HVWXGLRV VREUH OD HFRQRPLD HVSDÔROD

The redistributive effects of the EU budget: An analysis and some reflections on the Agenda 2000 negotiations

Angel de la Fuente Rafael Doménech

(((

http://www.fedea.es/hojas/publicado.html

Angel de la Fuente

Instituto de Análisis Económico (CSIC)

and

Rafael Doménech*

Universidad de Valencia

March 1999

Abstract

This paper analyzes the redistributive effects of the EU budget during the period 1986-97 and advances a proposal for setting member countries' budgetary balances in a manner consistent with a simple equity criterion based on relative income per capita.

Keywords: European Union, fiscal flows

JEL Classification: H11, H77, H87

* We would like to thank Juan Varela, Antonio Maudes and Alberto Cerdá for their helpful comments. We gratefully acknowledge financial support from the European Fund for Regional Development (through the projects "Determinants of growth at the regional and national level" and "Economic growth, productivity and labour markets in the Spanish regions") and from the Spanish Ministry of Education (under grants DGICYT PB95-0130 and CICYT SEC96-1435).

Correspondence to: Angel de la Fuente, Instituto de Análisis Económico, Campus de la Universidad Autónoma de Barcelona, 08193, Bellaterra, Barcelona, Spain.Tel: 34-93-580-6612. Fax: 34-93-580-1452. E-Mail address: delafuente@cc.uab. es.

Non-technical summary

The current negotiations on the European Union's financial perspectives for the period 2000-2006 (the so-called Agenda 2000) have brought out into the open wide discrepancies among EU members concerning budgetary matters. In the months preceding the 1999 Berlin summit, discussions have focused on the national and functional allocation of EU revenues and expenditures. While most countries seem to agree on the need to maintain the current ceiling on expenditure (set at 1.27% of the Union's GDP), there is considerable disagreement among them on how this revenue should be raised and spent. The richer members of the EU, led by Germany, have been pressing hard for a reduction of their net contributions, while the poorer ones have tried to preserve their positive balances. Disparate national interests are also clearly discernible in the existing proposals for the restructuring of farm and structural programmes and in the ongoing discussions about their funding. With a weakened Commission and national governments bent on the defense of their narrow interests, progress in building a consensus has been rather slow. In the present situation, there is a real danger that the matter may not be closed within the current year, or that the final agreement may be perceived as unfair by a significant minority of states, thereby weakening the legitimacy of EU institutions.

In this context, the present paper provides a positive analysis of the redistributive effects of the EU budget and advances a proposal for the setting of member countries' budgetary balances in a manner consistent with a simple equity criterion based on the principle of cohesion. In the first part of the paper, we extend a standard methodology for the analysis of the impact of fiscal flows that involves running regressions of appropriately normalized budget items on relative incomes per capita. The estimated coefficients and the residuals of these regressions are then used to construct annual indicators of the degree of redistribution induced by the overall EU budget and by each of its expenditure and revenue items, as well as measures of total and partial national "excess" balances (corrected for income per capita). These indicators, moreover, have some convenient additivity properties and are easier to interpret and to work with than the Lorenz curves often used to describe the redistributive effects of Union programmes.

Our results for the period 1986-97 show that the redistributive impact of fiscal flows across Union members has been considerable, particularly given the small size of the budget, that it has tended to increase over time, and that it is mostly due to the effect of structural programmes. We also find significant differences in net contributions across countries with similar wealth levels. In this sense, the current concern about the unfairness of the distribution of net burdens does seem to be justified, although the list of net beneficiaries that emerges after correcting for per capita incomes is very different from the one based on unadjusted balances.

The regressions we estimate can also be seen as an attempt to extract from the data a simple allocation rule for budget resources that is consistent with the principle of cohesion and presumably reflects the consensus on the desirable extent of redistribution implicit in the status quo. In the second part of the paper we argue for the explicit use of such a rule (linking net balances to relative incomes per capita) as a framework for EU financial decisions and for the adoption of a budget-setting procedure in three stages: The Commission and the member states would first agree on the overall expenditure volume, the mechanisms for its financing and the desirable redistribution coefficient. National budgetary balances would then be set on the basis of per capita income prior to the negotiation of specific expenditure programmes. Any discrepancies between the outcome of the previous stage and the desired net balance would be eliminated by a compensation mechanism that could take the form of a rebate or surcharge on GNP contributions. In practice, we would advocate that this procedure be applied in an incremental fashion, taking as given the allocation rule implicit in the status quo and gradually eliminating "excess" balances or adjusting the redistribution coefficient as required by an evolving consensus.

In our opinion, this approach would have several important advantages over current practices. Perhaps the most important one is that it would provide a transparent way of implementing the principle of cohesion and reduce the scope for real or perceived inequities in the allocation of budget resources by forcing member states to negotiate over a single, easily interpretable, parameter rather than about a host of specific items that may add up to apparently arbitrary allocations. In addition, setting net balances in advance of expenditure programmes will in effect force national governments to bear the marginal cost of community expenditures, thereby increasing the incentives for a more efficient allocation of resources.

In the same section, we also examine the implications of such a rule and compare it with those of some plausible outcomes of the current negotiations. Taking as a reference the situation in 1997 and the "revealed" redistribution coefficient for that year, the elimination of "excess" balances would imply sizable transfers from France (-3,500 Mecu), Ireland (-2,700), Denmark (-900) and Greece (-830) to Germany (+3,900), Spain (+1,750), the Netherlands (+1,450), Sweden (+1,280) and the UK (+900 Mecu). By contrast, a proportional reduction in unadjusted net balances financed by a reduction in either agricultural subsidies or structural spending would involve a very different pattern of financial flows and would significantly reduce the degree of redistribution of the system without improving its horizontal equity.

1.- Introduction

This paper analyzes the redistributive effects of the EU budget during the period 1986-97 and advances a proposal for setting member countries' budgetary balances in a manner consistent with a simple equity criterion. In the positive part of the paper, we extend a standard methodology for the analysis of the impact of fiscal flows that involves running regressions of appropriately normalized budget items on relative incomes per capita. The estimated coefficients and the residuals of these regressions are then used to construct annual indicators of the degree of redistribution induced by the overall EU budget and by each of its expenditure and revenue items, as well as measures of total and partial national budgetary balances corrected for income per capita. The results provide a compact and fairly complete characterization of the redistributive impact of fiscal flows across union members over the last decade. Our indicators, moreover, have some convenient additivity properties and are easier to interpret and to work with than the Lorenz curves often used to describe the redistributive effects of Union programmes.

The regressions we estimate can also be seen as an attempt to extract from the data a simple allocation rule for budget resources that is consistent with the principle of cohesion and presumably reflects the consensus on the desirable extent of redistribution implicit in the status quo. In the second part of the paper, we argue that the explicit adoption of such a rule (relating national budget balances to relative incomes per capita) would be useful in focusing and simplifying EU budget negotiations, in increasing their transparency, and in improving horizontal equity across member states. In addition, our proposal would change the incentives faced by national governments in a way that should improve the efficiency of EU expenditure programmes. Using as a reference the 1997 situation, we explore the implications of the proposed rule and compare the results to those of some plausible outcomes of the current budget negotiations. The exercise suggests that there is plenty of room for redistributing resources across member states in a way that will mitigate perceived inequities without reducing the overall degree of redistribution.

2.- Measuring the redistributive impact of the EU budget

This section describes the methodology used to analyze the effects of fiscal flows across EU states. As a number of authors have done in the related literature,¹ we will estimate a series of regressions of various budget items on relative "gross" income per capita (income per capita before EU transfers and contributions) in order to obtain a "redistribution coefficient" that measures the fraction of the income differential with the EU average which is eliminated by the relevant fiscal flow. The residual for a given country in this regression will measure the deviation from the implicit allocation rule described by the fitted regression line, and will accordingly be interpreted as an indicator of the adjusted "fiscal treatment" of the country under the relevant programme. It should be noted from the outset that we regard these regressions simply as a convenient way to summarize the data, and not as a way to test hypotheses. Hence, we will work only with point estimates and disregard any discussion of statistical significance or intervals of confidence, even though we will be concerned with the dispersion of the residuals as an inverse indicator of the degree of fairness of the allocation, relative to the average degree of redistribution implicit in the estimated equation.

Our starting point will be a decomposition of each country's per capita net budget balance with the Union (nb) as the algebraic sum of a series of per capita transfers and contributions:

\[\mathrm {(1) n b _ {i} = t.tot_ {i} - c.tot_ {i} = (t.str_ {i} + t.agr_ {i} + t.other_ {i}) - (c.vat_ {i} + c.gnp_ {i} + c.own_ {i})}\]

where t stands for transfers or expenditures and c for contributions, both in per capita terms. On the EU revenue side, we distinguish between contributions linked to VAT revenues (c.vat) and national GNPs (c.gnp) and "traditional own" resources, comprised mostly by net tariff revenues (c.own). Expenditures are grouped into three categories: i) structural expenditures (t.str), comprised by the Regional, Social and Cohesion Funds, the Fisheries instrument and the guidance section of the Agricultural Fund, ii) agricultural expenditures (t.agr), which includes farm subsidies under the guarantee section of the Agricultural Fund, and iii) other expenditures (t.other), which includes expenditure under horizontal policies (e.g. R&D programmes), transitory payments to new member states and various miscellanea. This last item also seems to include a small fraction of international cooperation programmes as allocated among member countries by the Court of Auditors. Although in principle the expenditure data exclude the administrative costs of the various Union institutions, on the basis of the results for Luxembourg we cannot exclude the possibility that part of these costs may be included in t.other.

Our budget data are taken from the Annual Report of the European Court of Auditors and are originally expressed in ecus.² These figures are converted into national currencies and adjusted, together with incomes per capita, for differences in purchasing power. The resulting values are then normalized by average EU income per capita and expressed in differences with the corresponding (weighted) EU average.³ Thus, nb_i is the difference between country i's net balance per capita and the average EU per capita net balance, measured as a fraction of average real EU income per capita in gross terms (i.e. before contributions to or transfers from the Union budget). Notice that the average EU net balance will generally be different from zero because i) some EU expenditure is not allocated among member countries and ii) Union payments may differ from EU receipts in any given year (even more than budgeted expenditure) due to time lags in the execution of the budget.

¹ See for instance Sala-i-Martin and Sachs (1992), von Hagen (1992), Bayoumi and Masson (1995), de la Fuente (1997), Martin (1997) and Doménech, Maudes and Varela (1998) on the analysis of fiscal flows within federal or quasi-federal systems. The details of the specification vary. Some authors regress income after tax and transfers (instead of these last two variables) on gross income. Specifications in levels with cross-section data averaged over "long" periods are generally used to try to assess long-term redistributive effects, while specifications in differences with time-series data for each territory are used to determine the stabilization effects of the federal budget.

² See Doménech, Maudes and Varela (1998) for a more detailed description of this data set.

Using the transformed data, we will regress member countries' relative net balances per capita and their various components (total transfers, total expenditures and each of the individual budget items) on relative gross income per capita normalized in the same way as the other variables (i.e. measured in percentage deviations from average EU income per capita in purchasing power units). Since we are interested in redistributive (rather than stabilization) effects, we will run our regressions in levels, using cross section data, and repeat the exercise for each year in the sample period in order to examine the evolution of the system over time. Hence, for each year between 1986 and 1997, we estimate cross-section regressions of the form

\[\mathrm {nb_ {i} = - \rho_ {b} y_ {i} + \varepsilon_ {bi}} \quad (3) \mathrm {t_ {ki} = - \rho_ {tk} y_ {i} + \varepsilon_ {tki}} \quad \text {and} \quad (4) \mathrm {c_ {ki} = \rho_ {ck} y_ {i} + \varepsilon_ {cki}}\]

where y_i is the relative gross income per capita of country i, t_k and c_k are the k-th transfer and contribution items, and ρ the corresponding ("partial") redistribution coefficients and the regression residuals. Since these residuals capture the deviation from the amount that would correspond to each country on the basis of its income per capita under the allocation rule implicit in the regression line, we will refer to them as "adjusted" or "excess" partial balances. Notice that we have specified the regressions so that is positive if and only if the k-th budget item is redistributive, that is, if on average poor countries contribute less to the Union or receive more transfers from it than rich ones. Thus, is the relative net per capita transfer from the EU as a fraction of the income differential with the EU average in a typical country in the sample. The composition of the sample varies over time, reflecting the accession of new entrants and German unification.

Equations (2)-(4) are estimated by weighted least squares (weighting the observations by population size). Failure to weigh the observations would give misleading results due mostly to the effect of Luxembourg (see Figure 1 below). Our normalization of the data (in relative deviations from weighted EU averages) implies that the sum of the weighted residuals in the estimated equations will be zero. Hence, the exercise of redistributing net balances across countries in a way that (approximately)⁴ adds up is simpler than it would be in the unweighted case, where a counterfactual budget designed to eliminate deviations from the fitted line will typically not be feasible (because a reduction in per capita transfers to Ireland by an ecu will not pay for a reduction of the same amount in the per capita German contribution). The definition of our variables also ensures that the regression constant will be zero. This feature is convenient for two reasons. The first one is that it allows us to measure the progressivity of the system through a single parameter with a straightforward interpretation. The second is that, if we interpret the estimated regression as an allocation rule, this rule will specify that the budget should be neutral for countries of average income per capita. This seems sensible if we want to make normative recommendations as to the "fair" allocation of net contributions.

³ See the Appendix for a more detailed description of the normalization procedure and the construction of the variables used in equation (1).

⁴ See footnote 7.

Another convenient feature of our specification is that the partial redistribution parameters (ρ ) associated with the different programs and the adjusted partial balances (i.e. the residuals εik) will add up, across equations for a given year, to the overall redistribution parameter (ρ_b) and the adjusted net balance obtained from the overall redistribution equation.⁵ This property allows us to construct a simple indicator of the contribution of each budget item to redistribution across member states and to trace the sources of a country's net fiscal balance to the effects of specific programmes.

3.- Results for 1986-97

Figure 1 displays the relationship between per capita net budget balances and relative income per capita across the member states of the EU in 1997. The negative slope of the fitted regression line indicates that the net effect of the EU budget is redistributive, as net contributions are, on average, positive in the poorer countries and negative in the rich ones. The estimated redistribution coefficient is 0.0535. Given the definition of our variables, this parameter indicates that, for a typical European citizen, the net effect of the EU budget is equivalent to a flat tax of 5.35% levied on the difference between his gross income (adjusted for purchasing power differences) and the EU average -- or to a subsidy of the same magnitude if his income is below average. Although this figure is very small compared with the degree of regional redistribution achieved by national fiscal systems (which typically lies around 0.30 for European and North American countries), it is achieved with a volume of expenditure only slightly above 1% of GDP (as compared to figures between 30 and 60% of GDP for national fiscal systems). In this sense, then, the EU budget is a rather effective instrument for redistribution among member states.

The figure shows that the net expenditures of the European Union are distributed rather unevenly across its members. While an important part of this variation (around two thirds) is systematically related to per capita incomes, the large deviations we observe from the regression line indicate that countries with similar income levels are often treated very differently. The most obvious outliers are Ireland and Luxembourg, whose net balances are, respectively, 3.9 and 2.6 points higher than they "should" be on the basis of the allocation criterion implicit in the fitted regression line. Moreover, the distribution of adjusted net balances given by the residuals of the regression is very different from that of unadjusted balances. After correcting for income, for instance, Denmark and France display relatively large positive net balances, while those for Germany and Portugal are close to zero and Spain's balance is negative. For ease of comparison, Figure 2 displays the gross and adjusted net balances corresponding to 1997.

⁵ See the Appendix for a derivation of these properties.

Figure 1: Relative net national balances per capita vs. relative per capita incomes, 1997

- Legend: Gr = Greece, Po = Portugal; Sp = Spain; Ir = Ireland; Fin = Finland; Sw = Sweden; UK = United Kingdom; It = Italy; Nl = Netherlands; Fr = France; Ge = Germany; Ost = Austria; Be = Belgium; Dk = Denmark; Lux = Luxemburg. - Note: estimation by OLS with observations weighted by population size. The fitted equation is of the form n

Figure 2: Gross and adjusted relative per capita net fiscal balances with the EU, 1997

- Note: The gross balance is the net surplus per capita expressed as a fraction of average EU income per capita. The adjusted net balance is the residual of a regression of the previous variable on relative income per capita, defined as income per capita corrected for PPP and measured in percentage deviations from the EU average

Figure 3: Evolution of the redistributive impact of the EU budget (overall redistribution coefficient, ρ_b) Figure 4: Evolution of the dispersion of adjusted net budgetary balances

- Note: weighted standard deviation of the residuals of the redistribution regressions for net budget balances.

As we back in time, the pictures we have just examined do not change very much in qualitative terms. The most noteworthy development is the sharp increase in the redistribution parameter over most of the sample period. As shown in Figure 3, the value of increases almost threefold between 1986 and 1993, and levels off thereafter. The evolution of the dispersion of the adjusted net balances, which can be interpreted as an indicator of the "arbitrariness" of budget allocations (relative to the redistributive criterion implicit in the estimated equation), is shown in Figure 4. On average across the period, the typical country is about half a percentage point of per capita EU income away from the net balance dictated by its income level. This figure increases significantly from 1988 to 1993 and declines thereafter.

Figure 5: Gross and adjusted net fiscal balances, averages over the entire sample period

Working with period averages, the distribution of net balances across countries resembles the one observed in 1997, although with some changes in the ranking. As shown in Figure 5, Ireland, Luxembourg, Denmark, Greece and France are the main beneficiaries of the EU budget after adjusting for per capita income. On the other end of the scale, Sweden, Spain, the UK, Finland and Germany display the largest negative balances.

The redistributive effects of the EU budget come from both its expenditure and revenue sides. As shown in Figure 6, where we decompose the total redistribution coefficient into its transfers and contributions components, the first of these parameters was typically much larger than the second one. Indeed, EU expenditures have become increasingly progressive with the growth of structural spending, which being heavily concentrated in poorer countries, is the most redistributive of all budget items. In 1997, this programme accounted for over 60% of the total redistribution effect with about one third of total expenditure. The farm subsidy programme is also redistributive, but only because the weight of agricultural employment is considerably larger in the cohesion countries. Payments per farmer, on the other hand, vary widely across countries and, with the exception of Ireland, do not favour the poorer states. (See Figure 9). EU revenues, on the other hand, are roughly proportional to income.⁶ Hence, they are somewhat redistributive (because the rich pay more per capita in absolute terms), but their impact is quite limited.

⁶ See Doménech, Maudes and Varela (1998).

Figure 6: Decomposition of the overall redistribution coefficient: transfers vs. contributions

Figure 7: Decomposition of the redistribution coefficient for total transfers structural vs. agricultural transfers and other expenditures Figure 8: Decomposition of the redistribution coefficient for total contributions VAT, GNP and own resources

Figure 9: Agricultural subsidies per farm worker, 1997 thousands of US dollars - Source: The Economist, Feb. 20th 1999, p. 30, with data from the European Commission.

Figures 10 and 11 show the sources of national net balances working with averages over each country's period of membership in the Union since 1986. Figures 12 to 20 (included at the end of the text) show the year-by-year evolution of the adjusted total and partial balances for each country in the sample. A few items deserve special comment. The UK's net VAT contribution is very small because a large fraction of it is rebated under the "British check" in order to reduce this country's net deficit with the EU. Holland and Belgium display large positive residuals in their contributions to the Community's "own" resources (c.own) because a disproportionate share of imports from outside the EU comes through their ports, generating a large amount of tariff revenue. Luxembourg, on the other hand, presents exactly the opposite situation (because very few non-EU imports come directly into the country) but displays a very large VAT contribution, presumably because of taxes on financial transactions residenced there for fiscal reasons. Large tariff revenues in Ireland, finally, may be due in part to the importation of intermediate goods by multinationals who have production facilities in this country and export final goods to the rest of the Union.

Figure 10: Adjusted partial fiscal balances: expenditure side of the EU budget averages over each country's period of membership in the Union

Figure 11: Adjusted partial fiscal balances: revenue side of the EU budget averages over each country's period of membership in the Union

On the expenditure side, farm support programs have benefited Ireland, Greece, Denmark, the Netherlands, Belgium and France at the expense of Portugal, Spain, the UK, Sweden and Finland. The Dutch and Belgian figures, however, are probably inflated by rebates on exports shipped through their ports but originating in other Union countries. Adjusted structural expenditure has been highest in Luxembourg, Ireland and Portugal and lowest in Sweden, Finland, Spain, Holland and the UK. Finally, other expenditures have gone mostly to the new entrants as well as to Belgium and Luxembourg. As noted above, the size of these payments makes us suspect that this item may include some administrative expenses.

In summary, our analysis shows that the overall redistributive impact of the fiscal flows across EU members is considerable, given the small size of the Union budget, and that it comes mostly from its structural expenditure programmes. The degree of redistribution, moreover, has increased significantly over time, although it has tended to stabilize in recent years. On the other hand, the distribution of net contributions across countries shows large and persistent deviations from the average pattern of redistribution. To the extent that this pattern embodies a consensus on what may be considered a fair allocation of the fiscal costs and benefits of Union membership, the existence of such large "excess" surpluses and deficits is a serious matter for concern and would seem to lay at the root of the ongoing budget dispute. While it is certainly true that net fiscal transfers capture only a part (and probably a small one) of the benefits of membership in the European Union, they are also the most visible and easily quantifiable aspect of such benefits. As a result, public and government views on the overall fairness of the system are likely to depend critically on the extent to which their allocation is perceived to be equitable. Ensuring that it is so should be an important goal of Community budget policy.

In this context, the analysis we have just conducted may be of interest as a tool for making operational a simple equity criterion derived from the Union's commitment to the principle of cohesion. Our procedure, in particular, allows us to decompose net national balances into two parts: one which may be attributed to wealth differences, taking as given the "revealed" average degree of redistribution in the system, and a residual or adjusted balance that can be interpreted as the deviation from the allocation rule implicit in the regression line fitted to the observed net balances. It seems likely that progress in budget negotiations will be easier to make if these two different components are discussed separately.

3.- Some reflections on the current budget negotiations

The current negotiations on the Agenda 2000 have brought out into the open wide discrepancies among EU members concerning budgetary matters. In the months preceding the 1999 Berlin summit, discussions have focused on the national and functional allocation of EU revenues and expenditures. While most countries seem to agree on the need to maintain the current ceiling on expenditure (set at 1.27% of the Union's GDP), there is considerable disagreement among them on how this revenue should be raised and spent. The richer members of the EU, led by Germany, have been pressing hard for a reduction of their net contributions, while the poorer ones have tried to preserve their positive balances. Disparate national interests are also clearly discernible in the existing proposals for the restructuring of farm and structural programmes and in the ongoing discussions about their funding. With a weakened Commission and national governments bent on the defense of their narrow interests, progress in building a consensus has been rather slow. In the present situation, there is a real danger that the matter may not be closed within the current year, or that the final agreement may be perceived as unfair by a significant minority of states, thereby weakening the legitimacy of EU institutions.

In addition to the unavoidable conflict between national interests, perhaps the most important obstacle to a budget consensus is the lack of a shared and operational equity criterion for the allocation of net budget contributions. One of the objectives of this section is to propose such a criterion and examine its implications in the current context. Our second objective is to explore the implications of various hypothetical alternatives that may throw some light on the likely effects of existing budget proposals.

In essence, our proposal involves approaching budget negotiations in several stages. First, an agreement should be reached on the overall amount of resources available to the Union, on the mechanisms for its financing and on the desired degree of redistribution. Second, indicative net balances should be computed for each of the member states on the basis of their per capita incomes and an agreed-upon redistribution coefficient. Third, budget allocations to specific expenditure programmes should be made on the basis of the relevant policy objectives. Finally, structural spending and, if necessary, compensatory payments or rebates should be set so as to obtain the desired redistributive outcome. The procedure will probably work best if it is applied in an incremental way, i.e. as a tool for identifying the desired direction of change from the status quo.

In the current situation, it is probably sensible to take as given the overall ceiling on expenditure and the current degree of redistribution and concentrate on the gradual elimination of "excess" balances. The rationale for this objective is clear, as there is very little justification on equity grounds for a EU budget which does not treat equally countries with similar levels of wealth. This, of course, applies only to net balances. Specific budget items, with the possible exception of structural spending, should be set and evaluated in terms of the specific policy goals involved, and not in terms of their redistributive effects. But the principle of cohesion, and more importantly, the political feasibility of the Union and its legitimacy, require that the distribution of net benefits be inversely related to income levels. If this objective cannot be reasonably well met under the existing budget structure, it may be advisable to introduce a compensatory mechanism (in the form of rebates or surcharges on GNP contributions, for instance) which ensures overall equity.

We believe this procedure will also improve the quality of budget decisions by changing in a positive way the incentives faced by national governments. Under the current situation, with national contributions largely set in advance, community expenditures are perceived as having a zero marginal cost to national governments. Hence, ministers will enter council meetings (especially sectoral ones), with the incentive to fight for every possible increase in spending favouring their own countries -- and this includes programmes they would probably not be willing to finance with their own budgets. If net balances are set in advance, so that expenditure gains in any given programme will have to be financed at the margin by national budgets, this perverse incentive disappears.

Our analysis of the 1997 budget shows that the there is ample scope for the reallocation of EU revenues and expenditures in a way that, without reducing the overall degree of redistribution, should go a long way towards meeting the demands of countries that have good reasons to argue that their net contributions are too high. Table 1 shows the adjusted relative per capita net balances of the different member states, their shares of the Union's population and the implied "excess surpluses" in millions of current ecus and pesetas. As can be seen in the table, moving to the "fair" allocation described by the estimated redistribution regression would involve a transfer of 9,400 million ecus across member states. In absolute terms, the main beneficiaries would be Germany, Spain, the Netherlands, Sweden and the United Kingdom, at the expense of France, Ireland, Denmark, Greece, Italy and Belgium. As shown in the table, the required transfers would flow mostly across rich countries, rather than from the "Cohesion four" to the rest.

Table 1: Adjusted net fiscal balances with the EU in 1997

These estimates should be refined in a number of ways before they can be used as the basis for an equitable reallocation of resources.⁷ First, all EU expenditures should be allocated among member states in some reasonable way. The reason is that our procedure implicitly allocates such expenses on a per capita basis (through the average EU net balance, which enters into the computation of the net relative balances used in the regressions). On equity grounds, it seems preferable to allocate most "overhead" expenses, including foreign aid, in proportion to GNP. In the case of operating expenses (such as salaries and purchases of goods and services), the allocation should also be related, at least in part, to the location of the various EU institutions and to the nationality of its personnel for, while it is true that their benefits accrue in principle to all EU citizens, all countries would prefer to have such institutions located in their own territory. Finally, export rebates and the cost of price support interventions in agricultural markets should probably be allocated in proportion to the production of the relevant agricultural commodities. On the revenue side, tariff receipts should be allocated in proportion to GNP, rather than on the basis of the port of entry of non-EU imports. Finally, our computations should be adjusted in order to take into account the need to gradually free resources in order to finance the accession of new member states.

⁷ In addition to the issues discussed in the text, there is an additional question that is a bit more technical in nature. To simplify calculations and ensure that they add up to zero, the total balances shown in Table 1 are calculated using average EU income per capita in ecus, without correcting for PPP differences. This creates an

While the reallocation of net balances in greater accordance with equity criteria is one of the central issues in the current negotiations over the Agenda 2000, this objective seems to have been approached in a rather roundabout way. Rather than focusing directly on net balances, member states have argued mostly about how much to reduce expenditure on agricultural and structural programmes (relative to the Commission's original proposals) in order to finance a reduction of the fiscal deficits of net contributors to the system. While the effect of these proposals is difficult to evaluate without detailed knowledge of their fine print, some indicative calculations based on two not unlikely scenarios may give us a rough idea of the range of possible outcomes under existing proposals and of how they compare to the results obtained through a partial elimination of adjusted net balances.

Taking as a baseline the current situation, our first scenario (str) involves a uniform 20% reduction in all structural expenditure (including the Cohesion Fund) relative to observed 1997 levels, and the use of these funds (5,148 million ecus) to reduce the deficits of net contributors in proportion to their (total unadjusted) net balances. The second scenario (agr) is identical except in that the (same) expenditure cut comes from the reduction of agricultural subsidies (at a uniform rate of 12.67% for all countries). Our last scenario (nb) involves a redistribution of the same total amount of money in proportion to adjusted net balances, i.e. the reduction of "excess" deficits or surpluses by 54.77% (= 5,148/9,400).

Table 2: Redistributive impact of various budget scenarios

inconsistency with the rest of our procedure and will work against poor countries that see their net surpluses reduced under the proposed adjustment, because the real cost for them of each "nominal ecu" is higher.

Figure 21: Changes in adjusted net balances induced by the different scenarios (adjusted balances are computed as the residuals from the regression line fitted to observed 1997 data)

The right-hand side of Table 2 shows the baseline values of total net balances in millions of 1997 ecus and the changes induced by each of our three scenarios. Germany, for instance, would see its net contribution reduced by over 2,000 million ecus under all three proposals. The left-hand side of the table shows the adjusted relative per capita net balance for each country in each of the four scenarios, computed in each case as the residual from the corresponding regression line. At the bottom of the table we show the weighted standard deviation of each of these sets of residuals, and the redistribution parameter obtained from each of the regressions. These figures show that cuts in either structural or agricultural spending will reduce the average redistributive effect of the EU budget relative to the current situation, without significantly mitigating horizontal inequities. Our hypothetical cuts in structural expenditure, in particular, would reduce the overall redistribution coefficient of the EU budget by one third.

To visualize the net effects of the different scenarios on each member country, Figure 21 shows the changes they induce (relative to the baseline) in adjusted relative net balances, computed in all cases as deviations from the line fitted to the 1997 data (as opposed to the figures shown in the Table, where the reference is a different regression in each case). The figure shows that the effects on individual countries differ widely across scenarios, except in the case of Germany. Cuts in structural spending would transfer resources from the cohesion countries to all the rest. Reductions in agricultural subsidies would hurt the cohesion countries (although not as much as the previous alternative) as well as Italy, France and Denmark. In the third scenario, by contrast, both the winning and the losing groups include a mixture of rich and poor countries.

5.- Summary and conclusions

In the first part of this paper we have developed a simple extension of a standard methodology for the analysis of the redistributive effects of fiscal flows within a federal or quasi-federal system. Using an additive decomposition of net budget balances and a weighted regression specification, we obtain a set of partial redistribution coefficients that add across expenditure and revenue items to an overall redistribution parameter, and estimates of residual or excess balances after correcting for income per capita that add up in the natural way across programmes for a given country and across countries for a given programme. Applying this methodology to EU budget data, we have analyzed its redistributive effects over the period 1986-1997. We find that the redistributive impact of fisca flows across Union members has been considerable, particularly given the small size of the budget, that it has tended to increase over time, and that it is mostly due to the effect of structural programmes. We also find significant differences in net contributions across countries with similar prosperity levels. In this sense, the current concern about the unfairness of the distribution of net burdens does seem to be justified, although the list of net beneficiaries that emerges after correcting for income includes several rich countries -- at least one of which has been quite vocal in pressing for reductions in its net contributions to the Union

The redistribution equations we have estimated can also be interpreted as a one-parameter family of allocation rules describing how budgetary balances should be set as a function of relative income levels. We have argued for the explicit use of such a rule as a framework for EU budget decisions and for the adoption of a multi-stage budget-setting procedure: The Commission and the member states would first agree on the overall expenditure volume, the mechanisms for its financing and the desirable redistribution coefficient. National budgetary balances would then be set on the basis of per capita income prior to the negotiation of specific expenditure programmes. Any discrepancies between the outcome of this last stage and the desired net balance would be eliminated by a compensation mechanism that could take the form of a rebate or surcharge on GNP contributions. In practice, we would advocate that this procedure be applied in an incremental fashion, taking as given the allocation rule implicit in the status quo and gradually eliminating "excess" balances or adjusting the redistribution coefficient as required by an evolving consensus.

In our opinion, this approach would have several important advantages over current practices. Perhaps the most important ones are that it would provide a transparent way of implementing the principle of cohesion and that it would reduce the scope for real or perceived inequities in the allocation of budget resources by forcing member states to negotiate over a single, easily interpretable, parameter rather than about a host of specific items that may add up to apparently arbitrary allocations. In addition, setting net balances in advance of expenditure programmes will in effect force national governments to bear the marginal cost of community expenditures, thereby increasing the incentives for a more efficient allocation of resources.

In the last part of the paper, we have examined the implications of our proposed allocation rule, taking as given the redistribution coefficient implicit in the 1997 data, and compared the effects of its partial implementation with those of two alternative scenarios involving flat-rate cuts in either structural or agricultural programmes used to finance a proportional reduction in the budget balances of net contributors. Our calculations show that either of these two alternatives (and especially the first one) will reduce the degree of redistribution of the system without improving its horizontal equity.

Appendix

i.- Normalization of the budget data

Let and denote country i's income per capita before and after contributions to or transfers from the EU budget. These two variables are related by the following equation

\[\mathrm{(A.1)} \quad \mathrm {Q_ {i} = Y_ {i} + NB_ {i} = Y_ {i} + T_ {i} - C_ {i} = Y_ {i} + \mathfrak {R} _ {k} T_ {ki} - \mathfrak {R} _ {k} C_ {ki}}\]

where is the net EU budget balance of country i, defined as the difference between total transfers from the Union (T ) and total contributions to it (C ), both measured in per capita terms. These two variables can be expressed as sums of various per capita expenditure and contribution items and

Let Y, Q, C and T be EU-wide per capita weighted averages of the same variables (i.e. total amounts divided by total EU population). If we assume that the EU budget is balanced and that all expenditure is domestic (which is roughly true as an approximation), then and . Dividing both sides of (A.1) by , we can express all quantities as proportions of average EU income per capita. Letting lower-case letters denote these new normalized variables, we have

\[\mathrm{(A.2)} \mathrm {q_ {i} = y_ {i} + nb_ {i} = y_ {i} + t_ {i} - c_ {i} = y_ {i} + \mathcal {R} _ {k} t_ {ki} - \mathcal {R} _ {k} c_ {ki}}\]

for country i and

\[\mathrm{(A.3)} \mathrm{q} = \mathrm{y} + \mathrm{t} - \mathrm{c} = \mathrm{Y} + \Re_ {\mathrm{k}} \mathrm{t} _ {\mathrm{k}} - \Re_ {\mathrm{k}} \mathrm{c} _ {\mathrm{k}}\]

for the average of the EU. Letting denote the deviation of each variable x from its EU weighted average and subtracting (A.3) from (A.2) we obtain

\[\mathrm{(A.4)} \widetilde {\textbf {q}} _ {\mathbf {i}} = \widetilde {\textbf {y}} _ {\mathbf {i}} + \widetilde {\textbf {t}} _ {\mathbf {i}} - \widetilde {\textbf {c}} _ {\mathbf {i}} = \widetilde {\textbf {y}} _ {\mathbf {i}} + \widetilde {\textbf {n b}} _ {\mathbf {i}} = \widetilde {\textbf {y}} _ {\mathbf {i}} + \Re_ {\mathbf {k}} \widetilde {\textbf {t}} _ {\mathbf {k i}} - \Re_ {\mathbf {k}} \widetilde {\textbf {c}} _ {\mathbf {k i}}\]

That is, the relative disposable income of country i (in % deviations from average disposable income) is the sum of the country's gross relative income and the algebraic sum of its contributions to and transfers from the EU budget, measured as proportions of average EU income and expressed as deviations from their average values in the sample. Equation (1) in the text is derived from equation (A.4) after subtracting from both sides in order to leave nb on the left-hand side. Notice that the notation changes slightly, as the tildes are omitted in the text.

ii.- Estimation procedure and additivity properties of the redistribution coefficients and adjusted balances

As noted in the text, we estimate equations of the form

\[\mathrm{(A.5)} \mathrm {nb_ {i}} = - \rho_ {\mathrm{b}} \mathrm {y_ {i}} + \varepsilon_ {\mathrm{bi}} \quad \mathrm{(A.6)} \mathrm {t_ {ki}} = - \rho_ {\mathrm{tk}} \mathrm {y_ {i}} + \varepsilon_ {\mathrm{tki}} \quad \mathrm{and} \quad \mathrm{(A.7)} \mathrm {c_ {ki}} = \rho_ {\mathrm{ck}} \mathrm {y_ {i}} + \varepsilon_ {\mathrm{cki}}\]

using the normalized data and a weighted least squares procedure. Focusing on equation (A.5) for concreteness, we choose so as to minimize the weighted sum of the squared residuals,

\[\mathrm{(A.8)} \mathrm{R} = \mathfrak {R} _ {\mathrm{i}} \mathrm{w} _ {\mathrm{i}} \varepsilon_ {\mathrm{bi}} ^ {2} = \mathfrak {R} _ {\mathrm{i}} \mathrm{w} _ {\mathrm{i}} (\mathrm{nb} _ {\mathrm{i}} + \rho_ {\mathrm{b}} \mathrm{y} _ {\mathrm{i}}) ^ {2}\]

where is the share of country i in the total population of the EU. Setting the derivative of R with respect to ρ_b equal to zero and solving for this parameter we obtain the following WLS estimator:

\[\rho_ {\mathrm{b}} = - \frac {\Re_ {\mathrm{i}} \mathrm{wi} \mathrm{nb} _ {\mathrm{i}} \mathrm{y} _ {\mathrm{i}}}{\Re_ {\mathrm{i}} \mathrm{wi} \mathrm{y} _ {\mathrm{i}} ^ {2}}. \tag {A.9}\]

By a similar argument, the partial redistribution coefficients for total transfers and contributions are given by

\[\rho_ {\mathrm{t}} = - \frac {\Re_ {\mathrm{i}} \mathrm{wi} \mathrm{tiyi}}{\Re_ {\mathrm{i}} \mathrm{wi} \mathrm{yi} ^ {2}} \quad \text {and} \quad (\mathrm{A}. 1 1) \rho_ {\mathrm{c}} = \frac {\Re_ {\mathrm{i}} \mathrm{wi} \mathrm{ciyi}}{\Re_ {\mathrm{i}} \mathrm{wi} \mathrm{yi} ^ {2}} \tag {A.10}\]

respectively. Hence, we have

\[\rho_ {\mathrm{t}} + \rho_ {\mathrm{c}} = - \frac {\Re_ {\mathrm{i}} \mathrm{w} _ {\mathrm{i}} \mathrm{t} _ {\mathrm{i}} \mathrm{y} _ {\mathrm{i}} - \Re_ {\mathrm{i}} \mathrm{w} _ {\mathrm{i}} \mathrm{c} _ {\mathrm{i}} \mathrm{y} _ {\mathrm{i}}}{\Re_ {\mathrm{i}} \mathrm{w} _ {\mathrm{i}} \mathrm{y} _ {\mathrm{i}} ^ {2}} = - \frac {\Re_ {\mathrm{i}} \mathrm{w} _ {\mathrm{i}} (\mathrm{t} _ {\mathrm{i}} - \mathrm{c} _ {\mathrm{i}}) \mathrm{y} _ {\mathrm{i}}}{\Re_ {\mathrm{i}} \mathrm{w} _ {\mathrm{i}} \mathrm{y} _ {\mathrm{i}} ^ {2}} = - \frac {\Re_ {\mathrm{i}} \mathrm{w} _ {\mathrm{i}} \mathrm{nb} _ {\mathrm{i}} \mathrm{y} _ {\mathrm{i}}}{\Re_ {\mathrm{i}} \mathrm{w} _ {\mathrm{i}} \mathrm{y} _ {\mathrm{i}} ^ {2}} = \rho_ {\mathrm{b}} \tag {A.12}\]

so the partial redistribution coefficients add up to the total redistribution parameter. Using (A.12), and (A.5) - (A.7), moreover, we have

\[\begin{array}{r l} \mathrm {nb_ {i}} & = \mathrm {t_ {i}} - \mathrm {c_ {i}} = - \rho_ {\mathrm{t}} \mathrm {y_ {i}} + \varepsilon_ {\mathrm{ti}} - \rho_ {\mathrm{c}} \mathrm {y_ {i}} - \varepsilon_ {\mathrm{ci}} = - (\rho_ {\mathrm{t}} + \rho_ {\mathrm{c}}) \mathrm {y_ {i}} + (\varepsilon_ {\mathrm{ti}} - \varepsilon_ {\mathrm{ci}}) \\ & = - \rho_ {\mathrm{b}} \mathrm {y_ {i}} + (\varepsilon_ {\mathrm{ti}} - \varepsilon_ {\mathrm{ci}}) \end{array} \tag {A.13}\]

from where it follows that , that is, that the adjusted partial balances add up across programs to the net budget balance for each country.

Finally, we show that the weighted sum of the residuals is equal to zero in any given equation. Working with (A.5) and reintroducing the tildes as needed to emphasize that we are working with normalized variables in deviations from EU-wide averages, we have:

\[\begin{array}{r l} \mathfrak {R} _ {\mathrm{i}} \mathrm{wi} \varepsilon_ {\mathrm{bi}} = & \mathfrak {R} _ {\mathrm{i}} \mathrm{wi} (\widetilde {\mathrm{nb}} \mathrm{i} + \rho_ {\mathrm{b}} \widetilde {\mathrm{y}} \mathrm{i}) = \mathfrak {R} _ {\mathrm{i}} \mathrm{wi} \widetilde {\mathrm{nb}} \mathrm{i} + \rho_ {\mathrm{b}} \mathfrak {R} _ {\mathrm{i}} \mathrm{wi} \widetilde {\mathrm{y}} \mathrm{i} \\ = & \mathfrak {R} _ {\mathrm{i}} \mathrm{wi} (\mathrm{nb} _ {\mathrm{i}} - \mathrm{nb}) + \rho_ {\mathrm{b}} \mathfrak {R} _ {\mathrm{i}} \mathrm{wi} (\mathrm{y} _ {\mathrm{i}} - \mathrm{y}) = 0 + \rho_ {\mathrm{b}} 0 = 0 \end{array}\]

where the variables without a subscript denote (weighted) EU averages. Hence, nb = nb , and the desired equality follows. Notice that the result follows from the normalization of the data and holds independently of how is estimated.

References

1. Bayoumi, T. and P. Masson (1995). "Fiscal flows in the United States and Canada: Lessons for monetary union in Europe." European Economic Review 39, pp. 252-74.
2. de la Fuente, A. (1997). "Desigualdad regional en España: la incidencia de la imposición directa sobre las familias y el sistema de protección social." Documento de Trabajo PT 59.97, Instituto de Análisis Económico (CSIC), Barcelona.
3. Doménech, R., A. Maudes and J. Varela (1998). "Fiscal flows in Europe: the redistributive effects of the EU budget." Mimeo, Ministry of Economics and Finance, Madrid.
4. Martin, C. (1997). España ante la nueva Europa. Alianza Editorial, Madrid.
5. Sala-i-Martin, X. and J. Sachs (1992). "Fiscal federalism and optimum currency areas: Evidence for Europe from the United States," in M. Canzoneri, V. Grilli and P. Masson, editors, Establishing a central bank: Issues in Europe and lessons from the U.S. Cambridge University Press, Cambridge, pp. 195-219.
6. The Economist (1999). "European farming: Against the grain." Feb. 20th 1999, pp. 30-1.
7. von Hagen, J. (1992). "Fiscal arrangements in a monetary union: Evidence from the U.S." in D. Fair and Ch. Boissieux, editors, Fiscal policy, taxes and the financial system in an increasingly integrated Europe. Kluwer, Deventer.