PUBLIC GOODS AN PUBLICLY PROVIDE PRIVATE GOODS




PUBLIC GOODS AN PUBLICLY PROVIDE PRIVATE GOODS
  INTODUCTION

This Lecture deals with the public provision of goods and services. We are concerned with four basic questions:
1.      How do we characterize those goods that are, or ought to be provided publicly?
2.      If the government knew the preference of all members of society, how ought the supply of each of the public goods to be determined?
3.      How are the supplies of public goods in fact determined, and how does this contrast with the optimal provision?
4.      How can the government ascertain the preferences of the members of society regarding the provision of public goods?

The questions are considerable importance and have generated a great deal of controversy. There are those who claim that the government is engaged in supplying goods that ought to be privately marketed, for instance, that education ought to be privately rather than publicly provided. There are others who claim that public programmes receive insufficient funds and that there are activities at present privately supplied that ought to be provided by government. What we shall have to say here does not resolve these controversies, but we believe that a careful consideration the kinds of issue treated in this Lecture will have focus the debate.
At the outset we need to make an important distinction, between public production and public provision. The two are often confused, though both logically and in practice they are distinct. The government provides for the national Defence, yet much of the production of the goods purchased for national defence is within the private sector. The government has, in many countries, amonopoly of the mail service, yet in charges for the use of mail a manner little different from that of private enterprise. In the previeus Lecture we dealt with the pricing of publicly produced commodities; here are concerned with goods and services that are provided freely, perhaps a rationed amounts, to all members of society. (we are also at this stage concerned with public goods are discussed in the next Lecture) .

Characteristics of Publicly Provided Goods

The free provision of goods may be seen as the limiting case of subsidization. i.e., the delivery to consumers of commodities at price below the cost of production. In this sense, the analysis of this Lecture, and that of public sector pricing, are aspect of the same subject. There is however a distinct feature of public provision which approach does not capture and which is focus of much of our discussion: with public provision there is not necessary any monitoring of usage, whereas with any price, positive or negative, usage must be recorded.
            The issue of monitoring usage introduces the first aspect that is relevant to characterizing those goods that are or ought to be publicly provided: it may be impossible, or extremely costly, to charge for the use of specified commodity. In other words, it may not be possible to exclude non-contributors. This is essentially a technical question, and depends on the available technology. In the case of television, calculation of the extent of use depends on it being possible to determine from outside whether the receiver is in operation or no the employment of scrambling devices. It has been suggested that automatic metering devices could be installed to record the passage of vehicle through the highways system and that with large scale computer networks it would be feasible to charge for actual usage. For some goods, such as national defence, it is hard to imagine that even future developments in information processing will allow individual benefit to be determined; so that for these exclusion is indeed impossible.[1]
            Where exclusion is not technically impossible it may still be decided to supply the good publicly, for reasons parallel to those discussed earlier in other contexts. The first is that it may not be desirable on efficiency grounds to use prices to govern the usage of commodity. The effects of charging depend on (1) the conditions of demand and (2) the conditions on which the good can be supplied to additional individual. If the demand is high inelastic, there is no efficiency loss from not charging for the commodity (although there may be other arguments, such as raising revenue, as we have seen). Many place do not charge for the quantity water used, because it is judged that the benefits for metering would be relatively small, demands not being very elastic, and insufficient to warrar the installation of metering devices. (there may also be external economic in consumption-at least, that was an important historical reason for public provision).
Standard discussions tend in effect to focus on the second aspect- that usage by one person does not reduce the amount that other can consumer. In the words, the cost of supplying a fixed quantity to another individual is zero. Examples typically given include television programmes (my listening to a TV programme transmited over the airwaves does no deract from other listening); information (my knowing something does no deract from others knowing the same thing); and national defence. These are extreme cases, and are referred to as pure public goods, where “each individual’s consumption” (Samuelson, 1954, p. 387). More generally, there is arrange of commodities that have the property that an increase in the person’s consumption (keeping aggregate expenditure on the commodity constant) may not decrease the consumption of the other people by the same amount. If one person travels on a little-used highway, the benefits of the road to others are reduced only slightly.
On this view, private goods are at one extreme of a spectrum, where an increase of one unit in the consumption by Mr X reduces the consumption available to the others by one unit; and pure public goods at the other extreme, where an increase in Mr X’s consumption leads to no reduction for others. These polar cases are sometimes characterized in the following way. Let
 be the consumption by household h of the ith commodity. Then for private goods,
                                                                                                  (16-1)

Where Xi is the aggregate supply. In contrast, for a pure public good,
                                              
                                               all h                                               (16-2)

It may be noted that his assumes no free disposal. For many public goods, such as defence, this may not be an unreasonable assumption; on the other hand, for goods such as television, free disposal is possible, and (16-2) should be replaced by
                                               all h                                             (16-2’)

The intermediated case are somewhat harder to characterize, and various approaches have been suggested in the literature. One is to write the consumption possibility frontier for economy as being for good i:

                                                                             (16-2)
with


Private good
Pure Public good
450

Consumption of good i by household k

Figure  Public and private goods.

These are illustrated in fig, 16-1 and the reader is invited to consider how intermediate cases can be handled. An alternative approach is in term of consumption externalities (as discussed in Lecture 14), and this has been developed by Samuelson (1969). In this case the purchase of good i by household h may enter the utility function of the other individuals.
            In both cases, we have a problem of defining what it is that is being consumed, and how it is to be measured. For instance, for television and radio broadcasts, the obvious unit to measure consumption is “programmes listened to”. In this case, the first approach seems more natural. On the other hand, if individuals privately purchase protective services (e.g., police guards), utility may be a function of the level of “safety” in the community which may be function of the aggregate expenditure on protective services, as well as on the private level of protection. Individuals, in providing protection for themselves (and thus lowering the return to crime), are providing a public good (safety), and the consumption externalities representation seem natural. This problem can however be reformulated in terms of our first approach, although one must be careful how this is done. For instance, if P represents the total number of policemen available, and Ph household h, then  = P, and police appear to be a private good, yielding consumption externalities. If however what is consumed (negatively) is the expected number of crimes suffered by household h, denoted by Ch, then we have a consumption possibilities curve

                                               all h                                               (16-4)

Where an increase in the number of policemen reduces the crimes committed.
            The third set of reasons for public provision relates to distributional objectives. This may stem either from a general distributional goal, for example embodied in a social welfare function, or from principles of specific egalitarianism as discussed in Lecture 11. Thus, distributional reasons are probably the primary rationale for the public provision of education-either because it reduces inequality of endowments, or because access to at least a minimum level of education is an objective in itself this commodity an optimal non linear price function. For certain goods, that function may have the characteristic that no price is charged for consumption below a specified minimum.
            We have tried to bring out some of the features that characterize goods that may be publicly provided. In determining whether or not they are supplied in this way, the various factors are likely to be of differing importance. In table 16-1 we have listed some of the goods that are commonly, but not necessary universally, publicly provided. In each case, one can ask whether exclusion is feasible (at reasonable cost), what are the properties of demand, what are the cost of supplying to the individual, and whether there are likely to be distributional arguments. For the first six, we have suggested our own judgement; the reader may like to consider how far he agrees, and to complete the remainder.
            In what follows, we concentrate particularly on the cases that are at the extreme ends of the spectrum for the cost of indivisual spply. In sections 16-2 and 16-3, we consider the provision of the pure public goods; in 16-4 we take the opposite extreme of publicly provided private goods. These sections are concerned with the arguments regarding the optimum level of provision, and-in the case of publicy provided private goods-its allocation among individuals, on the assumption that the government has full information about individual preferences and endowments. The actual procedures by which public spending decision may be effected, and preferences revealed, are the subject of Section 16-5 and 16-6.
Table  Characteristic of publicy supplied goods
Table  Experimental evidence on willingness to pay

Costly
exclusion
Demand irresponsive
Low cost of individual supply
Distributional arguments

?
?
?
?
National defence
Roads and bridges
TV and radio
Education
Water
police
Yes
Yes
Yes ?


Yes
Yes



Yes

Yes
Yes ?
Yes


Yes



Yes
Yes ?
Yes

Medical care
Fire protection
Legal system – criminal case
                      – civil cases
Leverage and rubbish
National park





16-2 OPTIMUM PROVISION OF PURE PUBLIC GOODS-EFFICIENCY
In this section we consider the optimum level of provision of a single, pure public good, consumed in quantity G by everyone. There is an aggregate production relationship :
                                             F (X,G) = 0                                                 (16-5)
Where X denotes the vector of total private good production.

Firs-Best Allocation
The goverment of fully controlled economy is assumed to choose the level of G, and the allocation of private Xh to household h (where h = 1…….H) to maximize an individualistic social welfare function.[2] If the individual utility function is Uh(Xh.G). then the social welfare function ma be written as
                                              [U 1,…, U h,…, U H]                                (16-6)
where  is assumed to be a twice differentiable, concave function an to be increasing in all arguments. If we form the Langrangean
                                                 =  λF (X, G)                                          (16-7)
the first-order conditions are
                                              (X,G) = 0                                           (16-8a)
                                             F (X,G) = 0                                                 (16-8b)
The condition (16-8a) yields the standard first-best welfare conditions (equality of marginal rates of substitution and transformation). The new condition is (16-8b).
From (16-8a) we can see that (i.e., the left-hand side is the same for all h). we can then divide the hth term in the sum on the left-hand side of (16-8b) by giving

This is the basic condition for the optimum supply of public goods : the sum of the marginal  rates of substitution between the public good (and some private good) must equal the marginal rate of transformation (∑MRS = MRT). There is a clear intuitive interpretation of these conditions for a full optimum. The marginal benefit of an extra unit of a public goods is the benefit that person 1 gets, plus the benefit that person to gets, etc. in contrast, an extra unit of a private good is either given to person 1 or given to person 2.
The solution may be illustrated diagrammatically foe the case where there are two individuals and two goods (X = private good, G = pure public good). Figure 16-2 shows in the upper part in the indifference curves for citizen I and the production constraint AB. Suppose we fix citizen I on the indifference curve UI. the possibilities the citizen II are shown in the lower part of Fig. 16-2 by CD (the different between AB and UI). Clearly, pareto efficiency requires the marginal rate of substitution of the second individual be equal to the slope of the curve CD (i.e., at point E). but this is just the difference between the marginal rate of transformation (the slope of the production possibilities schedule) and the marginal rate of substitution of the first individual (the slope of his indifference curve). Thus, we have
MRS11 = MRT-MRS1


A
B
Public good
D
C
Public good
XII
UI
G
XI







Figure 16-2 optimum provision of public goods-two-person example.
MRS1 + MRS11 = MRT
The sum of the marginal rates of substitution must equal the marginal rate of transformation.[3]
The analysis so far has been conducted in terms of a fully controlled economy. It is however equivalent  to the situation in a competitive economy where the government is able to levy first-best lump-sum taxes, both to finance the expenditure and to redistribute income. As in earlier Lectures, we need to ask aht happens when first-base taxation is not possible. In the remainder of this section, we consider the efficiency aspect, taking for this purpose the case where individuals are all identical ; in the next section, we take up the issue of redistribution.

Financing of Public Goods by Distortinary Taxation
When the public expenditure is financed by taxes that generate an excess burden, it appears likely on intuitive groundsthat the rule of equating ∑MRS with MRT will lead to too high a level of spending. As it was put by Pigou,
The raising of an additional £ of revenue... inflicts indirect damage on the taxpayers as a body over and above the loss they suffer in actual money payment. Where there is indirect damage, it ought to be added to the direct loss of satisfaction involved in the withdrawal of the marginal unit of resources by taxation, before this is balanced againts the satisfaction yielded by the marginal expenditure. [Piqou, 1947, pp. 33-4]
Piqou’s intuitive argument is not, however, necessaril correct. In order to explore this, let us take the case of two private goods-consumption (X) and labour (L)- and one public good. We take leisure (= minus labour) as the numeraire, and dotonate the producers price of the consumption good by p, that of the public good by pG. For convenience, we assume a linear production constraint :

If all individuals are identical, and are treated identically, this can be written

(where X , L now detonate the individual level of consumption).
In order the examine the effect of different methods of financing, we assume that the public good is financed partly by uniform lump-sum tax T on all individuals and partly by a spesific tax at rate t on the consumption good. The individual budget constraint is therefore (there is no profit income)

And the first-order conditions for individual utility maximization,

Where detonates the private marginal utility of income. From these we can have the individual demand and labour supply functions of, p, t, T, and G. The goverment aims to maximaize welfare measured by HU, subject to production constraint. The Langrangean can therefore be written :

The necessary conditions for optimalty involve

From this it follows that goverment expenditure should be carried to the fint where :[4]
The left-hand side represent the sum of the marginal rates of subtitution between G and the numeraire good (leisure), while on the right-hand side p G correspond to the marginal rate of transformation.
From this expression we can see that the existance of indirect taxes modifies the conventional ∑MRS = MRS formula in two ways :
1. To the extent that an a increase in G leads to an increase in the consumption of taxed goods (), this reduces the revenue to be raised (throught the term ). The right-hand side is therefore lower than with the conventional formula, or vice-versa . If, for the example, the provision of a further television channel increases the demand for television sets, and these are subject to an indirect tax, it may be socially optimal to carry provision to a point where the sum of the marginal rates of substitution is less then the marginal rate of transformation, even though the expenditure has to be financed by distorionary taxation.
2. The conventional formula is based on the assumption that raising $1 extra revenue would have a social cost equal to the marginal utility of income. However, where there are non-lump-sum taxes this is no longer true. The social cost of raising $1 () may in fact be greater or less than the private marginal utility of income ()
The intuition behind these result is that the goverment whises to the marginal rates of subtitution equal to the marginal economic rate of transformation (as in earlier Lectures). With taxes that are not lump-sum, the marginal economic rate of transformation is in general different from the marginal physical rate of transformation. The different arises from the fact that, when there is distortionary taxation, the changes taxes required to raise he extra revenue to finance the addition to public expenditure affect the deadweight loss (Stiglitz and Dasgupta, 1971). The relationship between and may be seen from the condition for the choice of t :

Using the fact that , we obtain

X, L
(the second step following from differentiating the individual budget constraint). Where t = 0, we have to allow for the effect on revenue (R) of the change in X. Substituting into (16-16), we obtain

Exercise 16-1 Carry out the same analysis where there ia a tax on wage income and no indirect tax. What differences are there in the result and how can they be explained? (See Atkinson and Stern, 1974)

Comparison with Lump-sum Taxation

The analysis so far has considered the effect of non-lump-sum taxation on the ∑MRS = MRT rule ; it is important to emphasize that he result do not tell us anything about the optimum level of provision for public goods-whether the optimim provision in the case of distortinary taxation is larger or samller than where lump-sum taxation can be employed. One cannot in general make deductons from the first-order conditions about the behaviour of the optimum quantities-a point that is often confused. (For example, the form of the first-order conditions depends on the choice of the untaxed god, but this has no implications for the optimum level of G.)
In order to investigate how the optimal quantity of the public good may be affected by the method of financing, we assume that the utility function is additively separable between “private utility” u (X,L), and the public good :
U = u (X,L) + g (G)                            (16-20)
Where g’ > 0, g” < 0 and u is strictly concave. The goverment constraint is

U+ g (G) = constant
G
Lump-sum
(t = 0)
Distortionary
(T = 0)
Q









Figure 16-3 Provision of pblic goods with distortionary
Wen by : [5]
H (tX + T) = p G G                  (16-21)
In the case t = 0 (lump-sum financing), we can tarce out the transformation curve between u (X,L) and G, with slope given by (see Fig. 16-3).[6] The optimum level of provision is found by maximizing the welfare of a representative individual, which with contours as shown in Fig. 16-3 gives the point . the concativity of u implies that is a declining function of lump-sum income, and hence that the frontier is concave to the origin.
Let us now consider the case ot the indirect tax, with T = 0. The level of private utility is given by
And the goverment budget constraint
HtX = pGG                  (16-23)
The slope of the transformation frontier is therefore

This frontier is illustrated by the curve nearer the origin in Fig. 16-3, althought it should be noted that there is no necessary reason why it should be concave.
Optimality again requires a tangency between the social welfare function and the transformation curve. The slope of the social welfare function is –g’ and is a function simply of G. If the distortionary tax tarnsformation curve is steeper than the no-distortionary tax transformation curve, then this implies that at the level of G that was optimal with lump-sum taxation, the distortionary tarnsformation curve cuts the indifference curve from below ; i.e., optimality requires a smaller level of production of public goods. This situation is illustrated in Fig. 16-3. However, while “on average” the distortionary curve is steeper, and hence there may be a presumption that expenditure wil be reduced, it is not necessarily so and global result cannot be deduced. At the same time, sufficient conditions can be given for the level of G with indirect taxation to be lower than that with lump-sum taxation, for example, for small levels of t and G. (N.B.: the tranformation fontiers have identical slope at G = 0.) it is also possible to establish that a small reduction in the possibilities for lump-sum taxation from the first-best optimum (t = 0) leads to a fall in the optimim quantity of the public good (Atkinson and Stern, 1974, p. 124).

Exercise 16-2 for the Cobb-Douglas utility function

Describe the transformation frontiers with t = 0 and T = 0. What conclusions can be drawn about the optimum quantity of G in the two situations?

16-3 OPTIMUM PROVISION OF PURE PUBLIC GOODS-DISTRIBUTION

In this section we examine how the conditions for the optimum supply of public goods are influenced by distributional considerations, paying particular attention to situations in which there are restriction on the set of feasible taxes.

Redistribution and Non-Distortionary Taxation
In the previous section we derived the firs-best allocation rule ∑MRS = MRT, where the optimum could be attained b the use of lump-sum from taxes and transfers. Typically, the government does not enjoy complete freedom in its choice of lump-sum taxes, and indeed we have earlier argued that these may be restricted to a uniform poll tax or subsidy.  Where this is, the ∑MRS = MRT condition ia no longer necessarily applicable. To see is, let us suppose that the government can levy tax Th on household h, where Mh is the (fixed) income. There is oneprivate good (quantity h = Mh- Th) and one public good (G). the government chooses G and X to maximize

Subject to

And a set of restriction on feasible tax rates.
The solution depends on the nature of the restrictions. If Th can be varied freely, the first-order condition imply
Which imply ∑MRS = MRT, the result used earlier. If government is constrained to set Th = T all h, it follows that T = pGG/H, and the first-order condition may be seen to be

Or the social marginal utility of income, we obtain

Where is the mean value, in other words, the appropriate measure of benefits is wighted sum of marginal rates of substitution, the wheigts being proportional to the social marginal utility of income and summing to unity.
This corresponds to the distributional wheight sometimes used in cost-benefit analysis (see, for example, Weisbrod, 1968). An alternative way of writing the rule is

Where cov [A,B] detones the covariance between A and B. the ∑MRS = MRT, rule is therefore modified by taking account of the covariance between the social marginal utility of income and the marginal rate of substitution. As in the earlier optimal tas discussion, this may be written in thrms of the “distributional characteristic”, (see Arnott, 1978) :

Where
If  falls with Mh, this means that for public goods that are valued more highly by the poor than by others the level of provision will be taken to a point where ∑MRS is less than MRT, (as before, one cannot draw conclusion about quantities from the first-order conditions.)
Distortionary Taxation and Redistribution
The taxes considered above are not distortionary ; we now consider the conbined implications of dead loss and distributional objective. For this purpose, we take the case where there are two private goods (consumption, X, and labour, L) and one public good. Individuals have identical utility function, but differ in their wage rate, detonated by wh.
The government is assumed to determine the level of indirect taxation, t, and public goods, G, to maximize [V(t,T,G,w)], where V denotes the vector of indirect utility functions, and T denotes a uniform lump-sum tax (for the present assumed to be zero). We form the Langrangean

Where the backet gives the revenue constraint.
The firs-order conditions are : [7]

Where denotes the mean consumption. Rearranging, this gives :

Is the ditributional characteristic for the private good. The ∑MRS = MRT rule has therefore to be modified for the distributional effect of public goods (on left-hand side) and for the indirect tax (distibutional characteristic on right-hand side), in addition to the corrections for distortionary taxation (compare Eq. (16-19)). If provision of the public good is more progressive than consumption of the private good, in the sense tahat , this raises the relative weighting of the benefit side.
Suppose now that the goverment can levy a uniform poll tax-as with the linear income tax. There is the further first-order condition :

Using this to subtitute for , the right-hand side of (16-34) becomes
The marginal economic rate of transformation exceeds pG where and consumption is a normal good.
16-4 PUBLICLY PROVIDED PRIVATE GOODS
In this section we consider goods tahat, as far as cost of supply to an individual are concerned, are exactly like private goods, but are publicy provided. The examples most commonly given, such as education and medical care (in some countries), may not strictly have these properties (e,g., because of externalities). However, just as we considered an idealized version of public goods in the preceding sections, so here we tahe the pure case of a private good supplied at zero charge in a specified quantity. We begin with the situation where all individuals are identical and there is uniform provision of the good ; we then exted the analysis to that where pepole differ, but the goverment is again required to provide an identical allocation t each individual. Finally, we consider the case where the goverment can provide for different individuals a different quantity of the good in question. We assume troughout that the good cannot be traded ; the reader should consider the impliction of this assumption.
Uniform Public Provision
With identical individuals, and financing via a poll tax (T), the optimum  public provision of the private good, detoned by E, must satisfy a first-order condition identical to that implied by individual choice if we had used a price system.[8] Suppose that there is private consumption, X, and labour, L, in addition to the publicy provided good. The price of private consumption is p, that of E is pE, and labour is the numeraire. The goverment maximizes :
HU (X,L,E) 
Subject to T = pE E. The first-order condition is that
HU E = HpE (-UL)                   (16-35)
Or, where MRSh = UE/(-UL)
                                    MRSh = pE                                          (16-36)
This is identical to the first-order condition for invidual utility maximization when faced with a price pE. As one would expect, there is in this case no summation of individual marginal rates of subsitution.
Where the public provision is financed by distortionary taxation, the first-order conditions have to be modified in the sme way as before. The marginal physicak rate of transformation, represented in (16-36) by pE, has to be replaced by the marginal economic rate of transformation. This is left as an exercise.
Exercise 16-3 suppose that the method of financing is an indirect tax on X at rate t. Show that the first-order condition for public provision is

What conclusions can be drawn abaout the optimum public spply compared with that in a parivate market?
If people differ, then uniform public provision is closer to a public good in the sense taht we have to determine a single level for all individuals. Suppose first taht the goverment has complete freedom in the use of lump-sum taxes, Th, on individual h. The goverment’s problem may be formulated in terms of the Langrangean :
The first-order condition may be written :


Since
Be average MRS should be equated to the MRT. Clearly, the success of public provision depends on he extent of divergence of tastes for consumption about this average. Where there is not full freedom to vary Th, when there is a distributional adjusment ; for example, with = T all h, where is an additional covarience term, cov , as in the discussion public goods.
Optimum Allocation of Publicly Provided Private Goods
The model just discussed, the good is provide uniformly ; we need however to ask whether this is social optimal. This is a question taht is frequently debated. Should educational resources be concentrated more heavily on gifted children, should the goverment provide aqual inputs, or should it aim to ensure equal outputs (e.g, as measured by earning ability)?
There were free use of lump-sum redistributive taxes and transfers, when-in an otherwise firs-best world-the criterion would be straight-forward : we allocate the good according to private demand. In other words, we would not have a uniform level, absed on an average MRS, but would allocate so that individual MRS equal the cost of provision. However, in the absence of such freely flexible lump-sum taxes, the conditions for optimum public provision will, in general, be different from hose arising from private alocation.
The illustrate the way in which public provision may differ, suppose that the social welfare function is the sum of identical individual utility functions which are additive in he publicly provided god. Then the optimal provision is uniform, and unless the tax policy equates the marginal utility of income, this involves differing marginal rates of subtitution across individuals.
To set this out formally, and to see what happes when we relax the assumption of additivity of the utility and welfare functions, supose that individuals differ only in health needs, where . the firs-order conditions for an optimim allocation, Eh, is (for an interior solution)
Where is the the multiplier associated with the total expenditure caonstraint. This model is that examined by Arrow (1971a), who is concerned with the utilitarian objective . he defines a policy as “input-progressive” where > 0 ; i.e., people with greater health needs context-a natural conclusion ; however, from (16041) we can see that, differentiating totally with resoect to (where =1)

If UhEE < 0, the optimum policy is input-progressive if and only if . in other words, resources are allocated according to their effect at the margin. If a worsening of a person’s medical condition makes health care less productive, then the optimal input may decline.
We can also ask wheter the allocation is less than or more than fully compensatory. Since

It follows that an equal input policy is stiil not fully compensatory. In the light of our earlier discussions, it should not come as a surprise to see that a utilitarian policy does not ensure a fully compensatory policy. The role played by the form of thr social welfare function may be seen by alowing for. Differentiating (16-41) and re-arranging then gives
The terms in are both positive, and tend to make it more likely that the optimum policy is input-progressive. (The reader may like to consider the implications of a Rawlsian objective function)

Different Abilities and the Allocation of Education
We noe examine the case where individuals difer in ability, as in earlier Lectures, but where this “raw” ability () is supplemented by education (E) so that the wage of an individual is given by w = E). As in Lectures 12-14, we assume that ability is distributed continuously with density function and this is the only chracteristic in which people differ.
The goverment s assumed to determine the optimum allocation E () and the optimum method of financing. If lump-sum tax, T (), related to quality is not feasible,[9] then the alocation departs from that which would be made on grounds of allocative efficincy. To see this, let us compare the use of the lump-sum tax with a linear income tax, at rate t and such an income gurantee G*. The goverment’s problem may be formuated as maximizing

Subject to

The first-order condition for the choice of E () and T () are (introducing the multiplier for the constraint)

Where V denotes the derivative with respect to the after-tax wage. Since the use of lump-sum taxation, with t = 0, implies

An other words, the first-order conditions imply that education should be allocated so that its amrginal contribution to wages (weighted by L) is equalized. (A full treatment should consider the form of the function (E) and possible non-convexities.)
Where, however, there are restriction on lump-sum taxation, this “efficiency” condition for the allocation of education ceases to aplly. This can be seen from (16-47a). Where we can no longer simplify by using (16-47b) and . In this situation one cannot in general treat the allocation of publicly provided goods separatelly from the redistibutive tax policy, even when the latter parameters (t and G*) are optimally chosen. (The reader may like to check that this is also true with a nonlinear income tax-see (Ulph, 1977). In determining in allocation, one has to balance te fact that-with the assumption made here-the more able can use education more efectively, againts the redistributive factors embodied in , taking account of the contribution to goverment revenue.

Choice Between Public and Private Provision.
Many goods can be alocated either publicly or privately ( i.e., they can be supplied without charge, or they can be allocated on the private market); moreover, there are cases where there are private goods (e.g., security agencies) that are a close substitute for public goods (plice). Thus, the required services can be supplied in either a public or a private mode, and we have to consider which gods are to be provided publicly.
The problem may be seen in general terms as that of designing the optimal price schedule, and the way in which it is likely to work may be describe heuristically. Where public provision of a uniform quantity, E*, is vertically (i.e., there is an infinite price of further units). Where public provision of unlimited quanities is desirable, then the optimal price is zero troughout. Where private provision is preferable, then the optimal price of the first unit of public provision is infinite. In this way, we can see the relation to the earlier analysis of the optimal indirect tax and public sector price schedule.
There are two features of this general problem that should be noted. First, the solution may depend sensitively on the range of instrument at the goverment’s disposal-as we have stressed earlier. Thus, consider the finding that a uniform allocation of he publicly provided goods is socially optimal where the welfare function is utilitarian and the individual utility function additive (page 499). This appears to be in conflict with the earlier result (Lecture 14) that, where individuals differ in wages, and the utility function is locally separable in labour and commodities, the optimal price schedule is linear for commodities. However, we did not allow for the choice between public and private provision-it was assumed that the good was supplied by the goverment.
The second feature of the problem is that non-convexities are likely to be most important. We have seen in the discussion of optimum taxaxion that there is no reason to expect the maximization problem to be well behaved, and in the present case there are further reasons to expect serious non-convexities. A clear example is provided by administrative costs. There are good reasons to suppose that these are likely to be less in certain fields with public provision, for example, because of economies  of scale or because monitoring of individual usage is not required. It may well be necessary therefore to make global comparisons, reliance on local optimally conditions not being sufficient.
In the comparison of public and private provisin, several factors are likely to be influential. In addition to the administrative costs just mentioned, diversity of tastes and distributional objectives play a major role. To illustrate the effect of tastes, suppose that individual differ in their preferences (detonated by ) concerning the good, but not otherwise, and that utility functions have the spesial form (as in Weitzman, 1977)

Where Xh is the (composite) consumption of other goods, taken as the numeraire. With private suply of the good at its production price, p, the demand by household h is

And the resultant level of utility is (up to a constant) :

The good is supplied publicly in uniform quantity E, financed by  a pool tax, the level of utility is (again dropping the constant)

As we have seen earlier, E is optimally chosen according to the average of the marginal rates of subtituton (in this case ccording to (16-50) with . The provision of a uniform quantity involves a loss of total utility, Which may be calculated (16-52) and (16-52), with . the loss
Which can be shown to reduce to H times the variance of . the loss in efficiency of allocation from public provision is therefore governed by the variance of the taste parameter.
In making the comparison between private suply and uniform public supply, the efficiency loss has to be balanced againts any advantage in administrative costs on the side of public supply (and distributional actors). Suppose, for example, that the costs of administration are less in the public sector to the extent of units of the good per person ; then the condition for private supply to be superior to public (with uniform provision) is that :[10]

Where denotes . a sufficient conditions for this to hold is that the coeficient of variation in the taste parameter is greater than. On the other hand, if we consider quantities at zero price (e.g., with water supply) can consume unlimited poll tax, then the condition for private supply to be
Superior is that :[11]

In this case, differences in tastes are provided for, and the efficincy loss arises on account of consumption in excess of that demanded at price equal to marginal costs. Where this excess consumption is relatively small ( is small) it requires only a small advantage in terms of administrative costs for public provision to be justified.
Teh full-scale analysis of the choice between private and pblic provision in a general model is beyond the scope of these Lectures. The example of education may however serve to illustrate how the problem may be formulated in more general terms. People are supplied with a uniform level of public education (which may be zero), and may purchase private education in addition (they are not alternatives as assumed in Lecture 10). The goverment can impose a tx on the private purchase of education in, addition to an optimal linear income tax. They key question for policy are whether the uniform level of public provision is strictly positive and wheter or not the tax on private education should be ser at a prohibitive level. The working out of the details of this analysis are left as an exercise, but it can be shown that, even with an optimally chosen income tax, free public provision of education may be desirable on distributional grounds. (this result can be demonstrated from local properties, but for a fuller characterization of the optimum, care needs to be taken to allow for possible non, convexities.)
Exercise 16-4 Consider the model set out on pages 500-501 and show how it can be extended to include individual purchase of additional education, so that a person’s outlay on education is  where is the public provision, te is the rate education, and . the level of public provision is assumed niform fao all. The goverment can vary the tax rates te, t (on income) and the poll subsidy, G*. Formulate the maximazation problem and write down the first-order conditions under which a stricly positive level of state provision is desirable. Show tht under certain conditions it is optimal to levy a tax on education. In wht circumstances should this tax be prohibitive?
The argument discussed above are not intended to capture all important features of the case for public provision of education. It has for example been assumed implicitly that an individual has no constraints on the access to resources to finance education, whereas in practice capital markets are less than perfect, and parental wealth may be a major determinant of access to eduaction. The scope for the expression of individual preferences and the extent of response of the school authoritiee may be important. These , and other broader considerations, need to be borne in mind in assesing the case for the public provision of educatation; and similar issues arise with other publicly provided private goods such as medical care.

16-5 EQUILIBRIUM LEVELS OF PUBLIC EXPENDITURE

The preceding sections have been concerned with the optimum provision of pure public, or publicly provide private, goods; we now ask how this compares with the levels of provision likely to emerge from actual procedures for determining public spending. We consider three classes of models. In the firts, we ask, what would be the equilibrium supply if there were no government? This is sometimes reffered to as a ‘’subscription equilibrium’’ , where each individual voluntarily contributies the amount he wishes to pay. In the second set of ‘’political’’ models, we discuss the level of public expenditures that arises in a voting equilibrium. Is the public budget likely to be too small in a democracy?cthe third model is known as the ‘’Lindahl’’ equilibrium. This represents an attempt to devise a mechanism that attains a Pareto –Efficient allocation of public goods and has certain formal similarities to the competitive equilibrium model. We confine our attention to the case of pure public goods.

No- Government Equilibrium

Even in the absence of government, individuals may contribute to public goods; and indeed, in actual economies, with substantial public provisions, there is still extensive private support for such items as medicine, education and research. The motives for this support are varied, and are perhaps not adequately captured by the kind of individualistic utility functions we have positied for most of the analysis. But it is worthwhile analysing what the economy might look like with such individualistic functions in the absence of government provision of the public good.
            We take the conventional approach of analysing the Cournot-Nash equilibrium of the economy: each individual takes the others’ supply of the public good as given, independent of his own purchase. For convenience, we assume that there is a single public good, which can be produced at constant marginal cost, PG . There is a single private good, X, taken as the numeraire of which iindividual h has endowment Mh. The individual purchases  Gh of the public good, chosen to maximize


Each individual treats the sum (over i ≠h as fixed; we then obtain the first-order condition :


In other words, he determines his expenditure such that his own MRS is equal to the marginal rate of transformation. The aggreagate MRS is therefore H times PG, and hence greater than PG at this level of public goods.
The solution is illustrated in Fig. 16-4. The upper part shows the choice of Gh given ∑i≠h  Gi;i.e., the person chooses S on QR. By varying the sum, we can generate tthe reaction curve indicated. If we now assume that all individuals are identical, then it is a condition of overall equilibrium that X=M-PG G/H. The interactions P with the reaction curve in the lower part of the diagram gives the Nash equilibrium. It is clear that the individual indifference curve  cuts the line X=M-PG/H at P, and that the maximization of social welfare (with lump-sum taxation) indicates aa higher level of G. The condition ∑MRS= MRT is in fact satisfied at P*.
The comparison of the optimum with the no-government solution is less straightforward where individuas differ and where the government cannot levy freely variable lump-sum taxes. In fact case, the level of public expenditure in the optimum depends on the social weights associated with different groups. It is clearly possible  that the Nash equilibrium leads to greater spending if thr welfare-maximizing solution attaches particular weight to people who do not like the public good iin question. On the other hand, it is possible to make some statements. Suppose that the government provides a equqntity G of the public good, financed by a uniform poll tax. The level of social welfare is given by


Differenting with respect to G, and evaluating at G=0, it may be seen that social welfare is locally increasing in the level of public provision, where (after some rearrangement)



(we have used the fact that MRSh = PG  at the Nash equilibrium, and have divided by pG). The left –hand side is positive, where an increase in G increases the total rpovision. Where this is the case, a sufficient condition for social welfare to be locally increasing in government spending is that the more’’deserving’’ (according to h) individuals reduce the private provision by more.

Nash
Reaction
curve
G
Xh
Mh
Gh
Q
A
S
R
Nash
Reaction curve
G
R
X
M
X = M- pG G/H
P*
P


















Figure 16-4 Nash equilibrium for public goods

Voting over Public Goods

The determination of public spending by means of voting was disccused in lecture 10, where we drew attention to a number of the problems that arise. When there is a single decision variable then resctrictions on preferences of individuals such as
 how resonable such assumptions are(the example was given there a public and private education, where theses are alternatives), and once move to two or more dimensions the conditions on preferences necessary to ensure a voting equilibrium are extremely restrictive. On the other hand determine outcome may be ensured if there are exogenously determined rules governing the agenda.
            Here we simply make the comparison between the level of expanditure determined by majority voting under conditions in which and equilibrium exist. In particular, we assume that there is a single decision—that the concerning the level of a public good financed by a uniform poll tax. It is assumed that individuals votev ‘’sincerely’’(see Lecture 10). The utility of the h-individuals is
                          (Mh


And, as we saw in Lecture 10, the individual valuation of different levels of G is a single-peaked function. The votinf equilibrium is therefore characterized by the preferred level of a median voter (denoted by m) :
           



Does the voting outcome lead to a higher or lower level of G than is socially optimal? Suppose that we consider a small increase in G above the level determined by the median voter. The effect on social welfare is given by (this tells us whether or not social or not social welfare is locally increased by an increased by an increase in spending above the amount in the voting equilibrium)


It is clear that there is no presumption that this it either positive or negative. If the government has no distributional preferences, the ∑MRS may be greater or less than PG at the median voter equiibrium, as is illustrated by Exercise 16-5. If  ¥h. UhX  differs across individuals, then whether (16-62) is positive or negative depends on the weights attached to different individuals and their position relative to the median. If, for example, the preferred level of government spending is a strictly declining function of income, and the government is concerned solely with the welfare of the lowes income group, then the social optimum involves a higher level of public spending than in voting equilibrium.    
Exercise 16-5 for the special case of the utility function


Show that the social optimum may involved more or less government speending than voting equilibrium where the government has  no distribuional preferences does it make if the public good is financed by a proportional tax oon imcome? (See Stiglitz, 1974a.)
            The analysis just given should be treated simply as a counter-example to the proposition that majority voting leads to a level of government spending that is below( conversly baove) the social optimum. The example shows that there is no persumption either way. In order to reach more positive conclusions,it is ncessary to specify more fully the political machinery and procedures lyiing behind public spending dicisions. This mould need to make account of the conditions that ensure a determinate outcoome, and role played by legislators ad bereaucrats in addition to that of voters.

 Lindah Equilibrium
The inefficiency of the Nash equilibrium aries because each consumers i s faced with a price equal to that of the public good, whereas some of the benefit accrues to others. By anology with the case of comsumption externalities, we can seek therefore a set of corrective subsidies, whic will in general have to vary across individuals-we need ‘’personalized’price. This procedure was discussed by Samuelson (1969) as a ‘’pseudo-demand algorithm’’ to calculate the optimal level of public goods supply, but was proposed as an actual allocation process by Lindahl (1919)
            The essence of the Lindahl procedure is that individuals ‘’demand’’ a total quantity of public goods on the basic of a specified distributions of the tax burden (see Johansen, 1965,Ch.6). thus each individual faces a tax share h of the expenditure , where ∑h h= 1, and these tax share erform the function of personalized prices. Reffered to as ‘’Lindahl prices everyone demands the same level of each public good. In the case of a single public and a single private good, individual h maximizes.
 The Lindahl equilibrium satisfies the necessary condition for a Pareto- officient of public goods in full conditions.

D1
G
D2
Lindahl equilibrium









Figure 16-5 Lindahl equilibrium : two person example.
The  Lindahl equilibrium is illustrated in Fig. 16-5 for the case where there are two (types of) individuals. The share of the first is denoted by and the  demand is shown by D 1the share of the second is given by 1-15 and the demand is shown by D2 The intersection is the Lindahl equilibrium.
            The general properties of the Lindahl equilibrium have been extensive discussed in the literature. These include the Pareto efficiency reffered above, and converse of this result that, under certain conditions, every Pareto-efficient allocation can  be generated by a Lindahl equilibrium with suitable lump-sum taxes and transfers.[12] A particular questions that have recieved considerable attention is the relationship between Lindahl equlibria and the core. An allocation (of goods among individuals) is to be in the core if no coalition of iindividuals can together propose an alternative allocation of its own resources that make at least one member better off and no member worse off- they cnnot in this sense improve upon the allocation.[13] For a two –good, two person economy, the core is simply the set of Pareto-efficient points that represent and improvement for both individuals over their no trade position. It is shown. In the standard. Edgeworth  box diagram, by the points on the contract curve between the indifference curves through the inditial endowment.
            In an example economy with no public goods,complete maekets and full information, there are two basic theorems concerning the relationship between the core and the competitive economy :
1.      The competitive economy is contained in the core;
2.      The core ‘’shrinks’’ to the competitive economy as the number of traders increase.
 The first proposition is trivial. Since the competitive economy represents an improvement for all individuals(or at least no dcrease in welfare) relative to the no trade point, and since the competitive economy is Pareto-efficient, it is clearly contained in the score. The second proposition may be illustrated as follows. Suppose that there are two types of individual, but a large number of each type. The first step is to show that any allocation in the core must be symmetric;cie,. Everyone of the same type gets the same comsumption bundle.[14] The second step is to show that, if we can repliciate the economy by any arbitrary factor, then  the only possible core allocations are the competitive equilibria. For this, we need only to consider points on the contact curve. Suppose that we consider a point such as Q in Fig. 16-6, which is not a competitive equilibrium. A line from the initial endowment point to Q, inteesects at least one of the indifference curves through Q, and there exist points such as P where advantageous trades can be made. A group consisting of individuals of type I and II in approciate ratio can improve on the allocation at Q, and replication ensures that this is feasible(with integral numbers of individuals).
             When we introduce public goods, the natural parallel result would be for the lindahl equilibrium to belong to the core, and the core to shrink to the save of Lindahl equilibria ask the number of traders increases. The latter would particularly significant, since.
One could them ague that, no matter by what sytem (public) goods actually are allocated, if..... we assume that trade and production will take place among agents as long as it is advantageous, then any allocation that actually arose could have been achieved by the Lindahl price mechanism. (Roberts.1974b,p 38)
 However , although the first result- that any Lindahl equilibrium is in the core-can be demostrated under certain conditions (Milleron.1972), the second result does not hold. For instance, muench (1972) gives an example where the Lindahl equilibrium is unique but the core is very large (see also Milleron, 1972). The reason for this can be seen to lie in the requirement.
Initial
endowment
Indifference
Curves of individual
II
Indifference
Curves of individual
I
Contract
curve
Q
P
A

Figure  Competitive equilibrium and the core
that the coalition must be able to make its members better off irrespective of the actions of the remaining group. In the context of public goods, this means that they must be better off even if the remaining individuals decide to produce no public goods. It thus  becomes difficult for small groups to improve upon the proposed allocation: the core of the economy is likly to be bigger.[15]
Since much of the appeal of the concept of a Lindahl equilibrium, the results just describe reduce the strenght of the claims that can be made. We must therefore reconsider either either the cincept of the core as applied to a public good economy, or the status of the Lindahl equilibrium. In any event the Lindahl equilibrium is probably best regarded as an analytical benchma
RELEVATION PREFERENCES
Throughout the previous sections, we have assumed that the governmnet knows the preferences of individuals (in the voting model) that individuals vote for their ‘’true’’ preferences. This raises two closely related questions: how can we the government learn the preferences of consumers, and how can we be sure that in any actual procedure for determining the provision of public goods,individuals will be have ‘’honestl’’?if we start from a presumption that individuals reveal the truth,unless it is their interest not so to do, then this means examining the icentivies for lying of for providing the government with false information.
            That the demand for public goods may provide people with such incentivies has been recognized for a long time. If the amount an individual has to pay for the public good is related in some way to his demand. As Samuelson expressed it in his classic paper, ;;it is in the selfish iterest of each person to give false signals, to pretend to have less interest in a given collective consumption activity than he really has’’ (1954, pp. 888-9). This is sometimes referred to as the ‘’ free-rider problem’’ ;and it arises in a variety of contexts apart from public goods. Unions claim tthat the reason that all individuals  should be required to contribute dues is that there exsist a free-rider problem. They provide a collective good (negotiating better terms with the management); and any individual disclaiming interest in the good has the advantage of enjoying the benefit without paying the cost. The  general problem is the same as that of incentive compatibility in a (finite) competitive private economy, which we referred in Lecture 11. In that case, the incentive for individuals to misrepresent their preferences disappears as the economy becomes ‘’Large’’. For public goods however , the incentives do not improve as the number of people increases (see Roberts,1976)  and in this respect  there is indeed a contrast between the allocation of public and private goods.[16]

Mechanisems for the Revelation of Preferences
A general class of mechanisms can be describe as follows. The hth individual is asked to report a valuation of public goods, zh (G). The government announces that the tax shares of the hth individual and the supply of public goods will be a function of all statements according to some rule :
           
In designing the mechanism, the government may seek to scure properties such as the following :
1.      Thee Nash equilibrium (where everyone takes the announcements of others as given) is Pareto-efficient where each person chooses his announcement maximize his own welfare.
2.      In the Nash equilibrium, everyone reports truthfully their valuation of public goods(for Pareto efficiency thi is not necessary; all that is required is that the government can’’translate’’ the announced valuations).
3.      Trurhful reporting is a dominant strategy ( thatit, it pays each individual to report zh (G) accurately regardless of the announcement of others).

There  have been a number of attemps to devise mechanisms that have some or all of these propoties. The earliest of these mechanism was that of Vickrey (1961), who developed a procedure for a public marketing agency faced by monopolistic buyers and sellers. He showed that it would be possible to motivate individuals to give correct iinformation by paying themt the net increase in the sum of producer and consumer surpluses of the other persond in the market that resulted from the supply or demand curve revealed. This procedure was then independtly discovered and developed by Clarka(1971,1972) and  Groves (1970.1973; Groves and Loeb, 1975). (See also the discussion of elititation functions in Kurz, 1974)
The procedure may be described in a partial equilibrium model where utility functions are of the form
      Uh =gh(G)+Mh

It is assumed that lump-sum transfer of income Mh can be made freely. The stated valuation function of individual h (note that it is a function and not simply a single value) is zh (G). The level of public provision, G*, is chosen to maximize ∑hzh(G). And individuals are taxed in a lump-sump way according to the schedule

Where the last term is an arbritary function of the vector z, excluding zh. The level of individual utility is


With the procedure, the dominant strategy is for each person to reveal the true marginal valuation. To see this, suppose that is function of G and some variable (so that we can think of his answer as being represented by the choioce of ). A variation in has no direct effect via dG* / d . By the conditions determining the choice of G*, this has no effect on the underlined term in (16-69), and  the variation  is therefore proportional to ghG-z hG. With the optimal choice of , this is zero, so zh (G) must equal gh (G) up to the addition of a constant.[17]
This preference revalation  mechanism has therefore certain attractive properties, and Green and Laffont (1977a) have shown that this is the only class of mechanisms such that stating one’s true preferences is a dominant stretegy and that outcome is Pareto-efficient. It is however limited, both by the assumptions made and by the fact that the mechanism does not guarantee a balanced  budget for the government (on this, see Groves and Ledyard, 1977). It does not allow for collusion between individuals, and coalition incentive compatibility further issues(see Green and Laffont, 1979). Finally, no equity considerations are allowed for.

Empirical Significance of Free-Riding
Preference revalation and incentive compatibility is an active area of research. This is undoubtedly a valuable antidote to much of the earlier literature, which, with axceptions such as Samuelson (1954) and Buchanan(1968), has tended to ignore the problem-as in previous sections of this lecture. On the other hand, there are those who argue that there  is little evidence to suggest that the problem of correct revalation of preference has been of empirical significance:
We have a lot of public goods around probably more than we would expect on the basis of the thory of the free-rider tedency...and there are also many groups and individuals around who by no means appear to conceal their preferences for public goods. (Johansen, 1977,p.148)
There  are two principal reasons for questioning the importance of the free-rider problem . the first is that honesty may itself be a social norm, rather than simply the outcome of maximizing utility :
Economic theory, in this as well as in some others fields, tends to suggest that people are honest only to the extent that they have economic incentivies for being so....the assumption can hardly be true in its most extreme form.(Johansen, 1977,p.148)
In societies where honesty is a social norm, one would not expect misrepresentation of preferences unless the pay-off to dishonesty reaches a threshold level. Where there is too complicated consuming and resort to telling the truth:’’since i cannot find a way to beat  the system, I had just as well tell the truth’’ (Bohm, 1971,p.56). The secomd reason  why the revalation of preferences may be less important is that the decision os not made directly by individuals but typically through elected representatives. Johansen argues that in this case misrepesentation is unlikely to pay, either in terms of eloctoral success or in terms of decisions made by legislative assemblies. This brings us back to some of the issues discussed in Lecture 10.
            There have in fact been experimental studies of individual preference revalation under different incentive schemes. For example, Bohm (1972) carried out an experiment at the Swedish Radio –TV Company, where 211 people were asked to express their willingness to pay to see a new programme, nnot yet shown to the public. They were paid  on arrival 50 Kr (approximately) $10) for taking part, and then asked to specify how much they would contribute to see the programme under a specified  payments sructure. They were told that the programme would be shown  if the total sum stated exceeded the costs (500 Kr). The main results are set out in Table 16-2. Although there are some differences in the means and medians between the different incentive schemes, none of the differences are significant at the 5 per cent level. This experiment is clearly on a small scale, and intended more to assess the feasibility og the method, but ia is none the less interesting that so little diffrence emerges.
Table  Experimental evidence on willingness to pay
Number of
cases
Payment scheme
Amount mean willing to pay  (Kr)


Mean
Median
23
29


29



37
39
(I)  The amount stated by respondent
(II) A percentage of the amount stated (so that collected = total cost)
(III) One of four possibilities determined by a lottery (with equal probabilities)-designed to represent the case of “uncertainty
(IV) Five Kr
(V) Nothing
7.61
8.84


7.29



7.73
8.78
5
7


5



6.50
7








Decentralization and Information
 The models employed in earlier sections asssume that the government  knows not only the preferences of the individuals, but also the production possibilities of all firms. In, fact the government does not have at its disposal all the requisite information and allocation, nor does it have the ability to solve all the problems of production and allocation simultaneously.
            This is one of the main motivations for organizing governments in a decentralized manner, i.e., having branches responsible for different activities os functions. Thus, Musgrave’s (1959) division of the branches of government into the stabilization, allocative and distribution branches  may be thought of as more than just an analytical device. On the other hand, the sense in which the different branches can carry on their business seperarately from one another is not made clear in Musgrave (or in most of the subsequent literature) , and the conditions under which various schemes of decrentralization lead to  a full optimum are not spelled out.
            This may be illustrated by reference  to the provision  of public goods. Suppose that lump-sum taxes may be frelly used, and that public instructed to maximize social welfare subject to the bugget (public goods being charged for at the producer prices). The agency will then equate the ∑MRS to MRT where is the multiplayer associated with the budget constraint, and if us chosen correctly the social optimum is reached. (we are ignoring here the problem of revelation of preferences.) where, however, there are non-lump-sum taxes, we have to allow both for the fact that varying G may affect government revenues and for the distributional  effects of public goods.  There is than fundamental interdependence between decision about the relative quantities supplied of various public goods and the stucture of for the finance of these goods. As a consequence,the marginal  rate of subtitution between two public goods is not in general equal at the optimum to their marginal rate of transformation (the ratio of the producer prices).[18] Moreover, the benefits for public goods need to be weighted according  to the social marginal utility of income,and these weights depend on other aspect of distributional policy.
            There is therefore a persumption that decentralization  will entail certain costs, which have to be balanced against the administrative and informational advantages. The rigorous analysis of this problem, taking account of such factors as the motives of those who administer government programmes and political power, is clearly a major task. In the next Lecture, we consider one particular form of decentralization- where individuals form local communities for the supply of local public goods

READING
Key references on the optimal provosion of pulic goods are the papers by Samuelson (1954,1955,1958b,1969) . a valuable survey of the area is provided by Milleron (1972). On tthe public provision of private goods,see  Arrow (1972a)and the subsequent literature. The discussion of voting on public goods draws on Stiglitz (1974b). A useful review of dynamicc procesess for the provision of  public goods iin provided by Tulkens (1978). The revalation of preferences is dealt with in depth by Green and Laffont (1979), and the  references contained therein.






[1] It should be noted that term “exclusion” is being used in a slightly different sense from for example, that employed by Musgrave. He refers to the exclusion principle as indicating that a person “is the included from the enjoyment of any particular commodity or service unless he is willing to pay the stipulated price” (1959, p. 9). This however reflects a choice about the method by which the good is to be allocated. Our definition relates solely to the technical possibilities.
[2] The modern general equilibrium treatment of the optimum provision of public good lates from Samuelson (1954), he has returned to the subject in Samuelson (1955, 1958b, 1969)
[3] As recognized by Samuelson (1954), his treatment was a general equilibrium version of the earlier partial equilibrium analysis of Lindahl (1919) and Bowen (1943). In that case, the “total demand” is found by adding up the dermand curves ; but unlike private goods, where we add horizontally the total dermand at a given price), for public goods we add vertically (the total amount that all individuals are wailing to pay for the given amount of the public good)
[4] Using the fact that (p + t) obtained from differentiating the individual budget constraint (16-12)
[5] This can be obtained by summing the individual budget constraints (p+t) HX = HL - HT and subtracting the agregate production constrain (16-11)
[6] Define
then
[7] We have used facts that MRSh
[8] Note that we are assuming uniform provision. The reader should consider wheter there can be situation where the goverment would want to allocate different amounts to identical
[9] As in earlier Lectures, this assumes that information may be used by the goverment for certain purpose but not others. It is assumed that a is observed by he education authorities with sufficient accuracy to allocate E, but that this information either is not available to the tax officials or else is not acceptable as a basis for determining tax liabilities. Although a more complete treatment of information would be desirable, we feel that this captures an important feature of the problem.
[10] The utility loss person from the costs of administration is
[11] The welfare loss from public provision is then
[12] On this, see, forexample. Foley (1970) discussion of the existence of Lindahl equilibria. See Milleron (1972) and Roberts (1974b) Reference should also be made tobe planning procedures for public goods, such as those proposed by Malinvaad (1971a. 1971b) and Dreve and de la Vallee Poussin (1971). A useful survey of  work in this area contained in Tulkens (1978)
[13] The reader should note that we have read Shapley (1973) and have resolved not to use the term ‘’blocking’’
[14] if not , the ‘’uderdogs’’ can form a coalition. For an exposition, see Hildenbrand and Kirman (1976.Ch.1) or Varian (1978,p.181-2).
[15] The assumption that the best cost of the public good is totally independent of the size of the economy may be questioned or discussio of the question of ‘’returns to group size’’ an ‘’semi-public goods’’ see Roberts 1974b).
[16] The problem of incentive compatibility may arise quite widely where there is government intervention in the economy. For example, we have seen in Lecture 13 that the first-best redistributive tax many involve utility being a declining function of ability, and that this  would give people an incentive to  misrepresent their ability.
[17] the Clarke procedure has a rather simpler form with the arbitary function being replaced by the valuation of public goods to the remaining H-1 people, at level G chosen to maximize ∑. hzh (G)-pG G. See Tideman and Tullock (1976).
[18] the condition under which such decentralization remains possible despite the existence.

Komentar

Postingan populer dari blog ini

Surat Babtis (Tardidi) di Gereja HKBP

Peta