Revenues Fall—or Increase

Exemption breaks the VAT chain. Whether this increases or decreases the net revenue raised by the VAT depends where in the chain of supply the break occurs. If the exemption occurs immediately prior to final sale, the consequence is a loss of revenue since value added at the final stage escapes tax.

If the exemption occurs at some intermediate stage, on the other hand, the consequence is actually an increase in net revenues: the cascading of tax on inputs means that, as the price charged by downstream firms using the exempt item rises in order to cover their increased costs, so the tax on output downstream increases.⁷⁸ Thus value added prior to the exempt stage is effectively taxed more than once.

Distorted Input Choices

The exemption of items used as inputs into production removes the key feature of the VAT, discussed in Chapter 2, of preserving undistorted the production choices that firms make. The unrecovered taxation of some intermediate inputs that is implied by exemption will induce producers to substitute away from those inputs.

The distortionary consequences of the initial exemption, it should be emphasized, can spread far beyond the sectors most directly affected. Exempting the production of steel, for example, will not only distort the production decisions of machine tool manufacturers using steel products as an input; the consequent impact on the price for machine tool services will distort the prices of tooled products, disadvantaging items and production methods that make intensive use of such inputs. Exemption may thus render the impact of the VAT system far more opaque, with effective rates of VAT—in the sense of the difference between the price at which goods finally sell and the value of the underlying resources used in their production—potentially differing greatly, and in nontransparent ways, from the statutory rates of VAT applied to final output.

This concept of the “effective rate of VAT”—a term that has been used in the literature in rather different senses—is discussed in detail in Appendix II. Even describing the distortions on input choices that exemptions create is far from straightforward, as is shown there. Conceptually, the effective rate of VAT—referred to here as the type-I effective rate, to highlight the focus on input-choice distortions—can be viewed as the excess of the price producers actually pay for a commodity over the price that would prevail (assuming unchanged factor prices) in the absence of the VAT. If there are no exemptions, the type-I effective VAT rate on any commodity is simply the rate of VAT applied to that product. In the presence of exemptions, however, the type-I effective rate reflects the precise pattern of input-output relationships between sectors. Information on these quantities is especially hard to come by in developing countries, so that the calculation of type-I rates will often be problematic. Assessing the welfare cost of the distortions associated with these effective rates is even harder.

Incentive to Self-Supply

By introducing cascading, exemption creates incentives for the avoidance of tax by vertical integration, commonly referred to by VAT practitioners as “self-supply.” To elaborate, exempt traders have an incentive to supply taxable items to themselves rather than purchasing them and incurring irrecoverable VAT. For many exempt items, economies of scale or the specialized nature of the activity may be such that self-supply is hardly feasible; it is services produced by relatively unskilled labor and relatively suited to small scale production that seem most susceptible to self-supply. Banks producing exempt financial services, for example, may find it worthwhile producing security services in-house rather than purchasing them from outside companies that must charge VAT, which the bank cannot recover. While such self-supply mitigates the production efficiency problem associated with exemption, it evidently does so only at some revenue cost.

Final consumers, of course, are an especially important group of exempt persons. Their incentive is to avoid tax on final sales by supplying themselves with some of the value added embodied in final sales: they may escape tax on the value added by painters touching up their woodwork, for example, by doing the work themselves. (By the same token, of course, painter and consumer together have an incentive simply to conceal the transaction, with the painter charging no output tax and refraining from reclaiming input tax.) The essential effect here is the bias that any indirect tax creates for consumers to enjoy their purchasing power in the form of untaxed leisure—broadly interpreted—rather than as paid work.

The incentive to self-supply may be greater than that implied by the statutory VAT rate on final output in the presence of rate differentiation. Assume, for example, that, as in the United Kingdom, food is zero-rated but restaurant meals are taxed at 10 percent. Assume also that a prepared meal uses $0.2 of food and sells at $1 (both amounts net of tax). Since food is zero-rated, the full tax of 10 cents is attributable to the value added in the restaurant; a tax rate on valued added in that sector is not 10 percent but 10/80 = 12.5 percent. Since the net effect is to tax more heavily than might at first seem the value added in sectors whose output is relatively heavily taxed, the incentive for self-supply of such goods is greater than might at first seem. The reverse would also be true.

The measurement of this incentive to self-supply gives rise to a concept of the effective rate of VAT on a commodity ⁸⁰ that is quite distinct from the type-I effective rate above (which focuses on input choice distortions). This alternative concept, referred to here as the type-S effective rate, so as to call to mind “self-supply,” is simply the net revenue paid in respect of the production of some item expressed as a proportion of value added. The type-S effective rate will thus be high for lightly taxed outputs produced from heavily taxed inputs. This notion of the effective rate, and its relation to the type-I effective rate introduced above, is also discussed in Appendix II.

The appropriate policy response to self-supply is not clear-cut. The artificial grouping of activities that results implies that organizational forms are in part being dictated by other than purely commercial considerations, and to that extent is undesirable. Self-supply can itself, however, be a response to the distortion associated with unrecovered input tax, and so may serve a socially useful purpose in mitigating the distortion of input-choices that might otherwise arise.

Perhaps in part as a reflection of this ambiguity, policy responses to the self-supply issue vary. Some revenue authorities reserve powers enabling them to charge tax (and deny credit) on self-supply to avoid nonrecoverable input tax. One test, for example, may be whether the services in question are also available for hire outside the firm. More widely, the question arises to how firms are to be grouped for VAT purposes. The United Kingdom, for example, has recently proposed that companies not eligible to reclaim input VAT not be allowed to join VAT groupings; and grouping rules are under consideration more widely in the EU.

Compromising the Destination Principle

Exemption compromises the destination principle for the taxation of items entering international trade. While exported items that would otherwise be exempted are in practice typically zero-rated, it is not possible to remove the consequences of exemption at an earlier stage in the production chain. In the European context, for example, the Sixth Directive enables banks selling financial services directly to countries outside the EU to reclaim VAT levied on their inputs. But the exports of companies that make use of financial services provided by banks in their own jurisdiction, which are exempt rather than zero-rated, bear, indirectly, unrecovered input tax. To some degree, exemption thus undoes what has often been seen as a principal merit of the VAT.

By the same token, firms using inputs that are exempt have an incentive to import those inputs—which will be zero-rated rather than exempted in the country of export—instead of purchasing tax-laden items from exempt domestic producers. Indeed, there is an incentive for exempt producers to artificially export their output (and so have it zero-rated) in order that domestic producers can escape indirect taxation through the input into the exempted sector. These difficulties have led a number of countries (such as Colombia in 1999) to consider introducing countervailing tariffs so as to eliminate the advantage of imports of exempted items over domestically produced substitutes.

Partially Exempt Traders

Complications arise in respect of traders who sell both taxable and exempt outputs. For recovery purposes their input tax payments must be allocated between the two kinds of sales. This is typically in proportion to the values of the two types of sales, which clearly hold the potential to do rough justice in the presumed allocation of inputs to outputs, and to distort traders’ decisions as to the composition of their sales depending on how such treatment differs from the underlying reality.

Exemption Creep

One of the key features of exemptions is the way in which they feed on one another, giving rise to a process of what might be called “exemption creep.” The point here is distinct from the concern that each exemption provides a general precedent for others. Rather, it is that each exemption creates direct pressures for further exemptions, both upstream and downstream:

• Creating one exemption in order to lighten the tax load on a particular item or group creates pressure for exemption (or zero-rating) of commodities used to produce that exempted item. If the government wishes to remove the burden on that item, lobbyists for upstream suppliers might reasonably argue, surely it should remove the tax on its inputs that would otherwise be unrelieved?

• Creating an exemption at an intermediate step in the chain increases the potential return to downstream users of that input from lobbying to secure their own exemption. To see this, suppose that firm B buys some input from firm A. If A is taxable, then the gain to B⁸⁰ from becoming exempt is that B’s own value added escapes tax. If A is exempt, however, then since output tax levied on B captures value added by both firms, the gain to B from becoming exempt would be that both firms’ value added escape tax.⁸¹

The first kind of creep has been especially apparent in practice, a prime instance being in the agricultural sector: the exemption of basic foodstuffs has in a number of countries bolstered pressure for the exemption of agricultural inputs (the wider issues here being discussed in detail in the next chapter).

Avoidance

The distortions that, as just described, can be induced by exemptions are, in a sense, forms of avoidance. But one response to exemptions may be avoidance of a more transparent kind. De Wit (1995) gives the example of characterizing a lease of property (exempt, suppose) as an agreement for storage of goods (taxable, suppose, and hence preferable for a taxpaying lessee).

Output Is Hard to Tax

In some cases it may seem impractical to apply the VAT to output; either practicalities or revenue needs make exemption preferable to the alternative of zero-rating.

A prime instance concerns the treatment of small traders, where administrative and compliance costs can effectively preclude their inclusion in the VAT system. In this case zero-rating is for the same reason not a realistic option. Standard practice is thus to exempt traders below a certain size. Quite where that line should be drawn is an important issue, and addressed at some length in Chapter 12. For the present, the important point is that exemption—in effect, excluding from tax the value added by such traders—may be an acceptable compromise between the desire to save implementation costs and the desires not to unduly jeopardize revenue and/or favor small traders in competing with large.

A second key instance is that in which output is sold at prices below true market value. The most prominent examples arise in connection with outputs sold by the public sector in competition with private enterprises. State-financed educational institutions, for instance, may provide output at low prices, and in competition with private enterprises subject to VAT. Complete parity between the two could be achieved by zero-rating these services, but may present opportunities for abuse. Though still placing the public sector at an advantage, exemption may be preferable.

There are also sector-specific cases in which difficulties in identifying the appropriate output to tax have been used to justify exemption as a means of ensuring that taxation is not avoided altogether. The leading instance of this is in the taxation of intermediation services, most prominently in respect of financial services. The difficulty here is that the taxable output is in principle the intermediation service. While the aggregate value of this can be inferred from the difference between buying and selling prices, the allocation between buyer and seller—which will matter for the appropriate crediting of the tax—is not immediately revealed by the market.

Exemption as a Less Costly and Administratively Convenient Substitute for a Reduced Rate

When both outputs and inputs can be observed, the alternative to exemption is to subject the commodity to the VAT at a reduced or even zero-rate. If that is possible, and given that exemptions are associated with production inefficiencies, can it ever be the case that granting an exemption would be the preferred course of action? On balance, this is not obvious, even though exemptions do avoid some problems. On the positive side, exemption avoids the administrative difficulties associated with the payment of refunds as a result of zero or reduced rating (discussed further in Chapter 16). On the negative side, exemption does not avoid either of two problems already noted with rate differentiation—namely, that such differentiation can encourage definitional disputes and set the stage for further base erosion. There is, of course, an important sense in which exemption is simpler than rate differentiation, in that it is not necessary to monitor either output tax or the recovery of input tax. This, however, bears on the risk that exemption is used precisely as an opaque device to favor certain commodities or interest groups—multiple nominal rates are in an important sense more transparent than multiple effective rates induced by exemptions.

Public Sector

Standard advice and practice is for commodities provided by publicly owned bodies in competition with private enterprise to be fully taxable, except for items, principally those listed below, which are generally exempted. More contentious is the proper treatment of services provided by the public sector either free of charge to the final user—defense and other classic public goods fall into this category—or more generally on a not-for-profit basis. These services are the focus here. (Essentially the same issues also arise in the treatment of nonprofit bodies and fraternal organizations more generally.)

Policy toward the public sector adopted in the EU appears to be broadly representative of practice in most developed countries. The Sixth Directive effectively exempts noncommercial services provide by public bodies. This indeed is a prominent example of the first rationale for exemption noted above, namely, it is hard to tax output that is given away other than by taxing inputs into its production. In effect, the public sector is regarded as the final consumer of the inputs used to produce services it provides on a noncommercial basis.

Little systematic information is available, however, on the treatment of the noncommercial activities of the public sector in developing countries (no question on this was included in the survey conducted for the IMF study). Practice appears to vary: in some cases, for example, sales to public bodies seem to have been exempted.

Does it make any difference whether the public sector is exempt or fully taxable? Absent an explicit charge on output, with respect to goods and services provided for a fee to members of the public, output tax revenue will be zero even if in principle taxable; and any revenue that the government collects by denying the recovery of input tax on purchase by public bodies is exactly offset by an increased cost to the government of financing those bodies’ activities. Closer inspection reveals, however, that exempting the public sector can have the same adverse effects that are associated with exemption of private activities. Specifically:

• Insofar as the agency does not anticipate that it will be reimbursed for the input tax that it incurs, the effect of exemption on the input prices that it faces will distort production decisions just as it does for profit-maximizing private firms. Public agencies are presumably required, among other things, to minimize the costs associated with their activities: and it is this cost-minimization, rather than profit maximization, that generates a production inefficiency when input prices are distorted.

• For public bodies engaged in both commercial and noncommercial activities, the difficulties of partial exemption arise.

• Self-supply biases also arise: for activities that can be contracted out to the private sector, exemption of the public sector will mitigate in favor of retaining production in the public sector.

• Classification disputes may arise from, for instance, the requirement that public bodies be taxable on activities that do, or might, compete with the private sector.

The difficulties have given rise to various ad hoc responses. One strategy, adopted by Canada and a number of European Union countries, is to rebate VAT payments to public bodies in respect of their exempt activities. In effect, this converts the exemption into a zero-rating. This eliminates both the potential distortion of input choices and the bias against outside contracting. It also eases the problem of distinguishing commercial and other activities: in either case, input tax is recovered, albeit by different administrative means (within the regular VAT system for commercial activities, outside it through the rebating).

A more systematic response is simply to treat public sector bodies as fully taxable in respect of all their activities. This option has attracted increasing attention in recent years. Notably, it has been adopted by New Zealand, and has been proposed in the context of the European Union by Aujean, Jenkins, and Poddar (1999). While raising no new issues of principle in relation to the crediting of input taxes, there is the matter of identifying the taxable sale. This is straightforward in respect of fees charged to users of a commodity. More generally, the logic of this approach is to regard all sums received by the producer that are tied to production, whether as user charges or as subsidy payments, as, in effect, taxable sales. The net revenue raised by levying output tax in this way on subsidy payments received from other branches of government is of course zero: the revenue collected by the supplier exactly matches the increased cost to the deemed purchasing agency. The advantage is that the VAT chain is preserved through the public sector and its interface with the private sector.

The additional question is what tax rate should be applied. This takes us back to the discussion on rate differentiation. Interestingly, Aujean, Jenkins, and Poddar argue that the sensitivity of the goods in question warrant a reduced rate: they mention a rate in the order of 3-5 percent for the EU.

Turning back to practice, the notion that basic services traditionally provided by the public sector on a not-for-profit basis should be exempt from VAT appears to underlie two of the exemptions widely found in both IMF advice and practice: those for education and health.

Education

Standard advice and practice is to exempt basic education services, and tax at some “normal” rate more specialist training provided on a commercial basis.

The appropriate taxation of education is a complex matter. External benefits associated with education mean that it may even be optimal to subsidize some kinds of training. Leaving these arguments aside, however, the focus is on why it might be appropriate to exempt basic education provision. Two features of the education sector are important to this question: it is a significant component of national income; and basic educational services are commonly delivered at zero or subsidized prices, in competition, to some degree, with private producers.

The last feature implies that to subject education to a VAT in the normal way would exacerbate the competitive distortion between private and publicly provided forms. One way of leveling the playing field between public and private providers, at least to some degree, would be by zero-rating. This however raises the difficulties associated with refunds (discussed in Chapter 15). While manageable in many developed countries, making zero-rating an option there—and Australia, for example, does zero-rate education—refunds continue to be problematic in developing countries. Moreover, the first feature (the considerable size of the education sector) makes the revenue cost of zero-rating basic education expenditure considerable.

Where, however, publicly provided education is provided at cost, as, for example, may increasingly come to be the case in the tertiary sector, the case for subjecting education to VAT in the usual way, perhaps at a reduced rate to reflect the growth externalities that increasingly seem to be associated with education expenditure, becomes more powerful.

Health

As with education, it is only the basic services that, typically, are exempted: professional services of registered doctors and dentists, supply of prescribed medicines, and the like. The arguments in this case are very similar to those for education. There may again be externalities from some kind of health care that warrant some degree of subsidization. Again, there would be some logic to zero-rating basic health services: Australia does this, for example, while Uganda apparently zero-rates some medical supplies and the United Kingdom zero-rates aids for the disabled. Revenue cost and administrative difficulties, however, weigh against wide-ranging zero-rating the core supplies of the sector.

As private provision grows relative to public, and the latter moves closer to market pricing, so the extent of properly taxable nonbasic health care services can be expected to grow, though this is likely to be some way off in many developing countries.

Financial Services

The rationale for the commonplace exemption of financial services is different, resting on technical difficulties arising from the nature of value added in financial intermediation. For financial services provided on a fee-paying basis, such as safe-keeping services and financial advice, VAT can be charged in the usual way. The difficulty arises for services charged for in the margin between the return paid to lenders and that charged to borrowers.⁸² Even though the aggregate value added created by intermediation can be identified, to the extent that the financial services are used by registered firms, one needs further to allocate the aggregate value added between the two sides of the transaction. This is difficult.

Suppose, for example, that a bank pays its depositors 5 percent and charges its borrowers 15 percent. Clearly the value added by the bank is 15 - 5 = 10 percent of deposits (less any material inputs, and assuming too there is no risk of default). This should be taxed. If all loans were to final consumers, this would be the end of the matter. Assume instead that the borrower is a registered firm. How much of the 15 should be creditable? The standard conceptual approach on this issue has been to imagine a hypothetical “pure” interest rate at which the lender could have lent (but without enjoying the ancillary services (clearing, etc.) offered by the bank) and at which the borrower could have borrowed (had they been able to find suitable lenders without the help of the intermediary). If this pure rate is 12 percent, for instance, then the value added provided to the borrower is 15 - 12 = 3 percent of the loan, and the remaining 12 - 5 = 7 percent is value added provided to the lender. On a loan of $1,000 and at a VAT rate of 10 percent, the appropriate outcome is thus for the borrower to be charged VAT of $3 and the lender VAT of $7, the total payable thus being 10 percent of the aggregate value added on 10 percent of $1,000. These VAT payments would be creditable, in the usual way, if lender or borrower is registered. It is, however, the difficulty of bringing about this outcome, which would appear to require, in particular, identifying a “pure” interest rate, that has led most countries to exempt financial intermediation.

Some countries (Israel, for example) have instead taxed value added in financial services by the addition method (described in Chapter 2): that is, by levying tax directly on the sum of wages and profits.⁸³ A potential alternative is afforded by the subtraction method (as was at one point proposed in Canada). Either method is capable of taxing aggregate value added in this sector, and indeed either would in principle be perfectly adequate within a wider VAT system based comprehensively on the addition or subtraction method. Neither method, however, sits well with the application of the invoice-credit method in the rest of the VAT system. Neither enables the identification of embodied VAT on a transaction-by-transaction basis, and hence neither allows the systematic crediting of financial services provided to registered traders.

In principle, these difficulties can be circumvented by applying VAT on a “cash flow” basis.⁸⁴ Under this system, all inflows of funds, including the receipt of a loan, and of interest payments, would be treated akin to sales, and be taxable if the recipient is registered; and all outflows, including the repayment of loans, or payment of interest, would attract credit if the payer is registered. For example, consider a loan of $1,000 to a registered trader (at 15 percent) financed by a deposit (paid 5 percent) from a consumer. Assume as before that the tax rate is 10 percent. Under the cash flow VAT:

• The bank is liable to pay $100 on the deposit but this is exactly offset by a credit of $100 on the loan itself. When the loan is repaid, the bank has a net inflow of 10 percent of $1,000 (its spread on the loan) and so owes tax of $10.

• When the loan is made, the business pays tax of $100. When it is repaid, it receives a credit of 10 percent of $1,150 (principal plus interest).

If the government is able to earn the pure rate of interest of 12 percent on its receipt from the business of $100, it is left with net revenue of $112 + 10 - 115 = $7, which is exactly 10 percent of the services of $70 provided to the consumer.⁸⁵

So long as the interest rate available to the government is identified with the pure interest rate, the cash flow approach achieves the theoretically correct allocation of value added, and allows a proper crediting of input tax. Intuitively, by crediting inflows and outflows at the same rate the system ensures that the present value of the revenue raised from registered traders is zero; revenue is ultimately raised only to the extent of that part of the margin which falls on unregistered traders.

The treatment of pure insurance—insurance with no savings element—is straightforward under the cash flow VAT, an example being given in Box 8.1.

Outside the case of pure insurance, however, the cash flow scheme is cumbersome administratively. Some measures can be taken to alleviate this (for example, by suspending the payment of tax, and refunds, associated with the initial receipt of loans and deposits). But even within the EU, where the scheme has been closely considered, there are doubts as to its practicability.⁸⁶ For developing countries, it seems likely to remain overly complex for some time yet.

It should be emphasized too that it is by no means certain that bringing financial services fully into tax will generate an increase in VAT revenues. If financial services were fully taxable, revenue would be collected only on sales to final consumers. When they are exempt, in contrast, tax is collected on all inputs into the sector.⁸⁷ Which will lead to greater revenue is an empirical question. Subjecting financial services to VAT is unlikely to be the fiscal panacea for cash-strapped governments that it may appear to be.

Real Estate and Construction

Real estate is the durable good par excellence. It yields services over more than one period, and it is commonly resold. Thus the issues that arise in its VAT treatment are simply those that apply to all such goods, but writ large.⁸⁸

The ideal treatment of durable goods would be to tax the flow of services in each period, with a corresponding credit if the services are used as a business input. This is easy to do when the services are traded in the marketplace and hence readily valued. Thus the leasing of real estate for commercial purposes is often subject to VAT, with a credit to the lessee (if not in an exempt activity). Where it is exempt, so that a risk of cascading arises insofar as the lessee is a taxpayer, provision is generally made to allow the lessor to opt for registration and payment of the tax (and recovery of input tax).

In many cases, however, real estate services are self-supplied, and so have no observable market value. For services used as business inputs this poses no particular difficulty, as the tax that the enterprise should charge itself on those services would in any event carry an exactly offsetting credit. Owner-occupied housing is problematic, however, because this involves final consumption on which one would like the tax to “stick.” While attempts have been made in the past to impute value to services enjoyed from owner-occupation for the purposes of income tax, the experience has not been a success and is now rarely made. Thus services enjoyed from owner occupation are—with no exception that we know of—exempt from VAT. To avoid distorting the choice between house ownership and renting, the commercial leasing of residential property is commonly also exempt.

There is, however, another way in which services from owner occupation can be taxed. This is by the “prepayment” method, which simply means levying VAT on owner-occupied properties at the time of purchase. Since the value of the property capitalizes the value of future services, the effect of this is to levy tax in advance of the enjoyment of those services. This, implicitly, is the general method applied to durable goods under the VAT. It is now also the generally recommended method for taxing residential properties.

The question that then arises is how to tax resale of property under the prepayment method. In principle, resales should be fully taxed to the purchaser (to capture the future services to be enjoyed by the purchaser) and fully refunded to the seller (to give, in effect, a credit in relation to services taxed at purchase but not enjoyed during their ownership).⁸⁹ Net revenue would be zero. This treatment is generally feasible, and should be adopted, for commercial properties.⁹⁰ Given a possible ownership period of some decades, however, this is unlikely to be practicable for owner occupation; it is simpler—and apparently universal practice—to exempt resale of owner-occupied housing.

Construction services and inputs should generally be fully taxed, and creditable only for those undertaking construction as a business activity. Taxing such inputs acquired by owner-occupiers is a rough and ready way of taxing the enhanced consumption services to which they presumably lead.

Taxing the sale of new residential properties does raise transitional issues at the time the provision is introduced. Unless also applied to first sales of preexisting properties—which, though feasible, apparently has never been done—it confers some windfall gain on owners of that initial stock, who will now be able to sell their properties at prices reflecting the increased VAT-inclusive price of new properties (though their benefit is mitigated by the increased house prices they will face if they choose to purchase another residence). This measure will also be seen as disadvantaging first-time house-buyers, a politically sensitive group in many countries. Thus Australia, for instance, provided some relief for such buyers at the time of introducing its GST. More generally, exemption from VAT for housing services is frequently cited as necessary to reduce regressivity. It is not clear, however, why housing should be favored over other forms of “necessary” consumption. Further, and especially in developing countries, the wealthy are frequently heavy consumers of housing relative to their incomes, as compared to the poor. In any event, if the impact of the VAT is to increase costs of housing for the poor, it would be far more efficient to give them direct subsidies.