The past decade has seen the adoption of the VAT by a number of countries that—in terms of population—would generally be regarded as small. Even more strikingly, it was seen in Chapter 1 that many of the countries still without a VAT are also small. The suitability of the VAT for small countries, and small islands in particular, is an issue that arises with increasing frequency. Does country size matter is assessing the advisability of adopting a VAT?
Background
Table 16.1 lists the 36 small countries that currently have a VAT, with “small” for this purpose being defined as a population of under five million. All but seven of these adopted the VAT since 1989. The list includes, moreover, seven small island economies144 whose distinctive character may give rise to distinct concerns: no such economy had a VAT prior to 1989.
Small Countries with a VAT
Ratio of VAT revenues to private consumption divided by the standard rate: see Chapter 4. Figures are those available to IMF staff in February 2000.
VAT abolished in 1999.
C-efficiency is for 1995.
Small Countries with a VAT
| Country and Date of Introduction of VAT | Population (millions) | C-efficiency1 |
|---|---|---|
| Albania (1996) | 3.5 | 71.3 |
| Armenia (1992) | 3.8 | 29.0 |
| Barbados (1997) | 0.3 | 101.1 |
| Belize (1996)2 | 0.2 | 56.2 |
| Congo Republic (1997) | 2.5 | 11.7 |
| Costa Rica (1975) | 3.7 | 87.4 |
| Croatia (1998) | 4.8 | 111.4 |
| Cyprus (1992) | 0.6 | 100.1 |
| Estonia (1992) | 1.5 | 79.7 |
| Fiji (1992) | 0.8 | 84.1 |
| Gabon (1995) | 1.4 | 74.9 |
| Iceland (1990) | 0.3 | 64.8 |
| Ireland (1972) | 3.6 | 66.2 |
| Jamaica (1991) | 2.5 | 83.5 |
| Kyrgyz (1992) | 4.8 | … |
| Latvia (1992) | 2.6 | 69.0 |
| Lithuania (1994) | 3.8 | 58.9 |
| Luxembourg (1970) | 0.4 | 82.9 |
| Malta (1995 and 1999)3 | 0.4 | 67.0 |
| Mauritania (1995) | 2.6 | 46.2 |
| Mauritius (1998) | 1.1 | … |
| Moldova (1992) | 4.5 | 71.0 |
| Mongolia (1993) | 2.4 | … |
| New Zealand (1986) | 3.6 | 103.2 |
| Nicaragua (1975) | 4.8 | 34.6 |
| Norway (1970) | 4.4 | 78.8 |
| Panama (1977) | 2.7 | 67.3 |
| Papua New Guinea (1999) | 4.4 | … |
| Samoa (1994) | 0.2 | … |
| Singapore (1994) | 3.0 | 107.9 |
| Slovenia (1999) | 2.0 | … |
| Togo (1995) | 4.1 | 31.8 |
| Trinidad and Tobago (1990) | 1.4 | 46.8 |
| Turkmenistan (1992) | 4.1 | … |
| Uruguay (1968) | 3.2 | 50.9 |
| Vanuatu (1998) | 0.2 | … |
Ratio of VAT revenues to private consumption divided by the standard rate: see Chapter 4. Figures are those available to IMF staff in February 2000.
VAT abolished in 1999.
C-efficiency is for 1995.
Small Countries with a VAT
| Country and Date of Introduction of VAT | Population (millions) | C-efficiency1 |
|---|---|---|
| Albania (1996) | 3.5 | 71.3 |
| Armenia (1992) | 3.8 | 29.0 |
| Barbados (1997) | 0.3 | 101.1 |
| Belize (1996)2 | 0.2 | 56.2 |
| Congo Republic (1997) | 2.5 | 11.7 |
| Costa Rica (1975) | 3.7 | 87.4 |
| Croatia (1998) | 4.8 | 111.4 |
| Cyprus (1992) | 0.6 | 100.1 |
| Estonia (1992) | 1.5 | 79.7 |
| Fiji (1992) | 0.8 | 84.1 |
| Gabon (1995) | 1.4 | 74.9 |
| Iceland (1990) | 0.3 | 64.8 |
| Ireland (1972) | 3.6 | 66.2 |
| Jamaica (1991) | 2.5 | 83.5 |
| Kyrgyz (1992) | 4.8 | … |
| Latvia (1992) | 2.6 | 69.0 |
| Lithuania (1994) | 3.8 | 58.9 |
| Luxembourg (1970) | 0.4 | 82.9 |
| Malta (1995 and 1999)3 | 0.4 | 67.0 |
| Mauritania (1995) | 2.6 | 46.2 |
| Mauritius (1998) | 1.1 | … |
| Moldova (1992) | 4.5 | 71.0 |
| Mongolia (1993) | 2.4 | … |
| New Zealand (1986) | 3.6 | 103.2 |
| Nicaragua (1975) | 4.8 | 34.6 |
| Norway (1970) | 4.4 | 78.8 |
| Panama (1977) | 2.7 | 67.3 |
| Papua New Guinea (1999) | 4.4 | … |
| Samoa (1994) | 0.2 | … |
| Singapore (1994) | 3.0 | 107.9 |
| Slovenia (1999) | 2.0 | … |
| Togo (1995) | 4.1 | 31.8 |
| Trinidad and Tobago (1990) | 1.4 | 46.8 |
| Turkmenistan (1992) | 4.1 | … |
| Uruguay (1968) | 3.2 | 50.9 |
| Vanuatu (1998) | 0.2 | … |
Ratio of VAT revenues to private consumption divided by the standard rate: see Chapter 4. Figures are those available to IMF staff in February 2000.
VAT abolished in 1999.
C-efficiency is for 1995.
The average population of those countries without a VAT—excluding India and the United States (both exceptional cases)—is 7.6 million, compared to an average of 38.3 million for those countries with a VAT. Over a quarter of the countries without a VAT, moreover, are small island economies.
Performance of the VAT in Small Countries
The final column of Table 16.1 reports the C-efficiency ratios145 of the VATs implemented in small countries. They are strikingly high. Indeed, at 65 percent, average C-efficiency for these countries is comparable with the highest of any of the regions identified in Table 4.1. The implication is that the VATs in these countries are relatively well designed and well implemented.
Malta and the VAT
Malta introduced a VAT in January 1995. Initial performance was strong, the VAT raising revenue of about 6.8 percent of GDP and having C-efficiency of around 73 percent.
In December 1996 a new government, committed to repealing the VAT, came to power. As noted in the White Paper For a Better Tax System (January 1997), three main criticisms were leveled at the VAT:
-
“The reality of VAT in a state like Malta is that much of it is a disguised customs duty” (paragraph 2.7). It was argued that a return to explicit tariffs would be superior, recognizing that “… in the particular geographic and economic situation of Malta, customs duties need not be replaced by VAT…to ensure a nondiscriminatory system” (paragraph 2.8).
-
Compliance costs were perceived to be high, especially for medium-sized enterprises.
-
The quite extensive zero rating of domestic consumption, including food other than catering, introduced considerable complexity.
On the first of these, Malta is indeed one of the few countries in which imports exceed consumption, so there is some merit to the argument. In relation to the second criticism, an unusual feature of the initial VAT was a threshold of zero. On the third, widespread zero rating on domestic transactions, of course, is not inherent to small countries.
The VAT was replaced during 1997 by a system relying instead on a broad-based tariff, a single-stage sales tax (the “Excise Tax on Products,” ETP), and selective excises on services. The ETP included crediting and other arrangements to avoid cascading. These replacements were introduced sequentially during 1997, complicating evaluation of the revenue effects of the reform. There seems, however, to have been a noticeable fall in revenue, in the order of 1.5 percent of GDP.
Following a further change of government, the VAT was reintroduced in January 1999 by the party that had initially introduced it. The main structural difference from the initial VAT was an increase in the threshold from zero to, for goods, about $38,500.
Experience with the VAT has inevitably been mixed. Malta, for instance, removed the VAT it introduced in 1995; it was then reintroduced at the start of 1999 (Box 16.1 provides an account of this episode). In other cases, however, such as Barbados, the VAT has been generally seen as a success.
The strong revenue performance of the VAT in these countries is associated with the general observation, developed in Chapter 4, that the performance of the VAT is stronger, all else equal, in economies with a heavier reliance on international trade. As discussed and documented by Alesina and Wacziarg (1998), there is a strong negative correlation between country size and the importance of trade.146 Thus, small economies tend to rely more on trade and so would be expected to have higher C-efficiency.
The econometrics reported in Table 16.2 explores the relationship between country size and the revenue performance of the VAT more closely. The dependent variable is in each case the (log of the) ratio of VAT revenues to private consumption. The first column reproduces a parsimonious specification from the general discussion of VAT performance in Chapter 4 above (column B of Table 4.2). As discussed there, the importance of trade has a significant positive effect. Interestingly for present purposes, the small island dummy appears with a strong positive effect: even conditional upon their reliance on trade, the VAT performs well in island economies. Column B of Table 16.2 repeats the same specification but with the inverse of population replacing openness. As expected, this enters with a strong positive effect. The small island dummy becomes insignificant, but this may merely reflect that these tend also to have the smallest populations, and so effectively correspond to the higher values of the size variables. In column C, the dependent variable is the ratio of VAT revenues to GDP rather than consumption: efficiency again emerges as higher in smaller economies, with small islands now performing well even controlling for their size.
Size, Importance of Trade, and the Performance of the VAT
Size, Importance of Trade, and the Performance of the VAT
| A | B | C | |
|---|---|---|---|
| In(SR) | 0.68* | 0.73* | 0.55* |
| (17.58) | (24.31) | (21.04) | |
| OPEN | 0.006* | ||
| (2.39) | |||
| 1POP | 0.096† | 0.08* | |
| (1.81) | (2.59) | ||
| AGE | 0.008† | 0.008† | 0.009* |
| (1.84) | (1.88) | (2.63) | |
| AF | -0.33* | -0.35* | -0.19 |
| (2.17) | (-2.21) | (-1.38) | |
| SI | 0.34* | 0.20 | 0.27* |
| (3.22) | (1.35) | (2.84) | |
|
|
0.31 | 0.36 | 0.40 |
| N | 89 | 90 | 97 |
Size, Importance of Trade, and the Performance of the VAT
| A | B | C | |
|---|---|---|---|
| In(SR) | 0.68* | 0.73* | 0.55* |
| (17.58) | (24.31) | (21.04) | |
| OPEN | 0.006* | ||
| (2.39) | |||
| 1POP | 0.096† | 0.08* | |
| (1.81) | (2.59) | ||
| AGE | 0.008† | 0.008† | 0.009* |
| (1.84) | (1.88) | (2.63) | |
| AF | -0.33* | -0.35* | -0.19 |
| (2.17) | (-2.21) | (-1.38) | |
| SI | 0.34* | 0.20 | 0.27* |
| (3.22) | (1.35) | (2.84) | |
|
|
0.31 | 0.36 | 0.40 |
| N | 89 | 90 | 97 |
By the standard criteria, the VAT has thus performed well in small countries, with some sign that it has performed especially well in small islands. The natural interpretation is that the importance of international trade for these economies enables them to collect considerable amounts of VAT on imports, and that the geographical remoteness of small islands insulates the tax base, to some degree, from smuggling. This general strong performance does not imply, however, that the VAT is an especially appropriate tax for these economies. Thus:
-
It may be that other taxes would also prove good revenue raisers in these economies. Indeed economies where international trade is important tend to have higher tax yields irrespective of whether they have a VAT or not (as seen in Chapter 4). Given the association between the importance of trade and size, the same will be true of small countries. Moreover, very small economies may have social structures that facilitate tax enforcement.147 Finally, the protection against smuggling which may come with remoteness would presumably help secure tariff revenues as much as it does VAT revenues.
-
These empirical results relate only to the yield of the VAT. It is also important to look to collection costs, again relative to those of other taxes.
Both considerations point to the need to make comparisons with other taxes. In general, there is no single alternative tax to the VAT. For many small economies, however, leading alternatives include a broad-based tariff and a retail sales tax.
Comparing VAT with a Uniform Tariff
Collection costs aside, it is well-known that for a small economy—here using the term in the technical sense of one unable to affect the world prices at which it trades—an appropriately designed consumption tax (such as, but not only, a VAT) is superior to any system with nonzero tariffs. Even if the object of policy is simply to maximize revenue, the consumption tax is in this sense intrinsically superior. Intuitively,148 tariffs mean that private producers face prices that differ from those of world markets, so that their decisions do not lead them to maximize the value of national output at those prices. The country is therefore poorer than it need be. Indeed, it is possible to replace tariffs by consumption taxes in such a way that consumers benefit (as a result of more efficient production) and government revenue simultaneously increases.149
In practice, however, there are collection costs to be weighed against this gain. To assess the likely balance of effects, it is helpful to consider more closely the consequences of moving from a regime in which only tariffs are deployed to one in which only destination-based consumption taxes are deployed. It is easy to show (see Appendix III) that there are three effects to consider:
-
The change in public revenues. Since tax and tariff revenues are simply transfers from public to private sector, this revenue effect should be valued at the marginal efficiency cost of raising revenue; that is, the amount by which the social value of $1 of tax revenue exceeds $1.
-
The change in the deadweight loss that arises from distorting the prices that consumers face away from those at which products must be traded on the world market.
-
The gain in production efficiency from aligning the prices faced by producers with those on world markets. Tariffs on intermediate products, for example, will induce domestic manufacturers to substitute away from those intermediates, implying production techniques which, valued at the world prices that ultimately determine the opportunities open to the economy, are less efficient than they might be.
What do these considerations imply for the suitability of VAT in small countries?
It is always possible to raise at least as much revenue from a consumption tax as from a tariff.150 With a properly designed reform, the first of these effects is thus always positive. How large might it be? This will depend on the rates at which the consumption tax and tariff are set. It will also depend on the size of the respective bases. All else equal, the potential revenue gain in moving from a tariff to a VAT is thus likely to be greater the higher consumption is relative to imports.151 Broadly speaking, one would expect the ratio of consumption to GDP to be independent of country size; the import share, on the other hand, has been seen to be strongly negatively related to size. Thus the ratio of consumption to imports would be expected to increase with size, and this is indeed the case.152 On this account, the advantage of VAT over a broad-based tariff is indeed likely to be less in smaller economies.
The sign of the second effect is ambiguous: deadweight loss could rise in moving to a consumption tax. Taking as a benchmark the case in which it is desirable to tax all goods at the same rate,153 one key consideration will be the breakdown of final consumption between imports and domestically produced items. If much final consumption is imported, then a broad-based tariff will function in large part as a broad-based consumption tax. In so far as small economies would be expected to be relatively undiversified in their production, much of their consumption is likely to be imported. Moreover, much of the consumption that is not imported may be of a kind that either would not in any event likely be subject to VAT, (notably basic agricultural products), or could conveniently be taxed by a well-targeted excise (locally produced alcohol, for instance, or hotel rooms). These observations again suggest that the gains from a VAT will typically be greater in large economies.
The third consideration always tells in favor of the consumption tax, since in a competitive economy eliminating tariffs always increases the value of output at world prices in a competitive economy. The gain is likely to be larger of course, the higher the initial tariffs. This consideration also directs attention to the balance of imports between intermediate and final consumption goods: the greater the share of intermediates in imports, the greater the production inefficiencies induced by the tariff are likely to be. These inefficiencies can be mitigated by duty drawback arrangements (that enable tariffs on intermediates to be reclaimed insofar as they are identifiably used to produce exports) or by outright exemption of intermediates, albeit at the cost of some revenue both directly and perhaps indirectly through the scope for abuse thereby created. In any event, there seems no particular reason to suppose the composition of imports between final consumption and intermediate items—or, hence, the importance of this third consideration—to vary systematically, one way or the other, with country size. Rather, the variation may be more a matter of the state of a country’s development.
On balance, there are grounds for presuming that the gain in moving to a VAT from a broad-based tariff will be less in smaller economies, both because the probable widening of the tax base is likely to be less and because more consumption is likely to be caught, in any event, under the tariff. The basic truth remains, however, that, collection costs aside, the gain is positive: it is better to raise revenue by taxing consumption properly than by distorting trade. Moreover, there may be benefits in putting a VAT in place relatively early in the development process so as provide a solid foundation for future growth.
Thus it can only be collection costs that turn the balance in favor of the tariff strategy. While the collection costs of a VAT were discussed in Chapter 5, the issue for present purposes is how those costs compare, for small economies, with those of collecting tariffs. Unfortunately, there appears to be even less evidence on the costs of collecting tariff revenue than there is on the costs of collecting VAT. It does though seem to be widely supposed that—being collected at a single stage, from a limited number of importers, traditionally heavily reliant on physical verification, and, not least, familiar—tariffs are relatively cheap to administer and comply with.
Fixed collection costs may be a particular concern in small economies. As discussed in Chapter 5, there is strong evidence that there are substantial fixed elements in complying with the VAT, which consequently bears more heavily on smaller firms. Insofar as the average size of firms above the appropriate VAT threshold tends to be smaller in smaller economies—which one might expect to be the case—so collection costs will consequently be more burdensome. Presumably there are fixed components to tariff collection costs too. Their extent, however, is unclear.
In the absence of firm estimates of the likely magnitude of such key quantities as differential collection costs between VAT and tariffs, illustrative calculations can be helpful to give some sense of likely magnitudes. Consider a country thinking of moving from a 10 percent tariff to a 10 percent VAT. Assume that the share of consumption in GDP exceeds that of imports by 20 percentage points, and that under both regimes, half of the base will, for various reasons, be exempt. Assume also that the marginal social value of $1 of tax revenue is $1.25. Finally, assume that the efficiency gains from shifting to the VAT will amount to 0.5 percent of GDP. Then shifting to the VAT will be desirable unless its costs of collection exceed those of the tariff by more than 1.75 percent of GDP,154 which is a large amount. It should be stressed, however, that it may be even more desirable, depending on circumstances to shift the balance not to a VAT but to a simple sales tax combined with excises on a few key commodities.
Conclusions
This discussion points to a number of lessons:
-
Size itself does not seem to be a key concern in determining the desirability of a VAT, except insofar as a smaller average size of taxable firms means that collection costs bear more heavily than they would under other taxes. Little is known about the collection costs of, in particular, tariffs, and how these vary with scale.
-
The importance of trade in many small countries, and the remoteness of many small island economies, are conducive to the effective functioning of the VAT. But these same advantages are also likely to apply to other taxes, not least to tariffs.
-
The advantages of a VAT over a tariff will be greater, all else equal, the lower the share of imports in GDP, the smaller the proportion of consumption that is imported and the larger the share of intermediate goods in imports. The first two of these considerations suggest that the gains from moving to a VAT will indeed be lower in smaller economies.
-
Crude calculations suggest, nevertheless, that the collection cost differential would in many cases have to be implausibly large to outweigh the efficiency and base-broadening gains of moving to a VAT.
-
When there are few stages in domestic production, the VAT may offer little gain over a retail sales tax or broad-based tariff. As the economy develops, however, the merit of the VAT in preserving the chain of taxation will grow. As a longer-term strategy, the introduction of the VAT may serve a useful role in developing familiarity with the tax.
As regards the best advice for these countries:
-
It is appropriate to temper general support for the introduction of a VAT into smaller economies with a note of caution. The gain is likely to be lower in such countries than elsewhere. In some cases, there may be little gain, or even a loss, relative to a broad-based tariff155 or a retail sales tax.
-
It is important clearly to articulate and fully incorporate these considerations into pre-introduction policy analysis.
-
Articulation is especially important, given the general lack of favor with which tariffs are viewed: the point needs to be made that in the absence of domestic production there is no economic difference between a consumption tax and a tariff.
-
There are a few key magnitudes upon which the comparison between tariff and consumption tax turn: the share of imported consumption in total consumption; the share of intermediates in total imports; and the number of stages in domestic consumption. The potential gains from introducing a VAT can be quantified, at least roughly, and it would be useful to do so. Less tangible arguments—such as the wisdom of building a tax system suited for a more sophisticated future—are important, but do not reduce the value of simple calculations.
-