2. Diagnosis of the major problems with the Arnona and the financing of local governments in Israel
Chapter 2 provides a diagnosis of the most important problems with the way that Israel finances its local governments. The major focus is on the Arnona, the Israeli property tax, which is the major source of revenue raised directly by local governments. The analysis draws on data and information presented in Chapter 1 and on best practices found in other OECD countries. Although, as an area-based system, the Arnona is quite different from property tax systems found in most other countries, it is similar to all property taxes in that it is a recurring tax on immovable property. The chapter begins with a discussion of the advantages and the disadvantages of the property tax as a local government source of revenue. It then evaluates the Arnona relying on a standard set of criteria used by economists. Special attention is paid to land-use incentives created by the tax and to the distribution of tax burdens across households. The chapter then considers the adequacy of local government services, with a focus on public education and social services. It concludes with an assessment of the role of grants in reducing fiscal disparities among local governments.
There is a wide agreement among public finance experts that a recurrent tax on immovable property is a highly appropriate source of tax revenue for local governments. Although the property tax should not serve as the only source of local government revenue, it works well as the core for a revenue system that includes a mix of taxes, non-tax revenues, such as user fees, and fiscal transfers from higher levels of government.
Economists point to a number of reasons why property taxes are particularly well suited as a revenue source for local governments. These include:
1. By its very definition, immovable property cannot flee the tax collector; and the tax base is easily allocated between different subnational units without requiring adjustments for cross-border transactions or flows.
2. Because the supply of land is fixed, a property tax levied on land will not distort taxpayer behaviour, and thus is considered by economists to be highly efficient. At modest rates, a property tax on improvements, i.e. buildings, is not very distortive, and hence is considered quite efficient, especially when compared to other possible local government taxes.
3. Most local government public services provide services directly linked to properties, such as water supply and sewers, or to property owners and occupants.
4. A value-based property tax system captures for local governments some of the increases in the value of land and buildings that are attributable to local government expenditures. This capitalization of public services leads many economists to classify the property tax as a benefits tax, effectively serving as the price residents pay for the local public services they receive.
5. As long as local governments have some control over property tax rates, the property tax promotes local fiscal autonomy.
6. Compared to most other taxes, the property tax is a highly visible tax. It is generally paid in large lump-sum payments several times per year. This contrasts with local consumption taxes, which are included in the purchase price of goods and services, or earnings taxes, which are generally withheld by employers from regular pay checks. Because the property tax is visible, local taxpayers are more likely to hold local government officials accountable for the efficient and effective operation of local government.
7. In most decentralized countries, the property tax base of most local governments is large enough to generate, with moderate rates, sufficient revenue to cover a significant portion of their fiscal needs.
8. Property tax bases are generally more stable over the course of a business cycle than the bases of income and consumption taxes. This is particularly true for the base of an area-based property tax system.
9. Because it is generally much easier to make annual changes in property tax rates than to make changes in income or consumption tax rates, property tax revenues can provide a stable revenue source for local governments during economic downturns.
There are additional advantages to the property tax that are not specific to local governments. These include:
1. Economists do not agree on the incidence of the property tax. The recurrent property tax could be considered a tax on capital, an income tax, or a consumption tax depending on whether one considers it a tax on an asset, a tax on the flow of services that derive from owning or renting a property, or a tax on the imputed rent from owner-occupied housing.
2. Compared to other taxes, property taxes are considered the least harmful to economic growth. This is particularly true for property taxes levied on the value of unimproved land.
3. Property taxes can underpin sustainable land use. A tax on the value of land can help contain urban sprawl by providing developers with an incentive to convert developed land rather than develop vacant land on the urban periphery. The land-use effects of property taxes, which tax both land and buildings, are more ambiguous. Specifically designed “green” property taxes, such as oil-sealing taxes, and development charges, can further help internalize land-use externalities.
4. Information on land, buildings, and market prices collected in the course of administering taxes on immovable property becomes part of a valuable collection of information that has numerous governmental and private uses.
5. If up-to-date and publicly available, this information can facilitate orderly functioning of real property markets.
Despite their advantages—or perhaps because of some of them—recurrent property taxes are often underutilized as a source of local government tax revenue across OECD countries. In 2016, local government recurrent immovable property tax revenue as a percentage of GDP varied between 0.1% (in Luxembourg, Switzerland, and Mexico) and 2.9% (in Canada), with an average of 0.9% (OECD, 2018[1]). Israel’s Arnona revenue, at 2.0% of GDP was over twice as high as the OECD average.
Although property values have risen sharply in most OECD member countries over the past couple of decades, on average property tax revenue as a share of total local government tax revenue has been quite stable at around 40%. In 2016, the OECD average property tax revenue as a share of local tax revenue was 41.6%.
There are several possible reasons why the property tax is not more heavily utilised in OECD countries.
1. The high visibility of the property tax tends to make it quite unpopular among the public (Cabral and Hoxby, 2012[2]; Blöchliger, 2015[3]).
2. This unpopularity has led to the enactment in a number of countries of various restrictions imposed by higher level governments on local government property tax rates, revenues, or the growth of property tax bases. To prevent economic hardship on households with low incomes, on the disabled, and on the elderly, most property tax systems include various exemptions, credits or deductions, or discounts targeted to needy taxpayers.
3. Although the extent to which a well-functioning property tax system is regressive is subject to some controversy among economists, a poorly administered system with infrequent reassessment of property values is likely to place heavier burdens on those with limited ability to pay then on those with higher incomes.
4. As property tax liabilities are unrelated to income or cash flows, the property tax may cause hardships for some taxpayers, especially farmers and the elderly. Many countries address this issue with specific policy instruments, such as exemptions, credits, or reliefs.
5. Relative to income or consumption taxes, property taxes are costly to administer. The establishment of a cadastre and a system for assessing the value of property requires a considerable initial investment. In countries with low reliance on the property tax, administrative costs can be high relative to property tax revenues.
6. In a number of countries, the property tax is administered by the central government, with revenues being returned to local jurisdictions. The staff of the national tax administration offices are more used to collecting so-called “modern” taxes such as the value added tax, personal income taxes, and business taxes, all of which are levied on economic flows. Many tax officers appear to have a “cultural” bias against the property tax, which is levied on a stock. Officials of national tax collection agencies are often evaluated on the basis of efficiency indicators, such as collection ratios. They often see the property tax as a tax that is costly to administer and collect relative to the revenues raised, and thus they focus their main efforts on the more modern central government taxes (United Nations Human Settlements Programme, 2013[4]).
7. The buoyancy of the property tax base with respect to economic growth depends on the procedures used to define and update the property tax base. Buoyancy will be relatively high if the assessment of property is based on market values, and assessments are frequently updated. The property tax will be only moderately buoyant if the property tax base is based on rental values that are infrequently updated. The buoyancy of the base will be low if property is reassessed very infrequently, or if the property tax is unrelated to market values, for example, based on area. Israel addressed the issue of revenue buoyancy by adjusting tax rates annual to reflect increases in the consumer price index.
8. Local governments that rely heavily on the property tax risk having to raise property tax rates in periods when market values decline in order to maintain the delivery of local public services. Implementing these tax increases is likely to be politically extremely difficult.
While property tax is a cornerstone of local taxation in many countries in the world, its implementation and management face many obstacles and reforms are particularly difficult to implement. Because of these debates and difficulties, the importance of property taxation began to decline in the early 1980s (OECD/KIPF, 2016[5]). But it regained prominence in the wake of the 2008 global crisis, which provided an opportunity to optimise revenue from this tax and to redesign property taxation so that it is economically and fiscally successful. Since, property tax reforms, including cadastral and land registry’s reforms, are on the rise in many countries in the OECD (Ireland, Greece, Italy, Slovenia, Denmark, etc.) and the world (Box 2.1).
Numerous OECD countries enacted property tax reforms over the last ten years. Reforms are diverse, including creating a new property tax, imposing a property tax on a new category of owners (legal persons or individual) or assets (land or building), providing local governments with more leeway on tax rates or bases, changing the rates or brackets, reforming land or property registries, modifying valuation methods towards rental or market values or rental value and revaluation processes, etc.
In Ireland, a local property tax on residential properties (LPT) was introduced in 2013, in addition to the existing property tax levied commercial properties (commercial rates) 80% of LPT revenue are retained locally for the funding of basic public services, while the remaining 20% are allocated to the equalisation fund. In 2015, new powers were conferred to local authorities to vary the LPT annual rate on valuation by up to 15%, with the aim of increasing their financial autonomy and funding sustainability. As of 2016, 11 local authorities had taken the opportunity to vary the LPT rate, and LPT represented overall 7% of SNG local revenue. Each individual local authority is responsible for the levying and collection of these taxes. The Annual Rate on Valuation (ARV), applied to the valuation of each property as determined by the Valuation Office, is set by the elected members of each local council as part of their annual budget.
In France, one of the most notable reforms is France’s progressive repeal of the residence tax (taxe d’habitation), one of the taxes that make up recurrent property taxation in France with the land tax, the property tax on building, and the business real estate tax (part of the part of the territorial economic contribution). The residence tax is due annually by the household who occupies a dwelling (whether as owner or tenant). The residence tax has largely been perceived as both horizontally and vertically unfair. It is horizontally unfair because the level of the tax varies widely across municipalities and because the notional rental values upon which the tax is based have not been updated since the 1970s. It is vertically unfair because the dwelling tax is regressive with respect to income. In a first attempt to lessen this regressivity, the household residence tax charge has been subject to a ceiling with respect to taxable household income. In a second step, the abolition of the residence tax has been decided which is phased in over three years so that eligible households will see a 30% reduction in their taxe d'habitation bill in 2018, a 65% reduction in 2019 and a 100% reduction in 2020. In May 2018, the government published a report on reforming local taxes, which stated that the taxe d'habitation would be completely abolished by 2021 (instead of 80% of households as initially announced). Overall, before the reform, property taxes stood for 59% of subnational government revenue tax revenue and 30% of their total revenue in 2016, representing a major source for municipalities and departments. They amounted to 3.4% of GDP one of the highest levels in the OECD. This high level will, however, decrease with the abolition of the residence tax,
In Portugal, a new municipal property tax surcharge of 0.7% and 1% on real estate of a total taxable value respectively above EUR 600 000 and EUR 1 000 000 was introduced in 2016. This surcharge, which replaces the stamp duty on residential urban properties or land for construction of a taxable value above EUR 1 million, aims at enhancing fairness. The reform also introduces a single rate of 0.4% levied on companies’ total real estate. Cadastral value of 4.9 million properties were also re-evaluated to underpin the new property tax regime. Revenue from property tax accounted for 0.8% of GDP, 32.4% of SNG tax revenue and 13.2% of SNG total revenue in 2016.
In Italy, the property tax on primary residences (ICI) was abolished in 2006 but reintroduced in 2013 under a new single municipal tax (IMU), which incorporates three taxes: 1/ the IMU (imposta municipale propria), which is a real estate tax paid by owners of secondary residences only. The taxable base is determined in connection with the value of the property according to the cadastre. The regular tax rate is 0.76% of the taxable base, but municipalities may increase or reduce the rate, with a maximum of 0.3%. 2/ the TASI or “tax for indivisible services” which is a supplementary real estate tax, which is supposed to meet the expenses for the delivery of lighting, street cleaning, green areas and services that are provided equitably by municipalities to all citizens; 3/ and the TARI (waste tax) which must cover the cost of the service of collection and treatment of waste. Both the IMU and the TASI were repealed on primary residences (except luxury homes) in 2014 and 2015. A reform of cadastral values is still pending to fully exploit the potential of the property tax. In 2016, the recurrent property tax accounted for 1.1% of GDP, in line with the OECD average.
In Greece, a new property tax (Unified Property Tax, ENFIA) was established in 2014 to replace two previous property taxes, the real estate based wealth tax (FAP) and the property tax that is collected through the Electric company utility bill. The new tax applies to individuals and legal entities owners of land and buildings. The ENFIA comprises a “main tax” and a “supplementary tax”. The main tax applies for buildings, plots, fields and so on. The main tax is calculated on the basis of “objective values” based on several criteria such as the location of the property, its size, use, age, the floor on which the property is located, etc. The supplementary tax (so called sumptuary tax) is imposed on very expensive property. The tax rate ranges from 0.1 to 1% of the assessed value. In 2018, the government reassessed the “objective values”, resulting in a decrease of paid taxes for 23% of property owners (23%), a stability for 62% and an increase for the remaining 15% of tax payers.
In Latvia, the property tax is a “national tax” but fully redistributed to municipalities. It is levied on all land and buildings, either used for residential (since 2010) or commercial purposes. Since 2013, municipalities have leverage over the property tax of which they can now set the rates instead of the central government within limits set by the State and provide rebates. In 2016, it accounted for 15% of local tax revenue, 9% of local revenue and 0.9% of GDP.
In Germany, the property tax is currently being reformed, as its valuations date back to 1964 (Western Germany) and even 1935 (Eastern Germany). The Constitutional Court has recently ruled that valuations must be updated by 2024 in an effort to make the tax system more efficient. The Constitutional Court has given the federal government until the end of 2019 to come up with a new method of calculating the tax.
In the United Kingdom, until 2013, the structure of local tax funding in England, Scotland, Wales and Northern Ireland was relatively similar until recently. Over the last years, it has become more divergent ever since then. traditionally, in England, Scotland and Wales, local councils are funded essentially by the council tax (a property tax based on the rental value of individual property paid by the resident) and a share of the business rates (property tax levied on non-residential properties, whose receipts are pooled and then redistributed by the UK government in England or the devolved nations on a per capita basis).
In England, the Local Government Finance Act 2012 introduced major changes in the English system. Local government taxing power increased in 2013 through the localisation of the local council tax support scheme (LCTS) and of the business rates retention scheme (BRRS). In Scotland, the council tax support was localised to Scotland. The Scottish Government has made changes to the Council Tax multiplier. In Northern Ireland, councils are now legally required to set domestic and non-domestic district rates.
In Slovenia, the attempts to introduce a market value-based real estate tax were held to be unconstitutional in 2014. A new reform was however introduced through the Property Tax Act in 2017 to improve its efficiency and to provide for greater fiscal autonomy. The reform broadened the tax base through a new system of valuation of properties, through the Act on mass property valuation, entered into force in January 2019, to be effective in 2020. The tax is levied at different progressive rates depending on the type and value of the premises. They range from 0.10% to 1%. Rates for properties used for business or recreational purposes are higher, at up to 1.5% of value.
Sources: OECD/UCLG (2019[6]), 2019 Report of the World Observatory on Subnational Government Finance and Investment, http://www.sng-wofi.org; OECD (2018[7]), Tax Policy Reforms 2018: OECD and Selected Partner Economies, https://doi.org/10.1787/26173433.
Box 2.2 provides a set of principles developed by the OECD to guide efforts to enact property tax reforms.
1. The taxation of immovable property should be designed so as to be coherent with the overall tax system. In particular, the immovable property tax should be aligned with other forms of capital taxation and with user fees. Taxing immovable property through the income tax system (imputed or actual rent) might reduce the need for a specific immovable property tax, although moving from property taxation to income taxation tends to change overall progressivity of the tax system and the sub-central share of overall tax revenues.
2. Immovable property taxation should cover all types and forms of land use. These include improved and non-improved land for residential and businesses purposes, agricultural land, forests and land dedicated to special infrastructure purposes. The number of untaxed items should be kept to a minimum. The property tax may be paid by property owners, property occupants or both.
3. The property tax should be based on the market value of immovable property. For that reason the fair assessment of property values is crucial for a property tax system to be widely accepted. Property values should be updated frequently, in order to avoid unequal treatment of properties, and taxable values should be determined by reference to assessed values. The use of house price, construction price or consumption price indices can be used as an auxiliary measure to up-rate properties in between re-evaluations.
4. Sub-central governments can be given some autonomy to set property tax rates, and be assigned the revenues from property taxation. The property tax is closely linked to the benefits taxpayers receive from the consumption of sub-central public services, and moreover it is the least prone to sub-central tax competition and tax base erosion. However, upper-level governments should in any case keep a say on the tax base, such as by determining rules for property valuation or tax allowances.
5. Differences in property tax rates between households and businesses should reflect differences in government services provided. While households tend to be the main beneficiaries of sub-central public services, firms also benefit from them. Business property taxation may also act as a backstop against incorporation of households and tax avoidance. On the other hand, the business sector currently faces higher property tax rates than households in most countries.
6. Ideally, land values should be taxed at a higher rate than the values of improvements built on that land (two-tier property tax). On economic grounds, a pure land tax can be justified by the fact that it captures an economic rent and hardly distorts the work, save and investment choices of households and firms. As such, taxing improvements might discourage investment and infrastructure up-grading, especially for businesses. But as it is hard to separate the land value from that of improvements, a pure land tax might be difficult to implement.
7. Low-income households can be entitled to tax allowances and other forms of tax relief and tax liability reductions (means-testing). Also, a lump-sum might be deductible from assessed property values to allow for the uncertainty around the true value of a property. Both measures would make the property tax progressive without the need to introduce progressive tax rates, and are likely to broaden political support for property tax reform. However, the overall consistency of such allowances with other parts of the tax and transfer system must be carefully checked.
8. Property tax reform may provide an opportunity for wider public finance reforms. Higher property tax revenues help upper-level governments in reducing intergovernmental transfers, thereby making local public finance more accountable to taxpayers. Moreover, property tax reform could go together with reforms aimed at improving the quality of sub-central public services, which again may broaden support for property tax reform.
9. Property tax reform should be phased in to avoid large jumps in tax liabilities and to ensure long-term political success. During transition, tax liabilities should gradually converge to those resulting from fair market valuation. If recurrent property taxation is to replace property transaction taxes, former transaction tax payments could be credited against recurrent property tax payments. Liquidity-constrained households can be allowed to postpone property tax payments until a property is sold or bequeathed (deferral, escrow).
10. Property taxation should rely on a good administrative and judicial framework, and administrative reform might have to go hand in hand with property tax reform. Property taxation should rely on a well-functioning property identification system (cadastre). While the set-up costs of a cadastre may be high, the costs tend to decline once the system is established. Identification, valuation, assessment and enforcement of liabilities should be considered effective and fair by taxpayers. Also, appeal processes should be simple and fast.
Sources: OECD (2016[8]), “Property tax principles”, CTPA/CFA/WP2(2016)3, Unpublished, OECD Working Party No. 2 on Tax Policy Analysis and Tax Statistics, OECD, Paris; OECD/KIPF (2016[5]), “Fiscal federalism 2016: Making decentralisation work”, in Reforming the Tax on Immovable Property.
Economists use a standard set of criteria to evaluate tax systems. A list of these criteria includes the following:
Revenue sufficiency: A tax should generate enough revenue to justify its inclusion in a mix of taxes and yield sufficient revenue relative to the costs involved in its administration. Taxes that generate little revenue are often referred to as nuisance taxes.
Economic efficiency: To what extent does a tax effect (distort) the behaviour or actions of individuals or businesses. An efficient tax leads to little or no change in behaviour.
Equity: Considerations of equity involve comparing the amount of taxes individuals (or households) pay relative to some standard such as their income, the size or value of their housing unit, their age or other personal characteristic. Vertical equity compares the tax paid, generally relative to income, by individuals at different levels of income. A tax is considered progressive if the tax paid relative to income rises with income, and regressive if the tax paid relative to income is higher for lower incomes. Horizontal equity exists if individuals in similar situations, such as living in houses of equal size or equal value, pay the same amount of taxes.
Administrative and compliance costs: Taxes differ in the amount of money required to administer the tax. Costs include those involved in defining the tax base, determining tax liabilities for individual taxpayers, and collecting the tax. Taxes, especially individual income taxes and many business taxes, place substantial compliance costs on individual taxpayers.
Revenue buoyancy and stability: The revenue generated by most taxes varies over time. Revenues from some taxes, especially taxes on income and business profits, vary in proportion to the strength of the economy. During periods of prosperity, revenues grow, but during economic downturns, revenues can fall dramatically. In contrast, for some other taxes, tax revenues remain largely unchanged over time, whereas the costs of providing public services generally rise. A well-functioning tax system will find a mix between excessive buoyancy and stability.
Transparency and simplicity: Both the structure and administration of taxes should be sufficiently simple and transparent so that taxpayers have faith that the tax is being fairly administered. Collection rates tend to be lower, when taxpayers have no idea how a tax is being administered and whether they are being treated fairly.
Tax autonomy: For local government officials to be accountable to local residence, it is important that local officials can respond to local public service preferences and needs by being able to adjust the tax rates on local taxes.
The design of a tax system involves making compromises. No tax system performs well in terms of all the criteria. To conduct a full analysis of the Arnona system based on the criteria listed above would require a great deal of detailed data on the operation of the Arnona within each of Israel’s 255 local jurisdictions. The evaluation of the Arnona that follows is based on the limited information available and on the experience of property tax systems in other countries. It will highlight both the strengths and the weaknesses of the Arnona system.
Revenue sufficiency
The Arnona is the most important source of local government revenue, generating almost two-thirds of the own revenues of regular budgets of local authorities (see Figure 1.8). In 2016, 81.2% of the total tax revenue raised by local governments came from the Arnona, a share that is nearly double the OECD average ( (OECD, 2018[1]). Combining the important role played by revenue from taxes in local government finance with the country’s heavy reliance on property taxes as a source of local tax revenue, it is not surprising that the share of total local government revenue from property taxation is more than double the OECD average. Revenue from the Arnona is certainly sufficient. In fact, given the heavily reliance on the Arnona, the government might consider authorizing local governments to use additional taxes and fees.
Economic efficiency
A tax is considered to be inefficient to the extent that it effects the actions and behaviours of businesses, households, or local governments. The question is not whether a tax influences behaviour, but rather the size or importance of the effects. The government is concerned that the Arnona is providing a strong incentive for local governments to discourage new housing constructions and devote an excessive amount of municipal land to commercial and industrial development.
As the Arnona is an area-based tax system, it may encourage households to live in smaller housing units and businesses to operate with smaller offices, retail space, or factories. To our knowledge, there has been no evidence that at current rates, the Arnona has reduced the consumption of physical space for either residential or non-residential uses.
Arnona rates for all types of property vary across local jurisdictions. These rate differences may encourage both households and businesses to move to a jurisdiction that provides a more desirable mix of public services and Arnona rates. There is a large literature that shows that these local government fiscal conditions are at least partially capitalized into higher or lower property values. Although the issues are complex, these fiscally-motivated moves may be efficiency enhancing, especially if they encourage local governments to operate more efficiently and respond more effectively to the preferences of their residents.
Arnona rates within any given local jurisdiction differ substantially by type of property. The differences among non-residential properties are particularly large. For example, in 2016, the average rate on parking lots was NIS 25 (per square meter), on hotels, NIS 61, on industry, NIS 71, on office and retail trade, NIS 109, and on banks and insurance companies, NIS 850. The reason for these rate differences in unclear. It is difficult to imagine an economic argument that could be used to justify or explain these differences. Because land is more or less in fixed supply, a tax on land values is highly efficient as it will not distort behaviour. This has led economists to argue that higher tax rates on land are desirable because they will encourage development (Oates and Schwab, 2009[9]). This argument suggests that parking lots, which are essentially vacant land usually located in central urban locations, should be taxed at high rates. The average Arnona rates on parking lots are, however, quite low.
The observed Arnona rate differences across types of non-residential property are sufficiently large to create locational inefficiencies. Although the OECD was not provided with parcel-level data on land use, it is reasonable to hypothesize that, for example, high Arnona rates on banks and insurance companies discouraged banks and insurance companies from locating in certain jurisdictions, especially those with especially high Arnona rates. The result is both inefficient and it creates hardships on local residents in need of banking or insurance services.
Equity
Most discussions of the equity of the property tax focus on the tax paid by households relative to their ability to pay and relative to the tax paid by households residing in housing units of similar value. Israel has developed an area-based system as a proxy for market value. Horne and Felsenste in (Horne and Felsenstein, 2010[10])describe the area-based system as an “alternative assessment” method designed to provide a substitute for property valuation. Assessing how well the Arnona system mimics a value-based property tax system is extremely difficult. Because of limited parcel-level and household data, there is no comprehensive evidence on the link between Arnona payments, housing values, and household incomes.
Two academic studies, however, shed some light on the issue. Portnov et al. (2000[11]) do not directly address the question of tax equity, but present evidence that differences in per capita Arnona collections “reflect factors that would normally be expected to influence property market values.” These factors include the population size of local authorities, their distance from the center of the country, average income, and their ethnic and socio-religious composition. In a more recent study, Horne and Felsenstein (Horne and Felsenstein, 2010[10])draw on detailed data from the Israeli Household Expenditure Surveys from 1997 through 2005 to analyse the relationship between property values, household income (measured by total household expenditures), and the actual Arnona tax paid. Whereas they find that property tax payments are strongly correlated with property values, they also find that a percentage point increase in both expenditures and in housing values was associated with a less than percentage point increase of Arnona payments. This result leads them to conclude that the residential Arnona is a regressive tax.
It is important to note that several aspects of the Arnona system are designed explicitly to reduce regressivity. The government has established several categories of “needy” individuals and households that are entitled to exemptions and discounts to their Arnona charges. In 2017, local governments issued NIS 3.4 billion in exemptions, releases, and discounts. Although no detailed data are available, these reductions in Arnona tax liabilities went primarily to households with relatively low incomes. First, in some jurisdictions, spatial zones are defined to include areas with concentrations of luxury housing. Residents of these zones are subject to higher Arnona rates. While these local authority efforts to apply higher Arnona rates to certain residential sub-classifications (subject to ministerial approval) undoubtedly increase tax progressivity, Horne and Felsenstein (Horne and Felsenstein, 2010[10]) argue that within jurisdiction residential rate differences are not large enough to convert the Arnona into a progressive tax.
One reason that it is difficult to increase progressivity using an area-based system is that households living in identically-sized apartments in the same neighbourhood or in the same building face the same tax liability independent of their income or the value of their housing unit. Some residents will do major renovations, such as new kitchens and bathrooms, while others will fail to do regular maintenance. The market value of apartments on the top floors of residential apartment buildings are generally substantially higher than the value of identically-sized apartments on lower floors. None of these differences in property value will be reflected in Arnona liabilities.
Horizontal inequities in a value-based property tax system exist when within the same jurisdiction, two households in similar valued housing units face different tax bills. In an area-based system, intra-jurisdictional horizontal inequities exist if taxpayers with similarly sized housing units or businesses face substantially different tax liabilities. Some jurisdictions are divided into zones, with different tax rates applied to residents of each zone. Urban areas in particular tend to be quite heterogeneous in terms of income, property values, and socio-economic characteristics. As a result, within one jurisdiction, identical households (along any number of dimensions) who live in similar sized housing units will face different Arnona rates if they live in different zones.
One contributor to inter-jurisdictional horizontal inequities is that different jurisdictions follow different approaches to the measurement of taxable area. In some municipalities, taxable area includes a portion of the common space in a building, such as staircases, the lobby, or a common storage room. In other local authorities, the taxable area excludes common areas but includes the footprint of the external walls of the building. A third approach is based on the usable space of an apartment, excluding both wall footprints and common space. The use of the different methods of area measurement results in 14% to 29% differences in the taxable area of the same apartment (Daran, 1999[12]).
Administrative costs
The costs of administrating an area-based property tax system are lower than the costs of administrating a value-based system. While all property tax systems require the creation and maintenance of a cadastre, the sending out of tax bills, and efforts involved in collecting taxes from delinquent taxpayers, a value-based system, by definition, must undertake the process of both determining and updating property values.
Local governments in Israel are responsible for establishing and maintaining records of the households and businesses that occupy each parcel of land within jurisdiction boundaries. Local jurisdictions must determine both the criteria for measuring area, and undertake the measurement of area according to the established criteria. The calculation of the tax levied on each parcel also requires information on the location of the property, the age of the property, and how the property is being used. While this process appears to be simple, the fact that most local governments utilize many sub-categories of property, especially for non-residential property, makes the Arnona system more complex, and undoubtedly adds to the cost of administration.
Unfortunately, no data from individual local governments on the costs of administrating the Arnona were made available to the OECD mission. However, Dan Daran (1999[12]), a former deputy mayor of Tel Aviv, reported that in 1997 the cost of implementing the Arnona system in Tel Aviv was approximately NIS 24 million. This amount was equivalent to 1.6% of the Arnona revenue in that year. Officials in the Ministry of Finance suggested that on average the costs of administering the Arnona were between 0.4% and 0.5% of total Arnona charges.
Without detailed data on the administrative costs associated with the Arnona, it is difficult to assess whether some jurisdictions are substantially more efficient in the administration and collection of the Arnona compared to others. It would also be useful to know whether there exist substantial scale economies in the administration and collection of the Arnona, whether regional collaboration in Arnona administration may be cost effective, and whether the use of private collection companies reduces collection costs. Knowledge of the relationship between collection rates—measured as Arnona revenue as a percentage of net Arnona charges—and spending on tax administration would also help in the design of policies to improve collection rates. Finally, in considering possible reforms to the Arnona, it would be important to assess by how much administrative and collection costs would rise if Israel moved from an area-based property tax system to one based on market values or rental values.
The costs involved in moving from an area-based to a value-based system are clearly lower when a country already has, as in Israel, a well-function property market. Ron Rakow (2009[13]), one of the leading experts on property tax administration in the United States, argues that “the difference in administrative costs between an area-based system versus a value-based system is not that great. If Israel has a good system for recording property transactions, including sales prices, and a digitized land records system linked to GIS, then it is well on its way to having the ingredients for a value-based system.” Property assessors argue that the two most important factors in determining the value of property are location and size. As long as both of these are known, it is not terribly difficult or expensive to develop a reasonably accurate assessment valuation model.
Box 2.3 provides information on how recent advances in technology, including the use of artificial intelligence (AI) are dramatically reducing the costs of implementing value-based assessment systems. Box 2.4 highlights some information on the organisation and implementation of property assessment in France and the United States.
The most common tool used by many jurisdictions to determine the value of residential property is the implementation of automated valuation models (AVMs). These models use techniques, such as geographically weighted regressions to produce estimates of market values. Quite accurate estimates can usually be made using a relatively limited amount of data. Artificial intelligence (AI) is also being used with AVMs, and has been producing excellent results at a relatively low cost.
The technology for collecting data on structures has also advanced. Light Detection and Ranging (LIDAR), a remote sensing methodology that uses light in the form of a pulsed laser, when combined with AI can provide a low cost method for collecting data attributes for the exterior of buildings. Drones are also beginning to be used for data collection for assessment purposes. These techniques can identify physical changes, such as property additions, swimming pools, parking lots or even roof changes. Assessors can use these techniques to generate a priority list of properties that should be targeted for special attention using more traditional assessing and appraisal techniques.
In unitary countries, the assessment (and collection) of local government taxes is often the responsibility of the central tax administration, sometimes with local officials being responsible for providing property assessments. The assessment and the collection of the local property taxes entail administrative costs, which are sometimes charged to the local governments. The central government retains a portion of the revenue raised to cover administrative costs. Often these costs are not itemised, by the central tax administration, leading local governments to suspect that these costs are overestimated. In France, the issue of over-charging for the administrative costs of tax collection was the subject of a recent report by the French Cour des Comptes (2017).
In the United States, individual state governments determine how property will be assessed and how property tax collected. In most states, the assessment and collection of the property tax is a local government responsibility, often with some state government oversight. Because of different institutional arrangements and the large variation in the size of local governments, one observes large differences in the cost of assessment.
In 2013 the International Association of Assessment Officials (IAAO) conducted a membership survey on the costs of property tax assessment in the U.S. The survey asked whether assessment was carried out by a state agency, a public agency serving either a single county/parish, or a single municipality/city, or a single town. The survey also asked whether assessment offices served multiple local authorities. The survey results, reported in Walters and IAAO Research Committee (2014[14]), showed that there are substantial economies of scale in the assessment of property. The number of employees per parcel decreases by 25% from small assessment agencies to the largest ones. Not surprisingly, the larger assessment organisations have better access to technology than the smaller ones. As a result, the total cost per employee increases with the size of the assessment agency, but the average cost per parcel is 25% lower in larger units than in smaller ones. The cost per parcel does vary with the type of assessment agency; it is two times higher in state or city agencies than in private agencies serving multiple jurisdictions. However, the private agencies have better access to technology, and their employees benefit from higher salaries than employees in public assessment offices.
The survey also reported that the average assessing cost per parcel of property was approximately $26. Based on U.S. Census Bureau data from the American Community Survey, the median property taxes paid on owner-occupied homes was $2,150 (a five-year average statistic for the years 2011-2015). On the basis of these numbers, the costs of assessing residential property are about 1.2% of the revenue raised. This percentage would probably be lower in locations where assessments are carried out less frequently.
Sources: Walters, L.C. and IAAO Research Committee (2014[14]), “Staffing in assessment offices in the United States and Canada: Results of 2013 survey”, Journal of Property Tax Assessment & Administration, Vol. 11/2, pp. 5-62; Cour des Comptes (2017).
Buoyancy/stability
As compared to property values, which can grow rapidly in boom periods, and decline during recessions, taxable areas are relatively stable over time. To assure that Arnona revenues grow as the costs of providing public services increases, Arnona rates are increased annually to reflect changes in the Israeli cost-of-living index. The resulting modest annual increases in the Arnona charges faced by households and businesses are likely to enhance the political acceptance of the tax. Also, linkage of Arnona rates to the cost-of-living index assures the steady flow of income for local governments even during downturns in property markets (Daran, 1999[12]).
Transparency and simplicity
At first blush, the Arnona is a simple and transparent tax, depending only on a few easy- to-measure factors, area, age, locations, and type of use. Complexity arises when local governments, in an attempt to increase the fairness of the tax, add additional sub-categories to the type of use classification scheme, add spatial zones with separate tax rates, or special discounts for specified groups of taxpayers. It is the understanding of the OECD team that information on each parcel of property, such as size, age, and Arnona charge is neither publicly available, nor reported to the Ministry of Finance and/or the Ministry of the Interior. This lack of transparency makes analysis of the implementation of the Arnona very difficult if not impossible. In many parts of the United States, physical characteristics of each property, including age and size, and the assessed market value of the property are all easily available on a public website maintained by local governments.
Autonomy
Since the mid-1990s the OECD has been conducting periodic studies of the degree of tax autonomy among state/provincial and local governments in OECD member countries. The OECD has developed a taxonomy to describe the degree of tax autonomy that exists at the subnational level in each country. As part of the taxonomy, a country whose local governments set their own tax rates and tax reliefs without having to consult a higher level government is assigned code “a1”. A country where the central government sets local government tax rates and defines tax bases is coded as “e”. Varying degrees of limitations to full autonomy, including tax sharing, are given “b”, “c”, or “d” codes. In the latest available tax autonomy analysis, over 95% of the tax revenue raise by local governments in Israel is assigned a code of “e”, reflecting the fact that Arnona rates are set by the central government and any changes to those rates must be approved by the Ministries of Finance and Interior (OECD, 2019[15]).
Despite their very limited autonomy with respect to tax rates, local governments have discretion over how they choose to measure the area subject to the Arnona, the definition of sub-classifications of the major types of property, and some degree of discretion over the awarding of Arnona exemptions and discounts. As discussed above, this limited autonomy comes at the cost of increasing horizontal inequities in the Arnona system both within and across local jurisdictions. It also increases the complexity of the system and reduces transparency.
Summary evaluation
The Arnona is the main source of tax revenues for local governments. It provides a growing, yet stable source of revenue for local governments, and is relatively inexpensive to administer. Compared with other possible sources of local government tax revenue, the Arnona is a quite efficient tax. However, as will be discussed in the next section, the Arnona creates an incentive for local governments to discourage new housing development in favour of, in some cases, excessive non-residential development. Although the absence of data prevents a complete evaluation of tax equity, it appears that the Arnona is a regressive tax and that Arnona payments differ substantially among households in similar economic situations. If the Arnona were a more transparent tax, it is likely that these inequities would increase public opposition to the tax.
There is a widely-held view that the Arnona system has adverse effects on the land-use and planning decisions of local governments in Israel (Eckstein et al., 2014[16]; Gruber, 2014[17]; Fitoussi, Yakir and Sarel, 2016[18]). The basic argument is that the Arnona incentivizes local governments to discourage new residential development, particularly for low and moderately priced housing units, and encourages local governments to reserve too much land for commercial and industrial use.
Although precise calculations are hard to find, it appears that, with the exception of high-income households, the costs of providing public services to new residents exceed the revenue associated with these new residents. The additional revenue comes primarily from the Arnona paid by new residents and from increases in formula-based central government grant revenues attributable to an increase in local government population. The gap between the costs of providing public services and the revenue generated by new residents is particularly large for residents who qualify for Arnona exemptions, reliefs, or discounts. Low-income households, the disabled, and the elderly, are not only eligible for discounts, but are likely to have above-average needs for local government-provided public services.
In contrast, the local tax revenue generated by new commercial and industrial development is likely to substantially exceed the costs of the public services required by the new development. A major reason that commercial and industrial development generates a fiscal surplus is that, as shown in Table 1.4, non-residential Arnona rates are much higher than residential rates.
The fiscal calculus described above provides an incentive for local governments to use the planning and land-use tools available to them to discourage new residential development and to encourage the expansion of non-residential developments. While responding to these incentives may strengthen the short-run fiscal condition of local governments, from a national perspective, these actions are likely to have serious adverse consequences.
Over the past five years, Israel’s population has grown at an annual rate of about 2.0%. This population growth has exacerbated problems of inadequate housing supply in many local communities. A consequence of inadequate housing supply is a rise in housing prices and a growing problem of housing affordability. The fact that the additional local government revenues associated with new residents is inadequate to cover the costs of providing services for these new residents discourages local governments from developing new housing. For example, the City of Netanya, despite a shortage of affordable apartments, has, in recent years, focused on developing large, luxury apartments designed to attract new high-income residents (OECD, 2017[19]).
The potential fiscal benefits of non-residential development have encouraged many local governments to aggressively develop industrial parks or commercial-industrial zones. If these zones are successful in attracting new commercial and industrial development, they will likely generate a substantial amount of local government tax revenue that can support a robust level of public services. However, developing an industrial park is no guarantee that it will attract new enterprises. In recent years, a number of local jurisdictions have discovered that the supply of land available for commercial and industrial development exceeds the market demand. Especially in the North and South, away from the Tel Aviv-Jerusalem axis, many established commercial-industrial parks remain largely empty. The result is that land will remain unused and unavailable for alternative uses, including for new housing development.
Although Israel’s Arnona system is quite unique, the fact that local government taxes can influence land use and planning decisions is a widely observed phenomena throughout the world. Local governments nearly everywhere have an incentive to assure that they have enough revenue available to provide the public services for which they are responsible. Different local tax systems will imply different patterns of land-use incentives. For example, local governments with access to a sales (or consumption) tax, have an incentive to zone land for retail development.
The term, the fiscalisation of land-use, is often used to describe local government land-use decisions that are influenced by the perceived impact of economic development on the future revenues and spending of local governments. In a number of countries, local governments routinely conduct “fiscal impact analyses” to help them determine what kind of land use to encourage and what kind to discourage. The results of these analyses are often used to discourage or zone out moderate-income housing, or housing units that are appropriate for families, under the assumption that the extra property tax revenue from these new housing units will be insufficient to finance the extra costs, especially for public education, associated with new residents.
Fiscal impact analyses are frequently criticised for their overly narrow and short-term perspective. While housing units for families may result in increased education costs, in some circumstances additional students result in substantial increases in grants for higher-level governments in support of education. Zoning out moderate-income may be short-sighted, as it may make it harder for businesses to find workers, and ultimately end up discouraging commercial and industrial development. Blöchliger et al. (2017[20])provide a number of examples from various countries of the influence of fiscal policies on the land-use decisions of local governments. For example, they cite a study that showed that after the United Kingdom shifted the taxation of commercial property from local governments to the central governments, local governments took steps to discourage new commercial development.
In Israel, the substantially higher Arnona rate on non-residential property as compared with residential property encourages local governments to strongly favour new commercial-industrial development to new residential development. However, evidence from the United States indicates that fiscal incentive to favour non-residential development exist even when residential and non-residential property face identical property tax rates. In some U.S. states, for example, in Wisconsin, the state constitution requires that within a local jurisdiction, residential and non-residential property tax rates must be identical. Despite this uniformity requirement, local governments often prefer new business development because they believe that new non-residential development generates more revenue and requires less spending on services than most new residential development.
Compared with other countries that utilize the property tax as a source of local government funding, local governments in Israel have little control over their property tax revenues. They have limited ability to adjust Arnona rates and to determine criteria for awarding Arnona discounts. As a result, the fiscal benefits of new residential development are modest, and may well be negative. In countries that utilize value-based property tax systems, successful new residential developments generate increased property tax revenues as property values grow. One can think of a value-based property tax system as a mechanism for local governments to capture some of the increased property value that is attributable to local government decisions with respect to zoning, the issuance of building permits, and the creation of public infrastructure that enables the new development. With an area-based system, the potential revenue benefits of new development are limited. This is especially true if high housing prices lead households to prefer smaller housing units.
A recent report published by the Kohelet Policy Forum (Fitoussi, Yakir and Sarel, 2016[18]), an analysis of the Israeli housing market by Noam Gruber (2014[17]), and an OECD (2017[19]) study of spatial planning and policy all discuss several barriers to increasing housing supply in Israel. Although they highlight the negative incentives towards housing development attributable to the Arnona system, they also discuss several other factors that deter increases in housing supply.
Over 90% of land in Israel is publicly owned. To determine the use of this land, to ensure adequate land for agriculture, and to protect environmentally sensitive land from development, Israel has a highly centralized planning system. Although recent legislation has decentralized some planning authority, the overall planning system remains complex, hierarchical, and highly bureaucratic (OECD, 2017[19]; Gruber, 2014[17]). As a consequence, the process of obtaining planning approval and a building permit for a new residential development took over a decade (Gruber, 2014[17]). This is in sharp contrast to the situation in many OECD countries, where the approval process is often completed in a matter of months.
As a mechanism for increasing housing supply and for addressing the long delays in the planning and approval process, the government in 2013 authorized the establishment of “umbrella agreements” (Heskem Gag) with local governments (Eshel and Hananel, 2019[21]). Under the terms of these agreements, the government would agree to sell land to local governments for the express purpose of building large-scale housing developments. As part of the agreements, the planning approval process and the issuance of building permits would be “fast tracked”. As with all new development, local governments would be able to levy development fees that would cover a portion of the public infrastructure needed to support the new housing developments.
The decision to establish umbrella agreements appears to be a recognition that the complex and very slow process of gaining planning approval for new residential development was a critical factor in limiting housing supply. Furthermore, the inclusion of a substantial per housing unit grants to some local governments signals that the government recognizes that some local governments need additional resources to finance the physical infrastructure necessary to support new residential development. This infrastructure includes water and sewer lines, streets and street lights, plus new public facilities such as school buildings, recreation facilities, and parks.
Although over 90% of land in Israel is publicly owned, about 50% of new construction is occurring on privately-owned land (Gruber, 2014[17]). Nevertheless, substantial hurdles exist for increasing housing development on privately-owned land. Even with recent reforms, the process of obtaining building permits is complex and time-consuming. These problems appear to be particularly severe in Arab-dominated communities, where most of the land is privately owned. Requirements that land cannot be sold for development unless a portion is given over for public use further constrains the development of new housing.
Reforms to the Arnona system have the potential to increase local governments’ incentives to expand the development of new housing. However, the impact of any reform is likely to be limited unless it guarantees local governments sufficient revenues – from development fees or from government grants – to finance the infrastructure and operating cost of the public services needed to support the new development.
A complete analysis of the financing of local governments includes not only a discussion of local government revenues, but also consideration of the delivery of public services for which local governments are responsible. This section will focus primarily on the regular budgets of local authorities. While the construction of capital infrastructure is crucially important in local governments’ ability to provide public services, we lack data on both the details of the irregular budgets of local authorities and on their inventories of capital infrastructure. Also, the Arnona, the central focus of this report, plays only a minor role in financing infrastructure.
The discussion in this section starts with an analysis of the pattern of per capita spending from the regular budgets of local governments on the major types of expenditures. That discussion is followed by a closer look at local governments’ two major delegated functions, education and social welfare. Available information on the number of local government residents served, namely students and social welfare clients, allows for a fuller discussion of the adequacy of funding for these two important public services.
Local government per capita expenditures by function
Table 2.1 presents descriptive statistics on per capita spending from regular budgets by major expenditure functions. Average per capita expenditures in 2017 was NIS 7 854, but median per capita spending was significantly lower (NIS 7 172), revealing that a number of local authorities have relatively low per capita expenditures. Whether based on comparing standard deviations to averages or observing the range of per capita spending, the variation in spending across local jurisdictions is quite large. For total per capita expenditures, the standard deviation is 38% of the average, and the range is from NIS 3 481 to 22 741.
The bottom panel of Table 2.1 illustrates average spending differences across socio-economic clusters. Except for the top cluster, which has a very high level of per capita total spending, the difference in average per capita spending across clusters are modest. However, within clusters there is a substantial amount of variation in per capita spending across all expenditure functions.
Table 2.2 illustrates the differences across socio-economic clusters in how per capita spending varies among expenditure functions. The differences are quite large. The share of expenditures devoted to delegated functions (mainly education, social welfare and culture) is on average 54.7% of total expenditures. In cluster 1, however, it represents more than 76% of total spending, and in cluster 10, only 38%. Among delegated functions, 60% of total spending goes towards education in cluster 1, 39% in cluster 9, and only 16% in cluster 10. The share of total expenditures devoted to social welfare in the highest clusters is about half of the share in the bottom clusters. On average, spending on statutory functions (mainly general administration, water and sewage) represent approximately one third of total expenditure. The dispersion on general administration expenses per capita is relatively low suggesting that the costs of managing local authorities are more or less proportional to population across the clusters. The same pattern applies to per capita expenditures on water and sewage.
The category of spending that varies the most across socio-economic clusters is “other statutory functions.” While only about 6% of total spending goes to this category in cluster 1, nearly a quarter of total spending in cluster 9 and over a third of spending in cluster 10 are devoted to other statutory functions. This pattern reflects the fact that per capita own revenue is substantially higher in higher-numbered clusters than in lower-numbered clusters. In addition, fiscal needs related to delegated functions use up nearly all available resources in many jurisdictions in the low-numbered clusters. The result is that jurisdictions in low-numbered clusters end up with very little discretionary revenue to spend on other statutory or discretionary functions.
The differences in per capita expenditures illustrated in Table 2.1 and Table 2.2 arise for several reasons. The majority of local government spending goes to delegated functions. Thus, local government must perform these functions, and the amount of spending on these functions will depend to a large degree on the minimum amount of money needed to perform these functions. Economists refer to these minimum amounts as costs. Take the example of education. The Ministry of Education defines a set of educational standards and then provides local authorities with a grant that funds most, but not all, of the spending that the ministry believe each local government will need to meet its educational standards. The amount of money will depend on the number of students and on a set of other factors that reflect differences in per pupil costs of achieving the standards. Resource costs required to educate students are likely to depend on whether they come from Hebrew-speaking families, have mental or physical disabilities, or are being raised in poor families. Together these factors, most of which are out of the control of local authorities, will determine the fiscal needs of each jurisdiction. Actual spending may differ from fiscal needs for a number of reasons. Some local governments will choose to provide education in ways that are more expensive than in other communities, some local jurisdictions will waste resources, while other local governments, particular those with higher levels of own resources, and thus in higher socio-economic clusters, will choose to provide higher quality or supplemental education for their students.
As noted in Table 2.1, except for cluster 10, average expenditures per capita are quite similar across clusters. In contrasts, as shown in Figure 2.1, both per capita Arnona revenues and other own-source revenues rise as socio-economic conditions improves. Arnona revenues per capita in cluster 1 are only NIS 296, but they rise steadily as one moves to higher clusters. In cluster 9, average per capita Arnona revenues are NIS 3 513. Other revenues raised by local governments average NIS 769 per capita in cluster 1 and rise to NIS 1 937 per capita in cluster 9. Average per capita spending, however, is actually NIS 700 higher in cluster 1 than it is in cluster 9. There are two reasons for this pattern of own revenue and spending. First, grants from the central government play a much more important role in funding expenditures in jurisdictions in low socio-economic clusters, with some grants being targeted to jurisdictions with low socio-economic index values. Second, except for local governments in the highest socio-economic clusters, spending by local authorities is almost entirely on delegated functions and on required statutory functions. Local authorities thus have very little spending autonomy, and given the distribution of transfers, almost no options for using own resources to increase spending levels.
The funding of public education
The central government delegates to local governments the responsibility for providing primary and secondary education. At the same time, the central government gives local governments categorical grants that must be spent on public education. Local governments operate schools. Subject to available resources, they supplement central government education grants with their own revenue raised from the Arnona or form other local sources. In this section, we ask the following questions:
Do government grants provide enough resources to local governments to compensate for limited Arnona revenue?
Although Israel’s investment in public education has increased substantially in the past few years, and average student academic performance has risen, it is widely accepted that large inequities continue to exist within the current system (Blass, 2018[22]; Ben-David, 2018[23]; Gruber, 2017[24]; Wolff, 2017[25]). While it is recognized that standardized testing provides an imperfect measure of the quality of education, both national tests (Meitzav) and international tests (TIMMS and PISA) show both relatively low performance of Israeli schools relative to schools in other OECD countries, and large differences in academic performance between Hebrew and Arabic schools. For example, Figure 2.2 provides a comparison of the performance of 15-year old students on the PISA mathematics test.
The most recently available PISA results show clearly that academic performance in Israel is substantially below the OECD average. The results in Figure 2.2 are from the mathematics test taken by 15-year old students. Qualitatively similar results are found for the reading and the science assessments. In 2015, there were wide differences in the mathematics performance of students in Hebrew schools (state or religious schools) and student in Arabic schools. The students in the Hebrew schools performed at about the OECD average, whereas the performance of students in the Arabic schools was below the average achievement level in all OECD countries. Student testing within Israel finds that within the Hebrew and Arabic groups, large differences exist in the academic performance of Haredi students and other Jewish students, and between Bedouins and other Arabic-speaking students.
Source: OECD (2015[26]), Programme for International Student Assessment (PISA), 2015 Mathematics Assessment, OECD, Paris.
These wide differences in academic performance are not surprisingly reflected in the distribution of skill levels of Israeli adults. The OECD has studied the variation in training and skill levels within OECD countries. As shown in Figure 2.3, the within country variation in skills is higher in Israel than in a broad selection of OECD member countries. These skill disparities translate directly into differences in labor productivity and income. Although not the sole determinant, these skill differences are an important contributor to poverty and income inequality in Israel (OECD, 2018[27]). Skill differences are particularly pronounced in some communities. Large wage gaps between Israeli-Arabs and Haredim, and the rest of the population can be for the most part explained by differences in skill levels. Israeli labor market is highly polarized between sectors supplying high-quality jobs for high-skilled workers and low productivity sectors, which attract workers with limited skills.
Note: The measure of variability is the interquartile range, which is the difference between the third and first quartiles in the distribution of skills. Higher values reflect more variability. Data indicated as Belgium correspond to Flanders; GBR1 = England and GBR2 = Northern Ireland.
Source: OECD elaboration based on OECD (2018[27]), OECD Economic Surveys: Israel 2018, https://dx.doi.org/10.1787/eco_surveys-isr-2018-en.
These disparities in educational outcomes are related to the segregated education system in Israel comprised of three different streams for Hebrew-speaking students (Haredi, state-religious and state) and one Arabic stream. This fragmented education system weakens skills formation of Haredi and Arabs (Wolff, 2017[25]; Blass and Shavit, 2017[28]). Since merging all these streams into a single curriculum is politically infeasible, improving the system will require more linkages between the different streams with the objective of raising the outcome of low achievers.
Although not all education reforms require more resources, it is hard to see how the government can raise the quality of education for students who are not currently meeting national and international norms without investing more resources in their education. In order to reduce inequalities in student academic performance, the government must either reallocate existing school budgets towards under-performing schools or increase central government funding of primary and secondary education. In 2015, the latest year for which comparative spending data are available, primary school education spending per student as a share of per capita GDP stood at 22%. As shown in Figure 2.4, this level of primary school spending is at approximately the median of OECD countries. However, as shown in Figure 2.5, upper secondary school spending per pupil in Israel relative to GDP per capita is significantly below the OECD median and average.
Source: OECD (2015[26]), Programme for International Student Assessment (PISA), 2015 Mathematics Assessment, OECD, Paris.
Source: OECD (2015[26]), Programme for International Student Assessment (PISA), 2015 Mathematics Assessment, OECD, Paris.
As described previously, schools are jointly financed by the central and local governments, with in most cases the largest portion of funding coming from a grant to local governments from the Ministry of Education. Box 2.5 provides an outline of how the amount of the MOE education grant to each local government is determined. Funding varies between schools according to the school stream and the local government’s willingness and ability to provide schools with additional funding. Parents generally are required to pay some fees associated with the schooling of their children. The Ministry of Education also provides schools with categorical grants for teacher services and infrastructure, as well as support for transport services for school children, when needed. Schools in the Arab education stream tend to be underfunded relative to other schools, as they are often located in less affluent areas.
The budgeting of schools is determined by the sum of three main budgeting sources:
A basic allocation (“the basic standard”) is the main component of the school budget. It is meant to guarantee the appropriate number of teacher work hours (i.e. annual cost of one teaching hour) necessary to cover the required curriculum. It is determined by three factors:
Dedicated “baskets” based on the school’s student profile. These budgets are automatically transferred to the schools according to fixed formulas. They are usually dedicated funds, and are “earmarked” for specific purposes. The main baskets of this type are the:
1. Nurture basket — intended to reduce education and academic gaps between population groups, based on socioeconomic criteria.
2. Mainstreaming basket — for mainstreaming students with special needs in regular classes within regular schools. This basket, which is automatically allocated to official schools, can be considered part of the basic standard.
3. Prayer time basket — specifically for Hebrew State-religious schools and schools with specialized Jewish studies. Resources are provided to fund teacher-supervised prayer every morning.
4. Immigrant integration basket — intended for new immigrants and returning residents.
Other allocations (“baskets”) There are additional discretionary baskets (such as those for road safety programs or life skills programs) based on a school’s inclusion in specific projects, or that stem from a recognition of a special need. Allocation is at the discretion of the Ministry of Education based on ministry guidelines. Discussion with local officials, including principles, and/or pressure from public entities can influence allocations.
Source: Blass, N. and H. Bleikh (2018[29]), “The determinants of the educational budget: Per student and per class”, in Weiss, A. (ed.), State of the Nation Report-Society, Economic Policy in Israel, Taub Center for Social Policy Studies in Israel.
As shown in Table 2.3, public education spending per student varies substantially across local governments. The differences across socio-economic clusters are striking. Education spending per student rises as one moves from lower to higher clusters. Spending is roughly two times higher in clusters 5 to 9 than in clusters 1or 2. This pattern reflects in part the lack of Arnona resources in the lowest clusters. Arnona revenue per student is 36 times higher in cluster 9 relative to revenue in cluster 1. The higher per student spending levels in higher clusters are also partially due to the fact that the per student Ministry of Education grant is higher in upper clusters than in lowest clusters. The ratio of the Ministry of Education grant to per student education spending decreases from 94% in cluster 1 to 78% in cluster 4.
The ratio is approximately constant for clusters 5 to 8 at around 70%. It is striking that the per student MOE grant is 50% higher in cluster 9 than in cluster 1. Blass and Bleikh (2018[29]) describe the incredible complexity of the grant allocation formulas and describe in some detail how some policy decisions may partially counteract the goal of providing larger per student grants to lower socio-economic clusters.
In all clusters, per student education spending is higher than the Ministry of Education grant. These gaps must be filled from own revenues or from other grant revenues, for example from the Balance Grant. In cluster 1, almost all Arnona revenue would be needed to fill the gap between the MOE education grant and education spending. However, in higher clusters, although the size of the nominal per student gap is larger, the gaps can be filled by an increasingly smaller share of Arnona revenues.
Despite the relatively low levels of Arnona revenue per student in the lowest clusters, it is striking that the balancing grant per student are below NIS 4 000 in the two lowest clusters, and at NIS 5 474, highest in cluster 5.
Table 2.4 highlights the ethnic/religious dimension of the education spending issue. Per student spending was 12% above average in the (predominantly) Jewish local authorities, while predominantly Arabs local authorities spent 36% less than average, and mixed/other local authorities 18% below average. The Ministry of Education grant finances two-thirds of the education expense in the Jewish or the mixed/other sectors, but more than 88% in the Arab sector. Nevertheless, the average grant par student in the Jewish sector is above the average grant in the Arab sector by 8%.
The pattern of education spending and MOE grant per student varies across socio-economic cluster and by ethnic/religious classification. In the Jewish sector, the largest per student grants go to local jurisdictions in clusters 5 to 9, and the average per student grants in clusters 9 and 10 and nearly three times larger than the average grants in clusters 1 and 2. The distribution of the Ministry of Education grant differs completely in the Arab sector. 80% of Arabs local authorities are located in the three lowest clusters. They receive relatively high per student grants, although these grants are lower than the average per student grants in predominantly Jewish clusters 5 through 9. Students in the three highest Israeli-Arab clusters (5 through 7) receive lower average grants than students in the same clusters who live in Jewish or mixed/other jurisdictions.
The data in Table 2.3 and Table 2.4 suggest that substantial inequities exist in the funding of education in Israel. The current allocation of grants does not adequately compensate for the unequal distribution of Arnona revenue, with the result that per student support for education lags in many Israeli communities, especially Arab Israeli communities and all communities with poor socio-economic status. These data support the conclusion of the OECD (2018[27]) Economic Survey of Israel that “disadvantaged schools need much greater funding” (p. 47).
The funding of social welfare
More than 75% of the budget of the Ministry of Labour and Social Welfare is earmarked for social services in local authorities. Local social welfare services are jointly funded by the central government through the Ministry of Labour and Social Welfare and by the local authorities. The local contributions come primarily in the form of a 25% matching rate to the social welfare grant. Social service activities include services provided directly by local authority-employed social workers and by services outsourced to various agencies. Many social welfare services are provided to clients in their homes or in local social service agency facilities, although some services, such as long-term care, are provided elsewhere.
The Ministry of Labour and Social Welfare grants are allocated to local authorities using a set of formulas related to various social services provided and financed by the Ministry. Allocations depend in part on eligibility conditions for those in need. Based on these formulas, the Ministry determines the budget earmarked to each local authority at the start of each budget year. This initial budget is re-examined over the course of the year, in accordance with new needs that may have arisen, and the local authority’s willingness to provide its share of the funding. Toward the end of the year, the Ministry transfers remaining budget sums (some of which come from localities that did not utilize the sums that the Ministry had earmarked for them) to localities in need of additional funds and that are able to match the additional levels of spending. The Ministry also transfers to the localities additional budgets to fund new social welfare programs that were started during the year. In addition, local authorities are authorised to expand the social services provided within their area of jurisdiction through self-funding. These local funds may come from local resources such as Arnona revenue or from external non-governmental funding, such as from non-profit foundations.
While social welfare spending as a percentage of total budgetary expenditure tends to decline as locality socio-economic status rise, the data in Figure 2.6 shows that total social welfare spending per social service client rises as socio-economic status rises. The average social welfare expenditure per client in local authorities belonging to clusters 1-3 (most of the Arab Israeli authorities are concentrated in these lower socioeconomic clusters) is lower than the average expenditure in localities belonging to clusters 4-10. These disparities are even more marked when one compares the budgets of affluent and poor local authorities, and of Jewish and Arab Israeli localities (Gal, Shavit and Bleikh, 2017[30]). These spending gaps are related to differences in the types of client populations within the various communities, variations in local authority’s service provision patterns, and the willingness of stronger localities to increase social welfare spending beyond the allocations provided by the Ministry of Labour and Social Welfare. The spending disparities also appear to be rooted in the weaker localities’ inability to commit to funding their share of out-of-home services for their residents, and in a shortage of out-of-home institutions serving needy Arab Israelis. These funding gaps worsen already-existing inequalities in Israeli society and harm the country’s weakest population segments.
Figure 2.7 shows that after accounting for the financial contributions to social welfare service from local authorities, the funding gaps per client between different groups of jurisdictions grow even larger than the gaps observed in grant allocations. For example, the Jewish local authorities’ total per-client expenditure was NIS 7 318 in 3024— 2.2 times the Arab Israeli authorities’ average expenditure. We have seen that the Haredi local authorities exhibit a relatively high degree of additional municipal expenditure given their socioeconomic standing; their total per-client expenditure amounts to NIS 8 750 — NIS 2 500 more than the total average per-client expenditure for all local authorities. However, this disparity spending per client between the Haredi local authorities and the other groups begins with their higher grant allocation from the Ministry of Labour and Social Welfare at the start of the year.
Source: Adapted from Figure 4 in Gal, J., M. Shavit and H. Bleikh (2017[30]), “Social service budgeting in Israeli local authorities”, in Weiss, A. (ed.), State of the Nation Report-Society, Economic Policy in Israel, Taub Center for Social Policy Studies in Israel.
Note: Forum 15 is group of 15 local governments that do not receive government development or balance grants. Haredi jurisdictions are those with a majority of Haredi Jewish residents.
Source: Adapted from Figure 15 in Gal, J., M. Shavit and H. Bleikh (2017[30]), “Social service budgeting in Israeli local authorities”, in Weiss, A. (ed.), State of the Nation Report-Society, Economic Policy in Israel, Taub Center for Social Policy Studies in Israel.
The role of grants in reducing local government fiscal disparities
Large differences exist across local governments in their per capita revenues from both the residential and non-residential Arnona. As illustrated in Figure 2.1, average per capita revenues differ among socio-economic clusters. Average per capita Arnona revenues are lowest in jurisdictions in cluster 1 and increase steadily for local governments in higher clusters. These differences reflect differences across jurisdictions in taxable area, in tax rates, and in the administration of the tax. The square meters of developed property and land used for various purposes defines the tax base of each jurisdiction. Although Arnona tax rates are set by the central government, actual rates by type of property can vary because local governments can define sub-classifications of property types and apply different rates to various sub-classifications. Local governments can also request permission from the Ministry of Finance to change rates. Local governments also have some discretion in determining how the Arnona is administered. They can vary the level of discounts granted certain taxpayers, and by their actions and inactions influence their Arnona collection rates.
In nearly all countries, higher-level governments respond to differences in the fiscal conditions of their local governments by allocating grants-in-aid, usually in a manner designed to reduce fiscal differences. A major concern in the design of intergovernmental transfer systems is to avoid the creation of incentives for inefficient behaviour on the part of local governments. For example, providing larger grants to local governments with low levels of Arnona revenue would provide those governments an incentive to issue more discounts and reduce efforts to collect Arnona bills. To prevent these responses and to avoid creating soft budget constraints for local governments, most equalising grant programs base grant allocations on the revenue (or tax) capacity of local governments. Changes in local government tax behaviours or efforts will affect tax revenues but will have no impact on its tax capacity. As described in Box 1.11 (Chapter 1), the allocation of balance grants depends in part on a sophisticated tax capacity measure known as potential revenues.
A commonly used and straightforward measure of tax capacity is the representative tax system (Chernick, 1998[31]). As applied to the Arnona it multiplies the per capita tax base for each type of property in each local government, e.g., the total square meters of residential property per capita, by the average tax rate applied to that type of property across all local governments. The Arnona tax capacity of each local government is then calculated by summing across all types of real property.
Observing the distribution of per capita tax capacities across local governments provide one measure of the degree of inequality that would exist across local jurisdictions if there were no transfers from the central government. For a complete measure of the fiscal disparities among local governments one needs to also consider variations across local governments in their expenditure needs (Boadway and Shah, 2007[32]). As explained in Box 1.11 (Chapter 1), this concept takes the form of normative expenses in the calculation of general balancing grants. As the development of a measure of expenditure needs is beyond the scope of this report, here we explore several measures of tax capacity inequality across local governments, and then assess the degree to which measured inequality is reduced by Israel’s current system of grants to local governments.
Table 2.5 displays several inequality measures that are applied to residential, non-residential, and total Arnona tax capacity and tax revenue, all measured in per capita terms. The first measure is the coefficient of variation, which is simply the standard deviation of the distribution of capacity or revenue divided by the average (or mean) value. The second measure is the range, which is the difference between the minimum and maximum values, and the third measure of inequality is the Gini coefficient. A Gini coefficient of zero represents perfect equality, while a value of one represents maximum inequality.
As would be expected, the inequality in residential tax capacity is relatively small. While home values vary substantially across the country, the variation in housing consumption, measured in housing area, is much smaller. The Gini coefficient is quite low and the coefficient of variation is modest. Inequalities in non-residential tax capacity are much greater, however. All three measures of inequality show that there is a high degree of variation across local governments in their capacity to raise property tax revenue. Adding together residential and non-residential capacity reduces the measures of inequality observed for the non-residential Arnona. The coefficient of variation at 0.85 is quite high, as is the Gini coefficient, at 0.33.
Comparing the distribution of actual revenue to the distribution of tax capacity, one finds that inequalities are substantially larger for the residential Arnona, while they are smaller for the non-residential Arnona. For the average local government in clusters 1 through 7, actual residential revenues are substantially lower than residential capacity. This probably reflects a combination of a couple factors. As illustrated in Table 1.7, discounts relative to gross charges are higher in the low-numbered clusters, and the collection rates are also substantially lower. These factors lead to greater variation in actual residential Arnona per capita revenues relative to variations in tax capacity. For the non-residential Arnona, the differences between capacity and actual revenue are much smaller. While average non-residential revenues are lower than capacity in the bottom eight clusters, non-residential revenues are larger than capacity in the top two clusters, suggesting that clusters with the highest socio-economic ranking are utilizing above-average tax rates. Differences in the distribution of total Arnona capacity and total Arnona revenue are muted. The inequality measures provide a mixed message. Inequalities in tax capacity are larger than inequalities in actual revenues when measured by the coefficient of variation and the range, but they are smaller than inequalities in actual revenue when the Gini coefficient is used. This difference may be explained by the fact that the Gini coefficient is less influenced by the extreme values of a distribution than is the coefficient of variation.
To determine whether the spatial pattern of growth and development in Israel has increased or decreased fiscal disparities over time, we calculated local government fiscal capacity related to the Arnona using data for 2009. We then compared the coefficients of variation and the Gini coefficient for residential and non-residential Arnona for 2009 and 2016. The data show that while the distribution of the residential Arnona capacity remained essentially unchanged, the inequalities in non-residential tax capacity were substantially reduced. The coefficient of variation went from 1.98 to 1.64, and the Gini coefficient dropped from 0.690 to 0.577. Thus, although the disparities in non-residential tax capacity remain large, the direction of change has been towards equalisation. It is important to emphasize that rapid economic development in the past few years in the central part of Israel may have a substantial impact on both residential and non-residential fiscal disparities within Israel.
More important than the size of equalising grants is their equalising performance (or “effectiveness”) in reducing fiscal disparities among subnational governments. Using two measures of fiscal disparities—Gini coefficients and the ratio of highest to lowest tax capacity--an OECD report explores the impacts of equalisation grants on fiscal disparities among subnational governments (OECD, 2013[33]). The report finds that equalisation grants in some countries, for example, Australia, Italy, Norway, and Sweden, were particularly effective in reducing fiscal disparities. On the other hand, equalisation grants in Germany and Switzerland had smaller impacts on the reduction of fiscal disparities.
As equalising grants systems are often composed of various grants of different sizes and allocated using different formulas, understanding the overall equalising performance of the grant system requires analysing the marginal contribution to equalisation of each component grant, and diagnosing how individual grants interact. This can be accomplished using appropriate statistical indexes to measure inequality across local governments. Overall equalising performance results from the combination, for each of the equalising grants, of a “size component”, accounting for the magnitude of each grant, and of a “marginal equalising component,” which measures the effect (if any) on an inequality index of the distribution of each grant. See Gilbert and Guengant (2004[34]; 2005[35]), for a detailed description of the methodology and an application to the highly complex equalising grant system in France.
Sources: OECD (2013[33]), Fiscal Federalism 2014: Making Decentralisation Work, https://dx.doi.org/10.1787/9789264204577-en; Gilbert, G. and A. Guengant (2005[35]), “Evaluation de la performance péréquatrice des concours financiers de l’Etat aux communes”, Economie et Statistique 373, pp. 81-108; Gilbert, G. and A. Guengant (2004[34]), “The equalizing performance of the central government grants to local authorities; the case of France”, Papers and Proceedings, pp. 192-202.
To assess the extent to which the Israeli system of intergovernmental grants has reduced the underlying fiscal disparities among local governments caused by differences in Arnona tax capacity, we recalculated our inequality measures after accounting for the distribution of grants. For easy reference, the descriptive statistics for total Arnona tax capacity are repeated in the first column of Table 2.6. The second column displays descriptive statistics for the sum of Arnona capacity and the per capita balance grant. As the balance grant was explicitly designed to reduce fiscal disparities among local governments, it is not surprising that the variation in resources is more equally distributed after receipt of the balance grant. The coefficient of variation declines from 0.85 to 0.68, and the Gini coefficient falls from 0.33 to 0.255. The final column presents the results from adding Arnona tax capacity and all government grants going to the regular budget. Both the coefficient of variation and the Gini coefficient are reduced by adding other central government grants.It is important to note that the balancing grant is much more equalising than the other grants going to local governments, even though, as shown in Figure 1.9, the general balancing grant made up only 15% of the total government grants going to the regular budgets of local governments. The conclusion is that aside from the very small, equalisation fund grants, the largest grant program, the education grant, has only a small impact on reducing fiscal disparities among local governments.
The data in the final column of Table 2.6 demonstrate that whereas government grants have been successful in reducing fiscal disparities that exist among local governments, there remain substantial differences across local government in their capacity to provide the public services for which they are responsible. These differences in resources will continue to translate into differences in the education, the skills, and the opportunities available to Israelis depending on where they live. Moreover, these data are likely to understate the true magnitude of fiscal disparities among local governments because they focus only on resource differences and not on differences in the expenditure needs and the costs of delivering public services in different communities.
References
[23] Ben-David, D. (2018), The Education System That’s Endangering Israel’s Future, Haaretz, https://www.haaretz.com/opinion/.premium-the-education-system-that-s-endangering-israel-s-future-1.6431884.
[22] Blass, N. (2018), “The Israeli education system: An overview”, in Weiss, A. (ed.), State of the Nsation Report-Society, Economic Policy in Israel, Taub Center for Social Policy Studies in Israel.
[29] Blass, N. and H. Bleikh (2018), “The determinants of the educational budget: Per student and per class”, in Weiss, A. (ed.), State of the Nation Report-Society, Economic Policy in Israel, Taub Center for Social Policy Studies in Israel.
[28] Blass, N. and Y. Shavit (2017), “Israel’s education system in recent years: An overview”, in Weiss, A. (ed.), State of the Nation Report-Society, Economic Policy in Israel, Taub Center for Social Policy Studies in Israel.
[3] Blöchliger, H. (2015), “Reforming the Tax on Immovable Property: Taking Care of the Unloved”, OECD Economics Department Working Papers, No. 1205, OECD Publishing, Paris, https://dx.doi.org/10.1787/5js30tw0n7kg-en.
[20] Blöchliger, H. et al. (2017), “Local taxation, land use regulation, and land use: A survey of the evidence”, OECD Economics Department Working Papers, No. 1375, OECD Publishing, Paris, https://dx.doi.org/10.1787/52da7c6a-en.
[32] Boadway, R. and A. Shah (2007), Intergovernmental Fiscal Transfers: Principles and Practice, Public Sector Governance and Accountability Series, The World Bank, Washington, DC.
[2] Cabral, M. and C. Hoxby (2012), The Hated Property Tax: Salience, Tax Rates, and Tax Revolts, National Bureau of Economic Research, Cambridge, MA, https://doi.org/10.3386/w18514.
[31] Chernick, H. (1998), “Fiscal capacity in New York: The city versus the region”, National Tax Association, Vol. 51/3, pp. 531-540.
[12] Daran, D. (1999), “A politician’s appraisal of property taxation: Israel’s experience with the Arnona”, Lincoln Institute Working Paper, No. WP99DD1, Lincoln Institute of Land, Cambridge, MA.
[16] Eckstein, Z. et al. (2014), “Do local authorities face a negative incentive to increase the population under their jurisdiction?”, Policy Paper-GGA, Gazit-Globe Real Estate Institute, http://gazit-globe.idc.ac.il/en/files/Policy%20Paper%20-%20Local%20Authority%20Incentives%20-%20March%202014%20-%20GGA2014_1.pdf.
[21] Eshel, S. and R. Hananel (2019), “Centralization, neoliberalism, and housing policy central-local government relations and residential development in Israel”, Environment and Planning C: Politics and Space, Vol. 37/2, pp. 237-255.
[18] Fitoussi, S., I. Yakir and M. Sarel (2016), “The roles and preferences of the local authorities in the development of residential real estate”, Removing the Barriers to Housing Development in Israel, No. 27, Policy Paper, Kohelet Policy Forum, http://en.kohelet.org.il/wp-content/uploads/2016/05/KPF055_eReal_estate_240416.pdf.
[30] Gal, J., M. Shavit and H. Bleikh (2017), “Social service budgeting in Israeli local authorities”, in Weiss, A. (ed.), State of the Nation Report-Society, Economic Policy in Israel, Taub Center for Social Policy Studies in Israel.
[35] Gilbert, G. and A. Guengant (2005), “Evaluation de la performance péréquatrice des concours financiers de l’Etat aux communes”, Economie et Statistique 373, pp. 81-108.
[34] Gilbert, G. and A. Guengant (2004), “The equalizing performance of the central government grants to local authorities; the case of France”, Papers and Proceedings, pp. 192-202.
[24] Gruber, N. (2017), “Why Israel does poorly in the PISA exams: Perceptions versus reality”, Shoresh Research Paper, Shoresh Institution for Socioeconomic Research.
[17] Gruber, N. (2014), “The Israeli housing market”, in State of the Nation Report 2014: Society, Economy and Policy in Israel, The Herbert M. Singer Annual Report Series, Taub Center for Social Policy Studies in Israel, Jerusalem.
[10] Horne, R. and D. Felsenstein (2010), “Is property assessment really essential for taxation? Evaluating the performance of an ‘Alternative Assessment’ method”, Land Use Policy, Vol. 27/4, pp. 1181-1189, https://doi.org/10.1016/j.landusepol.2010.03.008.
[9] Oates, W. and R. Schwab (2009), “The simple analytics of land value taxation”, in Dye, R. and R. England (eds.), Land Value Taxation: Theory, Evidence and Practice, Lincoln Institute of Land Policy, Cambridge, MA.
[15] OECD (2019), Tax Autonomy Index - Fiscal Decentralisation Network, OECD, Paris, https://www.oecd.org/tax/fiscal-decentralisation-database.htm (accessed on 24 April 2019).
[27] OECD (2018), OECD Economic Surveys: Israel 2018, OECD Publishing, Paris, https://dx.doi.org/10.1787/eco_surveys-isr-2018-en.
[1] OECD (2018), Revenue Statistics, OECD, Paris, https://stats.oecd.org/ (accessed on 16 April 2019).
[7] OECD (2018), Tax Policy Reforms, OECD Publishing, Paris, https://dx.doi.org/10.1787/26173433.
[19] OECD (2017), Spatial Planning and Policy in Israel: The Cases of Netanya and Umm al-Fahm, OECD Publishing, Paris, https://dx.doi.org/10.1787/9789264277366-en.
[8] OECD (2016), “Property tax principles”, CTPA/CFA/WP2(2016)3, Unpublished, OECD Working Party No. 2 on Tax Policy Analysis and Tax Statistics, OECD, Paris.
[26] OECD (2015), Programme for International Student Assessment (PISA), 2015 Mathematics Assessment, OECD, Paris.
[33] OECD (2013), Fiscal Federalism 2014: Making Decentralisation Work, OECD Publishing, Paris, https://dx.doi.org/10.1787/9789264204577-en.
[36] OECD (n.d.), Tax Policy Reforms, OECD Publishing, Paris, https://dx.doi.org/10.1787/26173433.
[5] OECD/KIPF (2016), “Fiscal federalism 2016: Making decentralisation work”, in Reforming the Tax on Immovable Property.
[6] OECD/UCLG (2019), 2019 Report of the World Observatory on Subnational Government Finance and Investment, OECD, Paris, http://www.sng-wofi.org.
[11] Portnov, B. et al. (2000), “Property tax, social equity and efficiency of local authorities in Israel”, Journal of Property Tax Assessment and Adminsitration, Vol. 5, pp. 33-59.
[13] Rakow, R. (2009), Email Communication, April 2.
[4] United Nations Human Settlements Programme (2013), Property Tax Regimes in Europe, The Global Urban Economic Dialogue Series.
[14] Walters, L.C. and IAAO Research Committee (2014), “Staffing in assessment offices in the United States and Canada: Results of 2013 survey”, Journal of Property Tax Assessment & Administration, Vol. 11/2, pp. 5-62.
[25] Wolff, L. (2017), “Education in Israel: Divided society, divided school”, No. 13, Joseph and Alma Gildenhorn Institute for Israel Studies, University of Maryland.