The World’s Largest Open Access Agricultural & Applied Economics Digital Library

This document is discoverable and free to researchers across the

globe due to the work of AgEcon Search.

Help ensure our sustainability.

Give to AgEcon Search

AgEcon Search
http://ageconsearch.umn.edu
[email protected]

Papers downloaded from AgEcon Search may be used for non-commercial purposes and personal study only.
No other use, including posting to another Internet site, is permitted without permission from the copyright
owner (not AgEcon Search), or as allowed under the provisions of Fair Use, U.S. Copyright Act, Title 17 U.S.C.
 The Impact of Fiscal Policy on Poverty in Ethiopia: A
Computable General Equilibrium Microsimulation
Analysis1

Daniel Abraham Mengistu2

Abstract
Ethiopia has implemented various fiscal policy reforms in the past decade. Most
of these reforms center on indirect taxes and pro-poor expenditure patterns. This
study investigates the economy-wide impacts of these fiscal policy changes on
poverty. To this effect, the study used a static computable general equilibrium
(CGE) model linked to a microsimulation (MS) model. The CGE model used the
2005/06 social accounting matrix (SAM) and the MS model used the 2004/05
Household Income, Consumption and Expenditure (HICE) survey to investigate
household poverty by way of the consumption expenditure changes from the
CGE model. The fiscal policies simulated are domestic indirect taxes,
government consumption expenditures, and government transfers to households.
The findings of the study suggest that the increase in revenue from indirect taxes
has worsened the poverty state of households. The results from the CGE model
have all shown decline in real GDP, sectoral output, employment and welfare. In
contrast, the study found improvements in the poverty state of households as a
result of the introduction of various short-run expenditure measures. However,
examination of the net effect revealed worsening poverty at the national level in
general and for rural households in particular. On the other hand, poverty
tended to decline among urban households. The major conclusion is that the tax
policy has dominant adverse effect on poverty in the short-run. Thus, policy
makers need to take into account these adverse effects and come up with pro-
poor spending policies that would protect households from the negative strains
while the financing policies go along.

Keywords: Fiscal policy, poverty, indirect taxes, government consumption expenditure,

government transfers, social accounting matrix, computable general equilibrium, household
income, consumption and expenditure survey, microsimulation
JEL Classification: H30, H53, I32

1
I sincerely thank my thesis advisor Dr Tadele Ferede for his meticulous
professional advice. I am also grateful to Ermias Engida, Sinshaw Tamiru, Alekaw
Kebede, Tewodros Tebekew, and Mathias Assefa for their helpful comments,
suggestions, and support.
2
Lecturer at Arba Minch University. Tel. +251 911 938303; Email: [email protected]

25
Daniel Abraham Mengistu: The Impact of Fiscal Policy on Poverty in Ethiopia:…

1. Introduction

The state of poverty in Ethiopia is among the worst in the world measured by
most socio-economic and human development indicators. Over the period
2005-2010, the Human Development Report (HDR) ranked the country as
157th, out of 169 countries, in human development index (HDI) and second
highest in multidimensional poverty index (MPI)3 next only to Niger
(UNDP, 2010). Though the country is reported to have improved its HDI
rank in the report, there is little doubt that a long and rough way awaits as
poverty dominates the center stage in the endeavor of change in the country.

Poverty reduction is one of the principal development challenges facing low

income countries. The challenges and impediments to reduce poverty are
difficult in countries like Ethiopia where poverty is persistent and
widespread, vulnerability to shocks is high, and income is extremely low
(Abebe, 2005). Hence, addressing the problems of poverty has become the
priority of development policy and Ethiopia considers poverty reduction as
its primary development goal (MoFED, 2010).

Like most African countries, Ethiopia has implemented the policy directions
of the World Bank and the International Monetary Fund in the 1990s and
2000s. The 1990s saw orientation of the earlier development approaches in
the form of the Structural Adjustment Programs (SAPs). According to
Alemayehu and Alem (2006), these policies mainly opted to the welfare
improvements through the liberalization and conservative macro-policies. A
series of economic reforms took place to take the country from a command
to a market economy which opted to bring macroeconomic stability and
workable business climate. The country also adopted the Agricultural
Development Led Industrialization (ADLI) strategy which was considered as

3
The UNDP (2010) introduced MPI for the first time to complement money-based
measures by considering multiple deprivations and their overlap. The MPI is 0.582
for Ethiopia and 0.64 for Niger.

26
 Ethiopian Journal of Economics Vol. XXII No 1, April 2013

a necessary step towards economic growth, poverty reduction, and industrial

development.

In the new millennium, as a way to join the Highly Indebted Poor Countries
(HIPC) initiative of the World Bank, Ethiopia embarked on new policy
directions by developing an Interim Poverty Reduction Strategy Paper (I-
PRSP) in the year 2000 (AFRODAD, 2005). In 2002, the country launched the
full-PRSP known as Ethiopia’s Sustainable Development and Poverty
Reduction Program (SDPRP). As the Millennium Development Goals (MDG)
initiative started in 2000, the SDPRP targeted economic growth averaging 7%
per annum in order to reduce poverty by half in 2015. A second phase of the
PRSP process, a Plan for Accelerated and Sustained Development to End
Poverty (PASDEP) was launched in 2005 as a guiding framework for the
period 2005-2010. Most of its strategic directions were continuations of the
SDPRP in relation to human development, rural development, food security,
and capacity building, but it added new directions like intensifying agricultural
commercialization, private sector participation, and scaling up the efforts to
achieve the MDGs (MoFED, 2010).

Though poverty reduction is a forefront agenda, as Agenor, 2004 argued, the

policies to be pursued to its attainment are complex and openly contested.
Understanding the direct and indirect impacts of macroeconomic policy on
poverty still remains a key policy challenge (Aziz, 2008; Mallick, 2009).
IMF (2001), for instance, stated that there is a large literature on issues of
poverty and poverty reduction while there is a lack of detailed understanding
of the relationships between macro- policies and poverty. In Ethiopia,
likewise, despite the abundant discourse on poverty, the rigorous economy-
wide studies to address its link with fiscal policies are scarce. Mentions can
be made of two studies. Munoz and Cho (2003) focused on the poverty
impacts of the 2003 E.C. tax reform using incidence analysis and didn’t find
major adverse effects on the poor. Kasahun (2003), on the other hand,
focused on the reductions in government consumption and import tariff, and
nominal exchange rate devaluation using a CGE – MS analysis and found

27
Daniel Abraham Mengistu: The Impact of Fiscal Policy on Poverty in Ethiopia:…

out poverty reducing results. The literature, however, could be extended

further by examining the economy-wide impact of fiscal policies on poverty
using a recent and comprehensive data.

Fiscal policy is one of the few and most important instruments available to
governments of poor nations in fighting poverty (Johannes et al., 2006; Kiringai
et al., 2006). The interest of this study on fiscal policy, among the
macroeconomic policies, emanates mainly from the fact that it can play a role in
poverty reduction as an indirect intervention besides being one of the influential
direct interventions targeting specific groups or pro-poor sectors which are
vulnerable to economic or natural shocks (Damuri and Perdana, 2003).

However, fiscal policies that were designed as pro-poor might in fact have
no impact on poverty or sometimes even worsen the poverty situation if the
direct and indirect effects of the link are not well articulated. Ethiopia has
implemented fiscal policy reforms in the past decade mainly in relation to
indirect taxes and pro-poor expenditures. These policies have short-run and
long-run implications on the poverty state of households. Thus, the central
research question of the study is: what are the short-run impacts of fiscal
policies on poverty of households in Ethiopia?

Among the variety of policy analysis tools, CGE models are widely used
because of their ability to illustrate the feedback effect between different
markets, and produce disaggregated results at the sectoral or microeconomic
level within a consistent macroeconomic framework (Wang et al., 2010). As
Cury et al. (2010) argue, formal assessments on the poverty effects of
economic policies using CGE models took shape in the 1970s and 1980s. As
a result, a bulk of empirical studies was conducted to examine poverty
mainly by linking CGE with MS models in many developing countries (Cury
et al., 2010; Wang et al., 2010).

In this line, the objective of the study is to analyze the short-run impact of
fiscal policy on poverty in Ethiopia. To this effect, scenarios of changes in

28
 Ethiopian Journal of Economics Vol. XXII No 1, April 2013

domestic indirect taxes, government consumption expenditure, and

government transfer expenditures were used. Policy combinations that
represent the net effects are evaluated for scenarios with respect to total
government recurrent expenditure (government consumption plus transfers)
and combined impact of the financing (revenue) and spending schemes
applied together. The remaining part of the paper is organized as follows.
Section 2 reviews the literature briefly. Section 3 focuses on overview of
fiscal policy in Ethiopia. Section 4 introduces the data base (SAM) and
specifies the theoretical framework for the CGE and MS models. Section 5
discusses the results from the CGE and MS models. The final section
concludes and provides implications for policy and future research.

2. Literature Review

The common understanding in policy making was to evaluate macro policies

based on their macroeconomic objectives. But this notion was gradually
replaced, mainly in developing countries, in the sense that the policies are
also to be judged based on their impact on poverty and income distribution
(Agenor, 2004). Much progress has been achieved in recent years in
understanding the various transmission mechanisms despite the difficulties
due to the multidimensionality of poverty.

The 1980s and 1990s were periods where macro adjustment policies were
implemented with the intention of achieving a wide range of macroeconomic
objectives. But most literature criticized the adverse impacts of adjustment
policies on poverty and income distribution. Agenor (2004) identified direct
and indirect channels through which macro policies could adversely impact
poverty in times of such macroeconomic adjustment. The major indirect
effects identified operate through aggregate demand, the rate of economic
growth, distributional effects, employment and the like. Contractionary
policies affect aggregate demand and employment (and thus poverty)
through reductions in transfers and subsidies, and expenditure cuts (mainly
capital spending). Moreover, reductions in public spending have divergent

29
Daniel Abraham Mengistu: The Impact of Fiscal Policy on Poverty in Ethiopia:…

negative effects on private spending. Rates of growth are affected as well

when the poor are constrained of their public transfers which deplete
economic savings. In addition, due to the complementary effects of private
and public spending, decreases in public spending may lead to the same on
private spending that have negative impact on the rate of growth. In relation
to income distribution, the argument was that initial levels matter to the
extent and nature of subsequent growth and redistribution (Agenor, 2004).

In most developing countries, the beginning years of the new millennium

saw shift in policy away from structural adjustment programs. The recent
approach, as Lipton and Ravallion (1993) attest, has evolved in the sense that
the government could intervene to determine the pattern of growth. In other
words, emphasis is evolving away from the link between economic growth
and poverty reduction to explore what policies, beyond growth itself,
contribute to poverty reduction and income distribution. With respect to
fiscal policy, this refers to expansionary policies that target most of the
expenditure to basic social services like education and health (or more
elaborately) on primary education, preventive health, safety net programs
(transfers to households) and the like. This approach attempts to cut overall
expenditures without affecting the main entry points of such pro-poor
expenditures. The expenditure reallocation came up with distinctions
between levels of expenditure and composition (nature) of expenditure.

Figure 1 depicts the short run transmission mechanisms of fiscal policy on

poverty. The framework in this study focuses on the transmission
mechanisms of the impacts on poverty of government consumption and
transfers to households on the spending side and (domestic) indirect taxes on
the revenue side (Damuri and Perdana, 2003). When we look into
government expenditure financed by increased indirect taxation (increase in
VAT, excise or service taxes), this brings issues of who bears the tax burden
since indirect taxes are mostly imposed on consumable commodities. As
Gunter (2005) states, there is a broad agreement on the beneficial effects on
the poor of increase in the budget share of priority sectors and better

30
 Ethiopian Journal of Economics Vol. XXII No 1, April 2013

targeting of public expenditures. But the more we come to indirect (mostly

price related) effects, the more it gets blurred on how fiscal policy could lead
to positive impacts on poverty. Most studies agree that the labor market
constitutes the most central mechanism through which macro policies are
transmitted to poverty (Islam, 2001; Mulat et al., 2003; Agenor, 2004;
Heintz et al., 2008).

As depicted in the figure, the principal transmission mechanism envisaged in

the study is pointed by the “block arrows”. It starts from the indirect taxes
that are the source of revenue to government budget which is spent on
consumption and transfer expenditure schemes. These revenue and
expenditure changes bring about employment, price, income and expenditure
adjustments by households that lead to changes in poverty. The study didn’t
extend the analysis to government investment expenditure, effects of direct
taxes, external borrowing and inflation. Damuri and Perdana, 2003 arrived at
similar results with this study for Indonesia though it differs in its
incorporation of government investment into the analysis.

To this effect, the study used a CGE – MS model. We can site numerous
advantages of CGE modeling to analyze indirect effects. In CGE models,
general equilibrium effects can be accounted for, interactions of different
measures can be investigated, complex micro-macro relationships can be
performed better, and constraints of linearity can be reduced to the minimum
(Iqbal and Sidiqqui, 2001). Besides, such models have ability of examining
variety of incidence assumptions and socioeconomic divisions including
various welfare measures and behavioral responses (Gemmel and Morrisey,
2002). These models are also consistent with generally accepted
microeconomic theory, have significant structural detail, and their general
equilibrium nature - changes in one area of economic activity affecting the
rest of the economy- elevates their influence for economic analysis (Bibi et
al., 2010).

Although CGE models are among the most influential tools in applied
economics and have provided unique insight into the policy-poverty debate,

31
Daniel Abraham Mengistu: The Impact of Fiscal Policy on Poverty in Ethiopia:…

they have also raised the sophistication of prediction in the policy debate
(Iqbal and Sidiqqui, 2001). The literature raises various limitations of CGE
models in relation to quality and intensity of data requirement, choice of
parameters, choice of functional forms, calibration of the model, unrealistic
assumptions of neoclassical theory, lack of sensitivity analysis, validity of
predictions for policy etc. One fact that most researchers agree on is that the
analysis is still evolving, incorporating new dimensions and methods in
which we can see development of software packages like GAMS (General
Algebraic Modeling System) in simplifying the complex model simulations.
Apart from this, we find that other modeling approaches are confronted with
critical limitations as well. For instance, some consider the criticisms against
CGE models as part of the wide debate concerning the issue of contributions
of empirical economics in general (Iqbal and Sidiqqui, 2001).

Figure 1: Transmission mechanisms of fiscal policy impacts on poverty

(Domestic) indirect taxes,

eg. VAT, excise tax, service
tax External borrowing
Direct taxes, e.g.
income taxes
Pressure on inflation
Government budget
Work-leisure
preference switch
Government expenditure

Government consumption Government investment Direct cash transfers

expenditure to households

Employment and price adjustments Income adjustment

Adjustments in households’ income and expenditure

Poverty
Source: Modified version of Damuri and Perdana (2003)

32
 Ethiopian Journal of Economics Vol. XXII No 1, April 2013

Some argue that CGE models are more appropriate than traditional partial
equilibrium models as the latter do not account for the economy-wide
multiplier effects (don’t allow knowledge of who gained and who lost) and
as they overestimate sectoral benefits ignoring the negative repercussions
evident in a general equilibrium framework (Ahmed and Donoghue, 2004).
The underlying cause of this is that, partial equilibrium models implicitly
assume fixed-prices where as CGE models give due consideration to relative
price changes in commodity and factor markets (Wobst, 2001). Advances in
computer technology and numerical simulation soft-ware exercise have
allowed the transformation from such partial equilibrium approach to a general
equilibrium approach which can very modestly incorporate many more sectors
and complex behaviors (RTI, 2008). After the 1970s, the general equilibrium
approach became advanced enough to incorporate imperfect information,
increasing returns, price rigidities, and many extensions addressing various
markets and institutions (Sinha and Latigo, 2003).

3. Characterizing Fiscal Policy in Ethiopia (1999/00 – 2009/10)

For the decade from 1999/00 to 2009/10, MoFED and NBE reports from
EEA (2009) data base show that the real GDP has grown by 129% in
constant 1999/00 prices. As the reports disclose, it has shown sluggish real
growth around the down of the new millennium (with a negative growth in
2002/03) but started to consistently record double digit figures after 2003/04
except 2008/09 in which a relatively lower 8.8% real growth was recorded.

Though the reports show that government revenue and expenditure have
shown major increases, the fiscal balance as a proportion of GDP continued
to record negative figures. In this period, fiscal policy was aimed at
maintaining the deficit at a sustainable level besides increasing pro-poor
expenditures. The financing aspect of these huge expenditures has been of
great concern since the budget couldn’t be covered from domestic revenue
collection alone. To improve on this, reforms took place with the aim of

33
Daniel Abraham Mengistu: The Impact of Fiscal Policy on Poverty in Ethiopia:…

strengthening domestic borrowing, domestic revenue collection capacity,

and mobilizing external aid and borrowing.

Comparing 1999/00 and 2009/10 periods, government reports claim that

capital expenditure, including external assistance, has shown significant
increases in comparison to the current expenditure. It has shown increases up
to five times its 1999/00 levels. Capital expenditure as a ratio of GDP has
risen from 3.8% to 10.3% and total expenditure as a ratio of GDP has risen
from 20% to 55.3%. Capital expenditure on social development in education,
health, social welfare and the like has also increased covering 2.6% of GDP
in 2009/10. Current expenditure, on the contrary, has decreased as a share of
GDP and total expenditure from 20.6% and 80% to 8.4% and 44.7% in the
ten year period. However, in real terms, current expenditure has grown by
10% in 2009/10 from 2005/06 values whereas its basic component, final
government consumption expenditure, has risen by a mere 4%. Notably, the
share of social services like education and health has shown increases in
2009/10 like the growths in social development for the same sectors. Capital
expenditure showed constant increases in the new millennium replacing
current expenditure as the largest component of total expenditure (Annex 1).

We can resort to the 2005/06 Ethiopia SAM for data on government transfers
to households as a component of general government recurrent expenditure
which also includes government consumption expenditure on goods and
services and government savings. Government consumption takes the lions
share in this account (68.5%) where as government transfers plus external
interest payments cover about (8.5%) and the remaining is government
saving. Of the government transfers, transfers to households take about
6.6%.

When we come to the revenue side, tax and non-tax revenues have increased
in the ten year period with larger shares recorded by the tax component
compared to the non-tax component. Tax revenue as a share of total revenue
and grants has increased from 54.6% to 65.4% from 1999/00 to 2009/10

34
 Ethiopian Journal of Economics Vol. XXII No 1, April 2013

where as non-tax revenue declined. In the same period, direct taxes have
increased from 19.9% to 22.5% whereas indirect taxes have increased by a
larger amount from 34.7% to 42.9% (Annex 2). As a ratio of total revenue
and grants, the size of government revenue has recorded constant increases
in which most of the changes are results of the tax reform introduced. In
2009/10, domestic indirect taxes have increased by 34% as proportions of
total revenue and grants compared to the 1999/00 period.

The Ethiopian government has set out to achieve a large sum of revenue
collection in aggregate during the course of a five-year Growth and
Transformation Plan (GTP) from 2010/11 to 2014/15 thereby raising the
country’s annual domestic revenue to GDP from 14% to 17.1% and annual
tax revenue to GDP from 11.3% to 15% (MoFED, 2010). On the expenditure
side, resource allocation is planned towards growth enhancing (mainly
agriculture and infrastructure) and social sectors (mainly education and
health). In the five year GTP; increases as percentages of GDP are
anticipated for total expenditure (18.6% to 23.7%), capital expenditure
(10.3% to 14.4%), recurrent expenditure (8.4% to 9.3%), and total poverty
oriented expenditure (12.3% to 17.3%). Expenditures in the social sectors
show great increases in the plan which reveals tendency to continue from the
PASDEP. Also, the GTP targets to reduce total poverty head count from
29.2% to 22.2%.4

When we consider poverty-oriented spending as a share of total expenditure,

the trend doesn’t show significant changes in the five year period from
2005/06. In fact, the trend is one of decrease in the cases of agriculture,
education and social welfare, whereas it fluctuated in the case of health
expenditures. Road construction is the only spending, in this case, to have
constantly increasing share of the total expenditures with most of the funds
allotted to construction of rural roads or roads for poor areas in general. A
point to note here is that education expenditures are the largest, about a

4
The base period for all these forecasts of the GTP is 2009/10.

35
Daniel Abraham Mengistu: The Impact of Fiscal Policy on Poverty in Ethiopia:…

quarter, whereas health and social welfare spending constitutes the smallest
among the poor-related expenditures (Annex 3).

Domestic indirect taxes have been the center of focus in the past decade.
Since its introduction in 2003 with the objectives of reducing distortions by
other indirect taxes, VAT has been one of the principal sources of revenue
for the government and yet there is a large unexploited potential in the use of
this tax. The share of local and import VAT as a share of total indirect taxes
and GDP has been at around 50% and 4% respectively over the five year
period examined. Domestic indirect tax revenue has increased by 80% in
2009/105 from the 2005/06 values though the share to real GDP showed
minor increase from 2.6% to 2.8%. Excise taxes, both domestic and foreign,
have also steadily increased in the period. The share of most of the taxes has
fluctuated over the period though most showed minor increases from the
2005/06 period. Total indirect tax revenue to the government, which is the
sum of domestic indirect tax revenue and foreign trade tax revenue, has
shown an overall 48% increase in real terms in 2009/10 compared to the
2005/06 period though share to real GDP stagnated (Annex 4).

The new five year GTP is already underway. But the huge task ahead is how
to maximize overall domestic tax and non-tax revenue using tax reforms like
the VAT without constraining the lives of consumers. Pro-poor spending
schemes have been constrained by this issue of financing as they are dubbed
to lack the financial requirements and possible (investment) sources.

4. Methodology
4.1 The Social Accounting Matrix

The benchmark data used for calibration in CGE modeling is the Social
Accounting Matrix (SAM) (Thurlow, 2004). A SAM is a comprehensive and
consistent, economy wide data framework or set of accounts that has detailed

5
There was around 50% increase in domestic indirect tax revenue collected in
2009/10 compared to 2008/09.

36
 Ethiopian Journal of Economics Vol. XXII No 1, April 2013

quantification for economic flows of incomes and expenditures in an

economy, usually a nation, for a given period of time, mostly a year
(Decaluwe et al., 1999; Lofgren et al., 2002).

In this study, we used the 47×47 aggregated SAM prepared by EDRI (2009).
In this 15 sector SAM,6 productions and incomes of the various agro-
ecological zones were aggregated into one account before further
aggregations were made. The matrix has 14 activities,7 15 commodities, 4
factors of production (labor, land, livestock and capital), 7 institutions (an
enterprise, a government, 4 households, and a rest of world or ROW), 3 tax
accounts (direct tax, sales tax and import tax), transaction costs (total
margins), stock changes,8 and S – I account.

The four household categories distinguished are rural-poor, rural non-poor,

urban-poor and urban non-poor. The sales tax account incorporates local
VAT, domestic excise tax and service taxes whereas the import tax account
incorporates import duty, sur tax, import excise tax, import VAT, and
withholding tax.

6
The production activities are for teff, maize and wheat, non-traded agriculture,
exportable cash crops, livestock, food processing, chemicals, machinery, other
manufacturing, construction, utility, domestic trading, private services; and
government services. These activities are basically aggregations and disaggregations
from the agriculture, industry and service sectors. The agricultural sector includes five
production activities: teff, maize and wheat, non-traded agriculture, exportable cash
crops, and livestock. The industrial sector includes five production activities:
construction, food processing, other manufacturing, chemicals and machinery. And,
the service sector includes four production activities: utilities, domestic trading, private
services and government services. There are three activities that produce more than
one commodity. These are cash crop production activity which produces cash crops
for export and non-traded agricultural commodities, livestock activity which produces
food products and raw materials for further production, and activities for utility which
produces utilities and machinery.
7
The commodity account for fuel (cfuel) does not have activities account as Ethiopia
is non-oil producing nation. Thus, the 14 activities produce 14 commodities with
some combinations, as mentioned above, but commodity fuel doesn’t have domestic
production activity.
8
A stock change represents inventory investment by sector of origin (Lofgren et al., 2002).

37
Daniel Abraham Mengistu: The Impact of Fiscal Policy on Poverty in Ethiopia:…

4.2 The Computable General Equilibrium Model

The static CGE model used is based on a comparative static standard

neoclassical-structuralist model developed in IFPRI by Lofgren et al. (2002).
It follows the neoclassical – structuralist modeling tradition with additional
features included like treatment of transaction costs, household consumption
of non-marketed (home) commodities, and separation between production
and commodities (Lofgren et al., 2002; Thurlow and Seventer, 2002). The
model incorporates both paradigms to better represent the real world features
that are found in developing countries.

The standard CGE model is a system of simultaneous linear and non-linear

equations. The equations of the model include non-linear first order
optimality conditions for production and consumption decisions which are
driven by maximizations of profit and utility respectively. The CGE model
also includes equations for closures. The term ‘closure’ implies the way
adjustment is made in the economy to ensure equilibrium or an indicator on
how the model gets solved (Tadele, 2010). The choice of closures, hence,
provides the macroeconomic settings to conduct the policy simulations
(Damuri and Perdana, 2003). The structure of the CGE model is divided into
four major blocks: price, production and trade, institutions and system
constraint blocks.

The production in the economy takes place in each activity to yield the
commodities produced domestically. The producer is assumed to maximize
profit (subject to a production technology) which is the surplus after
payments are made to (primary) factors and intermediate inputs. The
production technology connotes a multi-level production function. It chooses
between a constant elasticity of substitution (CES) and Leontief technology
at the top level of the technology nest. In this study, the technology at the top
level is a Leontief function of the quantities of value-added and intermediate
inputs that yield commodity outputs in the production process. The value-
added part is a CES function of primary factors. This CES specification for

38
 Ethiopian Journal of Economics Vol. XXII No 1, April 2013

value-added makes producers respond to dynamics in factor returns by

substituting among available factors (Thurlow, 2004). The aggregate
intermediate input part is a Leontief function of composite commodities
partly domestic and partly imported (Lofgren et al., 2002).

The notion of closure rule implies equality of equations and endogenous

variables which requires fixation of some variables for the model to have a
solution. The choices on closures do not have impact on solutions of the base
simulation but affect other simulations, and in addition, the choice per
closure doesn’t constrain the choice for the other closures (Damuri and
Perdana, 2003). This study selected the model closures that are applicable to
the Ethiopian economy.

The standard CGE model has closures for factor markets and the
macroeconomic system. In this study, the factory closures are that labor is
unemployed and mobile across sectors; land is fully employed and mobile
across sectors, and capital is fully employed and activity specific. In our
model, labor is not disaggregated into skilled, semi skilled and unskilled. A
cumulative of these subdivisions is made to follow the labor market
characteristics of the large proportion of the labor force in Ethiopia,
unskilled labor. Land and capital are fully employed and hence have fixed
supply whereas labor is unemployed and its employment is flexible. On the
other hand, labor and land are mobile across sectors implying that they can be
employed in different activities. But capital is activity specific as its use is
usually immobile across sectors in Ethiopia.

The macroeconomic closures (balances) are based on the government

balance, the external (current account) balance and the saving-investment (S-
I) balance. In this study, we follow the government closure in which direct
tax rates are exogenous and it is the changes in government savings that
equilibrate the economy. For the external (ROW) balance, foreign saving is
fixed, and thus it is real exchange rate that plays the equilibrating role. In the
S-I balance; we follow a saving-driven investment closure in which we have

39
Daniel Abraham Mengistu: The Impact of Fiscal Policy on Poverty in Ethiopia:…

flexible capital formation but fixed propensities to save for all non-
government domestic institutions. We follow this closure in this model in
which investment adjusts to ensure equilibrium.

In our model, the DPI is the numeraire and hence is fixed whereas the CPI is
made flexible. The CPI is made flexible in order to adjust the expenditures
we used in the microsimulation model. Since price is normalized to one in
the CGE model, the changes in CPI indicate consumer prices changes that
bring about equilibrium within the model.

4.3 The Microsimulation Model

Though the frequently used measure of the extent of poverty has been the
head count index, the FGT measures are considered to be the standard as
they combine the head count index with the poverty gap index and the
squared poverty gap index (Yesuf, 2007; MoFED, 2008).

Foster, Greer and Thorbecke (1984) lumped these measures into one formula
that incorporates the three consistent and additively decomposable (by
income class or region) poverty indices.

The formula for the FGT index is given as:


 z  yi 
q
1
P   
n i 1 

z  ; α ≥ 0, y < z, i=1, 2,…, q; [ y1 < y2 <…< yq
< z < yq+1 <…≤ yn]

Where z is the poverty line, i is the sub-group of individuals with income

below the poverty line, yi is the value of poverty indicator chosen
(consumption expenditure below the poverty line in increasing order)9, n is
the total population size, q is the total number of poor people in the

9
If gi = Z – Yi, then gi represents income (consumption) short fall of the ith individual
(household) and this is assumed to be zero for those above the poverty line (Abebe, 2005).

40
 Ethiopian Journal of Economics Vol. XXII No 1, April 2013

population, and α is the poverty aversion parameter. By setting the value of α

to zero, one, two respectively,10 the FGT poverty formula delivers a set of
poverty indices. Setting α equal to zero, P0, reduces to the head count index
( q n ) measuring the incidence (prevalence) of poverty. Setting α equal to
one, P1, becomes the poverty gap measuring the depth or intensity of
poverty. Setting α equal to two, P2, will be the squared poverty gap
measuring the severity of poverty (the weighted sum of the poverty gaps
themselves)11 (Yesuf, 2007).

In this study, we link the CGE and the MS models in a top-down (sequential)
manner as can be seen from Figure 2. In this top-down approach, the CGE
model is linked with 21,594 households in the 2004/05 HICE survey of CSA
(CSA, 2008). The changes in the fiscal policies introduced in the CGE model
bring about economy-wide changes in the consumer price index and
consumption expenditures of households once we solve the CGE model
using GAMS software. These simulation results for the before and after
shock period are later fed into the MS model using distribution analysis12
(DAD) software that yields the FGT poverty indices.

Figure 2: The Top-down (sequential) approach

Output
A vector of changes in
CGE Model Input  Consumption expenditure
 Consumer price index
Output Poverty indicators

MS Model

Source: Modified version of Colombo (2008) and Dejene et al. (2007)

10
Some refer to the three measures: incidence, intensity, and inequality as the
“Three “I”s of Poverty” (Gemmel and Morissey, 2002).
11
“α” denotes the weight given to the poorest of the poor and so the higher the value
of α, the more is the concern for the poorest (Abebe, 2005).
12
The DAD (distribution analysis) software is “designed to facilitate the analysis
and the comparisons of social welfare, inequality, poverty and equity across
distributions of living standards.” (Duclos et.al., 2010)

41
Daniel Abraham Mengistu: The Impact of Fiscal Policy on Poverty in Ethiopia:…

As per EDRI, 2009, the households in the SAM were categorized into four
as rural-poor, rural non-poor, urban-poor and urban non-poor. The bottom
40% of the households are taken as poor after the households are arranged in
descending expenditure levels. In other words, the bottom two quintiles (4th
and 5th) were considered as poor. Based on this treatment of the HICE
survey in the 2005/06 SAM, this study used the consumption expenditure
level at the demarcation of the top 60% and bottom 40% of the total
households to be the poverty line in estimating the FGT indices (EDRI,
2009).

5. Simulations and Results

5.1 Description of Simulations

Baseline simulation
This scenario is used as a reference point where the economy is evaluated at
times of no policy change or at times where the present policy environment
is maintained.

Simulation 1: Increasing sales tax rate by 80%13

The 15 sector SAM used has a sales tax aggregate account which includes
the domestic indirect taxes of local VAT, local excise taxes and service
taxes. Local VAT contributes the largest to the account followed by service
and excise taxes. The 2005/06 SAM represents the early years of
intensification of the indirect tax reform that come of age after January, 2003
which marked the period in which VAT was introduced. To deal with the
80% huge increases in sales tax revenue in 2009/10 compared to 2005/06,
we opted to a proxy increase in the sales tax rate that would represent the
revenue increase. The underlying fact of the matter is that revenue increased
due mainly to changes in the tax base (not the rate). It is rather difficult to
capture the (price related) impacts on the poor of this expansions of the tax

13
This percentage is based on calculations using EEA (2009) data base as discussed
in section 3 above.

42
 Ethiopian Journal of Economics Vol. XXII No 1, April 2013

base or tax collection and thus we proxied this by changes in the sales tax
rate ( tq ) in the CGE model.
c

Simulation 2: Increasing government consumption expenditure by 4%

This simulation is in relation to government consumption expenditure
policies. As discussed in section 3, the MoFED data retrieved from EEA
(2009) data base shows that final government consumption expenditure has
increased by close to 4% in 2009/10 compared to 2005/06 in real terms
which runs in opposite direction to Kasahun, 2003 that decreased
government consumption by 20% as a result of structural adjustment. In line
with this, we used this 4% value to introduce the shock, with no financing
changes. Government consumption expenditure is explicitly accounted for in
QG
the CGE model as c and thus we look into the impacts of a 4% increase

in government consumption. In the SAM, government consumption

spending is reported for public administration, education and health.

Simulation 3: Increasing government transfers to households by 20%

Another fiscal policy variable of our interest is the transfers from the
government to households. This simulation implies government expenditure
on transfers with no financing changes. The EDRI (2009) report states that it
used the 2004/05 HICE survey for the government transfers to households.
An important point to note here is that we have introduced this policy change
to all poor and non-poor households in the SAM as anti-poverty policies are
designed to hinder households from slipping into extreme poverty and to
minimize the length of the poverty spell once they fall into it (Bigsten and
Abebe, 2007). Thus, based on the 2004/05 HICE survey and the static nature
of our CGE analysis, we have included the non-poor into the transfer scheme
as well. In the HICE survey, for example, the 3rd quintile which is just in the
consumption category above the bottom two poor quintiles (4th and 5th) has
households with a mere 500 birr difference in consumption expenditures
with the national poverty line. Thus, we introduced the shock to each
household category assuming transfer of equal percentage.

43
Daniel Abraham Mengistu: The Impact of Fiscal Policy on Poverty in Ethiopia:…

It is useful to know the transfer schemes of the government to introduce a

shock to the model but these values are difficult to get and are not consistent
with those in the model. Thus, we introduced an increase in these transfers to
look into their importance in reducing poverty if government gives more
focus on to these mechanisms. We assume a 20% DPI indexed increase in
trnsfr
government transfers, hgov
, to all households in the 2005/06 SAM.

Simulation 4: Sim 2 + Sim 3

This simulation is a combination of consumption expenditure (sim 2) and
government transfer (sim 3) policies of the government. In the EDRI (2009)
report, these two cover the largest part of government recurrent expenditure
with the remaining part allotted to government savings. Thus we examine the
combined impacts of increases in government consumption expenditure and
government transfers by 4% and 20%, respectively.

Simulation 5: Sim 1 + Sim 2 + Sim 3

This simulation is the one that helps us to evaluate the combined impact or
the net effect of the fiscal policy options employed together. Domestic
indirect tax is contributing one of the largest to government revenue, but
likewise, it may adversely contribute to price related impacts on the
consumer. On the expenditure side, we have stated that the government has
invested heavily on the poor sectors using domestic and foreign finance. As
the realistic economy usually pursues simultaneous fiscal measures, the
intent in this simulation is to investigate the combined (net) impacts on the
revenue side of an 80% increase in the sales tax rate that proxied the revenue
increase (sim 1) vis-a-vis 4% and 20% respective increases in government
consumption (sim 2) and transfer expenditures (sim 3).

Analysis of Results
In this section, we analyze the results of the simulations. We give central
focus to the poverty impact of the fiscal policies with three related effects:

44
 Ethiopian Journal of Economics Vol. XXII No 1, April 2013

macroeconomic, sectoral and welfare. The analysis is based on the changes

brought about by the five policy simulations.14

Effects on Macroeconomic Indicators

In Table 1, we report the simulation results of selected macro indicators. In
simulation 1, the macroeconomy has recorded negative changes in most
cases. An 80% increase in government indirect tax revenue brought about a
0.67% decrease in real GDP at factor cost. Absorption has shown a 0.58%
increase due mainly to the strong increases in investment (4.13%) that
offsets the negative changes in private consumption by 1.75%. This decrease
in private consumption could also be affirmed by the decreases in
consumption expenditure of households. Also government recurrent
expenditure has increased by 0.64% contributing to increases in absorption.
Due to the increase in taxes, government income has increased by a huge
margin of 14.03%. This explains most of the changes in investment as
government investment is larger. The CPI has risen by 0.67% indicating the
general increases in prices when government imposes taxes on consumers.

In simulation 2, the results are mostly opposite compared to the first

simulation. Real GDP has shown a 0.18% growth at factor cost. Absorption
has grown by 0.15%, like the first simulation, but in this case, the causes are
11.89% increase in government recurrent expenditure and 0.4% increase in
private consumption that offset the 3.26% negative changes in investment.
The major reason for the decrease in investment is the decline in government
income by 0.44%. CPI has changed negatively by 0.11% which explains
why private consumption has increased.

Simulation 3 results in an increase in real GDP by 0.05% at factor cost. Like

the second simulation, the increase in absorption by 0.03% could be
explained by the increases in government recurrent expenditure (2.58%) and
private consumption (0.19%) though investment has declined by 0.58% due
mainly to decreases in government income by 0.08%. The increase in private
14
This percentage changes calculate the changes from the base year, 2005/06.

45
Daniel Abraham Mengistu: The Impact of Fiscal Policy on Poverty in Ethiopia:…

consumption of households could directly be associated with increases in

purchasing power of the households as a result of the transfers. Hence, most
of the increment in the GDP is related to increases in domestic absorption. In
this case, CPI has increased by a small 0.07% which could be linked to the
direct impact of the transfers on purchasing power of households.

In simulation 4, real GDP has grown by 0.22% at factor cost. Absorption has
increased by 0.19% due mainly to increases in government expenditure and
private consumption which have grown by 14.48% and 0.59%, respectively
to offset the 3.88% decline in investment that is caused by decrease in
government income. Since the simulation principally is a shock to recurrent
expenditure in general, consumption expenditure has dominated the transfer
changes in explaining the 0.05% decrease in the CPI.

Table 1: Summary of results of macroeconomic effects of simulations (%

changes)
Base value
(billions
birr)

Macro Indicators Sim 1 Sim 2 Sim 3 Sim 4 Sim 5

Real GDP (factor cost) 122.22 -0.67 0.18 0.05 0.22 -0.43
Absorption 162.48 0.58 0.15 0.03 0.19 0.77
Investment 28.18 4.13 -3.26 -0.58 -3.88 0.6
Private Consumption 114.75 -1.75 0.4 0.19 0.59 -1.21
Government Income 17.45 14.03 -0.44 0.08 -0.37 13.75
Government Expenditure 12.09 0.64 11.89 2.58 14.48 15.19
Consumer Price Index 1.13 0.67 -0.11 0.07 -0.05 0.65
Source: Simulation results from the CGE model

In the last simulation, we find interesting results that resemble the results of
the first simulation. We find that the real GDP has decreased by 0.43% at
factor cost. Domestic absorption has increased by 0.77%. Parts of domestic
absorption, investment and government recurrent expenditure, have
increased by 0.6% and 15.19%, respectively, though private consumption

46
 Ethiopian Journal of Economics Vol. XXII No 1, April 2013

has declined by 1.21%. The small increase in investment, unlike the first
case, could be explained by the fact that government now channels most of
its increased income (13.75%) to recurrent expenditure. The CPI has now
increased by 0.65% which follows mainly from the increases in indirect
taxes.

Sectoral Effects

For reporting purposes, we classified activities into two; agriculture and non-
agriculture. Table 2 and 3 present the simulation results based on the mean
growths of the agricultural and non-agricultural activities. Table 2 presents
the results for sectoral growth of output.

In simulation 1, sectoral output has shown negative growth both in the

agricultural and non-agricultural sectors. Domestic output in the agricultural
sector has declined by 1.2% whereas non-agricultural output declined by
0.5%. The underlying reason for these changes could emanate from the price
related effects when indirect taxes are imposed. For instance, factor returns
have decreased in all cases and that the output purchasing power of
consumers decreases as the CPI increases. A peculiar change we can
appreciate in non-agriculture is the 3.6% output increase in construction
services which could be associated with the increases in fixed investment as
a result of increase in government income.

In simulation 2, sectoral output has increased by 0.05% and 0.003% for

agriculture and non-agriculture in that order. When government consumption
for goods and services increases, it could bring in producers who want to
gain profits from the increase in exogenous government demand for goods
and services. The very small change in level of output in non-agriculture
emanated from the large decline in construction services by 2.79%.

In simulation 3, output level in agriculture has increased by 0.14% whereas

output level in non-agriculture has declined by 0.04%. The transfers are

47
Daniel Abraham Mengistu: The Impact of Fiscal Policy on Poverty in Ethiopia:…

provided to both rural and urban households. Most of the rural households
are expected to engage in agriculture. This exogenous increase in household
income helps such households to purchase more (increasing their
consumption demand) and could allow them buy more raw material for
further production. But in the non-agricultural sector; production activities of
machinery, construction services and other-manufacturing have shown major
decreases that offset the increases in the remaining non-agricultural
activities. This is probably because government has shifted the expenditure
away from such industrial and service activities.

In simulation 4, similar changes have resulted whereby agricultural output

has increased by 0.19% and non-agricultural output has declined by 0.05%.
As this is a combined simulation to represent recurrent expenditures, it has a
demand and supply side effect on output production following from the
changes in government consumptions and transfers.

The last simulation has replicated the first simulation as we have seen in the
behaviors of macro indicators. Production in both sectors has shown
declines. Agricultural production has declined by a relatively stronger
margin of 1.1%. In the non-agricultural sector, similar results of 0.5%
decrease have resulted.

Table 2: Effects of simulations on sectoral output (% changes)

Sectors Base Sim 1 Sim 2 Sim 3 Sim 4 Sim5
Agriculture 12.9 -1.2 0.05 0.14 0.19 -1.1
Non - Agriculture 13.6 -0.51 0.003 -0.04 -0.05 -0.5
Source: Simulation results from the CGE model

The other way to approach the changes in the production activities is by

evaluating the changes in the demand and supply of factors (labor). Table 3
presents the results. The simulations have only produced results for the
employment (supply) of labor input combining the changes in the

48
 Ethiopian Journal of Economics Vol. XXII No 1, April 2013

agricultural and non-agricultural sectors. Labor employment has declined,

especially in the first and last simulations by 1.6% and 1.3%, respectively.
This might be due to decreases in returns to labor, as output production
shrinks, which decreases labor supply. But when we come to the other
simulations, labor employment has increased for the second (0.4%), third
(0.1%), and fourth (0.49%) simulations. As we discussed in table 6, the
increase in output for these simulations could be associated with the increase
in labor inputs for production as a result of increases in returns.

For the demand for factors, we can examine labor as the only flexible input
of production. In the first simulation, demand for labor has declined for both
agriculture and non-agriculture by 1.6% and 1.09%, respectively. The
decline in labor demand is mainly related to the contraction in output
production as a result of the tax increases.

In the second simulation, the demand for labor has increased by 0.05% in the
agricultural sector. However, demand for non-agricultural labor has declined
by 0.32%. We could associate these declines to the negligible change in
output production in the aggregates for agriculture and non-agriculture. In
the third simulation, the increases in transfers have brought about increased
demand for labor by 0.18%. In contrast, the demand for labor has decreased
by 0.15% in the non-agricultural sector. The intuition is related to the use of
the cash transfer in rural areas. The benefited households (most of which are
farmers) could aspire to produce more which requires inputs. In agriculture,
labor is a principal input of production which validates the increase. In non-
agriculture, output production has declined on average which is related with
shift in use of government resources. This could lead to reduction in surplus
inputs. In the fourth simulation, similar patterns of change have been
observed. Demand for agricultural labor has increased by 0.22%, while non-
agricultural labor has decreased by 0.46%. This decline could also follow
from the decreases in production of output in the non-agricultural sector.

49
Daniel Abraham Mengistu: The Impact of Fiscal Policy on Poverty in Ethiopia:…

In the fifth simulation, we also find similar results like the first simulation.
Demand for agricultural labor has decreased by 1.34%. Like the case of
agricultural labor, demand for non-agricultural labor has declined by 1.6% in
this last simulation. The increase in tax seems to have strained the use of
labor in both agriculture and non-agriculture.

Table 3: Effects of simulations on labor employment and demand by

sector (% changes)
Sectoral Indicators Base Sim 1 Sim 2 Sim 3 Sim 4 Sim5
Changes in labor employment (both agriculture and non-agriculture)
Labor employment 60.29 -1.33 0.4 0.1 0.49 -0.83
Changes in demand for labor by activity
Agricultural labor 8.86 -1.6 0.05 0.18 0.22 -1.34
Non-agricultural labor 1.78 -1.09 -0.32 -0.15 -0.46 -1.6
Source: Simulation results from the CGE model

Welfare Effects

The most important welfare indicator used in the literature for CGE models
is the equivalent variation (EV). Since policy shocks are usually followed by
major price adjustments, the EV measures the level of income (in money
terms) that the consumer needs to (presumably) pay before the shock to
leave him as well off at the equivalent level of utility loss after the price
increase. Since the consumer is harmed prior to the policy change by paying
the price equivalent in income, negative EV changes represent welfare
(utility) loss as a result of the policy shock. Figure 3 presents the welfare
effects of the policy shocks.

In the first simulation, the instant increases in domestic indirect tax collected
by government seems to have negatively affected both the urban and rural
households as the negative values for EVs would suggest. The EVs have
declined by 0.4%, 1.2%, 0.1% and 0.4% for rural-poor, rural- non-poor,
urban-poor and urban-non-poor, respectively. Comparatively, the EVs

50
 Ethiopian Journal of Economics Vol. XXII No 1, April 2013

indicate that the non-poor receive much of the welfare strain in both rural
and urban areas. This may be indicative of the fact that tax impositions have
adversely affected these well off household groups, dragging their
livelihoods downwards. Probably, the major culprit behind this is the price
increasing effect of taxes which is mainly evident in urban areas in relation
to VAT imposition. Comparing urban and rural households, however, we
find that rural households face larger welfare loses.

In the second simulation, we see positive EVs for the increases in

government consumption indicating welfare improvements. The EVs
showed increases for all household groups by 0.1% (rural-poor), 0.3% (rural-
non-poor), 0.01% (urban-poor) and 0.1% (urban-non-poor). The welfare of
rural households improved larger than that of the urban counterparts with
non-poor households reaping the relative advantage in both areas. This could
mainly be because, in the SAM, recurrent government expenditure is spent
on public administration, education, and health most of which constitute
wage and non-wage payments that the non-poor are characterized with.

Figure 3: Effects of simulations on welfare (EV) of household groups

(% changes)
1

0.5

0
% changes

Sim 1 Sim 2 Sim 3 Sim 4 Sim 5

-0.5

-1

-1.5

-2

-2.5
Policy simulations
Rural Poor Rural non-poor Urban Poor Urban non-poor Total
Source: Simulation results from the CGE model

51
Daniel Abraham Mengistu: The Impact of Fiscal Policy on Poverty in Ethiopia:…

In the third simulation, the results show that welfare increased by 0.1%,
0.03%, 0.03% and 0.1% for rural-poor, rural-non-poor, urban-poor, and
urban-non-poor, respectively. The outcomes seem to favor the rural-poor and
urban-non-poor. This seems to suggest that the rural-poor will need to be
targeted in such programs whereas the urban-poor may not get the benefits
expected from such transfer programs. Since welfare has improved for all
households, though by small amounts, transfers could be one of the policy
instruments of government to improve welfare.

In the fourth simulation, we have combined sim 2 and sim 3 to have an

overall impact of government recurrent expenditure on welfare. The results
are improvements in welfare by 0.1%, 0.3%, 0.04% and 0.2% for rural-poor,
rural-non-poor, urban-poor and urban-non-poor, respectively. The
improvement in welfare is found to be larger for non-poor households in
urban and rural areas. Since this simulation is a combination, the influence of
government consumption looks larger as the non-poor seem more benefited
from the policy shock. The welfare effects on the poor, however, are positive
but small.

The last simulation shows that the effect of combined policy shocks are
almost similar with the first simulation. The rural-poor, rural-non-poor,
urban-poor and urban-non-poor have all recorded negative welfare changes
by 0.3%, 0.9%, 0.03% and 0.2%, respectively. Household welfare seems to
be strained for both urban and rural households but the non-poor seem to
receive the bigger blow. These results suggest that heavy tax collection
schemes of the government have brought about a net negative impact on
welfare of all the household groups. Interestingly, the urban-poor are found
to have the lowest decrease in welfare. The CGE results depicted in figure 3
also show total welfare changes for households in each simulation. We see
that the welfare loss in the last simulation is lower than the first simulation
due to the offsetting effects of the spending schemes.

52
 Ethiopian Journal of Economics Vol. XXII No 1, April 2013

Impacts on Poverty

To analyze poverty, we used the percentage changes in consumption

expenditure of household groups that are taken from the CGE model. It is
these economy-wide changes in consumption that we used as the source of
link with the MS model in conducting the poverty analysis using the three
poverty indicators. Figure 4 presents the CGE results for changes in
consumption.

In the first simulation, consumption shrank for all the household groups by
1.3%, 1.2%, 1% and 0.8% for rural-poor, rural-non-poor, urban-poor and
urban-non-poor, respectively. The likely explanation is that the increases in
price of consumption commodities forced households to adjust their use of
income. For instance, when VAT is imposed on commodities, it is imposed
on the price paid by the consumer which increases the “menu price”. Hence,
consumption expenditure has to fall assuming fixed incomes.

For the three spending simulations, consumption expenditure increases. In

the second simulation, consumption has increased by 0.3% for rural
households whereas it has increased by 0.2% for urban households. The
reason could be the effect of increases in wage and non-wage payments that
give households more income to expend. In the third simulation, the rural-
poor and non-poor experienced consumption expenditure increases by 0.3%
and 0.1%, respectively. But the urban- poor and non-poor experienced
higher increases by 0.8% and 0.5%, respectively. As can be seen from the
figure, both rural and urban-poor have larger increases in their income
compared to the non-poor mainly because the transfer covers a larger
proportion of their total income. In the fourth simulation, the consumption
expenditure increased for all household groups by a larger amount compared
to the previous two individual simulations of government spending schemes.
Consumption increased in all cases for rural-poor (0.6%), rural-non-poor
(0.5%), urban-poor (1%), and urban-non-poor (0.7%). The explanations we

53
Daniel Abraham Mengistu: The Impact of Fiscal Policy on Poverty in Ethiopia:…

give to the previous two simulations apply in the combination when

recurrent expenditure increases.

In the last simulation, consumption expenditures of rural households have

decreased by 0.7% (poor) and 0.8% (non-poor). Likewise, the urban-non-poor
have recorded negative consumption growths by 0.2%. In contrary, urban poor
seem to have a slight increase in their consumptions (0.1%) which emerge as
an interesting net impact of the fiscal policy combinations. In the welfare
results from the CGE model, we also noticed that welfare losses are smallest
for the urban-poor in this fifth simulation. The possible explanation could be
that urban poor in the survey are less affected by the tax increases or that the
government consumption and transfer expenditures have a stronger positive
impact as can be seen from the results in the third and fourth simulations. The
government transfers seem to have largely benefited the urban-poor. A study
by Obi, 2007 for Nigeria found out different results in which transfers and
subsidies to firms were found to be more productive than transfers to
households.

To investigate the impacts of this consumption changes on poverty, the FGT

poverty estimates are generated for households at national, rural and urban
levels. Among the 21,594 households at national level; we found that 9,493
were rural and 12,101 were urban.15 Since the SAM has delineated poor and
non-poor households based on levels of consumption expenditure with the
bottom 40% as poor, we needed to separate the poor and non-poor in rural
and urban areas in the same manner. Rural and urban households were each
divided into two categories taking the bottom 40% as poor and the top 60%
as non-poor. As shown in figure 5, since consumption expenditures
separately change for these four household categories, we made necessary
adjustments on the consumption expenditures in the 2004/05 HICE survey
using the CGE results for each policy simulation.

15
Though the number of rural households is smaller in the survey, one point of note
is that the number of people each sample rural household represents (weights) is
very large compared to urban households.

54
 Ethiopian Journal of Economics Vol. XXII No 1, April 2013

Figure 4: Effects of simulations on consumption expenditure growth of

household groups
1.5

0.5
% changes

0
Sim 1 Sim 2 Sim 3 Sim 4 Sim 5
-0.5

-1

-1.5
Policy simulations
Rural Poor Rural non-poor Urban Poor Urban non-poor
Source: Simulation results from the CGE model

To elaborate more, based on additional information on adult equivalent

household size for the 2004/05 survey from CSA, we changed the
consumption expenditures in the survey to consumption expenditure per
adult equivalent. After this, we took approximate levels of consumption
expenditure at the demarcation of the bottom 40% (two quintiles) to
represent cut points for national, rural and urban-poor and non-poor
households. So among rural households, we got 3,861 as poor and 5,632 as
non-poor whereas among urban households, we got 4,751 as poor and 7,350
as non-poor. We then introduced the consumption changes on the base
values of the four households after simulations from the CGE model
(Savard, 2003). The new values we calculated are the ones we used in DAD
to compute the FGT indices.

An important point here is the disparity that may be created in the poverty
results when we compare it with the official levels of the poverty measures
gathered from the 2004/05 HICE of CSA. The main reason is that the CSA
used 1075.03 poverty line that is CPI indexed to compute the FGT indices.
But in this study, we are guided by the mechanism followed in the 2005/06

55
Daniel Abraham Mengistu: The Impact of Fiscal Policy on Poverty in Ethiopia:…

SAM that is to arrange the households based on expenditure in descending

order and take the bottom 40% as poor. The consumption expenditure level
at the demarcation of total households in the survey, 1782.98, is taken as the
poverty line. This national poverty line is adjusted for the CPI in each
simulation based on CPI values generated from the CGE model. After this
basic ground work, we came up with the results for the poverty measures.

Table 4 presents the results for the head count index. At the base run, the
proportion of the poor from total population is 40.9% at the national level.
But at rural and urban levels, the index becomes 42.8% and 30.8%. In the
first simulation, we see that poverty incidence has increased by 1.5%, 1.4%
and 1% at national, rural and urban levels implying worsening of the poverty
situation in the short-run. Since this simulation is related to tax imposition,
the indication is that when government squeezes money out of the pockets of
households for various purposes, it has a short-run adverse impact on their
well being. Similar results were found by Wong et al., 2008 for the case of
Ecuador.

In the second, third and fourth simulations, the head count poverty has
shown slight decreases. In the second simulation, the national, rural and
urban head counts declined by 3.2%, 3.3% and 1.6%. The explanation could
be that government consumption has increased on sectors that benefit the
poor. If consumption of government increases on public administration,
education and health, then this may result in short-run decreases in head
count poverty. In the third simulation, the results are similar with the second
that are decreases in poverty incidence by 2.5% (national), 2.6% (rural) and
1.9% (urban). The likely implication is that government transfer schemes to
households can be used as a policy in reducing poverty in the short-run.
Also, in the fourth simulation, the proportion of poor has shown decreases by
3.4% (national), 3.5% (rural) and 2.3% (urban). These figures are relatively
larger compared to the separate spending simulations. Most of these
decreases in the poverty head count are related to the increases in income
and real consumption that come with the government recurrent expenditures.

56
 Ethiopian Journal of Economics Vol. XXII No 1, April 2013

In the last simulation, we find that at national and rural levels, the poverty
incidence increased by 0.5% and 0.7%, respectively. But for urban
households, poverty incidence decreased by a slight 0.3%. From these results,
we see the dominant straining effects from the first simulation which has
increased the poverty head count only for total households and rural
households. In case of urban households, a peculiar decrease in head count
poverty is found. The explanation for this could be the increase in
consumption expenditure of urban-poor that we have examined in figure 4.
When we compare these values with the results from the first simulation, we
find that they are moderate indicating that the welfare loss from the increases
in tax collection could be overturned by government spending schemes that
bring benefits to households.

Table 4: Effects of simulations on poverty head count index ( P0 ) (%

changes)
Base Sim 1 Sim 2 Sim 3 Sim 4 Sim 5
National 0.409 1.5 -3.2 -2.5 -3.4 0.5
Rural 0.428 1.4 -3.3 -2.6 -3.5 0.7
Urban 0.308 1 -1.6 -1.9 -2.3 -0.3
Source: Microsimulation results

Table 5 presents the results for the poverty gap index. The results imply
similar changes like the head count index in the sense that the index
increases in the first simulation for all the three categories from the base.
Poverty depth has increased by around 2.5%, 1.7% and 1% at national, rural
and urban levels, respectively. Rural poverty gap has increased by a
relatively larger margin compared to urban poverty gap. The implication is
that when the increases in domestic indirect taxes press the consumption
power of households downwards, the mean level of consumable goods that
the households need to get out of poverty increases worsening their poverty
state or pushing them down to chronic poverty.

In the second, third and fourth simulations; the poverty gap has declined by
relatively higher percentages. In the second simulation, poverty gap has

57
Daniel Abraham Mengistu: The Impact of Fiscal Policy on Poverty in Ethiopia:…

declined by 3.5%, 4.3% and 3.2% for national, rural and urban households,
respectively. The indication is that increase in government consumption
reduces the income shortfall of poor households from the poverty line. In the
third simulation, poverty gap has declined by 3.5%, 3.4% and 4.2% for
national, rural and urban households, respectively. Compared to the effect of
changes in government consumption, changes in transfers to households
seem to have larger impact on urban households compared to rural
households. In the fourth simulation, a stronger decrease in poverty depth
has resulted for all the households. At national, rural and urban levels, the
poverty depth declined by 4.4%, 5.1% and 5.3%. The evident explanation
could be that the poverty gap decreases are a cumulative effect of the second
and third simulations.

In the last simulation, we observe that the poverty gap index has shown
negligible changes. Poverty gap has slightly increased by a mere 1% at
national and urban levels. However, we find that poverty gap has not
changed for rural households. The possible explanation is that the net impact
of the fiscal policies employed together did offset each other for the poverty
depth of rural households leaving the poor households unaffected.

Table 5: Effects of simulations on poverty gap index ( P1 ) (% changes)

Base Sim 1 Sim 2 Sim 3 Sim 4 Sim 5
National 0.113 2.5 -3.5 -3.5 -4.4 1
Rural 0.117 1.7 -4.3 -3.4 -5.1 0
Urban 0.095 1 -3.2 -4.2 -5.3 1
Source: Microsimulation results

Table 6 presents the results for the poverty severity (squared poverty gap)
index. In the first simulation, we see a 4.8%, 2.3% and 2.7% increase in
poverty severity at national, rural and urban levels, respectively. The
implication of this is that the inequality among the poor has risen due to the
imposition of the domestic indirect taxes. Poverty severity has increased by
higher margins for urban households compared to rural households.

58
 Ethiopian Journal of Economics Vol. XXII No 1, April 2013

In the second, third and fourth simulations; poverty severity has declined. In
the second simulation, poverty severity has been reduced by 4%, 6.8% and
5.4% at national, rural and urban levels, respectively. We can see from the
result that increases in government consumption expenditure reduce the
inequality among the poor. This is more pronounced for rural households
compared to urban households. Likewise, for the third simulation, poverty
severity has declined in all cases by 3.8%, 4.5% and 5.4% for national, rural
and urban households. In this simulation, poverty severity in urban areas has
declined by larger amounts compared to rural areas. Similarly, in the fourth
simulation, we see stronger declines in the inequality among the poor.
Poverty severity has declined by 5.2%, 6.8% and 8.1% at national, rural and
urban levels, respectively. What we can infer from the results of the three
FGT indices is that the combined effects of the government expenditure
measures have stronger impacts on reducing poverty compared to the
separate policy options.

Table 6: Effects of simulations on poverty severity index ( P2 )

(% changes)
Base Sim 1 Sim 2 Sim 3 Sim 4 Sim 5
National 0.042 4.8 -4 -3.8 -5.2 2.4
Rural 0.044 2.3 -6.8 -4.5 -6.8 0
Urban 0.037 2.7 -5.4 -5.4 -8.1 -2.7
Source: Microsimulation results

In the fifth simulation, we find mixed results. At the national level, poverty
severity has shown a 2.4% increase. But it has shown no change for rural
households which follows from similar results for rural poverty gap.
However, urban poverty severity has declined by 2.7%. This result seems to
follow from the results that we reported for changes in welfare and
consumption expenditure for this simulation.

59
Daniel Abraham Mengistu: The Impact of Fiscal Policy on Poverty in Ethiopia:…

6. Conclusions and Implications

6.1 Conclusions

In this study, we set out to investigate the economy-wide impacts of fiscal

policy on poverty. We looked into the impacts of government tax measures
and expenditure schemes. The crux of the matter is that the Ethiopian
government is intensifying domestic revenue collection usable for various
spending plans.

The results of increasing domestic indirect taxes indicate negative changes in

the macro economy represented by the declines in real GDP. The sectoral
effects also show reductions in output production and labor employment.
The results generally indicate welfare loss, decline in household
consumption expenditures and worsening of poverty in the short-run when
domestic indirect tax increases are imposed on commodities.

When we come to the short-run expenditure policies, we see opposite results.

In these cases, real GDP has grown positively though its various components
exhibited various changes across the simulations. The sectors showed
increments in labor employment and output production. The expenditure
schemes also revealed welfare gains to households. Moreover, household
consumption expenditure has increased in all cases. Though these increases
were different in magnitude, the impacts of these changes were also visible
on poverty. The poverty measures revealed improvements in the poverty
status of all households in the short-run.

In the last simulation, we found out that the combined tax and expenditure
interventions lead to declines in real GDP. Sectoral output and labor
employment were also reduced as a result of the policy combinations. The
results show net welfare loss and worsening of the poverty in the case of
majority of households. In this simulation, consumption expenditure has
decreased for all households except the urban-poor. Due to this, the poverty
state of urban households has revealed improvements though small in

60
 Ethiopian Journal of Economics Vol. XXII No 1, April 2013

magnitude. Overall, this scenario resulted in many indicators that had similar
changes like the first simulation implying dominant adverse effects of
changes in domestic indirect tax policies over the changes in government
consumption and transfer expenditure policies.

6.2 Implications

This study has some useful implications for policy and future research in
relation to the link between fiscal policy and poverty in Ethiopia. Firstly, the
Ethiopian government has been expanding the tax base through improved tax
collection principally from domestic sources. This trend seems to even be
widened further as we can see from the Growth and Transformation plan
(GTP) for the period 2010/11 to 2014/15. The results in this study, however,
indicate that government policy towards domestic sources has repercussions
on poverty in the specific case of domestic indirect taxes, with all other
anticipated changes retained at the base level.

Secondly, the government consumption spending schemes are poverty

reducing. The study results revealed that the government policy of increasing
consumption expenditure does not worsen the poverty situation in the short-
run though inflation in the economy was not examined.

Thirdly, government transfers to households are poverty reducing. The

results from the study indicate that increasing transfers to households
improves the poverty status of both poor and non-poor households. Hence,
the government can use these policies as an alternative in addressing the
poverty problem in the short-run. However, this argument shouldn’t be taken
at face value. Practically, transfer schemes require preliminary justifications.
For example, given the dynamic nature of poverty, transfers can take various
forms based on the type of poverty state.

Fourthly, in relation to the fiscal policy combinations, we found out that the
tax policies have a dominant short-run negative impact on poverty. This

61
Daniel Abraham Mengistu: The Impact of Fiscal Policy on Poverty in Ethiopia:…

implies that in the financing plans that government formulates that use
domestic indirect taxes, households could be negatively affected. So, to
protect households from such unintended strains of the fiscal plans, the
government has to also prepare short-run spending policies like safety nets
schemes. As stated, the spending policies we examined are poverty reducing,
hence policy makers need to exploit such policies that would improve the
status of households while the financing policies go along.

Finally, we could raise two agenda worth investigating in future research.

First, since we only focused on ex-post analysis based on the 2005/06 data, it
is difficult to extrapolate to the future. Further studies in this area could
extend this study by using ex-ante analysis using recent SAM to predict the
poverty impacts of financing plans like the GTP. Second, it would be
difficult to have a full picture of the net impact of the fiscal policy changes
without in depth analysis on long term capital expenditure schemes. Poverty
reduction is a principal long term development objective and the Ethiopian
government has designed a number of poverty reduction strategies in which
pro-poor expenditure policies are at the center. Since the study is a static
analysis, it remains for future research to examine the impacts of government
financing and spending policies in the long-run. Future researchers will
however be faced with the implicit nature of government capital investment
in the Ethiopian SAM and the IFPRI CGE model. In this regard, a
suggestion is that the Ethiopian SAM should be prepared in a manner that
could allow a separate analysis of private and government investment. This
will be helpful in investigating the long – run effects of government and/or
private investments using dynamic CGE models. Dynamic CGE analysis that
examines the anticipated returns from government capital investment could
add a lot to the literature in the country.

62
 Ethiopian Journal of Economics Vol. XXII No 1, April 2013

References

Abebe, S. (2005). “Essays on Poverty, Risk and Consumption Dynamics in

Ethiopia”, Economic Studies, Department of Economics, School of
Economics and Commercial Law, Goteborg University (155).
AFRODAD (African Forum and Network on Debt and Development). (2005).
“The Impact of the Poverty Reduction and Growth Facility on Social
Services in Ethiopia”, Study Report.
Agenor, P. R. (2004). “Macroeconomic Adjustment and the Poor: Analytical
Issues and Cross–Country Evidence”, Journal of Economic Surveys, 18
(3): 351 – 408.
Ahmed, V., and O’Donoghue, C. (2004). “Using CGE and Micro-Simulation
Models for Income Distribution Analysis”, A Survey: Preliminary
Version.
Alemayehu, G. and Alem, A. (2006). “Macro Policy Reform, Labor Market,
Poverty and Inequality in Urban Ethiopia: A Micro Simulation
Approach”, final report of phase II AERC project on “Poverty, Income
Distribution and Labor Market Issues in Ethiopia”.
Aziz, I. (2008). “Macroeconomic Policy and Poverty”, Discussion Paper No.
111, Asian Development Bank Institute.
Bibi, S., Fofana, I., Chatti, R., Corong, E., and Bouazouni, O. (2010). “Gender
and Employment Impacts of Taxation Policy in the MENA Region: A
Comparative Analysis of Algeria, Egypt, Morocco, and Tunisia”, Paper
Presented at the International Conference on “Women and Youth in Arab
Development”, (22-24 March), Cairo, Egypt.
Bigsten, A. and Abebe, S. (2007). “Poverty Transition and Persistence in
Ethiopia: 1994 – 2004”, School of Business, Economics and Law,
Goteborg University.
Colombo, G. (2008). “Linking CGE and Microsimulation Models: A
Comparison of Different Approaches”, Discussion Paper No. 08-054,
Centre for European Economic Research.
CSA (Central Statistical Agency). (2008). “Household Income, Consumption
and Expenditure Survey 2004: HICE 2004/2005 Metadata”, CSA, Addis
Ababa, Ethiopia.

63
Daniel Abraham Mengistu: The Impact of Fiscal Policy on Poverty in Ethiopia:…

Cury, S., Pedrozo, E., Coelho, A. M. and Callegari, I. (2010). “The Impacts of
Income Transfer Programs on Income Distribution and Poverty in Brazil:
An Integrated Micro Simulation and Computable General Equilibrium
Analysis”, PEP Network, MPIA Working Paper, 2010 –20.
Damuri, Y. R., and Perdana, A. A. (2003). “The Impact of Fiscal Policy on
Income Distribution and Poverty: A Computable General Equilibrium
Approach for Indonesia”, CSIS Working Paper Series, No. 068.
Decaluwe, B., Patry, A., Savard, L., and Thorbecke, E. (1999). “Poverty
Analysis within a General Equilibrium Framework”, CREFA Working
Paper 99-06, Universite Laval.
Dejene, A., Belay, F. and Sindu, W. (2007). “Trade Liberalization, Poverty and
Inequality in Ethiopia: A CGE Microsimulation Analysis”, Paper
presented at the 6th PEP Research Network General Meeting, (14-16
June), Lima, Peru.
Duclos, J. Y., Araar, A. and Fortin, C. (2010). DAD: A Software for Distributive
Analysis (Analyse Distributive), MIMAP Programme, International
Development Research Centre, Government of Canada and CRFA,
Universite Laval.
EDRI (Ethiopian Development Research Institute). (2009). “Ethiopia: Input
Output Table and Social Accounting Matrix”, Ethiopian Development
Research Institute (in collaboration with the Institute of Development
Studies, University of Sussex).
EEA (Ethiopian Economic Association). (2009). EEA: Statistical Data Base,
Ethiopian Economic Policy Research Institute/Ethiopian Economic
Association.
Foster, J., Greer, J. and Thorbecke, E. (1984). “A Class of Decomposable
Poverty Measures”, Econometric Society Publications, Econometrica, 52
(3): 761 – 766.
Gemmell, N. and Morrissey, O. (2002). “The Poverty Impacts of Revenue
Systems in Developing Countries”, Final Report to the Department for
International Development, University of Nottingham.
Gunter, B. G., Cohen, M. J., and Lofgren H. (2005). “Analysing Macro-Poverty
Linkages: An Overview”, Development Policy Review, 23 (3): 243 –
265.

64
 Ethiopian Journal of Economics Vol. XXII No 1, April 2013

Heintz, J., Oya, C. and Zepeda, E. (2008). “Towards an Employment-Centered

Development Strategy for Poverty Reduction in the Gambia:
Macroeconomic and Labor Market Aspects”, UNDP, International
Poverty Centre, Country Study No. 16.
IMF (International Monetary Fund). (2001). “Panel on Macroeconomic Policy
and Poverty Reduction”,
http://www.imf.org/external/np/res/seminars/2001/poverty/041301.htm
(accessed 30 February 2011).
Iqbal, Z. and Siddiqui, R. (2001). “Critical Review of Literature on Computable
General Equilibrium Models”, Pakistan Institute of Development
Economics. MIMAP Technical Paper Series No.9, Islamabad, Pakistan.
Islam, R. (2001). “Poverty Alleviation, Employment, and the Labor Market:
Lessons from the Asian Experience”, Paper presented at the “Asian and
Pacific Forum on Poverty: Reforming Policies and Institutions for
Poverty Reduction”, (5-9 February), Asian Development Bank,
Philippines.
Johannes, T. A., Akwi, T., and Anzah, P. E. (2006). “The Distributive Impact of
Fiscal Policy in Cameroon: Tax and Benefit Incidence”, PEP Network,
PMMA Working Paper, 2006-16.
Kasahun, T. (2003). “Macroeconomic Policy and Poverty Reduction in Ethiopia:
A General Equilibrium Approach”, Master’s Thesis, School of Graduate
Studies, Addis Ababa University, Addis Ababa, Ethiopia.
Kiringai, J., Wanjala, B., and Mathenge, N. (2006). “Feminisation of Poverty in
Kenya: Is Fiscal Policy the Panacea or Achilles’ heel?”, MPIA Network
Session Paper presented during the 5th PEP Research Network General
Meeting, (June 18-22), Addis Ababa, Ethiopia.
Lipton, M. and Ravallion, M. (1993). “Poverty and Policy”, World Bank, Policy
Research, Working Paper Series 1130.
Lofgren, H., Harris, R. L., Robinson, S., with El-Said, M. and Thomas, M.
(2002). “A Standard Computable General Equilibrium (CGE) Model in
GAMS”, Microcomputers in Policy Research: 5, International Food
Policy Research Institute; Washington DC, USA.
Mallick, S. (2009). “Macroeconomic Policy and Poverty Reduction in India”,
Indian Institute of Management Bangalore, IGIDR Proceedings/ Project
Report Series, 62 (15).

65
Daniel Abraham Mengistu: The Impact of Fiscal Policy on Poverty in Ethiopia:…

MoFED (Ministry of Finance and Economic Development) (2008). “Dynamics

of Growth and Poverty in Ethiopia (1995/96 – 2004/05)”, DPRD/
MoFED; Addis Ababa, Ethiopia.
_______. (2010). “Growth and Transformation Plan (2010/11 – 2014/15): Main
Text”, Vol. 1 (November 2010), Addis Ababa, Ethiopia.
Mulat, D., Fantu, G. and Tadele, F. (2003). “Growth, Employment, Poverty, and
Policies in Ethiopia: An Empirical Investigation”, Issues in Employment
and Poverty, Discussion Paper 12, Recovery and Reconstruction
Department; ILO, Geneva, Switzerland.
Munoz, S. and Cho, S. (2003). “Social Impact of a Tax Reform: The Case of
Ethiopia”. IMF Working Paper 03/232, Fiscal Affairs Department, IMF.
Obi, B. (2007). “Fiscal Policy and Poverty Alleviation: Some Policy Options for
Nigeria”, AERC Research Paper 164, African Economic Research
Consortium, Nairobi, Kenya.
RTI (Research Triangle Institute). (2008). “EMPAX-CGE: Model
Documentation”, Interim Report Prepared for US Environmental
Protection Agency, RTI Project Number 0209897.002.041.
Savard, L. (2003). “Poverty and Income Distribution in a CGE - Household
Micro Simulation Model. Top-Down/ Bottom UP Approach”,
International Development Research Centre; CIRPEE, Working Paper 03-
43.
Sinha, A. and Latigo, A. (2003). “Pro-poor Growth Strategies in Africa. A
Proposed Gender - Aware Macroeconomic Model to Evaluate Impacts of
Policies on Poverty Reduction”, Economic Policy Research Center,
Economic Commission for Africa, Kampala, Uganda.
Tadele, F. A. (2010). “Economic Growth, Policy Reforms, Household
Livelihoods and Environmental Degradation in Rural Ethiopia: Towards
an Integrated Model of Economic Transformation”, Dissertation for the
degree of doctor in Applied Economics, Department of Economics;
University of Antwerp.
Thurlow, J. (2004). “A Dynamic Computable General Equilibrium (CGE)
Model for South Africa: Extending the Static IFPRI Model”, TIPS,
Working Paper 1 – 2004.

66
 Ethiopian Journal of Economics Vol. XXII No 1, April 2013

Thurlow, J. and Seventer, D. E. (2002). “A Standard Computable General

Equilibrium Model for South Africa”, International Food Policy Research
Institute, TMD Discussion Paper No. 100.
United Nations Development Programme (UNDP). (2010). “The Real Wealth of
Nations: Pathways to Human Development”, Human Development
Report, UNDP, 20th Anniversary Edition.
Wang Li, Li X., Wenbo W., and Guangbao Z. (2010). “Fiscal Policy, Regional
Disparity and Poverty in China: A General Equilibrium Approach”, PEP
Research Network, MPIA Working Paper, 2010 – 11.
Wobst, P. (2001). “Structural Adjustment and Inter-sectoral Shifts in Tanzania:
A Computable General Equilibrium Analysis”, International Food Policy
Research Institute, Research Report 117, Washington DC.
Wong, S., Arguello, R., and Rivera, K. (2008). “Fiscal Policies and Increased
Trade Openness: Poverty Impacts in Ecuador”, A PEP Project for
presentation at the 2nd Regional Meeting on ‘Computable General
Equilibrium (CGE) Modeling: Contributions to Economic Policy in Latin
America and the Caribbean’
Yesuf, M. A. (2007). “Vulnerability and Poverty Dynamics in Rural Ethiopia”,
MSc. Thesis, Department of Economics, University of Oslo.

67
Daniel Abraham Mengistu: The Impact of Fiscal Policy on Poverty in Ethiopia:…

Annexes
Annex 1: Current and Capital Expenditure as % of Total Expenditure
in 1999/00 constant prices

Ethiopian fiscal year

1999/00

2001/02

2003/04

2005/06

2007/08

2009/10
ending July 7

Current Expenditure 80 63.2 58.5 52.3 48.4 44.7

Economic services 4.7 6.3 6.8 6.8 6.7 5.6
Social services 12.2 16.8 16.3 17 18.5 17.4
General services 48.4 27.8 25.3 22.2 19.5 17.9
Other 14.6 12.4 9.6 6.3 3.7 4.7
Capital Expenditure 20 36.7 41.5 47.7 51.6 55.3
Economic development 12 20 24.8 35.5 37.9 38.3
Social development 3.6 6 10.4 10.5 11.6 13.8
General development 2 4.2 6.4 1.8 2.1 3.2
External assistance 2.4 6.7 5.3 7.5 8.6 7
Total Expenditure (,000 birr) 17,181 18,535 18,794 22,630 23,383 28,462
Source: Own computations based on MoFED data retrieved from EEA (2009) data base

Annex 2: Tax and Non-tax Revenue as % of Total Revenue & Grants in

constant 1999/00 prices

Ethiopian fiscal year

1999/00

2001/02

2003/04

2005/06

2007/08

2009/10
ending July 7

Tax Revenue 54.6 61.4 61.2 60.8 59.9 65.4

Direct Taxes 19.9 34.2 18.2 19.1 17.7 22.5
Indirect taxes 34.7 37.3 43 41.8 42.3 42.9
Domestic Indirect Taxes 12.1 11.6 12.3 13.4 12.8 16.2
Foreign Indirect Taxes 22.5 25.6 30.7 28.4 29.5 26.7
Non Tax Revenue 30.9 19.8 15.5 23.1 15.1 15.9
Total Revenue(exc. grants) 85.5 81.2 76.7 83.9 75.1 81.3
External Grants 14.5 18.8 23.3 16.1 25 18.7
Total Revenue & Grants (,000 11,872 14,337 16,214 17,865 19,853 26,495
birr)
Source: Own computations based on MoFED data retrieved from EEA (2009) data
base.

68
 Ethiopian Journal of Economics Vol. XXII No 1, April 2013

Annex 3: Share of public expenditure on poverty-oriented sectors at constant

1999/00 prices
2005/06 2007/08 2009/10

Sectors
Current

Current

Current
Capital

Capital

Capital
Total

Total

Total
Agriculture 10.2 23.9 16.7 10.3 16.1 13.3 6.9 12.1 9.8
Education 25.4 17.7 21.7 29.2 14.1 21.4 30.3 18.4 23.7
Health 5.3 3.7 4.6 6.6 7.7 7.3 6.6 6.2 6.4
Social Welfare 0.9 0.4 0.7 0.6 0.04 0.3 0.4 0.03 0.2
Roads 1.3 24.5 12.4 1.2 33.6 17.9 1.3 36.5 20.8
Total poverty
43.1 70.2 56.1 47.9 71.7 60.2 45.6 73.2 60.8
targeted spending
Source: Own computations based on MoFED data retrieved from EEA (2009) data
base

Annex 4: Indirect tax revenue by component at constant 1999/00 prices (% of

GDP)

Ethiopian fiscal year

2005/06 2006/07 2007/08 2008/09 2009/10
ending July 7
Domestic indirect taxes 2.6 2.6 2.1 2.2 2.8
VAT and TOT on local goods 1.4 1.4 1.1 1.0 1.2
VAT and TOT on services 0.5 0.6 0.6 0.8 1.2
Excise tax 0.5 0.4 0.3 0.3 0.3
Stamp duty 0.2 0.2 0.1 0.1 0.1
Foreign trade taxes on imports 5.0 4.6 4.7 3.5 4.6
Customs duty 2.3 2 1.6 1.2 1.5
Import VAT 2.3 2.1 1.6 1.2 1.7
Excise tax 0.5 0.5 0.4 0.4 0.5
Sur tax -- -- 1.1 0.7 0.9
Total Indirect tax revenue 7.6 7.1 6.8 5.6 7.5
Source: Own aggregations and computations based on regional and federal data
from MoFED.

69
Daniel Abraham Mengistu: The Impact of Fiscal Policy on Poverty in Ethiopia:…

70