fact that US local governments rely much more on property taxation than is generally the case in Europe. Another explanation may be that renters comprise different subsets and shares of the general population in the US and Europe. But these are speculations. To settle the question, we first and foremost need more direct analyses of renters’ and home-owners’ perception of property taxation. Otherwise, we will never be sure whether renter illusion is part of the cause.
Notes 1. 2.
For the contents of Table 1, I am much indebted to the comprehensive survey by Dollery and Worthington (1996). See the consultancy firm’s website for more detailed information about their monthly omnibus survey: http://ve.survey.pls.dk/ survey/default.asp. A test of differences in -2 log L shows that the set of dummy variables measuring education and occupation have a statistically significant impact on the dependent variable, whereas this is not true of the set of dummy variables measuring political affiliation. An incremental F-test shows that the set of dummy variables measuring occupation has a statistically significant impact on the dependent variable, whereas this is not true of the set of dummy variables measuring education and political affiliation. In technical terms, renters’ and home-owners’ estimates of the local land tax rate are evenly spread around (approximately) the same mean, but with a larger variation among the renters’ estimates. A test of differences -2 log L shows that the set of dummy variables measuring education and political affiliation have a statistically significant impact on the dependent variable, whereas this is not true of the set of dummy variables measuring occupation. The two metropolitan municipalities of Copenhagen and Frederiksberg are excluded since, as the only municipalities in the country, they function as both municipal and county governments. The municipality of Farum is excluded because of its special situation. In the 1990s, a charismatic mayor embarked on an ambitious tax reduction strategy. Tax reductions were combined with spectacular increases in service levels which were financed by controversial sale-andlease-back arrangements. In the end, the strategy led to the mayor’s downfall in 2002 when the local council took the, in Denmark, unprecedented step of ousting the mayor.
References ALLERS , M., HAAN , J. DE and STERKS , C. (2001) Partisan influence on the local tax burden in the Netherlands, Public Choice, 106, pp. 351–363. BARR , J. L. and DAVIS , O. A. (1966) An elementary political and economic theory of the expenditures of local governments, Southern Economic Journal, 33, pp. 149–165.
BECK , J. N. (1984) Nonmonotonic demand for municipal services: variation among communities, National Tax Journal, 37, pp. 55–68. BERGSTROM , T. C. and GOODMAN , R. P. (1973) Private demands for public goods, American Economic Review, 63, pp. 280–296. BORGE , L.-E. and RATTSØ , J. (2002) Income distribution and tax structure: empirical test of the Meltzer –Richard hypothesis. Unpublished working paper. BOWMAN , J. H. (1974) Tax exportability, intergovernmental aid, and school finance reform, National Tax Journal, 27, pp. 163–173. BRAZER , H. E. and MC CARTY , T. A. (1987) Interaction between demand for education and for municipal services, National Tax Journal, 40, pp. 555–567. BUCHANAN , J. M. (1967) Public Finance in Democratic Process. Chapel Hill, NC: The University of North Carolina Press. CHICOINE , D. L. and WALZER , N. (1986) Factors affecting property tax reliance: additional evidence, Public Choice, 49, pp. 17–28. DOLLERY , B. E. and WORTHINGTON , A. C. (1996) The empirical analysis of fiscal illusion, Journal of Economic Surveys, 10, pp. 261– 298. DOLLERY , B. E. and WORTHINGTON , A. C. (1999) Fiscal illusion at the local level: an empirical test using Australian municipal data, Economic Record, 75, pp. 37 –48. GEMMELL , N. MORRISSEY , O. and PINAR , A.. (2002) Fiscal illusion and political accountability: theory and evidence from two local tax regimes in Britain, Public Choice, 110, pp. 199–224. GRONBERG , T. J. (1980) The interaction of markets in housing and local public goods: a simultaneous equations approach, Southern Economic Journal, 46, pp. 445–460. HANUSHEK , E. A. (1975) The demand for local public services: an exploratory analysis, in: J. E. JACKSON (Ed.) Public Needs and Private Behavior in Metropolitan Areas, pp. 121–138. Cambridge, MA: Ballinger Publishing Co. HEYNDELS , B. and SMOLDERS , C. (1994) Fiscal illusion at the local level: empirical evidence for the Flemish municipalities, Public Choice, 80, pp. 325–338. HOGAN , T. D. and SHELTON , R. B. (1973) Interstate tax exportation and states’ fiscal structure, National Tax Journal, 26, pp. 553–564. KOMMUNAL AARBOG (2001) Kommunal Aarbog [Municipal Year Book]. Copenhagen: Kommunal Aarbog. LOVELL , M. C. (1978) Spending for education: the exercise of public choice, Review of Economics and Statistics, 60, pp. 487–496.
MARTINEZ -VAZQUEZ , J. (1983) Renters’ illusion or savvy?, Public Finance Quarterly, 11, pp. 237–247. MAY , P. (1982) Local tax burdens, benefit levels and fiscal illusions, National Tax Journal, 35, pp. 465–476. MINISTRY OF THE INTERIOR (2002) Indenrigsministeriets kommunale nøgleal [The Interior Ministry’s key municipal statistics] (www.noegletal.dk). MOOMAU , P. H. and MORTON , R. B. (1992) Revealed Preferences for Property Taxes: An Empirical Study of Perceived Tax Incidence, Review of Economics and Statistics 74, 176–180. NORDSTRAND , R. (1978) Primærkommunernes valg af skattekilder [Municipalities’ choice of tax sources], Nationaløkonomisk Tidsskrift, 116, pp. 43–55. OATES , W. E. (1988) On the nature and measurement of fiscal illusion: a survey, in: G. BRENNAN ET AL. (Eds) Taxation and Fiscal Federalism: Essays in Honour of Russell Mathews, pp. 65–82. Sydney: Australian National University Press [Reprinted in W. E. Oates (1991) Studies in Fiscal Federalism, pp. 431–448. Worcester: Edward Elgar]. PETERSON , G. E. (1975) Voter demand for public school expenditures, in: J. E. JACKSON (Ed.) Public Needs and Private Behavior in Metropolitan Areas, pp. 99 –121. Cambridge, MA: Ballinger Publishing Co. PRUD ’HOMME , R. and NAVARRE , F. (1992) Local property taxation in France, in: D. KING (Ed.) Local Government Economics in Theory and Practice, pp. 127–137. London: Routledge.
SCHOKKAERT , E. (1987) Preferences and demand for local public spending, Journal of Public Economics, 34, pp. 175– 188. SCHWAB , R. M. and ZAMPELLI , E. M. (1987) Disentangling the demand function from the production function for local public services, Journal of Public Economics, 33, pp. 245–260. SJOQUIST , D. L. (1981) A median voter analysis of variations in the use of property taxes among local governments, Public Choice, 36, pp. 273–285. STATISTICS DENMARK (2001) Statistiske efterretninger. Offentlige finanser. Copenhagen. STATISTICS DENMARK (2002) Danmarks statistikbank [Denmark’s Statistical Bank] (http:// www.statistikbanken.dk). SØRENSEN , R. (1992) Fiscal illusions: nothing but illusions?, European Journal of Political Research, 22, pp. 279–305. SØRENSEN , R. (2002) Should the “Little Leviathan” be constrained? Tax policy in Norwegian local government. Working Paper, Norwegian School of Management, Oslo (http:// home.bi.no/faq89001/). THIRSK , W. R. (1982) Political sensivity versus economic sensibility: a tale of two property taxes, in: W. R. Thirsk and J. Whalley (Eds) Tax Policy Options in the 1980s, pp. 384–401. Toronto: Canadian Tax Foundation. WINTER , S. and MOURITZEN , P. E. (2001) Why people want something for nothing: the role of asymmetric illusions, European Journal of Political Research, 39, pp. 109–143.