5 Collection Costs and the Complexity of VAT

Liam Ebrill, Michael Keen, and Victoria Perry
Published Date:
November 2001
  • ShareShare
Show Summary Details

One charge commonly leveled at the VAT is that it is an intrinsically complex tax, cumbersome to both authorities and the taxpayer and, consequently, ill-suited to developing countries in which familiarity with even basic record-keeping requirements may be low. This chapter considers that argument.

Assessing the Suitability of VAT for Developing Countries

The econometric analyses reported in Chapters 3 and 4 suggest that the VAT performs less well in less developed economies. Care must be exercised, however, before drawing any policy conclusion from this.

The results in Chapter 3 suggested that gains in total revenue associated with the presence of a VAT are greater, all else equal, in more developed economies. More specifically, the analysis in Chapter 4 found, quite robustly, that VAT revenues are higher, all else equal, in countries with higher literacy rates. Insofar as literacy proxies the administrative capacity of taxpayers and tax collectors in such matters as record-keeping, the implication is that the VAT does fare significantly less well where that capacity is weakest. This may reflect both less effective collection of any given VAT system and the impact of administrative capacity on the design of VAT features not controlled for, such as the prevalence of exemptions.

The key question, however, is not whether developing countries gain less from the VAT than do more developed. Nor is it whether the VAT itself performs better where administrative sophistication is greater: that will presumably be true of almost any tax. Rather, the question is whether, at lower levels of development, the VAT fares worse than other taxes theoretically capable of raising similar amounts of revenue.

Administration and Compliance Costs

There are two broad types of resource cost associated with the operation of any tax: administration costs incurred by the tax authorities, and compliance costs incurred by taxpayers. Taken together, they are referred to here as collection costs. Administration costs are akin to a reduction in tax revenues and appropriately viewed as such. Compliance costs, however, should be weighted rather less heavily than administration costs when evaluating the costs and benefits of a tax: the purpose of taxation being to transfer resources from the private to the public sector; $1 left in the private sector is worth less than $1 accruing to the public sector.59

High collection costs are not necessarily a sign of a bad tax, but may simply reflect inefficient or corrupt administration. Nor are low administrative costs (even relative to revenue raised) necessarily a sign of a good one, since one could, for instance, raise substantial revenue under a VAT with low administration costs by assigning few resources to the payment of refunds. Indeed such is the potential role of the VAT as a catalyst for organizational and cultural change both within the tax collection agencies (spearheading, for instance, the modern use of information technology and the development of audit methods) and among taxpayers (developing a culture of record-keeping) that there is an important sense in which a successful VAT is in some circumstances bound to involve substantial collection costs, especially in its early years (see Chapter 12).

Collection cost figures must thus be interpreted with care. The even more fundamental difficulty, however, is obtaining such figures. In terms of administration costs, the organizational structure and reporting practices of tax authorities may make it difficult to isolate costs attributable to the VAT. The exercise may be especially difficult in function-based organizations, where VAT is administered along with other taxes, by the same people in the same offices. Even when a separate VAT department is in place, there may be genuine common costs arising, for instance, from joint auditing for VAT and income tax purposes. In terms of compliance costs, a tailored survey of taxpayers is needed to extract accurate information. These difficulties mean that even for developed countries hard numbers on these matters are relatively difficult to come by.60 For developing countries, it is almost entirely unavailable. It has not been possible, within the resources of this study, to rectify this important gap.

Absent such information, some guidance is nevertheless provided by various studies of VAT collection costs that have been conducted for OECD countries (usefully reviewed by Cnossen (1994).

  • Administration costs for a broadly “best-practice” VAT are put by Cnossen at around $100 per registrant per annum. These costs vary widely, with the general perception that they are significantly lower where the VAT is simpler: they are estimated at around $50 per registrant in New Zealand (with a single nonzero VAT rate) and $200 in the United Kingdom (where there are two rates, and substantial zero-rating).

  • Compliance costs for a best-practice VAT are estimated by Cnossen at around $500 per registrant per annum. For Singapore, Jenkins, and Khadka (1997) estimated continuing compliance costs to be about $700. Since these are largely fixed costs, independent of the amount of tax payable, they fall especially heavily on smaller traders: Cnossen (1994) puts them in the order of 2 percent of turnover for those with turnover below $50,000, falling to about 0.3 percent for those with turnover above $500,000.

These dollar figures will overestimate the corresponding costs in developing countries, since they largely reflect labor costs—including those of the taxpayer. Nevertheless, it is clear that compliance costs in particular can be significant for smaller traders. Suppose the threshold is set at $50,000, for instance, and that traders at this level do indeed incur compliance costs of 2 percent tax on turnover. With a profit margin of 20 percent on sales, this is equivalent to a 10 percent tax on income. This implicit tax may be to some degree passed on to consumers in the form of higher prices, though insofar as small traders naturally tend to have relatively little market power much of the burden will remain with them. There is thus the possibility of some regressive impact. Not surprisingly, small traders are among the most vociferous opponents of the VAT,61 a point reflected in the discussion of the VAT threshold in Chapter 11.

One clear implication of the evidence for developed economies is that the collection costs associated with a VAT are likely to be significantly affected by the design of the tax. A simple structure with a single rate, few exemptions, reliance on self-assessment, and with a high threshold appears conducive to relatively low collection costs.

It should be borne in mind that it is not the collection costs of the VAT in itself that matter, but those costs relative to those of alternative taxes. The weight of evidence for developed countries suggests, for instance, that the VAT is less costly than an income tax. One natural question, however, is whether the VAT is more or less costly than alternative forms of sales tax. The question, in part, underlies the next section.

Is the VAT More Complicated than the Taxes It Has Replaced?

How do the VATs adopted in developing countries compare, in terms of complexity, with the taxes they replaced? Were those preexisting taxes simpler and easier to administer and comply with? To address this, Tables 5.1 and 5.2 describe the key features of the predecessor taxes to the VAT, and the VAT itself, for six Francophone and six Anglophone African countries.

Table 5.1.VAT and Predecessor Taxes in Six Francophone African Countries
BeninBurkina FasoGabon
Pre-VATVAT (1991)Pre-VATVAT (1993)Pre-VATVAT (1995)
TypeTurnover taxes: general; (ICAI); on imports; on tourism. Other indirect taxes on importsInvoice-credit VATTurnover taxInvoice-credit VATTurnover taxes: imports and domestic purchases; domestic comsumption tax; tax on services; tax on importsInvoice-credit VAT
Rates (in percent)ICAI: 25 on production, 29 on services; several rates on others18Three rates18Several rates18
BaseMany exemptions (including retail and capital goods)All activitiesMany exemptions (including retail and capital goods)All activitiesMany exemptions (including retail trade)All activities
Measures to relieve cascadingComplex crediting for ICAI on manufacturingInvoice-credit VATComplex crediting on manufacturingInvoice-credit VATComplex creditingInvoice-credit VAT
Pre-VATVAT (1996)Pre-VATVAT (1995)Pre-VATVAT (1995)
TypeDistinct turnover taxes on production, services and importsInvoice-credit VATDistinct turnover taxes on traders, services, and importsInvoice-credit VATTurnover taxInvoice-credit VAT
Rates (in percent)Several18Three rates for traders (0, 4, 10); two for services (6, 16); five for imports (0, 2, 5, 10, 20)185, 14, 3018
BaseMany exemptions (including retail trade)All activitiesRetail trade exempt; customs duties excluding from baseAll activitiesn.a.n.a.
Measures to relieve cascadingComplex crediting for ICAI on manufacturingInvoice-credit VATComplex crediting on manufacturingInvoice-credit VATInvoice method but: for traders, credit limited on purchases of goods to be sold; no credit for service providersInvoice-credit VAT
Note: Information on the predecessor to the VAT is generally as at the time of the introduction of the VAT; information on the VAT is the most recent available; n.a. indicates not available.
Note: Information on the predecessor to the VAT is generally as at the time of the introduction of the VAT; information on the VAT is the most recent available; n.a. indicates not available.
Table 5.2.VAT and Predecessor Taxes in Six Anglophone African Countries
Pre-VATVAT (1990)Pre-VATVAT1Pre-VATVAT (1989)
TypeManufacturers’ sales tax and tax on manufactured imports (including capital goods)Invoice-credit VATImports taxed unless taxed by South Africa; services taxed selectivelyManufacturers’ level sales tax, applied also to imports (with 20 percent uplift for latter)Invoice-credit VAT
Rates (in percent)4 rates12, 1612 (following South Africa)25 domestic, 30 on imports20, 10
BaseMany exemptionsAll activitiesExemptions for basic foodstuffs, agricultural inputsMany exemptions, including raw materials and intermediate goods (but not “investment goods”)To manufacturing and import stages
Measures to relieve cascadingSpecified exemptions for intermediate goods and raw materials; ring system before 1975, then credit methodInvoice-creditRing system for selected sales between registered traders; imports are in the ringIntermediate exemptions, plus ring system by exemption or creditInvoice-credit VAT
Pre-VATVAT1Pre-VATVAT (1996)Pre-VATVAT (1995)
TypeGeneral sales tax on domestic and imported goodsManufacturers’ sales tax, with uplift of 10 percentInvoice-credit VATManufacturers’ sales tax on specified domestic goods and all imports subject to tariffInvoice-credit VAT
Rates (in percent)5, 10, 20 (alcohol and imported cigarettes)Five different ad valorem rates ranging from 0 to 30; different rates for imports and domestic goods18Six rates for domestic manufactured items (10, 15, 20, 25, 30, 50); imports taxed at 25 with 25 uplift on tariff-inclusive price17.5
BaseExemptions included manufacturing and agricultural inputs, plant and machineryMany inputs exempt, including building materials and other equipmentAll activitiesDomestic manufactured items not on positive list exemptAll activities
Measures to relieve cascadingExemptions aboveExemptions above and ring systemInvoice-credit VATRing systemInvoice-credit VAT
Source: IMF staff compilation.

VAT not yet introduced.

As noted in Chapter 1, there is a semantic issue about when Malawi adopted a VAT. We here take 1989 to be the date of its introduction.

Source: IMF staff compilation.

VAT not yet introduced.

As noted in Chapter 1, there is a semantic issue about when Malawi adopted a VAT. We here take 1989 to be the date of its introduction.

They indicate that, while there are important cross-country differences, the predecessors to the VAT were far from simple and far from being broad-based taxes on consumption. In West Africa, the VAT generally replaced turnover taxes of the kind prevalent in France before the gradual adoption there of the VAT. In East Africa, it generally replaced a manufacturers’ level sales tax on the old Commonwealth model.62 Beyond that basic structural difference, however, there were many similarities across the predecessor taxes: numerous and sometimes narrowly differentiated rates (these frequently served the purpose of luxury excises); many specific exemptions (or, alternatively, positive lists of taxable items); exemption for retail trade; and, in some cases, differential treatment of domestic sales and imports (using the general consumption taxes for domestic protection). A further complicating feature in some of the East African systems was the fact that the tax was imposed at the manufacturers’ level, but upon a constructed price using an “uplift” factor over the manufacturers’ price, rather than the actual sale price. In some cases, the “uplift” price and a hypothetical retail controlled price were alternative bases, depending upon which one was higher.

Especially notable, however, is that all of these predecessors to the VAT included complicated methods for avoiding the cascading of tax inherent in turnover taxes. In the VAT, this role is played by the mechanism for crediting tax on all inputs, and it is frequently this feature that gives rise to the accusations of inappropriate complexity. It is striking, however, that the anticascade features of the preexisting turnover taxes were themselves neither simple, in most cases, nor easily administered. In West African systems, cascading was generally mitigated through a complex subtraction method for inputs: that is, tax was levied on output net of allowable inputs. In the Eastern and Southern African systems, this role was played by a combination of a “ring” system for exemption of the sales of certain items between registered traders only, and exemption of some categories of products viewed as generally being intermediate goods. In many countries, both Francophone and Anglophone, “fixed investments,” or “capital equipment” and “raw materials,” were exempt. Naturally such systems, involving in some cases the categorization of borderline items, and in others, identification of end-users, are difficult to administer correctly.

The complexity of some predecessors to the VAT is quite stunning. In Mauritania it involved three distinct turnover taxes, each levied at multiple rates. Relief against cascading was provided by subtraction. In Uganda, the predecessor tax was a multiple-rate sales tax at the manufacturers’ level on a positive listing of items, which charged domestic products and imports at different rates and provided relief for cascading both by widespread exemption and the operation of a ring system.

The VATs that have replaced these taxes in the 1990s also differ in important respects from one another. But they have important common features: they generally use the credit/invoice method and credit all inputs; have broader bases than the predecessor taxes (even where there are still a questionably large number of exemptions); exempt agricultural production; have a single, or few, positive rates; zero-rate exports; include all levels of production but exclude much retail trade and other small businesses by means of a threshold; and treat domestic production and imported goods the same.

It is thus hard to argue that, for these countries at least, the VATs currently in place are inherently more complex or, hence, more costly to collect, than the consumption taxes they have replaced. This bears on the related arguments of whether a VAT increases the risks of underreporting and fraud, and whether tax administrations in most developing countries have the capacity to cope with these risks.

Basic sources of fraud in a sales tax include nonregistration of businesses (especially small or medium-sized), underreporting of gross receipts, abuse of multiple rates, and nonremittance to the tax administration of the tax that has been paid to the taxpayer by its customers. These are also the primary risks for revenue that a tax administration must address when a VAT is implemented. In this connection, experience in many developing countries shows that a VAT has the advantage of making it easy to determine when a trader selling only in the domestic market is underreporting taxable gross receipts: the monthly VAT returns will show when the trader is continuously reporting more input VAT than output VAT. This is, in fact, the most common fraud and is obviously facilitated by the extensive use of cash payments for most transactions.

There are, however, some types of fraud under a VAT that are different from, and more sophisticated than, fraud under other indirect taxes. Examples include the use of false invoices (to abuse the credit system), the representation of domestic sales as exports (to benefit from zero-rating), and the claim of VAT credits for non-creditable purchases (such as purchases for private consumption). These types of fraud can take some time to develop after a VAT is introduced—though there have been fast learners—and even when they eventually appear, they are no more difficult to deal with than basic corporate income tax types of fraud such as transfer pricing, fictitious payments to phantom companies, or reporting as business expenses the purchase of goods or services for private use. However, false export schemes and other excess credit claims, which give rise to entitlement to actual refunds, have proven to be quite troublesome in countries with weak tax administration capacity, particularly in the audit area. This is explored at greater length in Chapter 15.


There is evidence that the gain from adopting a VAT is less marked, all else equal, in less developed countries. The important question, however, is whether the gain is positive even in such countries. While there are signs of a revenue gain (recall Chapter 3), there is little hard evidence on the costs of administering and complying with the VAT in developing countries. This is an important area of ignorance about the VAT.

What does seem clear, however, is that the taxes the VAT has replaced were not simple. Terkper (1995) puts the point more generally and succinctly in arguing that “… the worst form of VAT may still be a better option for raising revenue than most traditional forms of taxation.” Further, predecessor taxes were often marked by complicated arrangements to avoid cascading, and in that sense were made complex out of a desire to achieve indirectly precisely the effect that VAT achieves directly. In particular, it should be emphasized that the use of crediting mechanisms to alleviate cascading has been widespread in the predecessor systems of many countries. The logic and modalities of the invoice-credit VAT will thus in many cases have been broadly familiar to taxpayers and the authorities. The key novelty in adopting a VAT is often less in the nature of the tax itself than in the methods of its administration and its modernizing influence more generally.

These observations, combined with the key lesson that simple VATs are cheaper to collect than more complex ones, imply that in many developing countries a simple VAT with a high threshold will in many respects be a simpler tax than those that it has replaced.

    Other Resources Citing This Publication