Financing the United Nations: international taxation based on capacity to pay.

Broadbent, Simon

Introduction

The United Nations faces a financial crisis because some of its members fail to pay their contributions on time. By far the most important of these is the United States, which is responsible for more than a quarter of the budget and seeks a reduction in this share.

The UN's regular budget (1995) is $1.2 billion a year - half that of the Metropolitan Police - and peacekeeping costs another $3.5 billion. In total this is about 0.025 per cent of world income - far below national tax rates (typically 20-40 per cent) and other international taxes, such as the European Union's 1-2 per cent.

These costs are shared among member countries by a complex formula which is supposed to reflect their capacity to pay (CTP). Historically poorer countries get a large discount. All but 24 countries get a further discount on peacekeeping costs at the expense of four of the five permanent members of the Security Council, which, like several others, also contribute in kind.

The effective tax rate for individual countries ranges from 0.001 per cent of historic income, through 0.03 per cent for the main contributors, to as much as 0.3 per cent for a few micro-states. The range is even wider in relation to present day income. In cash terms, the poorest and largest countries pay as little as 0.1 cents per head to the UN each year, while the richest pay more than $7 a head. The US and Europe both contribute about $5 a head. Several countries' contributions have not risen in line with their growing prosperity. Singapore and Israel, for example, pay about $1 a head, despite higher income per head than Ireland and New Zealand, which pay $3 a head.

Since the sums involved are trivial for all but a handful of countries, it is clear that unwillingness rather than incapacity lies behind the failure to pay contributions on time. There are several reasons for this, including scepticism about the UN's role, criticism of its efficiency, the wish to limit its total expenditure, and a belief that the system for sharing the financial burden is unfair. The politics of these issues are discussed in a recent article by Hannay (1996). This system is supposed to be based on 'capacity to pay'. This article examines what this means, how it can be applied in practice, and how far it is reflected in the current system. It draws extensively on the Report of an Intergovernmental Working Group (UN(1995)) established by the UN General Assembly, of which the author was a member. The views expressed here are however his own.

Background

Under Article 17(2) of the UN Charter the General Assembly has the duty of apportioning its expenses among Members. In 1946 the General Assembly accepted the recommendation of the Preparatory Commission (UN((1946)) that this should be done 'broadly according to capacity to pay'. This principle has been reaffirmed in successive resolutions, most recently on 23 December 1993 when it set up the Ad Hoc Working Group to review 'all aspects of the scale methodology with a view to making it stable, simpler and more transparent, while continuing to base it on reliable, verifiable and comparable data....the principle of capacity to pay as the fundamental criterion in determining the scale of assessments'. The Preparatory Commission noted that it was 'difficult to measure capacity merely by statistical means, and impossible to arrive at any definite formula. Comparative estimates of National Income would appear prima facie to be the fairest guide.' Other factors to be taken into account to prevent anomalies included income per head, temporary wartime dislocation and ability to secure foreign currency. It warned against the dangers of Members seeking both to reduce their contributions unduly and also to inflate them for reasons of prestige. The responsibility of recommending the actual scale of contributions from year to year was given to an expert standing Committee on Contributions reflecting geographical balance etc, but the General Assembly has from time to time given the Committee specific instructions which have not necessarily been consistent with CTP.

The scale of assessment for the regular budget was established as a share rather than a rate of contributions. This inevitably led to a focus of attention on the result of the system of assessment rather than its rationale. Initially there was little dispute that national income should be the basis of assessment. All but three members of the United Nations had their shares reduced by up to 40 per cent because their income per head was below a threshold of $1000 a head. As a result the United States, faced an assessment of 49.89 per cent, and it sought a ceiling of one third. This was implemented in 1954 after a period at 39.89 per cent. A minimum contribution share of 0.04 per cent was also established at the outset. In 1974 this was reduced to 0.02 per cent and in 1977 to the present 0.01 per cent of the regular budget, i.e. $120,000 in 1995.

Over the years numerous other modifications to the system were introduced, usually at the instigation of members who believed themselves to be bearing an unfairly high share of the financing burden. A detailed account is provided in Chapter 2 of Renniger et al. (1982). National income was criticised as taking insufficient account of the position of developing countries, and proposals were made to adjust for non-monetary income, accumulated wealth, remittances, basic needs, socio-economic indicators, depletion of natural resources, environmental effects and debt repayment. Only the last of these adjustments was implemented. However the degree of relief for countries with incomes below the average was steadily increased.

Although - indeed partly because - the scale of assessments was in principle fixed for three year periods from 1955, complaints from oil producers about the sharp increases in their assessments in the 1970s led to further measures to increase the inertia in the system, first by lengthening the reference period used for the calculations from 3 to 7 to 10 years (and then back to the present 7.5 years) and later by a 'scheme of limits' to limit changes in any one year.

Finally, there has long been concern over the exchange rates used to convert national incomes to the common US dollar basis used for calculating the scale and for making payments. Historically, this concern arose mainly in relation to non-market determined exchange rates, typically in countries with high domestic inflation. Latterly it has been suggested that exchange rates derived from purchasing power parity (PPP) calculations would provide a more satisfactory basis than market rates for comparing countries' capacity to pay.

It should be noted that constraints such as the floor, the ceiling, longer base periods and the scheme of limits arise from political or practical concerns such as avoiding dominance by a single country, financial convenience etc, and that all of them constitute potential violations of the principle of CTP. Evaluating the extent of this, however, requires first that CTP is defined without such constraints.

The present system

The scale of assessment is expressed in percentage terms to two decimal places (i.e. 10,000 points of 0.01 per cent of the total), using a base period of 7.5 years (the mean of 7 and 8 years) ending three years before the year in which the scale comes into effect; it is then fixed for three years. Thus the scale for 1995-97 on data for a reference period covering 1985-92. The starting point is national income (i.e. net national product - GDP plus net factor income from abroad less depreciation of fixed assets) in national currency at current prices, as reported by member countries (or, where data are missing, estimated by the UN Statistical Office on the basis of the data that are available). These figures are converted to US dollars, normally at the market exchange rates for the relevant years, and summed to give world income. A debt burden adjustment (DBA) is then applied to member states with per capita incomes below $6000, on the basis of the notional cost of repaying outstanding long-term public and private debt over 8 years; the cost of this relief is absorbed pro rata by all countries.

A low per capita income adjustment (LPCIA) is then applied to member states with a per capita income below the world average (the 'cut-off' - $3198 in 1986-92) of 85 per cent (the 'gradient') of the percentage by which the member's per capita income falls below the world average. The cost is absorbed pro rata by countries above the cut-off. Country shares below 0.01 per cent are then raised to this floor, and the share of least developed countries and of the largest contributor reduced to the floor and ceiling respectively. The net cost of these processes is redistributed among the remaining countries. The complex 'scheme of limits' is then applied to constrain changes from one scale and another within predetermined levels, and the net cost of this is redistributed on the basis that the scheme of limits is not violated as a result. The result is called the machine scale and is shown in Table 1 for a selection of countries affected in different ways by these adjustments. Finally, there are discretionary adjustments ('mitigation') to the machine scale financed by voluntarily higher assessments. At each stage the costs or savings arising from the various adjustments are redistributed to ensure that the total shares sum to 100 and thus that the agreed budget is financed - so long as member states actually meet their assessments.

[TABULAR DATA FOR TABLE 1 OMITTED]

For example, China's income in the base period ($400 per head after the debt burden adjustment) was 87 per cent below the cut-off, and so received a LPCIA of 74 per cent (i.e. 85 per cent of 87.5 per cent). This was further reduced as a result of redistributing the savings arising from the imposition of the floor on, for example Vanuatu and Malawi, then increased to allow for the ceiling applied to the United States, and reduced again as a result of applying the scheme of limits. Malawi, on the other hand, received no LPCIA despite an income per head 95 per cent below the cutoff because Malawi's income share was already below the 0.01 per cent floor.

Peacekeeping

Peacekeeping operations - which in 1995 cost nearly three times as much as the regular budget - have been financed by a separate system based to some extent on the regular scale. For this purpose member countries are divided into four Groups, based on their political and economic status on joining the UN, although members may voluntarily 'graduate' into a higher group. Group D (broadly the poorest countries) are assessed at 10 per cent of their regular budget share. Group C at 20 per cent and Group B at their regular budget shares. Group A (the five permanent members of the Security Council) bear the cost of the 80-90 per cent relief given to Groups C and D as a surcharge of on their regular budget shares, which in 1995 worked out at about 25 per cent. Peacekeeping expenditure in 1996 is likely to be lower then in 1995, reducing the surcharge to about 15 per cent.

Alternative systems

Other international organisations use a variety of systems based in varying degrees on capacity to pay and/or value of services received; these are surveyed in Renniger et al. (1982). Other UN institutions in the UN system, with combined budgets of about $1.6 billions (i.e. a third more than the UN's regular budget) follow the UN system exactly, after adjusting for differences in membership. The IMF, World Bank and European Union all use systems based in part on income, but differ in two important respects. First, other considerations, such as voting strength, are more explicitly a consideration as well. Second, they have their own sources of revenue which to some extent bypass the need to obtain subventions from national governments annually. It is sometimes suggested that the UN should have its own source of revenue; possible candidates are discussed in ODI (1996), but it seems unlikely that national Treasuries would readily relinquish either the revenue or the degree of control which thus would imply.

Capacity to pay

Principles

As the Preparatory Committee recognised, capacity to pay is not a precise concept susceptible to exact measurement. It includes normative elements. Should it be based only on income when ownership of capital assets (whether accumulated investment or natural resources) manifestly increases a country's command over resources as compared with one without these advantages? To what extent should it give priority to satisfying the basic needs of the populations of very poor countries? Most domestic tax systems contain an element of progressivity by which richer taxpayers are supposed to pay a higher proportion of their income in tax, usually because the first tranche of income is free of tax. Should countries with balance of payments or domestic budgetary difficulties pay less than those that have, through better luck or better policy, avoided these problems?

Although the calculations are based on more or less objective data, the answers to all of these questions are to some extent a matter of judgment. This leads some to suggest that it is impossible to improve on the accumulated wisdom of the Committee on Contributions and the General Assembly as embodied in the agreed scale. However, some general principles can be suggested against which to test the results of this process. Horizontal equity requires that countries with the same income per head (on whatever definition is agreed) should be assessed for the same contributions per head. Progressivity requires that no country's per capita contribution should be greater than that of another country with a higher per capita income. (Strictly speaking, this is a principle of non-regressivity.) Although these two tests are relatively weak and innocuous - they say nothing, for instance, which would constrain the gradient or cut-off in the LPCIA - they are viewed with suspicion by some who fear that they may be a potential lever for increasing their own contributions. Two alternative extremes would be consistent with these principles - either for the country with the highest income per head to bear the entire burden of UN financing, or for members to pay in strict proportion to income shares, without any relief for low income.

Another consideration is that the degree of refinement attempted in the calculation should have regard to the accuracy, availability and comparability of the data to be used. In many of the elements of the calculation there is a conflict among theoretical correctness, accuracy of measurement and timeliness. The object is to derive a measure that is as simple, transparent and objective as is consistent with reasonable equity.

Basic income measures

Capacity to pay can be defined in terms of income and/or wealth, or expenditure. The various income measures commonly used and the relationship between them is set out in Table 2, together with an indication of how they are derived and their reliability. From the beginning, the UN has used Net National Product, apparently on the grounds that this best approximated to the Hicksian notion of sustainable consumption. In fact, as Sefton and Weale (1996a) demonstrate, it does no such thing. However, in the context of international comparison, the relevant measure is the level of spending which can be undertaken after allowing for the investment necessary simply to maintain the capital stock - i.e. depreciation. NNP doesn't measure this either, because it excludes net transfers from abroad, which are also available to be spent; including them gives net national disposable income; estimates are not easily available, but in all but a handful of countries the difference is small. A more important source of error is the fact that measurement of depreciation is highly unsatisfactory - it is a notional adjustment. As a practical matter the gross equivalent of national income, GNP, is preferable; it is simple, widely available, data-based, and comparable across countries.

Does the omission of accumulated capital or depletion of natural resources invalidate these measures of spending power? The main reason for omitting any further [TABULAR DATA FOR TABLE 2 OMITTED] adjustment in respect of accumulated capital is that the income generated by this capital is already incorporated into the income measure - in the form either of domestic income arising from investment at home or as factor income from investment abroad. The income figures also include the notional income arising from owner-occupation of real property. Like depreciation, this is roughly proportional to the relevant component of the capital stock and, as such, hard to measure and to some extent arbitrary.

As for depletion of non-renewable natural resources, it can be argued that this should be treated in the same way as depreciation of fixed capital assets. Indeed the new 1993 System of National Accounts (SNA) implies this. However, there are large practical and conceptual problems in making such estimates (particularly over valuation) and credible figures are not available. Similar problems arise over adjustments relating to environmental degradation. Moreover, the SNA treats exports of natural resources incorrectly (Sefton and Weale (1996b)).

One capital adjustment which has, however, been incorporated into the UN's income measure is the debt burden adjustment (DBA). The rationale for this is that debt repayments are a particularly serious problem for many developing countries, reducing their capacity to pay. Debt interest obligations are of course already subsumed in the GNP measure (incidentally on an accruals basis, i.e. whether or not the payments are actually made). The DBA of 12.5 per cent of the outstanding stock of certain categories of debt deals with the notional cost of repaying this debt. Since such debt is typically refinanced, it is hard to justify singling out this element of the capital account for special attention. Moreover, the income qualification applying to the DBA differs from that applied to the LPCIA, both in amount and definition. In practice the DBA makes very little difference to the scale: the $10 million a year saved by the beneficiaries is a tiny fraction of the relief granted annually through the Paris and London Clubs.

An alternative approach which avoids many of these difficulties is to use an expenditure measure, on the grounds that what is actually spent is identically a measure of capacity to do so and, for small amounts at least, of taxable capacity. The appropriate measure would be Final Domestic Demand - GDP minus inventory change and exports plus imports. Indeed, inventory change could be excluded since its trend value is typically a very small proportion of GDP and closely correlated with it. Measurement is thus relatively straightforward, since GDP and the goods and services balance are more widely available than the income measures discussed above. It is unnecessary to worry about capital adjustments: a country spending above its current income may increase its current capacity to do so by net borrowing abroad or by eating into its capital stock or natural resources, but at the cost of reducing future capacity. In due course this will be reflected in lower spending than would otherwise have occurred. These arguments for an expenditure measure are similar, but not identical to the familiar arguments for domestic expenditure taxes (discussed, for example, in Kay and King (1983) pp. 70-73). The main attractions are relative precision of measurement, incorporation of benefits arising from use of capital and ease of application progressively; the last two are more difficult to handle domestically. On the other hand, minimising economic distortions is not really an issue at the rates under discussion in the UN context, as it is domestically.

Alternative income measures

Apart from the various adjustments discussed above, it has been suggested from time to time that conventional national accounts do not adequately reflect real differences among countries. Historically one of the main reasons for this was that the accounts of centrally planned economies (CPEs) were drawn up on a Net Material Product basis, with different concepts of coverage and valuation from the SNA. An exposition of the main differences is given in IMF et al. (1991), pp. 137-141. Following the reforms in these countries, the SNA has become the standard, and these difficulties have begun to fall away. It is clear that in SNA terms income in the CPEs tended to be overvalued, although the exchange rate and prices rather than coverage appear to have been the main reasons for this. Two factors perpetuate the effects of this overvaluation; the present base period still includes several years valued on the old basis; and, by implementing the 1993 SNA, with its wider coverage, at the outset, CPE accounts include some categories of income not appearing in other countries' figures.

That said, the uneasiness over using the SNA continues to be manifest in suggestions that it would be better to compare countries on the basis of various real indicators such as levels of health, literacy and/or real income as measured by purchasing power parity (PPP). An example of this is the composite 'Human Development Index' appearing in UNDP (1994), illustrated in Table 3. Such composite socio-economic indicators (SEIs) are inevitably arbitrary with respect to the items included, their weights and their variance. They are often strongly correlated with one another and with GNP; indeed, indicators have sometimes been used in order to make GNP estimates where none is available (e.g. Beckerman and Bacon (1967)).

It is not clear whether SEIs are supposed to modify the basic income measures or to supplant them. If it is the former, this would either involve double counting (since GNP reflects all or some of the factors in the SEIs) and/or calling into question the LPCIA, which is supposed to allow for the special problems of poorer countries. It would be surprising if advocates for less developed countries envisage the latter, since SEIs typically differentiate income levels less sharply than SNA measures.

Table 3. The human development index

The HDI consists of the mean of the following components, scaled to
1 with a fixed maximum and minimum.

 Weight Min Max

Life expectancy (years) 33.33 25 85
Adult literacy (%) 22.22 0 100
Schooling (mean) (years) 11.11 0 15
Real income/head (ppp$) 33.33 200 40000

Some examples:

 HDI HDI GNP/
 value rank head/
 rank

Japan 0.929 3 3
US 0.925 8 9
UK 0.919 10 19
Singapore 0.836 43 21
Russian Federation 0.858 34 48
China 0.644 94 143
India 0.382 135 147
Malawi 0.260 157 156

Source: (UNDP(1994) Tables 5.1 to 5.3).

By contrast, using PPP does not involve a departure from SNA accounting concepts, but instead revalues the various components of national income on a common price basis ('international dollars' - the average worldwide price in US dollars for each detailed expenditure component) in order to compare volumes rather than values across countries, in much the same way, conceptually, as real rather than nominal GNP growth rates are compared over time within countries. A PPP exchange rate can be derived by dividing dollar GNP obtained in this way by GNP expressed in a country's national currency at current prices.

Summers and Heston (1991) and others argue that, in principle, such volume figures provide the best basis for measuring countries' shares of world income. It should be noted, though, that 'international dollar' prices approximate to those prevailing in middle income countries such as Turkey and Hungary. In richer countries, the relative price of services to tradeable goods tends to be higher, while the reverse is true in poorer countries (op tit. Table 1). This has two implications. First, the volume measures may be less satisfactory in countries in which relative prices differ from the international dollar relative prices because such countries might display different expenditure patterns if they had a price structure closer to the one assumed. This could either increase or decrease their measured real incomes (see Marris (1984)). Second, as with the SEIs, the difference between richer and poorer countries will tend to be narrower when measured in real terms rather than in nominal terms, as Table 4 implies - the richer countries have lower income shares when income is measured in real rather than nominal terms.

Table 4. Income shares in real and nominal terms, 1992

 per cent of world GDP

 Nominal(a) Real(b)

Industrial countries 74.0 54.6

United States 25.2 22.4
Japan 14.1 8.0

European Union 26.7 18.6
Germany (c)6.9 4.3
France 5.1 3.6
Italy 4.8 3.4
United Kingdom 4.4 3.3

Developing countries 18.3 37.0
Africa 1.7 4.1
Asia 7.2 19.8
NICs 2.3 2.7
Middle East/Europe 4.8 4.6
Western Hemisphere 4.6 8.5

Former Soviet Union 6.4 6.2
Central Europe 1.4 2.2

Notes:

(a) Calculated at market exchange rates.

(b) Calculated at purchasing power parities.

(c) West Germany only.

Source: Guide and Schulze-Ghattas (1993): table 3.

Although PPP based income figures may provide the best theoretical measure of world income shares, this may still not be the appropriate basis for UN assessments. These are calculated and paid in foreign currency (i.e. US$) obtained (for convertible currencies) at market exchange rates (MERs) which reflect the domestic resource cost of securing that foreign currency. For the poorer country, obtaining a unit of foreign currency by selling traded goods will be more costly in terms of the volume of services income foregone than for the richer country. They do not have the option of paying their UN contributions by buying US dollars at a PPP exchange rate.

In practice, PPP data are simply not available quickly enough for sufficient countries to allow their use in the UN scale - even if this were desirable. The monumental International Comparisons Programme(described in UN (1992)) covers a growing but varying range of countries, but there are still only 80 - less than half the UN membership - and figures are available only at five yearly intervals, well in arrears.

Base Period

The most accurate measure of capacity to pay in a given period would be based on income in that period, even though this can only be calculated some time later. This is, indeed, the practice of many national tax authorities, which collect most of the income tax due in the year of assessment on the basis of provisional estimates, with later adjustment when fuller information becomes available. In the absence of such a system at the UN, the most recent single year for which GDP data are available (normally after one year) provides the best estimate of current capacity to pay. Waiting a year or more for revised figures cannot be expected to improve overall accuracy. Revisions are normally no more than 2 per cent and not consistently above or below first estimates; against this, many countries' GDP can grow (or fall) by 5 per cent or more in a year. Including earlier years in the base period obviously worsens the latter distortion. It is a fallacy to suppose that averaging the data for a run of years improves the accuracy of the estimates: it simply makes them more out of date. Thus the 1995-97 scale, based on the average of 1985-92 and 1986-92 will be more than nine years out of date by 1997, even after the scheme of limits is eliminated.

There is more reason to consider a longer base for the exchange rate used to convert GDP estimates from national currencies to a common currency. Market exchange rates fluctuate considerably from day to day, and it is hard to dispute that in any given year the MER may have departed from its 'underlying' value; but it is even harder to determine what this 'underlying' value may be. For this reason, it is usually assumed (e.g. by the IMF and by national finance ministries) that, for convertible currencies at least, the actual rate is the best indicator of the underlying rate. Using single year MER figures to convert single year GDP figures might result in a degree of fluctuation in assessments. If it were thought desirable to reduce this, there would be the option of using the MER for the year before and the year after as well as the year in question to get a better idea of the underlying rate. The mean of the three years would be the simplest way to do this. A more satisfactory method would be to adjust the two surrounding years for price changes in the country in question relative to world dollar prices.

In practice the Working Group recommended reversion to a three year base period for GDP, with each year's figures converted at MER for that year - i.e. the system in force throughout the 1960s; this offers a reasonable approximation to current capacity to pay (albeit with a lag of some 3-5 years if assessment shares are held constant for three years at a time). This will result in a one-time shift in assessments of about 1.5 percentage points in all, once the scheme of limits has been removed. Rolling the calculation forward by a year each year would cut the lag to a consistent three years and would reduce the extent to which assessments changed in any one year.

Conversion to a common currency

Following the discussion above, it is clear that conversion of national GDP numbers should normally be made at market exchange rates (MER). Contributions are paid in foreign currency (i.e. US$), and, for countries with convertible currencies MERs, reflect the cost in domestic resources of securing that foreign currency. If payments were made at the same exchange rates as those used for assessments, it is irrelevant that the currency may be temporarily misaligned. However, there is in practice a lag of several years between the base period and the years of assessment and payment. During this period a relatively fast growing country can be expected to have a rising real exchange rate, compounding the measurement error arising from the use of out-of-date figures. There may, too, be changes in a country's price level relative to others'; these might be expected to be largely offset by exchange rate changes over time, but this may not be the case in particular periods.

These problems become more acute in the case of non-convertible and non-market determined exchange rates. When there is rapid inflation in a country, exchange rate movements may not be synchronised with domestic price changes, with the result that international income shares may be seriously distorted. One solution is to use the UN's Price Adjusted Exchange Rates (PARE), which maintain the real rate constant at the rate in a 'normal' base year. However, this is critically dependent on identifying a 'normal' year, not to mention adjusting for changes within a year. Moreover, it denies the possibility of change in the real exchange rate. Another possible approach is to use PPP rates - where they are available - as a consistency check to identify cases of misalignment, since there appears to be a strong correlation and rank correlation between income estimated on a PPP and MER basis. A typical relationship, using IBRD (1995) data for 128 countries is:

[y.sub.ppp] = 3.258 + 0.674[y.sub.mer] [R.sup.2] = 0.948 (13.0) (48.0)

where y is the logarithm of income per head using PPP and market exchange rates respectively. In the case of countries for which there are good PPP data and a questionable exchange rate, an estimated value for the MER might be inferred from the equation. However, in such cases the relevant observations would have to be excluded from the dataset and the equation re-estimated.

An SDR numeraire?

Some would like to calculate the scale in terms of SDR rather than US dollars. Whilst this seems logical at least in presentational terms, the choice of numeraire should not make any difference because the cross rates should be mutually consistent. This may not hold exactly in practice, if the annual series are arithmetic averages of monthly or daily data. Even if the annual figures are consistent, taking three year averages of income shares using a numeraire which is weakening in relation to its rivals consequently gives greater weight to the most recent year's data and vice versa. An example is given in Table 5, which assumes two countries, Japan and country X with incomes of 2500 yen and $100 dollars each year respectively. The Japanese share in each individual year is the same whether calculated in SDR or $, but is marginally higher when calculated in $ for the three years taken together because the $ is weaker in 1992.

The Low Per Capita Income Allowance (LPCIA)

The rationale for the CTP appears to be twofold; it can be regarded as a general catch-all for the items excluded from the basic income measure, such as sustainable development, environment considerations etc; and it gives effect to of the principle of progressivity. The adjustment, which has been a feature of the UN scale from its inception, currently benefits some 132 UN members with per capita income below the (weighted) average during the reference period. They accounted for over 80 per cent of the world's population but less than 20 per cent of its income as measured by the UN. The effect of the adjustment is to reduce their contribution share from this figure by 8.9 percentage points, the cost of which is borne, pro-rata, by a surcharge on the contributions of the remaining countries (except the United States, which benefits from a ceiling on its contributions, also at the expense of the remaining countries). Before 1979 the cost was spread over the whole membership, including LPCIA beneficiaries. The effect of the LPCIA before and after this change is illustrated in Chart 1, which shows contributions per head at various levels of income per head. Chart 2 expresses the same information as an effective tax rate. A curious feature of the post-1979 system is the jump in contributions at the cut-off, where members pass from being beneficiaries of the LPCIA to benefactors. The pre-1979 system avoided this, and with it the risk that small data errors might put a country on the wrong side of the threshold.

However, the key parameters of the LPCIA are the level of the cut-off and the gradient. The principle of CTP does not point to any particular values, although it is obvious that the higher the cut-off, the more widespread will be the benefit and the narrower the remaining tax base available to finance it. The original cut-off was relatively high ($1000 in 1946), although its overall cost [TABULAR DATA FOR TABLE 5 OMITTED] was kept down by a 40 per cent gradient. The cut-off subsequently fell in real terms and was then increased from time to time in line with US inflation. Latterly it has been set at (weighted) world average income, implying more or less perpetual redistribution to the existing beneficiaries, irrespective of their growing prosperity. An alternative would be to revert to the earlier practice of maintaining the cut-off in real terms, but using a more appropriate deflator than US inflation, such as the implicit world price deflator obtained by dividing the growth in world nominal income by growth in real income as measured in PPP terms. This would allow countries to 'graduate' and concentration of relief on the poorest. The gradient has risen from 40 per cent to 85 per cent over the years, with the result that the contribution rates of several significant countries are negligible - in many cases far below what they spend on maintaining their New York missions. This is, however, an essentially political rather than a CTP consideration.

Finally, it is of some interest to consider the possible effects of applying the existing LPCIA system to a dataset expressed in PPP terms. Table 6 compares the effects of using MER and PPP data from the World Bank data referred to above, with missing observations estimated from the equation. Column (1) shows the effect of applying the present system. 122 countries lie below the $4569 average income cut-off, accounting for 82 per cent of population but only 16 per cent of total income. With a gradient of 85 per cent their combined LPCIA reduces their income share by 51.5 per cent to 7.8 per cent of world income. Financing the LPCIA would require a 9.8 per cent surcharge on the countries above the threshold.

Table 6. Effects of low per capita income allowance using ppp

 mer [ppp.sup.*] ppp
 US$ int'l $ int'l $

 (1) (2) (3)

cut-off (US$) 4569 7635 5697
no. of countries below cut-off 122 116 105

in countries below cut-off:

% of total income 16.0 36.6 32.5
% of total population 82.0 80.8 77.3
LPCIA as % of income (g=.85) 51.5 44.0 38.0
LPCIA as % of world income 8.2 13.7 10.5
share of total income after LPCIA 7.8 22.9 22.0

in countries above cut-off:
% surcharge required to finance LPCIA 9.8 21.6 15.6

Note:

(1) Converted to US$ at market exchange rates

(2) Converted to US$ at ppp: line 1 = ppp equivalent of mer average.

(3) Converted to US$ at ppp: line 1 = average of ppp figures.

Data source: IBRD (1995).

Column (2) assumes that the cut-off is set at $7635 - the equivalent point in PPP terms to $4569 in MER terms, as given by the equation above. If the relationship between PPP and MER was exact, the same number of countries would lie below the cut-off in each case. In fact, six of them now lie above it. Moreover, because the PPP-based income distribution is more even than the MER one, countries below the cut-off now account for 36.6 per cent of total income. For the same reason, the combined LPCIA is a smaller percentage of these countries' income, but a larger percentage of total income, requiring a large surcharge to finance it. Column (3) assumes a cut-off set at the PPP based average income of $5697. Only 105 countries lie below this point, and benefit from a lower percentage LPCIA.

The floor and ceiling

The case for a floor and ceiling is primarily a political, not a CTP matter. The largest violations of CTP result from the floor, and could largely be eliminated by calculating the scale to three decimal places and lowering the floor to 0.001 per cent. Against this, the political case for a floor is that the smallest countries have one vote as do the largest.

Conclusions

Do UN contributions 'broadly reflect CTP'?

It is clear from Table 7, which is arranged in order of income/head in a three year base period 1990-92, that several countries with similar incomes are taxed at widely different rates, and that several countries are assessed at much higher rates than others above them in the table. This is particularly pronounced in the peacekeeping scale, where the main reason for the anomalies is the group system which allows several middle-higher income countries to make exiguous contributions. Table 8 indicates the main sources of the anomalies in the regular scale. Column 6 shows the wide variation in actual contribution shares in relation to GNP shares, resulting primarily from using a distant base period. Column 7 shows the effect of rebasing the 1997 scale to the three year base period recommended by the Working Group, without, however removing the scheme of limits. This largely vitiates the effect of the reform. Column 8 shows the effect of using the 1990-92 base, and makes it clear that the scheme of limits has been the key source of distortion.

The figures in column 8 also show that on this basis the system would largely satisfy the criteria suggested in the discussion above. The tax ratio falls for poorer countries, as a result of the operation of the LPCIA. There remain, however, two major anomalies which violate CTP. The US would pay much less as a percentage of income than all other countries above the LPCIA threshold. And the floor catches small countries. Neither Malawi nor Vanuatu benefit at all from the LPCIA, despite (in the former case) being one of the poorest countries in the world.

Thus the inescapable conclusion is that the scale is not consistent with CTP. So far as the regular budget is concerned keeping the scale up to date by shortening the base period to three years, eliminating the scheme of limits, eliminating the ceiling and the floor would restore the consistency with CTP. Whether it would eliminate members' reluctance to pay the resulting assessments is less clear. As for the peacekeeping scale, the key anomaly is the group system; only if this were based strictly on a specified range of income per head for each group could it be made consistent with CTP.

The largest source of difference (as distinct from anomaly) in tax rates is the LPCIA. It could certainly be modified (for instance by changing the gradient and/or cut-off) without destroying the principle of progressivity, while remaining consistent with CTP. At a technical level, an unsatisfactory feature of the LPCIA is the discontinuity by which its cost is borne only by countries above the cut-off; restoration of the original system would be one way to deal with this. However, the main argument for any change in the LPCIA is inevitably the political question of whether the UN's and its members' interests would be better served by increasing significant ldcs' engagement and influence by making larger contributions to its finances. But the political will for this seems to be lacking at present.

[TABULAR DATA FOR TABLE 7 OMITTED]

[TABULAR DATA FOR TABLE 8 OMITTED]

REFERENCES

Beckerman, W. and Bacon, R. (1967), 'International comparisons of income levels: a suggested new measure', Economic Journal, vol. LXXVI, September, pp.519-536.

Gulde, A.M. and Schulze-Ghattas, M. (1993), 'Purchasing Power Parity based weights for the World Economic Outlook' in Staff Studies for the World Economic Outlook, IMF, Washington, December 1993.

IBRD (1995), World Bank Atlas 1995: Washington.

IMF, World Bank, OECD and EBRD (1991), 'A study of the Soviet economy vol. 1', OECD, Paris, February.

Hannay, D.H.A. (1996), 'Paying for the UN. A case for emergency treatment' The World Today, June 1996.

Kay, J.A. and King, M.A. (1983), The British Tax System, third edition: Oxford University Press, Oxford.

Marris, R.L. (1984), 'Comparing the incomes of nations: a critique of the international comparisons project', Journal of Economic Literature, vol. XXII, March 1984, pp.40-57.

ODI (1996), 'New sources of finance for development', Overseas Development Insitute Briefing Paper, London, February 1996.

Renniger, J.P., Donaghue, D., Geib, P., Perez, H., Sevenaer, W. van N. (1982), 'Assessing the United Nations scale of assessments', Policy and Efficiency Studies, no.9, UNITP, New York, 1982.

Sefton, J.A. and Weale, M.R. (1996a), 'Real national income', National Institute Economic Review, no. 155, February, pp. 90-97.

Sefton, J.A. and Weale, M.R. (1996b), 'The net national product and exhaustible resources: the effects of foreign trade', Journal of Public Economics (forthcoming).

Summers, R. and Heston, A. (1991), 'The Penn World Table (Mark 5),' Quarterly Journal of Economics, vol. CVI, May 1991, pp. 327-368.

UN (1946), Report of the Report of the Preparatory Commission of the United Nations),United Nations: New York.

UN (1992), Handbook of the International Comparison Programme, Studies in Methods series F no.62, United Nations, New York.

UNDP (1994), Human Development Report 1994, Oxford University Press, New York.

UN (1995a), Report of the Ad Hoc Intergovernmental Working Group on the Implementation of the Principle of Capacity to Pay (A/49/897), United Nations, New York: 8 May 1995.

UN (1995b), 'Report of the Committee on contributions 1995', Official Records of the General Assembly, Fiftieth Session, Supplement no. 11, (A/50/11), United Nations, New York.

UN (1995c), Working Group on the UN Financial Situation, (WGFS/19-19A), July.