Sie sind auf Seite 1von 12

Forest Policy and Economics 9 (2007) 916 – 927

www.elsevier.com/locate/forpol

Economic dependence on forest resources: A case from


Dendi District, Ethiopia
Getachew Mamo a,1 , Espen Sjaastad b,⁎, Pål Vedeld b,2
a
SOS Sahel Ethiopia, P. O. Box 3262, Addis Ababa, Ethiopia
b
Norwegian University of Life Sciences, Department of International Environment and
Development Studies (Noragric), P. O. Box 5003, N-1432 Ås, Norway
Received 13 September 2005; received in revised form 27 July 2006; accepted 4 August 2006

Abstract

This paper examines variation in dependence on forest resources among rural households in Chilimo, Ethiopia, and the income-
equalizing effects of such resources. Data were collected through a systematic questionnaire survey of 102 households, randomly
selected from two peasant associations in the area. Forest income contributed 39% of the average household income, roughly equal
to agriculture, which contributed 40%. Forest income was more important than all other income sources combined for the poorest
40% of households and contributed more to household income than agriculture for 65% of households. While forest income
represents 59% of the total household income for the poorest quintile, the contribution drops to 34% for the wealthiest quintile. On
the other hand, the rich households derive a larger absolute income from forest resources than the poor households. Forest
resources have an important income-equalizing potential among the rural households. Reduced access to forest resources would
greatly affect the welfare of the rural population and increase wealth differentiation among rural households in the study area.
© 2006 Elsevier B.V. All rights reserved.

Keywords: Forest income; Economic dependence; Income distribution; Rural livelihoods; Ethiopia

1. Introduction on environmental resources. In recent years, there


has been increasing interest in the contribution that
There is a shared understanding among researchers, natural forests make to local rural employment and
policy makers and development practitioners that many income (Arnold and Townson, 1998). With a few
rural households in developing countries are dependent notable exceptions, however, there has been little em-
pirical research on the level of dependence across dif-
ferent socio-economic groups.
⁎ Corresponding author. Tel.: +47 64 96 55 11; fax: +47 64 96 52 01. Cavendish (2000) has noted that the contribution of
E-mail addresses: gmheth@yahoo.com (G. Mamo), forest resources to the welfare of rural households is
espen.sjaastad@umb.no (E. Sjaastad), pal.vedeld@umb.no
(P. Vedeld).
regularly overlooked in poverty surveys. Empirical
1
Tel.: +251 11 4160391; fax: +251 11 4160288. investigation of dependency on forest resources may
2
Tel.: +47 64 96 53 07; fax: +47 64 96 52 01. help to improve macro-level poverty estimates and
1389-9341/$ - see front matter © 2006 Elsevier B.V. All rights reserved.
doi:10.1016/j.forpol.2006.08.001
G. Mamo et al. / Forest Policy and Economics 9 (2007) 916–927 917

improve policy planning and execution. The quantifi- higher incomes, within relevant ranges of income in
cation of environmental income may serve as an input developing countries, are likely to increase the
into conservation policy, and particularly establish- pressure on forests. This argument seems to suggest
ment of protected areas, by determining the potential that the relationship will remain linear and positive in
loss to rural dwellers of reduced access to environ- poor communities.
mental resources. Apart from household internal factors, other con-
Arguments and counter-arguments over the link ditions influence household decisions in important
between poverty and environment proliferate. These ways. Greater access to forests and markets may often
links are, however, commonly discussed at macro and accelerate forest extraction (Angelsen and Kaimowitz,
intermediate levels. Here, the focus is on intra-com- 1999). The World Bank (1992), however, suggested
munity variation in villages that are uniformly poor in that markets enhance conservation when there are se-
a macro context. A key issue here is thus to investigate cure property rights. A study from an Amerindian
both forest use and dependence on forest resources community in Honduras (Godoy et al., 1997) reported
among different groups at a local level and its im- both positive and negative impacts of markets for wage
plications for policy. labour and annual crops. When market integration
Several factors may influence household utilization takes place through wage labour employment, it posi-
of freely available forest resources. Some of these tively influences forest resource stocks by diverting
concern household characteristics while others are people from extractive activities. High prices for agri-
contextual. Larger households clear more forest both cultural crops, meanwhile, give incentives to clear
because they have more workers and more mouths to more land. Concerning distance from forest, it is ap-
feed (Dove, 1995; Almeida, 1992; Godoy et al., 1997). parent that proximity to the forest allows the household
Educated individuals may be in a better position to tap to extract more environmental resources by reducing
into income flows from natural stocks. Godoy and labour and transportation costs. Remoteness impedes
Contreras (2001), however, found that a higher level of extraction not only through increased extraction costs
formal schooling is associated with less forest cutting. but also through lower probability of entitlement to the
Adhikari et al. (2004) argued that a higher level of resource. In the context of distribution and depen-
education makes fuelwood collection increasingly dence, it is also important that such factors frame
unprofitable due to higher opportunity costs of labour. profitability aspects differently for different groups of
The age of the head of household may be positively households.
related to forest resource utilization until a peak of The proxy that we use for dependence on natural
physical strength is reached and children move away forest resources is simply the share of absolute house-
(Piland, 1991, cited in Godoy et al., 1997). On the hold income, or the “relative income”, that is derived
other hand, older people may possess superior knowl- from such resources. This is in line with common
edge about various forest resources, and may utilize practice in the field (Vedeld et al., 2004). According to
more medicine plants and wild foods. Wunder (2001) being resource-poor indicates asset and
Total household income (and wealth) is often found welfare poverty.3 Rural households in a state of welfare
to have a nonlinear relationship with forest resource poverty can adopt available coping strategies, includ-
extraction. On the one hand, improved off-farm ing forest extraction activities. Empirical evidence
employment opportunities (Angelsen and Kaimowitz, from several sources (Reddy and Chakravarty, 1999;
1999) and access to credit (Godoy et al., 1997) may Cavendish, 2000; Escobal and Aldana, 2003) indicates
reduce forest clearance as a gap-filling activity. On the that poor households derived a relatively large share of
other hand, better asset endowments allow households their income from forests and wild lands compared to
to exploit more forest resources. While Escobal and better-off households within the same community. The
Aldana (2003) found that the demand for forest study in Nicaragua by Godoy et al. (1995) also
resources increased with income level, Godoy et al.
(1997) found that the relationship between total 3
The concept of “asset poverty” relates to restrictions in choice of
income and forest clearance resembled an inverted economic activities while “welfare poverty” is an absolute
U-shape. Angelsen and Kaimowitz (1999) argue that restriction in household consumption.
918 G. Mamo et al. / Forest Policy and Economics 9 (2007) 916–927

concluded that as income rises, the importance of non- In terms of wealth distribution and forest use,
timber forest products in the household economy the distinction between absolute and relative forest
shrinks. The explanation was that the economic income – and how these vary across groups – is of
importance of other income sources such as agriculture, crucial importance. Generating a large proportion of
wage employment and self-employment would rise absolute household income from forest resources does
relative to the income from environmental resources. not necessarily mean that overall extraction is high.
Swinton and Quiroz (2003) suggested that the Indeed, based on empirical evidence from Zimbabwe,
probability of engaging in tree extraction was reduced Cavendish (2000) concluded that richer households
when people have better access to farmland and credit. use greater quantities of environmental resources in
As the value of labour rises with increasing wealth, absolute terms while the poorer households are more
the opportunity cost of continuing to spend time ga- dependent on the resource. Similarly, a recent study
thering food, rather than purchasing it, becomes in- from Peru by Escobal and Aldana (2003) found
creasingly unattractive (Byron and Arnold, 1999). The that demand for natural resources grows with absolute
argument is that many forest-based activities are income but that the poor depend more on natural
arduous and generate low returns that can be easily resources than the wealthy. In a study covering the
dropped as alternatives become available or household African continents, Arnold and Townson (1998) re-
labour becomes scarce. Similarly, it appears that well- ported that where people have had relatively unre-
educated households depend less on forest resources of stricted access to forests, the poor usually derive a
low economic return. It was reported by Godoy et al. greater share of their overall needs from forest pro-
(1996) that households with more education seem to ducts and activities than the more wealthy, but the
reap higher crop yields, earn higher income, have wealthy are often the heaviest users. This general
fewer children, and suffer from fewer idiosyncratic feature was also found in a recent meta-study of more
shocks than households with less education. Education than 50 cases from Africa, Asia, and Latin America
could also create opportunities for off-farm employ- (Vedeld et al., 2004).
ment, self-employment and facilitate out-migration for What are the implications of the above findings for the
better jobs that reduce dependence on forest resources. role of forest resources in equalizing income? Cavendish
Adhikari et al. (2004) found that larger families have (2000) reported a 30% reduction in income inequality
a greater demand for natural resources and more labour due to inclusion of forest resources in household surveys
to fulfil this demand. However, it appears that house- in Zimbabwe. Empirical evidence from Uganda (Aryal,
hold composition, gender and age structure are more 2002) and Malawi (Botha, 2003) similarly revealed that
important than the mere numbers. Adult labour avail- forest income significantly contributed to reduced in-
ability perhaps places the household in a better position come inequality among rural households. In the World
to liquidate communally owned natural stock than Bank meta-study (Vedeld et al., 2004), the average
labour-poor households. Alternatively, ample adult increase in the Gini-coefficient when forest income was
labour could motivate the household to invest more in excluded was 0.13 (from 0.36 to 0.49).
agriculture or rural employment that can fetch higher What picture emerges from these studies? First,
incomes than gathering activities. In male-dominated while poor households in general exhibit greater
rural settings, forest-based low-return cash activities dependence on natural forest resources than wealthy
are often taken up by female-headed households who ones, the latter derive greater absolute values from the
cannot make a significant living from agriculture due to resources and thus bear a greater responsibility for any
absence of male labour for plowing. Aging may affect degradation that occurs. Second, failure to include
dependence on environmental resources negatively forest resources in poverty surveys will exaggerate
since people lose strength to engage in labour- income inequalities. Third, and by implication, re-
demanding jobs. In contrast, Cavendish (2000) argued duced access to forest resources will have an adverse
that older people have difficulty carrying out arduous affect on income equality.
agricultural tasks and may turn to experience-based Based on primary data from Dendi District in
resource collection activities that demand less physical Ethiopia, this paper explores the role of forest resource
labour and that are free of entry barriers. use in local livelihoods. It examines variation in the
G. Mamo et al. / Forest Policy and Economics 9 (2007) 916–927 919

level of dependence on forest resources across income households gained from peasant association secre-
groups and the potential equalizing effect of forest taries, which indicated a total of 406 households in
resources. The study area and the methodology used Galessa and 477 households in Gare-arera. The random
are covered in Section 2. Results are presented in sample yielded 36 households from Galessa and 66
Section 3, these are then discussed in Section 4, and from Gare-arera.
Section 5 concludes the paper. A questionnaire survey was administered to the 102
households during fieldwork from October to Decem-
2. Study area and methodology ber 2003. For the purpose of this research, a household
was defined as a unit whose members live, cook, and
2.1. The study area eat together. Quantitative data was collected on
household income, expenditure, household character-
The study was conducted in two Peasant Associa- istics and assets holdings. In addition to conventional
tions (PAs) living adjacent to Chilimo National Forest household income, particular attention was given to
Priority Area in Dendi district, Ethiopia. Chilimo different forms of forest income accrued through both
forest, located about 90 km from Addis Ababa to the consumption and cash. To qualify as an environmental
west along Jimma road, is among Ethiopia's few utilization, a resource must be freely provided by
remaining montane forests. The mean annual rainfall is natural processes, i.e. it is “natural bounty” (Cavendish,
1225 mm/year and the altitude ranges from 1700 to 2000). Secondary data were collected from Farm
3200 m above sea level. The forest has gone through Africa and government offices.
different management and ownership regimes and was Collection of primary data rested mainly on a
designated as a national forest priority area in 1982 as detailed seven-page questionnaire, which included sec-
part of the government's increased forest conservation tions on basic household variables, agricultural pro-
efforts. According to Farm Africa (1999), cited in duction, livestock, forest extraction, off-farm activities,
Melaku (2003), 90% of the Chilimo forest is disturbed, and support services. Each of the sections covering
with 35% heavily disturbed, and total forest cover was income sources included detailed questions on all
reduced by 50% between 1991 and 2001. The Ministry relevant inputs and outputs, allowing us to compute net
of Agriculture was given a mandate to conserve the incomes (market and home consumption) for not only
remaining forest, which is currently in the process of the major categories of income but also subcategories
transferring management and use rights to the sur- (e.g. the different forest sources, different crops, dif-
rounding community. Farm Africa, a British NGO, ferent animals). Data on incomes, costs, assets and
plays a facilitating role. The main livelihood sources in activities related to the Ethiopian year 1995, which ran
the area are agriculture, forest-related activities, and from 11 October 2002 to 11 September 2003 in the
livestock husbandry. Agricultural extension is provid- Gregorian calender; maximum recall time for respon-
ed both by the local government and by Farm Africa. dents was thus roughly 14 months.
The reliability of the data is subject to the usual
2.2. Sampling techniques and data collection caveats that apply when information is based on
recollection. Unless there is systematic under- or over-
Seven peasant associations live adjacent to the reporting, however, imprecise responses will enter as
Chilimo National Forest Priority Area. Limited re- white noise. There was no conceivable incentive to
sources necessitated a two-stage sampling process. over-report. Given the limited-use nature of the forest,
First, Galessa and Gare-arera peasant associations were however, there could be a slight danger of under-
identified as broadly representative in terms of distance reporting. We believe that this danger was minimized
to market and proximity to the forest. Galessa is settled by the familiarity of the interviewer and first author
within the range of 11–16 km from Ginchi town, with the area. The high percentage contribution of
whereas Gare-area is within the range of 3–7 km forest resources to local livelihoods found would seem
from town. A total of 102 households were randomly to support this. Strictly speaking, given the nature of
selected from Galessa and Gare-arera using random the two-stage sampling method, the data are represen-
number tables. The sampling frame was the list of tative only for the two peasant associations sampled.
920 G. Mamo et al. / Forest Policy and Economics 9 (2007) 916–927

2.3. Environmental income valuation dence ratio of 1.42. People within the low and high age
brackets do, however, contribute labour, although not
The value of a forest can be divided into use and equal to the full adults equivalent. Family labour is a
non-use values (Emerton, 2001; Campbell and Luck- basis for both on- and off-farm employment activities.
ert, 2002). The use value is further divided into direct Labour is often shared reciprocally among neighbors
and indirect values. Non-use values, such as e.g. and relatives to offset labour bottlenecks during peak
existence value, are of importance to resource valuation seasons in agricultural production. Of the 102 sample
exercises but of limited importance to local household households, 78 were male-headed and the remaining
surveys of environmental income (Vedeld et al., 2004) 24 female-headed. Oromo (81%), Amhara (11%) and
where the focus is on local and private benefits. Indirect Gurage (8%) were the three ethnic groups represented
use values, including environmental services such as in the sample. As much as 58% of household heads
stabilization of water flows and erosion control, are of were illiterate. None of the household heads had
great relevance but beyond the scope of this study. completed high school studies to qualify for engage-
Valuation and household income accounting meth- ment in paid public services or other formal employ-
ods used in this study draw on Cavendish (2002) and ment schemes.
Vedeld et al. (2004). In short, all products were valued Livestock, land, houses, and savings constitute the
according to local market prices except fodder, in main household asset endowment. The target commu-
which case price was calculated through value- nity themselves identified number of oxen, land size,
equivalent substitutes.4 Both consumption and cash savings and housing as wealth indicators during focus
income were calculated for the different goods and group discussions. Oxen ownership was emphasized as
income sources. Material expenditures were deducted the main wealth indicator, since oxen are not only status
from gross income, as were fines and damages inflicted indicators but also an important source of draught
on crops by wildlife. The basic measure of income used power for crop production. Land is held on usufruct
is value added rather than profit or rent; that is, all terms with an average holding of 2.1 ha per household.
inputs except family labour were deducted from gross Size of landholding varied between 0 and 7.5 ha, with
income. This is acceptable when wage labour markets 4.9% of surveyed households being landless. Farmers'
are absent or, as in the study area, relatively uniform perceptions about their own farmland fertility status was
across the sample (Sjaastad et al., 2005). All monetary reported as fertile (16.5%), medium (61.9%), and poor
units were converted from Ethiopian Birr to PPP- (21.6%). Extension services rendered by the Office of
adjusted United States Dollars (USD) by dividing the Agriculture reached 40% of sampled households. Farm
former with the 2000 conversion figure of 2.315. Africa provided small credit services for petty trading
and sheep rearing to 46% of the households.
3. Results Ecological, institutional, and policy factors are
shared by people in the study area and are not con-
3.1. Basic sample characteristics sidered in this analysis. The household food security
situation, distance to market, and proximity to the
The average household size was 6.2 and the average resource were seen as contextual factors identified for
age of household head was 45 years.5 In terms of analysis. On average, the settlements were at a distance
household age composition, 53% of the sample of 1.3 h walk and 9.2 km from forest and nearest town
population were in the 0–14 years bracket, 44% in respectively. Both of the sampled peasant associations
the 15–64 years – or working age – bracket, and 3% in had equal access to the all-weather road that crosses
the 65+ years bracket. This gave an average depen- through the forest. Only 17.6% of the sampled house-
4
holds satisfied their consumption needs through own
Since local market prices for fodder were unavailable, the value farm production. The remaining 82.4% faced seasonal
of fodder was calculated by converting grazing land rental fees into
the yield equivalent.
food shortages for an average of 4.5 months of the year.
5
“Household size” denotes the number of individuals in the Of these, 36% reported firewood sales as a the primary
household; in the regressions in this section, “adult labour” indicates consumption gap filler, 26% reported rural wage
household adult equivalent units (see Cavendish, 2002). employment, 15% reported sales of small ruminants,
G. Mamo et al. / Forest Policy and Economics 9 (2007) 916–927 921

3% reported rural self-employment, and 2% reported unskilled on- and off-farm jobs while rural self-em-
remittances. ployment encompasses petty trade, local liquor pro-
cessing and sales. The mean per capita income of the
3.2. Income levels and sources sample was 344 USD, somewhat lower than the
national average of 410 USD.
Agriculture is the main livelihood in the study area, The area under study is thus found in a fairly high-
with crop cultivation and livestock providing 40% of potential agricultural area, densely populated, with
the average total household income among sampled reasonable urban market access and situated close to a
households. The farming system of the study area can rather substantial source of forest resources. People's
be classified as mixed crop and livestock subsistence livelihoods are firmly connected to agriculture and to
farming. Barley, teff, enset, wheat, vegetables, pulses, natural resource use; they are generally poor and a
and maize are the main crops grown in order of eco- majority report food shortages several months per
nomic importance. The “false banana” enset (Ensete year.
ventricosum) is important as an energy source to
smooth out consumption fluctuations (Ferro-Luzzi 3.3. The importance of forest income
et al., 2001). Teff serves both as a subsistence and cash
crop. The largest share of land is devoted to barley The average contribution of forest resources to
(33%), followed by teff (27%) and wheat (14%). Sale household income was 39%. Firewood, fodder,
and consumption of livestock and livestock products honey, and construction materials are the major
are also important sources of income. sources of forest income (Fig. 2). Firewood constitu-
Income from sale and consumption of forest re- tes the largest proportion of forest income (59%),
sources is almost as important as crop cultivation and utilized for both home consumption and for sale.
livestock combined, constituting 39% of mean annual Construction materials for houses, storage facilities,
income. Rural wage employment (12%), rural self- fences, furniture, and farm implements represented
employment (5%), goods from forest plantations (3%), 21% of forest income, followed by fodder grazed and
and remittances (1%) are other sources of income browsed (18%). Honey contributed the remaining 2%
(Fig. 1). Rural wage employment covers an array of of annual forest income.

Fig. 1. Contribution to annual total household income by source; means and standard deviations (N = 102). Source of income: 1 = agricultural
income; 2 = forest income; 3 = rural wage employment; 4 = rural self-employment; 5 = planted tree income; 6 = remittances.
922 G. Mamo et al. / Forest Policy and Economics 9 (2007) 916–927

Fig. 2. Contribution to annual household forest income by source; means and standard deviations (N = 102). Source of forest income:
1 = fuelwood; 2 = construction materials; 3 = fodder; 4 = honey.

The communities depend entirely on local forest re- Per capita forest income is positively related to
sources for domestic energy needs, construction household size (⁎P b 0.05) and the number of food
materials, farm implements and wooden furniture. In deficit months (⁎P b 0.05) and negatively related to
terms of fodder, the cut and carry system is not com- adult labour ( ⁎⁎ P b 0.01), distance to forest
mon in the study area, except for a very few people (⁎⁎P b 0.01), and distance to market (⁎⁎P b 0.01).
who sell grass grown on plots left for short fallow. Neither sex, age (a square term was included), or
About 85% keep their cattle in the forest from two to education of the household head yielded significant
12 months of the year. 42% reported that they are results. Nor did total per capita income and asset
entirely dependent on the forest for grazing and endowment at the ⁎P b 0.05 level (though weak trends
browsing. Depending on livestock holding and prox- are observable through coefficients significant at the
imity to the forest, the value derived from fodder 10% level).
varied between 0 and 603 USD per year.
3.5. Determinants of forest income dependence
3.4. Determinants of forest income
We used relative forest income – the share of total
From a household economic perspective, there are household income contributed by forest extraction – to
several reasons why one would expect that house- measure the degree of dependence on forest income
holds with different socio-economic characteristics sources. In order to analyse relationships between
adapt differently relative to utilizing forest resources. forest income dependence and household level factors,
These are related to production capabilities, con- we ran an ordinary least squares regression of relative
sumption motives, and how households respond to forest income (RFI) against household characteristics,
challenges beyond their immediate control. Table 1 household income, asset endowment, and various con-
presents the result of an ordinary least squares textual variables. The results are presented in Table 2.
regression where forest income was regressed against Among five explanatory variables in the category of
household characteristics, per capita income, com- household characteristics, household size (⁎⁎P b 0.01)
posite wealth indicators and selected conditioning and adult labour (⁎⁎⁎P b 0.001) were found to be
factors. statistically significant. The signs were the same as
G. Mamo et al. / Forest Policy and Economics 9 (2007) 916–927 923

Table 1 matching the result arrived at for absolute per-capita


OLS regression of household per capita forest income (in PPP- forest income.
adjusted USD) against socio-economic characteristics
Variable Exp. Coef. SE t- p- 3.6. Forest income, forest income dependence, and
sign value value
wealth distribution
Per capita forest income na 288.510 65.702 4.39 0.000
Household size + 25.171 10.099 2.49 0.015
We regressed per-capita forest income (FI) and
Adult labour + − 40.790 12.959 − 3.15 0.002
Age of household head – − 3.013 2.676 − 1.13 0.263 relative forest income (RFI) against per capita total
Age of household ? 0.038 0.028 1.41 0.163 income (PCI). While per capita forest income is
head square positively and significantly related to total per capita
Sex of household head + − 13.616 14.035 − 0.97 0.335 income (R2 = 16.8, ⁎⁎P b 0.01), relative forest income
Education of household – 2.065 5.827 0.35 0.724
is inversely and significantly correlated with per capita
head
Per capita income + 0.025 0.013 1.96 0.053 household income (R2 = 76.4, ⁎⁎P b 0.01). The strong,
Total asset endowment + 0.004 0.003 1.74 0.086 inverse relationship indicates that while comparatively
Forest distance in – − 40.873 12.238 − 3.34 0.001 wealthy households extract greater values they rely
walking hours less on forest resources.
Market distance in km – − 4.473 1.509 − 3.03 0.003
Number of food deficit + 4.365 1.911 2.28 0.025
months in the family Table 2
N =102; S =116.9; R-Sq=0.550; R-Sq(adj) =0.495; F =9.99; P =0.000. OLS regression of household relative forest income against socio-
economic characteristics
Variable Exp. Coef. SE t- p-
those for forest income and the implications are similar. sign value value
Large families are constrained by income options but
Relative forest na 0.796380 0.09240 8.62 0.000
families with high worker–consumer ratios are, income
independent of family size, better able to utilize such Household size + 0.089550 0.02766 3.24 0.002
options. There is also significant and negative correla- Adult labour – −0.128180 0.03505 − 3.66 0.000
tions between forest income dependence and income Age of household + −0.000289 0.00114 0.88 0.383
from other sources; agriculture (⁎⁎⁎P b 0.001), wage head
Sex of household – −0.040610 0.03758 − 1.08 0.283
employment (⁎⁎⁎P b 0.001), and self-employment head
(⁎P b 0.05). Households with a lack of other options Education of – −0.008600 0.01760 − 0.49 0.626
appear more dependent on forests. Asset endowments household head
were, however, not found to have a significant in- Agricultural – −0.000343 0.00006 − 5.79 0.000
income
fluence on dependence.
Rural wage – −0.000343 0.00007 − 5.09 0.000
Distance to the forest is, as expected, negatively employment
and significantly associated with forest dependence Rural self- – −0.000336 0.00016 − 2.11 0.038
(⁎⁎P b 0.01). In contrast to the negative association employment
between forest income and distance to market, how- Land size – −0.005860 0.01075 − 0.54 0.587
(hectares)
ever, there is no significant relationship between de-
Livestock value – 0.000006 0.00001 0.65 0.520
pendence and market distance. Value of buildings – −0.000001 0.00001 − 0.15 0.883
We observe a significant (⁎⁎P b 0.01) positive rela- Savings – 0.000093 0.00006 1.47 0.145
tionship between dependence on forest resources and Forest distance – −0.103360 0.03509 − 2.95 0.004
number of food deficit months over the year. It is in walking hours
Market distance + −0.004547 0.00459 − 0.99 0.324
expected that food-insecure households not only turn
in km
to the natural environment for help to a greater overall Number of food + 0.016845 0.00504 3.34 0.001
extent than food-secure ones but that their dependence deficit months
on the forest is greater. in the family
Age, sex, or education of household head had no N = 102; S = 0.1365; R-Sq = 0.674; R-Sq(adj) = 0.618; F = 11.87;
significant effect on forest income dependence, P = 0.000.
924 G. Mamo et al. / Forest Policy and Economics 9 (2007) 916–927

In order to investigate differences in wealth, data on income is more important than all other income
per capita household income were divided into five sources added together for the poorest 40% of the
quintiles where the bottom and top quintiles are re- community and that the contribution of income from
ferred to as the poorest and the wealthiest group re- agriculture (crop and livestock) is less than that made
spectively. Quintile analysis supports the results of the by forest income for about 65% of the community.
regression analysis. The average annual household Firewood, construction materials, fodder and honey
income for the poorest group (160 USD) was only a were the main contributors to household forest in-
quarter of that of the wealthiest group (636 USD). The comes. Firewood is the main source of income for poor
annual forest income for the wealthiest group (191 households; fodder and construction materials are
USD) was roughly twice that of the poorest group (95 important for households in all income groups, but the
USD). Thus, while forest income represents 59% of overall value extraction is particularly high for the
poorest households' income the proportion declines to wealthiest households. In the study area, utilization of
30% for the wealthiest households. construction material comprises house building, fenc-
We employed ratios to compare absolute value ex- ing, granary construction, and production of furniture
traction and dependence between the poorest and and farm implements.
wealthiest groups, both in terms of overall forest Grass and shrubs are important inputs in livestock
extraction and in terms of the most important use cate- production. Exploitation of fodder resources from
gories (fuelwood, construction materials, fodder, and communal lands is obviously dependent on livestock
honey). The ratios are given in Table 3. The relative endowment. In the study area, the cut and carry system
ratio describes dependence (RFI) by the wealthiest is not developed, and the wealthy benefit more from
quintile divided by that of the poorest quintile; the fodder resources. Apiculture enterprise does not con-
absolute ratio describes absolute value of extraction by tribute much income to the household economy.
the wealthiest quintile divided by that of the poorest To further explore the distributional effects of forest
quintile. income, a Gini coefficient, indicating the level of
Examination of forest income by sources confirmed inequality (see Section 2), was computed for total per
that the wealthiest quintile utilized a higher quantity of capita income. Forest income was then deducted from
all forest resources; the absolute ratios for construction total income and a new Gini coefficient was computed.
materials (4.19), Fodder (3.48), and firewood (1.35) The resulting Gini coefficients were then compared
were all greater than unity (the ratio for honey was also to determine if forest incomes contribute to income
above 1, but the figures here were insignificant). equalization among rural households.
Conversely, dependence was found to be higher for Income inequality, as measured by the Gini
the poorest group, as indicated by a relative ratio coefficient, is 0.28 when forest incomes are included.
smaller than unity, for all forest resources except con- In general, the income inequality among households in
struction materials. Further analysis reveals that forest the study area is not pronounced compared to the
national Ethiopian Gini coefficient index (0.40) and to
Table 3 that of other developing countries like Uganda 0.55
Relative and absolute use ratios (wealthy/poor) for different forest (Aryal, 2002) and Malawi 0.41 (Botha, 2003).
goods When forest incomes are excluded, however, the
Per capita RFI FI Ratio coefficient rises to 0.41.
income by
Poor Wealthy Poor Wealthy Relative Absolute
source
4. Discussion
Fire wood 0.44 0.18 70 94 0.41 1.35
Construction 0.07 0.08 12 49 1.19 4.19
Results based on questionnaire surveys – and thus
materials
Fodder 0.08 0.07 13 47 0.89 3.48 on the recall, accuracy, and intentions of respondents –
Honey 0.00 0.00 0 2 na na are subject to various biases. Recall periods in the
All sources 0.59 0.33 95 191 0.58 2.01 survey on which this paper is based were, however,
RFI = relative forest income; FI = absolute forest income (in PPP- quite normal (maximum 14 months) and problems of
adjusted USD). recall will under some circumstances show up as white
G. Mamo et al. / Forest Policy and Economics 9 (2007) 916–927 925

noise. While the familiarity of one of the authors with and distance to market. This reflects the broad and
the survey area may have reduced strategic answers essential role of such markets also in rural areas;
(particularly with respect to illegal harvesting), how- market-remote households not only have to struggle
ever, such bias cannot be entirely eliminated. Figures with reduced opportunities for sales of forest products
on forest incomes and dependence should thus be but also fewer alternative income-generating options
treated as conservative estimates. in terms of agricultural marketing, self-employment,
The results of this study with regard to forest use and wage labour. They are, in general, poorer.
and dependence support findings from many other Asset endowments are often important for utiliza-
studies. In particular, the study supports the notion that tion of forest resources, particularly with regard to
many rural populations are highly dependent upon livestock and fodder extraction. In this study, however,
forest resources for their livelihoods; in the study area, only a weak trend was observed between asset endow-
forest resources contributed on average 39% of all ment and forest resource extraction. Opposing forces
household income. Both overall extraction of and de- may be an explanation for this: while wealthy house-
pendence upon forest resources were found to increase holds often are characterized by superior opportunities
with household size, indicating that larger households to exploit alternative income sources, and thus will be
face increasing constraints in terms of alternative in- less dependent on forest resources, they will also often
come sources. Greater availability of adult labour was have large herds of livestock dependent on forest
found to reduce both extraction and dependence, in- fodder resources.
dicating on the one hand that children and youth may Seasonal food deficits are positively and signifi-
play a significant part in forest resource extraction but cantly correlated with forest incomes. This is not sur-
also that households with higher worker–consumer prising since natural resources represent an important
ratios are better able to circumvent constraints in terms fallback option for resource-poor rural households,
of livelihood diversification. particularly where there are few alternative employ-
In terms of determinants of forest income, the ment opportunities. As mentioned earlier, 36% of the
positive relation between per capita forest income and survey area households depend on firewood sales
household size is of interest, particularly when juxta- during seasonal food shortage periods.
posed with the result that such income is negatively In terms of the links between forest resource use,
related to availability of adult labour. Larger families dependence, and wealth distribution, this study sup-
may naturally need to clear more forest because they ports conclusions reached elsewhere: while the com-
have more mouths to feed; the increase in per capita paratively wealthy extract greater overall values from
forest utilization as household size increases, however, the forest the poor remain more dependent on these
would seem to indicate that larger families may find it resources. In Dendi District, the wealthiest quintile
difficult to access alternative sources of subsistence, a extracts roughly twice as much value as the poorest
constraint that clearly overwhelms any economies of quintile; but the poorest remain roughly twice as de-
scale in consumption. On the other hand, greater avail- pendent as the wealthy (measured in terms of relative
ability of adult labour may cause the household to contribution). This broad picture can be disaggregated
diversify their economic portfolios to rural wage into different categories of forest goods. Thus, while
employment, self-employment and other activities that the poor disproportionately cut trees for firewood, the
drain labour from extractive activities. The results may wealthy rely to a greater extent on construction
indicate, first, that children and youth are engaged in materials, a higher-value output.
extractive forest activities and, second, that households In general, this study confirms the important role of
with high worker–consumer ratios are less affected by forests in securing livelihoods among the poorest rural
constraints in availability of alternative income sources. dwellers. When forest incomes are removed, the Gini
Increasing distance to the forest edge and to the coefficient rises from 0.28 to 0.41, a substantial change
nearest town market reduced overall extraction of in the context of a uniformly poor rural area. This is
forest resources. While dependence also decreased identical to the increase of 0.13 found in Malawi
with increasing distance to forest edge, however, there (Botha, 2003), higher than the increase of 0.06 found in
was no significant relationship between dependence Uganda (Aryal, 2002), and comparable to the increase
926 G. Mamo et al. / Forest Policy and Economics 9 (2007) 916–927

of 0.10 found by Vedeld et al. (2004) for a collection of crucial importance to natural resources and rural
studies from different developing countries. A study livelihoods that are based on complete ignorance of
from rural Zimbabwe (Cavendish, 1999) reported a the link between the two.
30% reduction of measured inequality among rural The political economy of forest use and livelihoods
households due to inclusion of forest income. Although in rural areas of poor countries is faced with two
there is variation from case to case, the general trend fundamental dilemmas. First, use of and dependence
indicates that forest resources have considerable on forest resources are often high in terms of physical
potential for reducing wealth differentiation among extraction rates and relative economic contribution,
rural households, and our study supports this. particularly among “the poorest of the poor”. Such use
Note, however, that the true equalizing effect of and dependence may lead to overexploitation in the
forest resources will depend on the optional activities sense of reducing long-term supplies of direct-use
faced by households; how they would have acted in the values as well as environmental services such as water
absence of forests and the income thus derived. In a flow stabilization, erosion control, carbon sequestra-
realistic scenario where the income-earning potential tion, and biodiversity. To pursue conservation agendas
of optional activities increases with wealth, the Gini by restricting access to forest resources may, however,
coefficient differential may either underestimate or relegate rural people to levels of livelihood deprivation
overestimate the true effect. It will be correct only that entail health problems, loss of life, distress mi-
when potential income from optional activities tends to gration, social disaffection and conflict. Second, the
be proportional to existing non-forest income. In short, solution of increasing agricultural productivity or pro-
the Gini coefficient differential will produce fairly fitability, no simple task in itself, may – in closed
good results whenever the general scenario of better economy settings – relieve pressure on forest re-
options for the wealthy holds. When this scenario does sources, but will more often lead to in-migration and
not hold, however, the method is dubious. increased forest clearing (Angelsen, 1999).
Among rural populations in poor countries, forest Neither solution – reduced forest access, increased
resources can serve several purposes (see Section 2). agricultural productivity or profitability – is realistic in
The picture that emerges from the study area is one the context of Dendi District, Ethiopia, and areas with
where the forest is not merely a safety net, to be utilized comparable natural and social environments. Greater
in times of predictable or unpredictable downturns in local participation in decision making, promotion of
other income-earning opportunities, but an essential, improved forest management practices, and enhanced
perennial livelihood source for poor and “wealthy” marketing infrastructure may contribute positively but
households alike. The distinction between poor and will ultimately have limited effect. Instead, expanding
wealthy mainly relates to the type of goods extracted the range and profitability of livelihood options other
(determined in part by asset endowment) and the ability than forest extraction and agriculture appear as the only
to access additional income sources. But the forest is viable alternatives. This implies a policy shift away
universally important; even for the wealthiest quintile, from the resource itself to the broader local economy
relative forest income reaches 30%, well above the and alternative rural livelihoods. Recent studies from
average (22%) found for all groups across more than 50 Africa and other continents have tended to emphasize
cases worldwide in a study by Vedeld et al. (2004). the diversity of rural livelihoods and the exaggerated
importance often accorded land and natural resources
5. Conclusion in the past. The case from Dendi reminds us that in
some areas, land and natural resources remain abso-
In light of the above discussion, an important point lutely critical, to the possible detriment of both the local
obtains with regard to poverty surveys, which often population and the resource on which it depends.
overlook income from wild resources such as natural
forests, particularly subsistence income. One effect of References
this is the overestimation of rural poverty in absolute
terms; more problematically, policies based on such Adhikari, B., Falco, S.D., Lovett, J.C., 2004. Household character-
incomplete surveys may culminate in policies of istics and forest dependence: evidence from common property
G. Mamo et al. / Forest Policy and Economics 9 (2007) 916–927 927

forest management in Nepal. Ecological Economics 48, Farm Africa, 1999. Rapid forest assessment of Chilimo Forest. Final
245–257. Report. CNRASD–Consultant, Addis Ababa.
Almeida, A.L.O., 1992. The colonization of the Amazon. University Ferro-Luzzi, A., Morris, S.S., Tafesse, S., Demissie, T., D'Amato,
of Texas Press, Austin. M., 2001. Seasonal undernutrition in rural Ethiopia: magnitude,
Angelsen, A., 1999. Agricultural expansion and deforestation: correlates, and functional significance. International Food Policy
modelling the impact of population, market forces and property Research Institute Research Report, vol. 118. Washington, D.C.
rights. Journal of Development Economics 58, 85–118. Godoy, R., Contreras, M., 2001. A comparative study of education
Angelsen, A., Kaimowitz, D., 1999. Rethinking the causes of and tropical deforestation among lowland Bolivian Amerindians:
deforestation: lessons from economic models. The World Bank forest values, environmental externality, and school subsidies.
Research Observer 14, 73–98. Economic Development and Cultural Change 49, 555–574.
Arnold, M., Townson, I., 1998. Assessing the potential of forest Godoy, R., Brokaw, N., Wilkie, D., 1995. The effect of income on
product activities to contribute to rural incomes in Africa. ODI the extraction of non-timber tropical forest products: model,
Natural Resource Perspectives 37 (November). hypothesis, and preliminary findings from the Sumu Indians of
Aryal, B., 2002. Are Trees For The Poor? A Study from Budogo Nicaragua. Human Ecology 23, 29–52.
Forest, Uganda. Agricultural University of Norway, Department Godoy, R., Groff, S., O'Neill, K., 1996. The Role of Education in
of Economics and Social Science, Ås. MSc Thesis. Neotropical Deforestation. Mimeo. Harvard Institute for Inter-
Botha, N.B., 2003. Looking for A Path Out of Poverty: The Study of national Development, Cambridge, Mass.
Chimaliro Forestry Reserve, Malawi. Agricultural University of Godoy, R., O'Neill, K., Groff, S., Kostishack, P., Cubas, A.,
Norway, Department of Economics and Social Science, Ås. MSc Demmer, J., McSweeney, K., Overman, J., Wilkie, D., Brokaw,
Thesis. N., Martinez, M., 1997. Household determinants of deforestation
Byron, N., Arnold, M., 1999. What future for the people of the by Amerindians in Honduras. World Development 25, 977–987.
tropical forests? World Development 27, 789–805. Melaku, B., 2003. Forest Property Rights, the Role of the State, and
Campbell, B.M., Luckert, M.K. (Eds.), 2002. Uncovering the hidden Institutional Exigency: The Ethiopian Experience. PhD Thesis,
harvest: valuation methods for woodland and forest resources. Department of Rural Development Studies, Swedish University
Earthscan, London. of Agricultural Sciences, Uppsala.
Cavendish, W., 1999. Poverty, inequality and environmental Piland, R.A., 1991. Traditional Chimane Agriculture and its Relation
resources: quantitative analysis of rural households. Working to Soils of the Beni Biosphere Reserve, Bolivia. Masters thesis,
Paper Series 99-9. Centre for the Study of African Economies, University of Florida, Gainesville.
Oxford. Reddy, S.R.C., Chakravarty, S.P., 1999. Forest dependence and
Cavendish, W., 2000. Empirical regularities in the poverty– income distribution in a subsistence economy: evidence from
environment relationship of rural households: evidence from India. World Development 27, 1141–1149.
Zimbabwe. World Development 28, 1979–2000. Sjaastad, E., Angelsen, A., Vedeld, P., Bojö, J., 2005. What is
Cavendish, W., 2002. Quantitative methods for estimating the environmental income? Ecological Economics 55, 37–46.
economic value of resource use to rural households. In: Swinton, S.M., Quiroz, R., 2003. Is poverty to blame for soil, pasture
Campbell, B.M., Luckert, M.K. (Eds.), Uncovering the hidden and forest degradation in Peru's Altiplano? World Development
harvest: valuation methods for woodland and forest resources. 31, 1903–1919.
Earthscan, London, pp. 17–65. Vedeld, P., Angelsen, A., Sjaastad, E., Berg, G.K., 2004. Counting
Dove, M.R., 1995. Swidden agriculture in Indonesia. Mouton on the environment: forest incomes and the rural poor. World
Publishers, New York. Bank Environment Department Working Paper, vol. 98.
Emerton, L., 2001. Why forest values are important to East Africa, Washington, D.C.
2001. ACTS Innovation 8, 1–5. World Bank, 1992. World development report. Oxford University
Escobal, J., Aldana, U., 2003. Are nontimber forest products the press, Oxford.
antidote to rainforest degradation? Brazil nut extraction in Madre Wunder, S., 2001. Poverty alleviation and tropical forests — what
De Dios, Peru. World Development 31, 1873–1887. scope for synergies? World Development 29, 1817–1833.

Das könnte Ihnen auch gefallen