Administrative and compliance issues unique to VAT : lessons from two periods of British experience

 DISCUSSION PAPER Report No.DRD192 ADMINISTRATIVE AND COMPLIANCE ISSUES UNIQUE TO VAT LESSONS FROM TWO PERIODS OF BRITISH EXPERIENCE By Cedric Sandford and Mich~el Godwin Bath University Centre for Fiscal Studies September 1986 Development Research Department Economics and Research Staff World Bnnk The World Bank does not accept responsibility for the views expressed herein which are those of the author(s) and should not be attributed to the World Bank or to its affiliated organizations. The findings, interpretations, and conclusions are the results of research supported by the Bank; they do not necessarily represent official policy of the Bank. The designations employed, the presentation of material, and any maps used in this document are solely for the convenience or the reader and do not imply the expression of any opinion whatsoever on the part of the World Bank or its affiliates concerning the legal status of any country, territory, city, area, or of its authorities, or concerning the delimitations of its boundaries, or national affiliation.  ADMINISTRATIVE AND COMPLIANCE ISSUES UNIQUE TO VAT LESSONS FROM TWO PERIODS OF BRITISH EXPERIENCE By Cedric Sandford and Michael Godwin Bath University Centre for Fiscal Studies September 1986 The World Bank does not accept responsibility for the views expressed herein which are those of the author(s) and should not be attributed to the World Bank or to its affiliated organizations. The findings, interpretations, and conclusions are the results of research supported by the Bank; they do not necessarily represent official policy of the Bank. The designations employed, the presentation of material, and any maps used in this document are solely for the convenience of the reader and do not imply the expression of any opinion whatsoeveL on the part of the World Bank or its affiliates concerning the legal status of any country, territory, city, area, or of its authorities, or concerning the delimitations of its boundaries, or national affiliation.  ABS'!RACT The paper updates the 1977-78 study on administrative and compliance costs in the U.K. It begins by outlining the results of the 1977-78 study,and then compares them with the estimates for 1984-85. The paper concludes from the analysis cf these findings and estimates that there is no a priori reason for procla~ming that the VAT is better or worse than the next most comparable alternative, the retail sales tax (RST). The authors conclt1de that, on balance, lin RST is to be preferred if the rate is low, but the VAT has the advantage if the rate is to be high. This paper was prepared for the Conference on Value Added Taxation in Developing Countries, sponsored by the Public Economics Division, Development Research Department, The World Bank.  ADMINISTRATIVE AND COMPLIANCE ISSUES UNIQUE TO VAT LESSONS FROM TWO PERIODS OF BRITISH EXPERIENvE Table of Contents Page No. The Administrative and Compliance Costs of VAT: Early Experience ••••••••••••••••••••••••••••••••••••••••••••• 2 Compliance and Administrative Costs of the U.K. VAT - 1977/78 • • • • • • • • • • • • • • • • • • • • • • • • • • • • a o e e e e • • • • • • 4 Some Comments and Qualifications •••••••••••••• c •••••• 8 • • • • • • • • ., 6 The Effect of Zero Rating e at e e e • e e e e e II e e • e e II e e e e e e e e e e e e e e e e e e e e 6 Offsetting Ben~~fits to Traders ............................. , ..... 7 Compliance and Administrative Costs of the U.K. VAT: Experience in 1984/85 ••••••••••••••••••••••••••••••••••••••'• 9 Some Conclusions on Administration and Compliance Costs ••••••••••••••••••••••••••••••••••••••••• 12 Ho~ Effectively Does VAT Ensure that Tax Does Not Enter Business Costs and Hence Export Prices ••••••••••••••••• 18 How Superior is VAT in Controlling and Minimizing Evasion? •••••••••••••••••••••••••••••••••••••• 21 Conclusion 26 ················"·~································~· Acknowledgement ................................................ 29 Table 1: Estimates of Costs of Operating the United Kingdom VAT 1977-78 and 1984-85 c••••••••••• 11  ADMINISTRATIVE AND COMPLIANCE ISSUES UNIQUE TO VAT: LESSONS FROM TWO PERIODS OF BRITISH EXPERIENCE Cedric Sandford and Michael Godwin This paper focuses on issues of tax administration and compliance cost for the most common type of VAT: the consumption type. The British VAT, adopted in 1Y73, conforms to general EEC standards: the tax is a consumption type VAT, employing the destination principle using the invoice system. Comparisons, implicitly or explicitly, are with the practical alternative of a retail sales tax (RST), rather th~n with other forms of VAT. In terms of administration, the ~ •.~":,damental feature::; of a VAT are its collection at every stage of the production and distribution process and its rebate or tax credit system by which a 'registered trader' (a convenient term for any VAT taxable person, what2ver the stage of production) deducts tax on his purchas~s ('input tax') from tax collected on his sales ('output tax') handing over the difference to the revenue authorities; where, for any reason, i~put tax exceeds output tax, the trader is entitled to a refund. Both multi- stage collection and the rebate system distinguish VAT from an RST; and the rebate system distinguishes VAT from a turnover cascade tax. The result of these two distinguishing features of VAT is that the number of registered traders will be larger than under an RST. (It is also likely to be larger than under a cascade tax which necessarily employs •  - 2 - exemptions rather than zero rating); and that both tax rrdministrators ~nd registered tr~ders are, in principle, concerned with records of both inputs and outputs. This paper seeks to explore the following issues: 1. The main disadvantage ,of VAT, viz the level and distribt~tion of administrative and compliance costs. 2. The ms:.. n benefits which are alleged to flow from the administrati.;."e struci.;; ·~L"e, viz. a) That it is a highly effective way of ensuring that tax does not enter production costs and export pricese b. That the tax is relatively evasion-free. The Administrative and Compliance Costs of VAT: Early Experience As far as we are aware:, the only large-scale study of the administrative and compliance costs of VAT has been that carried out in the United Kingdom at the Bath University Centre for Fiscal Studies.[! 1/ C.T. Sandford, M.R. Godwin, PeJ.W. Hardwick and M.I. Butterworth, Costs and Benefits of VAT (London: Heinemann Educational Books, 1981). A smaller study in the Netherlands (M.A. Snijder, De invloed von de fiscus op het midden-en klein bedrij£ (Deventer: Fed, 1981) show a similar reduction in proportionat~ costs as firm size increases, on a sample of 286 firms~ Snijder's costs were somewhat lower for small firms than Sandford's, presumably because, unlike the UK revenue department, the Dutch authorities allow firms a tapering rebate of tax according to size.  - 3 - The study related to the year 1977/78 but the present authors have attempted to update it. Because the survey data is more reliable than the updated figures and because a comparison of the two years offer~ some useful lessons, we start by outlining the findings for 1977/78 and then compare them with estimates for 1984/85. In the study, information on administrative costs was derived mainly from published sources, with some supplementation from survey data. The data for estimating compliance costs was derived from a mail sample survey of VAT registered tr~d€rs supplemented by interviews with traders and accountants and other professional advisers. At the time the survey was undertaken, the standard rate of VAT in the United Kingdom was 8 percent and there was a higher rate of 12 1/2 percent applying to petrol, most domestic electrical goods (except cookers, space heaters and water heaters), furs, jewellery, boats, aircraft and most caravans. Besides exports a wide range of products were zero rated, including food (excluding confectionery and restaurant meals), fuel and power (except petrol), new construction, books, maps and musicp newspapers and magazines, public transport, prescription drugs and children~s clothing. Exempted from VAT were rent and land, insurance and banking servives (unless exported), education, health and postal services (if provided by the post office) betting and gaming and funeral servives. In addition, small firms with an annual turnover of under 57,500 in October 1977 (raised to 510,000 in April 1978) were not required to register. Because of the zero rating and exemptions, a positive VAT rate extended to not much more than half the value of total consumption.  - 4 ~ Compliance and Administrative Costs of the U.K. VAT - 1977/78 The year to which the survey related was one in which no major changes took place in the conditions or coverage of VAT: hence the compliance costs can properly be taken to measure the regular costs of compliance, unaffected by temporary increasese The difficulties of this kind of survey are such as to generate healthy scepticism about the findings. A fuller account of the methodology of the survey may be readily found in S~ndford et al (1981). The main points are summarised as follows. In order to ensure the repres~ntativeness of the survey, the sample had to be drawn from the VAT register. This was done for the researcher~ by customs and excise in such a way as to ensure anonymity. The total sample size was just over 9,000 stratified by size and business sector. A response rate of 31 percent yielded just under 3,000 usable returns. Various internal and external checks served to establish the general validity of the responses. Aggregate compliance costs were estimated by grossing up the mean compliance costs by size group in each of nine business sectors. The total population in each size by sector group was known from the VAT register. Compliance costs consisted mainly of fees paid to staff doing VAT work, the time proprietors or their families spent over VAT work, valued appropriately, and fees to l'lccountants or other professional advisers for work on VAT. The main findings from this part of the study were as follows:  - 5 - 1 Aggregate compliance costs were 5392 million or about 9 percent of tax revenue. 2. The burden of compliance differed considerably amongst businesses. It was affected by a number of factors including the 'rate mix' of the firm's output, the business sector concerned and the system of accounting employed. The predominant influence, however, was size of firm. Proportionately, compliance costs bore much more heavily on small firms than on large. Measuring size of business by taxable turnover {to tie in with the national statistics) the smallest firms had compliance costs which, proportionately, were over 30 time~ as r~gh as the largest firms. Thus, in 1977/78, on average, it cost firms in the 510,000-520,000 turnover range 512.30 in compliance costs for every 51,000 of goods sold. The corresponding cost to firms of over tl million taxable turnover was only 40p for every 51,000 of goods sold. 3. The existence of the higher rate of tax was estimated to add 8-9 percent to the costs of compliance compared with what the cost would have been if all higher-rated goods had been standard-rated. 4. The costs of administration (as published by Customs and Excise) were s85 million, 2 percent of tax revenue.  - 6 - 5. On the basis partly of published data /2 and partly of data arising from the survey (relating to the freq·uency and length of VAT control visits to firms of different sizes) a rough and ready allocation of administrative costs as between different sizes of firm suggested that, like compliance costs, administrative costs were disproportionately attributable to small firms which generated little revenue. 6. The total costs of operating the VAT system (i.e. the combined compliance and administrative costs) amounted to 11 percent of tax revenue. Whilst the figures must be regarded as very approximate, it appears that 40 percent of the compliance costs and over 50 percent of the administrative costs were incurred by or in respect of some 69 percent of traders (those with a turnover of less than 550,000 in 1977/78) who between them generated less than 5 percent of the revenue. Some Comments and Qualifications In considering these figures, we need to take account of certain features: The Effect of Zero Rating. A high proportion of both administra~ive and compliance costs relate to the zero-rated products, apart from exports, which yield no revenue. The zero-rated products (listed above) constitute more than one quarter of domestic consumption in the United Kingdom and nearly 40 percent of the value of the goods and services in the VAT base. Of course, 2/ Review of Value Added Tax, Cmnd 7415 (London: HMSO, 9978).  - 7 - some zero-rated goods are sold by traders who sell standard-rated goods also, and have a net tax liability. Some exporters will require refunds. So there is no precise tie-liy between the proportion of zero-rated goods and the proportion of refunds. In fact, the total of refu~!a to UK traders is approximately one ~hird of VAT payments. Traders entitled to refunds have to be registered like other traders; their returnu must be monitored with particular care as payments are being made to them; and, in the United Kingdom, if regular repayment traders, they are allowed to submit returns monthly instead of quarterly. They therefore create more returns for vetting than other traders, yet they bring no revenue return from the administrative and compliance cost generated. Offsetting Benefits to Traders. We need to distinguish between 'gross' and 'net' compliance costs. In other words, we need to recognise some benefits accruing to firms as a result of compliance which constitute an offset to the figures set out above. These benefits are of two main kinds: (a) cash flow benefits and (b) managerial benefits. (a) Under the United Kingdom provisions, traders are required to collect tax over a three-month period (the 'collection period') and submit their return and pay over the net tax liability to customs and excise by the end of the following month (the 'grace' period). Regular repayment traders are allowed to submit monthly returns and can export to recover their refund about two weeks after subruission. The trader who is not a repayment trader will, therefore, assuming an even flow of payments and receipts, obtain a cash flow benefit which, on  - 8 - average throughout the year, is equal to 2 1/2 months of VAT or 5/24ths of annual VAT payments. This i~ made up as follows: 3 months' collection period, averaging 1 1/2 months' VAT or 3/24 V. 1 month grace period (3 months' VAT held for 1 month every 3 months =1 month's VAT every month) = 1/12 V. (Where V = annual VAT payment), If, as is often the case, traders delay paying VAT beyond the legal limit and do not incur offsetting penalties, their cash flow benefit will be more./3 Several pointa should be noted about this cash flow benefit. Its value will depend on the level of interest rates and the credit/debit situation of the individual trader. The distribution of the benefit depends partly on the commercial credit conditions (as, save with cash sales, the tax point is the invoice, not the payment date). The cash benefit, whilst a real advantage to the individual trader, is not a real resource saving to the economy as a whole; it is a transfer -- in effect an interest-free loan from the government to the business sector. For regular repayment traders there is a cash flow detriment (unless they can hold back on paying their bills until after they have received the refund on their invoices), In 1977/78 the net cash flow benefit overall was valued in the study at 573 million. For many large businesses, the cash flow benefit exceeded the compliance cost leaving them with a net benefit. (b) Some firms, especially smaller firms, gained managerial benefits. VAT requirements compelled many firms to keep bett,:,r and fuller 3/ Committee on the Enforcement Powers of Revenue Departments (Keith Committee Report, Cmnd 8822. London: HMSO, 1978.  - 9 - records thaP they would otherwise have done. Some of them used these records to improve decision-making. Of the firms with a turnover of s 10,000 - s 100,000 in 1977, 78, 42 percent agreed that their purchase records, and 30 percent that their sales records, were better kept as a result of VAT. Over~ll, 32 percent of traders claimed that their purchase records were better kept and 26 percent made the same claim for sales records. Whilst many respondents to the survey did not consider that they got any benefit from improved record keeping, 25 percent claimed to save money by doing more of their own accounts, 8 percent enjoyed improved stock control, 6 percent claimed discounts more frequently, 5 percent had fewer bad debts and 4 percent claimed other miscellaneous advantages. Whilst it is difficult to put a realistic money value on the managerial VAT, they clearly should not be ignored as an offset t~ c2mpliance costs; moreover, unlike the cash benefit, the managerial benefits represent a saving in real resources. Compliance and Administrative costs of the U.K. VAT - Experience in 1984-85 Since 1977/78, a number of changes have taken place which have affected administrative and compliance costs. Where has been an increase in the use of risk analysis by customs and excise: traders in high risk categories are now visited more frequently, by higher-grade officers, for more thorough visits than traders in lower risk categories. Credibility and control checks have been stepped up, while purely educational visits to traders take place only on request.  - 10 - Even so, there has been a steady deterioration in trader compliance, largely attributable to the recession (together with hi~h interest rates), which has led traders to cling to cash until the last possible moment. A system of fixed scale penalties, which will apply automatically to persistent rate payers, is now in the process of introduction. It is expected that these penalties will reduce the late payment problem substantially, so that fewer expensive criminal proceedings will need to be taken. On the compliance cost side, some chances will have tended to reduce costs: there has been a simplification of VAT procedures and documentation, and the higher positive rate of VAT has been abolished while, at the same time, the standard rate was substantially increased. On the other hand, the special provisions allowing postponement of accounting for imported items have been withdrawn. Import VAT is now subject to provisions similar to customs duties; payment guarantees are required, and documentation has become more ~omplex. This latter change will have increased the compliance costs of importers and reduced the cash flow benefit to the trader population because some VAT is now payable earlier in the production chain. The full effects of this measure, which was introduced in November 1984, have yet to show up in published statistics. Changes in inflation and interest rates will also have had important effects on total revenue, total costs and the value of VAT cash flows. Table 1 compares the 1977/78 figures with official administrative costs for 1984/85 and estimated compliance costs for the same year. The  - 11 - Table 1: ESTIMATES OF COSTS OF OPERATING THE UNITED KINGDOM VAT 1977-78 and 1984-85 (1) (2) (3) (4) (5) (6) (7) (8) Years Revenue Admin Compl Value Net Admin Compl Net from costs costs of compl costs costs compl VAT cash costs /b costs benefit /a (3-4)- As percent of revenue isbn J.mt Em Ism Em 1977/78 4.2 85 392 73 319 2.0 9.3 7.6 1984/85 18.5 191 940 495 435 1.0 5.1 2.3 a/ Calculated using interest rate of 7 percent, 1977/78 (average bank minimum lending rate) and 10.75 percent, 1977/78 (average bank base rate). /b No allowance has been made for managerial benefits.  - 12 - method of estimating has been to take the 1977/78 figures, allow for the saving arising from the abolition of higher rate, and then, because labor is the main component in the costs, update the figures in proportion to the increase in average earnings, adjusting at the same time for the int:rease in the number of traders. The figures make no allowance for the other changes mentioned above, nor for the managerial benefits derived from VAT record-keeping. Some Conclusions on Administration and Compliance Costs The more recent levels of aggregate administrative and compliance costs for VAT in the United Kingdom at 1 percent and 5 percent respectively are not unreasonable,/~ especially as no allowance has been made for managerial benefits. If we include the cash flow benefits, the aggregate compliance costs are reduced to 2.3 percent. However, the cash flow benefit is not a real resource saving but a transfer; and, in any case, the 1984/85 figures are inflated by the very high interest rates. ~/ Administrative costs of the UK revenue departments were given in the departments. Most recent annual report~ as 5786.7 million for inland revenue in 1983/84 equivalent to 1.72 percent of revenue; and 5393.5 million for customs and excise in 1984/85 equivalent to 1.11 percent of revenue. Relatively few attempts have been made to measure compliance costs: t~o examples are C.T. Sandford, Hidden Costs of Taxation (London: Institute for Fiscal Studies, 1972), in which compliance costs of U.K. personal direct taxes in 1970 were estimated at between 2.5 and 4.4 percebt of revenue, with up to 4 percent more for tax work done for private firms not billed as such: J. Slemrod and N. Sorum, ''Compliance Costs of the United States Individual Tax System," National Tax Journal, vol 37, no 4~ pp. 461-474 estimates u.s. personal income tax compliance costs to be between 5 percent and 7 percent of revenue in 1982.  - 13 - The reduction in administrative and compliance costs between the two years as a percentage of tax revenue has arisen primarily because of the increase in the standard rate of VAT (8 percent in 1977/78, 15 percent in 1984/85). Clearly, VAT is not a tax to impose at low rates. Changes in the tax structure, such as the introduction of additional rates, may have substantial effects on costs; but apart from the initial re-adjustment, once the collection machinery is in place, the cost to the registered trader and to the revenue department remains much the same whatever the tax rate. Whilst aggregate costs may now be reasonable, the burden on small firms remains disproportionately high. The market power of large firms may enable them to obtain an undue share of the cash benefit of VAT throug~ the commercial credit terms they can impose on their suppliers and their customers, whilst the record-keeping burden for small firms remains considerable. As a consequence, the large firm often enjoys a net benefit from VAT (the value of its cash flow benefit exceeds its compliance costs) whilst the small firm faces significant net cost (although it is small firms who stand to gain most managerial benefits from VAT). It should be borne in mind that a figure which may look small as a proportion of turnover (eg 1.2 percent in the 510,000-520,000 range in 1977/78) will represent a higher proportion {sometimes a much higher proportion) of net profit. The question therefore arises: 'Who actually pays the compliance cost?' 'What is its effective incidence?'. If all businesses had compliance costs proportional to turnover, we would expect the effective incidence to be identical to that of the tax  - 14 - itself, which, in most cases, we would expect to be wholly or largely shifted forward to the customere However, the compliance cost is not proportional; and where small firms are competing in the same market with large firms, the small firms will be placed at a distinct competitive disadvantage. They are likely to have to take a cut in profits or in the leisure time of the proprietor or his family. That small firms in the United Kingdom see VAT as a significant burden is indicated by the persistent criticism of VAT compliance costs from the small firm lobby ever since the tax was introduced. A recen.t survey, Burdens on Business},/5 found that small firms continued to cite VAT more frequently than any other government regulatory activity as an unwelcome extra burden. The competitive disadvantage which small firms face as a consequence of VAT is a government or state-created disadvantage; there is therefore a particular case for granting concessions to small firms. The various possibilities, such as the forfait system, are reviewed elsewhere in this volume by Professor John Due. However, it is worth stressin~ here that it is the relative position of small and large firms which is important. Thus, measures such as differential return and payment periods (adopted, for 5/ Department of Trade and Industry, Burdens on Business (London: HMSO, 1985}.  - 15 - example, in Belgium and Germany) /6 can assist small firms simply by putting large firms on shorter periods and thus reducing their cash flow advantage. This procedure may be suitable for developing countries where small local producers compete with large multinationals. Another possibibity for developing countries is to put the obligation to make VAT calculations on small firms' suppliers (as with the Belgian equalisation tax)./6 The principal concession to small firms in the United Kingdom has been the exemption level, which is the highest in Europe, standing in 1985/86 at s19,500. Raising the registration threshold to a turnover of s50,000 or even 519,500 would not involve much revenue loss./7 However, raising the exemption limit does not automatically mean that firms below the new threshold will wish to deregister; indeed, Customs and Excise already devote some energy to persuading small firms which are registered voluntarily to withdraw their registration./8. The reasons a small firm may wish to remain registered are: (a) VAT registration may be seen as a mark of respectability for a trader (important for his credibility in the eyes of suppliers and/or customers). M.R., Godwin andc. T. Sandford, "Simplfying VAT for Small Traders", Accounting and Business Research, vol. 13, no. 52 (autumn 1983), pp. 279- 288. Hansard (4 December 1985), Written answers, cola 253. Unpublished statistics supplied in private communication from BM ~1stoms and Excise.  - 16 - (b) Registered customers may prefer to purchase from a registered supplier, from whom all input tax is reclaimable, rather than from an unregistered firm where VAT inputs will enter into costs; indeed, sometimes registration is made a condition of entering a trading relationship. (c) Traders with substantial zero-rated outputs will be better off if they can reclaim input VAT. Whatever the level of the threshold, it will create a competitive imbalance between firms above and firms just below the threshold. Indeed, some sectors in the United Kingdom (especially the building industry, restaurants and hotels) currently advocate a lower limit fer exemption. The competitive imbalance is more pronounced the higher the rate of VAT. Even the small firm lobby is not of one mind on the merits of raising the threshold. Its largest spokesman, the National Federation of the Self-Employed and Small Businesses, has opposed an increase partly because it fears it would deter business expansion and might lead to compulsory deregistration. As already indicated, zero-rating raises administration and compliance costs very significantly, especially in proportion to revenue and compared with a situation in which such goods were standard-rated. The United Kingdom is very unusual in the extent of its zero-rating; of the European countries, only Ireland has anything like the same amount. On this count, therefore, one could expect other countries to have lower administrative and compliance costs. The main argument used by governments for zero-rating has  - 17 - been distributional. However, it can be strongly argued that zero-rating particular commodities is an inefficient way of assisting the poor and th~t, given an income tax which embraces most of the adult population and a flexible social security system, there are more effective ways of achieving the desired objective. Such ways, however, may not be open to developing countries. The distributional effects of VAT are the subject of another paper. All that need be said here is that the main d:l..sadvantage of VAT -- its high administrative and compliance costs -- is significantly reduced, probably in absolute terms and certainly in proportion to revenue, if zero-rating is confined to exports. The argument for a single positive rate instead of multi-rates is similar to, but not quite as strong as, the case against zero rating on cost grounds, beoause zero-rating requires the extra administrative work of repayments. What of the arguments against exemptionso Where traders are producing wholly exempt outputs, no administrative and compliance costs are incurred. But where traders are producing output which is part exempt and part taxable, it becomes necessary to distinguish exempt from non-exempt inputs. Moreover, as with multi-rating, the existence of exempt outputs ct•eates borderline problems which put up administrative and complianoe costs. There is no doubt that a single-rate VAT on as broad a base as possible, besides having advantages of neutrality, is the form which minimises administrative and compliance costs. As a final comment on the administrative and compliance costs of VAT, the United Kingdom study does bring out the importance of the choice of collection and grace periods. Compared with EEC countries generally, the United Kingdom is generous in the periods allowed (except, perhaps, to its  - 18 - smallest traders). Any country adopting VAT needs to give careful thought to the periods allowed and also to arrangements to control and minimise late payments. At high rates of interest there is much at issue. For example, if the United Kingdom were to reduce the return and payment periods for the largest 6 percent of firms from three months to one month, there would be a once and for all gain to the Exchequer of over 51.5 billion. We turn now to the advantages which ~~re claimed for VAT arising from the tax credit or rebate system on the basis of invoices. How ~ffectively Does VAT Ensure that Tax Does Not Enter Business Costs an~ Hence Export Prices1 VAT enters business costs and, hence, export prices where it is non- deductible. There are a number of ways in which this may ocour: (1) where an exempt service is directly exported; (2) where an exempt trader directly exports: (3) where exempt services are bought by an exporting business: (4) where an exporting busine$S buys from an exempt trader; (5) where VAT invoi.ces are not granted {so-called 'restricted goods'), lost, mislaid or forgotten, or where some 'exports' to tourists are not zero rated. Where an exempt service is directly exported, no VAT can be reclaimed on inputs subject to VAT. In the United Kingdom th~ kind of services (eg postal, education 9 burial and cremations) exempt from VAT are unlikely to figure significantly in exports save for financial services. But overseas banking and insurance are zero-rated.  - 19 - Similarly, an exempt trader cannot recover input tax, but traders exempt by virtue of their low turnover are unlikely to be engaged in the export business. The most significant element of VAT in business costs is likely to enter from the purchase by an exporting firm of exempt supplies which include an element of VAT. In the United Kingdom virtually all exemptions are in the service industries and include financial and insurance services, pro?-crty rental and postal services. Clearly almost all businesses are likely to incur rental, banking, insurance and postal costs. These activities will bear non- deductible input tax and the amount of VAT which gets passed on from them into business costs (part of which will go into export prices) will depend on the aggregate cost of VAT to the suppli~rs of exempt services and how far they are able to pass on these VAT costs to their customers. Where an exporting business buys from an exempt trader the same considerations apply as in (3), but this situation is relatively unimportant. Businesses will normally prefer to buy from other registered traders precisely so that they can obtain their VAT rebateo Where they do buy from exempt traders, the latter will have to offer a price competitiv~ ~ith the net of tax price of registered traders. Finally, VAT may enter export prices where exporters, although themselves registered, and purchasing from registered traders, cannot claim VAT on certain inputs which are disallowed. Thus, the United Kingdom and Ireland do not allow any reclaims for input tax on cars (except for car  - 20 - dealers)e Belgium and other countries allow only partial relief on cars. France does not allow businesses to reclaim hotel or restaurant expenses even for subsistence purposes, nor business travel, motor fuels or passenger cars. Other common disallowances are expenditures on business gifts, business entertainment and company aircraft. Claims for input tax have to be made on the strength of an invoice. If invoices are forgotten, lost or mislaid, input tax cannot be claimed (except that, in the United Kingdom, if the loss is realised, copy invoices are acceptable). Finally, some exports taking the form of sales to tourists to take home may not be zero-rated. In the United Kingdom, there is a provision by which customers can reclaim VAT by sending back to the seller a document stamped by customs and excise at the port of embarkation: the seller can then refund the VAT and use the stamped document as proof of export sale. However, although open to all businesses, some retailers do not take advantage of the scheme because of the extra paperwork; hence some VAT enters export price~. Whilst it is clear from the foregoing possibilities that some VAT does find its way into business costs and export prices, the amount is relatively trivial. It seems likely that the rebate mechanism of VAT is superior to that of any other tax in ensuring this end. The relatively precise mechanism of the rebate system of VAT for excluding tax is especially important in the context of customs unions or international agreements where, without it, countries may adopt ad hoc methods of rebating tax on exports  - 21 - which carry the suspicion of overcompensation and covert subsidiaation. This is one of the aspects of VAT which commended it to the EEC. How Superior is VAT in Controlling and Minimising Bvasion? It is part of the conventional wisdom that the tax credit or invoice system of VAT is a relatively effective means of controlling evasion. But the validity of the argument cannot be taken for granted and needs exploring in some depth. One argument has been that the invoice system provides a means by which the revenue authorities can control evasion bJ~ cross checking. With trade between registered traders, one man's tax receipt is another's tax credit or refund. Thus the invoices can be used to check the accuracy of returns. This proposition cannot, of course, apply to tr4de with non- registered persons; hence, it does not apply at the retail stage. Yet, by common consent, the retail stage, which includes a large number of small , businesses, is the most vulnerable. However, it is argued that under VAT {in contrast to RST) only a proportion of tax is collected at the retail stage; hence, both the incentive to evade at any stage, including retail, and the amount of tax at risk at that stage, is less than with an RST. A complementary argument, which does not depend on activity by the revenue authorities, is that VAT is largely self-policing~ If, as a registered trader, I wish to evade VAT~ then I will seek to understate my output tax (the t6x collected on my sales) or/and ~tate my input tax (the  - 22 - tax on my purchases which constitutes a c~edit). Moweve~, in trade between registered traders it is clegr that one trader's interest in respect of evasion is directly contrary to that of his supplier or customer. So collusion to evade between registered traders is more difficult. Again, the argument does not hold at the retail stage, where the member of the public doing the purchasing is only interested in the price he actually pays. He c~nnot reclaim VAT and often the pa}~ent will be by cash without any invoice. However, as before, it is argued that, with VAT, only a proportion of tax is collected at the retail stageo Leaving aside for the moment the question of the retail end, how valid are these arguments of cross-checking and self-policing? Clearly the revenue authorities cannot, as a matter of course, cross check all invoices of suppliers and customers. How·ever, some cross checking does take place. Doubtless practice differs between countrieso In the United Kingdom customs and excise make cross checks where there is a suspicion that something is amiss. A strong suspicion leads to an immediate cross check; where the suspicion is weak, a note is recorded on the file of the supplier (or custc~er) of the trader under suspicion .and the matter is investigated at the next routine inspection of the supplier. The invoices do, therefore, provide a definite means of c<Jntrol. As for the self-policing argument, this too, carries conviction. Clearly VAT frauds between registered traders require more elaborate collusion. However, more careful examination needs to be made of the final  - 23 - component of both arguments - ie that, because only a proportion of revenue is collected at the vulnerable retail stage, tt~ soope and incentive for evasion is less. This statement can be questioned. VAT is a tax on domestic consumption and the whole of VAT has to be extracted from the final consumer. The retailer is collecting the whole of the tax. He is not handing all of it to the revenue authorities because he is first deducting what he has paid on his inputs. But the retailer does have to collect the full VAT. Given a tax credit system which allows a trader to reclaim any balance of input over output tax, in theory it could be argued that there is as much tax at risk at the retail stage under VAT as under an RST. A retailer under VAT collects the whole tax on his outputs and, by claiming to be zero-rated, could seek a refund of his input tax. Whilst true, however, this must be regarded as a somewhat unrealistic argument; the practice could not long continue without being spotted. The fact that, under VAT, retailers have paid some tax on their inputs is significant in reducing the tax at risk at that stage. But the argument does illustrate several points. VAT offers scope for fraud by fake claims for rebates which does not affect RST, and the more zero-rated goods the more the scope for such fraud. Multiple rates and exemptions also facilitate fraud. If a VAT trader is selling goods all of which are standard- rated, to understate his output tax he needs to understate his aggregate sales also; if he is selling at different rates of VAT, he can understate his output tax by understating his sales of higher-rated goods within an unchanged aggregate. Another consideration is that, because VAT requires the maintenance of records of both inputs (purchases) and outputs (sales), the revenue  - 24 - authorities have a more solid basis for checking the accuracy of returns than if only sales figures were supplied. As we have seen, a registered trader will not wish to understate his purchases (on which he gets a tax credit). Revenue authorities develop guidelines about the mark-up in particular trades and how much output can be expected from a particular flow of inputs. Of course, this relationship will vary according to whether a firm is altering the level of its stocks or equipment: but a persistent irregularity in the relalionship between inputs and outputs can give rise to suspicions ~hich can be followed up by inspections. Amongst the ECC countries there has been a distinct move towards joint income tax/VAT audits which have helped to control evasion of both taxes. In the United Kingdom, where the Inland Revenue Department is responsible for income tax and customs and excise for VAT, an experiment with exchange of information between the two departments has been fruitful; the 'Keith' Committee on Revenue Powers /9 has recommended an extension of the practice. The United Kingdom and some other countries try to ease the burden of compliance at the retail stage by offering special schemes by which, for retailers, tax liability is determined solely in relation to inputs. Nonetheless, both input and output data still have to be collected: the special scheme is imposed on top as an easier means of calculating VAT. In 9/ Committee on Enforcement Powers of the Revenue Department~, op cit, p. 465.  - 25 - any case, however, because the schemes are based on inputs, the cross-check with suppliers sales remains. One form of evasion is the failure of small businesses to register even though their turnover is above the limit. At least, as compared with an RST, such traders pay VAT on whatever inputs they require. In comparing VAT with an RST in controlling or minimising tax evasion, there is one important consideration which favours the latter. Because lt is a tax at only one stage, an RST involves fewer registered traders than a VAT. The difference in numbers is not so large as might be thought because an ideal RST should relate to all retail sales, so that wholesalers, manufacturers and farmers who make some retail sales, if only to their employees, ought to be included. But there is, nonetheless, a significant difference in size. In the United Kingdom, an RST would have required the registration of 72 percent of those r~gistered for VAT, according to a 1979 Working Party Report./10 For a smaller number of registered traders a more intensive surveillance can be carried out for the same cost. There can be no certainty that estimates of the extent of evasion are wholly accurate and judgments about the relative proneness to evasion of different forms of tax are fallible. In the United Kingdom a recent official estimate is that 'undiscovered under-declarations' of VAT by registered 10/ "Elimination of VAT on Transactions Between Registered Traders" (unpublished Working Party Report, 1979).  - 26 - \ traders amount to between 5300 million and 5500 million (or around 2 percent of VAT revenue yield)./11 But this estimate taxes no account of revenue losses from traders who have falsely claimed exemption nor does it include losses from major frauds involving organised crime: there have been such problems with the importation of gold, with fraudulent refund claims and with liquidations. Measures have been taken to attempt to deal with these offenceso Conclusion The British experience suggests that there is no a priori reason for proclaiming that VAT is better or worse than an RST. Our tentative judgement in that there is much in the conventional wisdom that an RST ie, on balance, to be preferred if the rate is low, but VAT has the advantage if the rate is to be high. The unique feature of VAT, the rebate system with multi-stage collection, puts up administrative and compliance costs, but does provide a more accurate and reliable mechanism for excluding tax from business costs and export prices and is probably more effective in controlling evasion. What can be said with complete confidence, however, is that a structure with minimum exemptions, a single positive rate, and zero-rating only for exports will mimimise the administrative and compliance costa and maximise the advantages claimed for VAT. This paper draws primarily on United Kingdom experienceo Other references have been more to developed than developing countries. To what 11/ Hansard (16 January 1986), Written Answers, col. 648.  - 27 - extent is the experience of VAT in developed countries relevant to developing countries? In seeking to answer this question we must not forget that 'developing countries' is a term covering a whole range of different stages of development. It is clear that many of the problems arising from the unique administrative and compliance features of VAT are accentuated in developing countries. For their economic size, developing countries tend to have many more small farmers and businessmen than developed ~ountries. If small businesses in the United Kingdom have difficulty with record-keeping for VAT, how much more is that true of small businesses in a country where literacy and numeracy are much less widespread. There are particular difficulties with payments of refunds in many developing countries. If a developed country, for distributional reasons, hesitates to apply a single rate of VAT across the board, how much more reason is there for a developing country to hold back when it faces a sophisticated social security system to compensate the poor for the higher prices of necessities. If the disadvantages of the VAT structure are thus accentuated, the potential advantages of the VAT structure are less likely to be achieved in a developing country. The capacity of VAT to resist evasion rests in part on relatively sophisticated cross checking and crymputerised checks as were as honest tax officers and both may be lacking in developing countries where evasion is, in any case, more prevalent. Likewise, the completeness and precison with which tax can be excluded from export prices is reduced the more exemptions the structure permits -- and, for practical reasons, exemptions are likely to be more necessary in developing countries.  - 28 - All this does not mean that VAT does not have important merits for a developing country, but the expectations must be set lower. Essentially, the tax must be trimmed to administrative capacity. In more precise terms, it is particularly necessary to keep out of the standard VAT structure the myriads of small businessmen and peasant farmers who would find it impossible to keep adequate accounts. This might be done, in principle, by one or more of the several methods illustrated by Due in chapter * of this volume. Such measures ordinarily involve little revenue loss but considerable savings in costs of administration and compliance. The United Kingdom experience and experience elsewhere /12, /13 suggests that VAT can be introduced most easily where it replaces a cascade tax, or at least some general form of sales tax, which has already accustomed small firms to record-keeping. This paper has so far alluded only to the regular costs of compliance and administration the cQntinuing costs when VAT is running relatively unchanged. But, when the tax first comes in, there are once and for all commencement co~ts for the revenue authorities, such as United Nations Department of Economic and Social Affairs, Sales Tax Administration: Ma·or Structural and Practical Issues with S ecial Reference to the Needs of Developing Countries New York: United Nations, 1976). 13/ G.E. Lent, M. Casanegra and M. Guerard, "The Value Added Tax in Developing Countries", International Monetary Fund Staff Papers, vol. 20, no 2 (1973), ppo 319-378.  - 29 - training their own officers and educating traders: and costs for traders in buying new calculators, tills and other equipment as well as familiarising themselves with their obligations under the tax./14 Even after this there will be temporary learning costs for some time until revenue officers and traders have become thoroughly accustomed to the tax. Especially where VAT is not replacing some other form of sales tax which has provided relevant experience, there is a case for restricting the scope of the initial imposition with a view to extending the tax later as experience is gained. Finally, we would stress a point made earlier which applies to developed as well as developing countries. VAT is not a tax to be operated at low rates. The 'overheads' with VAT are such that it should only be contemplated if the intention is to use it as a major revenue yielder. Acknowledgement The authors gratefully acknowledge the assistance received from H M Customs and Excise in the preparation of this paper. 14/ M.Rs Godwin, "VAT -- the Compliance Cost to the Independent Retailer," Accountancy, vol. 87, no. 997 (September 1976), PP• 48-60.  Some Recent DRD Discussion Papers 177. China's Economic Reforms in a Comparative Perspective by Bela Balassa, April 1986 .. 178. Economic Prospects and Policies in Mexico by Bela Balassa, May 1986. 179. Calibration of Macroeconomic Models with'Incomplete Data: A Systems Approach by Michael J. Hartley» April 1985. 180. Structural and Stabilization Aspects of Fiscal and Financial Policy in the Dependent Economy, by Willem Buiter, September 1986. 181. Domestic Credit and Output Determination in a "New Classical" Model of a Small Open Economy with Perfect Capital Mobility, by Peter Montiel, September 1986. 182. Stability, Budget Deficits, and the Monetary Dynamics of Hyperinflation, by Miguel A. Kiguel, September 1986. 183. Public Debt Guarantees and P~ivate Capital Flight, by Jonathan Eaton, September 1986. 184. Saving Behavior in an Economy Without Fixed Interest, by Nadeem Ul Haque and Abbas Mirakhor, September 1986. 185. International Differences in Industrial Robot Use: Trends, Puzzles, and Possible Implications for Developing Countries, by Kenneth Flamm, July 1986. 186. The Determinants of Savings in Developing Countries: Theory, Policy and Research Issues, by Arvind Virmani, August 1986. 187. Measuring Income Tax Incentives Under Evasion: Effective Marginal and Average Rate Schedules, by Arvind Virmani, September 1986. 188. Evaluating Participation in an African Monetary Union: A Statistical Analysis of the CFA Zone, by Shantayanan Devarajan and Jaime de Melo, October 1986. 189. The Treatment of Foreign Trade in Computable General Equilibrium Models of Small Economies, by Jaime de Melo and Sherman Robinson, June 1986. 190. Problems, Development Theory and Strategies of Latin America, by Vittorio Corbo, September 1986. 191. Criteria for Choice Among Types of Value-Added Tax, by Carl s. Shoup, September 1986. 192. Administrative and Compliance Issues Unique to VAT Lessons from the Two Periods of British Experience, by Cedric .Sandford and Michael Godwin, September 1986.