C h a p t e r  5

Ethiopia
Xinshen Diao, Alemayehu Seyoum Taffesse, Paul Dorosh, 

James Thurlow, Alejandro Nin Pratt, and Bingxin Yu

Agriculture is perhaps the most important economic sector for sustainable 
development and poverty reduction in Ethiopia. More than 85 percent of 
the population live in rural areas and depend crucially on agriculture for 

their livelihoods. The sector also accounts for more than 40 percent of national gross 
domestic product (GDP) and 90 percent of exports, and it provides basic needs and 
income to more than 90 percent of the country’s poor population. In recent years 
the agricultural sector has driven economic growth in the overall economy while also 
improving food security and reducing poverty (World Bank 2007). 

Although Ethiopia remains an agrarian economy, its agricultural development 
has been challenged by a large and rapidly growing population combined with 
limited and deteriorating land resources. Agricultural growth has also fl uctuated 
considerably and, until recently, positive growth was rarely maintained for more 
than three consecutive years. There have thus been many doubts about agriculture’s 
potential contribution to economic growth and poverty reduction in the country. 
Although the role of agriculture in leading industrialization and economic transfor-
mation has been debated for many years, it has remained fairly political and has not 
been well served by rigorous empirical evidence. As the second largest country in 
Africa, one of the key questions dominating the policy debate has been the con-

This chapter is based on research conducted under the project “The Economic Importance of 
Agriculture for Sustainable Development and Poverty Reduction,” managed by the Organization for 
Economic Cooperation and Development. That study was fi nanced by the Government of the 
Netherlands. Any errors are the responsibility of the principal authors. 



108      XINSHEN DIAO ET AL.

straints that the diminishing size of farm plots is imposing on agriculture’s growth 
potential. Moreover, although most of the population, particularly the poor, lives 
in rural areas, some observers doubt whether small-scale agriculture can continue to 
signifi cantly reduce poverty. 

Even among agriculture’s proponents debate exists about what kind of agri-
cultural growth should be pursued. Should the government promote large-scale 
agriculture, which is perhaps more likely to adopt modern technology and thus be 
more productive and competitive? Alternatively, should the government focus on 
smallholder agriculture, which might benefi t the rural population more directly? 
More specifi cally, should specifi c policy supports or interventions focus on export-
led agriculture targeting niche foreign markets, or should staple crops and livestock 
be emphasized to encourage broader-based agricultural growth? This chapter con-
tributes to this ongoing debate in Ethiopia by evaluating the role of agricultural 
growth in future economic transformation and poverty reduction. The analysis 
below quantitatively measures agriculture’s economywide linkages and the impact 
of growth in different agricultural subsectors on poverty reduction. As with other 
chapters in this volume, the analysis makes use of a spatially disaggregated recursive 
dynamic computable general equilibrium (DCGE) model (see Chapter 2).

The chapter is structured as follows. We fi rst review Ethiopia’s national develop-
ment strategies and the recent performance of its agricultural sector. We then 
describe the structure of the Ethiopian DCGE model and the data sources used 
during calibration. The model results are then presented for the baseline growth 
scenario and the accelerated agricultural growth scenarios. Unlike other case studies 
in this volume, no detailed investment analysis is done for Ethiopia. However, the 
fi nal section draws together our fi ndings and summarizes their implications for 
future national and agricultural development strategies in Ethiopia.

Agriculture in Ethiopia

Overview of Agricultural Policies and Development Strategies
Although Ethiopia has experienced three major political regime changes in recent 
history, the importance of agriculture has been recognized by each government 
throughout this period. However, the policies pursued by successive administrations 
have resulted in very different outcomes. Only over the past two decades has the 
country’s agricultural development strategy become more focused and policies more 
consistent with promoting sustainable improvements in agricultural productivity. 
The Derg regime (1975–1991) was characterized as an agrarian socialist regime 
with widespread government controls in all economic spheres, including agricul-
ture. After overthrowing the imperial regime of Haile Selassie, the Derg announced 



ETHIOPIA     109

an agrarian reform program in which all rural land was made the property of 
the state. Almost all other assets in the industrial and services sectors were national-
ized, including factories, fi nancial institutions, large hotels, and many residential 
buildings. 

The agrarian reform prohibited all tenancy relations and provided a large num-
ber of rural households with access to land for cultivation. However, restrictions on 
plot size per family, prohibition of hired agricultural labor, intensifi cation of col-
lectivization, establishment of large-scale state farms, and a series of other antimarket 
and state-controlled economic instruments created negative incentives for farmers 
and distorted market mechanisms guiding land allocation and productivity improve-
ments. Although this central planning development strategy identifi ed agriculture 
as a key engine of growth and targeted improved food security and agricultural 
productivity, most of these targets remained on paper, and few were achieved. As a 
result, the economy was approaching a macroeconomic crisis by the end of the Derg 
regime—a fact that was compounded by the onset of Ethiopia’s worst famine on 
record in 1984. 

Weak policies and political repression during the Derg period not only gener-
ated disastrous economic outcomes but also led to civil confl ict. The Derg regime 
collapsed in 1991, and the Ethiopian People’s Revolutionary Democratic Front 
(EPRDF) assumed power. The years that followed witnessed a radical shift in gov-
ernment policy. Both the transitional government (1991–94) and the EPRDF 
government initiated extensive economic reforms, including market liberalization 
and a structural adjustment program. At the macro level, import protection was 
reduced, foreign exchange controls were eased, and state-owned enterprises were 
privatized, including banks. In agriculture, cooperative membership was made 
compulsory, grain delivery was discontinued, and fertilizers subsidies were discon-
tinued. Consequently, the direct role of the state in economic activity declined 
dramatically.

The most important development strategy that emerged under the transitional 
government was the adoption of Agriculture Development Led Industrialization 
(ADLI), which has remained a central component of the current government’s 
development program until recent years. ADLI focuses on the productivity growth 
of small farms and on labor-intensive industrialization. This strategy has been justi-
fi ed by agriculture’s large contribution to national income, employment, and 
exports, and by the gaps between rural and urban incomes. 

Consistent with ADLI, in the mid-1990s, the government refocused its policy 
reforms from market-based mechanisms (“getting the prices right”) to more active 
public investment in agricultural extension aimed at boosting productivity through 
the widespread introduction of modern technology (MOFED 2002). An extensive 
extension program—the Participatory Demonstration and Training Extension 



110      XINSHEN DIAO ET AL.

System (PADETES)—was implemented, and through this system, the government 
delivered off-the-shelf packages of fertilizer, improved seed, and credit, as well as 
information on input use and better agricultural practices to most rural small-
holders. The promotion of the credit–fertilizer package was accompanied by further 
liberalization of fertilizer markets. By 1997, fertilizer subsidies were completely 
removed, and retail prices were fully liberalized, which also resulted in higher fertil-
izer prices. Fertilizer use increased, although diffusion and adoption rates remained 
disappointing, despite the vigorous promotion of the credit–fertilizer packages. 
Overall, agricultural output continued to fall behind population growth. 

Acknowledging the limited success of PADETES, the government revisited the 
program and formulated an integrated rural and agriculture development strategy 
that was launched in 2002. The new strategy—offi cially known as the Sustainable 
Development and Poverty Reduction Program (SDPRP)—centered on the goal of 
poverty reduction (MOFED 2002). In line with SDPRP, the government intro-
duced fi scal decentralization, judicial and civil service reform, and public sector 
capacity building. After the continuing evidence of widespread food insecurity dur-
ing the drought of 2002/03, the government also implemented social safety nets, 
programs to build the assets of food-insecure households, land resettlement, and soil 
and water conservation (especially water harvesting). 

The SDPRP, which covered the three years between 2002/03 and 2004/05, was 
the fi rst full Poverty Reduction Strategy Paper (PRSP) developed and implemented 
by the Ethiopian government. It was followed by the current PRSP titled “Plan for 
Accelerated and Sustained Development to End Poverty” (PASDEP). PASDEP 
formed Ethiopia’s guiding strategic framework for the fi ve-year period 2005/06–
2009/10 (MOFED 2005). PASDEP aims to signifi cantly accelerate growth via the 
commercialization of agriculture and the promotion of private sector development. 
It focuses on several areas in setting its targets and designing interventions, including 
adopting a geographically differentiated strategy, addressing population growth, 
expanding and improving infrastructure, managing risk, and creating employment 
(MOFED 2005). The instruments to achieve these goals include farm-to-market 
roads, agricultural credit, extension services, national business plans tailored to 
specialized export crops (for example, spices, cut fl owers, fruits, and vegetables), 
irrigation and multipurpose dams, land tenure security and provision of land for 
large-scale commercial farming, and reforms to improve the availability of fertilizer 
and seeds. 

Recent Agricultural Performance
In recent years policy reforms, agricultural investments, and public service provi-
sions have boosted agricultural production, particularly cereals. After 2003, Ethiopian 
agriculture witnessed its most rapid growth period in history. As Figure 5.1 indi-



ETHIOPIA     111

cates, this rapid growth has continued for at least six years, until 2009 (the last 
year for which data are available). Even excluding the highest growth rate of 17 
percent in 2004, which refl ects recovery from a major drought in 2002/03, the 
average annual growth rate during 2005–09 is 9.5 percent. The accuracy of these 
production numbers is of some concern, and a rigorous, comprehensive inves-
tigation into the sources of such rapid agricultural growth has not yet been con-
ducted. Some evidence to support rapid agricultural growth comes from the 
agricultural sample surveys, whose measured cereal production growth is consistent 
with agricultural GDP growth rates. This recent performance has reversed the three-
decade downward trend in per capita agricultural GDP.

Cereals are the dominant staples for most Ethiopians. They account for 62 
percent of average daily calorie intake and for 45 percent of food expenditure for an 
average household. Thus, cereals, including barley, maize, teff, sorghum, and wheat, 
are the most important crops in Ethiopian agriculture. Although 64 percent of 
agricultural value-added comes from crops, more than 70 percent of cropland is 
devoted to cereal production. More than 11 million smallholders engage in cereal 
production, and total cereal production was 13.6 million metric tons in 2007/08—
an increase of 4.8 million metric tons over production in 2003/04 (Table 5.1). 
Total cropland area allocated to cereals also expanded by 27 percent, from 6.8 mil-
lion hectares in 2003/04 to 8.8 million hectares in 2007/08. At the same time, 
average cereal yield exhibited a 22 percent growth from 1.3 mt/ha in 2003/04 to 
1.6 mt/ha in 2007/08.

Growth per capita (constant 2000 US$)

80

50

65

70

55

60

75

20062001199619911981 1986

Per capita agricultural GDP
Trendline R2 � 0.656

Figure 5.1—Per capita agricultural GDP in Ethiopia, 1981–2009

Source: Authors’ calculations using World Bank (2011).
Note: GDP = gross domestic product.



112      XINSHEN DIAO ET AL.

Ta
bl

e 5
.1—

Ce
re

al 
pr

od
uc

tio
n 

in
 E

th
io

pi
a, 

20
03

/04
 an

d 
20

07
/08

 
20

03
/04

 
20

07
/08

 
Ar

ea
 

Ou
tp

ut
 

Yi
eld

 
Ar

ea
 

Ou
tp

ut
 

Yi
eld

 
(th

ou
sa

nd
s 

(th
ou

sa
nd

s o
f 

(m
et

ric
 to

ns
/  

(th
ou

sa
nd

s 
(th

ou
sa

nd
s o

f 
(m

et
ric

 to
ns

/
Ce

re
al 

of
 h

ec
ta

re
s)

 
 m

et
ric

 to
ns

) 
 h

ec
ta

re
) 

 o
f h

ec
ta

re
s)

 
 m

et
ric

 to
ns

) 
 h

ec
ta

re
)

Pr
od

uc
ed

 w
ith

 an
d w

ith
ou

t m
od

er
n i

np
uts

 
 

 
 

 
Ba

rle
y 

91
1 

1,0
71

 
1.1

8 
98

5 
1,3

55
 

1.3
8

 
Ma

ize
 

1,3
00

 
2,4

55
 

1.8
9 

1,7
67

 
3,7

50
 

2.1
2

 
Te

ff 
1,9

85
 

1,6
72

 
0.8

4 
2,5

65
 

2,9
93

 
1.1

7
 

W
he

at 
1,0

75
 

1,5
89

 
1.4

8 
1,4

25
 

2,3
14

 
1.6

2
 

Mi
lle

t 
30

3 
30

4 
1.0

0 
39

9 
53

8 
1.3

5
 

So
rg

hu
m 

1,2
42

 
1,6

95
 

1.3
6 

1,5
34

 
2,6

59
 

1.7
3

 
To

tal
 

6,8
16

 
8,7

86
 

1.2
9 

8,6
75

 
13

,60
9 

1.5
7

Pr
od

uc
ed

 w
ith

 m
od

er
n i

np
uts

 
 

 
 

 
Ba

rle
y 

39
0 

49
2 

1.2
6 

49
9 

68
8 

1.3
8

 
Ma

ize
 

78
3 

1,5
18

 
1.9

4 
1,1

06
 

2,4
41

 
2.2

1
 

Te
ff 

1,0
59

 
91

9 
0.8

7 
1,7

63
 

2,0
76

 
1.1

8
 

W
he

at 
68

7 
1,1

08
 

1.6
1 

1,0
24

 
1,7

43
 

1.7
 

To
tal

 
2,9

19
 

4,0
37

 
1.3

8 
4,3

92
 

6,9
48

 
1.5

8
Pr

od
uc

ed
 w

ith
ou

t m
od

er
n i

np
uts

 
 

 
 

 
Ba

rle
y 

52
1 

57
9 

1.1
1 

48
6 

66
7 

1.3
7

 
Ma

ize
 

51
7 

93
7 

1.8
1 

66
1 

1,3
09

 
1.9

8
 

Te
ff 

92
6 

75
3 

0.8
1 

80
2 

91
7 

1.1
4

 
W

he
at 

38
8 

48
1 

1.2
4 

40
1 

57
1 

1.4
3

 
To

tal
 

2,3
52

 
2,7

50
 

1.1
7 

2,3
50

 
3,4

64
 

1.4
7

So
ur

ce
: A

uth
or

s’ 
ca

lcu
lat

ion
 us

ing
 da

ta 
ag

gr
eg

ate
d f

ro
m 

Et
hio

pia
n A

gr
icu

ltu
re

 S
am

ple
 S

ur
ve

y (
Et

hio
pia

, C
SA

 20
04

, 2
00

8a
, 2

00
8b

).



ETHIOPIA     113

Teff is the favored staple crop for both rural and urban consumers at all income 
levels and has high income elasticity of demand relative to other cereals (see Table 
5A.1 in the appendix to this chapter). Not surprisingly then, more than 30 percent 
of total cereal land in 2007 was allocated to teff production. The next most impor-
tant foodcrop is maize, which occupies 20 percent of total cereal land, followed by 
sorghum (18 percent), wheat (16 percent), and barley (12 percent). Although most 
cereal crops are used as staple foods, barley is also used for local alcohol production. 
In terms of volume of production in the same year (2007), maize actually ranked 
fi rst, with 3.8 million metric tons of output (teff is 3.0 million metric tons). Despite 
occupying 30 percent of cereal land, teff output is only equivalent to 22 percent of 
cereal output. Thus, teff is much less productive in terms of land use. Indeed, the 
national average yield of maize is 2.1 mt/ha in 2007, whereas teff yields are only 1.2 
mt/ha—the lowest yield among all major cereal crops. Teff is only grown in a few 
countries, and its yield response to fertilizer is relatively limited, given that the 
technology needed to develop high-yield teff varieties is more diffi cult to develop 
than that for other cereal crops, which are more broadly grown and researched 
around the world. This inconsistency between the technology potential and con-
sumer preference presents a potential challenge for Ethiopian food security.

Under ADLI, the government emphasized intensifi cation to increase agricul-
tural production, especially through a centralized extension system to push techno-
logical packages that combined credit, fertilizers, improved seeds, and better farm 
management practices. Under this program, fertilizer use almost doubled between 
1990 and 2000 to 290,000 metric tons (World Bank 2007). However, the intensity 
of fertilizer nutrient use per hectare stagnated in the latter half of 1990s and the 
present decade. According to World Bank (2007), only 37 percent of farmers use 
inorganic fertilizer, and application rates remain low at 16 kg/ha of nutrients (about 
33 kg/ha for commercial products).

Following the 2002/03 drought, a large-scale food security program was scaled 
up in poor and vulnerable areas, amounting to a signifi cant share of total public 
spending. More recently, weaknesses in the marketing system have been recognized, 
and a new marketing strategy is being developed based on scaling-up cooperatives 
and establishing a commodity exchange (and its associated institutions). As a result 
of these various programs, public spending on agriculture, natural resources, and 
food security increased from 5 percent of the total government budget in 1997/98 
to more than 10 percent in 2003/04.  

Concerted efforts to promote intensifi cation seem to be showing some positive 
results in recent years. Consistent with growth in total agricultural GDP, 2003/04–
2007/08 registered record cereal production growth in Ethiopian history. Total 
cereal production increased by 54.9 percent in four years (that is, more than 10 
percent per year). Against the historical trend in which almost all increased produc-



114      XINSHEN DIAO ET AL.

Ta
bl

e 5
.2—

Ce
re

al 
gr

ow
th

 an
d 

gr
ow

th
 co

nt
rib

ut
io

n,
 20

03
/04

–2
00

7/0
8

 
Gr

ow
th

 co
nt

rib
ut

io
n 

(p
er

ce
nt

)

 
Sh

ar
e o

f c
ro

p’s
 o

ut
pu

t g
ro

wt
h 

 
Sh

ar
e o

f t
ot

al 
ce

re
al’

s o
ut

pu
t g

ro
wt

h
 

du
e t

o 
in

cr
ea

se
 in

 
du

e t
o 

in
cr

ea
se

 in

Ce
re

al 
Ar

ea
 

Yi
eld

 
Ou

tp
ut

 
Ar

ea
 

Yi
eld

 
Ou

tp
ut

 
Ar

ea
 

Yi
eld

 
Ou

tp
ut

Ba
rle

y 
8.1

 
17

.0 
26

.5 
30

.6 
69

.4 
10

0.0
 

3.0
 

6.9
 

10
.0

Ma
ize

 
35

.9 
12

.4 
52

.7 
68

.1 
31

.9 
10

0.0
 

18
.8 

8.8
 

27
.6

Te
ff 

29
.2 

38
.5 

79
.0 

37
.0 

63
.0 

10
0.0

 
8.1

 
13

.9 
22

.0
W

he
at 

32
.6 

9.9
 

45
.6 

71
.4 

28
.6 

10
0.0

 
12

.1 
4.9

 
17

.0
Mi

lle
t 

31
.7 

34
.4 

77
.0 

41
.2 

58
.8 

10
0.0

 
1.6

 
2.3

 
4.0

So
rg

hu
m 

23
.5 

27
.0 

56
.9 

41
.3 

58
.7 

10
0.0

 
8.1

 
11

.5 
19

.5
Al

l c
er

ea
ls 

27
.3 

21
.7 

54
.9 

49
.7 

50
.3 

10
0.0

 
49

.7 
50

.3 
10

0

So
ur

ce
: A

uth
or

s’ 
ca

lcu
lat

ion
 us

ing
 da

ta 
ag

gr
eg

ate
d f

ro
m 

Et
hio

pia
n A

gr
icu

ltu
re

 S
am

ple
 S

ur
ve

y (
Et

hio
pia

, C
en

tra
l S

tat
ist

ica
l A

ge
nc

y 2
00

4, 
20

08
a, 

20
08

b)
.

To
ta

l g
ro

wt
h 

(p
er

ce
nt

)



ETHIOPIA     115

tion was accounted for by area expansion, Table 5.2 shows that about half of recent 
growth has been due to yield increases. The table also presents the growth of indi-
vidual crops and decomposes this into area expansion and yield increases. Among 
the four main cereal crops, there are two crops, teff and sorghum, for which growth 
in production is a dominant result of yield increases, which contributes to 63 and 
58.7 percent of teff and sorghum output growth, respectively. Even though area 
expansion is the main source of growth for maize and wheat, increases in yields still 
account for 32 and 29 percent of their output growth, respectively.1

An important characteristic of Ethiopia’s agricultural sector is the high geo-
graphic concentration of cereal cultivation. Almost 80 percent of total area under 
cereals is in the Amhara and Oromia regions (that is, northwest, west, southwest, 
and south of Addis Ababa). This area includes a diverse set of conditions for agri-
cultural production. The spatial conditions for crop production and access to 
markets are characterized in Diao and Nin Pratt (2007) and Taffesse (2007).

Table 5.1 also separates production of the four main cereals by whether modern 
inputs are used in their production. Recently, Ethiopia has witnessed a rapid increase 
in modern input use, particularly fertilizer. The government’s fertilizer-promotion 
policy currently focuses on the four main cereal crops, which are relatively more 
responsive to fertilizer and have higher producer prices. Table 5.1 shows the result of 
this program: in 2003/04, fewer than 3 million hectares used to grow barley, maize, 
teff, and wheat (over 50 percent of the total area devoted to these crops) used 
modern inputs; by 2007/08, the area that used modern inputs had increased 
to more than 4 million hectares (over 60 percent of the total area devoted to these 
four crops). In these four years, total harvest areas of these four crops increased 
by 28 percent, but the areas not using modern input remained constant at about 
2.3 million hectares. 

Unfortunately, the observed yield difference between production with and 
without modern inputs is fairly modest for all four cereals crops. With the exception 
of wheat, the average yield when using modern inputs is less than 15 percent higher 
than the average yield without such inputs. Even for wheat, the yield gap was 30 
percent in 2003/04 but has since fallen to 19 percent in 2007/08. Although many 
factors affect fertilizer effi ciency, increased use of fertilizer without use of high-yield 
seed varieties seems to be the most important factor. The combined use of fertilizer 
and improved seed is extremely low for teff and wheat farmers. Moreover, the low 
yield response to combined seed–fertilizer use indicates that the so-called “improved” 
seed are not really high-yielding varieties. Further assessment is needed to under-
stand farmer’s behavior and the constraints to promoting modern input use in 
Ethiopia.

Agricultural performance in recent years also refl ects the new direction taken 
in the country’s development strategy (that is, PASDEP), which targets market-



116      XINSHEN DIAO ET AL.

driven diversifi cation and commercialization and the increase of private sector 
exports. After improving incentives for private investment in the fl ower industry, 
for example, more than 100 new investors entered the industry, and exports increased 
to nearly US$13 million by 2005. Other investments in high-value products and 
supply chains include exports of green beans to Europe and provision of higher 
quality milk and poultry to local urban centers. Several of these emerging industries 
involve outgrower or contract arrangements with small farmers, often linked to an 
emerging indigenous entrepreneurial class of farmers and agribusinesses. Exports of 
oilseeds and pulses (two traditional cash crops) have also experienced impressive 
growth—increasing their value by a factor of 10 between 1997 and 2006 and dem-
onstrating their increasing competitiveness and the uptake of new technologies. 
Although coffee is still the most important export crop in Ethiopia, the combined 
exports of other crops and leather has sometimes surpassed the value of coffee 
exports in recent years.

The Ethiopian DCGE Model
We apply the DCGE model described in Chapter 2 to assess the contribution of 
agricultural sectors to economywide growth and poverty reduction. The DCGE 
model for Ethiopia was developed by Dorosh and Thurlow (2009) based on an 
adapted version of the social accounting matrix developed by Ahmed et al. (2009). 
Given that the Ethiopian economy is still dominated by agriculture and that more 
than 80 percent of the population lives in the rural areas, special attention has 
been paid to the structure of agriculture in the model. There are 24 agricultural 
sectors (see Table 5A.2 in the appendix). Crop production is the most important 
agricultural activity, hence the model defi nes fi ve broad crop groups: 

1.  Cereal crops, which are separated into teff, barley, wheat, maize, and sorghum 
and millet.

2.  Pulses and oilseeds, both of which are aggregated crop groups in the model. In 
Ethiopia, the most important oilseed crops are sesame and groundnuts. 

3. Horticulture, which is separated into fruits, vegetables, and enset.

4.   High-value export-oriented crops, which are separated into coffee, cotton, cut 
fl owers, sugarcane, tea, and tobacco. 

5. Other crops, including chat and root crops. 



ETHIOPIA     117

The model also identifi es four livestock subsectors: cattle, milk, poultry, and 
other animal products. The two remaining agricultural subsectors are fi shery and 
forestry. Within its general equilibrium framework, the model also contains detailed 
information on 44 nonagricultural sectors. 

To account for geographic factors, such as agroecological conditions, popula-
tion distribution, production patterns, and market locations and connections, the 
Ethiopian model further disaggregates each agricultural subsector into four sub-
national regions: Zone 1 covers the humid cereals region; Zone 2 covers the humid 
enset region; Zone 3 is the drought-prone region; and Zone 4 is the arid pastoralist 
region (see Figure 5.2). 

Different agroecological conditions determine crop and livestock production 
patterns in the four zones. Zones 1 and 2 are located in the highland region, where 
rainfall is reliable and therefore crop growing periods are longer in these two zones 

SUDANSUDAN

SOMALIASOMALIA

KENYAKENYA

ERITREAERITREA

DJIBOUTIDJIBOUTI

UGANDAUGANDA

AsAsososa

GambelGambela

AsAsayayitita

S o m a l iS o m a l i

O r o m i aO r o m i a

A f a rA f a r

A m h a r aA m h a r a

SNNPSNNP

T iT i g r ag r a y

G a m b e l l aG a m b e l l a

B e n i s h a n g u lB e n i s h a n g u l
G u m u zG u m u z

D iD i r er e  D a w D a w a

H a r a r iH a r a r iAd d id d i s  s  Ab a b ab a b a

ArArbaminbaminchch

DeDebrbre e BirBirhanhan

DeDebrbre e MaMarkosrkos

NeNekekemtmte

AsAsseselala

AkAkakaki

SUDAN

SOMALIA

KENYA

ERITREA

DJIBOUTI

UGANDA

Asosa

Gambela

Asayita

S o m a l i

O r o m i a

A f a r

A m h a r a

SNNP

T i g r a y

G a m b e l l a

B e n i s h a n g u l
G u m u z

D i r e  D a w a

H a r a r iAd d i s  Ab a b a

Cities � 20,000

20,406–30,000
30,001–50,000
50,001–100,000
100,001–150,000
150,001–2,738,248

Road type

Trunk
Main access

Agroecological zones

Drought prone

Moisture reliable

Moisture reliable
–enset area

Pastoralist

Arbaminch

Debre Birhan

Debre Markos

Nekemte

Assela

Akaki

Figure 5.2—Agroecological zones of Ethiopia 

Source: This figure is adapted from a map that previously appeared in EDRI (2009). The map was 
created by Emily Schmidt, a program coordinator for the Ethiopia Strategy Support Program II of the 
International Food Policy Research Institute. The map is based on Ethiopian Roads Transport Authority 
(2006).



118      XINSHEN DIAO ET AL.

than elsewhere. Because Ethiopia’s agriculture is mainly rainfed, suitable rainfall 
conditions explain why these two zones have high agricultural potential. In fact, 
Zone 1 is the country’s grain basket: it produces 60–70 percent of the country’s 
cereals. Although Zone 3 is larger than Zone 1, it has less reliable rainfall and so 
only produces 25–30 percent of total production for most grain crops. The excep-
tions are sorghum and millet, which are relatively drought resistant. Hence, Zone 
3 produces about 50 percent of these crops. Zone 4 is called a “pastoralist region,” 
but only about one-fi fth of national livestock is produced in this zone because of 
harsh agroecological conditions and low population densities. In contrast, more 
than 40 percent of livestock is produced in Zone 1, refl ecting the close relationship 
between crops and livestock production among smallholders in Ethiopia. 

The household structure in the DCGE model allows us to conduct growth–
poverty linkage analysis. In the model, there are 12 representative household groups 
that are disaggregated by the four rural zones, small and large urban centers, and 
income level (poor or nonpoor status). “Poor” is defi ned as all households falling 
into the lowest two per capita expenditure quintiles (that is, the poorest 40 percent 
of the population defi ned at the national level). Household income elasticities are 
based on estimates from the 2004/05 Household Income and Expenditure Survey 
(HICES) (Ethiopia, CSA 2005; see Table 5A.1 in the appendix). Each household 
questioned in HICES 2004/05 is linked directly to the corresponding representative 
household in the model through a top-down consumption-based microsimulation 
model (see Chapter 2). In this formulation of the model, changes in representative 
households’ consumption and prices in the DCGE model are passed down to the 
survey, where total real consumption for each survey household is recalculated. This 
new level of per capita total consumption for each survey household is compared to 
a poverty line, so that poverty headcount can be recalculated. 

Both total and sector GDPs are endogenous variables, and their growth comes 
from three major dynamic factors incorporated in the model: (1) growth in factor 
supply, (2) change in total factor productivity (TFP), and (3) accumulation of capi-
tal through investment. By design, the supplies of skilled and unskilled labor, agri-
cultural land, and livestock stocks are fi xed in each time period, and annual growth 
rates are exogenously imposed. In contrast, the supply of agriculture-specifi c labor 
(that is, family workers) and workers with elementary occupations are fl exible (that 
is, they are endogenous) and depend on supply elasticities with respect to wage rates, 
which endogenously change over time. Changes in TFP at the subsector level are 
exogenous, whereas capital accumulation as an endogenous variable is the out-
come of past investment, which is fi nanced by private, public, and foreign savings. 
Foreign savings is an exogenous variable chosen as a macroclosure to balance the 
foreign account in the model. Public savings is a residual term between government 



ETHIOPIA     119

revenue (which is an endogenous variable) and government expenditure (which is 
an exogenous variable chosen as the second macroclosure to balance the government 
account in the model). Finally, private savings is an endogenous variable, although 
the saving rate is exogenous.

Baseline Growth Scenario 
The DCGE model for Ethiopia is fi rst used to design a baseline scenario, against 
which a series of accelerated growth scenarios are compared later in the chapter. 
Although economic growth rates in Ethiopia were extremely high during 2004–08, 
the recent global fi nancial crisis is expected to have dampened the country’s growth 
performance during 2009–10. Given this possibility and frequent weather shocks 
on agricultural production, we adopt a conservative annual growth rate as a target 
for the baseline scenario. Specifi cally, the average growth rate during 1997–2007 is 
chosen to calibrate the baseline, during which time total GDP grew at 6.1 percent 
and agricultural GDP at 4.1 percent per year on average (World Bank 2011).

We choose growth rate in labor supply (except for the supply of agricultural 
family workers and unskilled labor) consistent with population growth of 2.7 per-
cent per year. The same growth rate is also chosen for the supply of livestock capital. 
The agricultural land expansion rate varies between 2.3 and 3.3 percent for the four 
zones (with lower growth in the two highland zones and a higher growth rate in the 
other two zones). TFP growth in each agricultural subsector is chosen so that the 
subsector’s growth rate tracks production trends in recent years. Overall, agricultural 
GDP grows 4.2 percent in the baseline.

To understand how the choice of exogenous growth rates affects the baseline, 
we present some aggregate results in Table 5.3 to explain the relationship between 
factor growth and TFP change (mostly exogenous) on the one side, and resultant 
economic growth (endogenous) on the other. The fi rst row of the table reports 
annual growth rates for total GDP, agricultural GDP, total value-added for cereal 
production, total value-added for export crop production, and nonagricultural 
GDP. These are all endogenous results from the baseline scenarios. 

Under the heading “Role of labor,” the table presents the share of labor in value 
terms in fi ve types of GDP and sector value-added. For example, labor accounts for 
30.4 percent of GDP (column 1), 34.2 percent of agricultural GDP (column 2), 
and 26.7 percent of nonagricultural GDP (column 5). Thus, the agricultural sector 
is more labor-intensive than the nonagricultural sector. The table also presents 
labor’s annual growth rate. Total labor at the national level grows at 2.7 percent in 
the baseline (column 1), which is consistent with recent population growth rates. 
However, labor employed in the agricultural sector grows at 2.3 percent (column 2) 



120      XINSHEN DIAO ET AL.

and 3.1 percent in the nonagricultural sector (column 5). It should be noted that 
sector-level labor growth is an endogenous result of the model, because most catego-
ries of workers are able to migrate between farm and nonfarm activities. 

Although the nonagricultural sector is less labor intensive, more rapid growth 
in this sector (7.5 percent) requires that more labor be employed in nonagricultural 
production. The last row under “Role of labor” presents the contribution of labor 
to growth. Labor accounts for 13.8 percent of total GDP’s annual growth (column 
1) and 18.7 percent of agricultural GDP’s growth (column 2). Labor has a much 
smaller contribution to nonagricultural GDP growth (11.3 percent), because this 
sector is less labor intensive. 

The last component of Table 5.3 (fi nal two rows) presents TFP growth rates 
and the contribution of TFP to economic growth. Although subsector-level TFP 
growth rates are exogenously chosen in the model, calculated total TFP is actually 

Table 5.3—Decomposition of growth in baseline scenario (percent)

 Total Agricultural Cereal Export crop Nonagricultural
 national GDP GDP value-added value-added GDP
Category (1) (2) (3) (4) (5)
Annual GDP growth rate 6.1 4.2 4.8 5.0 7.5
Role of labor     
 Share of value-added 30.4 34.2 51.2 51.2 26.7
 Annual growth rate 2.7 2.3 2.2 2.2 3.1
 Contribution to growtha 13.8 18.7 23.3 22.5 11.3
Role of land     
 Share of value-added 18.2 37.2 48.8 48.8 
 Annual growth rate 3.4 3.4 2.6 2.6 
 Contribution to growtha 10.7 30.5 26.9 26.0 
Role of capital     
 Share of value-added 51.4 28.6   73.3
 Annual growth rate 5.0 2.8   5.7
 Contribution to growtha 47.3 19.1   63.6
Total factor productivity (TFP)     
 Annual growth rateb  1.6 1.3 2.3 2.5 1.7
 Contribution to growth 28.1 31.6 49.8 51.5 25.1

Source: Ethiopia dynamic computable general equilibrium model results.
Note: GDP = gross domestic product. TFP = total factor productivity. Blank cells = not applicable.
aContribution of a factor to growth is the ratio of the factor’s growth to economic growth (either growth of total GDP or a par-
ticular sector’s GDP, for example, agricultural GDP) weighted by the share of the factor in an economic sector or in the whole 
economy.
bTFP growth rate is derived from gTFP = g – ∑wt ̇ gt , where gTFP is TFP annual growth rate; g is economic growth rate for 
 t
GDP or agricultural GDP; wf is the share of factor f in GDP or sector GDP; and gf is this factor’s annual growth rate. Growth in 
TFP is the difference between economic growth and summation of growth rates for all factors (∑t) employed in the economy.



ETHIOPIA     121

a residual term, and hence its growth is affected by a series of endogenous variables 
in the decomposition process. As shown in the fi nal row of the table, TFP contrib-
utes 28.1 percent to overall economic growth in the baseline (column 1) and 49.8 
percent to cereal growth (column 3). This result is consistent with the observation 
made earlier in this chapter that yield improvements accounted for half of additional 
cereal production in recent years. Even though yield measures the productivity of 
land, this comparison is still meaningful, because yield increases are closely related 
to the improvement in labor effi ciency in labor-intensive agriculture. 

One of the main objectives of this chapter is to quantitatively measure the 
relationship between economic growth and poverty reduction in a general equilib-
rium setting. Therefore, it is necessary to further evaluate whether the poverty results 
in the baseline scenario are consistent with historical trends. National poverty rates 
are available from the three most recent HICES. Table 5.4 fi rst presents these 
poverty estimates and then calculates annual changes (in the fi nal three columns). 
Poverty reduction between the two most recent surveys (between 1996/2000 
and 2000/05) is more rapid than that between the fi rst two surveys. The table 
reports the annual growth in per capita GDP for the same periods. Using the histori-
cal data, Ethiopia’s poverty–growth elasticity is calculated as the ratio of the annual 
poverty reduction rate to the annual growth rate of per capita GDP. The elasticity 
of –0.35 in the fi nal row of the table indicates that during 1996–2000, 1 percent 
annual growth in per capita GDP led to a 0.35 percent decline in the national 
poverty headcount rate. The elasticity was higher during the second period (–0.71). 
Assessing the differential growth rates between agricultural and nonagricultural 
GDP indicates that when agriculture grows slowly (for example, it is 0.03 percent 

Table 5.4—Relationship between poverty and growth in Ethiopia’s history

 Value Annual change (percent)

Measure 1995/96 1999/2000 2004/05 1996–2000 2000–05 1996–2005
National poverty rate  45.5 44.2 38.7 –0.58 –2.62 –1.61
Per capita GDP 937 1,017 1,219 1.66 3.69 2.67
Per capita agricultural  474 474 532 0.03 2.33 1.17
 GDP 
Per capita nonagricultural  463 543 687 3.24 4.83 4.03
 GDP  
Poverty–growth elasticity      –0.35 –0.71 –0.60

Source: Authors’ calculations using poverty data (Ethiopia, CSA 2005) and GDP data (World Bank 2011).
Notes: The national poverty rate is the proportion of the nation’s population with per capita consumption below 
the poverty line. It is measured in percent. Per capita GDP and per capita sector GDP are measured in constant 
birr. GDP = gross domestic product. Blank cells = not applicable.



122      XINSHEN DIAO ET AL.

per year in the fi rst period), the value of the poverty–growth elasticity is also low. 
Thus, generally speaking, the size of the poverty–growth elasticity appears to be 
related to agricultural growth rates. Later in this chapter we more rigorously examine 
the relationship between agriculture- and nonagriculture-led growth and poverty 
reduction.

The same method of calculating poverty–growth elasticities is used for the 
baseline scenario and for other simulations discussed later in the chapter. Although 
per capita GDP growth rates can be directly obtained from the DCGE model, 
this is not the case for poverty headcounts, whose calculation requires the detailed 
information in the household survey. As discussed above, the representative 
household groups in the DCGE model are aggregates of the households appearing 
in the 2004/05 HICES. The simulated changes in these household groups’ con-
sumption of each commodity in the DCGE model are used to adjust the consump-
tion of individual households in the 2004/05 HICEs. We then recalculate poverty 
headcounts and obtain a new poverty rate for each year in the simulation period. 
An important caveat of this process is that the model captures differential changes 
in income and consumption across, but not within, the 12 household groups. 
Bearing this caveat in mind, the baseline produces a poverty–growth elasticity 
of –1.13. Thus, a 1 percent growth rate in per capita GDP reduces the poverty 
rate by 1.13 percent. Although this elasticity is higher than historical trends, it is 
fairly consistent with results from other African countries reported in this volume 
and elsewhere.

Accelerated Growth Scenarios 
The DCGE model is now used to assess the role of agriculture in economic growth 
and poverty reduction. We conduct six counterfactual simulations that differ 
according to the sectors driving an accelerated growth process. The fi rst scenario is 
the “cereal-led growth,” in which additional TFP growth is assumed for the fi ve 
cereal subsectors. The assumptions on TFP growth in the other subsectors remain 
unchanged from the baseline, as do the exogenous growth assumptions for labor 
and land. The second scenario is “export crop–led growth,” in which TFP growth 
is accelerated only in the seven agricultural subsectors that are export oriented. 
Similarly, the third scenario is “livestock-led growth.” The fourth scenario com-
bines the fi rst three into an overall “agriculture-led growth” scenario. Conversely, 
the fi fth scenario focuses on “nonagriculture-led growth,” and the fi nal scenario 
accelerates growth in all economic subsectors. The results and major conclusions 
from these six scenarios are presented below, with our discussion focusing on four of 
them: cereal-led growth, export crop–led growth, agriculture-led growth, and 
nonagriculture-led growth.



ETHIOPIA     123

Subsector (Crop-Level) Growth, Employment and Poverty Impacts
We fi rst discuss the scenario results of the model and analyze growth linkage effects 
in a general equilibrium setting. In a static CGE model, productivity growth in a 
shocked sector affects other sectors via various channels—the most important of 
which is resource reallocation. Resource allocation effects can be further enhanced if 
supply of some factors is fl exible (for example, the existence of unemployed labor). 
In our dynamic model, capital accumulation may also be affected by sectoral pro-
ductivity growth when this growth causes income and savings to change, resulting in 
changes in investment and new capital stocks. For this reason, when we discuss 
growth-linkage effects, we analyze not only the growth multiplier but also employ-
ment creation effects. This is important, because Ethiopia is a labor-abundant econ-
omy, and so released farm labor must fi nd employment outside of agriculture.

As mentioned above, there are two types of labor with endogenous labor supply 
in our model: agricultural family labor and unskilled labor. Table 5.5 therefore pays 
particular attention to these two labor categories. We note that, cereal-led growth 
lowers demand for the farm labor within agriculture but does not change demand 
for unskilled labor in the economy as a whole. When labor and land become more 
productive in cereal production, then to produce the same amount of cereals 
requires fewer of these resources. Without demand and land constraints, cereal 
production could continue to increase and hence possibly create additional demand 
for labor. However, cereals are produced mainly for domestic markets in Ethiopia, 
and import substitution is only possible for wheat (that is, there are currently no teff 
or sorghum imports, and maize imports are minimal). Thus, it is impossible for 
cereal production to grow beyond the growth in domestic demand. Moreover, land 
supply is fi xed, and when land is released from cereal production, it is reallocated to 
other crops, which creates growth-multiplier effects in the process. However, if other 
crop production is less intensive in the use of family or unskilled labor, then overall 
demand for these labor groups may decline. As shown in the second part of Table 
5.5, the annual growth rate of cereal value-added increases to 6.6 percent in the 
cereal-led growth scenario—up from 4.8 percent in the baseline. Similarly, the growth 
rate of agricultural GDP rises to 4.8 percent from 4.2 percent in the baseline. How-
ever, there is only a modest acceleration in nonagricultural growth, which explains 
why total unskilled labor demand did not rise when demand for agricultural labor 
fell.

The fi nal panel of Table 5.5 presents the growth multipliers for each scenario. 
Our second major fi nding from the cereal-led growth scenario is that cereals gener-
ate strong growth-multiplier effects. A one unit increase in real cereal value-added 
in 2015 causes a 1.14 unit of increase in total GDP in the same year, 0.14 of which 
is the multiplier effect. However, our third major fi nding that most of this multiplier 
effect is from increased production in other agricultural subsectors, which, as a 



124      XINSHEN DIAO ET AL.

Ta
bl

e 5
.5—

An
nu

al 
fa

ct
or

 g
ro

wt
h 

ra
te

s a
nd

 si
m

ul
at

ed
 G

DP
 g

ro
wt

h 
ela

st
ici

tie
s a

nd
 m

ul
tip

lie
rs

 

 
GD

P 
sh

ar
e

 
in

 20
05

/06
  

Ba
se

lin
e 

 
Ex

po
rt 

 
Al

l 
 

Co
m

bi
ne

d
Ca

te
go

ry
 

(p
er

ce
nt

) 
sc

en
ar

io
 

Ce
re

als
 

cr
op

s 
Li

ve
st

oc
k 

ag
ric

ul
tu

re
 

No
na

gr
icu

ltu
re

 
sc

en
ar

io
La

bo
r s

up
ply

 gr
ow

th 
ra

te 
(p

er
ce

nt)
 

 
 

 
 

 
 

 
Ag

ric
ult

ur
al 

fam
ily

 la
bo

r  
 

2.3
1 

2.3
0 

2.3
3 

2.3
3 

2.3
5 

2.4
0 

2.4
5

 
Un

sk
ille

d l
ab

or
 

 
3.5

3 
3.5

3 
3.5

4 
3.5

7 
3.5

9 
3.7

2 
3.7

8
 

Ca
pit

al 
  

5.3
2 

5.3
2 

5.3
2 

5.3
2 

5.3
2 

5.5
7 

5.5
7

To
tal

 G
DP

 gr
ow

th 
ra

te 
(p

er
ce

nt)
 

10
0.0

 
6.1

 
6.3

 
6.2

 
6.3

 
6.8

 
7.2

 
7.8

 
Ag

ric
ult

ur
al 

GD
P 

45
.2 

4.2
 

4.8
 

4.5
 

4.7
 

5.9
 

4.4
 

6.0
 

 
Ce

re
als

 
13

.2 
4.8

 
6.6

 
4.8

 
4.8

 
6.7

 
4.8

 
6.7

 
 

Ex
po

rt 
cro

ps
 

4.5
 

5.0
 

5.0
 

7.6
 

5.0
 

7.6
 

5.0
 

7.6
 

 
Liv

es
toc

k 
13

.3 
3.9

 
4.0

 
3.9

 
5.6

 
5.7

 
4.0

 
5.8

 
No

na
gr

icu
ltu

ra
l G

DP
 

54
.8 

7.5
 

7.5
 

7.5
 

7.5
 

7.5
 

7.6
 

9.2
 

Ma
nu

fac
tur

ing
 

5.3
 

7.7
 

7.6
 

7.7
 

7.8
 

7.7
 

9.7
 

10
.0

 
 

Ag
ro

pr
oc

es
sin

g 
2.4

 
6.9

 
6.9

 
7.0

 
6.9

 
7.1

 
8.4

 
8.7

 
Pr

iva
te 

se
rvi

ce
s 

32
.8 

8.0
 

8.0
 

8.0
 

8.0
 

8.1
 

10
.0 

10
.0

El
as

tic
ity

 of
 la

bo
r s

up
ply

–G
DP

a  gr
ow

th 
 

 
–0

.02
 

0.1
5 

0.1
1 

0.0
6 

0.1
0 

0.0
9

Gr
ow

th 
mu

ltip
lie

r b
y 2

01
5 

 
 

1.1
3 

1.0
4 

1.0
6 

1.0
3 

1.0
6 

Co
ntr

ibu
tio

n t
o t

ota
l G

DP
 in

cre
as

e b
y 2

01
5 (

pe
rce

nt)
 

 
 

 
 

 
 

 
Sh

oc
ke

d s
ec

tor
sb  

  
 

88
.34

 
96

.50
 

94
.04

 
96

.77
 

93
.92

 
 

Ag
ric

ult
ur

e r
ath

er
 th

an
 sh

oc
ke

d s
ec

tor
 

 
 

10
.29

 
0.3

3 
2.5

3 
 

6.0
8 

 
No

na
gr

icu
ltu

ra
l s

ec
tor

s 
  

  
1.3

7 
3.1

7 
3.4

2 
3.2

3 
  

 

So
ur

ce
: E

thi
op

ian
 dy

na
mi

c c
om

pu
tab

le 
ge

ne
ra

l e
qu

ilib
riu

m 
mo

de
l re

su
lts

.
No

te:
 G

DP
 =

 gr
os

s d
om

es
tic

 pr
od

uc
t. B

lan
k c

ell
s =

 no
t a

pp
lic

ab
le.

a La
bo

r s
up

ply
–G

DP
 el

as
tic

ity
 is

 de
fin

ed
 as

 pe
rce

nta
ge

 ch
an

ge
 in

 ba
se

lin
e t

ota
l la

bo
r s

up
ply

 ov
er

 th
e b

as
eli

ne
 G

DP
 gr

ow
th 

ra
te,

 bo
th 

for
 20

15
.

b Sh
oc

ke
d s

ec
tor

s a
re

 th
os

e l
ab

ele
d a

s c
olu

mn
 he

ad
s i

n t
he

 co
rre

sp
on

din
g s

ce
na

rio
s. 

Fo
r e

xa
mp

le,
 in

 th
e “

Ce
re

als
” c

olu
mn

, s
ho

ck
ed

 se
cto

rs 
ar

e c
er

ea
ls.

Ac
ce

ler
at

ed
 ec

on
om

ic 
gr

ow
th

 le
d 

by



ETHIOPIA     125

whole, explains 10.3 percent of total GDP increase of 1.14 in that year (see the 
second-to-last row of Table 5.5). Insignifi cant growth-multiplier effects on non-
agricultural sectors are due to diffi culty in labor mobility between the agricultural 
and nonagricultural sectors. To create more employment opportunities in non-
agricultural sectors at a given wage rate requires improving productivity in these 
nonagricultural sectors, which was not considered in this scenario. 

We turn next to the second scenario, in which export crops are the driving force 
behind accelerated economic growth. The fi rst fi nding is that the small size of this 
subsector in the economy (4.5 percent of total GDP in the base year) implies that 
even high growth rates in export crop value-added have only modest impacts on 
national GDP. The growth rate of agricultural GDP rises from 4.2 to 4.5 percent 
per year, and that for total GDP increases from 6.1 to 6.2 percent. These modest 
effects on the overall economy indicate that, although there is considerable potential 
to expand some export-oriented crops, its expected impact on the overall economy 
would be rather limited.

The second major fi nding from this scenario is that the growth multipliers for 
export crops are smaller than for cereals. An additional 0.04 units of GDP are cre-
ated by a one unit increase in export-crop production. Moreover, the impact chan-
nels are quite different. Unlike in cereal-led growth, the nonagricultural sector is 
positively affected by export crop–led growth, but the effect on other agricultural 
sectors is insignifi cant. Results indicate that 96.5 percent of increased GDP is a 
direct result of export-crop growth. Of the remaining impact channels, only 0.33 
percent is attributable to nonexport agriculture, whereas 3.17 percent is due to 
nonagriculture. 

The third fi nding is that export crop–led growth creates more employment 
opportunities, as growth rates in both family farm and unskilled labor increase from 
baseline levels. The reason is that export crops are produced for foreign markets, and 
demand in such markets is not constrained by domestic income. Without consider-
ing other constraints, such as increased trade barriers from importing countries and 
physical and institutional diffi culties in accessing markets, productivity growth in 
export crops leads to increased demand for family and unskilled labor. Measured as 
a labor-supply–GDP elasticity, a 1 percent increase in GDP led by growth in export 
crops is associated with a 0.15 percent increase in total labor supply (in Table 5.5 in 
the fourth column in the row for “Elasticity of labor supply–GDP growth”). This 
is the highest elasticity in all of the scenarios we consider.

A comparison of the results from the fi rst two scenarios indicates that the con-
tribution of an agricultural subsector’s growth to overall economic growth is mainly 
determined by the size of the subsector. Small subsectors may have high growth poten-
tials but can only generate modest increases in overall economic growth. Furthermore, 
although the economywide multiplier effect of a subsector is important, it should 



126      XINSHEN DIAO ET AL.

not be the only indicator used to explain growth linkages between agriculture and 
nonagriculture, as well as within agriculture. Given that the Ethiopian government 
has accorded a high priority to generating employment, promoting growth in some 
sectors can create more employment opportunities, even though they may not 
generate as much economywide growth. Those subsectors that face fewer demand 
constraints from domestic markets and that use unskilled labor more intensively will 
have greater potential to create jobs at the given level of growth.

Comparing Agriculture-Led and Nonagriculture-Led Growth
The fourth and fi fth scenarios compare agricultural and nonagricultural growth. In 
reality it is impossible to have either the agricultural or nonagricultural sectors grow 
independently of each other. However, these two scenarios are designed to assess 
possible differential effects on the overall economic growth and poverty reduction. 
As shown in the fi rst two rows of Table 5.5, growth in the aggregate agricultural and 
nonagricultural sectors creates more employment opportunities both for family 
farm labor and unskilled labor, whereas nonagricultural growth leads to a greater 
increase in labor demand and therefore more employment opportunities than does 
growth in the agricultural sector alone. Measured by the labor-supply–GDP elastic-
ity, a 1 percent increase in total GDP led by nonagricultural growth is associated 
with a 0.10 percent increase in labor supply, and the elasticity is 0.06 when agricul-
tural growth is the leading force. Moreover, the growth multiplier in the case of 
nonagriculture-led growth is higher than that of agriculture-led growth (1.06 versus 
1.03). Stronger growth-multiplier effects from nonagriculture-led growth are related 
to additional capital accumulation. Nonagriculture is more capital intensive than 
agriculture. Increases in nonagricultural productivity leads to faster investment 
growth, which generates additional demand for capital goods production. Faster 
capital accumulation creates more demand for labor that can be employed by both 
agricultural and nonagricultural sectors. This demand leads to higher growth-
multiplier effects in this scenario. Decomposition of the multiplier effect shows that, 
in the agricultural growth scenario, only 3.24 percent of increased GDP (1.03 units) 
occurs in the nonagricultural sector, whereas in the nonagricultural growth scenario, 
6.08 percent of the increase in GDP (1.06 units) comes from increased agricultural 
GDP.

Demand-side linkages are examined in Table 5.6. The fi rst panel in the table 
presents annual consumption growth rates in the various scenarios. The commodity 
groups shown in the table are aggregations from the 62 groups used in the DCGE 
model, with base year prices for each individual commodity used as weights. As 
shown in the table, the growth rate in real consumption is lower than the total GDP 
growth rate. This is because the change in relative prices and different income elas-
ticities across commodities cause demand growth to vary across commodities.



ETHIOPIA     127
Ta

bl
e 5

.6—
Co

ns
um

pt
io

n 
lin

ka
ge

s i
n 

th
e s

im
ul

at
io

n 
sc

en
ar

io
s

 
Ba

se
lin

e 
 

Ca
te

go
ry

 
sc

en
ar

io
 

Ce
re

als
 

Ex
po

rt 
cr

op
s 

Li
ve

st
oc

k 
Al

l a
gr

icu
ltu

re
 

No
na

gr
icu

ltu
re

 
Co

m
bi

ne
d 

sc
en

ar
io

Gr
ow

th 
in 

GD
P 

(p
er

ce
nt)

 
6.1

 
6.3

 
6.2

 
6.3

 
6.8

 
7.2

 
7.8

Gr
ow

th 
in 

co
ns

um
pti

on
, m

ea
su

re
d b

y r
ea

l e
xp

en
dit

ur
e (

pe
rce

nt)
 

To
tal

 
5.1

 
5.4

 
5.3

 
5.4

 
6.0

 
5.6

 
6.5

 
Ce

re
als

  
4.7

 
6.0

 
4.8

 
4.8

 
6.2

 
4.9

 
6.3

 
Ex

po
rt 

cro
ps

  
4.4

 
4.5

 
4.8

 
4.5

 
5.1

 
4.7

 
5.2

 
Liv

es
toc

k  
3.8

 
3.9

 
3.9

 
5.3

 
5.5

 
3.5

 
5.2

 
Ot

he
r f

oo
ds

 
3.7

 
3.8

 
3.8

 
3.7

 
5.0

 
3.9

 
4.9

 
No

nfo
od

 
6.3

 
6.3

 
6.5

 
6.4

 
6.7

 
7.2

 
7.6

Sh
ar

e i
n i

nc
re

as
ed

 to
tal

 co
ns

um
pti

on
, 2

01
5 (

pe
rce

nt)
 

Ce
re

als
  

 
76

.4 
10

.2 
3.9

 
29

.4 
5.2

 
21

.4
 

Ex
po

rt 
cro

ps
  

 
1.5

 
20

.5 
1.6

 
6.0

 
4.6

 
4.8

 
Liv

es
toc

k  
 

3.0
 

7.8
 

73
.1 

25
.2 

-6
.8 

13
.8

 
Ot

he
r f

oo
ds

 
 

7.4
 

7.8
 

1.8
 

19
.7 

5.3
 

13
.0

 
No

nfo
od

 
  

11
.7 

53
.8 

19
.5 

19
.6 

91
.7 

47
.0

Ch
an

ge
 fr

om
 ba

se
lin

e i
n 2

01
5 (

pe
rce

nt)
 

To
tal

 
 

2.1
 

1.2
 

1.9
 

6.3
 

3.4
 

9.3
 

Ce
re

als
  

 
9.0

 
0.7

 
0.4

 
10

.2 
1.0

 
11

.1
 

Ex
po

rt 
cro

ps
  

 
0.4

 
2.8

 
0.4

 
4.3

 
1.8

 
5.2

 
Liv

es
toc

k  
 

0.5
 

0.7
 

10
.6 

12
.1 

-1
.8 

9.8
 

Ot
he

r f
oo

ds
 

 
1.1

 
0.7

 
0.3

 
9.0

 
1.3

 
8.9

 
No

nfo
od

 
  

0.5
 

1.3
 

0.8
 

2.6
 

6.6
 

9.4
El

as
tic

ity
 of

 co
ns

um
pti

on
–G

DP
,a  20

15
 

To
tal

 
 

1.2
 

1.4
 

1.3
 

1.3
 

0.5
 

0.8
 

Ce
re

als
  

 
5.3

 
0.8

 
0.3

 
2.1

 
0.1

 
0.9

 
Ex

po
rt 

cro
ps

  
 

0.2
 

3.2
 

0.2
 

0.9
 

0.3
 

0.4
 

Liv
es

toc
k  

 
0.3

 
0.8

 
7.2

 
2.5

 
–0

.3 
0.8

 
Ot

he
r f

oo
ds

 
 

0.7
 

0.8
 

0.2
 

1.8
 

0.2
 

0.8
 

No
nfo

od
 

  
0.3

 
1.6

 
0.5

 
0.5

 
0.9

 
0.8

So
ur

ce
: E

thi
op

ian
 dy

na
mi

c c
om

pu
tab

le 
ge

ne
ra

l e
qu

ilib
riu

m 
mo

de
l re

su
lts

.
No

te:
 G

DP
 =

 gr
os

s d
om

es
tic

 pr
od

uc
t.

a El
as

tic
ity

 of
 co

ns
um

pti
on

–G
DP

 is
 de

fin
ed

 as
 th

e r
ati

o o
f th

e p
er

ce
nta

ge
 in

cre
as

e i
n c

on
su

mp
tio

n f
ro

m 
ba

se
lin

e t
o p

er
ce

nta
ge

 in
cre

as
e i

n G
DP

 fr
om

 ba
se

lin
e, 

bo
th 

for
 20

15
.

  
Ac

ce
ler

at
ed

 ec
on

om
ic 

gr
ow

th
 le

d 
by



128      XINSHEN DIAO ET AL.

Comparing consumption growth rates for the fi rst two scenarios reveals that 
consumption for a group of commodities grows more rapidly when the group is 
targeted in the scenario. For example, the growth rate of cereal consumption in-
creases from 4.7 to 6.0 percent in the cereal-led growth scenario. However, in the 
same scenario, livestock consumption growth only increases slightly. In contrast, 
when growth is led by additional livestock productivity, livestock consumption 
grows more rapidly, while cereal consumption growth rate rises only slightly. This 
differential demand effect by commodity groups is due to relative prices. Accelerating 
growth in a particular group of commodities causes its prices to fall relative to other 
commodity groups. This growth generates additional incomes for the farmers 
engaged in these subsectors, but increased consumption stemming from direct 
income effects is not limited to those products for which supply increases. On the 
contrary, lower prices induce more across-the-board consumption by both rural and 
urban households. When accelerated growth occurs across the whole economy, rela-
tive price movements become smaller, and growth in consumption is dominated by 
income effects. Thus, growth differences across commodity groups also decrease.

The second panel in Table 5.6 reports the contribution of each commodity 
group to the increase in total consumption. It shows that when accelerated growth 
is led by cereals, then 76.4 percent of increased total consumption is from cereal 
consumption itself. When growth is led by the livestock subsectors, then live-
stock consumption explains 73.1 percent of total consumption increases. More 
importantly, when growth is led by export crops then 53.8 percent of consumption 
increases are from nonagricultural goods. Such differential consumption outcomes 
will have effects on poverty reduction. For a low-income country like Ethiopia, most 
of the poor population lives in rural areas, where food consumption, particularly 
staple foods, is the most important component of household total expenditure 
baskets (including expenditure in kind). When economic growth increases income 
among rural low-income households, then staple consumption grows more rapidly 
than consumption of other commodities. This differential occurs because rural poor 
households have higher income elasticities for food demand than do urban and 
nonpoor rural households. In the poverty–growth linkage analysis below, we show 
that agriculture-led growth is more pro-poor than growth led by the nonagricultural 
sector. One important reason for this is due to the different effects on consumption 
growth across different scenarios shown in Table 5.6.

The fi nal panel of Table 5.6 shows the ratio of changes in consumption to 
changes in total GDP, which can be used to compare the marginal effect of growth 
led by different subsectors on consumption of the various commodity groups. The 
table shows that the consumption–growth elasticity is higher under agriculture-led 
growth than under nonagriculture-led growth. This trend is particularly evident for 



ETHIOPIA     129

food consumption and further emphasizes the importance of consumption linkages 
when growth is led by the agricultural sector.

To further examine price effects, Table 5.7 presents the change in agricultural 
product prices. The table shows that prices fall by 8–16 percent for the fi ve cereal 
products under cereal-led growth, whereas prices fall by 9–11 percent for the live-
stock products under livestock-led growth. When growth accelerates in both agri-
cultural and nonagricultural sectors, then the negative price effects on agricultural 
products become much smaller. In fact, prices even rise slightly for certain products. 
This result indicates that agricultural policy should avoid favoring only a few 
selected crops or livestock products, because there is potential to cause prices for 
these products to fall so much that farmers will benefi t little in terms of income 
generation. The results also reveal the importance of concurrent growth in agricul-

Table 5.7—Deviation in marketed commodity prices from baseline, 2015 (baseline = 1)

Scenario Teff Barley Wheat Maize Sorghum Oilseeds Enset
Cereals 0.92 0.87 0.84 0.85 0.90 1.00 0.97
Export crops 1.01 1.01 1.01 1.01 1.01 0.99 1.00
Livestock 1.02 1.02 1.02 1.03 1.03 1.02 1.01
All agriculture 0.96 0.91 0.88 0.89 0.95 1.00 1.00
Nonagriculture 1.06 1.06 1.06 1.07 1.07 1.05 1.05
All sectors 1.02 0.97 0.93 0.96 1.02 1.06 1.07

        Cut
Scenario  Vegetable  Fruits  Sugar  Tea  Cotton  Coffee flowers
Cereals 1.01 1.01 0.98 1.01 0.99 1.00 1.00
Export crops 1.02 1.00 0.84 0.91 0.68 0.96 0.99
Livestock 1.03 1.04 0.99 1.03 0.98 1.02 1.02
All agriculture 1.04 1.04 0.79 0.95 0.67 0.97 1.01
Nonagriculture 1.08 1.21 1.61 1.15 1.60 1.05 1.03
All sectors 1.13 1.26 1.44 1.10 1.04 1.02 1.04
       
     Meat and Dairy Milled Other
Scenario  Cattle  Poultry  Milk oil products products  grains  foods 
Cereals 1.00 1.00 1.00 1.00 1.00 0.99 1.00
Export crops 1.01 1.01 1.01 0.98 1.01 1.00 0.98
Livestock 0.90 0.91 0.89 1.03 0.90 1.02 1.03
All agriculture 0.92 0.92 0.91 1.01 0.91 1.02 1.01
Nonagriculture 1.09 1.09 1.09 0.99 1.09 1.00 0.98
All sectors 1.00 1.01 0.99 1.00 0.99 1.03 0.99

Source: Ethiopian dynamic computable general equilibrium model results.



130      XINSHEN DIAO ET AL.

tural and nonagricultural sectors for stabilizing agricultural prices—an important 
linkage effect that can only be observed in a general equilibrium analysis.

Decomposing Agricultural Growth’s Impact on Poverty
Accelerating economic growth is often the primary focus of policy interventions in 
low-income countries like Ethiopia, because only rapid growth can enable the 
country to catch up with the rest of the world. The previous subsection showed the 
importance of agriculture in generating overall growth and demonstrated that this 
importance is not only because of the size of agriculture. Rather, the importance of 
agriculture is also due to its strong linkages to the rest of the economy, particular 
consumption linkages, which allow growth to directly benefi t a majority of the popu-
lation and increase consumption levels. In this subsection, we focus on the impact 
of growth on poverty reduction, one of the most important goals for all developing 
countries. 

Although growth is generally good for the poor, opportunities for them to 
participate in growth varies, depending on the type of growth. The growth–poverty 
linkage analysis addresses this issue and measures what kind of growth creates more 
job opportunities for the poor. We use similar scenarios to those from the growth 
linkage analysis to evaluate the linkages between growth led by different subsectors 
and poverty reduction. Although income distribution infl uences the effectiveness 
of growth on poverty reduction, we are unable to analyze its effect in the current 
model structure. That is to say, we generally assume that the income distribution, 
particularly the distribution in each representative household group, does not 
change over time for different types of growth. Although this caveat of the model 
may lead to an overestimation of growth’s contribution to poverty reduction, it will 
not be a problem when the focus of the analysis is to compare growth–poverty link-
ages across different scenarios designed in the model. We explain this argument 
below.

To compare the effectiveness of growth led by different subsectors on poverty 
reduction, it is important to make sure scenarios are comparable. Because of the 
different sizes of the subsectors, the same growth rate at the subsector level can have 
very different impacts on national economic growth and poverty reduction. It is 
unlikely that a small subsector will cause large reductions in poverty, even if this 
small sector grows rapidly. To address this comparability issue, we focus on the link-
age between overall economic growth and poverty reduction. The poverty–growth 
elasticity is calculated using the same formula as before (the percentage change in 
the national poverty rate caused by 1 percent growth in per capita GDP). 

According to 2004/05 HICES (Ethiopia, CSA 2005), the country’s reported 
national poverty rate was 38.7 percent in 2005. This poverty rate is consistent with 
the standard US$1.25 per capita per day international measure of poverty. As 



ETHIOPIA     131

expected, the rural poverty rate is similar to the national poverty rate, because more 
than 80 percent of the population lives in rural areas. The rural poverty rate is 39.3 
percent, and the urban rate is slightly lower (35.1 percent). 

It is widely anticipated that Ethiopia is unlikely to meet the fi rst Millennium 
Development Goal of halving poverty by 2015, which requires the poverty rate fall 
to 23 percent by 2015. We fi rst use the Ethiopian model to simulate baseline poverty 
reduction, consistent with the baseline growth simulation. According to our results, 
the national poverty rate will fall to 28.6 percent by 2015 (Table 5.8). Obviously, 
Ethiopia needs to not only accelerate the level of growth but also fi nd ways to en-
hance the benefi ts of growth for its poor. In other words, it is important to identify 
the composition of growth that is most effective at reducing poverty. In this regard, 
it is necessary to consider the relative importance of agriculture and industry in 
helping the country signifi cantly reduce poverty.

Using the same six scenarios discussed above, the growth–poverty linkages are 
used to compare the effectiveness of additional growth in reducing poverty under 
different growth assumptions. The poverty–growth elasticity is calculated for each 
scenario based on growth in total GDP rather than on the sector’s GDP. Table 5.8 
shows the poverty rate by 2015 in each scenario. For example, the national rate falls 
to 28.6 percent in 2015 in the model’s baseline, and the rural and urban poverty 
rates fall to 29.6 and 23.5 percent, respectively. With one exception, all poverty rates 
fall more in the six accelerated growth scenarios. The national poverty rate falls 
to 23.9 percent in 2015 in the sixth growth scenario (all sectors combined)—a 
poverty rate that nearly achieves the country’s fi rst Millennium Development Goal. 

The second panel of Table 5.8 reports the additional number of people lifted 
above the poverty line by 2015 in each growth scenario. About 25 million people 
were living below the poverty line in 2005. The baseline scenario lowers this number 
to 24 million. Considering that the population will grow by 19.5 to 20 million over 
the 10-year simulation period (with a 2.7 percent population growth rate), the 
decline in the absolute number of poor under the baseline seems to be an optimistic 
scenario. The model results also indicate that about 60 percent of the reduction in 
the number of poor occurs in rural areas (not shown in the table). The model there-
fore reports more optimistic poverty reduction in urban areas, where the poverty 
rate falls to 23.5 percent by 2015—an 11.6 percentage point reduction from its level 
in 2005. The fi nal row of the table reports poverty–growth elasticity, which was 1.13 
in the baseline. We have mentioned in previous subsections that this elasticity is 
higher than that calculated using historical data, but it is comparable to those of 
other countries derived using similar methods. 

We now assess how different sources of growth affect poverty reduction rates dif-
ferentially. We start with the agriculture-led and nonagriculture-led scenarios. As dis-
cussed above, the annual GDP growth rate is 6.8 and 7.2 percent in the agriculture-led 



132      XINSHEN DIAO ET AL.

Ta
bl

e 5
.8—

Ch
an

ge
s i

n 
po

ve
rty

 ra
te

s a
nd

 p
ov

er
ty

–g
ro

wt
h 

ela
st

ici
tie

s

 
Ba

se
 ye

ar
 

Ba
se

lin
e 

 
Ex

po
rt 

 
Al

l 
 

Co
m

bi
ne

d
Ca

te
go

ry
 

va
lu

e 
sc

en
ar

io
 

Ce
re

als
 

cr
op

s 
Li

ve
st

oc
k 

ag
ric

ul
tu

re
 

No
na

gr
icu

ltu
re

 
sc

en
ar

io
Po

ve
rty

 ra
te 

by
 20

15
 (p

er
ce

nt)
 

Na
tio

na
l 

38
.7 

28
.6 

26
.1 

28
.1 

27
.9 

24
.8 

27
.0 

23
.9

 
Ru

ra
l 

39
.3 

29
.6 

27
.0 

29
.2 

29
.0 

25
.6 

27
.5 

24
.4

 
Ur

ba
n 

35
.1 

23
.5 

22
.1 

22
.9 

22
.5 

20
.7 

24
.5 

21
.5

Ad
dit

ion
al 

pe
op

le 
lift

ed
 ou

t o
f p

ov
er

ty 
by

 20
15

 re
lat

ive
 to

 ba
se

lin
e (

tho
us

an
d)

 
Na

tio
na

l 
 

 
2,0

75
 

42
2 

60
0 

3,1
94

 
1,3

57
 

3,9
22

 
Ru

ra
l 

 
 

1,8
71

 
32

8 
46

0 
2,7

99
 

1,5
02

 
3,6

29
 

Ur
ba

n 
 

 
20

4 
94

 
14

0 
39

5 
–1

45
 

29
4

Sh
ar

e o
f r

ed
uc

tio
n i

n a
bs

olu
te 

nu
mb

er
s o

f th
e p

oo
r b

y 2
01

5 (
pe

rce
nt)

 
Ru

ra
l 

 
 

90
.2 

77
.8 

76
.6 

87
.6 

11
0.6

 
92

.5
 

Ur
ba

n 
 

 
9.8

 
22

.2 
23

.4 
12

.4 
–1

0.6
 

7.5
Po

ve
rty

–g
ro

wt
h e

las
tic

ity
 

–1
.13

 
–1

.40
 

–1
.16

 
–1

.16
 

–1
.41

 
–1

.05
 

–1
.24

So
ur

ce
: E

thi
op

ian
 dy

na
mi

c c
om

pu
tab

le 
ge

ne
ra

l e
qu

ilib
riu

m 
mo

de
l re

su
lts

.
No

tes
: T

he
 po

ve
rty

 ra
te 

is 
the

 pr
op

or
tio

n o
f th

e p
op

ula
tio

n w
ith

 pe
r c

ap
ita

 co
ns

um
pti

on
 be

low
 th

e p
ov

er
ty 

lin
e. 

Bl
an

k c
ell

s =
 no

t a
pp

lic
ab

le.



ETHIOPIA     133

and nonagriculture-led scenarios, respectively (see Table 5.6). Although overall 
economic growth is higher in the nonagriculture-led scenario, more poverty reduc-
tion occurs in the agriculture-led scenario. As indicated in the fi rst row of Table 
5.8, the national poverty rate falls to 24.8 percent by 2015 in the agriculture-led 
scenario, but it remains at 27 percent in the nonagriculture-led one. A different way 
to present these differential effects of growth on poverty is to calculate the number 
of additional poor people lifted above the poverty line. The comparison shows that 
the agriculture-led scenario lifts 1.8 million more people out of poverty than does 
the nonagriculture-led scenario. A more accurate comparison is to compare poverty–
growth elasticities. The elasticity is –1.41 and –1.05 in the agriculture-led and 
nonagriculture-led growth scenarios, respectively. Economic growth driven by agri-
culture is therefore far more pro-poor than nonagricultural growth.

For nonagriculture-led growth, the poverty rate in urban areas rises by 2015 
from its baseline level in the same year. To understand this counterintuitive result, 
we compare changes in real disposable income, measured by total expenditure in 
real terms (that is, at the same level as base-year prices), for poor and nonpoor house-
holds in rural and urban areas. Table 5.9 shows that for nonagriculture-led growth, 
real incomes for poor urban households fall from their baseline levels in 2015, while 
those of rural poor households increase. Without simultaneous growth in the agri-
cultural sector, agricultural product prices rise in this scenario (fourth row in Table 
5.7). If the increase in nominal incomes for some poorer urban households does not 
keep up with rising food prices, then their real incomes will fall. This reduction leads 
to declining real consumption and causes some poor urban households, which were 
lifted out of poverty in the baseline, to now fall back into poverty. 

This result has important policy implications for the promotion of non-
agricultural growth. Ensuring the participation of the poor is important for any 
growth strategy, but it is equally important to pay attention to possible food price 
increases during the growth process. This perverse growth outcome has been recent-
ly observed in many developing countries where high nonagricultural growth has 

Table 5.9—Change in total real expenditure from baseline, 2015 (percent)

  Export  All  Combined
Category Cereals crops Livestock agriculture Nonagriculture scenario
Total 2.11 1.16 1.90 6.26 3.38 9.32
Rural poor 2.97 0.81 1.85 6.90 4.05 10.67
Rural nonpoor 2.17 1.47 1.73 6.95 4.60 11.21
Urban poor 1.89 1.03 2.81 6.34 –1.38 5.15
Urban nonpoor 0.90 0.76 2.18 3.41 0.49 3.46

Source: Ethiopian dynamic computable general equilibrium model results.



134      XINSHEN DIAO ET AL.

been accompanied by food price infl ation. Ethiopia appears to have shown this same 
outcome during its recent growth period, which may partly explain why rapid 
growth has not signifi cantly reduced poverty in urban areas.

Conclusions
Ethiopia has embarked on an agriculture-led growth strategy to meet the challenges 
of accelerating national growth and poverty reduction. Three sets of questions 
around the importance of the agricultural sector were highlighted as being relevant 
to the implementation of this strategy. Which sectors have large prospective link-
ages? What are the growth and poverty-reduction potential of these sectors? What 
policy interventions are needed to unlock their growth potential? We have applied 
the DCGE model described in Chapter 2 to address these three policy questions for 
Ethiopia. Several key messages can be drawn from our analysis. 

First, the importance of agriculture relates to its size in the economy. Although 
the agricultural sector is unlikely to grow faster than many nonagricultural sub-
sectors, due in part to its greater dependency on nature, its size is much larger than 
the nonagricultural sector in Ethiopia. The country’s economic history shows that 
GDP growth is strongly correlated with agricultural growth. The DCGE model 
showed that agricultural growth, even when it is slower than nonagricultural growth, 
directly contributes more to overall economic growth. Moreover, certain large staple 
subsectors drive the agricultural sector as a whole. In contrast, some nontraditional 
export crop sectors may be able to grow very rapidly, but this growth will be from 
a small base and so is unlikely to translate into signifi cant sectorwide growth. For 
this reason, overall agricultural growth must primarily come from growth in the 
staple subsectors.

Second, the importance of cereal production in overall economic growth is due 
not only to its size but also to its strong multiplier effects. The model shows that 
cereal-led growth has the largest multiplier effect and is a driver of growth in other 
agricultural subsectors.

Third, there exists a certain limitation to cereal-led growth for job creation. In 
fact, productivity-led growth in the cereal sector will release labor from agriculture, 
which requires that other sectors of the economy absorb more labor. Because Ethiopia 
is a labor-abundant economy with rapid population growth, it is necessary to pay 
attention to limited job creation potential when promoting cereal production. Export 
crops, in contrast, may have smaller growth multipliers, but they create more employ-
ment opportunities, because they face less stringent demand constraints and relatively 
stronger linkages with market activities (including trade and transport).

Fourth, the growth multiplier effect is higher in the nonagricultural sector than 
it is in the agricultural sector as a whole. These enhanced multiplier effects are related 
to the current structure of Ethiopia’s nonagricultural sector and the growth assump-



ETHIOPIA     135

tions of the model. In the model’s simulation, additional nonagricultural growth 
mainly comes from manufacturing and private services. Agroprocessing is one of 
the most important components of manufacturing in Ethiopia, and private services 
are primarily for meeting domestic demand. Thus, accelerated growth in these 
nonagricultural subsectors is expected to create demand for agricultural goods, both 
through production and consumption linkages. Moreover, such nonagricultural 
growth is good for job creation, because it increases labor demand in both agricul-
tural and nonagricultural sectors. However, more private investments and higher 
growth rates in capital accumulation are the preconditions for greater multiplier 
effects and job creation in the nonagricultural sector. Incentives for the private sector 
to invest can be discouraged in a business environment that has institutional barri-
ers. These will constrain nonagricultural growth led by the private sector and hence 
limit job creation. The model did not consider these constraints, which should 
ideally be taken into consideration for Ethiopia. 

Fifth, real prices for staple crops and livestock products are expected to fall when 
growth in these agricultural subsectors is faster than growth in other sectors of the 
economy. Urban consumers as well as farmers who are net buyers of foodcrops and 
livestock products will benefi t from such declines in prices. Conversely, farmers who 
are net sellers of these products will be hurt by lower incomes. Thus, it is essential 
to pay close attention to any potential negative effects on famer income induced by 
falling prices when agricultural policies favor only a few crops or livestock products. 
The simulation results also reveal the importance of concurrent growth of agricul-
tural and nonagricultural sectors for stabilizing agricultural prices—an important 
linkage effect that can only be observed though general equilibrium analysis.

Sixth, the impact of growth on poverty is larger when economywide growth is 
driven by agriculture rather than by nonagriculture. Agriculture-led growth can lift 
1.4 times more people out of poverty than the same amount of nonagriculture-led 
growth. 

Finally, nonagriculture-led growth can become less pro-poor even in urban 
areas when such growth causes domestic food prices to rise. This result has impor-
tant policy implications when promoting nonagricultural growth. Although the 
participation of the poor in the growth process is important for promoting any form 
of growth, it is equally important to pay attention to possible food price increases 
from neglecting agriculture. This kind of growth outcome has recently been observed 
in many developing countries where high nonagricultural growth has been accom-
panied by food price infl ation. It may also explain the slowdown in poverty reduc-
tion in Ethiopia’s urban areas.

This chapter quantitatively measured growth linkages and growth–poverty 
relationships by assuming that economywide growth is driven by accelerated growth 
in different sectors or subsectors of agriculture. Nevertheless, it is not the intention 
of this analysis to prioritize particular sectors for Ethiopia’s growth strategy. Rather, 



136      XINSHEN DIAO ET AL.

our analysis reveals differential roles for economic sectors in generating both growth 
and poverty reduction and hence displays their interdependence during the develop-
ment process. 

Government policies and public investments are important conditions for 
productivity-led growth in the economy. However, they are not assessed in this 
chapter. Moreover, to stimulate the same level of economic growth across different 
subsectors may require different levels of public investment. We did not use the 
Ethiopian DCGE model for cost–benefi t assessment. Moreover, growth at the sub-
sector level is not necessarily the result of public investment. Investment by the private 
sector is often the dominant force in generating growth in the nonagricultural sector 
and in export-oriented high-value agriculture. In this case, the existence of a dynam-
ic private sector requires policy supports and appropriate institutional environ-
ments. These factors are often more important for capital accumulation than public 
sector direct intervention. 

In the agricultural sector, both better incentives and improved production 
conditions are needed to provide farmers with more opportunities to diversify and 
consequently to transition many crops currently grown for subsistence into market-
able commodities and cash income opportunities. Policies and public investments 
that will help improve agricultural production and market conditions and encour-
age farmers to improve land and labor productivity are also necessary. The main 
elements of the Ethiopian government’s approach to improving market connectivity 
have focused on the crucial areas of liberalizing markets and improving roads, espe-
cially major highways. Liberalization has improved the functioning of grain markets 
after the interventionist policies of the Derg, particularly in regions that produce 
surplus grain (World Bank 2006). The government has recognized that a dramati-
cally improved road network is a prerequisite for enhanced growth in agriculture 
and for fostering urban–rural linkages. However, transport costs for farmers and 
traders remain high, despite a signifi cant effort over the past decade on road con-
struction. This is in part because most farmers lack feeder roads to the main net-
works, and roads are long-term and costly investments that do not improve yields 
or generate positive returns quickly. However, it is broadly accepted that growth in 
the agricultural sector is constrained by the service sector, including trade, transport, 
and fi nancial and business services. Improvements in these market-related services 
will strengthen intersectoral linkages and increase growth-multiplier effects. 
Moreover, reducing marketing costs benefi ts smallholders by increasing the net 
prices they receive for their goods, thereby raising incomes. Improving market con-
ditions also creates a more effi cient trading sector, which itself can increase non-
agricultural incomes without necessarily raising costs.

Looking beyond road investments, supportive policy measures are also needed 
to develop functioning markets. Institution building and attention to risk, informa-
tion, and distortions related to food aid are also critical (Gabre-Madhin 2001). The 



ETHIOPIA     137

government’s recent emerging-market strategy has resulted in a central commodity 
exchange to address problems of market information and transaction costs and risks 
and to exploit scale economies through cooperatives (Gabre-Madhin and Goggin 
2006). However, the outcome has so far not been as positive as expected, because 
other policy and institutional barriers have constrained market participation in the 
private sector. Many policy measures in other sectors, such as extending mobile 
phone coverage to rural villages or subsidizing the purchase of radios, could also 
have substantial impacts on market responsiveness and connectivity. Similarly, 
innovative approaches are needed to help farmers meet the new challenges of par-
ticipating in global markets—particularly meeting quality and safety standards. 

In conclusion, exploiting potential growth linkages and fostering greater pov-
erty reduction and structural transformation require a diversifi ed or balanced growth 
strategy that encompasses agricultural staples and export crops. The fi ndings in this 
chapter imply that the current emphasis on ADLI is, in principle, warranted. 
However, our results also show that an exclusive focus on agriculture, and particu-
larly on a few selected agricultural subsectors, is often counterproductive. Such focus 
would at best lead to suboptimal outcomes in growth and poverty reduction. The 
greater comprehensiveness of the new PASDEP strategy suggests that policymakers 
may have learned that lesson.

Appendix

Table 5A.1—Income elasticities of demand used in the Ethiopian dynamic 
computable general equilibrium model

Category Rural Urban Category Rural Urban
Maize 0.70 0.50 Cattle meat 0.80 0.80
Wheat 1.00 0.66 Poultry 0.75 1.15
Teff 1.10 0.70 Milk 0.75 0.75
Other cereals 0.70 0.50 Other meats 0.50 0.50
Root crops 0.50 0.70 Fish 0.70 0.70
Pulses 0.75 0.60 Processed foods 0.80 0.90
Oilseeds 0.80 0.50 Beverages 0.50 0.80
Enset 0.70 0.50 Textiles 1.20 1.00
Vegetables 0.50 0.75 Other manufactures 1.20 1.00
Fruits 0.50 0.75 Construction 0.90 0.50
Sugarcane 0.75 0.60 Utilities 1.00 0.80
Chat 0.70 0.50 Trade and transport 1.00 0.80
Coffee 0.60 0.60 Restaurants 0.80 0.50
Tobacco 0.70 0.50 Other private services 1.10 1.10
   Public services 1.20 0.90

Source: Authors’ estimates based on data from the 2004/05 Household Income, Consumption and Expenditure 
Survey (Ethiopia, CSA 2005).



138      XINSHEN DIAO ET AL.

Table 5A.2—Structure of the Ethiopian social accounting matrix

Agricultural sectors Teff; barley; maize; sorghum; wheat; pulses; enset; oilseeds; vegetables; fruit crops; 
   sugarcane and sugar beets; cotton; tea; chat; coffee; tobacco; cut flowers; other 

crops; cattle; poultry; other small livestock; raw milk; forestry; fishing
Industrial sectors Coal; natural gas; other mining; meat and oilseed products; dairy products; grain mill 
   products; grain mill services; sugar and sugar confectionary; tobacco products; 

manufacturing of tea; beverages; other processed food products; cotton lint; thread 
and yarns; fiber; other textiles; wearing apparel; leather products; wood products; 
paper products; publishing; petroleum; fertilizer; chemicals, rubber, and plastic 
products; nonmetal products; metal products; motor vehicles and parts; other trans-
port equipment; electronic equipment; machinery and equipment; other manufac-
tures; electricity; water; construction

Service sectors Trade and repair services; hotels and restaurants; transport services; communication; 
   financial services; business services; real estate and renting services; recreation 

and other private services; public administration and defense; education; health
Factors Professional workers; technicians and sales workers; unskilled workers; elementary 
   occupation workers; agricultural family workers; agricultural land; livestock stocks; 

physical capital
Households Poor and nonpoor households separated into four rural agroecological zones and small 
  and large urban centers
Regions Zone 1 (moisture-reliable region); Zone 2 (moisture-unreliable enset region); Zone 3 
  (drought-prone region); Zone 4 (pastoralist region)

Source: Authors.

Note
 1. It is worth noting that maize yields in Ethiopia are higher than that in many other African 
countries. According to World Bank (2007), the average maize yield rose from 1.0 to 1.8 mt/ha 
during 1986–2000 as the result of National Maize Research Programs.

References 
Ahmed, H., A. Amogne, T. Tebekew, B. Teferra, E. Tsehaye, P. Dorosh, S. Robinson, and D. Willen-

bockel. 2009. A Regionalized Social Accounting Matrix for Ethiopia, 2005/06: Data Sources and 

Compilation Process. Addis Ababa, Ethiopia: Ethiopian Development Research Institute.

Diao, X., and A. Nin Pratt. 2007. “Growth Options and Poverty Reduction in Ethiopia: An Economy-

Wide Model Analysis.” Food Policy 32: 205–228.

Dorosh, P., and J. Thurlow. 2009. Implications of Accelerated Agricultural Growth on Household Incomes 

and Poverty in Ethiopia: A General Equilibrium Analysis. Ethiopia Strategy Support Program 2 

Discussion Paper 2. Washington, DC: International Food Policy Research Institute. 

EDRI (Ethiopian Development Research Institute). 2009. Ethiopia: Input–Output Table and Social 

Accounting Matrix. Addis Ababa, Ethiopia. www.edri.org.et/Documents/SAM%20document

%20with%20list%20of%20tables.pdf.



ETHIOPIA     139

Ethiopia, CSA (Central Statistical Agency). 2004. Report on Area and Production of Belg Season Crops 

for Private Peasant Holdings. Statistical Bulletin 302. Addis Ababa, Ethiopia: Federal Democratic 

Republic of Ethiopia.

———. 2005. Household Income, Consumption, and Expenditure Survey 2004/2005. Addis Ababa, 

Ethiopia: Federal Democratic Republic of Ethiopia.

———. 2008a. Agricultural Sample Survey 2007/2008 (2000 E.C.): Volume I—Report on Area and 

Production Crops (Private Peasant Holdings, Meher Season). Statistical Bulletin 417. Addis Ababa, 

Ethiopia: Federal Democratic Republic of Ethiopia.

———. 2008b. Agricultural Sample Survey 2007/2008 (2000 E.C.): Volume III—Report on Farm 

Management Practices (Private Peasant Holdings, Meher Season). Statistical Bulletin 417. Addis 

Ababa, Ethiopia: Federal Democratic Republic of Ethiopia.

Ethiopia, CSA (Central Statistical Agency), EDRI (Ethiopian Development Research Institute), and 

IFPRI (International Food Policy Research Institute). 2010. The Population and Housing Census 

Atlas of Ethiopia 2007. Addis Ababa, Ethiopia; Washington, DC. www.csa.gov.et/pdf/atlas%20

cencus%202007.pdf. 

Ethiopian Roads Transport Authority. 2006. “Ethiopian Road Network.” Dataset. Addis Ababa, 

Ethiopia.

Gabre-Madhin, E. Z. 2001. Market Institutions, Transaction Costs, and Social Capital in the Ethiopian Grain 

Market. Research Report 124. Washington, DC: International Food Policy Research Institute.

Gabre-Madhin, E. Z., and I. Goggin. 2006. “Does Ethiopia Need a Commodity Exchange? An 

Integrated Approach to Market Development.” Paper presented at the Ethiopia Strategy Support 

Program Conference, Addis Ababa, Ethiopia, June 6–8.

MOFED (Ministry of Finance and Economic Development ). 2002. Ethiopia: Sustainable Development 

and Poverty Reduction Program. Addis Ababa, Ethiopia: Federal Democratic Republic of Ethiopia.

———. 2005. Ethiopia: Building on Progress: A Plan for Accelerated and Sustained Development to End 

Poverty (PASDEP) (2005/06–2009/10). Addis Ababa, Ethiopia: Federal Democratic Republic of 

Ethiopia.

Taffesse, A. S. 2007. “Product Flows and Growth Linkages in Ethiopia—A Note.” In Fostering New 

Development Pathways: Harnessing Rural-Urban Linkages (RUL) to Reduce Poverty and Improve 

Environment in the Highlands of Ethiopia: Proceedings of a Planning Workshop on Thematic Research 

Area of the Global Mountain Program (GMP) Held in Addis Ababa, Ethiopia, August 29–30, 2006, 

edited by Gete Zeleke, Peter Trutmann, Aster Denekew, 141–156. Lima, Peru; Addis Ababa, 

Ethiopia: Global Mountain Program.

World Bank. 2006. Ethiopia: Policies for Pro-Poor Agricultural Growth. Washington, DC.

———. 2007. Ethiopia: Accelerating Equitable Growth—Country Economic Memorandum. Washington, 

DC.

———. 2011. World Development Indicators. Washington, DC.