By

Asuman Altay

ABSTRACT: The traditional analysis of optimal taxation searches for a system of taxation that minimises the excise burden(welfare cost) of taxation. This literature does not offer any information to those politicians and bureaucrat’s selfish behaviour upon the tax structure. Additionally, traditional analysis of optimal taxation states that there are differences between alternative tax regimes according to social welfare functions. There are several approaches to optimal taxation, which is derived from political agent’s behaviour to tax, structure determination in political process. In this study examines traditional analysis and different approaches of optimal taxation as a survey.

Keywords: Optimal Taxation

Pareto Optimal

Dead –weight Cost of Taxation

Public Choice

Rent Seeking

*This study was made in Public Sector Economics Research Centre (PSERC), University of Leicester, between 15 March 1998-15 June 1998 as a visiting scholar which was supported by the Turkish Academy of Sciences.

**Asuman Altay, Associate Professor, Dokuz Eylul University, Faculty of Economics and Administrative Sciences, Department of Public Finance, Buca, 35160 Izmir, Turkey. e-mail: aaltay@sifne.iibf.deu.edu.tr.

The aspect of economic efficiency of a good tax system is to try to minimise the excess burden (including the effects on the supply of factors of production) at a given level of tax revenue. A good tax system can also be evaluated in terms of how fair or equitable to be. Optimal taxation literature has been interested in the integration of both “efficiency and “equity” criteria. But there are some difficulties and trade off between efficiency and equity criteria.

There are three different criteria for optimality of the tax system (Sandmo, 1976,37).

First, A good tax system is the one that minimises the resource cost involved in assessing, collecting and paying the taxes. It depends on frequently the efficiency and behaviour of tax administration.

Second, Alternative tax systems can be evaluated in terms of “justice” or “fairness”. Because taxpayer concept of justice may not be very precise in many cases.

Third, tax systems should be lined on the criterion of economic efficiency.

The third criterion is also the main point of departure from the economical theory of optimal taxation. In other words, the optimal taxation (or an optimal tax system) is the one which minimises the aggregate dead-weight loss for any given tax revenue or level of public expenditure. Optimal taxation theory has also been gradually extended to take account of distributional considerations.

Heady also pointed out three criteria for some taxation proposals which are analysed in terms of following criteria (Heady, 1993,15).

- Fairness;

- Minimise administrative costs;

- Minimise disincentive effects.

According to Heady, the literature of optimal taxation tries to combine these criteria into one criterion. Meaningless the literature of optimal taxation uses the concepts of individual (or household) utility and social welfare. Social welfare has become important yield of optimal taxation literature in 1970. Social welfare firstly depends on utilities of individuals then distribution of these utilities on the society.

Considering with an efficiency point of view of ideal tax system we have to consider a Pareto optimal allocation of resources. Optimal tax literature is also concerned with the “second-best problem”. Some important articles as mentioned above have been published by Learner (1970); Dictate (1970); Diamond and Mirrlees (1971). Especially Diamond and Mirlees have extended and generalised Ramsey’s formulation in their notable article.

The literature of optimal taxation is so rich and voluminous. There are several surveys' studies that provide an introductory survey to optimal commodity taxation, and Atkinson’s (1977) paper includes direct versus indirect tax controversy. A more technical and more general discussion on this issue appears in Atkinson and Stiglitz (1980). In addition very interesting and important paper was written in 1951 by Samuelson.

It should be noted that optimal tax analysis has concentrated on personal income taxes and commodity taxes. It has not dealt with company taxation, capital gain tax or inheritance taxes. The reason for this is probably that the effects of these taxes on behaviour and utility are less well understood than the effects of personal income tax and commodity taxes. Also, issues of administrative enforcement and costs are considered as more important for these taxes. This presents an extra difficulty in devising a suitable mathematical formulation of the tax design problem.

The literature of taxation is normative one. This normative view is based on “what is best for the community”. Traditional normative taxation literature has concentrated on the analysis of direct and indirect taxation. In direct taxation analysis examine the point of unequal of income taxation and to minimise its welfare losses. Indirect taxation analysis includes commodity taxation which levies on individual consumption of goods and services and some excise and sales taxation.

In the analysis of optimal taxation, attention is paid on “utility” instead of Adam Smith’s one of tax principle is tax payments in proportion to income (ability-to-pay). In addition to the literature is interested deeply in “disincentive effects” of taxation. The traditional analysis of optimal taxation searches for a system of taxation that minimises the excess burden (welfare costs) of taxation. This literature does not offer any information to those politicians and bureaucrat's selfless behaviour upon the tax structure. In addition, traditional analysis of optimal taxation states that there are differences between alternative tax regimes according to social welfare functions. There are several approaches to optimal taxation which derives from political agent's behaviour to tax structure determination in political process.

Optimal taxation analyses have some criticisms. For example, Brennan and Buchannan (1977,255) claim that this analysis is institutionally vacuous, Ricketts (1981,44) concluded that, “the literature on tax policy... is almost exclusively concerned with factors which are entirely missing from models of optimal taxation. There is also an interesting article by Broome (1975). He has criticised that the analysis has neglected several important aspects such as horizontal equity. Criticism on optimal taxation is collected as follows.

- The criticism that the analysis has neglected some aspects such as horizontal equity, evasion, administration and taxpayer preferences between different taxes are largely accurate.

The conclusions of optimal tax analysis could have been reached by intuitive argument without the need for extensive mathematical analysis.

A third criticism has been pointed out by Atkinson and Stiglitz (1980,chp:2). According to them the analysis does not lead to unambiguous policy conclusions. And so results depend on economic relationships about which there is little empirical evidence.

The aim of this paper is to seek out links between traditional optimal taxation analysis (normative optimal taxation) and optimal tax analysis in the political decision-making process (positive optimal taxation).

The next section presents the theoretical framework for normative optimal taxation from the view of first-best and second-best conditions, optimal income and commodity taxation and some critics on traditional optimal taxation. The section 3 examines different approaches on the optimal taxation literature. It calls positive optimal taxation approach and includes public choice approach, political equilibrium approach and rent seeking approach. The section 4 is the conclusion part of the study.

Optimal tax theory comes from “normative tax theory”. Famous Edgeworth (1897) and Pigou (1920) utilitarian approach is extended to include both efficiency and equity arguments in the welfare function. This utilitarian tax theory has basic assumptions; such as,

- The objective of society is to maximise its utility,

- that is done by maximising the utility of individuals,

- utility being cardinally measurable,

- interpersonal comparisons of utility permissible.

The “old” welfare economics developed by Edgeworth, Marshall and Pigou utility was considered cardinally measurable and interpersonally comparable. Whereas the “new” welfare economics have imposed greater restrictions. They claim that only ordinal measures of individual utility are permitted and that these are not interpersonally comparable.

In general, taxes should promote society’s microeconomics goals of allocational efficiency and distributional equity. There is a natural tension, however, between tax policy and the goal of allocational efficiency. Most taxes create distortions misallocate resources thereby creating allocational efficiency. From this point, one goal of normative tax theory is to describe how to design taxes that minimise these distortions for any given of revenue to be collected. The allocational theory of taxation that the guiding principle is Pareto optimality.

Pareto optimal is meaning that is impossible to reallocate resources such that one consumer can be made better of without making at least one other consumer worse-off. In the normative optimal literature that it is well known “first-best Pareto optimal conditions”. “First-best optimal condition” basically means that the government has a sufficient set of policy tools, for whatever problems may exist, to restore the economy to the bliss point on its first-best utility possibilities frontier (Tresch, 1981,46). First-best “bliss point” consists that if taxes promote society’s equity and efficiency norms, that are they support society’s quest for a social welfare maximum.

First-best analysis is really the only way to analyse, ceteris paribus, the particular allocation problems caused by breakdown in the technical assumptions and by market imperfections. Consider, first the role of “lump-sum” redistribution in this regard. Because any non-distorting tax, by definition, is a lump sum tax. It means that efficiency criterion is only achieved with a lump sum tax without distorting anytime in the economy. If lump sum redistribution are possible the allocational agency can ignore the existence of a social welfare function and analyse the externally in the context of the first general equilibrium model. General equilibrium model is specifically designed to find the set of Pareto optimal allocations consistent with society’s first-best utility possibility's frontier given the presence of an externality or any other imperfections.

The allocational view does not have to worry about social welfare. Because it knows that the distributional agency will be designing policies to ensure that social marginal utilities of income will be equalised along the first-best utility possibility's frontier. In this case “normative utility function” written as follows,

Utility (W)= U (X₁,...,X_n, L) (1)

Where x, is the consumption of good from I to n and L is the quantity of labour supplied.

If we are interested in only income taxation, the utility function can be drived as¹;

W=u (Y,L ) (2)

These functions use for quantitative applications. For the more general form (1), a complete demand system must be estimated. When determined the utility function of individual, then can be showed to aggregate utility to form of social welfare.²

Social Welfare= (3)

Where is the utility of individual (or household) h. This equation (3) does not include the distribution. If income distribution takes into account that transforms utility can be drived as follows;

Social Welfare= for 1 (4)

Social Welfare= log for 1. (5)

In function (4) that for =0 it is the same expression (3). So when = 0, there is no concern for inequality. And also these functions do not have disincentive effects in which it is assumed that labour supply is fixed. Actually the recent analysis of optimal taxation includes disincentive affects of taxation.

Heady noted that (Heady, 1993, 22) ...the changes in labour supply will then be used to calculate the change in tax revenue, while the changes in utilities will be used to calculate the change in tax revenue, while the affects changes in utilities will be used to calculate the change in social welfare. And also he added that ...the optimal tax system will be the one where it is impossible to increase social welfare without reducing overall tax revenue.

Relative to the first best optimum, what is the loss in social welfare associated with any given set of distorting taxes? First rigorous analysis of this issue in Hotelling’s article (1938), “The General Welfare in Relation to Problems of Taxation and of Railway and Utility Rates”, And also Harberger interested in this question in two separate articles (1964). “Taxation, Resource Allocation and Welfare” and “The Measurement of Waste”.

First-best models ignore a number of important real world phenomena that the policymaker can not ignore. Second-best optimal condition model is essentially a reaction to these first-best assumptions. Second-best theory recognises that no government redistributes income of a lump sum basis. Taxes are always distorting transfers. Therefore, they have an excess burden on society. However, in the first-best model lump sum taxes do not create an excess burden.³

Second-best analysis is concerned with the efficiency costs of distorting taxation. By the way, transfer payments are automatically included in the analysis because they can always be viewed as negative taxes. Since the second-best analysis many person’s efficiency considerations must be tempered by the relative inequities imposed by alternative distorting taxes so that second-best theory interested in “ability-to-pay” principles.

First-best and second-best analyses are a fairly recent phenomenon, however, it is clear that early ability-to-pay theorists were implicitly assuming first-best environment. Ability-to-pay as a sacrifice principal relates specifically to the goal of distributive equity.⁴

Understanding the trade-off between equity and efficiency in a second-best environment has become more important recently. Assumptions of second-best model as follows;

- consumers are assumed to maximise utility subject to a fixed price budget constraint,

- they have no monopoly or monopsony power,

- producers are typically viewed as perfectly competitive profit maximises.

There have been some studies on the second-best problem in the literature made by Bradford and Rozen (1976), Sandmo (1976). Also, Baumol and Bradford (1970) described this problem. However, this problem’s analytical formulations and solutions have been made by Ramsey (1927). After Ramsey’s study, Pigou examined this problem in his book on Public Finance (1947). However, a few people concerned with this subject in many years. It has been hardly mentioned in public finance textbooks.

Second-best problem which began with article by Lipsey and Lancaster (1956-57)”The General Theory of second-best”. But these writers did not refer the Ramsey-Pigou analyses. In 1970, some scholars had began writing some articles on the problem. These are Baumol and Bradford (1970); Lipsey and Lancaster were specifically interested in the following question; given invariant distortions in some markets, are the first-best Pareto optimal rules for other markets still consistent with social welfare maximisation? The answer, in general, turned out to be “no”.

Diamond and Mirrlees (1971) extended and generalisation Ramsey’s formulation in their notable article. They examine that the government must raise some tax revenue by means of distorting taxation. However the government is otherwise free to vary all price cost margins, exactly as in first analysis.

There are some different costs of taxation in the literature as follows;

First, Excess burden (dead-weight cost) of taxation; Taxes introduce distortions in the allocation of resources. It develops Smith’s point about the impediment of taxation to production, but extends it to include the distortions of consumer choice between goods that are actually produced. There are lot of theoretical and empirical work in economics that have been devoted to analysing and estimating the welfare costs (or excess burden) of the variety of taxes in existence.

Second, Administrative costs; it covers the burden to the public sector of administrating taxes. ⁵

Third, Compliance Cost; it describes that “cost incurred by taxpayers or third parties notably businesses, in meeting the requirements laid on them by a given tax structure”(Sandford, 1989,xi). Compliance costs have some form of the costs such as money costs, time costs and psychic costs (Sandford, 1981,163).

Fourth, Rent Seeking (Lobbying) Costs; Rent Seeking activities are using real resources to capture a pure transfer. And these kinds of activities cause some losses on society’s welfare. When the social cost of rent seeking activities are measured that welfare economics becomes a part of rent seeking theory.

Fifth, Uncertainty Costs; Taxes have coercive power on the society. When governments change in the elections then expected income of an individual may differ from her/his current income (Collins and Jones, l998, 176).

Optimal taxation literature always regarded with firs cost “excess burden of taxation”.

As well known taxes, clearly, transfer spending power from the taxpayer to the government. In addition, these transfers of resources, taxes may distort consumer’s choices between factors, and so impose an additional burden on the taxpaying community.⁶ Excess burden may be analysed further by looking at the effects of imposing a specific tax on a single commodity X. Suppose that the conditions of supply and demand for X are described in Figure-1.

Figure-1. Dead-weight Cost of Taxation

If there are no external effects the market works perfectly. Supply curve SS will reflect the social opportunity cost of producing X. Demand line DD indicates the benefits received by individuals from consuming X. If the market works perfectly, the level of output will move towards an equilibrium point of Q₁. At this point the marginal cost of producing X is just equal to the marginal benefit of consuming it.

P₂ AE P₁ shows a consumer surplus which is benefit (as shown by the demand curve DD) received from the consumption of units of X. P₁ EB P₃ indicates a producer-surplus which is the price ( P₃ ) that the producer receives lees the cost of production.

Now suppose that a tax of value t is imposed on every unit of X produced. As a result, the tax increases the cost of producing X by an amount t, and so the supply curve shifts upwards to S_tS_t. The market price consequently rises to P₂. However, the supply, who hands over the tax, only keeps an amount P₃ per unit which is the market price P₂ minus the tax t. Following the rise in the market price to P₂, the equilibrium level of output falls from Q₁ to Q₂.

In this case the area of P₃P₂ AB shows the revenue paid by the taxpayers and received by the government is the times the number of units sold. As a result of the price rises the consumers are worse-off by an amount P₁P₂ AC. Government receives only the area of P₁P₂ AE. In addition, AEC is a net loss of consumer surplus. Similarly, ECB is a net loss of producer's surplus. Indeed the excess burden of the tax is the aria of ABC.

As shown in Figure-1 the cost of taxation is distributed between the producers and consumers. And all taxes have some effects on the allocation of resources in nearly all circumstances. The only tax which could be claimed was neutral with respect to the working of the price mechanism is a lump-sum tax on each person.⁷ The concept of a lump sum tax will be useful in order to isolate certain characteristic of the other taxes.

Optimal tax depends on the gains in social welfare from redistributing income and the costs associated with taxation.

- The cost of income redistribution is the effect that the tax has on the supply of labour.

- The higher the marginal rate of tax, the greater the dead-weight losses created in the labour market.

The problem then is to determine a tax schedule which minimises these costs for any income distribution objective. While the optimal income tax literature is quite technical, however, it can be explained without technical framework. The first analysis of income taxation has been done by Mirrlees (1971). Mirrlees model has two basic assumptions:

1) The only disincentive effect of taxation is on the number of hours supplied by each worker, and

2) Differences between the wages of different workers are produced by differences in their fixed productivities.

Mirrlees separates the optimal income taxes with two sides;

-An optimal non-linear income taxation,

- An optimal linear income taxation

An important distinguishes between these sides are that the marginal rates are different in the first one, marginal rates are constant in the second one. Mirrlees model is that the optimal income tax schedule is approximately linear.

Mirrlee's model is the centre in the optimal income taxation literature. However there are several studies which their consequences offer some differences to the basic model. For example Tuomala (1990) and Allen (1982) and Atkinson (1973) have two interesting examples on this issue.

Allen is interested in the effects of dropping the assumption of fixed relative wages by using Mirrlees assumptions. His result that the poorer group will face a positive marginal tax rate while the richer group will face a negative marginal tax rate. In Mirrlees model income distribution at the top level that marginal tax rate equals zero. But Carrot, Heady and Ulph (1983) show this is negative (Heady, 1993, 29). And Heady says that Allen’s model casts some doubt on the desirability of linear income taxation. However, there is insufficient evidence of the degree of tax non-linearity that it would imply in practice.

Atkinson concerns a different type of disincentive. For example he says that when people have high level of education then they have to face high level of costs. Then income tax can have disincentive effects on educational choice. He extended model in the optimal linear income taxation.

Optimal non-linear income taxation faces an “ability-to-pay” argument. In this argument, the marginal rate of income tax “should” increase as income increase. In order to explain non-linear income taxation that it needs to look at the effect of chancing the marginal tax rate at any other incomes. There are three effects of the tax increase on tax revenue and welfare (Heady, 1993,25);

1) The tax payments of people with the increased marginal rate will probably fall,

2) The tax payments of people with income above the range of increase will rise, and

3) The utility levels of both groups affected by the tax increase will fall.

If the net effect of (1) and (2) is negative, there is no extra revenue available to fund an increase in tax allowances and so the increase in the marginal tax rate is clearly not desirable. This is most likely to occur if either effect (1) is large, because of a high compensated elasticity of labour supply, or effect (2) is small, because the number of people above the range of increase is small.

If the net effect of (1) and (2) is positive, the revenue gain from the tax increase must be weighed against the utility loss of effect (3) the utility levels of both groups affected by the tax increase will fall.

Heady notes that the net effect of changes of tax rates on social welfare will depend on four factors (1993,26);

1) The compensated elasticity of labour supply,

2) The degree of concern for inequality,

3) The degree of income inequality,

4) The proportion of the population above the range of the tax increases.

In the literature, the marginal income tax rates for the person with the highest income should be zero. But it has not provided information about what the tax schedule would look like at income levels up to the top level (Collins and Johns, 434,1998).

Figure-2. Marginal rate of tax from high-income individuals.

Source: based on Brown and Jackson (1994)

In figure-2, the top income earner has a zero marginal rate of income tax. The after tax budget line for the high-income earner is E₂4. In this case, marginal rates of tax must fall as income increases over a range up to the high-income earner.

Brown and Jackson (1994), and Heady (1988) note that there are wider problems in modelling taxpayer behaviour such as the problems of dealing with uncertainty and the difficulty of dealing with variation in the income of particular individuals over their life-cycle.

Scholars who study on the optimal income tax schedules concentrate on analysing optimal linear taxation. Because they have found that optimal income tax schedules are approximately linear. It is claimed that a negative income tax is efficient. If government wants to make income transfers from the rich to the poor, it means negative income tax.

In Figure-3, a linear income tax schedule is illustrated which includes a lump sum transfer (negative tax) and, then, a constant rate of tax on income. Tax revenues are a function of a constant (-a) and the marginal rate of tax t.

tax revenues= -a + tY

The value of (-a) implies to the lump sum payment that a government would make to individuals with zero income.

Figure-3. Linear Income Tax

Source; based on Collins and Jones (1998)

Point b would be a break-even point: at this point the taxpayer would pay in tax an amount equal to the lump sum payment already received. To determine the optimal linear income tax, it is necessary to minimise the excess burden associated with achieving a desired redistribution of income.

The most important study is that of Stern’s paper (1976) on “On the Specification of Models of Optimum Income Taxation”. He calculates optimal linear income tax rates and uses simulation analysis. In his analysis, potential efficiency loss is implied by the shape of the indifference map between income and labour captured in the elasticity of substitution. Also a certain amount of revenue(R) is required to cover the purchase of public sector goods and services in some of the simulations.

Stern’s results are shown in Figure-4 for the case when all tax revenue is used for redistribution purposes. In this graph that if there is no preference for equity v=1 and it can be seen that the optimum linear tax rate falls quite sharply as the elasticity of substitution rises. Stern’s higher the elasticity value captures in Figure 4.

Figure-4. Stern’s “trade-off between Income and Labour

Source-Stern(1976), "On the Specification...",pp.161-2

Elasticity of labour supply helps to determine welfare losses. In his analysis the lower is the t rate for any equity view incorporated in the e value. For any elasticity value a higher t value is associated with a greater preference for equity. With Rawlsian preferences, that is if we wish to maximise the welfare of the poorest person, v= - and it can be seen that the optimum tax rate declines much rapidly.

Stern concluded (162) that “...We should emphasise that the study of optimum income taxation is in its infancy, there is much work, empirical and conceptual as well as theoretical to do, and there for all our estimates and calculations must be viewed with circumspection and as attempts to understand the best model currently available rather than prescriptions for policy”.

As we before noted traditional approaches to the public finance focus on a trade-off between the effect of taxes has on different objectives. But we also noted that optimal taxation literature is based on the trade-off between “equity” and “efficiency”. This trade-off is the “keystone” in the optimal taxation literature.

We note that optimal taxation literature concentrates on optimal income taxation and optimal commodity taxation. This literature searches for a system of taxation that minimises the excess burden of income and commodity taxation. They are interested in politicians’ selfless behaviours and bureaucrats' selfless behaviours upon the tax structure. In addition, traditional analysis is that a social welfare function that will command wide support can be found to discriminate between alternative tax regimes.

Optimal taxation literature, however, has some critics from different point of view which derives from political agents behaviours to tax structure determination in political process. As follows;

First, Public Choice Approach:

- Brennan and Buchannan Model,

Second, Political Equilibrium Approach:

- Hettich and Winer Model, and

- Seiglie Model

Third, Rent-Seeking Approach:

- Lee and Tollison Model.

Public choice approach is final critique of the literature. Public choice perspective is completely different normative optimal taxation literature. Public choice scholar's interpretation on optimal taxation comes from political process. They are not interested in what is “best” for the community in contrast, they are regarded with what is the “best” for actors in the political process and how a tax system can constrain the self-seeking behaviour of those who get to design it (Collins and Jones,l998,374-78).

The public choice analysis of taxation is based on the model of Brennan and Buchanan. It is called “Revenue Maximising Government” or “Leviathan Government”(Brennan and Buchanan, l977,255-73)⁸. Their approach to optimal taxation is the point of view that government is a leviathan which tries to maximise their revenue. Government seeks to maximise revenue raising to increase public expenditure. But this does not mean that government has only taxation for its expenditure. Also government has budget deficit that allows government to spend without raising taxes. Of course, it causes “the inflationary impact upon the future.”

Brennan and Buchanan critic neo-classical normative tax theory the aspect of institutionally. They claim these theories do not include institutional matters in optimal theory analysis. They noted (255) that as follows:

“...normative criteria for a “good” tax system are derived in response to the problem posed by the requirement to raise some exogenously determined amount of revenue for governmental use within a single time period. Emphasis is placed on the familiar efficiency and equity characteristics of alternative tax instruments.”

One assumption on the model that median voter or his representative in a legislative assembly demands some public goods in post-constitutional or in-period budgetary decisions. Their concern is not to tell to governments how they should tax and how the taxing power should be utilised. They offer a positive analysis that governments how behave and how they may be predicted to behave. And bring to some limitations on revenue-maximising behaviour.

Their approach in Leviathan Government Model has constitutional framework that brings some constraints to revenue-maximising government. Tax institution and revenue demands placed on these institutions in the same time.

Figure-5. Income Yielding Activity

Figure 5 indicates the individual’s demand for income yielding activity. MR is a “marginal revenue” curve and the maximum revenue is given the marginal cost line (p$1). Y1 determines a post-tax equilibrium.

Buchanan and Brennan call their model that is a “monopoly theory of government”. The revenue-maximising tax rate, t*, can be derived as follows (264);

(1) R= t Y₁

(2) Y₀ – Y₁= Y₀ . h t, since h =

Where R is government revenue; t is the proportional tax rate and equals

h t), (3)

(4)

Setting (4) at zero, we have

t*= (5)

and, substituting tin (3), we have,

R*= (6)

They conclude that “maximum revenue is directly related to the initial size of the taxable base, and inversely related to the value of the elasticity”. It indicates that many progressive rate structures generate more excess burden than the equi-yield proportional tax. They used a method of an “equi-revenue” technique of comparison. Because, these equi-revenue comparisons involve separate base-rate combinations that yield the same “maximum revenue”. Buchanan and Brennan tried to apply excess-burden criteria and permitted the application of the equal-revenue methodology appropriately derived from the “constitutional perspective”.

They ask this question which is very important in their book “the power to tax”. “What sort of tax institutions would we expect the rational citizen-taxpayer to select as elements in the constitution?” Their focus on study is on the power to tax and on constrains on that power. In their model, individuals have a large protected domain over which the Leviathan himself has no influence, Although they have to take into account the governments objective function in devising the appropriate constrains to his power. Indeed the primary aim of the citizenry in designing the optimal constitution is to define their protected domain in the most comprehensive terms.⁹

Finally they suggested that the tax base and tax rate constraints imposed on governments at the constitutional level should, ideally, be such as to allow for the financing of some roughly efficient bundle of public goods and services. (Buchanan and Brennan, 271, 1977)

Their model based on the “political cost upon the taxation”(Hettich and Winer,l984,67-87). They follow Downs’ suggestion that maximising political agents while choose as optimal a tax structure that minimises the political costs (net loss in voters at the next election) of raising a given tax revenue.

“Tax Structure” covers the composition or pattern of public revenues and the division of such revenues among different tax sources. In general, a change in government thus leads to a change in the use of instruments, since politicians of different parties have differing ideological preferences for any given tax source or type of expenditure.

But Hettich and Winer’s approach is different from this approach. Their approach includes a general objective applying to all agents, regardless of political affiliation. Hettich and Winer develop five testable hypotheses in their analysis (71-74).

-The effective tax price;

H 1- Opposition to the use of a tax depends on effective (rather than nominal) tax prices. The government will use available opportunities to lower the affective tax prices of voters most likely to offer political opposition.

- Organisation cost;

H 2- The more revenue that is raised per dollar of potential tax base with a particular tax, the more political opposition (in total and at the margin) there will be to that tax.

- Opposition Effectiveness;

H 3- Opposition or political costs to the government growth at an increasing rate as revenue collection per dollar of potential base is raised.

- Competing Fiscal Jurisdictions;

H 4- The tax pattern incompetent political units serve as a constraint on fiscal structure.

- Tax-base Certainty;

H 5- Other things being equal, tax sources more subjects to fluctuations generate greater political opposition (in total and at the margin). Governments will try to reduce uncertainty about the yield of the revenue structure.

Hettich and Winer analysis on tax structure in which the composition of revenues is determined endogenously by the decisions of self-interested political agents. However the model is not determined by the characteristics of one decisive voter and any special interest groups.

As well known, in the analysis of public expenditure are suggested the nature of the fiscal process by the median voter model. In Hettich and Winer’s model comes from a single median voter. These different groups of voters most likely to offer opposition to a rise in their average tax rates.

This approach close to Laffer’s view on the relationship between tax rate and tax revenue (Laffer, 1981,1-22). In Laffer framework that each “short-run” curve would be drawn for a given fiscal structure and would show the relationship between average tax rates and total tax revenue. As average tax rates rise, the fall in tax revenue for a given increase in average tax rates along any “short-run” Laffer curve would be greater than the fall in revenue along the “long-run” Laffer curve which incorporates the political cost-minimising changes in tax structure that in the long-run would a company increases in average tax rates.

Seiglie (1990) develops an approach of the politically optimal tax where tax rates are endogenous and determined by force in the political market. His model extends on a commodity tax. This tax is on alcoholic beverages' taxes between states. Seiglie adopted this model with using cross-section data for 50 U. S. States.

The political behaviour in regulatory process plays the role the political process in the view of redistributing wealth. This approach is pointed out by Stigler (1971) and extended by Peltzman (1976) and Becker (1983; 1985). In Seigli’s model that an excise tax is the endogenous and determined by political factors. This kind of tax has two offsetting political impacts;

First, revenue can finance public expenditure and it provides gaining support from the beneficiaries of these expenditures.

Second, the incidence of the excise tax alters the support forthcoming from the affected groups.

Actually, optimal commodity taxation literature tries to balance these two opposing concerns. It is not based on political distributional considerations. But Seiglie develops a political distributional weight model for politically optimal commodity taxation (587). He pointed out some propositions;

1. if there are no externalities, the tax rate will never be set so as to place the consumption of a commodity in the elastic portion of the tax revenue curve;
2. the ownership structure of the alcoholic beverage industry will affect the level of the tax rate, more specifically, states with government liquor monopolies will tend to have higher tax rates;
3. The tax rate moves inversely with other demand or supply reducing regulations, e.g. the minimum drinking age.

Seiglie tries to apply these and other propositions to this model.

X= T.Y. (1)

In here, X is a public good determined more specifically by these legislative choices. T is an excise tax, Y is a private good like an alcoholic beverages.

In this equality, he assumes that no distortions exist in the market for goods, which are substitutes or complements to Y., and also he assumes that no administrative costs in collecting and distributing tax revenues. If there are some regulatory process from the government as follows:

Y= Y ( T, I⁰ , R⁰ ) (2)

The equilibrium quantity consumed of the commodity depends upon the tax rate, T, real income, I⁰ and a vector of regulatory parameters, R⁰, which affect the demand and/or supply of the commodity. And also it is decreasing in the excise tax and other regulations.

In figure-7, TR⁰ is tax revenue curve. It allows a maximum level of provision of G (G’) with the tax rate (T’). After 1933 in USA each state decided on the supply arrangement for the wholesale and retail industries. Every state started to determine the licensing of outlets or their placement under the state’s control. A “licence” state, using excise taxes and licence fees to raise revenue from private licensed suppliers of alcohol.

Since the government assumes the function of both wholesaler and retailer of liquor in control states, one would expect that the legislature’s support function does not include the wholesaler’s and retailer’s interest. In addition, if the liquor market and the political influence of consumers and those affected by the externality is the same in either type of state at any given T, the slopes of the control state’s support curves are less steep. Therefore it is tangent to the revenue curve at a higher T (597). In control states the absence of wholesalers and retailers with political power to wield has the effect of flattening so that a higher T and G is optimal at point 2.lowering the minimum drinking age shifts TR₀ to TR₀ as the from any given T rises. The optimal T and G will tend to rise as equilibrium is moved from point 1 on S₀ to point 3 on S₁. Lowering the legal drinking age increases PS and CS (for previous non-illegal consumers), allowing some gains to be taxed away so as to simultaneously increase G (Collins – Jones, 1998, 404).

Seiglie have got some empirical results. According to him, an average tax rate incorporates excise taxes, licence fees and the monopoly tax, which all them are serving to decrease consumers’ and/or producer’s surplus. Therefore, a higher tax rate results in higher revenues an implication of the model if legislators operate in the inelastic section of the revenue curve. In addition, the effect of income on the average tax rate on liquor is insignificant. And also, he found that interest group is positive and insignificant thing in this model. Finally, Seiglie adopted this model with using cross-section data for 50 U. S. States and achieved that some econometric support for his propositions that tax rates are either endogenous or determined by political factors.

Some government programs and policies cause to reduce wealth through rent creation. Rent seeking behaviour impact on the wealth of society. Rent seeking is damaging for several reasons( Murphy and others, 1995,506)1: First, increasing rent seeking activities occur to absorb labour and other resources and so reduce income. These effect of rent seeking has been often seen in Less Development Countries government bureaucracy. Second; growing rent seeking sector causes that the ability of enterprises are lower and also the rate of technological progress and of growth is likely to be lower.1¹⁰

Rent seeking approach belongs to the “public choice” school. Neo-classical economists have measurement “dead-weight losses (excess burden)” that come from pubic and private provision of goods and services. To neo-classical economists that the government has play a role an omniscient and benevolent in the society. However, public choice economists establish politically self-interest behaviour in the society. Political groups may organise lobby on government and bureaucracy for legislative changes or for licenses that would increase their future income stream.

Rent seeking analysis has firstly been made by Gordon Tullock which his well known article in 1967 “ The Welfare Cost of Tariffs, Monopoly and Theft”. First, Tullock demonstrates the orthodox measure of the welfare costs of monopoly. He concentrated on estimates of value of the Harberger’s welfare triangles. But he demonstrated that welfare loss from rent seeking is more than Harberger’s demonstrated. Second, Rent-seeking activities are growing when government restrictions are increasing. Anna Krueger showed this ties in her well-known article (1974) “The Political Economy of the Rent Seeking Society “Also she described these activities such as “rent seeking”. Although Tullock describes these activities before her but he did not give a name like “rent seeking”. There are a lot of contributions to rent seeking literature. Recently rent seeking has so wide literature. Especially Robert Tollison, Roger Conggleton and Charles Rowley make so important contribute to this lierature.¹¹

Rent seeking analysis claim that an economy’s resources can be wasted in special-interest political activity. And this analysis is considered with welfare losses from rent seeking activities. G. Tullock’s insight that expenditures made to capture an artificial created transfer represent a social waste suggested that the cost to the economy of monopoly and regulation is great than the simple Harberger (1954) dead-weight loss. Under Tullock’s original formulation and in the extensions of his work by Krueger (1974) and Posner (1975), rents are exactly dissipated at the social level. So that the total welfare loss from such activities is equal to the Harberger triangle plus the rectangle of monopoly profits.

Lee and Tollison contribute rent seeking with a model of excise taxation. According to them rent seeking includes in a model of excise taxation. We can explain their approach to the “optimal excise taxation” in third steps.

First, They present an optimal taxation model with the assumption that the supply curve is horizontal.

Second, They examined that the implications of rent seeking for the optimal tax pattern: the case where marginal costs are increasing in output is developed. And the efficiency implications of excise taxation and rent seeking in a both a competitive and monopoly environment are considered.

Third, Their model is in a very simple partial equilibrium setting. They think that this kind of equilibrium very suitable for their purpose which the relation between optimal taxation and demand elasticity rests on naive view of political behaviour.

They claim that does not exist between the tax and price elasticity of demand. In traditional analysis of optimal taxation exists, contrast, this relation between them. When rent seeking cost adds to the exceed burden of taxes cost becomes more than estimates. They offer firstly a conventional model on excise tax;

-They assume that n goods are subject to an excise tax.

They found the tax elasticity of demand for good i as follows;

Ei= -

then which expressed it as;

Ei =

l , is independent of i and represents the marginal cost, in terms of excess burden, of increasing tax revenue. Condition (1) requires that the tax elasticity of demand be the same for all goods. They arrived a rule, which is very-well known as the Ramsey Rule. This rule includes, in general, inelastically demanded commodities should more taxation than elastically demanded commodities.

In addition, Lee and Tollison examine social cost of taxation and rent seeking. Rent seeking cost is important because it causes waste of most of sources, which are use by political activities for to reduce burden of taxes or to prevent it from being made heavier. If rent seeking is exist then the cost of taxation is altered by this existence. Rent seeking increases the marginal social cost of excise taxation on commodities as to confound the traditional results (Lee and Tollison, 349)

They demonstrated that if commodity has more elastic demand it should taxed more heavily than inelastic demand commodities. They put rent seeking cost (Yi) to conventional model (1). As follows;

Ei

Lee and Tollison arbitrarily assumed that the marginal cost associated with the optimal taxation of all n commodities is.35. which yields a value of .60 for l in the rent-seeking model. It is further assumed that,

Y₁=.1

Y₂=.5

A reasonable estimated for the l (cost of the excess burden) is .25. Lee and Tollison uses some parameters value which are made in some studies. As follows; (These studies estimated on the marginal excess burden cost of commodity taxation).

Browning (1976); estimated the marginal excess burden cost of excise taxes in the United States, at .26.

Champbell (1975); estimated of the cost of commodity taxation in Canada, at .24.

Ballard-Shoven and Whalley (1985); estimated the marginal excise burden cost of consumer sales taxes in the US, at .256.

Lee and Tollison obtain for 1 and 2 commodities as follows;

E₁= .3125. and,

E₂= .0625.

It means T₁ reduce the quantity demanded of commodity 1 by 31.25 percent of the post-tax quantity. T₂ reduce the quantity demanded of commodity 2 by 6.25 percent of the post-tax quantity. These post-tax quantities are shown in figure-8 at Q₁ and Q₂ respectively.

T₁ and T₂ are taxes on commodities 1 and 2(in post-tax quantities) In here T₁ > T₂ or the optimal pattern of taxes requires the lowers tax on the relatively inelastically demanded tax base (344).

As mentioned above well known idea in traditional optimal taxation that excise taxes should be imposed on inelastically rather than elastically demanded products may not be true in Tollison and Lee’s analysis. Their result that excess burden plus rent-seeking cost of an excise tax will be greater when applied to a product with an inelastic demand curve than to a product with an elastic demand curve (348).

To conclude this survey, we have seen that the optimal taxation theories have two different approaches; the traditional approach and political decision-making process approach. Traditional optimal taxation theories neglected evasion, administration and horizontal equity, finally the analysis of social cost of taxation. In this literature politically decision-making process (politicians, bureaucrats, voters, interest and pressure groups etc. selfishly behaviours) have been ignored. Political decision –making process approach brings an opportunity for to understand easily different tax regimes. We need more searches on this new literature.