Content uploaded by Stein Terje Holden
Author content
All content in this area was uploaded by Stein Terje Holden
Content may be subject to copyright.
Economic Reforms and Soil Degradation
in the Ethiopian Highlands:
A Micro CGE Model with Transaction Costs
Stein Holden
Department of Economics and Social Sciences
Agricultural University of Norway
P.O.Box 5033, 1432 ÅS, Norway
e-mail: stein.holden@umb.no
Hans Lofgren
International Food Policy Research Institute
Washington, D.C., USA
Bekele Shiferaw
ICRISAT, Nairobi, Kenya
Abstract
The model is developed for a rural economy in the Ethiopian highlands. The economy is
characterized by transaction costs in the internal markets as well as in the markets
linking the economy to the external world. The model contains a representation of the
crop-livestock system in the area through nested constant elasticity of substitution (CES)
production functions with realistic elasticities of substitution, multiple inputs and
multiple outputs. Imperfections in the local markets are captured through market specific
transaction costs and price bands for some commodities and factors, and missing
markets for others. Market imperfections cause commodity and factor prices to be
household group specific and implies that production and consumption decisions are
non-separable for each household group. The model also captures the environmental
externality related to land degradation. The simulations indicate that both output price
increase (tax reduction) and fertiliser subsidy removal may have caused an increase in
the negative environmental externality in form of more rapid decline in land productivity.
A combination of a tax on marketed surplus of agricultural commodities and a fertilizer
subsidy appears to reduce the land degradation externality. Such a policy affects
household groups differently as it leads to a welfare improvement for the wealthiest
households and a welfare reduction for the poorest households.
2
1. Introduction
Typically macro economy-wide models incorporate the agricultural sector as a pure
producer sector, while the consumption side has been kept separate (pure consumers).
Production and consumption are closely linked in typical rural economies in low-income
countries, and may be better captured by separable or non-separable agricultural
household models (Singh et al. 1986). Farm households appear to be an extremely robust
and dominant decision-making unit in relation to production, natural resource
management and consumption in rural economies in Africa. Farm household modules
that represent groups of farm households with common characteristics are therefore
cornerstones of village economy models.
The links between macro policy changes and rural micro economies and the environment
may be complex. The main links between the economy and the environment in
agriculture-based poor rural economies go through the agricultural production activities
of farm households. Natural resources are depleted in the production process, reducing
the production potential of the agroecosystem unless a sufficient amount of productivity-
raising investments are carried out. The production and investment decisions of farm
households are endogenous responses to exogenous changes in policies and other
external factors, conditioned by household and farm characteristics.
Rural economies in developing countries are characterized by significant transactions
costs and imperfect information (Hoff et al. 1993, Sadoulet and de Janvry 1995). The
resulting gap between (higher) buying prices and (lower) selling prices induce farm
households to be only partially integrated into markets. Missing markets or non-
participation in markets may cause production decisions of farm households to become
non-separable from consumption decisions (Strauss 1986; de Janvry et al. 1991).
Transaction costs cause isolation of markets and inter-spatial and inter-temporal price
variation.
Holden et al. (1998a) developed a typology of village economies and village economy
models (Figure 2) based on the size of transaction costs in relation to trade and the degree
of differentiation in asset ownership within villages. The typology indicates that it is
relevant to use village CGE models only when significant transaction costs lead to
endogenous price determination in village markets. Some differentiation is required in
order for farm households to have incentives to trade with each other given that there are
transaction costs related to local trade as well. Holden et al. (1998) found that a remote
Zambian village, in which local trade was insignificant, could be modelled as a number
of non-separable farm household models. Taylor and Adelman (1996) modelled village
economies as consisting of a number of separable farm household models. Lofgren and
Robinson (1999) developed a CGE model that is written in a mixed-complementary
format and where households, in the presence of transactions costs, endogenously choose
between participation and non-participation in markets.
The most serious environmental problem in Ethiopia is land degradation, primarily due to
soil erosion and nutrient depletion. This leads to on-site and off-site external effects. The
on-site external effects are external because of high discount rates due to market
3
imperfections, poverty, and insecure or unspecified private property rights (Holden et al.
1998b). One consequence of imperfections in markets and tenure regimes is land
degradation, manifested in declining productivity, as users lack incentives to make
sufficient investments in the land that they operate. Such degradation may be irreversible.
The net present value of this permanent productivity loss, which may be called an inter-
temporal externality (Holden and Shiferaw 2002), may be considerable. In this study we
estimate the size of this externality and assess how it is affected by various policy
reforms. More specifically, in our model crop choice and the level of fertilizer use in one
year is permitted to influence land productivity in the following year.
Removal of policy distortions has been an important step in the right direction to make
the Ethiopian economy more dynamic. However, that does not necessarily mean that all
taxes and subsidies always are bad for an economy. The environmental and poverty
impacts of the changes in output tax and input subsidy policies have not been carefully
analysed. We therefore, in the spirit of Pigou (1924), assess whether such taxes and
subsidies alone or in combination could help reduce the land degradation externality
related to agricultural production in Ethiopia. Before we have used this model to assess
the impacts of such subsidies and taxes independently (Holden and Lofgren 2005). Here
we also look at the combined effects of agricultural output taxes and fertiliser subsidies.
Sterner (2003, p. 103) considers combining policy instruments for environmental and
natural resource management a fertile area. Such a tax may be used to fund a subsidy.
The tax could also be targeted at specially land degrading crops or activities while the
subsidy is used to stimulate ameliorating activities.
We have developed a micro CGE model for three villages for a high agricultural potential
area with relatively good market access in the Ethiopian highlands. We assess the impact
of a) an output price increase of 5% (reduction in output tax due to decontrol of prices),
b) removal of fertiliser subsidies, c) combining a) and b) and d) increasing the fertiliser
subsidy from 20 to 40% in combination with an additional tax of 5% on agricultural
output (Pigouvian tax and subsidy). We assess the impacts of these simulations on
household welfare for different household groups, on production, marketed surplus,
fertiliser demand, import of other goods, and the land degradation externality.
We present a brief background on the economic policy reforms in Ethiopia in part 2,
describe the case study area in part 3, and provide a model description with emphasis on
the unique characteristics of this model in part 4. The simulation results are presented and
discussed in part 5 and some conclusions are derived in part 6.
2. Economic policy reforms in Ethiopia
Market imperfections are common in rural markets in Ethiopia. This may partly be due to
historical factors since economic policy only recently (in 1991) was changed from a
socialist, top-down planning system to a more market-friendly regime. It may take time
before more efficient markets develop. The poor infrastructure also causes transaction
costs to be high. Ethiopian land reform in 1975 resulted in an egalitarian distribution of
land among farm households. While all land is state-owned, user rights have been
allocated to individual households through the land reform in 1975 and several land
4
redistributions in the years that followed. Land sales are illegal but there are active land
rental markets. The distribution of other rural assets, most importantly livestock, is less
egalitarian and this creates incentives for trade within villages, including the renting of
land and oxen.
With the regime change in 1991 and its replacement by a more market-friendly
government, Ethiopia embarked on structural adjustment policy reforms. These reforms
included devaluation of the exchange rate, removal of fertiliser subsidies, removal of
price controls for agricultural commodities (pan-territorial pricing), and privatisation of
public enterprises.
The development strategy, called ‘Agricultural Development-Led Industrialisation,’ is
focused on the development of labour-intensive industries that rely heavily on domestic
raw materials and inputs from smallholder agriculture. The new strategy aims to
stimulate market development, competition, and efficiency. Consequently, the
Government has dissolved producer cooperatives, reduced the role of state farms,
abolished compulsory food grain quotas, and removed price controls on agricultural
commodities and domestic trade restrictions.
Like in many African countries, Ethiopia followed a pan-territorial fertiliser pricing
policy and provided subsidies to smallholder farmers. These subsidies are often blamed
for creating wrong incentives to farmers although the universality of this claim has been
questioned (Holden and Shanmugaratnam, 1995). The fertiliser subsidy in Ethiopia was
15% in 1993, 20% in 1994, 30% in 1995, and 20% in 1996. Following the devaluation in
1992, fertiliser prices increased sharply, causing a decline in fertiliser consumption in that
year. Fertiliser subsidies were therefore introduced, but later reduced and then eliminated
starting from 1997. Fertiliser use has remained low in Ethiopia after the reforms. In terms
of nutrients the average rate of fertiliser application is 7 kg/ha in Ethiopia against 9 kg/ha
for SSA, and 65 kg/ha worldwide. A new fertiliser distribution policy was introduced in
1997. It called for elimination of fertiliser subsidies and pan-territorial pricing system for
fertiliser. The involvement of the private sector in importation and distribution of
fertilisers was also stimulated.
Although dependence on rain-fed agriculture and frequent droughts continue to pose
serious concerns, most macro indicators suggest that economic performance has been
strengthened since the introduction of the reforms. On average, growth seems to have
accelerated, both in agriculture and other parts of the economy, while overall inflation
has remained moderate. Both exports and imports have grown much more rapidly than
gross domestic product (GDP), thus drastically increasing the openness of the economy.
3. Case study area
The study area consists of the Hidi, Hora Kilole and Borer Guda peasant associations in
the Ada-Liben district in Showa region in the central highlands of Ethiopia. This area is
favourably located approximately 20 km from Debre Zeit, which is near the main
highway and only 50 km east of the capital, Addis Ababa. In addition to good market
5
access, the area enjoys a high agricultural potential. Ada-Liben district is a surplus
producer of teff (Eragrostis teff), the main crop (both in terms of consumption and
market sales) and the preferred cereal among Ethiopian consumers. The production
system is an integrated crop–livestock system where oxen provide traction power for land
cultivation, and straw from grain production is the main source of animal fodder. Very
little communal land exists as most of the land has been distributed to individual
households.
Land rental markets are active, particularly given that land cannot be sold or purchased.
Usually, fixed-rent contracts are used. Livestock, most importantly oxen, are the most
important privately owned asset in the study area. The distribution of this resource is less
egalitarian and is a good indicator of household wealth. Oxen ownership is also an
important indicator of farming capacity due to the crucial role of oxen. Rental markets for
oxen are less important because of moral hazard problems in relation to oxen
management and because proper timing of ploughing is crucial on the dominant heavy
black soils in the area. In this setting, characterised by a non-egalitarian distribution of
oxen and impediments to oxen rental, households without oxen rent out much of their
land. Typically, households with two or more oxen rent in land as they have excess
ploughing capacity. Households with one ox tend to exchange oxen with other one-ox
households as a pair of oxen is needed for ploughing. Average household and farm
characteristics for different oxen ownership groups of households are presented in Table
1. In 1993/94, 25% of the households had no oxen, 17% had one ox, 34% had two oxen
and 24% had more than two.
The local economy is highly agriculturally oriented, as the diversification into non-farm
activities is limited. However the area is a net importer of unskilled labour (seasonal
demand in crop production) but exports some skilled labour.
Holden et al. (1998a) found that the subjective discount rates of farm households in the
study villages were high and that they were influenced by wealth as poorer households
had higher subjective discount rates. This indicates that households are credit constrained
and that poorer households suffer most. Credit in kind (in the form of fertiliser), was
provided in the area but this credit also appeared to be rationed during our first survey
round in 1993-94 but credit access had improved in the two later survey rounds in 1997
and 2000.
Holden and Shiferaw (2002) estimated the farm households’ willingness to pay (WTP) to
sustain land productivity in the area as almost all farm households stated that land
productivity was declining over time. They also estimated farmers’ perceived average
rates of land productivity decline. Shiferaw and Holden (1999) used the Universal Soil
Loss Equation adapted to Ethiopian conditions to estimate soil erosion in the area, and
production functions adapted from experimental studies at other locations in Ethiopia to
estimate the impact of erosion on crop yields. The resulting estimates of average rates of
land productivity decline were about twice as high as those estimated based on farmers’
judgements.
6
4. Model description
4.1. General model structure
As the starting point for our village CGE model, we use a standard CGE model
developed by Lofgren et al. (2002). The model has been applied to a large number of
countries. Its flexibility is based on two main features: a. It separates the model from the
database, making it easier to apply to model to new settings, and b. It permits the user to
choose among alternative assumptions for how factor markets and macro constraints
operate (Lofgren et al., 2002).
The modelling structure has also been adapted for village-level analysis. The general
structure of the modelling system is kept separate from the specific structure and inputs
needed for a specific economy to be modelled. Relatively few changes are therefore
needed in its general structure. The specific structure is entered through an input file or
database containing most of the relevant elements and quantitative inputs for modelling
the specific economy. One important distinction for a village model from a country
model is that a village does not have a separate currency with an exchange rate and this
has implications for how the interactions of the village with the surrounding economy are
balanced. Another distinction is that ‘the rest of the world’, export, import, and ‘foreign’
have different meanings in village and country models. For the village model these terms
refer to outside the village, be it inside or outside the country.
The database of the model consists of the SAM, elasticity data for production,
consumption and trade, and possibly physical factor quantities. A simplified picture of
the model and its building blocks is provided in Figure 1 (Lofgren et al., 2002). Vi refer
to Lofgren et al. (2002) for a detailed mathematical description of the standard CGE
model.
4.2. Specific model structure
The general model was modified to accommodate the characteristics of the economy in
the study area. These modifications included fitting a village SAM to the requirements of
the CGE model. Market imperfections (in markets for: land, labour, oxen ploughing
services, manure, crop residues, and crop outputs) in the study area caused production
and consumption-decisions of households to be non-separable. The prices for land and
oxen services became endogenous to villages (closed village market only). Shadow prices
for non-traded factors (crop residues and manure) became endogenous to each of the
household groups. Transaction costs in the local factor markets caused household group
specific factor prices for factors traded within the village to depend on whether the
household group was a net seller or net buyer of the factor in the village market.
Household group specific and general commodity accounts were used to capture
transaction costs in relation to local and external trade.
As discussed earlier, oxen ownership is a good wealth indicator and is used as a basis for
household group classification. One additional reason for this is that oxen ownership
tends to drive the participation in land rental markets. In the CGE model and the
7
underlying village SAM, the households were therefore divided into three groups on the
basis of oxen ownership. Households with two oxen or more were pooled into one group,
while households with zero and one ox were kept as separate groups.
The modelling structure is quite flexible and this has been further developed in order to
handle a relatively complex farming system (crop–livestock system) and market
imperfections induced by transactions costs. Households typically produce a variety of
crops and livestock types. Agricultural production activities may have multiple inputs and
multiple outputs. Outputs from one activity may be inputs in another activity. Typically
crop residues are used as livestock fodder, while oxen are used for land preparation, and
animal manure is used as fuel. Production technology is captured by nested, two-level
CES-production functions, allowing substitution elasticities to be different at different
levels of the nest (Figure 3). At the bottom level, substitution elasticities between oxen,
land, capital and labour may be quite low (σ = 0.2) relative to the elasticities between
land and fertilisers at the higher level of the nest (σ = 0.9). This reflects the relatively
fixed relationship between land, oxen, capital and labour in relation to land preparation
and the much higher flexibility that exists in relation to fertiliser application.
As a result of transaction costs, selling prices are typically lower than buying prices. In
the base year, each household group is a net seller, self-sufficient, or a net buyer of
various factors of production (inputs) and commodities. This also determines the choice
of price when value added is imputed to the different factors in the SAM. Markets for
crop residues and manure are missing. Manure is used for fuel (an input into the chores
activity, representing miscellaneous tasks carried out inside the households). SAM cell
values for manure are based on the nutrient contribution of manure and the cost of
nutrients if they were bought as fertiliser (for nitrogen and phosphorus). The values of
crop residues and fodder from grazing land were determined residually after subtracting
the value of labour and animal stock from the value of livestock production.
A village SAM that was structured to match the requirements of the CGE model was
constructed. The SAM is based on a survey of the economy in 1993, carried out in 1994.
The basic structure of the village SAM is presented in Table 2. Transaction costs in
relation to village exports and imports are captured by separate accounts while
transaction costs in relation to internal village trade are primarily represented by the
labour time needed to carry out the transactions. This implies a considerable expansion in
the number of rows and columns in the SAM, making it too big for reproduction here (a
copy of the SAM used in this case study can be obtained from the authors upon request).
Land productivity declines have been estimated for the area for different types of soils
and crops (Shiferaw and Holden, 1999). Information about farmers’ perceptions on the
rates of land degradation is available from Holden and Shiferaw (2002). The rate of
productivity decline is reduced by the use of fertilisers that replace lost nutrients. The
model is calibrated such that the estimated rates of productivity decline (annual mean,
1.1%) are taken as an indicator of the rate of productivity decline when no fertilisers are
used, while the average rate of fertiliser use in the area is assumed to reduce the rate of
productivity decline to the rate that farmers perceived (annual mean, 0.55%).
8
Some of the equations that capture the link between the land productivity and land
degradation are described below. The land productivity decline per unit of land is a
function of land-type (A), crop choice (C), household type (H) such that:
( , ( , )) ( , )/(100(1 / ))
ach ach MAX MAX ach ach
F A C A C F Q
ψ ψ ψ ψ β
= = + (1)
where
ach
F
is the household, land and crop specific fertiliser use,
MAX
ψ
is the maximum
land productivity decline that takes place when no fertiliser (F) is added to the land,
ach
Q
is the current output of crop type c, on land type a by household type h, and
β
is a
calibration parameter. The rate of productivity decline without fertiliser use is specified at
two levels – high and low. The actual values used for
( , )
MAX
A C
ψ
for different crop types
and land types in the model are given in Table 3. The intensity of fertiliser use
)(
ach
F
is
household group (H) specific and depends on the price of fertiliser
)(
F
P
, crop choice (C),
crop price (P
C
) and land type (A);
( , , , , ,)
ach F C
F F H P P C A
= (2)
while the fertiliser price
)(
F
P
depends on the import price (including transportation costs),
(P
FI
), and the level of subsidy (S
F
);
FFIF
SPP −=
(3)
Optimal fertiliser use is determined through the first-order conditions for the production
functions, implicit in equation (2). The first order conditions equate marginal value
products to the prices of the respective inputs. An output price change will similarly
affect the first- order conditions and affect both input use and output supply. Other things
remaining the same, reduction in the fertiliser subsidy will cause the fertiliser price to
increase, the level of fertiliser use to go down, and the level of productivity loss from
land degradation to increase. This land degradation externality (LDEXT) is aggregated
across areas (L
ach
) of different land types, crop types, and household groups, assuming
that the process is irreversible,
1
with a social discount rate
δ
and that land use and output
prices
)(
Qc
P
are constant over time:
/
ach ach Qc
H A C
LDEXT L P
ψ δ
=
(4)
Household consumption is captured by a Stone-Geary Linear Expenditure System which
can handle broad commodity groups. Leisure is one of the commodities that is included
in the model, consistent with theoretical household models but unlike typical macro CGE
models. Agricultural production for home consumption is also included in the
1
Although some soil degradation may be reversible to some degree, soil formation is a slow process and
erosion rates have been found to be more than 10 times higher than the soil formation rates in Ethiopia.
9
expenditure system. All household groups are net sellers of agricultural commodities,
indicating the importance of the area as a surplus producer of food grains. The
agricultural production activities are given in Table 5.
5. Policy simulation results
For each simulation experiment, in the following sections we present the impacts on
household welfare, crop and livestock production, village exports, fertiliser use, and
environmental externality, see Tables 4 to10.
Because of the influence of market imperfections, the impacts of specific policies vary by
land type, crop type and household group. This is a consequence of the non-separability
of production and consumption decisions, making land use and land degradation a
function of household characteristics. This has been one of the major limitations of many
of the standard CGE models, which assume that production decisions are unaffected by
poverty or equity. The model developed here demonstrates how this assumption can be
relaxed to make CGE models capture poverty–NRM linkages more effectively.
5.1. Increase in output prices (output tax reduction).
The results from a 5% increase in the prices for all agricultural outputs (reduction of
output taxation) are examined first. The real income effects for the different household
groups are shown in Table 4. The 5% price increase leads to an increase in household
real incomes of 1.4–3.6% for the different household groups. The poorest group benefited
more (3.6%) from the output price increase than any other group. This is mainly because
the price increase makes production more profitable thereby allowing them to cultivate
more of their land themselves (rent out less). This is supplemented by increased fertiliser
use and livestock production. The impacts on production activities are shown in Table 5.
An increase in output prices has a positive impact on production of most commodities
although there is some variation across commodities and household groups. The price
policy had a stronger effect on production of lowland cereals (wheat and barley). The
lion’s share of the production comes from households who own at least a pair of oxen,
but the largest % increase in production was for the poorest household group.
Perhaps a surprising result is the impact on marketed surplus from the village (Tables 6
and 10). The output price increase reduced the marketed surplus of grains because the
income effect increased the demand for self-consumption more than it stimulated local
production. The profit effect (Singh et al. 1986) and a relatively high income-elasticity
for food consumption among the poor households explain the increase in food
consumption. Market imperfections and low elasticities of substitution for inputs may
also explain this low supply response. Similar results have been found by Bardhan (1970)
and de Janvry and Kumar (1981) in parts of India, and by Dorward et al.(2004) in
Malawi.
10
The effect of output price changes on fertiliser demand is shown in Table 7. The price
increase led to increased fertiliser use for cereal crops. At the aggregate level a 10% price
change led to an almost equivalent (9.7%) increase in fertiliser demand.
5.2. Removal of fertiliser subsidies
We simulated the effects of a removal of the 20% subsidy on fertiliser that was present up
to 1997. Table 4 shows that the reduction in fertiliser subsidy reduced household incomes
by 1.6–2.3%. The strongest relative change was for the wealthiest household group
(2.3%). This is opposite to the distributional impact of output price changes. The
wealthiest benefited relatively more from fertiliser price subsidies, while output price
increase seemed to be more pro-poor.
Table 5 shows that the removal of fertiliser subsidy caused a reduction in cereal
production in most cases, while it had mainly positive effect on the growing of pulses
(legumes). This is mainly because the price hike causes a shift to crops that are less
fertiliser-intensive or encourage planting of legumes usually grown without fertilisers.
The subsidy removal also had a negative effect on livestock production because fodder
production (crop residues) became more costly. The reduction in fertiliser subsidy caused
a decrease in the marketed surplus (export from the village) of teff and an increase in the
export of other cereals and of pulses (Table 6) while overall exports were reduced (1.3%).
There was a small reduction in the export of livestock products and an increase in out-
migration and outputs from small businesses.
The removal of fertiliser subsidy caused a fall by 18-24% in the demand for fertiliser in
the different cereal production activities (Table 7). The aggregate demand decreased by
20%. The village import of other commodities was also reduced (Table 8) and overall
import was reduced by 6.3%. We see from Table 9 that the land degradation externality
increased when the fertiliser subsidy was removed. It increased more than it did due to
the 5% output price increase, 6.7% for subsidy removal vs. 3.7% for output price
increase, see Table 10, in the case of low level of land degradation and low discount rate
(3%).
The results are also consistent with the findings of Croppenstedt et al. (2003) who found
that fertiliser demand stagnated in Ethiopia after the fertiliser subsidy removal even
though credit availability was improved. Our own panel data from the study area for this
model confirm the same; increased access to credit for farm inputs did not lead to an
increase in fertiliser use from 1994 to 2001 (Holden et al. 2005). Our panel data also
confirm a declining trend in marketed surplus that may be due to land degradation and
population increase.
5.3. Combining output price increase and fertiliser subsidy removal
This is in essence what happened based on the strong pressures to reduce taxation of the
agricultural sector (Krueger et al. 1991) and based on many arguments against fertiliser
subsidies (e.g. reviewed by Holden and Shanmugartnam 1995). We see from Table 4 that
the effects of this policy have mixed effects on the different household groups. The
11
poorest household group benefited (2.0%) while the wealthiest group lost (-1.0%) and the
overall effect was negative but small (-0.6%). We see also that the effects on production
were diverse but with a reduction in teff exports (-4.1%), an increase in exports of pulses
(+15%) and other cereals than teff (+1%), see Tables 5 and 6.
The effects on fertiliser demand for the different crops were negative and in the range –
12% to –18% with an overall reduction of –16.1% (Table 7). There was also a small
increase in imports of labour and agricultural commodities.
When it comes to the environmental externality we found that it increased by 10.9% that
is more than the sum of the effects of output price increase and fertiliser subsidy
introduced separately (3.7% and 6.7%), see Table 10. We see the overall effects in Table
10, including a reduction in overall private consumption, exports from and imports to the
village economy. This gives reasons to question the benefits of this policy.
5.4. Combination of increased fertiliser subsidy and an output tax
This is an opposite (“mirror”) experiment of the previous one and implies an increase in
the fertiliser subsidy from 20% to 40% and an increase in the tax on agricultural exports
from the village of 5%. The results are not exactly opposite of the previous experiment,
however, due to the non-linearity of many relationships.
We see from Table 4 that the main beneficiaries of this policy would be the wealthiest
household group that has a pair of oxen or more while the group without oxen will be
worse off. One of the reasons for this is the effect on the local land rental market where
wealthier households rent in more land from the poorer households as a response to the
policy change.
Production effects are diverse due to the various responses in input demand and
substitutions among crops (Table 5) but village export of teff will increase by 4.9% and
export of other cereals and pulses decline by 1.8% and 16.5% but the base values of these
are much lower than that of teff. The overall effect on export of livestock products was
very small and negative while livestock production increased (Tables 6 and 5).
The responses in fertiliser demand range from 19% to 29% across cereal crops with an
overall increase of 25.2% (Table 7) while there were small changes in other imports.
The most interesting result is that the negative environmental externality was reduced by
11.9% with this combination of output tax and input subsidy when the externality is
assumed to be low and we use a low social discount rate (3%). The overall effects on the
village economy were increased private consumption, increased village exports and
imports.
12
6. Conclusions
We have developed a micro CGE model that captures essential structures of a low-
income natural resource based rural economy. Such an economy is characterised by
substantial transaction costs that affect the functioning of internal and external markets.
The basic decision-making units are farm households that are both producers and
consumers and market imperfections cause their production and consumption decisions to
be interdependent. To capture the social differentiation of the economy farm households
are grouped in homogenous producer-consumer households. This basic structural
difference as compared to standard CGE models may lead to household price responses
that may go even in opposite direction of those of pure producers and pure consumers
(Singh et al. 1986; de Janvry et al. 1991). Our simulations demonstrate such effects.
Our model simulations for a village economy with high agricultural potential and fairly
good market access in the Ethiopian highlands indicate that both the output price increase
and the removal of fertiliser subsidies that were implemented in the late 1990s lead to
more rapid land degradation.
The blanket recommendation that all input subsidies are bad may not always hold. The
environmental economics policy perspective, drawing on Pigou, may be worth
considering in relation to environmental externalities. We have looked at the case of a
village economy in Ethiopia where agricultural production, as currently practiced, has a
negative effect on future land productivity. Fertiliser use may counteract the negative
effects of nutrient depletion and soil erosion. Imposing blindly a “polluter pays principle”
through an output tax may be questioned, however, as it would enhance poverty. The
innovative idea here is to impose a tax on agricultural output that may be ploughed back
into the economy to stimulate fertiliser use. Such a policy is shown to both have a less
negative effect on poverty and a stronger positive effect on the negative land degradation
externality. Still, we found that such a policy may have a negative effect on the poorest
household group without oxen.
We have to state that this kind of policy is insufficient to deal with the land degradation
and poverty problems in the Ethiopian highlands. There is in addition an urgent need for
an “Organic Green Revolution” that stimulates adoption of productivity-enhancing
conservation methods like reduced tillage, water harvesting and conservation, improved
fodder crops, intercropping of legumes and cereals, cover crops, and fodder trees in
combination with improved cattle and animal management systems (cut and carry
system).
References
Angelsen, A., van Soest, D., Kaimowitz, D. and Bulte, E. (2001) Technical Change and
Deforestation: a Theoretical Overview. In A. Angelsen and D. Kaimowitz (Eds.)
Agricultural Technologies and Tropical Deforestation. CABI Publishing, New York. Pp.
19-22.
13
Bardhan, K. (1970) Price and Output Response of Marketed Surplus of Foodgrains: A
Cross-Sectional Study of Some North Indian Villages. American Journal of Agricultural
Economics 52: 51-61.
de Janvry, A. And Kumar, P. (1981). The Transmission of Cost Inflation in Agriculture
with Subsistence Production: A Case Study in Northern India. Indian Journal of
Agricultural Economics 36: 1-14.
de Janvry, A., Fafchamps, M. and Sadoulet, E. (1991). Peasant Household Behaviour
with Missing Markets: Some Paradoxes Explained. Economic Journal 101: 1400-1417.
FAO (1986). The Highlands Reclamation Study – Ethiopia. Final Report. Volumes 1 and
2. Food and Agriculture Organization of the United Nations, Rome.
Croppenstedt, A., Demeke, M. and Meschi, M. M. (2003). Technology Adoption in the
Presence of Constraints: the Case of Fertilizer Demand in Ethiopia. Review of
Development Economics 7, 1, 58-70.
Dorward, A., Morrison, J., Robinson, S. and Tchale, H. (2004). Differential Household
Responses in a Poor Agricultural Economy: Implications for Agricultural Policy
Analysis. Unpublished.
Greenwald, and Stiglitz, J. E. (1986). Externalities in Economies with Imperfect
Information and Incomplete Markets. Quarterly Journal of Economics 101, 229-264.
Grepperud, S. (1996). Population pressure and land degradation: The case of Ethiopia.
Journal of Environmental Economics and Management 30: 18-33.
Holden, S. and Lofgren, H. (2005). Assessing the Impacts of Natural Resource
Management Policy Interventions with a Village General Equilibrium Model. In B.
Shiferaw, H. Ade Freeman and S. Swinton (eds.), Natural Resource Management in
Agriculture: Methods for Assessing Economic and Environmental Impacts. CABI
Publishing, pp. 295-318.
Hoff, K., Braverman, A. and Stiglitz, J. E. (eds.) (1993). The Economics of Rural
Organization. Theory, Practice, and Policy. Oxford University Press, Oxford. 590 pp.
Holden, S. T. and Shanmugaratnam, N. (1995). Structural adjustment, production
subsidies and sustainable land use. Forum for Development Studies 2: 247-266.
Holden, S. T. and Shiferaw, B. (2002). Poverty and Land Degradation: Peasants’
Willingness to Pay to Sustain Land Productivity. In C. B. Barrett, F. M. Place, and A. A.
Aboud (eds.), The Adoption of Natural Resource Management Practices: Improving
Sustainable Agricultural Production in Sub-Saharan Africa. CABI Publishing, New
York. pp 91-102.
14
Holden, S. T., Shiferaw, B. and Wik, M. (1998b). Poverty, Market Imperfections, and
Time Preferences: Of Relevance for Environmental Policy? Environment and
Development Economics 3: 105-130.
Holden, S. T., Taylor, J. E., and Hampton, S.(1998a). Structural Adjustment and Market
Imperfections: A Stylized Village Economy-Wide Model with Nonseparable Farm
Households. Environment and Development Economics 4: 69-87.
Krueger, A. O., Schiff, M. and Valdés, A. (1991). The Political Economy of Agricultural
Pricing Policy. Volume 3: Africa and the Mediterranean. The Johns Hopkins University
Press, Baltimore and London. 340 pp.
Lofgren, H. and Robinson, S. (1999). Nonseparable Farm Household Decisions in a
Computable General Equilibrium Model. American Journal of Agricultural Economics
81: 663-670.
Lofgren, H., Harris, R. L., and Robinson, S. (2002). A Standard Computable General
Equilibrium (CGE) Model in GAMS. Microcomputers in Policy Research, Vol. 5. IFPRI,
Washington, D.C.
Pigou, A. C. (1924). The Economics of Welfare. 2
nd
Edition. Macmillan, London.
Sadoulet, E. and de Janvry, A. (1995). Quantitative Development Policy Analysis. The
John Hopkins University Press, Baltimore. 397 pp
Shiferaw, B. and Holden, S. T. (1999). Soil Erosion and Smallholders’ Conservation
Decisions in the Highlands of Ethiopia. World Development 27(4): 739-752.
Shiferaw, B. and Holden, S. T. (2000). Policy Instruments for Sustainable Land
Management: The Case of Highland Smallholders in Ethiopia. Agricultural Economics
22: 217-232.
Sterner, T. (2003). Policy Instruments for Environmental and Natural Resource
Management. Resources For The Future, The World Bank and SIDA, Washington D.C..
Stiglitz, J. E.(1986). The New Development Economics. World Development, 14 (2),
257-265.
Stoorvogel, J. J. and Smaling, E. M. A.(1990) Assessment of Nutrient Depletion in Sub-
Saharan Africa:1983-2000. Volume I: Main Report. The Winand Staring Centre,
Wageningen.
Strauss, J. (1986). The Theory of Comparative Statics of Agricultural Household Models:
A General Approach. In Singh, I., Squire, L. and Strauss, J. (eds.) (1986). Agricultural
Household Models. Extensions, Applications, and Policy. The John Hopkins University
Press, Washington D. C. pp 71-91.
15
Taylor, J. E. and Adelman, I. (1996). Village Economies: the Design, Estimation, and
Use of Villagewide Economic Models. Cambridge University Press, Cambridge and New
York. 264 pp.
World Bank (1996). Towards Environmentally Sustainable Development in Sub-Saharan
Africa. A World Bank Agenda. The World Bank, Washington, D.C.
Table 1. Basic farm household characteristics in the survey area of Ethiopia, 1993
Variable Household category by number of oxen
0 1 2 >2 All
Share of population (%) 25 17 34 24 100
Female headed households (%) 27 7 7 3 11
Farm size (kert)
a
4.48 7.4 8.18 11.25 7.83
Total income (Birr)
b
2 992 4 893 5 792 12 279 6 489
Male work force (adult equivalents) 0.71 1.34 1.57 2.84 1.62
Female work force (adult equivalents) 0.91 1.01 1.12 1.68 1.18
Consumer units (adult equivalents) 2.47 3.76 4.12 6.47 4.2
Tropical Livestock Units 0.31 2.46 4.46 9.12 4.09
16
Table 2. A stylized SAM for the village economy
Activities Commodities Factors Households Enterprises Government Savings-
Investment Transaction
costs Rest of the
World TOTAL
Activities outputs activity
income
(gross
output)
Commodities
intermediate
inputs private
consumption government
consumption investment Transaction
costs exports demand
Factors value-added
transaction
costs
factor
income
from RoW
factor
income
Households factor
income to
households
surplus to
households transfers to
households transfers to
households
from RoW
household
income
Enterprises factor
income to
enterprises
transfers to
enterprises transfers to
enterprises
from RoW
enterprise
income
Government producer
taxes,
value-added
tax
sales taxes factor
income to
government,
factor taxes
transfers to
government,
direct
household
taxes
surplus to
government,
direct
enterprise
taxes
transfers to
government
from RoW
government
income
Savings-
Investment household
savings enterprise
savings government
savings foreign
savings savings
Transaction
costs Transaction
costs Transaction
costs
Rest of the
World
(RoW)
imports factor
income to
RoW
surplus to
RoW government
transfers to
RoW
foreign
exchange
outflow
TOTAL activity
expenditures
supply factor
expenditures household
expenditures enterprise
expenditures government
expenditures investment Transaction
costs foreign
exchange
inflow
17
Table 3. Maximum land productivity decline rates used in the model
a
Level of land degradation, annual % yield decline Land and crop type High Low
Upland, teff 4.1 1.2
Lowland, teff 0.38 0.25
Upland, other cereals 3.5 0.9
Lowland, other cereals 0.3 0.1
Upland, pulses 3.5 0.5
Lowland, pulses 0 0
a
Based on Shiferaw and Holden (1999, 2000) and Holden and Shiferaw (2002).
18
Table 4. The impact of alternative policies on real household incomes by household
group
a
Household group Base
results Output
price
+5%
(a)
Fertiliser
subsidy
Removal
-20%
(b)
Combine
(a)+(b)
Fertiliser
subsidy
+20%
Output
price –5%
Without oxen (H0)
263.4 3.6 -1.6 2.0 -1.8
With one ox (H1)
281.1 2.5 -2.0 0.5 -0.3
With two or more
oxen (H2) 2972.1 1.4 -2.3 -1.0 1.4
Total 3516.5 1.6 -2.2 -0.6 1.0
a
Base results in ‘000 Ethiopian Birr ; non-base simulation results as % change from base results.
19
Table 5. The impact of alternative policies on land use and crop and livestock
production activities for the three household groups
a
defined in Table 13.3.
Production activities by
household group Base
results Output price
+5%
(a)
Fertiliser
subsidy
Removal
-20%
(b)
Combine
(a)+(b)
Fertiliser
subsidy +20%
Output price
–5%
Upland teff
H0
36.5
2.9
-2.7
0.2
0.1
H1 99.5 0.8 -4.3 -3.4 4.3
H2 1098.7 0.6 -3.3 -2.6 3.3
Lowland teff
H0
30.2
2.9
-2.5
0.4
-0.1
H1 52.0 -0.4 -4.8 -5.0 6.2
H2 823.9 0.4 -3.0 -2.5 3.1
Upland other cereals
H0
8.5
1.6
0.2
1.7
-2.0
H1 19.8 0.8 0.3 1.0 -1.3
H2 193.9 0.3 -1.0 -0.7 0.8
Lowland other cereals
H0
5.6
4.0
-4.9
-0.7
1.5
H1 7.4 3.4 -6.5 -3.0 4.0
H2 53.2 3.8 -7.5 -3.9 5.5
Upland pulses
H0
2.9
0.9
4.3
5.3
-5.8
H1 9.5 -1.3 6.6 4.8 -5.2
H2 108.6 1.1 -0.5 0.5 -0.2
Lowland pulses
H0
6.2
1.1
1.1
1.9
-2.1
H1 12.5 -0.2 6.6 5.9 -6.2
H2 147.9 0.0 1.1 1.0 -0.9
Livestock
H0
8.2
2.3
-1.7
0.6
-0.4
H1 53.0 0.7 -1.4 -0.7 0.8
H2 622.3 0.5 -0.9 -0.4 0.4
a
Base results in ‘000 Ethiopian Birr; non-base simulation results as % change from base results.
20
Table 6. The impact of alternative policies on village exports
a
Factor/
trade/ activity Base
results Output price
+5%
(a)
Fertiliser
subsidy
Removal
-20%
(b)
Combine
(a)+(b)
Fertiliser
subsidy +20%
Output price
–5%
Teff 794.9 -0.5 -3.8 -4.1 4.9
Other cereals 9.7 -1.5 2.3 1.0 -1.8
Pulses 34.4 -4.5 20.5 15.0 -16.5
Livestock 211.4 0.4 -0.7 -0.1 -0.1
Business 113.7 -7.0 3.8 -3.2 3.0
Skilled labour 13.8 -36.5 45.6 -5.3 1.1
a
Base results in ‘000 Ethiopian Birr; non-base simulation results as % change from base results.
21
Table 7. The impact of alternative policies on fertiliser use by production activity and
household group
a
Production activities by
household group
Base
results Output
price
+5%
(a)
Fertiliser
subsidy
Removal
-20%
(b)
Combine
(a)+(b)
Fertiliser
subsidy
+20%
Output price
–5%
Upland teff
H0
5.5
7.6
-19.0
-12.8
19.9
H1 14.9 5.2 -20.5 -16.2 25.6
H2 161.4 4.8 -20.1 -16.2 25.4
Lowland teff
H0
4.6
7.6
-18.8
-12.6
19.7
H1 7.7 4.0 -21.0 -17.7 27.9
H2 121.8 4.6 -19.8 -16.1 25.1
Upland other cereals
H0
0.7
6.6
-17.9
-12.6
19.3
H1 1.5 5.7 -18.3 -13.7 21.0
H2 14.4 4.9 -19.6 -15.7 24.5
Lowland other cereals
H0
0.8
8.4
-21.4
-14.5
23.1
H1 1.0 8.0 -22.7 -16.2 25.8
H2 6.7 8.0 -23.9 -17.8 28.9
Aggregate change 341.0 4.9 -20.0 -16.1 25.2
a
Base results in ‘000 Ethiopian Birr; non-base simulation results as % change from base results.
22
Table 8. The impact of alternative policies on village import of labour and commodities
a
Commodity Base
results Output price
+5%
(a)
Fertiliser
subsidy
Removal
-20%
(b)
Combine
(a)+(b)
Fertiliser
subsidy
+20%
Output price
–5%
Unskilled labour 91.4 1.4 -0.5 1.0 -0.9
Fertiliser 341.1 4.9 -20.0 -16.1 25.2
Agricultural
commodities
246.9 3.6 -2.3 1.2 -0.9
Other commodities
1031.3 1.8 -1.7 0 0.2
a
Base results in ‘000 Ethiopian Birr; non-base simulation results as % change from base results.
Table 9. Sensitivity analysis of the impact of alternative policies on the village-wide land
degradation externality
a
Level of land degradation and social discount rates
High Low
Policies
3% 5% 10% 3% 5% 10%
Base results 1227.0 736.2 368.1 353.1 211.8 105.9
Output price increase 5% (a) 1272.2 763.3 381.7 365.4 219.2 109.6
Fertiliser subsidy removal (b) 1309.6 785.7 392.9 378.9 227.3 113.7
Combine: (a) + (b) 1360.4 816.2 408.1 393.0 235.8 117.9
Combine: –(a) – (b) 1080.6 648.4 324.2 309.0 185.4 92.7
a
In ‘000 Ethiopian Birr.
Table 10. Aggregate indicators for the economy
Commodity Base
results Output
price
+5%
(a)
Fertiliser
subsidy
Removal
-20%
(b)
Combine
(a)+(b)
Fertiliser
subsidy
+20%
Output
price –5%
Absorption 4262.1 1.4 -1.8 -0.5 0.8
Private consumption 3516.5 1.6 -2.2 -0.6 1.0
Marketed surplus (“export”) 1481.1 -1.4 -1.3 -2.7 3.1
“Imports” -1725.1 2.8 -6.3 -3.8 6.2
Externality, low, 3% disc.rate 353.1 3.7 6.7 10.9 -11.9
a
Base results in ‘000 Ethiopian Birr; non-base simulation results as % change from base results.
23
Source: Lofgren et al. (2002).
Activities
Commodity
Markets
Factor
Markets
Rest of the
World
Households Government Sav./Inv.
Factor
Costs Wages
& Rents
Intermediate
Input Cost
Sales
Private
Consumption
Taxes
Local Private Savings
Government
Consumption
Gov. Savings
Investment
Demand
ImportsExports
Foreign Savings
Transfers
Foreign Transfers
Exports
Imports
Figure 1. Structure of Payment Flows in the Standard CGE model.
24
Figure 2. Typology of village economy models
DifferentiationDifferentiationDifferentiation
Transactions
Costs
High
LowLow
Low High
Nonsepa-
rable farm
household
models
Separable farm household models
Village CGE-models with
nonseparable farm houshold
models
Village CGE-models with
separable farm household models
Figu re 3. T e chnolo gy tree in c ro p an d lives tock p rodu ctio n
L a n d C a p i ta l O x e n L a b o u r
F e r t il i ze r
C r o p o u t p u t
C r o p
r e s id u e s
G r a i n
C r o p
r e s id u e s C a p i ta l L a b o u r G r a zi n g
la n d
L i v e s t oc k o u t p u t
C o n s u m p t i o n S a v i n g s
CE S
(lo w )
(h ig h )
CE S
CE S
(lo w )
25
