Chapter

20 Ethiopia

Editor(s):
Teresa Ter-Minassian
Published Date:
September 1997
Share
  • ShareShare
Show Summary Details
Author(s)
Giorgio Brosio and Sanjeev Gupta 

The process of fiscal decentralization in Ethiopia began in 1992. Proclamations were issued during 1992–93 to establish federal and regional self-governments and to define both the revenue-sharing arrangements and the powers of the federal and the regional governments (henceforth referred to as states).1 These elements of a federal structure have been embraced by the National Constitution, which was approved in December 1994. The National Constitution lists nine states, but there are two other jurisdictions that currently have the status of provisional administration. Interstate boundaries reflect ethnic or linguistic groups, and states vary considerably in terms of land area and population.

Since the 1993/94 budget, expenditures devolved to the states have increased gradually to about 40 percent of total consolidated expenditures, whereas the federal government has continued to account for most of the revenues under the tax-assignment arrangements. Transfers from the federal government are thus the major source of support of states’ expenditures. Expenditure devolution has taken place in four core areas: agriculture, education, health, and road construction.

This chapter first describes the main characteristics of intergovernmental relations in Ethiopia, particularly the fiscal structure of Ethiopia, It then reviews the current grant system, highlighting the major drawbacks in its design and the way it is implemented. Finally, it presents options to strengthen it.

Main Characteristics of Intergovernmental Relations

The following characteristics of Ethiopia’ federal structure are particularly noteworthy.

First, revenues are categorized as central, state, and joint (Table 1). The most important revenues are assigned to the federal government. As the sharing of revenues jointly levied and collected by federal and state governments is still to be defined, they are collected by the federal government. The revenue assignment of the state governments is concentrated in direct tax revenue, including most personal income taxes; however, there is no historical evidence in Ethiopia of sustained growth of these taxes. As a result, there is a large and growing vertical fiscal imbalance between the state share of revenue and expenditures, which has widened from 14.7 percent in 1993/94 to 28.5 percent in the 1995/96 budget (Table 2). Under present trends, this vertical fiscal imbalance is likely to persist in the foreseeable future.

Table 1.Ethiopia: Revenue Sharing Between the Federal and State Governments
FederalStateJoint
Duties, taxes, and other charges on imports and exports.
Personal income tax of employees in federal government, international organizations, and those working in enterprises owned by federal government.Personal income tax of state government employees and those working in enterprises owned by states and in the private sector.Personal income tax of employees working in enterprises owned jointly by federal and state governments.
Profits tax and sales tax from federal government-owned enterprises and those operating across regional boundaries.1Profits tax and sales tax from the state government-owned enterprises.Profits tax and sales tax of enterprises owned jointly by federal and state governments.
Income tax, royalties, and land rent from small-scale mining enterprises,Profits tax, royalties, and rent from large-scale mining, petroleum, and gas enterprises that are incorporated,
Agricultural income tax from private and incorporated farmers.
Taxes on national lottery prizes and gambling.
Taxes from air, rail, and marine transport.Fees from water transportation within the state.
Taxes from rent of property owned by the federal government,Taxes from rent of property owned by the state government and income from private properties within the state.
Charges and fees on licenses and services of federal government; stamp duties; and rents of federally owned government houses and properties.Charges and fees on licenses and services of the state government; rents on state-owned houses and properties; and fees on the use of land.
Profits and sales taxes from individual merchants who are residents of the state.Profits tax on corporations and tax on dividends paid to shareholders.
Forest royalties.
Sources: Ethiopian Government Proclamation 33/1992; and Constitution of Ethiopia, December 8, 1994.

Excise taxes are not mentioned in the Proclamation and the Constitution.

Sources: Ethiopian Government Proclamation 33/1992; and Constitution of Ethiopia, December 8, 1994.

Excise taxes are not mentioned in the Proclamation and the Constitution.

Table 2.Ethiopia: Expenditures and Revenue of the Federal and State Governments1(In percent of total)
1993/94

Preliminary Actual2
1994/95

Preliminary Actual2
1995/96

Budget
Federal government
Expenditure share68.560.155.4
Revenue share83.285.383.9
Difference314.725.228.5
State governments
Expenditure share31.539.944.6
Revenue share16.814.716.1
Difference2−14.7−25.2−28.5
Sources: Ministry of Finance; and authors’ estimates.

The fiscal year runs from July 8 to July 7.

Excludes in-kind grants and project grants, data on which will become available after fiscal accounts are closed.

This difference is between revenue and expenditure shares.

Sources: Ministry of Finance; and authors’ estimates.

The fiscal year runs from July 8 to July 7.

Excludes in-kind grants and project grants, data on which will become available after fiscal accounts are closed.

This difference is between revenue and expenditure shares.

The principal reason for the large share of the federal government in total revenue is the dominance of indirect taxes in total revenue and the federal government’s high share of these taxes under revenue assignments. For instance, the federal government collects about 90 percent of all indirect taxes, with over 80 percent of sales taxes and 100 percent of import duties. One important revenue source, excises, is not legally assigned to either level of government, but is being collected by the federal government. Consequently, the average proportion of spending covered by a state’ own resources was about one-third in 1994/95. With the exception of the provisional administration of Addis Ababa, all states had deficits in 1994/95, and required federal transfers.

Second, revenue collections are highly concentrated in a few states; four states accounted for 84 percent of total revenue in 1994/95, while the smallest four contributed a mere 2.2 percent (Table 3). This illustrates the states’ widely divergent revenue-raising capacities. The per capita revenue effort in 1994/95 varied between Br 5.7 in Afar and Br 125 in Dire Dawa.

Table 3.Ethiopia: Revenue Shares of States and Per Capita Revenue Effort
State Revenue Shares

(In percent of total)
State Per Capita

Revenue Collection1

(In birr)
States and Provisional Administrations1993/941994/951993/941994/95
BudgetPreliminary ActualBudgetPreliminary ActualBudgetPreliminary ActualBudgetPreliminary Actual
Tigray (region 1)7.16.15.96.817.712.116.017.9
Afar (region 2)1.00.91.00.59.77.610.85.7
Amara (region 3)14.114.813.714.38.36.98.99.0
Oromiya (region 4)23.031.526.729.210.911.813.814.6
Somali (region 5)3.95.95.44.517.721.327.221.6
Benishangul/Gumuz: (region 6)0.60.50.50.69.36.79.910.6
SEPA210.310.110.911.57.96.19.19.3
Gambela (region 12)0.30.40.30.523.722.927.736.4
Harari (region 13)2.30.90.6209.787.255.1
Addis Ababa (region 14)34.527.632.529.3133.284.6137.4119.7
Dire Dawa (region 15)2.92.22.22.1169.9102.9140.7125.0
Sources: Ministry of Finance; and authors’ estimates.

Based on population data used in 1991/92 elections.

Southern Ethiopian People’s Administrative region (formerly regions 7, 8, 9, 10, and 11).

Sources: Ministry of Finance; and authors’ estimates.

Based on population data used in 1991/92 elections.

Southern Ethiopian People’s Administrative region (formerly regions 7, 8, 9, 10, and 11).

Third, the expenditure responsibilities assigned to the federal and state governments in the Constitution are close to what is found in a highly decentralized system. For instance, the Constitution calls on the federal government to establish and administer national defense; administer monetary policy, currency, and banking; formulate and implement foreign policy; be responsible for the development, administration, and regulation of air, rail, waterways, and sea transport and major roads, as well as for postal and telecommunications services; and regulate interstate and foreign commerce. Furthermore, the federal government is asked to formulate the country’ policies in respect of overall economic and social development, and draw up and implement plans and strategies of development. The Constitution also calls for the federal government to set national standards for policies in, inter alia, public health and education.

Table 4 shows state expenditures by function, as a percent of total consolidated expenditure, for 1993/94, 1994/95, and 1995/96. The share of the states in total public expenditure has increased steadily. Whereas in 1993/94 states were responsible for 31.5 percent of total consolidated expenditure, their share is budgeted to rise to 44.6 percent in 1995/96. The increase is composed of a 6 percentage point increase in the recurrent budget and a 24 percentage point increase in the capital budget. Thus, there has been a strong shift toward devolution of the capital budget to the states since 1993/94.

Table 4.Ethiopia: State Expenditures by Function(In percent of consolidated total expenditure)
1993/941994/951995/96
BudgetPreliminaryBudgetPreliminaryBudget
Total expenditure39.431.538.138.844.6
Total recurrent expenditure39.137.841.042.143.6
General service28.625.429.833.930.6
Organs of state83.956.081.976.177.0
Judiciary74.281.375.384.982.4
Defense
Public order and security67.964.371.265.671.4
General services27.728.529.738.332.3
Economic services60.957.861.664.366.2
Agriculture and natural resources78.381.181.883.585.0
Industry56.354.763.468.067.0
Mines and energy27.622.022.529.728.7
Trade and tourism53.054.461.969.667.3
Transport and communications75.827.277.992.288.8
Urban development and construction27.418.623.424.322.9
Social services79.872.979.475.381.4
Education and training86.082.787.486.888.3
Culture and sports58.957.659.264.163.8
Public health83.682.082.884.284.3
Labor and social welfare23.732.326.129.230.9
Relief and rehabilitation28.915.121.910.823.5
Pension payments58.659.555.857.752.7
Interest and charges
Miscellaneous16.015.2
Total capital expenditure39.822.034.933.046.0
Economic development29.816.124.123.438.8
Agriculture, water, and natural resources57.732.765.954.079.8
Mining, industry, commerce, and tourism26.26.40.35.20.6
Electric power0.90.70.4
Roads, transport, and communications24.99.39.416.025.1
Social development78.939.179.069.077.4
Education73.740.370.459.267.5
Public health87.937.595.385.688.1
Community services78.138.977.869.483.9
General services100.043.431.622.828.4
Compensation payments1.6
Source: Ministry of Finance.
Source: Ministry of Finance.

The shift in expenditure responsibilities between the preliminary outcome for 1993/94 and the budget for 1995/96 was greater in some areas of public expenditure. For instance, the states’ share of economic services in recurrent expenditures increased by 8.4 percentage points, owing to devolution in agriculture and natural resources, industry, trade, and tourism, as well as transport and communications. Furthermore, around four-fifths of total consolidated recurrent expenditure on the judiciary, agriculture and natural resources, transport and communications, education and training, and public health were incurred by states in 1994/95 and are budgeted for 1995/96.

The bulk of capital spending on agriculture, water and natural resources, education, public health, and community services is also undertaken by the states. For education, the state share is two-thirds, compared with 80 percent and more in the other three areas. There is a national plan to increase rural roads by 1,000 kilometers annually. The states’ 1995/96 capital budget for roads, transport, and communications is designed to meet this target.2

Fourth, the states vary widely in population, land area, and economic circumstances. For instance, Oromiya is estimated to have a population of 17.1 million and a land area of about 0.32 million square kilometers, while both Gambela and Harari have populations of only 0.1 million each and land areas of 27,300 square kilometers and 3,000 square kilometers, respectively. Severe economic disparities exist between war- and drought-affected states. This implies that expenditure needs also vary widely, as do revenue-raising and, hence, fiscal capacities.

Fifth, there are vast differences in per capita expenditure of states with an annual average of about Br 42 per capita for the recurrent budget and Br 19 per capita for the capital budget in 1994/95. The variation among states in per capita recurrent expenditure is between Br 17.3 in Afar and Br 300.3 in Gambela; for per capita capital expenditure, it is between Br 10.6 in Amara and Br 204.6 in Gambela (Table 5). This suggests that the present grant system does not fully capture relative expenditure needs. It further suggests that a reform of the grant system alone is likely to be inadequate to correct historical inequities in Ethiopia, and therefore, an adjustment in the revenue assignments would also be needed. However, such an adjustment may not be feasible in the near term, since the Constitution was approved only at end-1994.

Table 5.Ethiopia: Per Capita Expenditure by State and Function, 1994/951(In birr)
TigrayAfarAmaraOromiyaSomaliBenishangul/

Gumuz
SEPA2GambelaHarariAddis

Ababa
Dire

Dawa
Total expenditure96.237.241.153.757.9138.048.6508.5326.8187.6193.9
Total recurrent expenditure51.417.330.542.335.979.032.1303.8300.3111.5165.6
General service12.07.56.111.217.624.76.9139.467.944.339.2
Organs of state4.82.91.65.90.57.71.628.611.510.17.3
Judiciary1.12.30.50.85.22.00.812.55.83.02.2
Defense
Public order and security4.40.32.42.812.88.92.832.236.813.618.2
General services1.72.01.51.71.26.11.766.113.817.611.6
Economic services11.24.86.26.35.615.86.666.121.913.226.9
Agriculture and natural resources6.52.55.55.61.610.55.544.79.26.121.8
Industry0.10.40.10.11.10.40.22.72.30.90.7
Mines and energy0.10.40.10.10.60.11.81.20.3
Trade and tourism0.30.90.20.22.21.40.27.11.21.32.9
Transport and communications0.10.10.10.40.60.23.63.51.6
Urban development and construction4.20.50.20.40.21.10.46.34.63.11.5
Social services24.64.515.020.512.137.816.898.3149.653.572.6
Education and training15.02.510.515.80.819.812.444.780.632.242.1
Culture and sports0.30.30.20.20.21.00.42.72.32.31.5
Public health8.50.43.84.210.614.13.540.263.316.925.4
Labor and social welfare0.20.40.20.10.20.80.24.52.31.51.5
Relief and rehabilitation0.60.90.30.20.32.00.36.31.20.52.2
Pension payments3.60.51.14.10.60.81.761.00.526.9
Interest and charges
Miscellaneous0.1
Total capital expenditure44.819.910.611.422.059.016.5204.626.576.228.3
Economic development25.916.06.27.811.815.46.9110.818.421.68.7
Agriculture, water, and natural resources14.713.84.15.811.39.94.797.418.416.98.7
Mining, industry, commerce, and tourism0.10.10.32.23.6
Electric power0.1
Roads, transport, and communications11.22.31.91.70.63.22.113.41.1
Social development18.93.83.93.29.439.68.076.08.153.819.6
Education10.22.31.31.15.312.52.615.22.31.94.4
Public health5.31.01.60.92.68.32.731.34.60.35.1
Community services3.40.51.01.21.418.82.729.51.251.710.2
General services0.50.40.74.01.717.90.7
Sources: Ministry of Finance; and authors’ calculations.

Based on population data used in 1991/92 elections.

Southern Ethiopian People’s Administrative region.

Sources: Ministry of Finance; and authors’ calculations.

Based on population data used in 1991/92 elections.

Southern Ethiopian People’s Administrative region.

Sixth, state governments in Ethiopia are legally empowered by the Constitution to borrow domestically. The Constitution requires the federal government to determine conditions and terms under which states may borrow from domestic sources. The initial proclamation issued in 1992 set these conditions. Any form of external borrowing by the states is disallowed.

According to the 1992 proclamation, a state could borrow only for projects that have a feasibility study showing its ability to repay the debt, and submit their request for borrowing to the Ministry of Finance or the Ministry for Economic Development and Cooperation.3 The likely impact of such borrowing on the national budget is an important consideration in granting the approval.

The new Constitution and the 1992 proclamation are unclear on who the lender is (whether the National Bank of Ethiopia or any other commercial bank or financial institution) and what instruments of debt are open to states (for example, treasury bills or bonds). According to proclamations on banking and monetary matters (No. 83 and No. 84 of 1994), commercial banks are not permitted to lend to state governments; and the National Bank may not extend direct credit to any person other than federal government, commercial banks, or other financial institutions.

Finally, there is a wide variation in administrative capacities of the states. For instance, the provisional administrations of Addis Ababa and Oromiya have a relatively better infrastructure and personnel to formulate and implement public programs. By contrast, some states need to be assisted by the federal government.

Differing administrative capacities of the states have important implications for the federal government seeking to exercise adequate control over the country’ financial operations and macroeconomic policy. Expenditure devolution in Ethiopia has resulted in diffusion of financial information away from the center. States have, in turn, devolved much of the responsibility of the payment to the lower levels of government. But the communication between these levels and a state, and between the states and the center, is often less than satisfactory. The lower levels of government in some states lack both telephones and electricity. Reliability of service is also a problem where such facilities exist, and travel to some areas can take several days.

The Grant System

The Constitution gives state governments four distinct sources of revenue: own taxes, fees, and user charges; taxes shared with the federal government; grants from the federal government; and domestic borrowing. The states have relied principally on federal transfers to support their expenditures because states’ own taxes accounted for only 16 percent of consolidated total revenues in the 1995/96 budget, the sharing of revenues from taxes jointly levied and collected by the federal and state governments is still to be defined, and domestic borrowing by the states is yet to assume any role. Until now, only general, nonconditional grants have been extended. The grant system has been modified twice since 1993/94, following the devolution of staff, offices, and responsibilities to regional governments.

As state governments were not able to prepare their own recurrent and capital budgets in 1993/94, allocations were decided mainly by the federal government. Transfers for recurrent expenditure to the states were based on the number of bureaus and staff and on the expenditure needed to maintain the infrastructure. The interstate allocation for the capital budget was made on the basis of the ongoing projects transferred from the federal government to the state governments and an assessment of the states’ capacities to implement projects.

The method adopted in 1994/95 and 1995/96 for grants follows a pattern frequently found in other developed and developing countries, combining a formula with ad hoc adjustments. More precisely, the overall size of grants to states was determined by taking into account all important macroeconomic variables (Box 1). First, the size of total resources available to support expenditures of both the federal and the state governments was estimated. The resource envelope for these expenditures included total tax and nontax revenues, counterpart funds, and other foreign assistance to the states. The total resources were then divided between the federal and the state governments on the basis of existing expenditure assignments.

In 1994/95, allocations from the states’ overall expenditure ceiling to individual states were made separately for the recurrent budgets and for the capital budgets. No precise formula was used for the recurrent expenditure grant. Instead, five broad criteria were used; the number of existing “weredas” and “zones” in each state,4 the 1993/94 allocations, the length of rural roads, the number of state agricultural demonstration centers, and the 1993/94 state expenditure for education and public health. Thus, the authorities have been mainly concerned, it appears, with providing each state with no fewer resources than they received the previous year. A formula was experimented with for capital expenditure in the same year. It included five indicators, with weights shown in parentheses: population (30 percent), I-distance5 (25 percent), regional tax effort (20 percent), 1992/93 capital expenditure (15 percent), and area (10 percent).

Box 1.The Grant Framework

(a) Federal and state expenditure

  • FR + SR+DD + FFA + SFA = TE =

  • FE + SE FE = FR+DD + FFATR

  • SE = SR + SFA + TR

where,

  • FR is federal government tax revenue,

  • SR is state government tax revenue,

  • DD is domestic borrowing,

  • FFA is foreign assistance to the federal government,

  • SFA is foreign assistance to the states,

  • TE is Ethiopia’s total public expenditure,

  • FE is federal government expenditure,

  • SE is state government expenditure, and

  • TR are the net transfers from federal to state governments.

(b) The formula for allocating gross grants in 1995/96

where,

  • i stands for state i,

  • POP is population,

  • ID is the I-distance indicator,

  • SRB is budgeted revenues of state i.

In 1995/96, by contrast, the entire state allocation was distributed by means of a general formula, with no distinction between current and capital allocations. It included three variables: population, the state revenues budgeted, and the I-distance indicator weighted by population. An equal weight of 33.3 percent was assigned to each (see Box 1). Because of the concern for providing states with no less resources than in the preceding year, the historical levels of expenditures were not very much affected.

Net transfers, that is, actual disbursements from the federal treasury, are the difference between the gross grants determined by the formula and budgeted own revenues of states. The application of the formula means that if actual revenues fall short of budgeted revenues, the states’ budgets can face severe expenditure compression.

As in 1994/95, substantial differences between the allocation generated by the formula and the amounts allocated in the previous year emerged for a number of states in 1995/96. They were adjusted with supplementary discretionary allocations, so that the impact on state allocations would be gradual (Table 6). Allocations for 1995/96 show a redistributive pattern, where the poorest states receive a net per capita amount well above the national average; in part, this redistribution arises from a hidden factor, that is, the poorest states are also the smallest. The substantial gross grants that benefit Addis Ababa and, to a lesser extent, Dire Dawa are explained by their city-state characteristics, that is, by their high revenue-generating capacity.

Table 6.Ethiopia: Allocation of Grants to States, 1995/1996(In millions of birr unless otherwise indicated)
States and

Provisional

Administrations
According

to the

Formula
Difference

over

1994/95
Discretionary

Adjustment
Initial

Budget

Allocation
Additional

Allocation
Final

Allocation
From

Domestic

Sources
Foreign

Grants and

Loans
Final

Allocation

Per Capita (In birr)
Tigray295.9−31.932.6328.514.4342.9319.323.6105.5
Afar128.3−13.513.8142.16.3148.4121.626.8189.0
Amara811.870.41.9813.739.6853.3766.786.662.5
Oromiya1,076.436.92.61,079.052.41,131.4982.2148.666.5
Somali164.1−1.21.6165.77.9173.6151.122.598.7
Benishangul/Gumuz32.4−74.374.4106.81.6108.482.525.9219.1
SEPA1649.041.01.5650.531.5682.0615.666.464.8
Gambela9.6−80.680.690.20.590.772.718.0810.5
Harari15.0−25.725.740.70.741.438.03.4476.4
Addis Ababa480.1109.71.1461.223.4504.6397.8106.8241.1
Dire Dawa33.2−3.33.436.61.638.232.85.4277.4
Total3,695.8239.23,935.0179.94.115.03,580.9534.082.4
Sources: Ministry of Economic Development and Cooperation; and authors’ calculations.

Southern Ethiopian People’s Administrative region.

Sources: Ministry of Economic Development and Cooperation; and authors’ calculations.

Southern Ethiopian People’s Administrative region.

while the poorest states benefit from higher per capita transfers than the rest of the country, the overall redistributive impact of the present system is difficult to ascertain. This is because of the lack of information on extrabudgetary donor assistance and because redistribution and equity aspects have to be evaluated in Ethiopia according to inter-ethnic fairness, historical neglect of certain regions, and the differing impact of war, drought, and famine on various parts of the country.

Problems with the Existing Grant Formula

Establishing a grant system that is simple, fair, and transparent is difficult for any federal country. The problems are compounded in Ethiopia by the lack of adequate and up-to-date information, the shortcomings of the I-distance indicator, the lack of incentives for enhancing states’ tax efforts, the need for interstate equity, and the need for encouraging the implementation of minimum standards. The reform options examined in the following section seek to remedy these weaknesses.

Lack of Adequate and Up-to-Date Information

In Ethiopia, population plays a dominant role in the existing formula, but population data are outdated. For 1995/96 allocations, the population data from the 1991/92 elections were used. The mobility of the population and differences in the rate of population growth among states since the last census are not taken into account in the present grant system. The use of old population data thus favors static states and penalizes the states, including the cities, that experience inflows of migrants. The same problem carries over to most variables used for calculating the I-distance. Many of these variables are compiled through surveys, often conducted at long intervals.

Shortcomings of the I-Distance Indicator

The I-distance indicator in use in Ethiopia suffers from a number of shortcomings:

  • The variables included in the index are highly heterogeneous. Variables, such as per capita food crop and industrial production, are related to the level of economic development. Others reflect the existing level of public services, both national and states, such as post offices and hospital beds, thereby capturing expenditure needs. But the use of the share of urban population in total population has no clear meaning in the Ethiopian context. Urban development is an indicator of growth, but then expenditure needs also tend to be concentrated in urban areas.

  • A given indicator could have a different meaning for different states. A case in point is the road density in sparsely populated (desert) and highly populated (urban) areas;

  • Many needs-based indicators are excluded from the I-distance index, such as the share of population with access to clean water, the number of victims of war, the rate of infant mortality, and the proportion of population below the poverty line.

  • The construction of the index is not transparent and its application is discretionary. The formula, which is quite complex, is questionable on both economic and statistical grounds. Indeed, the final value of I-distance appears to be highly dependent on the choice of the first variable fed into the computer program.

Lack of Incentive for Revenue Enhancement

The formula does not necessarily stimulate the states’ revenue efforts. This is because greater revenue efforts will lead to higher gross grants, and, at the same time, this revenue will be deducted from the gross grants to determine the net transfer to each state.6

The present formula could also create competition among states to overstate their budgeted revenue. It could lead them to embark on ambitious expenditure plans to have a larger entitlement from the common pool. The cost of the revenue shortfall might then have to be borne by the whole country if the state exercised its right to borrow domestically. This problem would be obviated to some extent if actual revenue collected in the preceding year were used instead of budgeted revenue.

Need for Interstate Equity

This goal has the highest priority in the present grant system of Ethiopia. Each state should perceive that the formula provides an equitable distribution of national resources. As in most developing countries, Ethiopia’s effort to reduce disparities in levels of public services in different states is plagued by the problems of identifying indicators of need and by severe data limitations, compounded by the continuing redefinition of interstate boundaries.

Need to Encourage the Implementation of Minimum Standards

Ethiopia’s Constitution (Article 51) mandates that the federal government establish national standards in health and education. The federal government is also committed to foreign donors to spend the counterpart funds on social programs. In 1995/96, counterpart funds generated resources for the budget amounting to 3.3 percent of GDP. This would require some built-in incentive in the formula for the state to spend more on these areas. Since grants to the states in Ethiopia are general purpose, there is no mechanism at present to ensure that the minimum national standards are being adhered to. It can, however, be argued that, in Ethiopia’s case, even if certain standards were established by the federal government, their monitoring would be extremely difficult in view of its weak administrative capacity.

Options for Improving the Grant System

There are various ways in which many of the above-noted drawbacks of the grant system could be rectified, while ensuring that the grant formula is simple, fair, and transparent. A few options from a range of possibilities are presented below. For illustrative purposes, the impact of different reform options is compared with actual 1995/96 allocations.

Simple Indicators of Inequality and Need

It is easy to replace the present I-distance indicator with a more transparent, less discretion-prone indicator of inequality among states. Two possibilities are discussed here. The first is based on two proxies of the relative expenditure need of states, namely the ratio of students to total population and the number of medical personnel to total population. The second possibility uses proxies for economic development or wealth: the density of telephones and electricity consumption in a state. The formulas assign values of over one to states that are below the national average and values of less than one to those that are above.7 These values are multiplied by population, as in the present formula used in Ethiopia. Table 7 shows a state’s allocation and per capita gross and net transfers (that is, after adjusting for own revenues).

Table 7.Ethiopia: Options for Allocating Grants—Simpler Indicators of Inequality and Need1
States and

Provisional

Administrations
Total Existing

Allocations

(In millions of birr)
Per Capita TransfersModified

Transfer Based on
Gross Per Capita

Transfer Based on
Net Per Capita

Transfer Based on
GrossNetNeed2Wealth3Need2Wealth3Need2Wealth3
(In birr)(In millions of birr)(In birr)(In birr)
Tigray295.991.075.0218.4219.867.267.651.251.6
Afar128.3163.4152.549.547.063.159.952.249.0
Amara811.859.550.6873.4872.564.063.955.155.0
Oromiya1,076.463.149.21,174.61,195.168.870.055.056.2
Somali164.193.366.1151.5146.386.283.259.056.0
Benishangul/Gumuz32.465.555.730.228.861.058.251.148.3
SEPA4649.061.752.6648.2671.461.663.852.554.7
Gambela9.685.858.59.59.285.282.258.054.9
Harari15.0172.682.315.014.7172.9169.582.679.2
Addis Ababa480.1229.492.2491.5457.6234.8218.697.681.4
Dire Dawa33.2241.198.033.933.4246.4242.6103.399.6
Sources: Ministry of Economic Development and Cooperation; and authors’ estimates.

To include two indicators of need or wealth as defined in the text.

Defined as relative shares of students in population and the number of medical personnel in population, weighted equally.

Defined as the relative shares of telephone use and electricity consumption in population, weighted equally.

Southern Ethiopian People’s Administrative region.

Sources: Ministry of Economic Development and Cooperation; and authors’ estimates.

To include two indicators of need or wealth as defined in the text.

Defined as relative shares of students in population and the number of medical personnel in population, weighted equally.

Defined as the relative shares of telephone use and electricity consumption in population, weighted equally.

Southern Ethiopian People’s Administrative region.

Interestingly, the results are broadly similar to those arrived at by applying the existing formula. Amara and Oromiya gain somewhat at the expense of Tigray and Afar. There is no significant difference in the results based on need and wealth proxies. There is some advantage of using the latter, since information for its calculation is readily available from national electric and telephone companies.

Different Weights for the Same Variables

This possibility consists of changing the weights assigned to different variables (and replacing the I-distance indicator with a new proxy for inequality). In this option, the weight assigned to population is increased from the existing 33.3 percent. One of the objectives behind this simulation is to gauge the sensitivity of the grant formula to the weight assigned to the population variable. Using population as the major criterion for grant distribution amounts to assuming that expenditure needs are identical in per capita terms across all areas. Once again, two variants are presented here. In the first one, the population share is increased to 50 percent; in the second, the share is increased to 70 percent.

As expected, transfers to Tigray, Afar, Amara, Oromiya, Benishangul/Gumuz, and SEPA, which are among the poorest states, are higher than in the previous option (Table 8). Per capita gross and net transfers are now negatively correlated with the level of development. The negative correlation increases with the weight assigned to population. Grant allocations for Addis Ababa are clear evidence of this relationship. The gross per capita transfer decreases from Br 182.4 to Br 139.1 when the weight of population increases from 50 percent to 70 percent, compared with Br 234.8 and Br 218.6 in the previous options in Table 7. Similar results hold for Dire Dawa. Such a system—with a high weight of population at the expense of the weight of own revenue effort—would reduce the net per capita transfer amount for states with a strong revenue base such as Addis Ababa. This could, in turn, blunt their incentive to expand their tax collections.

Table 8.Ethiopia: Options for Allocating Grants—Using Different Weights
States and Provisional AdministrationsTotal Existing Allocations

(In millions of birr)
Per Capita

Transfers
Greater Weight for Population1Per Capita

Transfers
Greater Weight for Population2Per Capita

Transfers
GrossNetGrossNetGrossNet
(In birr)(In millions of birr)(In birr)(In millions of birr)(In birr)
Tigray295.991.075.0225.069.253.3231.271.155.1
Afar128.3163.4152.549.863.452.553.167.656.7
Amara811.859.550.6906.966457.5948.169.460.6
Oromiya1,076.463.149.21,211.971.057.21,232.272.258.4
Somali164.193.366.1142.380.953.7137.478.150.9
Benishangul/Gumuz32.465.555.730.762.152.333.166.957.0
SEPA3649.061.752.6698.166.457.2730.169.460.3
Gambela9.685.858.59.080.152.98.777.750.4
Harari15.0172.682.312.7145.655.310.2117.026.6
Addis Ababa480.1229.492.2381.9182.445.3291.1139.11.9
Dire Dawa33.2241.198.027.6200.557.420.6149.96.8
Sources: Ministry of Economic Development and Cooperation; and authors’ estimates.

Population with a weight of 0.5 and revenue and inequality with a weight of 0.25 each.

Population with a weight of 0.7 and revenue and inequality with a weight of 0.15 each.

Southern Ethiopian People’s Administrative region.

Sources: Ministry of Economic Development and Cooperation; and authors’ estimates.

Population with a weight of 0.5 and revenue and inequality with a weight of 0.25 each.

Population with a weight of 0.7 and revenue and inequality with a weight of 0.15 each.

Southern Ethiopian People’s Administrative region.

Additional Variables

The additional variables introduced in the simulations are area and a lump-sum grant for each state irrespective of its size or needs. The rationale for using area is that it accounts for differences in the cost of providing services. Public administrations in sparsely populated states face higher production costs than those in more populated areas. Higher costs for building infrastructure and for shipping supplies are some of the examples. Scale economies in the production of services such as health care, education, and water provide another justification for using area as an indicator.

The rationale for introducing a lump-sum grant relies on indivisibilities. That is, costs for some activities do not depend on the size of the population served, such as general administration, roads, and airports. It should be noted, however, that area and the lump-sum grant account for different factors. Whereas area is related to density of population, the lump-sum grant benefits states that are very small.

However, the lump-sum grant, when assigned a large weight in the formula, may stimulate formation of new inefficient jurisdictions. Some countries, such as Colombia, are presently confronted with this problem, because they give relatively high weight to lump-sum grants.

Two simulations are reported in Table 9. In the first, the inequality indicator is replaced by an area indicator with a weight of 20 percent. The weight for population is 60 percent, and that of own revenue is also 20 percent. In the second simulation, the weight of population falls to 55 percent, to make room for the lump-sum grant that has a weight of 5 percent. Table 10 reports the results of the simulations with a lower weight given to the own revenue effort variable (10 percent) than in the preceding simulation.8

Table 9.Ethiopia: Options for Allocating Grants—Additional Variables
States and Provisional AdministrationsTotal Existing Allocations

(In millions of birr)
Per Capita

Transfers
Including Area1

(In millions of birr)
Per Capita

Transfers
Including Area and Lump Sum2

(In millions of birr)
Per Capita Transfers
GrossNetGrossNetGrossNet
(In birr)(In birr)(In birr)
Tigray295.991.075.0237.172.956.9241.874.458.4
Afar128.3163.4152.5105.3134.1123.2119.2151.8140.9
Amara811.859.550.6807.859.250.3774.156.747.8
Oromiya1,076.463.149.21,165.968.354.51,119.665.651.8
Somali164.193.366.1285.9162.6135.4296.2168.5141.3
Benishangul/Gumuz32.465.555.759.1119.5109.774.1149.7139.9
SEPA3649.061.752.6630.760.050.8608.657.948.7
Gambela9.685.858.525.4227.2200.041.8373.7346.4
Harari15.0172.682.312.4142.552.228.9332.2241.8
Addis Ababa480.1229.492.2333.2159.222.0342.2163.526.3
Dire Dawa33.2241.198.033.1240.297.249.4358.5215.5
Sources: Ministry of Economic Development and Cooperation; and authors’ estimates.

Population with a weight of 0.60 and revenue and area with a weight of 0.20 each.

Population with a weight of 0.55, area and revenue with a weight of 0.20 each, and lump sum with a weight of 0.05.

Southern Ethiopian People’s Administrative region.

Sources: Ministry of Economic Development and Cooperation; and authors’ estimates.

Population with a weight of 0.60 and revenue and area with a weight of 0.20 each.

Population with a weight of 0.55, area and revenue with a weight of 0.20 each, and lump sum with a weight of 0.05.

Southern Ethiopian People’s Administrative region.

Table 10.Ethiopia: Options for Allocating Grants: Using Different Variables
States and Provisional AdministrationsTotal Existing Allocations

(In millions of birr)
Per Capita

Transfers
Including Area1

(In millions of birr)
Per Capita

Transfers
Including Area and Lump Sum2

(In millions of birr)
Per Capita Transfers
GrossNetGrossNetGrossNet
(In birr)(In birr)(In birr)
Tigray295.991.075.0239.473.657.7244.275.159.1
Afar128.3163.4152.5107.5136.91261121.4154.6143.8
Amara811.859.550.6858.162.954.0824.460.451.5
Oromiya1,076.463.149.21,193.569.956.11,147.167.253.4
Somali164.193.366.1278.9158.6131.4289.2164.5137.3
Benishangul/Gumuz32.465.555.760.7122.8112.975.7153.014.3.2
SEPA3649.061.752.6668.463.554.4646.361.452.3
Gambela9.685.858.525.0223.2196.041.4369.7342.4
Harari15.0172.682.39.8112.221.926.2301.8211.5
Addis Ababa480.1229.492.2228.7109.2−27.9237.7113.6−23.6
Dire Dawa33.2241.198.025.9187.944.842.2306.2163.1
Sources: Ministry of Economic Envelopment and Cooperation; and authors’ estimates.

Population with a weight of 0.7, area with a weight of 0.2, and revenue with a weight of 0.1.

Population with a weight of 0.65, area with a weight of 0.20. lump sum with a weight of 0.05, and revenue with a weight of 0.1.

Southern Ethiopian People’s Administrative region.

Sources: Ministry of Economic Envelopment and Cooperation; and authors’ estimates.

Population with a weight of 0.7, area with a weight of 0.2, and revenue with a weight of 0.1.

Population with a weight of 0.65, area with a weight of 0.20. lump sum with a weight of 0.05, and revenue with a weight of 0.1.

Southern Ethiopian People’s Administrative region.

The introduction of two new variables (area and lump-sum grant) allows a better adaptation of the formula to commonly perceived expenditure needs. For example, the Somali state, which is poor, spread out, and sparsely populated, would benefit from the new distribution compared with the baseline allocation for 1995/96. Small and poor states like Benishangul/Gumuz and Gambela are now situated well above the national average. Inclusion of area in the formula appears to give a relatively fairer distribution of grants compared with the baseline. The introduction of a lump-sum grant benefits particularly the smaller states (such as Afar, Benishangul/Gumuz, Gambela, and Harari), as well as the provisional administrations of Addis Ababa and Dire Dawa, compared with the previous option. This benefit disappears when the share of revenue in the formula is decreased.

Conclusions

The existing grant system in Ethiopia plays an important role in financing states and has been modified twice since 1993/94. Transfers to state governments in Ethiopia have consisted of general purpose block grants only. No specific, conditional grants are presently extended. Given the need to achieve multiple goals and the need to encourage the states to provide at least the minimum standard of services, it may be more appropriate to use a combination of general and specific grants, although in Ethiopia’s case the mechanism for monitoring the implementation of these standards may be lacking. General purpose grants, however, are preferred by decentralized units in a federal structure because they do not intrude on their decision-making authority. They also render the distribution of the national resources by the federal government more transparent.

The options for reform presented in this chapter do not consider the introduction of specific grants, assuming a continuing strong preference in Ethiopia for a system of block grants. These options seek to better equip the system to meet its objectives. The initial reform steps could include using more transparent and reliable variables in the present formula, altering the weights assigned to different variables and introducing new variables. The initial reforms could include the following:

  • Replacing the present I-distance indicator with a simpler and more transparent alternative.

  • Increasing the weight for population.

  • Reducing the weight for state revenue.

  • Including a state’s area.

  • Giving a small lump-sum grant, to take into account indivisibilities.

  • Applying the grant-determining framework to total resources net of the states’ own tax revenues.

The simulations indicate that it is difficult to assign a precise weight for each indicator. To some extent, this would have to be determined by the political process.

The establishment of an independent Grants Commission would also help, by providing the framework and necessary input for strengthening the grant formula. This commission could make recommendations on grant-related matters to the Federal Council, as envisaged under the Constitution.9

It should be noted that although the options discussed in the chapter represent improvements over the existing grant formula, they nevertheless amount to “gap filling,” that is, they seek to bridge the gap between the state’s actual revenues and expenditures rather than to provide transfers on the basis of relative expenditure needs and relative revenue capacities.

The authors wish to thank Ehtisham Ahmad, Benedict Clements, Dawn Rehm, Edgardo Ruggiero, Gerd Schwartz, and Teresa Ter-Minassian for many helpful comments, and Manfred Koch for his assistance and advice.

Proclamation numbers 7/1992, 33/1992, and 41/1993. The decentralized subnational units are referred to as states in the Constitution. Although proclamations preceding the Constitution label the decentralized subnational units as regions, this chapter refers to them as states, in keeping with the usage in the Constitution.

Ethiopia has about 15,000 kilometers of trunk and link roads and about 8,000 kilometers of rural roads. No major extension of trunk roads is planned by the federal government.

The study should also include a forecast of income generated by the project, based on realistic economic indicators.

The lowest level of subnational government in Ethiopia is weredas, followed by zones.

The I-distance indicator is a synthetic variable that seeks to capture the differences in the levels of social and economic development among the states. It is based on eight distinct, quite heterogeneous, variables: length of rural roads, share of rural population in total population, per capita industrial production, per capita crop food production, density of telephone lines, number of post offices, hospital beds in relation to total population, and pupils in elementary schools in relation to total population. These variables are combined by means of a rather awkward formula aimed at minimizing the intercorrelation between variables. This indicator is weighted by population. In other words, the choice of the I-distance indicator is dictated by the desire to achieve interstate equity in the distribution of grants, against the background of uneven revenue-generating capacities.

The following example clarifies this point. Suppose a state has no own tax revenue. In this case, gross grants will coincide with the net grant amount. As this state starts to collect revenue, it will receive larger gross grants. An increase in own tax revenue will have a negative impact on the net grants; this effect is greater than the positive impact on gross grants. States with rapidly expanding revenue will eventually end up with no net grants, that is, they will find themselves in a situation in which they have to finance their expenditures entirely out of their revenue.

Each indicator is calculated to measure, for each state, the difference from the national average. That is, for the first valiant:

where

  • Sti is the number of students in all schools in state i,

  • Hei, is the number of medical personnel in state i, and

  • Popi, is total population of state i.

  • St and He, and Pop are the corresponding values for the whole country. The two indicators are linked by multiplication.

  • For the second, the income-related indicator:

  • Ei = Popi/PopEli/El and,

  • Pi = Popi/PopPhi/Ph,

where

  • Eli is electricity consumption in state i,

  • Phi is the number of telephone lines in state i, and

  • El and Ph are the corresponding values for the entire country.

The negative net transfer for Addis Ababa in Table 10 implies that either Addis Ababa will have to transfer part of its resources to other regions or the federal government will have to find additional resources to offset the negative entry.

The Federal Council in Ethiopia is composed of members elected by the State Council.

    Other Resources Citing This Publication