Table of contentsIntroduction ...................................................................................................................... 5Benchmark ........................................................................................................................ 62.1 Theory .......................................................................................................................... 62.1.1 The concept........................................................................................................... 62.1.2 Normative tax system........................................................................................... 72.1.3 Measurement........................................................................................................ 72.2 Criticism of tax expenditure analysis ........................................................................ 93 Analysing tax expenditures............................................................................................ 103.1 Practical issues of tax expenditure analyses .......................................................... 103.2 Reporting and evaluation of tax expenditures ........................................................ 113.2.1 Classification...................................................................................................... 123.2.2 Important elements of reporting........................................................................ 143.2.3 Challenges in evaluating tax expenditures: Effectiveness and efficiency......... 153.3 Reporting and evaluation in other countries .......................................................... 173.3.1 In general............................................................................................................173.3.2 The German evaluation of tax expenditures..................................................... 174 Country overviews .......................................................................................................... 204.1 Introduction ............................................................................................................... 204.2 Main features ............................................................................................................. 204.2.1 Definition of tax expenditures and the benchmark system............................... 204.2.2 The volume of tax expenditure........................................................................... 214.2.3 Reporting, classification and evaluation of tax expenditures...........................234.2.4 Critique from National Audit Offices................................................................244.2.5 Tax expenditure and fiscal policy...................................................................... 254.3 Tax expenditures in Denmark ................................................................................. 264.3.1 Introduction........................................................................................................ 264.3.2 Definition of tax expenditures............................................................................ 274.3.3 Methods of calculation....................................................................................... 284.3.4 Reporting and evaluation.................................................................................. 284.3.5 Overview of the most important tax expenditures.............................................304.3.6 Challenges and future development...................................................................334.4 Tax expenditures in Finland..................................................................................... 354.4.1 Introduction........................................................................................................ 354.4.2 Definition of tax expenditures............................................................................ 354.4.3 Methods of calculation....................................................................................... 374.4.4 Reporting and evaluation.................................................................................. 374.4.5 Overview of the most important tax expenditures.............................................374.4.6 Challenges and future development...................................................................394.5 Tax expenditures in Norway .................................................................................... 404.5.1 Introduction........................................................................................................ 404.5.2 Definition of tax expenditures............................................................................ 403

4.5.3 Methods of calculation....................................................................................... 424.5.4 Reporting and evaluation.................................................................................. 424.5.5 Overview of the most important tax expenditures.............................................444.5.6 Challenges and future development...................................................................464.6 Tax expenditures in Sweden .................................................................................... 464.6.1 Introduction........................................................................................................ 464.6.2 Definition of tax expenditures............................................................................ 474.6.3 Methods of calculation....................................................................................... 484.6.4 Reporting and evaluation.................................................................................. 484.6.5 Overview of the most important tax expenditures.............................................494.6.6 Challenges and future development...................................................................495 Conclusions and suggestions for future work .............................................................. 516 Literature ........................................................................................................................ 537 Appendix - What do other countries report?............................................................... 55

1 IntroductionThe background for this work is partly the increased interest in tax expenditures from partiesoutside the ministries, for instance the National Audit Offices, but also a realisation that thereis a need to produce more reliable, transparent and maybe also comparable tax expenditureanalyses in the Nordic Countries. It is our ambition that this report can shed some light onsome issues related to tax expenditures and maybe also serve as an “easy entry” to the taxexpenditure concept. We also present a comprehensive overview of how tax expenditures aretreated in the Nordic countries.The main objective of tax expenditure analysis is increased expenditure control byidentifying tax provisions that escape the scrutiny applied to regular subsidies and transferson the expenditure side of the budget. Originally tax expenditure analyses ambitiously wereregarded as tools for revealing the inferior nature of tax expenditures as a way to appropriatepublic resources. It was believed that a thorough tax expenditure analysis would unveil that alot of tax expenditures violated the objectives of efficiency, fairness and simplicity in the taxsystem. Since the first tax expenditure analysis was presented in the US more than forty yearsago, the concept of tax expenditures has spread worldwide and today virtually all OECD-countries make tax expenditure analysis in one form or another.Using tax expenditure analysis in the tax policy work to pursue such idealistic principles ofthe tax system as mentioned above is highly ambitious and demands a definition of anormative tax system which reflects the principles of efficiency, fairness and simplicity. Thishas proven to be a difficult task, and have brought about extensive discussions and alsosubstantial criticism of the definitions and methodologies used in tax expenditure analyses,not the least in the US.The examination of tax expenditure analyses of the Nordic countries in this report shows thatthe levels of ambitions are fairly moderate compared to those mentioned above, and themethodologies are to a larger extent based on pragmatic choices and prevailing tax rules.These analyses are therefore less vulnerable to the general criticism regarding the normativeassessments of the benchmark, but have on the other hand to some extent been criticised forthe lack of transparency.The working group does not work out detailed recommendations for a reform of taxexpenditure analyses in the Nordic countries. However, we make some suggestions forpossible future improvements. One suggestion is that the Nordic countries endeavour toclarify the tax expenditure analysis. Another is that more emphasis could be placed onevaluatingtax expenditures. We believe that this could help to increase both transparencyand the accuracy of tax expenditure analysis.The report is structured in a theory part (chapter 2 and 3) and a practical part (chapter 4).Chapter 2 presents the theoretical background of defining a benchmark. In chapter 3 wediscuss some important elements of tax expenditure analyses, with a specific emphasis onreporting and evaluation. In chapter 4 the treatment of tax expenditures in Denmark, Finland,Norway and Sweden is described and discussed. Each country has been responsible for itsown description. The chapter starts with a summary of common features and points towardssome differences. The challenges concerning the treatment of tax expenditures, which are

The termtax expenditurerefers to provisions in the tax code that give favourable taxtreatment for an activity or a group of taxpayers. Tax expenditures may take a number offorms like exemptions, allowances, credits, preferential rates, deferral rules etc. They are, ineffect, policy instruments of government to promote specific social or economic policies andthus are closely related to direct spending programs. Negative tax expenditures are taxsanctions. A tax sanction is the result of levying tax at a higher rate than the norm.The identification of tax expenditures is a difficult task for many reasons. The tax code mayinclude a number of provisions that lead to revenue losses but nonetheless promote thestandard goals of a good tax system. They are integral parts of the tax system and are notjudged as tax expenditures. Neither can the tax expenditure analysis rely entirely onlegislative documents, since the objectives of tax provisions are not always expressed clearlyand openly there. So, there is a need for a systematic method for identifying tax expenditures.Such an approach was developed in the US in the late 1960s, well captured in the following(Surrey & McDaniel (1985)):“The tax expenditure concept posits that an income tax is composed of twodistinct elements. The first element consists of structural provisions necessaryto implement a normal income tax, such as the definition of net income, thespecification of accounting rules, the determination of the entities subject to tax,the determination of the rate schedule and exemption levels, and the applicationof the tax to international transactions. The second element consists of thespecial preferences found in every income tax. These provisions, often calledtax incentives or tax subsidies, are departures from the normal tax structure andare designed to favour a particular industry, activity, or class or persons. …Whatever their form, these departures from the normative tax structurerepresent government spending for favoured activities or groups, effectedthrough the tax system rather than through direct grants, loans, or other forms ofgovernment assistance.”Under this approach tax expenditures are defined as deviations from a normal tax structure,also called a benchmark or a reference tax system. With this concept we may understand a‘pure’ tax system as a system that focuses on raising revenue, while also fulfilling theprinciples of a good tax system, neutrality, equality and the requirement to be implementable.According to the quotation, other criteria for tax rules to be judged as tax expenditure are thatthey favour some narrow activities or tax payer groups and that they serve some specific

Tax expenditure analysis is based on the concept of a normative tax system. This concept isthe measuring rod to identify and to measure the amount of tax expenditures. The normativetax system captures the generally accepted elements of the tax system that are necessary toachieve the fiscal goals and to fulfil the objectives of equity, efficiency and simplicity.Tax expenditure analysis requires good understanding of the objectives and functions of theelements of a particular tax. Based on such understanding the normative structure can bedefined and the tax expenditures identified as deviations from the norm. This understandingcan be drawn from some idealized tax structures developed in tax research. Income tax andVAT are good cases in point. The so called Schanz-Haig-Simons economic income concept(SHS income) has been an influential tool when specifying the norm for the income tax basein the US and elsewhere. It defines one period’s income as consumption plus the change innet wealth during the period. The concept is of course abstract and leaves several importantaspects of a normative tax system open. Therefore the benchmark tax system usually is acombination of elements from the theoretical abstract and the actual tax system. Similarly, ageneral non-cascading sales tax could be a useful abstract concept behind the benchmark taxsystem of VAT or, even more broadly, behind all taxes on private consumption.Examples of elements of the actual tax system usually accepted in the benchmark are the taxrate schedule, the unit of taxation, the time frame of taxation, accounting principles,realization principle2as the basis of taxation and the potential non-elimination of inflationgains. Some allowances motivated by redistribution of income may be part of the benchmark.Similarly deductions or relieves that are grounded on simplification of tax administration orthe tax code may be defined as parts of the benchmark.That the benchmark tax is a compromise between the theoretical ideal and the actual taxsystem has rendered tax expenditure analysis subject to much criticism in recent years,especially in the US. The concern is that under this approach the norm cannot be definedrigorously enough to ensure that the identification of tax expenditures leads to an objectiveand reliable outcome. This criticism has led to at least two new ideas in tax expenditureanalysis. One is that the connections to the theoretical ideals (SHS income) are not usefulsince the choice of the norm is mainly a pragmatic exercise. The other is that classifyingrelevant provisions into subgroups could relieve the problems from branding all items asprovisions.2.1.3Measurement

The revenue foregone method is based on the assumption that the introduction of taxexpenditures does not affect the behaviour of taxpayers or the revenues from other taxes. It istherefore the easiest estimation method. In general, there are good reasons to believe thattaxpayers change their behaviour in response to the tax expenditure (increase their demandfor the tax-subsidised good or increase/decrease their demand for income). Therefore therevenue forgone method gives a narrow picture of the revenue effects of a tax provision.However, if the tax expenditure is re-estimated the behavioural effects will indirectly beaccounted for if the tax base changes due to the tax expenditure.The method of revenue gain takes the behavioural change and the change in tax interactioninto account. Of course this makes the method much more complicated to apply in practice.Although the method is superior in principle, many governments seem to assume that theaccuracy that may be gained is not worth the efforts required to apply the method.Outlay equivalence is a measure that leaves the net budget impact (on the surplus or deficit)and the after-tax incomes of taxpayers the same in the situation with tax expenditure and inthe situation with equivalent outlay but without tax expenditure. Outlay equivalence takesinto account the fact that regular transfers are sometimes estimated gross of the tax paid bythe recipient, whereas tax transfers are by definition net of tax. In order to estimate these taxexpenditures on the same basis as regular expenditures, it is necessary to add the tax that istypically levied upon the regular transfer. Otherwise, it appears as if the tax expenditure is acheaper way to get the same amount of cash into the hands of the recipient than the regularexpenditure.Methods of measurements are a much broader matter, however.. There are at least thefollowing additional aspects:••Which methods are used? E.g. present value calculations or implementation of microsimulation models.Is the calculation one periodic or does it take into account counterbalancing effects inlater years. One example here is the so called EET model of voluntary pensionsavings which deviates from a SHS income based norm in several respects. Theessential feature is that those deviations spread over years. Another example but withsome different features are the depreciation rules. Even if the main approach would be

The aim of tax expenditure analysis is widely shared, i.e. to improve the control of the use ofgovernment resources on the revenue side of the budget. The usefulness of the concept of taxexpenditure has, however, been questioned for years, almost since its introduction in the late1960s (JCT 2008).The most important part of the criticism is focused on the concept of a benchmark tax system.The main objection is that the concept does not have a sufficiently rigorous formal basis andis more or less a result of a series of subjective, pragmatic choices. As an example, it iscommon that the tax expenditure analysis officially refers to SHS income as a starting pointwhen the normal structure for income tax is defined. Here it follows the approach introducedby Surrey. In principle comprehensive income can be defined quite precisely. However, SHSincome is not an operational concept that could be measured exactly and easily enough to beused either as a basis for taxation or as a tool in tax expenditure analysis. Therefore thebenchmark applied is usually a compromise between SHS income and the actual tax systemor, put differently, an extension of the actual tax system towards the theoretical concept.Due to the vagueness of the benchmark, the classification of at least some tax expenditureshas weak grounds and their status will easily become subject to discussions. If there is nohard theory behind the benchmark it is very hard to defend the identification decisions. Thereis much experience from such a debate from the US.One line of criticism targets the transparency of the benchmark. It can be traced back toBittker (1969) who questioned Surrey’s way of officially connecting the normal tax system tothe SHS-concept but at the same time include several other elements into it withoutimplicating the theoretical grounds. Bittker considered these additions as subjective choices.According to Burman (2003) the US tax expenditure debate has had an ideological stance.Those who have favoured an income tax have supported the current way to identify andmeasure tax expenditures. On the other hand, those who have preferred consumption taxfeatures in the tax system have proposed changes towards savings relieves in the benchmarksystem.The criticism also includes the hidden reform agenda on the background of tax expenditurereporting and also that tax expenditure analysis implies an idea of a clean and apolitical taxpolicy that does not fit well with social decision making. It seems that this kind of debate andthe resulting credibility problem are, to some extent, a result of unclear theoretical grounds ofthe identification process.Another main subject of criticism has been the measurement methods (Burman 2003). Thetax expenditure estimates usually give a static first year revenue loss due to a particular taxprovision. They do not include any behavioural responses which may be incorporated in

Making tax expenditure analyses raises a lot of practical issues on how to define andoperationalise the appropriate benchmark, and there is no universal recipe on how to solvethem. The crucial task is to define a suitable tax code and a feasible and adequate tax basethat serves as a reference. Calculating tax expenditures requires a distinction of the normativecomponents of a particular tax from its tax expenditure component, and deciding what shouldbe part of the benchmark structure and income and what should constitute a tax expenditure.Such considerations have to be made for each individual tax, which hampers the directapplication of a general theory or methodology.Tax expenditure analyses around the world differ widely both when it comes to definitionsand applied methodology. This can make tax expenditure analyses from different countriesvirtually incomparable, and a comparison of the number and amount of tax expenditures canbe rather misleading. A general definition tends to generate a lot of tax expenditures, while amore narrow definition can lead to less or almost no tax expenditures at all. In between thereare systems based on various concepts that differ along several dimensions. Some are looselybased on a “norm” where the norm can change over time (as in France) and others are veryrestrictive and detailed but not necessarily strictly based on principles (as in Canada and to3 See Joint Committee on Taxation (2008).4 See e.g. OECD (2008/1) and OECD (2008/2).

some extent the US and UK). However, since most countries are claiming more or less thesame methodology and definitions at the overall level (e.g. revenue foregone method, incometax benchmark etc.), these variations are probably more due to how the methodology and thedefinitions are applied in practice, and maybe due to the lack of practical guidelines foundedon undisputed theory.The most difficult task seems to be related to choosing the right tax base. Many countries usethe Schanz-Haig-Simmons (SHS) income definition for tax expenditure purposes. The mainreason for this is probably that the SHS definition constitutes a tax base that neutralise the taxtreatment of consumptions and savings (while a conventional comprehensive incomedefinition gives a preferential treatment of consumption). However, it is by no meansstraightforward to apply this income definition in practice. A lot of income and expenditurecomponents have to be treated separately. How do you treat governmental transfers andgrants, consumption of public goods or what kind of expenses constitutes a cost of earnedincome and so forth? There are also difficult issues related to the appropriate taxable periodto be used. How do you treat unrealised gains or other tax credits or the missing opportunityto get negative deferred taxes? Which unit of taxation do you choose, families, couples orindividual taxpayers, and should you use equivalent scales to adjust for family size? How doyou measure imputed income when there is no “physical” profit or payout? The SHSdefinition of income is too rigid and demanding to be applied comprehensively in a nationalincome tax and a range of pragmatic choices has to be made.Choosing the right rate structure can also lead to practical challenges. The lack of a clearlydefined set of tax laws implies a vast amount of subjective judgement in the analyses, whichincrease the level of arbitrariness. This is especially true when it comes to choosing the taxrates that applies to deductible expenses and losses.3.2Reporting and evaluation of tax expenditures

Although tax expenditures, like direct expenditures, affect the government’s budget byreducing the resources available, tax expenditures typically are not included in the budget(especially after the year of enactment), are normally not limited to a maximum cost set, andoften continue permanently without regular evaluation or reauthorization. Furthermore, thelevel of tax expenditures can increase when new taxes are introduced or existing taxes arealtered. Direct expenditures do not necessarily have maximum cost set, but the level ofexpenditures appear more transparent in the budget. While direct expenditure programs aremore or less routinely reviewed and funded through the normal course of the annual statebudget process, such a process is often lacking for tax expenditures. As a result, taxexpenditures can become expensive subsidies that lead to considerable amounts of forgonerevenues without legislative action or even awareness.Apart from the fiscal deficit dimension, tax expenditures can also increase tax systemscomplexity and distort both its neutrality and the allocation of resources. Furthermore, taxexpenditures themselves are often non-transparent (e.g. concerning financial volume, targetattainment, and beneficiaries). For all these reasons, it is important that tax expenditures – aswell as the tax system itself and spending programs – receive (periodic) review to ensure thatthey are effective and are justified continued support from the public.In addition, a more comprehensive assessment of government activity leads to a betterunderstanding of the effects of providing tax expenditures, and helps to make tax policy moretransparent. Tax expenditure reporting is thus a contribution to the design of the whole tax11

As table 3.1 shows, one method of classifying tax expenditures is to categorise themaccording to their taxable base. For example, deviations from the benchmark tax on labourare defined as one category, deviations from the benchmark tax on capital as anothercategory, and so on. This simple classification system gives an overview of the taxexpenditures within each tax category and to some extent it gives an idea of who the directrecipients of the tax expenditures are, but gives limited additional information. In general,this method is used to demonstrate how different kinds of tax revenues are affected by taxexpenditures. It is particularly important in the context of tax revenue forecasts.Classifying tax expenditures according to their purpose and/or objective (that is budgetfunction, recipients, and related type of direct spending) is another widely used method. TheGerman Government, for instance, differentiates between tax expenditures which are directlyaimed at companies (business sector level) and others which are aimed at private households(but nevertheless indirectly to companies). Besides, for further categorisation the objectivetargets preservation, economic adjustment, and productivity are used.6However, the Joint Committee on Taxation in USA classifies according to tax transfers (e.g.refundable portion of the earned income credit and child tax credit), social spending (e.g.IRAs, fringe benefits, mortgage interest deduction) and business synthetic spending (e.g.energy subsidies and R&E credit).7Another approach to classify according to technicalreasons can be found by the OECD’s classification: “technical” tax expenditures, otherincome exemptions, tax expenditures which could not be repealed without being replaced bya spending program, and other tax expenditures.8Classifying tax expenditures in this manner gives additional information. For example, it canbe shown, which tax expenditures belong in a special context (e.g. make work pay,6 See German Government (2007), p. 8 f. and 112.7 See Joint Committee (2008), p. 17.8 See OECD (2008/2), p. 6.

retirement, employees benefits or education9) or who are the recipients of tax expenditures.However, one should bear in mind that the different kinds of classification not necessarilygive the full picture of incidence, i.e. the persons who benefit from the tax expenditure in theend. Furthermore, the fact that it is often unclear to which category a tax expenditure belongsand why the categories are chosen as they are, represents an important disadvantage of thiscategorising method. Besides, the volume of additional information generally increases withthe number of categories, while the clarity of the categorization decreases. All in all,objective judgements and comparisons are often difficult.Furthermore, tax expenditures can be classified according to what type of tax measure theyrepresent. That means, how the tax expenditure is designed from a regulatory point of view:allowances, exemptions, deductions, rate relieves, and tax deferrals.10This relatively directand simple approach which is widely used for research purposes contains only few categoriesand allows objective (comparable) judgements.All in all, the Nordic countries classify, as many other countries, first of all according topurpose and/or objective on the one hand and tax base on the other. For example, Norwayuses tax base and budget function. Denmark divides the tax expenditures into categoriesbetween purposes (e.g. education, health, business development etc.). In Sweden, taxexpenditures are divided into two broad categories: expenditures which would affect thebudget balance if they where abolished and expenditures which do not. In addition to thisdivision, the tax expenditures that affect budget balance are classified according to their taxbase and presented in subgroups. Moreover, the expenditures are also classified with respectto their general purpose. However, Finland is the only Nordic country which classifies inaddition to the budget function also to the type of tax measure.3.2.2Important elements of reporting

Reporting tax expenditures in one form or the other, helps keeping focus on this part of thetax system. When reporting, the following should be considered: what to report; how oftenand where to report.In the ideal world tax expenditure reports should be as comprehensive as possible. Dependingon benchmark, reporting should include tax incentives, tax credits, tax breaks, exemptionsand other measures. Therefore the report should include a complete list of all explicit andimplicit exemptions in all taxes (including personal income, corporate income and salestaxes), regardless of the fiscal impact. When present, tax expenditures in local governmenttaxation should also be included. In practice, there is often used a minimum level for a taxexpenditure to be reported/calculated.The ambitious report include detailed information about all tax expenditures (see for instanceGerman “Datenblätter” – tabular summary of each tax expenditure containing original andactual objectives, legal citation, possible time limitation and/or digressional design, fiscalvolume, and – if available – results from former evaluations).11The following information is valuable when reporting tax expenditures:9 See for example OECD (2008/1), p. 36 with target to compare tax expenditures and spending programs.10 See for example OECD (1996), p. 9, OECD (2003), p. 2, Kran (2004), p. 130, Thöne (2005), p. 55 f., Boss and Rosenschon (2008), p. 5.11 German Government (2007), p. 186 ff.

First of all it is important to describe the applied benchmark used to identify tax expenditures.If the definition is not transparent, it is not clear what kind of information you get from thereporting.As the main focus probably will be on the estimates of the annual revenue loss, in the senseof costs which the expenditures create, it should be clear which method is used whencalculating the tax expenditures (revenue foregone, revenue gain or outlay equivalence).Including cost in recent years and cost estimates for the future will give an indication of anytrends in the single tax expenditure. It might not be necessary to re-calculate all taxexpenditures every year. Instead, different parts/categories of the tax expenditures could bere-calculated e.g. every 2-4 year, or when the (groups of) tax expenditures are evaluated inconnection with other parts of the tax- or spending system.When reporting existing tax expenditures and re-estimating tax expenditures, one has to beaware that behavioural responses have an impact, even though the revenue foregone methodis applied. In the longer term the level of tax expenditures will reflect behavioural responsesdue to the tax expenditure itself. Change in economic fluctuations, market conditions etc. willalso effect the level of the existing tax expenditures. Changes in the general tax rates can alsohave an impact on tax expenditures, even though the specific legislation regarding the taxexpenditure has not changed.Besides listing all tax expenditures one by one, they can be listed according to tax base,purpose, who benefits from the tax expenditures etc. (see 3.2.1). Furthermore the reportingcould include legal citation, reasons for enactment (intended objectives), and year ofenactment for each tax expenditure. If there is a time limitation and/or degression this will beuseful information as well. In addition, information from (former) evaluations could besummarized in the report.Last but not least, tax expenditure reports should be timely and accessible. One target of taxexpenditure reporting is to give policymakers and voters the information which they need toevaluate spending through the tax code. This allows them to weigh tax preferences againstother spending and decide what really deserves funding. Tax expenditure reporting could, forinstance, be incorporated into budget process or in a permanent list on the internet. Wherevertax expenditures are reported, the reports should be up-to-date and easily (online) available.3.2.3Challenges in evaluating tax expenditures: Effectiveness and efficiency

analysis of tax expenditures less useful and therefore discredit and weaken them. In addition,evaluation criteria are needed. Designing principles as generally as possible can be seen asapplications of the substantive goals of equity, efficiency, and ease of administration. Due tocompletely different types of tax expenditures with nearly incomparable targets and effects,evaluating tax expenditures systematically is problematic. Nevertheless comparable resultsare necessary to give fruitful advice to the public and politicians. If methods necessarilydiffer, one possibility to ensure comparable results is to use a common evaluation framework.Apart from these evaluation scheme problems, there are a lot of practical difficulties inevaluating tax expenditures (e.g. data problems, interactions and behavioural effects) whichcan strongly affect the value of evaluations. One major challenge is that significant empiricaldata is required to evaluate the theoretical background of tax expenditures. Whereas data forcollected taxes typically are registered, data for tax expenditures which are taxes not collectedare not always recorded. The introduction of tax expenditures can even directly lead to a lackof information. All in all, widely data collecting and providing of data (by taxpayersbenefiting from tax expenditures) is necessary in order to evaluate the effectiveness of taxexpenditures. For purpose of comparison, tax expenditure data has to have the same standingand be of the same level of quality as spending data in the budget. An increased use ofmandatory reports from businesses and households can establish an improved dataset forevaluating tax expenditures. This extra information must be weighted against the cost, bothmonetary and in terms of integrity, of obtaining this information and possible politicalobjectives of administrative burdens.Further challenges result from interactions between different tax expenditures and/or betweendifferent political subdivisions when tax expenditures exist at both federal and state level. Inthese cases changing one tax expenditure can affect costs and effectiveness of another andthereby make it difficult to isolate the effects of individual tax expenditures. When analyzingthe distributional effect of a certain tax expenditure it is therefore important to analyze the taxsystem as a whole. Analyzing one tax expenditure in isolation might give a false impressionof the real distributional effect of all elements of the tax and expenditure system. This couldbe true in the case where the same group of persons or firms receive a tax deduction in onearea but have to pay an extra tax or fee in another area.In addition, there is often limited information regarding how taxpayers’ behaviour areaffected by the existence of tax expenditures. Therefore, the overall effect often differssignificantly from the first stage effect. Furthermore, looking at tax expenditures next todirect spending is not always an apples-and-apples comparison. Direct spending does e.g. notnecessarily take account of taxes, whereas tax expenditures mostly are stated after taxes.All in all, evaluating costs, efficiency, and equity impact of tax expenditures requires a highlevel of ambition. Comprehensive assessments will potentially require a lot of resources,which are not allocated today. Therefore, depending on the amount of resources allocated forassessments a more targeted approach, which focuses on individual tax expenditures ofspecial interest, could be considered. Evaluations of tax expenditures along with otherevaluations of the tax system in general or when specific areas are evaluated will could alsobe considered. Such an approach should include studies both by the administration itself aswell as by other institutions. A sunset provision is another possibility to ensure that (newly)created tax expenditures do not continue indefinitely unless merited.

In general, international comparisons of tax expenditures are difficult.13Definitions of taxexpenditures share a common core across countries. Every country defines tax expendituresas exceptions from some baseline standard for the entire tax. Although the benchmark isalmost identical, the countries differ in the purpose of the exercise of the identification of taxexpenditures and this cause differences in what is considered as tax expenditures in thespecific country. Japan, for instance, defines its benchmark in terms of its basic principles oftaxation: equity, neutrality and simplicity. This benchmark is expected to give a large numberof tax expenditures. In the other end of the spectre, the Netherlands considers its benchmarkto be the “primary structure” of the actual tax system in place, which is expected to inducerelatively few tax expenditures. Thus, with the choice of definition the number of identifiedtax expenditures can vary considerable. Few countries identify tax sanctions.Some countries have made it a legal requirement to report tax expenditures, and others havenot. Some report outside the budget and some in annexes to the budget. Reporting of taxexpenditures alongside similar outlay programs is seldom done. In general most countriesreport annually, but some countries report every two years or less frequent. Most countriesgenerally cover only a few years when reporting, but some countries like Canada, theNetherlands, and USA covers a period for 7-8 years.Most countries acknowledge the importance of evaluation, but no countries have systematicprograms to evaluate tax expenditure. Even in the few countries where evaluating programsare required by law, the evaluations only comprise of an estimation of the cost of taxexpenditure.143.3.2The German evaluation of tax expenditures

In 2007, the German Federal Ministry of Finance commissioned – based on the FederalGovernment’s report on subsidies – a systematic evaluation of the Federal Government’smost important tax expenditures.15The current research project, undertaken by the CologneCentre of Public Economies (FiFo), the Mannheim-based Centre for European EconomicResearch and Copenhagen Economics, determines if there are rational goals for these taxexpenditures, as well as effectiveness and efficiency in achieving objective targets.The tax expenditures in question reach from VAT reductions to income tax measures andinstruments with environmental objectives. These tax expenditures belong to completelydifferent categories with both incomparable targets and effects, and this makes it impossibleto evaluate them according to one method. Nevertheless a systematic evaluation is ensured byasking the same questions in a common evaluation framework for each of the taxexpenditures. The following structure is used:

Tax expenditures are generally defined as deviations from a benchmark tax system. In all ofthe countries discussed here, the underlying idea behind the definition of the benchmarksystem is the concept of comprehensive income taxation, a broad based system where allincome is largely taxable. For pragmatic reasons, the benchmark system is in all countriesmore or less based on the prevailing tax system. Sweden applies a more ambitious approachin defining the benchmark, as the benchmark is based on the idea of uniform taxation, i.e.each type of tax should be levied uniformly and without exemptions. The benchmarknevertheless allows for exemptions and can thus also be perceived as pragmatic. Besides taxexpenditures, all countries also calculate tax sanctions in cases of unfavourable tax treatmentof specific groups or activities. Denmark, however, only calculates tax sanctions when thereis a close link to the tax expenditures, like when a tax sanction reduces a tax expenditure. Thegeneral features of the benchmark systems are fairly stable, although as they are to a largeextent based on the prevailing tax systems, they do change when the general tax systemschange.17Defining the benchmark system can be complicated in practice, even though the benchmarkis based on the current tax system. Deciding which tax rules should be included in thebenchmark and which constitute tax expenditures is not always straightforward. Examples ofthese questions are the treatment of interest deductibility, earned income tax deductions,deductibility of pension income contributions, VAT exemptions and excise duties.All the Nordic countries use the dual income tax system where labour income and capitalincome are taxed separately, although with great individual variations. This system isgenerally included in the benchmark. Progressive taxation, including standard deductions, ispart of the benchmark in all countries. In Denmark and Sweden in-work tax deductions arealso considered part of the benchmark. Denmark includes deductions for many, but not allexpenses in the benchmark, whereas the other countries only allow for deductions of costs17 An example of these changes is the adoption of the shareholder model in taxation and in the benchmark in Norway in 2006.

directly linked to generation of income. Generally the deductions that are allowed for in thebenchmark are of general nature and do not favour specific groups or activities.All countries except Finland include interest rate deductions in their benchmarks. Imputedrent from owner occupied housing is considered as real income in all countries but the taxexpenditure is not calculated in Denmark due to difficulties in defining the benchmark. InNorway a wealth tax is included in the benchmark. Deviations from the standard rate orlower assessment values than real values are regarded as tax expenditures.All countries have a standard rate for corporate taxation and deviations are defined as taxexpenditures. In Norway real profit is taxed as ordinary income, with the same rate as otherincome sources.In value added taxation, the standard VAT rate (25 per cent in Denmark, Norway, andSweden, while 22 per cent in Finland) defines the benchmark in all of the countries, anddeviations from the standard rate create tax expenditures. In Finland all, and in Sweden some,VAT exemptions are also considered part of the benchmark, in Denmark and Norwayexemptions are considered to create tax expenditures.The countries define the benchmark with respect to excise duties in different ways and thescope of excise duties included varies as well. Finland has no special benchmark for exciseduties, but defines tax expenditures as deviations from the standard rate. The same applies forDenmark and Sweden, but included in the benchmark for energy is a split between where andto what the energy is being used. Denmark considers a new benchmark for environmentalexcise duties that reflect externalities. Norway on the other hand applies a benchmark that isbased on the theory of optimal taxation and thus a normative benchmark. The benchmark isdivided between fiscal and environmental excise duties. For fiscal excise duties theexemption of taxes on factors on production is part of benchmark and the environmentaltaxes are in line with external costs. Sweden and Finland only considers excise duties thatinvolve tax expenditures with a certain magnitude, whereas Norway and Denmark includeall, or most of the excise duties.

Comparing tax expenditures between countries and over time is not straightforward. Thecoverage of the tax expenditure calculations varies, as does the definition of tax expendituresin the first place. A more orthodox definition of the benchmark usually leads to a highernumber of tax expenditures. It should also be kept in mind that tax systems, and thus thebenchmarks, within each country change, and this will affect the evolvement of taxexpenditures over time.Moreover, the data used may be of varying quality. This, among other things, calls forcaution when interpreting tax expenditures. Nevertheless, the number of tax expendituresranges from 60 in Norway to 115 in Sweden. The total estimated volume of tax expendituresis by far the largest in Sweden, 25 billion euros, and smallest in Denmark, 5 billion euros. InFinland the estimated value of tax expenditures was 13 billion euros, and in Norway 16billion euros. The share of tax expenditures as per cent of GDP is largest in Sweden, 8 percent, and smallest in Denmark, 2.2 per cent. In Finland the share is slightly lower than inSweden, 7 per cent, and in Norway the share of tax expenditures is 5.4 per cent of GDP. As21

per cent of total tax revenue, the tax expenditures amount to 4.4 in Denmark, 17 in Sweden,16 in Finland and 13 in Norway.18As method of calculation, all countries apply the revenue-foregone method. Tax expendituresare calculated as the direct revenue effect of abolishing a specific tax rule that is not part ofthe benchmark system. Denmark and Sweden also use the outlay equivalent method. Thismethod calculates the corresponding amount needed on the expenditure side of thegovernment budget to reach the same effect. Norway calculates some tax expenditures aspresent value. This method is used when there is a time horizon, by estimating today’s valueof a future gain. The tax expenditures related to depreciation rates higher than actualdepreciation and tax expenditures related to employee premiums and contributions tooccupational pension schemes are calculated using this method. A common feature amongthe countries is that behavioural responses are not taken into account when calculating taxexpenditures. Moreover, interactions between different taxes are not taken into account. Thetax expenditure calculations rely on tax administration data and other statistical data. Inaddition, Finland and Norway utilise microsimulation models in calculating tax expendituresin personal income taxation. In Denmark and Sweden the Ministry of Taxation and Ministryof Finance, respectively, performs the calculations, and in Norway the work is dividedbetween the Ministry of Finance and Statistics Norway. In Finland the Government Institutefor Economic Research makes the calculations.Since the late 1990’s, the share of tax expenditures to GDP has declined in Denmark. This ismainly due to abolition of tax expenditures related to taxation on energyconsumption. Therehave also been some changes in the composition of the tax expenditures. The Danish Taxreform (Forårspakke 2.0) will reduce the tax expenditures with app. 0.2 billion euros. InFinland the share of tax expenditures to GDP has remained relatively stable during thisperiod, but it has come down remarkably since the mid-eighties, which is the start of thereporting period. Lately there has been an increasing tendency towards introducing new taxexpenditures in Finland, so generally the volume of tax expenditures can be expected toincrease rather than decrease in the years to come. Since the late nineties in Norway, anincreasing number of deductions and allowances have been defined as tax expenditures. Thishas contributed to an increasing volume of tax expenditures. One of the objectives of the2006 tax reform in Norway was to abolish several exemptions and allowances that werepoorly justified, but the reform was only partly successful in this matter. The general trend inNorway is that the reported tax expenditures are growing. A substantial part of this growthcan be attributed to growing tax bases, but some tax policy changes have also contributed tothis development. In Sweden the level of tax expenditures has been fairly constant over thelast few years. Although there has been some year-to-year variation it is hard to point in anydirection.In all of the countries, the composition of tax expenditures evolves over time when new taxexpenditures are introduced and old ones abolished. The distribution of tax expenditures todifferent policy areas is different in the Nordic countries, but currently tax expenditure onhousing seems to be among the most important tax expenditures in all of these countries. Themost important tax expenditures in Denmark are in the area of business development,corresponding to 30 per cent of total tax expenditures. Besides from business development,traffic and communication, housing and energy supply represent areas with a relatively highlevel of tax expenditures. Tax expenditures directed to business development and housing has18 The figures in this paragraph refer to 2008 or 2009.

increased substantially since the late 1990’s in Denmark, and at the same time taxexpenditures for energy supply has decreased. In Finland, the most important taxexpenditures are in the area of social security, housing and labour income taxation. The shareof tax expenditures directed to manufacturing has decreased since the late 1980’s, and sincelate nineties, the share of tax expenditures on labour income has increased. In Norway, themost important tax expenditures relate to lack of taxation of imputed income from, and lowassessment values on, housing. The regionally differentiated employers’ social securitycontribution and employee premiums and contributions to occupational pension schemes alsogive rise to major tax expenditures. In Sweden, the largest tax expenditure in 2009 isassociated with returns to housing as it is not taxed as other capital income. Three other largetax expenditures involve VAT, indirect taxation of labour income and energy taxes.4.2.3Reporting, classification and evaluation of tax expenditures

Tax expenditures reflect political priorities, and by reporting tax expenditures, these prioritiesare made visible. Tax expenditures have been reported since late 1990’s in all of the countriesexcept for Finland, where they have been reported since 1988. All countries except Denmarkreport tax expenditures yearly to the Parliament. In Denmark tax expenditures were reportedin an appendix to the Budget Proposal until 2006, when the Danish Government decided toexclude them. The Danish Ministry of Taxation publishes a list of changes to taxexpenditures due to legislation on its homepage. In Sweden tax expenditures are reported in aGovernment Communication in conjunction with the Spring Fiscal Policy Bill, and they arere-reported in the autumn as supplement to the Budget Bill. In addition to the actual numbers,Sweden also publishes the methodology of the calculations. In Norway tax expenditures arereported in the National Budget, and in Finland the main categories of tax expenditures arepresented in Report on the Central Government Final Accounts. Although there is nosystematic tax expenditure reporting in Denmark, new tax expenditures are explicitlymentioned when a bill includes tax expenditure. Revenue and distributional effects and thepurpose of the new tax expenditure are then presented. In Sweden, when a new taxexpenditure measure is introduced, this is explicitly pointed out in the tax expenditure report.In Finland, the status of the tax expenditure reporting is relatively low at the moment, and nospecial attention is given to new tax expenditures in tax expenditure reporting. Norway hasno systematic approach to new tax expenditures in the tax expenditure report, but in the latestreport (2008) new tax expenditures were pointed out.There are differences between the countries not only in how the tax expenditures arereported, but also in what is left out of the tax expenditure reports or not calculated. The mainreason for leaving something out is methodological difficulties. In Finland, real estatetaxation, inheritance tax or social security contributions are not covered in the taxexpenditure reports. Also most of the excise duties are not covered, as is also the case inSweden. In Norway all existing tax expenditures are included in the report, but not all arecalculated. This for example applies to the inheritance tax and several tax expendituresrelated to payments in kind. In Denmark tax expenditures are not calculated where abenchmark is difficult to establish, as for example in private pensions systems, where thetime horizon and correlation to public transfers complicates the matter.There are different ways of classifying tax expenditures (see chapter 3.2), and the informationgiven by the tax expenditure reports also depends on the classification. In Denmark andFinland, tax expenditures are classified according to operational categories. In Finland thisclassification was originally based on budgetary categories, but in that sense the classificationis no longer valid, since the budget categories have changed. Tax expenditures are also23

classified according to types of tax (income tax, indirect taxation etc.) in Finland. In Norwaytax expenditures in direct taxation are classified according to taxable base. In indirecttaxation, tax expenditures are classified according to both the type of tax (VAT, excises) andobjectives (fiscal, environmental, health). In Sweden, tax expenditures are divided into twobroad categories: expenditures which would affect the budget balance if they where abolishedand expenditures which do not. For the former category, which includes most taxexpenditures, both the revenue forgone and the outlay equivalent are calculated. For taxexpenditures not affecting the budget balance, i.e. tax exempted transfers, only the outlayequivalent is calculated. The revenue foregone is readily available on the expenditure side ofthe government budget. In addition to this division the budget balance affecting expendituresare classified according to their tax base and presented in subgroups. Moreover, theexpenditures are also classified with respect to their general purpose. In this respect, adistinction is made between technically or administratively motivated tax expenditures andpolitically motivated tax expenditures.Despite of the reporting, tax expenditures are not an integrated part of the budget process inany of the countries. In most cases tax expenditures are not reported in connection with directexpenditure targeted to the same activities or recipient groups, so it seems that whendiscussing the distribution of public finance to different policy areas, tax expenditures are notsystematically included. One exception is found in Norway, where direct and tax transfers todifferent industries are reported under the heading “Industrial support” in the NationalBudget.Some have pointed to the lack of yearly evaluation and assessment of whether the publicsupport given trough tax expenditures is increasing or decreasing, and how well the targets ofthe tax expenditures are achieved.There is no systematic evaluation of the effectiveness of taxexpenditures in any of the countries discussed here. But there are good examples ofevaluation though. For example in Norway, tax allowance for R&D expenses has beenthoroughly evaluated by Statistic Norway. Prior to the Norwegian tax reform in 2006, agovernment appointed Tax Committee also evaluated several tax expenditures. In Denmarktaxation and other regulation of emissions of CO2have been evaluated in 2007. The doubleregulation of CO2can be regarded as a form for tax sanction. In Finland, the tax credit forhousehold services introduced in 2001, has been evaluated for employment effects. Morecomprehensive evaluations of tax expenditures are often done in connection with tax reforms.This has been the case in Finland when a comprehensive tax reform was introduced in thelate 1980’s. A distinction can be seen between the treatment of new tax expenditures andexisting ones – new tax expenditures tend to be better evaluated, (in terms of for example ofemployment effects), than existing ones. There is often an ex ante evaluation of the effects ofnew tax expenditures when they are introduced. It should be noted that not all directexpenditures are evaluated on a yearly basis.4.2.4Critique from National Audit Offices

It is not easy to draw conclusions about the general atmosphere and discussion concerningtax expenditures in the Nordic countries. There clearly are different tendencies in the primaryobjectives behind tax expenditure reporting, the political interest in using tax expendituresand in the political and general discussion about the reporting of tax expenditures, and thetreatment of them in the budget process.Tax expenditures can be justified for various reasons, but there can also be less satisfactoryreasons behind the use of tax expenditures instead of direct expenditures. For the latter, theprinciple policy instrument, there are routines and systems designed for efficiency, controland fiscal discipline. In general such systems are missing for tax expenditures. For example,while a cap can be applied to direct expenditures, there is no direct limitation on taxexpenditures. Finland and Sweden apply budget caps to direct expenditures and it may beargued that such caps give incentives to give support through the budget income side, i.e.through tax expenditures. Denmark does not have a budget cap, but there is a long runbalanced budget goal. Due to the tax freeze since 2001, tax expenditures cannot be reduced.New tax expenditures can be introduced, which will reduce the room for direct expenditures.This is also the case in Norway, where the government operates with a balanced budget rule,i.e. a structural non-oil central government budget deficit corresponding to the expected realreturn (estimated at 4 per cent) on the Government Pension Fund – Global (the formerPetroleum Fund). Even though this implies that the government has a choice between taxexpenditures and direct transfers, it often seems like tax expenditures are regarded as easier toimplement than direct expenditures and also appears to be less costly. This probably stemsfrom a fact that applies to all the countries in question, namely that tax expenditures are notsubject the same scrutiny as direct expenditures. Although in Denmark changes in taxexpenditures are subject to the same scrutiny as other changes in the tax code. A fullintegration of tax expenditures into the budget process is hardly feasible. Lack of data,computational methods, benchmark choice and other methodological issues complicates thismatter.The Danish Government has decided to end tax expenditure reporting in the Budget Proposal.This does not mean that the tax expenditures are totally neglected though. The exclusion oftax expenditures is justified by that the fundamental objective with the Budget Proposal is toestablish the basis for next year’s direct expenditures and revenue. Distortions due to the taxsystem and direct regulation are not listed in the Budget Proposal, even though these costsexceed the costs due to tax expenditures. Even though collecting more information about taxexpenditures would improve transparency, it would also involve increasing costs andadministrative burden. The Danish Government has committed to decreasing administrativeburden, and so these costs should be carefully weighted against the gains from moreinformation. Yet the Danish Government agrees that the use of tax expenditures should notlead to open ended public spending.

In Finland the interest in using tax expenditures has been increasing recently, after beingrelatively limited for almost 20 years. There has been an increasing amount of requests fornew tax expenditures, and as a consequence, new tax expenditures have been introduced. Itcan be argued that this is due to budgetary spending limits on direct expenditures. At thesame time, the distance between tax expenditure reporting and the budget process has grown.The rationale behind each tax expenditure is discussed in a working group report published in1988, but they are not presented along the yearly reporting of tax expenditures. Following thecritique from the National Audit Office and the Parliamentary Audit Committee, the FinnishMinistry of Finance has appointed a working group with a mandate to evaluate, and suggestimprovements to, the treatment of tax expenditures.In Norway the stated purpose of reporting the tax expenditures is to obtain a greater degree oftransparency regarding political priorities and financial support to different groups oractivities. However, the pragmatic approach to the reference system used to identify taxexpenditures, limits the informational value of reporting tax expenditures. On the other hand,this approach implies that the issue of tax expenditures attracts only moderate attention inNorway, and the estimates are more or less undisputed.Despite of the differences, in can still be argued that the importance of discussing taxexpenditures in the Nordic countries is increasing rather than decreasing. In all of thecountries, the problems in defining the benchmark tax systems, and also in documentation ofthe calculation methodologies applied, limit the informational value of the tax expenditurereports. The lack of evaluation of tax expenditures also makes the matter more important.4.3Tax expenditures in DenmarkIntroduction

The first comprehensive survey of tax expenditures in the Danish tax system dates back to1996. After 1997, tax expenditures were included in an appendix to the Budget Proposal(Finansloven). The tax expenditures were not reported in connection with direct transfers orsubsidies. The tax expenditures were published in the Budget Proposal until 2006 andcovered the period up to 2009. The tax expenditures have not been updated with revisions inthe benchmark system the last couple of years, but were more or less a mere mechanicalprojection with the annual change in the GDP or consumption growth when relevant andtaking into account changes in the legislation with respect to tax expenditures. In 2006 theGovernment decided to stop publishing tax expenditures in the Budget Proposal. Instead,when a Bill implies tax expenditures, this must be accounted for explicitly. The revenueeffects, distribution, purpose etc. will thereby be transparent as is the case with other changesin the tax system.A list of changes in tax expenditures due to legislation is published on the Ministry ofTaxation homepage.19This reporting started in the financial year 2007/2008. The list will notupdate all tax expenditures every year, but will include new tax expenditures and changes inexisting tax expenditures.The benchmark for tax expenditures in Denmark is, in general, based on a pragmaticapproach. The tax expenditure is calculated as direct revenue when abolishing a special rule19 See www.skm.dk.

All types of taxes are included when calculating tax expenditures; also differences in thetiming of tax payment are covered. Tax expenditures are not calculated where a benchmark isdifficult to establish as for example owner-occupied accommodation or in private pensionssystems, where the time horizon and correlation to public transfers complicates the matter.The benchmark for tax expenditures is, in general, based on a pragmatic approach, where thetax expenditure is calculated as direct revenue when abolishing a special rule that is notconsidered part of the benchmark system.The applied benchmark tax base can be characterised as a Comprehensive Tax Base,including capital income and gains, labour income, fringe benefits and public transfers. Thebenchmark has some modifications. For instance, the dual income tax system, which hasbeen in place since 1987 is considered part of the benchmark. In addition, the progressive taxrates are part of the benchmark system for income taxation.Furthermore, deductions for expenses to acquire and maintain income are part of thebenchmark system. But differentiation between private expenses and expenses to acquire andmaintain income are not always simple in the real world. As a practical solution all expenses(some specified by law) are regarded as part of the benchmark system, while only deductionstargeted to specific groups are calculated as tax expenditures. For example, deductions ofinterest payments are part of the benchmark. If not, the level of tax expenditures wouldincrease substantially. Moreover, the in-work tax credit is regarded as part of the benchmark,even though it only concerns income from employment and not income from transfers. Thecredit has a max and is given to all employed. For higher incomes it doesn’t reduce themarginal tax rate, but only reduce the average tax rate.. All in all, this benchmark is inpractice relatively close to the overall structure of the Danish tax system.For VAT purposes the standard VAT rate (25 per cent) is the benchmark. Therefore lowerVAT rates and exemptions are considered tax expenditures. This includes exemptions thatfollow the EU VAT directive. For excise duties however there is no general benchmark.Instead, tax expenditures are defined as deviations from the benchmark for each excise duty.The Danish corporate tax rate amounts to 25 per cent, which is the general benchmark ratefor corporate taxation. There are some difficulties in defining which exceptions to the generalsystem have to be treated as tax expenditures. In the list below there are some considerationsregarding selected topics in the corporate taxation area:R&D tax incentives are treated as tax expenditures in Denmark. The legislation puts updifferent rules of preferential treatment regarding R&D expenditures, which makes it possibleto deduct expenditures to R&D right away and not following the normal rules ofdepreciation.Imputation systems for taxation of dividends are a part of the benchmark system in Denmarkand are therefore not treated as tax expenditures. Dividend incomes are taxed at preferentialrates (compared to interest income) at the shareholder level. The rate is set so that the overall

CIT + dividends equal the nominal tax on wages. Therefore the real tax rate on capitalincome might exceed the wage income tax. This could lead to a discussion about what thebenchmark system should be if the tax expenditure had to be calculated.Reduced tax rates for certain forms of capital income. Capital incomes are taxed differentlywhether it is positive or negative capital income. Negative capital income is taxed with a flatrate whereas positive capital income is taxed with varying rates depending on the type ofcapital income. The different tax rates on different types of capital incomes are not regardedas tax expenditures in the Danish system. Neither is taxation on an accrual basis (as opposedto taxation upon realisation) calculated or regarded as a tax expenditure.Tax expenditures are not defined by an independent organization, so the government hassome leverage when deciding whether changes in the tax system give birth to change or newtax expenditures. There has not been much debate over what to define as tax expenditures.The National Audit Office of Denmark has recommended a broader definition of taxexpenditures in line with recommendations from OECD. As the tax expenditures are definedas deviations from a norm this benchmark will change if the tax system undergoes reforms orchanges.Finally, tax sanctions are only calculated if there is a close link to tax expenditures. That istypically a tax sanction reducing a tax expenditure. Moreover, tax expenditures includegeneral and local taxation but tax expenditures are not split into different levels for purposesof reporting. Almost all tax expenditures are permanent.4.3.3Methods of calculation

In Denmark both therevenue foregone methodand theoutlay equivalent methodhave beenused, even though the revenue foregone method has been most widely used. Therevenuegain methodhas not been applied, although the estimated behavioural responses are includedin proposals for changes within the tax system. Even though behavioural responses are notdirectly taken into account when calculating tax expenditures, the level of tax expenditures inthe longer term, when recalculated, will reflect behavioural responses due to the taxexpenditure, market conditions etc. But this behavioural response is of course the reason whythe tax expenditure often will be different to the revenue if the tax expenditure were to beabolished.The individual tax expenditure is estimated under the condition that all other tax expendituresare unaltered. The possible interaction of different tax expenditures is thus neglected, as thetheory does not point to correct order of adding up.The Ministry of Taxation that manages the calculations does not use any special models toderive the tax expenditures. Tax expenditures are not calculated where a benchmark isdifficult to establish as for example in private pensions systems, where the time horizon andcorrelation to public transfers complicates the matter.4.3.4Reporting and evaluation

period from 1993 to 1998. The 1996 report was followed by an update in 1997.21This updateincluded new tax expenditures, and minor revisions in the reference system and changes dueto revised data sources. The 1997 report covered the period 1997 to 2001.After the 1997-update, tax expenditures was included in an appendix to the Budget Proposal(Finansloven). The tax expenditures were not reported in connection with direct transfers orsubsidies. The tax expenditures were published in the Budget Proposal until 2006 andcovered the period up to 2009. The tax expenditures have not been updated with revisions inthe reference system the last couple of years, but were more or less a mere mechanicalprojection with the annual change in the GDP or consumption growth when relevant.The Government decided not to publish tax expenditures in the Budget Proposal after 2006.Although tax expenditures are no longer reported in the Budget Proposal, this does not implythat tax expenditures are neglected. There is a greater focus on tax expenditures in thelegislation process. When a Bill implies tax expenditures, this must be accounted forexplicitly. Thereby the revenue effects, distribution, purpose etc. will be transparent as it isthe case with other changes in the tax system.There is a list of changes in tax expenditures due to legislation on the Ministry of Taxationhomepage.22This reporting started in the financial year 2007/2008. The list will not beupdated completely every year, but will include new tax expenditures and also revisions ofexisting tax expenditures. In addition, parts of the tax expenditures are reported separately indifferent contexts.In 2007 the National Audit Office of Denmark (NAO) evaluated the use of tax expendituresin Denmark.23The NAO main critique was a lack of transparency in the use of taxexpenditures. The NAO recommended proceeding with a yearly publishing of taxexpenditures in connection with the Budget Proposal. Every single tax expenditure should berevised and evaluated every year.The Danish Government agrees that the use of tax expenditures should not lead to openended public spending. The fundamental objective with the Budget Proposal though is toestablish the basis for next year’s direct expenditures and revenue. Distortions due to the taxsystem and direct regulation are not listed in the Budget Proposal as well, even though thesecosts exceed the costs due to tax expenditures.The NAO also noted that an increased collection of data would improve the quality of taxexpenditure estimates in certain areas. An increase in the collection of data will involve extracosts. Whereas data for collected taxes are registered, data for tax expenditures which aretaxes not collected, are not always recorded. An increased use of mandatory reports from21 ”Skatteudgifter i Danmark, 1997”, by the Danish central administration.http://www.skm.dk/tal_statistik/provenuoversigter/6731.html.23 Beretning til Statsrevisorerne om gennemsigtighed vedrørende skatteudgifter (fradrag mv.), 2007:22

businesses and households can establish an improved dataset for evaluating tax expenditures.This extra information must be weighted against the cost to get this information. TheGovernment has committed itself to reducing the administrative burdens on businesses withup to 25 per cent by 2010. An extended duty to report data will impose administrativeburdens on businesses and households, which conflicts government objective.Transparency of tax expenditures, and evaluation of the tax system and expenditures ingeneral, can be established outside of the Budget Proposal. This might not imply anevaluation of all aspects of the tax system etc. every single year, but directs resources tospecific parts of the system. In time the majority of direct as well of indirect spending will beevaluated.The tax burden was technically increased between 1993 and 1994 as a consequence ofrestructuring a number of social pensions from being fully or partially tax free to fullytaxable. Contributions before tax rose accordingly. This technical reform raises the taxburden and lowers tax expenditures (if the reference system claims that all income must betaxed) but has no real economic effects.4.3.5Overview of the most important tax expenditures

In this subsection there is a description of size, distribution and trends of tax expenditures inDenmark. It should be noted, that there are considerable uncertainties when estimating andcalculating tax expenditures. The estimates are, as described above, sensitive to changes inthe choice of benchmark and different benchmarks would generate different levels of taxexpenditures. There are also tax expenditures in a number of areas where the statistical andanalytical basis for a survey are not sufficient. Therefore the estimates of the total taxexpenditures are not complete and must be taken as such.Total tax expendituresThe number of tax expenditures in Denmark is approximately 90. Total tax expenditures canwith considerable uncertainty be estimated to around 37.8 billion DKK or 5 billion euros24in2009 using the revenue forgone method. This corresponds to around 2.2 per cent of GDP, andaround 4.4 per cent of total tax revenue.As figure 4.1 shows, tax expenditures have fluctuated between 26 and 38 billion DKK incurrent prices in the period from 1996 to 2009. From 1996 to 2000 there was a relatively highincrease in tax expenditures, which was mainly due to new legislation regarding health andbusiness development expenditures. The increase in tax expenditures from 2003 and forwardwas mainly due to economic growth and inflation.

tax stated as a nominal amount, the amount in DKK cannot be increased. Reductions in taxexpenditures are not within the restrictions of the tax freeze. The tax freeze is not establisheddirectly by law, but it commits the Government to abstain from raising any tax.However, the tax freeze does not completely exclude a necessary restructuring of the taxsystem. If there are compelling reasons to introduce a new tax or raise an existing one, therevenue resulting from this rise in taxation will have to be used fully to reduce another tax.The same principle will be applied if it becomes desirable, for environmental reasons, tointroduce a new environmental tax or raise an existing one.If Denmark is forced to lower a tax as a result of EU decisions or international agreements,there may be compensated for this reduction by increasing other direct or indirect taxes. Suchchanges will be required to leave the net revenue from taxes unchanged.The Government set up a tax commission in 2008 to draw up models for a total tax reform.The resulting tax reform has led to changes in some tax expenditures, especially basebroadening in the VAT system and general reductions of the marginal tax rates. The taxfreeze will be maintained after the tax reform.New Benchmark system due to quota system1 January 2005 marked the start of the EU greenhouse gas emissions trading scheme, whichintroduced a quota system for CO2-emissions across the 25 Member States of the EuropeanUnion. The quota system gives rise to reconsider the benchmark system with respect to theCO2tax.The way the ETS functions there will be no global reduction in CO2emissions if you on topof the price of emission rights is burdening emissions directly with a CO2tax on fuels orindirectly with a tax on products produced inside the ETS sector (e.g. electricity). However ifyou on top of the price of emissions rights is burdening emissions with different CO2taxes indifferent countries the consequence is, that the cost of reducing CO2emissions inside the ETSsector will be higher than necessary.A binding quota system will set the level of CO2emissions from within the emission scheme.Therefore the Danish policy is to abolish CO2taxes on emissions, which already on themargin is burdened with the price of emissions right. It has already been proposed to removethe CO2tax on fuels used for industrial process inside the ETS sector.The quota system gives an opportunity to change the benchmark for CO2taxes as well as fortaxes on other externalities. This implies that the benchmark is moved towards an optimaltaxation scheme. For practical matters the quota price is assumed to be equal to the correctestimate of the environmental cost of CO2emissions.Externalities and tax expenditures when no tax or regulations are applied?A new tax on emissions of nitrogen oxides (NOx) will be implemented in 2010. The harmfuleffects of NOxemissions have been acknowledged for decades. This gives rise to whetheridentified externalities should be regarded as tax expenditures when no tax is implemented atall. If “Pigou taxing all externalities” were the benchmark, then - theoretically - this wouldimply tax expenditures even when no taxes are in place. This approach has not been appliedin Denmark, though.

Tax expenditures have been reported in Finland since 1988, when the first report covered theperiod from 1984 to 1987. In recent years, the connection between the reporting of taxexpenditures and the budget process has been loosened. At the same time, the interest inusing tax expenditures for distribution of public financial support has increased. This hasbeen seen problematic, and thus the Ministry of Finance has appointed a working group witha mandate to evaluate and improve the treatment of tax expenditure. The term of the workinggroup is 1.5.2008-31.12.2009.The share of tax expenditures as per cent of GDP has decreased from 14 per cent in 1985 to 7per cent in 2007. The most significant tax expenditures are in the area of social security,housing and labour income taxation.Tax expenditures are defined as deviations from the benchmark tax system, which is largelybased on the prevailing tax system. The underlying principle is the concept of comprehensiveincome, where all income is taxable and the tax base is thus as wide as possible. Thebenchmark tax system is defined in personal income taxation, taxation of corporate incomeand other enterprise income, value added taxation and excise duties. In calculating taxexpenditure, the revenue foregone method is used.4.4.2Definition of tax expenditures

The definition of tax expenditures in Finland is based on a working group report published bythe Ministry of Finance in 1988. Tax expenditure refers to indirect support, which isincorporated in the tax system, i.e. when a tax provision departs from the basic structure oftaxation for support purposes. On the other hand, if a tax provision leads to a form of taxationmore strict than normal, this is considered a tax sanction. Tax expenditures are defined asdeviations from the benchmark tax system. The benchmark system is in principle based oncomprehensive income, so that all types of income should be taxed and the tax base should beas broad as possible. For practical reasons the benchmark system is nevertheless quite nearthe prevailing tax system. When the prevailing tax system undergoes changes, the benchmarksystem also changes.The benchmark tax system is defined in personal income taxation, taxation of corporateincome and other enterprise income and value added taxation. For excise duties there is nogeneral benchmark. Instead it is defined individually for each excise duty, for which taxexpenditures are calculated (currently there are only tax expenditures in motor car taxationand in no other excise duties). There is no benchmark tax system for real estate tax,inheritance tax or social security contributions. Tax expenditure calculations include generalgovernment taxation, municipal taxation, church taxation and contributions paid byindividuals to the Social Insurance Institution25.In personal income taxation, income is in principle largely taxable and tax liability is thus ascomprehensive as possible. The deductions are only related to actual costs. The benchmark25 Although there is no reference system for social security contributions, the contributions paid to the Social Insurance Institution can beincluded in the tax expenditure calculations, since they are included in the income taxation.

tax system also includes imputed income and income transfers. All exemptions from theincome tax are considered tax expenditures. Since 1993 Finland applies dual income taxationof capital and earned income. Dual income taxation is considered part of the benchmarksystem, including the splitting of income of the self-employed, partnerships and agriculturalincome. Progressive taxation of earned income, including a standard deduction for lowincome earners, is also considered part of the benchmark. The tax rate schedules areincorporated in the structure of the benchmark in such a form as they have been laid down atany one time.In income taxation, the prevailing tax rates define the benchmark. In capital income taxation,the prevailing tax rate is 28 per cent. The deduction of interest in personal income taxation isnot considered part of the benchmark system and thus creates tax expenditure. Capital gainstaxation in realisation without taking inflation into account is considered part of thebenchmark. Neither the capital gains taxation nor the treatment of losses create taxexpenditures or tax sanctions.The definition of tax expenditures in the taxation of business profits and income fromprofessional activities is based on the principle that income is largely subject to tax.Furthermore, the allocation of income and expenses is closely linked to the accountingprinciples. Losses can be carried forward 10 years and this is taken as part of the benchmark.Regarding to the depreciation of buildings and machinery over their technical economic life,currently the tax expenditure is calculated based on the depreciation of machinery. InFinland, the norm depreciation is 15 per cent whereas the prevailing provisions allow for 25per cent depreciation.In corporate income taxation the benchmark is the prevailing tax rate, 26 per cent. The taxexemption on public companies is considered part of the benchmark tax system. The situationis the same when it is question of communities that complete the operations of publicorganisations (e.g. national unemployment fund). The prevailing system of partial doubletaxation of dividends is considered part of the benchmark system. (That is 70 per cent of thedividends from listed companies are taxed as capital income. In unlisted companies the yieldof 9 per cent to the mathematical value of the share is tax exempt up to 90 000 euro, 70 percent of the rest being taxed as capital income. From the 9 per cent exceeding yield, 70 percent is taxed as earned income according to the progressive taxation.)In value added taxation the benchmark tax system is a broadly based, multi-stage (with inputVAT deduction) value-added tax, collected according to the destination principle (that is withexports exempted and taxation of imports). The standard VAT (22 per cent) rate is part of thebenchmark, and deviations from the standard VAT tax rate create tax expenditures.Exemption from VAT is considered part of the benchmark system (e.g. financing, education,public or publicly supervised social and health care). Tax exemption for enterprises withturnover below 8 500 euro, and without voluntary registration, as liable for VAT is notconsidered as tax expenditure. The graduated tax relief for enterprises with turnover between8 500-22 500 euros creates tax expenditure.Tax expenditures from excise duties arise only in departures from the standard level of exciseduties (only in motor car taxation; e.g. support directed to disabled people or taxi drivers).There is no special benchmark taxation system in the excise taxation. In energy taxation thereare certain provisions that could be considered tax expenditures, but they are not currentlyincluded in the tax expenditure calculations.

Tax expenditures have been reported in Finland since 1988. The first report covered theperiod from 1984 to 1987. Since then, the report has been published annually. In 1989-1999,the tax expenditure report was published as an appendix to the budget proposal. In 2000, itwas moved to the Parliament’s Report on the administration of government finances andfurther in 2005, to the Report on the Central Government Final Accounts, where a summaryof the tax expenditures in different categories is presented under the topic Fiscal Policy. Thelatest report (published in May 2008) covers tax expenditures in 2005 and 2006. At the webpage of the Government Institute for Economic Research, a more detailed report on taxexpenditures is presented, including estimates for 2008 and 2009. The Report ofgovernment’s financial statements is presented to the Parliament, but tax expenditures are notsystematically discussed as part of the budget process.The reports on tax expenditures have been utilised for tax reform planning (i.e. thecomprehensive income tax reform between 1989 and 1991). It enhanced transparency inpublic finances; created prerequisites for describing better the public subsidies received bydifferent sectors and facilitated the comparison among alternative forms of subsidies.There is no comprehensive reporting of public assistance for specific causes, where both taxexpenditures and direct budgetary expenditure would be included. The tax expenditures arenot reported in connection with subsidies or income transfers that serve the same purpose. Ithas not been thoroughly evaluated, whether the existing tax expenditures would better reachtheir objective if replaced by direct expenditure.The motivation of each tax expenditure is based on the tax expenditure report, which waspublished in 1988 (or 1989). The motivation for each individual tax expenditure is presentedin the tax law proposal when the tax expenditure in question is first introduced. There is nomention in the legislation about the treatment of tax expenditures in general.4.4.5Overview of the most important tax expenditures

The number of tax expenditures has remained almost the same in the past 10 years. Therewere 66 tax expenditures in 1998 and 60 in 2008. Some tax expenditures have been abolishedand new ones have been introduced. In the comprehensive tax reform 1986-1992 and thecapital income tax reform in 1993, a large number of tax expenditures was abolished andreplaced with direct income transfers. The majority of tax expenditures are permanent, butthere are a couple of temporary tax expenditures also (i.e. the reduced rates of value-addedtax for certain labour-intensive services 2007-2010).The tax expenditures are classified under 11 operational categories: General Administration(0.1 % of tax expenditures in 2008), Defence (0.0 %), Education, Science and Culture(1.3 %), Social Security (29.4 %), Health Care (1.0 %), Housing and Environment (30.0 %),Agriculture and Forestry (3.8 %), Transport and Communication (3.7 %), Industry (4.4 %),Other Expenses (0.0 %), and Miscellaneous Items26(26.5 %). The tax expenditures are alsocategorised according to the type of tax (Income Tax, Business Taxation, Indirect Taxation).The sum of tax expenditures in each operational category is also reported for the GeneralGovernment separately.The most significant tax expenditures in 2008 are the exemptions in the area of26 This category includes tax expenditures with non-identifiable allocation or distributions into several categories. The largest taxexpenditures in this category are the earned income tax allowance and earned income tax credit.

social security: deduction of employees’ statutory pension insurance contributions(1 310 mill. €)housing: tax exempt imputed rent of owner-occupied housing (1 900 mill. €) and taxexempt capital gains from selling own housing (900 mill. €), interest deduction ofmortgage etc. (900 mill. €)taxation of labour income: the earned income deduction in municipal taxation(1 400 mill. €) and earned income tax credit in state taxation (800 mill. €).The basis for calculating the tax expenditure of tax exempt capital gains from selling ownhousing is the nominal value of selling prises of houses and estimated holding time of houses.The capital income tax rate (28 per cent) is used. The tax expenditure from the imputed rentof owner-occupied housing is calculated on the basis of the net income from housing, asestimated by Statistic Finland.There are some problems with the calculation of some specific tax expenditures. Incalculating the tax expenditure on deduction of voluntary pension savings contributions, onlythe deduction (and not the future pension income) is taken into account because of lackingdata. This does not give the correct estimate of the total amount of the tax expenditure. Thesame problem concerns the deduction of employers’ statutory pension insurance premiums.The calculation of tax expenditure of depreciation gives also the wrong amount on the truesubsidy effect to the enterprises. It only tells us how much the public sector looses taxrevenues in a specific year.4.4.6Challenges and future development

In the past 20 years the attitude towards tax expenditures has been quite reserved. In recentyears the opinion on directing public expenditure as tax expenditure has however changedsomewhat. There has been an increasing amount of requests for new tax expenditures, and asa consequence, new tax expenditures have been introduced. At the same time, the distancebetween tax expenditure reporting and the budget process has grown.As the role of budgetary spending limits in controlling public expenditure has increased,using tax expenditures seems to have grown more attractive in recent years. This is, ofcourse, not the meaning of the spending limits since as well as direct budgetary expenditure,also tax expenditures are used in allocating public means for different targets, and thus shouldbe under public scrutiny. The National Audit Office and the Parliamentary Audit Committeehave both taken notice on the reporting tax expenditure, and there clearly is a need for a moreextensive discussion of tax expenditure. The Finnish Ministry of Finance has on the 1stMay2008 appointed a working group with a mandate to evaluate, and suggest improvements to,the treatment of tax expenditures. The work is due by the 31stof December 2009. Theworking group is expected to propose major changes in the way some central taxexpenditures (e.g. dividend taxation, fringe benefits) are defined nowadays. Totally new taxexpenditures will be introduced in the areas of excise duties, real estate tax, transfer tax,social security contributions and inheritance tax.

Norway has reported tax expenditures annually since 1999 in the national budget (St. meld.nr.1 (Nasjonalbudsjettet)). The purpose of reporting tax expenditures is to obtain a greaterdegree of transparency regarding political priorities and financial support to different groupsor activities. Tax expenditures (tax sanctions) are defined as exemptions from the generalrules in the tax system that imply lower (higher) tax revenue due to a more gentle (stricter)taxation of certain groups or activities. Norway has chosen a pragmatic approach by using thegeneral rules in the tax system as the reference system, instead of a more idealistic ortheoretical norm. This approach is probably one of the reasons that the issue of taxexpenditures attracts only moderate attention in Norway. On the other hand the informationalvalue of tax expenditures calculated in this manner is limited to a rough estimate of therevenue loss of the tax provision in question. Most tax expenditures can be considered as taxprovisions with a clear element of subsidy, and many of these could easily be replaced byspending programs. There are also a few examples of tax incentives (e.g. R&D expenses andsavings schemes).4.5.2Definition of tax expenditures

The definition of the reference tax system that constitutes the benchmark in the annual taxexpenditure analysis in Norway is rather vague and is based on the general tax system. HenceNorway does not operate with a normative tax system in order to identify tax expenditures.The general tax system is heavily influenced by the low rate/broad base approach and thefundamental principles of equality, neutrality and symmetry. This implies that all types ofincome and all assets should be taxed, and the tax base should be as close to real values aspossible. This approach has made tax expenditures a subject of very little controversy inNorway, because the definition of tax rules and tax bases that constitutes the reference systemis based on clearly identifiable tax rules and a straight forward definition of (otherwisetaxable) income. On the other hand it has probably also led to a minimal consciousness abouttax expenditures in the tax policy work.The reference system is defined in an appendix to the National Budget 2001(St.meld. nr. 1(2000-2001) vedlegg 1). The definition states that“..equal persons, equal activities, equalgoods etc., should be taxed according to the same principles. Exemptions from these generalrules are regarded as tax expenditures or tax sanctions, unless the deviation can be justifiedfor practical or other superior reasons.”This implies that the benchmark for personal income is that all advantages from working,including benefits in kind, and all other types of income should be taxed. The personalincome tax has two tax bases: personal income and ordinary income. Personal income isdefined as income from labour and pensions. Personal income is a gross income base fromwhich no deductions are made, and are subject to a progressive rate schedule. Ordinaryincome is subject to a flat rate of 28 per cent and includes all types of taxable income fromlabour, pensions, business and capital exempt any deductions or allowances. This is in linewith the principles of the dual income taxation.Personal income is subject to a social security contribution of 3 per cent and 7.8 per cent ofpension and wage income respectively. Contributions from self-employed are 11 per cent ofpersonal income from labour. Personal income for self-employed is a net income where costs40

directly related to generation of income are deducted. Deviations from these rates areregarded as tax expenditures. For example is the lower rate for agricultural incomeconsidered as a tax expenditure. An employer’s social security contribution of 14.1 per centof wage costs is also included in the benchmark. The contribution is geographicallydifferentiated with lower rates outside the central parts of southern Norway. These lowerrates are considered as tax expenditures. Personal income above a threshold (NOK 441 000 in2009) is subject to surtax. The progressive tax structure, which includes the surtax, isconsidered as part of the benchmarkOrdinary income is taxed according to a rate of 28 per cent, and this is the benchmark forpersonal income, capital income, dividends and corporations. There are several accessibledeductions and allowances in the computation of tax on ordinary income. Except for costsdirectly linked to the generation of income, these deductions and allowances are generallyregarded as tax expenditures. The basic allowances in income, which contribute to aprogressive tax structure, are considered as part of the benchmark.The benchmark for corporate and capital income taxation is that the real profit of aninvestment is taxed as ordinary income. “Real profit” is however not defined, at least not onan overall level. In the corporate income taxation all deviations from the ordinary tax rate of28 per cent are regarded as tax expenditures. Dividends and capital gains to persons are taxedaccording to the shareholder model, where dividends or capital gains exceeding a risk freeopportunity rate of interest is taxed as ordinary income. The tax allowance equal theopportunity rate of interest is part of the benchmark, and so are interest rate deductions asthey ensure neutrality in capital taxation.Wealth above a threshold (NOK 470 000 per person in 2009) is taxed at a rate of 1.1 per cent.In the computation of the wealth tax base, assessment values should be as close to real valuesas possible. Deviations from this principle are regarded as tax expenditures. The lowassessment value on housing, for example, is one of the largest tax expenditures in Norway.Inheritance tax with lower rates for close relatives is included in the benchmark. The tax baseis the assumed sales values of the inherited assets. Lower assessment values are regarded astax expenditures. However, due to technical difficulty this tax expenditure is not calculated.The benchmark for indirect taxes is split in two; excise duties for fiscal purposes andenvironmental taxes. The benchmark does not take into account differences in efficiencylosses, but treats each tax separately.Excise duties are treated individually which means that a reference rate is made for each tax.Deviations from the standard rate are either regarded as sanctions (if it is higher than thestandard rate) or expenditures (if it is lower). Produced factors of production should inaccordance with theory of optimal taxation not be taxed. These factors are not taxed in thebenchmark for fiscal taxes. Excise duties on production factors will hence normally betreated as sanctions.The benchmark for the VAT is the standard rate of 25 per cent, and any deviation(reductions) from standard rate is calculated as tax expenditures. Also zero rated sectors andsectors outside the VAT system are normally treated and calculated as tax expenditures. Thecalculation of tax expenditures from zero-rated areas and sectors outside the VAT system, isbased on the difference between levying zero rates and exemptions instead of the referencesystem with the standard rate.

Norway uses the revenue foregone27method when calculating tax expenditures. Only taxexpenditures in general government taxation are calculated, although the revenue loss willalso affect local government income as tax revenue from tax on ordinary income is dividedbetween central and local government.The actual calculation of tax expenditures in direct taxation is for the most part performed byStatistics Norway, but some tax expenditures are calculated by the Ministry of Finance.Statistics Norway uses a microsimulation model called LOTTE-Skatt when calculatingrevenue effects from changes in direct taxation and tax expenditures. This model is based ontax returns from a selection of households. Tax expenditures are calculated by calculating taxrevenue when one deduction or allowance is removed compared to a reference system whereall existing tax expenditures are included.LOTTE-Skatt calculates most of the tax expenditures in personal income taxation and incapital and wealth taxation. Tax expenditures that cannot be calculated by LOTTE-Skatt arecalculated by the Ministry of Finance (although a limited number of tax expenditures are notcalculated at all). The Ministry also calculates some tax expenditures with a time horizon.The tax expenditure related to employee premiums and contributions to occupational pensionschemes are calculated as a present value. The same method is used for tax expendituresrelated to depreciation rates higher than what is assumed to be the real depreciation rate.Tax expenditures related to indirect taxes are not calculated by models, but by ad hocmethods based on sectoral statistical information and information from tax administrationsystems. Tax expenditures related to environmental taxes are calculated by deviations fromthe reference rate that normally is set in accordance with the estimated external costs. Areasthat are exempted from taxes but subject to other significant measures, for example quotas,are not normally treated as tax expenditures.Challenges in calculating indirect tax expenditures are, among others, to decide to whatdegree goods are consumer goods and/or production factors and how to treat taxes thatcombine fiscal and environmental objectives.4.5.4Reporting and evaluation

calculated. In the following year more tax expenditures were reported and calculated, andseveral deductions and allowances that previously were not regarded as tax expenditureswere included. In the first report the tax expenditures were classified according to purpose.Since 2000 the tax expenditures in direct taxation have been classified according to taxablebase. Expenditures in indirect taxation are classified according to both the tax type (VAT,excises) and objectives (fiscal, environmental/health). Today the report contains a shortdescription of all exceptions and deviations from the benchmark system, and the revenueeffect is calculated when possible. The description is neutral and the tax expenditures are notjustified in any way.Norway operates with a balanced budget rule, which is formulated as a structural non-oilcentral government budget deficit. This deficit shall over time correspond to the expected realreturn, estimated at 4 per cent, on the Government Pension Fund – Global (the formerPetroleum Fund). However, the guidelines also allow fiscal policy to be used actively tocounter fluctuations in economic activity. This budgeting method implies that thegovernment has a choice between tax expenditures and direct expenditures as policyinstruments. It also implies that reduced taxes and tax expenditures imply less revenue to bedistributed on the spending side of the budget.The decision making process on new tax expenditures is the same as direct transfers, whichrequires a majority in the Parliament. But once the tax expenditures are in place they are notsubject to the same scrutiny. All the tax expenditures are permanent, and as opposed to directtransfers, where block grants are common, tax expenditures are always open-ended, whichmake them more difficult to project and control. Tax expenditures reflect political priorities,just as direct transfers do. By publishing the tax expenditures the amount of governmentalsupport to different groups becomes more evident. But the tax expenditures are for the mostpart not published side-by side with direct expenditures in the budget, and they are less likelyto undergo rigorous review and repeal. The exception is a chapter in the National Budgetwhere direct transfer and tax transfers to different industries are reported together under theheading “industrial support”.Even though the purpose of reporting the tax expenditures is to obtain a greater degree oftransparency regarding political priorities and financial support to different groups oractivities, there are no attempts to assess whether the targeted groups of the tax subsidiesreally are the beneficiary, or if the tax incentives alter behaviour in the desired direction.There are still some examples of more thorough evaluation of some tax expenditures. The taxallowance for R&D expenses, for instance, has been thoroughly evaluated by StatisticsNorway. The Tax Committee, appointed by the Government in order to evaluate theobjectives and principles applicable to the tax system prior the tax reform of 2006, alsolooked into several tax expenditures. The committee’s recommendation was to abolish a widerange of tax expenditures in order to secure a broad tax base and horizontal equality oftaxation (with equal taxation of persons with the same tax liability).In general the politicians are hesitant to use tax incentives and most tax incentives incorporate and business taxation where abolished in 1992. Although there are several relievesand allowances that can be regarded as tax expenditures, few of them can rightly be definedas tax incentives. None the less, some saving schemes (BSU, retirement saving) and theallowance for R&D expenses are examples of the opposite.

In 2008 Norway had 60 calculated tax expenditures and 5 calculated tax sanctions. The actualnumber was higher, as there are several tax expenditures that are not calculated due totechnical difficulty. When calculated by the revenue foregone method, the tax expendituresfor 2008 add up to almost 16 billion Euros. This corresponds to 12.6 per cent of total taxrevenue and 5.4 per cent of GDP. It should be stressed that these numbers must beinterpreted with care as neither behavioural effects nor interaction effects are taken intoaccount.During the last 10 years an increasing number of deductions and allowances have beendefined as tax expenditures. This, of course, has contributed to an increased number of taxexpenditures, although the actual revenue effect is the same regardless of whether they areregarded as tax expenditures or not. Some examples of such allowances are the childcareexpense deduction and the tax allowance for commuters’ daily work travel and visits to mainresidence, which were not regarded as tax expenditures until 1999. The same applies to thetax expenditure related to the geographically differentiated employer’s social securitycontribution. These tax expenditures are now some of the most significant tax expenditures inNorway.The tax expenditures are also growing in size, at least on a general basis. A substantial part ofthis growth can be attributed to growing tax basis and growing tax revenue, but some changesin tax policy have also contributed to this development. This particularly applies to taxexpenditures in capital income tax. The taxation of imputed income of owner occupiedhousing was removed in 2005. This policy change led to more than a doubling of the taxexpenditure related to income taxation on own housing and vacation property. The taxexpenditure related to the wealth tax on housing and vacation property also has grownsignificantly. Much of this is due to exceptional growth in marked values on properties,without an equal growth in assessment values. There has also been considerable growth in thetax expenditure linked to the tax relief on employee premiums and contributions tooccupational pension schemes. Increasing employment is the cause of a substantial part ofthis growth, but the introduction of compulsory occupational pension in 2006 has also beenof importance. The tax rules for shipping companies were changed in 2008, with effect from2007. The shipping companies are now exempted from ordinary capital income tax, whereasthey earlier had a tax deferral arrangement. The tax expenditure in personal income has beenmore stable, although some tax expenditures have grown and some have been reduced. Also,some tax expenditures have been abolished and new ones have been introduced.The Norwegian VAT-reform from July 2001 broadened the VAT-base substantially byincluding services in the VAT base on a general basis. Since the establishment of theNorwegian VAT-system in 1970, services in general were not inside the scope unless ifspecifically listed in the VAT-law. A low VAT-rate on food was also introduced as a result ofthe reform in 2001. The tax expenditures in the VAT system in 2002 were estimated to aboutNOK 11 515 mill. (1 300 mill. €). Several sectors have been included in the VAT-base since2002, e.g. accommodation, travel agencies, broadcasting and cinemas. These sectors havebeen included in the system with a reduced rate of 8 per cent, and are therefore still a sourceto tax expenditures. Recent increases of the VAT rate, latest from 24 to 25 per cent from 1.January 2001, has also contributed to increased tax expenditures.The main tax expenditures in the VAT arise because some sectors are (i) outside the VAT-system, (ii) inside the system with a lower rate, or (iii) inside the system with a rate equal to44

zero (zero-rate). The tax expenditures calculated for sectors outside the VAT-system areestimated to about NOK 1 410 mill. (160 mill. €) in 2008. The tax expenditures calculated forsectors with a lower rate are estimated to about NOK 12 225 mill. (1 380 mill. € ) the sameyear. The bulk of this expenditure is due to the reduced rate on food (14 per cent vs. thegeneral rate of 25 per cent). The tax expenditures due to zero-rated areas are estimated toabout NOK 3 250 mill. (370 mill. €). This is mainly related to zero rating of newspapers,books and periodicals.The main tax expenditures in the personal income tax include additional personal allowancefor one-income families and sole parents, childcare expense deduction and tax allowance forcommuters’ daily work travels and visits to main residence. All these tax expenditures arecalculated by LOTTE-Skatt and are estimated to about NOK 1 755 mill. (200 mill. €), NOK1 960 mill. (220 mill. €) and NOK 1 510 mill. (170 mill. €) respectively in 2008. The maintax expenditures in capital income tax include deductions for employee premiums andcontributions to occupational pension schemes, lacking income taxation on own housing andvacation property, lower assessment value than real value on housing and vacation propertyand tax exemption for ordinary capital income tax for shipping companies. The taxexpenditure related to pension premiums is calculated as if the deposit had been paid out aswage income and saved in a bank. The disbursement will be taxed as pension income, andthis is taken into consideration when calculating the present value of the tax savings. This taxexpenditure is an example of a tax incentive and was estimated to NOK 18.2 billion (2 billion€) in 2008. The calculation of the tax expenditure related to lacking income taxation on ownhousing is based on marked values on marketed houses over a five year period. Marketvalues on vacation property are estimated by assuming that market values are four timeshigher than assessment values. The tax expenditure is calculated by assuming a real return of5 per cent rate on the investment and was estimated to NOK 58 billion (7 billion €) in 2008.These calculated marked values are also used when calculating the tax expenditure related tolower assessment value than real value on housing and vacation property, which wereestimated to NOK 26 billion (3 billion €) in 2008. The favourable tax rules for shippingcompanies are estimated to constitute a tax expenditure of NOK 2.4 billion (270 mill. €) in2008. This estimate is based on annual accounts over a four year period. The regionallydifferentiated employers’ national insurance contribution also represents considerable taxexpenditures in Norway, estimated to almost NOK 10 billion (1.1 billion €) in 2008. Thecalculation is based on reported wage payments by employers.Tax expenditures from excise duties sum up to about NOK 16.5 billions (1.9 billion €) withapproximately NOK 10.5 billions (1.2 billion €) linked to environmentally and health relatedexcise taxes. The main sources are the excise tax on energy consumption, taxes on alcoholand tobacco, and a lower tax rate on diesel fuel than petrol.There are also calculated tax sanctions related to excise taxes on totally NOK 4.4 billions.The largest sanction is the CO2-tax on petroleum related activities which partly is committedto a quota-regime. In addition to taking account of alternative measures, corrections are donefor other reasons, for instance for activities that can be deducted from the special tax onpetroleum activities, and that part of the CO2-tax is paid by the governmental institution thatmanages the ownership of the Norwegian state. This illustrates the complexity of calculationsand the numbers of consideration linked to tax expenditures/sanctions.

The conceptual idea of producing tax expenditures is to provide a framework from which toevaluate the equity, efficiency and administrative issues raised by new or existing tax rules.However, as previously described in this report, practical, and to some extent theoretical,difficulties related to defining the “normal” tax system from which the definition of taxexpenditures arises, weakens this ambitious objective. The purpose of producing andreporting tax expenditure estimates in Norway is not very clear, and the list of taxexpenditures is presented each year in the annual budget for informational purposes. Thenormal tax system (the benchmark) is vaguely defined in general and is more or less based onwhat constitutes the general tax law. The methodology is somewhat unclear and lacks thenecessary stringency when applied to different tax expenditures. The absence of behaviouraleffects and accounting periods makes the magnitudes of the estimates unreliable. Hence theannual tax expenditure analysis in Norway must be considered as inadequate to fulfil suchambitious goals mention above. At best the reporting of tax expenditures in Norway gives arough overview of the relative importance of the main deviations from what can beconsidered as the “general” tax legislation. A tax expenditure estimate is only adequate forefficiency and distributional analysis if the reference system constitutes an efficient or fairlyredistributive tax system. So the informational value of tax expenditures in the sense ofequity or efficiency costs is only as good as the reference system. On the other hand, thegeneral corporate and capital income tax law in Norway is fairly neutral, and the taxation oflabour income and wealth is quite progressive compared to the majority of OECD-countries.This increases the informational value of the estimated tax expenditures. However, the linkbetween the methodology used for producing tax expenditures and the informational value ofthe tax expenditures needs to be clarified.4.6Tax expenditures in SwedenIntroduction

The Swedish Government has since 1996 reported tax expenditures (and tax sanctions) to theParliament annually in conjunction to the Spring Fiscal Policy Bill. The main objective withthe reports is to illuminate the implicit support given on the budget revenue side. The taxexpenditure reports may hence serve as a basis for prioritising among different policy areas.In the autumn, as supplements to the Budget Bill, the tax expenditures are re-reported withineach policy area. There is also an expressed desire to describe the degree of uniformityamong the tax rules. Such a description, as well as the illumination of implicit support,requires that a benchmark is specified. The Swedish benchmark for tax expenditures aims tobuild on the principal of uniform taxation. Deviations from the benchmark are considered taxexpenditures and tax sanctions, which are estimated and presented. In general both revenueforgone and equivalent outlay are estimated and presented.The current benchmark is relatively simple to apply and communicate. A strict application ofthe benchmark is rather inflexible, however. It does not consider the motives behind taxexpenditures. From an economic efficiency viewpoint it may, however, be warranted withtaxes deviating from the benchmark. The Swedish Government intends to initiate furtherdevelopment of the benchmark in a direction where economic efficiency and optimal taxationreceive greater attention.

The Swedish benchmark builds on the principal of uniform taxation, which was one of thecornerstones of the tax reform in 1990-91. The principle implies that each type of tax shouldbe levied uniformly and without exemptions. Deviations from the principle constitute taxexpenditures (or tax sanctions). The benchmark evolves with the tax system and is constantlyunder revision. For example, in 2003 there was a major overhaul of the excise dutybenchmark. The Swedish tax expenditures can be divided into four tax areas; incometaxation, indirect taxation of labour income, VAT and excise duties. The benchmark differsbetween the tax areas.28There is no single benchmark tax rate for income taxation. Labour income, capital incomeand income from businesses may be taxed at different rates. The benchmark does howeverstipulate that the taxable income should correspond to the SHS income concept, i.e. incomeequals all consumption expenditures and the change in net wealth. Empirically, the Haig-Simons income is hard to measure and in order to make it operational further clarificationshave been made. These include among others;•••••••Savings should be done with taxed incomes.Capital gains/losses should be taxed on accrual and not at realisation.The value of non-wage household work and leisure shall not be part of the tax base.Public payments are part of taxable income.The implicit return to owners of homes and condominiums is part of taxable income.Depreciation allowances in businesses should be based on true depreciation.Only interest payments made for loans where the investment return is taxable shouldbe deductable.

In addition, the benchmark stipulates horizontal uniformity rather than vertical as differentdegrees of progressivity in the tax schedule are considered consistent with the benchmark.Moreover, tax credits are included in the benchmark if they are general and do not favourspecific groups of tax payers. Tax reductions due to interest payments and the in-work taxcredit are, for example, part of the benchmark.The benchmark for indirect taxation of labour income is that Social Security Contributions(SSC) or SSC paid by the self-employed and the general pension contribution paid by theemployee29should be levied on all remunerations which give eligibility in the social securitysystem. Remunerations not giving eligibility are subject to a ‘special tax on certain earnedincome and pension costs’.30For each type of contribution/tax there is only a single rate inthe benchmark.The VAT benchmark consists of a single rate, the standard VAT rate 25 per cent. Allcommercial good and service transactions should be taxed at this rate. There are, however, afew exceptions to this benchmark. The VAT exemption on letting property is considered as apart of the benchmark since such taxation would imply unequal treatment of rented andowned housing. The VAT exemption of financial and insurance services is considered to bepart of the benchmark. Goods and services which are subsidised by public funds are not28 The Swedish Government Official Report “Förmåner och sanktioner” (SOU 1995:36) and the Spring Fiscal Policy Bill of 1997 discussthe benchmark more thoroughly.29 Allmän pensionsavgift.30 Särskild löneskatt.

taxable when the net VAT is negative and not considered as tax expenditure. Finally, forinternational transactions the principle of destination is considered the benchmark. VATshould hence be levied in the country where the good or service is consumed.For excise duties, only energy and CO2taxes including the tax on thermal effect levied onnuclear reactors (a tax sanction) are considered when tax expenditures are calculated. Prior tothe 2003 overhaul of the benchmark more duties were included. It was decided that onlyduties with large tax bases were to be included in the tax expenditure reports. If an exciseduty would, under a uniform rate, give rise to at least one per cent of the total tax revenue itsbase is considered large.The benchmark for energy tax is that all energy consumption should be levied the same taxper kWh. Electricity and other sources of energy may however be taxed at different rates.Moreover, it is consistent with the benchmark to differentiate between heating and fuelconsumption.31The rationale being that the tax on fuel also, besides serving fiscal purposes,captures some of the societies’ cost of road traffic, e.g. wear and tear on roads, noise andaccidents. The benchmark for electricity consumption is the normal tax rate on electricity.For other energy sources used for heating purposes the benchmark is the heating oil tax rate.For energy resources used as fuel, the tax rate of petrol constitutes the benchmark.The CO2tax should according to the benchmark be levied proportional to emissions. Thebenchmark tax rate is the normal CO2tax rate regardless the source of energy and usage.There are, however, exemptions to the benchmark tax rates. Fuel used as an input in theindustry is exempted from tax and this exemption is not considered tax expenditure. Theenergy tax on electricity is only applied to the usage of electricity and a tax exemption of fuelused in the production of electricity is therefore not considered tax expenditure.4.6.3Methods of calculation

In 2009 some 115 tax expenditures and tax sanctions were reported. The estimated taxexpenditures were around 25 billion euros. This corresponds to around 8 per cent of GDP and17 per cent of total tax revenues. These figures, as well as later figures, should be interpretedwith care, especially if different years are compared. For several reasons the reports of taxexpenditures are not complete. For example, data used for some calculations may beinsufficient. In addition, there may be tax expenditures not included in the reports. Typically,a couple of new expenditures are added each year due to changes in the tax system whereassome tax expenditures will cease to exist or simply expire. In 2008 the estimated taxexpenditures were 16 per cent of the total tax revenue, but the three preceding years theshares were only 11 to 13 per cent.Two of the largest tax expenditures in 2009 are associated with tax on capital income andowner occupied housing. First, the implicit real return to owners of homes is not taxed ascapital. Instead a municipal fee, designed as a tax, is paid. The revenue forgone due to thisdifferential treatment is estimated to around 2 billion euros. In order to estimate the taxexpenditure, real returns are calculated based on the market value of the properties and thereal government interest rate. The difference between the capital income tax on these returns(if they were to be taxed) and the fees paid under the current system amounts to the taxexpenditure. Second, capital gains from housing are taxed upon realisation. Under certaincircumstances the gains, and hence the tax, may be deferred even further. At the present,1.67 per cent of the deferred amount is taxed as capital income. The difference between theserules and the rules for taxation of deferred capital incomes is considered tax expenditure. Thedeviation from the benchmark is estimated to 1.7 billion euros in foregone revenue.Three other large tax expenditures involve VAT, indirect taxation of labour income andenergy taxes. For food items there is a reduced VAT. Instead of the normal tax rate of 25 percent there is a 12 per cent VAT levied on food items. The difference between the tax rates isconsidered tax expenditures, estimated to about 2 billion euros. For persons under the age of26 (at the beginning of the year) there is a reduction in social security payments. There is alsoa reduced energy tax on electricity used in industry production. These latter tax expenditureshave an estimated forgone revenue of 1.4 billion euros each.4.6.6Challenges and future development

not suited for continuous evaluation of different tax expenditures. According to NAO theGovernment ought to consider how tax expenditures should be treated in the fiscal process.These issues are considered in the continuous development of the treatment of taxexpenditures in order to make them more visible and an effective tool in the fiscal process.Besides this continuous developmental work a more fundamental methodological issue is onthe rise.The current benchmark is relatively simple to apply and communicate. A strict application ofthe benchmark is rather inflexible, however. It does not consider the motives behind taxexpenditures. From an economic efficiency viewpoint it may, however, be warranted withtaxes deviating from the present benchmark. For example, taxes that correct for marketfailure by internalising external effects. Moreover, certain tax expenditures may expand thetax base and in the end increase welfare. A lower – than the benchmark – tax may bemotivated if the deviation increases, directly or indirectly, hours worked. This would increasethe labour income tax base. Deviations like these, based on a principle of optimal taxation,could contribute to higher welfare. Under the current benchmark such deviations areconsidered tax expenditures. It could be questioned if such a rigid interpretation of thebenchmark and tax expenditures is appropriate. The Swedish Government intends to initiatefurther development of the benchmark in a direction where economic efficiency and optimaltaxation receive greater attention.

5 Conclusions and suggestions for future workIn section 2 we present the concept of tax expenditures, originally defined by Stanley Surreyin the late sixties. Surreys’ main objective was increased expenditure control by identifying arange of tax provisions that escaped the scrutiny applied to regular subsidies and transfers onthe expenditure side of the budget. This objective is still the core of any tax expenditureanalysis. An extensive review of all kinds of deviations from the general tax system is crucialin order to get a full picture of public provisions that consists of both direct public transfersand favourable tax treatments. Since the first tax expenditure analysis was presented in theUS more than forty years ago, the concept of tax expenditures has spread worldwide andtoday virtually all OECD-countries make tax expenditure analysis in one form or another.Using tax expenditure analysis in the tax policy work to pursue these idealistic principles ofthe tax system is highly ambitious, and would require a benchmark developed from theseprinciples. In other words ideally you need to define a normative tax system which reflectsthe principles of efficiency, fairness and simplicity, both in the rate structure and in thedefinition of income, and then be able to operationalise it in an estimation procedure. Ofcourse this is not an easy task, and in practice the ideal principles have been proven difficultto accomplish. As discussed in section 2, tax expenditures have been vividly discussed andcriticised among other things for the difficulties to establish a commonly acceptedbenchmark.The examination of tax expenditure analysis of the Nordic countries above shows that thelevels of ambitions are fairly similar. Norway aims at “obtaining a greater degree oftransparency regarding political priorities and financial support to different groups oractivities”. Sweden has the most ambitious goal, namely to “to illuminate the implicit supportgiven on the budget revenue side”, but also “an expressed desire of describing the degree ofuniformity among the tax rules”. Finland and Denmark do not express the goal explicitly, butDenmark seems to have a fairly pragmatic reference system, and Finland use a methodologywhich to a large extent seems to be comparable to the Norwegian methodology. The NationalAudit Offices in Finland, Denmark and Sweden have to some extent criticised the taxexpenditure analyses for not being sufficiently transparent.The working group thinks it is crucial to ask; what do the tax expenditure analyses in theNordic countries really tell us? Originally the ambitious goals of making tax expenditureanalyses formulated by Surrey were among other things to draw reliable conclusions abouteffectiveness, equity or other desirable features in the tax system. Conventional methods ofcalculating tax expenditures presented in this report are insufficient in fulfilling theseambitious goals, and in practice the goals with tax expenditures are much more modest asindicated above. The Nordic tax expenditure analyses have so far more or less been based onidentifying deviations from the basic principles of uniform taxation or the tax system inplace. These analyses are therefore less vulnerable to the general criticism regarding thenormative assessments of the benchmark. A uniform taxation implies a broad tax base butthere could still be problems if real income or real expenses are unobservable. What is thecorrect depreciation of (different kinds of) real capital? How do you divide business incomefor self employed in capital income and labour income? What is the correct imputed incomefrom owner occupied housing? Hence there are many examples of tax expenditurecalculations that must rely on data that are not always appropriate and relevant.

Potential areas for future developmentIt has been beyond the capacity of this working group to work out detailed recommendationsfor a reform of tax expenditure analyses in the Nordic countries. However, the group wouldlike to point at some potential areas where future development efforts may be placed. Basedon the findings in this report and also with a reference to the National Audit Offices criticismthat the tax expenditure reports to some extent lack transparency, the working group haspointed out some areas for further development, in order to achieve a higher degree oftransparency and probably increase the informational value of the tax expenditure analyses.One suggestion is that the Nordic countries could endeavour to clarify the tax expenditureanalysis in future reforms. Another is that more emphasis could be placed onevaluatingtaxexpenditures.A part of a clarification process could be to review the benchmark. As stated above we findthe benchmarks that constitute the foundations of the tax expenditure analyses to besomewhat vaguely defined in some cases. One way forward could be to base the benchmarkon easy identifiable general tax rules, and let the income definition be consistent with eitherSchanz-Haig-Simons or what would (otherwise) be the taxable income. This would reducethe ambition of making tax expenditure analysis toidentify deviations from the general taxsystem that generates less tax revenues than it would otherwise have done if the general taxlegislation were applied.This simple benchmark could be combined with efficiencypromoting measures in order to create a benchmark that better reflects an economicallyefficient tax system. A simplified tax expenditure analysis like this could be accompanied byseparate efficiency and equity analyses of tax provisions that escape the definition of taxexpenditures. This could be done first and foremost in a qualitative manner, and secondly in aquantitative manner if there is a feasible and credible way to measure the welfare lossgenerated by the tax provision.The group also suggests having more extensive and more frequentevaluationsof taxexpenditures, in form of periodical reviews of the effectiveness of the measures and theachievements of the intended goals. Both the actual tax provisions and the tax expenditureanalyses should be investigated. A comprehensive evaluation of all tax expenditures would ofcourse demand a vast amount of resources (as described in section 3), and is thereforeunrealistic. Evaluations should therefore be carried out on a smaller scale, but be specific andtargeted and make sure that all tax expenditures are subject to a proper evaluation at somepoint or another. It could for example be a good idea to introduce a “fixed evaluation date”for all new tax expenditure, say three years after introduction, to make sure that they aresubject to proper evaluation at least once, and maybe to increase the consciousness of thepoliticians about the specific tax expenditure when passing the budget.Finally the group will encourage properly and open discussions about tax expenditures. Apart of this could be to invite academics, tax experts, politicians etc. to comment or analysethe tax expenditures publicly and on a frequent basis. This could help to increasetransparency, make the politicians more aware of the tax expenditures and help the civilservants and technicians to improve the tax expenditures analyses.

6 LiteratureAustralian Treasury (2009), Tax Expenditure Statements,www.treasury.gov.au.Bittker, B.I. (1969) "Accounting for Federal 'Tax Subsidies' in the National Budget",National Tax Journal 22, 244-266.Bratić, Vjeskoslav (2006), Tax Expenditures: A Theoretical Review, Financial Theory andPractice, Vol. 30 No. 2, 113-127.Boss, Alfred and Astrid Rosenschon, (2008), Der Kieler Subventionsbericht: EineAktualisierung, Kieler Diskussionsbeiträge Nr. 452/453, Mai 2008, Institut fürWeltwirtschaft Kiel,www.ifw-kiel.de.Burman (2003): Is the Tax Expenditure Concept Still Relevant? National Tax JournalSeptember 2003.German Government (2007), Einundzwanzigster Subventionsbericht, Bericht derBundesregierung über die Entwicklung der Finanzhilfen des Bundes und derSteuervergünstigungen für die Jahre 2005-2008,www.bundesfinanzministerium.de.GSI (Global Subsidies Initiative) (2007), Germany announced review of federal taxsubsidies,www.globalsubsidies.org.Joint Committee on Taxation (2008), A Reconsideration of Tax Expenditure Analysis, JCX-37-08, Washington,www.house.gov/jct/x-37-08.pdf.Kraan, Dirk-Jan (2004), Off-budget and Tax Expenditures, in: OECD Journal on Budgeting,Volume 4 No. 1, Paris,www.oecd.com.LAO (Legislative Analyst’s Office) (2003), Tax Expenditure Programs: Reporting andEvaluation,www.assembly.ca.gov.OECD (1996), Tax Expenditures: Recent Experiences,www.oecd.com.OECD (2003), Working Party No. 2 on Tax Policy Analysis and Tax Statistics – SpecialFeatures for the 2003 Edition of Revenue Statistics, WP2(2003)2, Paris,www.oecd.com.OECD (2008/1), Working Party No. 2 on Tax Policy Analysis and Tax Statistics – TaxExpenditures in OECD Countries, WP2(2008)24, Paris,www.oecd.com.OECD (2008/2), Working Party No. 2 on Tax Policy Analysis and Tax Statistics – TaxExpenditures and Base Broadening, WP2(2008)25, Paris,www.oecd.com.Surrey, Stanley S. and McDaniel P.R (1985): Tax Expenditures, Harvard University Press.Thöne, Michael (2003), Subventionskontrolle: Ziele – Methoden – internationale Erfah-rungen, Edition Sigma, Berlin.Thöne, Michael (2005), Subventionen und staatliche Beihilfen in Deutschland, Forschungs-vorhaben Nr. 18/03: „Einheitliches Verfahren zur differenzierten Erfassung und Messung von53

7 Appendix - What do other countries report?See in detail the following overview which is based on OECD (2008/1), p. 39 ff.:

Country

Reporting

Location of Estimates

Frequency

Reporting of tax expenditures is notrequired by law, but Canada reports itstax expenditures each year – since 1997,reports have covered the report year, thefive preceding years, and two succeedingyears (thus eight years in total). Eachcycle’s tax expenditures for the threeearliest years for the individual incometax, and the four earliest years for thecorporate income tax, are developedusing final administrative data; the lateryears’ figures are estimates orprojections. Tax expenditure figures areproduced for calendar years rather thanfiscal years. Every four years, Canadaproduces a detailed enumeration anddescription of all tax expenditures.As noted above, tax expenditures arereported each year in the budget act andin the finance bill of social security. Boththe budget act and the budget for socialsecurity act report the cost of taxexpenditures for the budget year and thetwo prior years.There is no formal mechanism for taxexpenditure review by parliament orCabinet after provisions have beenapproved in a budget. However, taxmeasures are reviewed on an ongoingbasis within the Department of Finance(and the Canada Revenue Agency, withrespect to administrative matters), withtechnical input as appropriate from linedepartments. Some measures areevaluated more formally on adiscretionary basis, and the results arepublished.

Germany has begun a process of formalreviews of tax expenditures. The 20largest tax expenditures – accountingfor 92 percent of the total cost of all taxexpenditures – are to be evaluated. Theevaluations are charged to define theobjective of the tax expenditure,including macroeconomic motivationsor perceived market failures; determinewhether the tax expenditures areeffective and efficient, and whether thetax expenditure is the best public-policy instrument to pursue theobjective; and to find any side effectsfor the tax system broadly. Severalrespected outside research institutesperform the reviews, with the use ofmultiple reviewers seen as an importantguarantee of unbiased analysis. TheMinistry of Finance will comment onthe reviews, and report the findings tothe Parliament.Special Tax Measures are reviewedannually by tax officials of Ministry ofFinance, mainly focusing on those thatexpire in the next year due to sunsetclauses. Usually, the majority of theSpecial Tax Measures at the nationallevel are stipulated in the Special TaxMeasures Laws to have two- or three-year sunset clauses. These sunsetclauses have functioned effectively,because they force tax officials andother related parties to review thecontents of the Special Tax Measuresregularly. Negotiations between taxofficials and the requesting ministriesover the Special Tax Measures expiringin the next spring (usually the end ofMarch) begin in September, at thesame time as with the budgetexpenditure negotiations. In manycases, each ministry requests thecreation of new Special Tax Measuresfor their policy objectives. Thenecessity, effectiveness and efficiencyof the measures are scrutinized in the

The Tax Expenditure Report is releasedannually, in keeping with the language ofthe National Fiscal Act. The NationalFiscal Act also mandates that estimatesbe provided for the year prior to thebudget year, the budget year itself, andthe year after; however, the requirementis not considered binding until 2010, andprojections for the succeeding year havenot to date been provided. Theretrospective year is based on final data;the current year is a projection. Taxexpenditures hitherto have been reportedaccording to functional areas that havenot aligned with the functions for thereporting of spending. Recategorizing thetax expenditures, and providing estimatesfor the succeeding year, are among theleading tasks for the 2010 reporting andprocess reform.The Tax Plan and Budget Memorandumis presented every year. It provides taxexpenditure figures for the budget year,one prior year, and the five succeedingyears. The tax expenditures reported forthe year prior to the budget year are thefinal figures for that year. Changes inindividual tax expenditures – repeals,new provisions, increases and decreases– are presented.

Although there is no statutoryrequirement to produce a report on taxexpenditures, the Government stillestimates and reports all major taxexpenditures in the Tax Ready Reckonerevery autumn. And Chapter A of theannual Budget, Budget and PolicyDecisions of Financial Statement andBudget Report, contains a list ofproposed tax expenditures. Nocomprehensive historical report exists,but the Financial Statement and BudgetReport was first reported followingapproval of Parliament (for the purposesof Section 5 of the EuropeanCommunities Amendments Act) in 1993.The “Financial Statement and BudgetReport” has been published online onlysince 1997.Tax expenditures were presented for thefirst time in the FY 1976 Budget issuedin 1975. Since the late 1970s, the taxexpenditure tables show seven years ofestimates: the two years prior to the yearof the budget, the year of the budget, andthe four years following the year of thebudget. The estimates are currently basedon the economic forecast used for themid-year estimates of the budget andthey are not retrospectively revised orupdated. Each year’s Budget includeslistings of all new tax provisions enactedin the preceding year, but no separatelisting of new tax expenditures. When

Congressional Research Service of theLibrary of Congress; that review doesnot reflect the views of eitherExecutive or Legislative policymakers.Also, the governmental but non-partisan Congressional Budget Officeproduces a once-every-two-yearsvolume of potential policy changes toreduce the deficit; the ideas consideredinevitably included some reductions orrepeals of existing tax expenditures.Tax expenditures receive considerableattention whenever tax reform is on thepolitical agenda. In 2005, thePresident’s Advisory Panel on FederalTax Reform issued a report calling forthe comprehensive overhaul of the taxsystem, which would have altereddrastically many of the largest taxexpenditures. This effort at tax reformdid not lead to legislation, but thecentral place of tax expenditures in thereform options is typical of what ageneral tax reform would produce. Inthe FY 2008 Budget, and again in theFY 2009 Budget, the Presidentproposed major changes in the taxexpenditure for private healthinsurance.