Collective Resource Management and Labor Quota Systems for Sustainable Natural Resource Management in Semi-Arid Ethiopia

Abstract

Since the mid-1970s, natural resource management package programs have been implemented mainly in the northern Ethiopian Highlands (Amhara and Tigray regions), providing participants with food-for-work (FFW) supported by donor agencies. Meanwhile, the government has sporadically implemented such programs in the Ethiopian Lowlands, including the semi-arid Ethiopian Rift Valley (the study area). Local villagers took the initiative to manage various natural and life resources. In this study, the following factors were determined: (i) the type of village organization that manages common-pool resources (CPRs) and controls collective work, and (ii) the kind of institutional arrangements that should be implemented in participatory CPR management and small-scale village infrastructure development programs. These issues were investigated using mixed methods research, combining multivariate analyses, interviews, and field observation. The analyses were compared specifically with advanced participatory CPR management in Tigray, northern semi-arid Ethiopia. Tigray has an indigenous labor quota system (baito) and a collective grazing land management system (hizati) at the hamlet (qushet) level. Since 1991, the Tigray government has incorporated hamlets into the local administration system and supplied FFW and other incentives to participants with a high participation rate in the baito collective work. Those institutional arrangements helped reduce soil erosion rates and restore grass and tree biomass in the area. In the study area, user groups and youth and women’s associations were (and still are) institutionally fragile CPR organizations (e.g., no bylaws). In contrast, an iddir is a robust CPR organization at the hamlet (gott) level having a labor quota system (iddir system) and funds. The requirements for sustainable participatory rural development in the two regions of semi-arid Ethiopia are, first, to institutionalize a local administration system that links the district, village, and hamlet; and second, to use various incentives provided by donor agencies to strengthen their indigenous labor quota systems.

1. Introduction

Sustainable common-pool resource¹ (CPR) [1,2] management (including natural resources, such as forests and grasslands) has experienced trials and failures due to either government intervention (e.g., due to rent seeking [3]) or market mechanisms (e.g., due to externalities [4]). An alternative approach has been systematized by extensive fieldwork and reviews of a large number of empirical studies of the institutional system of CPRs undertaken by many researchers. Among these, Ostrom [1] concluded that self-sustaining organizations could maintain long-term sustainable use of resources. She also advocated design principles for CPR institutions (Table 1).

Ethiopia has undulating mountainous landscapes and volcanic soils. The rainfall in Ethiopia has large drop sizes and high kinetic energy [5]. These terrain properties and rainfall characteristics induce high water erosion in Ethiopia [6]. The mean annual soil erosion from actually measured catchments ranges from 15 Mt ha⁻¹ y⁻¹ [7] to 17 Mt ha⁻¹ y⁻¹ [8], which is higher than the global average of 12–15 Mt ha⁻¹ y⁻¹ [9]. Partly because of the intensive soil erosion, crop fields in Ethiopia have been losing soil fertility [10]. The population in Ethiopia is predicted to be 188 million in 2050, 216% of that in 2010 (UN medium-variant projection). The food demand to feed this population is estimated to be 237% of that in 2010 [11]. Forests in Ethiopia are home to various resources needed for people’s daily lives, such as fuelwood and construction materials. Between 1990 and 2010, Ethiopia annually lost an average of 1% of its total forest area (~1500 km²) [12]. The forest area in 2020 was 15% of the total national land area. Livestock in Ethiopia are used as the draft power to plow crop fields and thresh harvested crops. They also have value for meat and dairy products. Animal dung is used for making manure, construction materials, and fuels. Livestock are grazed in grasslands and harvested crop fields; however, many grazing lands are subjected to overgrazing conditions, involving area reduction mainly due to population increase [13], reduction in herbaceous species richness, reduction in each species’ abundance [14], and soil degradation [15].

Table 1. Design principles derived from studies of long-enduring institutions governing sustainable common-pool resources (CPRs).

Design Principles	Description
1. Clearly defined user and resource boundaries	Clear boundaries between legitimate CPR users and non-users are present. Clear boundaries that define a resource system and separate it from the larger biophysical environment are present.
2. Congruence of appropriation and provision rules 1 with local conditions	(i) The costs incurred by CPR users are proportional to the benefits they receive via their participation. The rules to determine (i), and (ii) deal with appropriation and provision problems, are congruent with local social and environmental conditions 2.
3. Collective-choice rules	Most individuals affected by appropriation (harvesting from CPRs) and provision (protection of CPRs) rules are authorized to participate in making and modifying their rules.
4. Monitoring users and resources	Monitors, who audit appropriation and provision levels of CPR users and biophysical conditions of the resources, are accountable to the users or are the users themselves.
5. Graduated sanctions	Sanctions for rule violations are initially very low but become stronger if a user repeatedly violates a rule.
6. Conflict-resolution mechanisms	There are rapid and low-cost local arenas to resolve conflicts among users or between users and officials.
7. Minimal recognition of rights by the government (Independence)	CPR users’ rights to make and devise their own rules are recognized and not challenged by external governmental authorities.
	For resources that are parts of larger systems:
8. Nested institutionalization	When a CPR is closely connected to a larger social-ecological system, governance activities (e.g., appropriation, provision, monitoring, enforcement, conflict resolution) are organized in multiple nested layers.

¹ Appropriation problems occur with unregulated excessive appropriation (harvesting) of CPRs. Provision problems are related to investment in the establishment and maintenance of CPRs and sometimes regulating harvesting rates to avoid adversely affecting CPRs [1]. ² Reviewing the successful cases of community-based natural resource management indicated that a harmonious relationship between cost-and-benefit distributions among CPR users is essential to deal with appropriation and provision problems on sites [16]. Sources: Ostrom [1], Cox et al. [16] (Table 4), Ostrom [2] (p. 33, Table 2.1), and Ostrom [17] (p. 653). Adapted by the author.

Table 1. Design principles derived from studies of long-enduring institutions governing sustainable common-pool resources (CPRs).

Design Principles	Description
1. Clearly defined user and resource boundaries	Clear boundaries between legitimate CPR users and non-users are present. Clear boundaries that define a resource system and separate it from the larger biophysical environment are present.
2. Congruence of appropriation and provision rules 1 with local conditions	(i) The costs incurred by CPR users are proportional to the benefits they receive via their participation. The rules to determine (i), and (ii) deal with appropriation and provision problems, are congruent with local social and environmental conditions 2.
3. Collective-choice rules	Most individuals affected by appropriation (harvesting from CPRs) and provision (protection of CPRs) rules are authorized to participate in making and modifying their rules.
4. Monitoring users and resources	Monitors, who audit appropriation and provision levels of CPR users and biophysical conditions of the resources, are accountable to the users or are the users themselves.
5. Graduated sanctions	Sanctions for rule violations are initially very low but become stronger if a user repeatedly violates a rule.
6. Conflict-resolution mechanisms	There are rapid and low-cost local arenas to resolve conflicts among users or between users and officials.
7. Minimal recognition of rights by the government (Independence)	CPR users’ rights to make and devise their own rules are recognized and not challenged by external governmental authorities.
	For resources that are parts of larger systems:
8. Nested institutionalization	When a CPR is closely connected to a larger social-ecological system, governance activities (e.g., appropriation, provision, monitoring, enforcement, conflict resolution) are organized in multiple nested layers.

¹ Appropriation problems occur with unregulated excessive appropriation (harvesting) of CPRs. Provision problems are related to investment in the establishment and maintenance of CPRs and sometimes regulating harvesting rates to avoid adversely affecting CPRs [1]. ² Reviewing the successful cases of community-based natural resource management indicated that a harmonious relationship between cost-and-benefit distributions among CPR users is essential to deal with appropriation and provision problems on sites [16]. Sources: Ostrom [1], Cox et al. [16] (Table 4), Ostrom [2] (p. 33, Table 2.1), and Ostrom [17] (p. 653). Adapted by the author.

In the 1980s, droughts and famine frequently affected Ethiopia. The socialist government (1974–1991) and donor agencies considered that the degradation of country-level natural resources aggravated famines, and as a result started national-level natural resource management package programs [18]. These programs were implemented by providing participants with food-for-work (FFW; in exchange for participation in work, grain or cash are supplied mainly by donor agencies)². The techniques used in the package programs were a combination of cropland soil and water conservation (SWC), hillside vegetation restoration, and gully control [19,20]. The main components of cropland SWC were the construction of soil bunds (Figure 1f) and stone bunds (Figure 1g). In Ethiopia, steep hillsides, unsuitable for crop fields, are considered communal lands for the village community. For hillside vegetation restoration, hillside exclosure (Figure 1d), hillside SWC (e.g., bench terrace) construction, and reforestation on exclosed hillsides have been selected. The main areas in Ethiopia have a free and uncontrolled grazing system [21]. To protect grasses, trees, and planted seedlings from animal feeding, exclosure is a necessary treatment. Large-scale gullies across several villages are commonly found all over Ethiopia (Figure 1a)³. By treating the evolved parts of the gully with check dams (gabions and stones), the road can be repaired, and the gully can be partly treated (Figure 1b,c).

These package programs have been implemented mainly in the northern Ethiopian Highlands (especially Amhara and Tigray regions; Figure 2), which are acknowledged by the government and donor agencies as the frontier of desertification [18]. However, hillside exclosure and reforestation sites implemented from 1985 to 1990 were extinguished in 1995 [18]. Of the cropland SWC measures constructed between 1976 and 1990, only 30% of the soil bunds and 25% of the stone bunds were used in 1994 [23]. These failures in the 1980s were attributed to top-down planning that did not reflect villagers’ opinions and ideas, and package programs that did not consider local realities [18,19].

Reflecting on the policy failure in the 1980s, participatory land and natural resource management, where the authority to decide on and use land and natural resources was partly or fully transferred to local communities or locally accountable institutions, began in sub-Sahara Africa in the 1990s [24]. Natural resource management package programs implemented in the Tigray region under the Ethiopian People’s Revolutionary Democratic Front (EPRDF; 1991–2019) since the 1990s are appreciated as a successful case of participatory land and natural resource management [25]. In Tigray, more local institutions’ participation is realized in the association between the regional government, district agricultural office, village, and hamlet (locally called qushet; Tigrinya) [25]. Because of the package programs implemented, area-specific soil loss caused by gully erosion in an actually measured catchment in Tigray reduced after 2000 (from 17.6 Mg ha⁻¹ y⁻¹ to 8.3 Mg ha⁻¹ y⁻¹) [26]. Multiscale assessment methods elucidated that sheet and rill erosion rates have decreased by ~68% [27]. Area exclosure and various SWC measures retarded 74% of the soil loss from a conservation catchment [7], and total runoff from the catchment was reduced to 19% [28].

The northern semi-arid Highlands (Tigray) and semi-arid Ethiopian Rift Valley (the study area; Figure 2) are characterized by recurrent drought. They are also vulnerable to the negative impacts of climate change, in addition to the arid regions of Afar and Somali [29]. In contrast to the semi-arid Northern Highlands (Tigray), where participatory natural resource management approaches were established and already attained considerable achievements, these package programs have not been implemented in other domestic areas (particularly Ethiopian lowlands occupying 61% of the Ethiopian landmass). However, area-specific soil losses from croplands in the semi-arid Ethiopian Rift Valley, part of the Ethiopian lowlands (i.e., 19.6 Mg ha⁻¹ y⁻¹ [22]; that by gully erosion was 16·2 Mg ha⁻¹ y⁻¹ [30]), were comparable with those in the Ethiopian Highlands. Even in the semi-arid Ethiopian Rift Valley, the natural resource management package programs were sporadically implemented [22]. The techniques used in the project area were more or less the same as those used in the Northern Highlands (Figure 1). Although those programs have not been significantly implemented by the government or donors, villagers collectively manage natural and life resources (water sources for daily life, grazing lands, forests, etc.), and conduct collective work for the management of CPRs and small-scale infrastructure construction and maintenance (e.g., school [31], road, and clinic construction) in the study area.

In contrast to abundant case studies on natural resource management in the Northern Highlands (Tigray), studies in the semi-arid Ethiopian Rift Valley have only been conducted since 2018 [32]. Thus, the first objective of this study is to elucidate the type of village organization that manages collective natural and life resources and controls collective work in the semi-arid Ethiopian Rift Valley. Both semi-arid areas use indigenous labor quota systems in the collective work: the baito system in Tigray [33] and the iddir system in the semi-arid Ethiopian Rift Valley [31]. The iddir system was analyzed and compared with the baito system.

Iddir is a funeral association that manages ceremonial occasions, marriages, funerals, etc., and informal insurance arrangements that provide deprived households with cash and food [34,35]). Regardless of gender, ethnic group, age, economic status, etc., household heads who want to join can sign up for an iddir. In all the iddirs in the Boset district of the Oromia region (part of the semi-arid Ethiopian Rift Valley, the study area), members meet every other Sunday morning in front of the iddir office to discuss various topics. An accountant collects 0.5–1.0 birr (Ethiopian currency; 1 birr = USD 0.05, 2013, the survey year [36]) from every member, enters this in the account book, and adds it to the iddir funds. Iddirs can be observed in all Ethiopian cities and villages [34]. Recently, iddirs have been involved in social and rural development activities. Iddirs have provided part of the expenses for the construction of small-scale infrastructure such as schools and clinics, micro-credit, and job training from their funds⁴ [31,38,39]. Iddirs have bylaws about their activities and organizations, admission restrictions, and penalties for violating their bylaws. Iddirs also have funds that enable shortages of finance to be overcome, which is the most critical factor affecting the performance of social development activities in developing countries [40]. Therefore, many researchers pointed out that iddirs could be a core organization for participatory social development activities in Ethiopia [31,40,41,42]⁵. However, these trials remained within individual pilot cases, and no proposals to increase the scale have been made by the government to offer broader coverage [40].

The second objective of this study is to consider the kind of institutional arrangements that should be adopted in villagers’ participatory natural resource management and small-scale village infrastructure development programs in the study area. It should first be borne in mind, that, in Ethiopia, the decentralization policy implemented by the former EPRDF regime transferred a large share of the budget (45% of the regional public expenditures) and authority to the district (woreda) administration [45]. Therefore, in comparison to the local administration system between two semi-arid regions in Ethiopia, those below the district (woreda) level, including villages (kebele) and hamlets (gott/qushet)⁶, are particularly important.

Second, because Ethiopia has been dependent on aid, it is affected by donors’ foreign aid policy. In Tigray and Boset, which are vulnerable to drought, a productive safety net program (PSNP) began in 2005. PSNP supplies food and cash to those who participate in public work such as the FFW programs. However, while previous FFW was inclined to community development approaches, PSNP emphasizes boosting individual/households’ livelihoods, influenced by neoliberal foreign aid policy [49]. Thus, compared with participatory rural development approaches in the two areas, the relationships between village organizations, district administration, and donor agencies should be analyzed with selected individual development cases.

Third, it is well-known that, in the humid Ethiopian Highlands, e.g., the Amhara region (Figure 2), cropland SWC measures, e.g., soil bunds, do not attract farmers in terms of crop yield, based on ten years of field experiments conducted at two sites (annual rainfall of 1372 and 1617 mm [50]). For a number of reasons (i.e., (i) setting soil bunds ruined ~15% of crop fields (so-called yield penalty), (ii) the structures encourage the spread of running grass and provide breeding grounds for rodents [19], and (iii) other reasons [19]), considering SWC construction and maintenance costs, the discounted incremental net gain (discounted flows of net benefits with and without SWC) was found to be negative⁷ [50]. Because of hillside exclosure, indigenous grass biomass increases within a few years (Figure 1d). On the other hand, hillside exclosure constrains villagers’ use of life resources (e.g., firewood and construction materials). In the long run, this approach may increase those life resources; however, this takes a long time. The survival rate of the seedlings planted on a communal hillside is 9.5% [54], and that at one year after reforestation is only 7% [55]. Due to hillside exclosure, grazing land areas for livestock in the village are reduced, and villagers are forced to look for other means to feed their livestock. Gully treatment generally requires large-scale collective earthwork involving from several dozen to more than one hundred workers over several days. Thus, cropland SWC in the humid Highlands, hillside conservation, and gully treatment are forms of natural resource management from which it is hard to generate short-term benefits for villagers. Instead, they often cause friction among villagers about land use and tenure [32]. In the meantime, conserving these natural resources helps reduce soil loss and runoff from catchments, which are socially valuable projects [56]. Therefore, the government should give farmers incentives to encourage them to participate in natural resource management programs [56]. Alternatively, society may have to look for other incentives (carrots and sticks) to persuade land users to implement conservation practices [57] (p. 750). In Section 2, the history of natural resource management policy in Ethiopia (Section 2.1.) and in the Tigray region (Section 2.2.) is briefly reviewed, focusing on how the administration at the time encouraged villagers to participate in the natural resource management program.

The third objective of this study is to assess whether the institutional arrangements observed in the two semi-arid Ethiopia regions fulfill the principles of sustainable institutions governing CPRs (Table 1). Furthermore, the support that would be needed in the case of failures is discussed.

This study used mixed methods research to determine the type of village organization that assumes CPR management and infrastructure construction, and controls collective work, in the study area [58], combining quantitative approaches to analyze the field data and qualitative approaches based on interviews and field observations at the sites. This study used multivariate analysis techniques to explore the field data. SPSS ver. 20 (IBM) was used for statistical analyses. To fulfill the second objective of this study, the results of the mixed methods analyses were compared with case studies in other parts of Ethiopia. Information on the other Ethiopian regions was obtained from a literature review.

2. History of Natural Resource Management in Ethiopia

2.1. Natural Resource Management in Ethiopia

In the Haile Selassie Imperial period (1930–1974), villagers had the privilege of using natural and life resources in communal lands in exchange for collectively conserving them. The Imperial regime institutionalized a feudalistic landlord-tenant relationship in rural Ethiopia [59]. In this land administration system, landlords paid taxes to the government, and most villagers cultivated croplands as tenants. Up to the end of the Imperial regime in 1974, a traditional land tenure and property right system called Irist (Amhara), based on a blood-related group of people, was the dominant system from north to central Ethiopia⁸ [61].

The socialist regime (1974–1991) that overthrew the Imperial government in 1974 prohibited private land ownership, confiscating land from landlords and distributing it to tenants. The socialist regime established an administration body, i.e., a peasant association (PA), in every village (kebele). Every villager over 18 registered as a PA member, and the PA gave them farmland. A land-reform proclamation issued simultaneously gave authority to PAs for communal natural and life resource management in the village. The land tenure and property rights system (Irist) in the Imperial regime was abolished. Hamlet dwellers were deprived of the privilege of communal land use. However, because the socialist government did not have any explicit policy on communal land use, and PAs could not substantially resolve land holding and tenure disputes, communal lands were gradually encroached upon, partly turning into individual croplands [62]. Villagers having critical feelings about vegetation reduction due to increases in human and livestock populations took the initiative and began natural and life resource management in various regions [63]. Its key feature was that hamlet dwellers in different areas continuously assumed communal land management. Grazing land management in Tigray, called hizati (Tigrinya), was undertaken by qushet (hamlet) [64]. In Amhara in 1999/2000, 53% of the villages had at least one communal grazing area managed by the village (kebele; 30%) and hamlet (gott; 70%) [47]. Villagers were motivated to participate in natural resource management programs of the socialist regime by a combination of campaign-based mass mobilization and FFW [18,32]. PAs selected who would join in the work [31]. FFW was supplied to the participants. However, many farmers did not necessarily willingly or voluntarily participate in the programs⁹ [65].

After the natural resource package programs with mass mobilization and FFW in the 1980s failed, the EPRDF regime (1991–2019), which defeated the socialist regime, introduced two policies following the recommendations of donor agencies. One was land registration, which was a neoliberal policy, and the other was a participatory natural resource management approach. The EPRDF began land registration and certification programs in 1998 [66]. The land registration program improved land use rights and tenure security for individual holdings [67]. A 1997 Ethiopian federal proclamation (law) devolved responsibility for land policy to the regions. Land use policies for personal land use, e.g., land registration, were more or less similar among the regional governments; however, considerable diversity in key policies was observed for communal land management among the regions¹⁰ [66]. Even in the EPRDF period, the Amhara and Oromia regional governments entrusted communal land management to PAs and did not implement any explicit communal land use policy. The Amhara agricultural office decided to exclose 61,178 ha of degraded hillsides between 1995 and 1998, and entrusted the exclosure hillside management to PAs; however, this trial failed, and encroaching started again [68].

A campaign-based watershed management program, which began nationwide in 2012, did not supply FFW to the participants. The program introduced three new approaches that were not seen in the 1980s’ package programs: (i) planning and implementation in the unit of the micro-watershed (200–500 ha); (ii) a micro-watershed association was organized at the PA level; and (iii) training was given to the participant farmers before the program implementation [32]. Practically, the micro-watershed association, which was composed of landless and deprived households, was organized¹¹. The conservation techniques used in the program were almost the same as those used in the 1980’s package program. Assefa [32] investigated three villages in the Boset district. The effectiveness of the hillside conservation was unclear¹² [32]. Conflicts were frequently observed between the micro-watershed associations and farmers who lived near the micro-watersheds and between the micro-watershed association and a youth association [32]. Assefa [32] concluded that farmers are less motivated to participate in the program or adopt conservation technologies that will not generate short-term benefits. They are motivated to contribute labor and working tools with pressure from local government actors.

2.2. Participatory Natural Resource Management in Tigray

The principal features of participatory natural resource management in the Tigray region are, first, the hamlet (qushet) manages communal grazing lands and forests. In Tigray, 90% of the villages (tabia; the same as kebele in other regions) had an average of four communal grazing lands managed by hamlets [70]. In Tigray, 88% of the villages had at least one exclosed communal forest, of which the village managed 30% and the hamlet managed 58% [25]. In other Ethiopian regions, communal forests and grazing lands were encroached upon, turning into croplands most in the EPRDF period (1991–2019). In contrast, natural resource management programs in Tigray made tremendous progress in the EPRDF period. Exclosure hillsides in Tigray expanded from 143,000 ha in 1996 to 895,220 ha in 2011 [71]. During these periods, based on the hizati system, which was continuously maintained since the Imperial period, hamlets managed communal grazing land¹³ [63]. In the 884 ha catchment studied by Ogawa et al. [77], the increase in exclosure communal forests, continuous communal grazing lands management, and livestock population increase resulted in (i) the disappearance of free-grazing areas (free grazing was allowed all year round), (ii) an increase in the rainy-season exclosure grazing land (free grazing was allowed only in the dry season), and (iii) an increase in exclosure communal forests (reforested areas where grazing and cutting trees were prohibited, but firewood collection from dead branches and cut-and-carry grass collection were allowed)¹⁴. The land use changes of (ii) above were based on mutual discussions between hamlets and those of (iii) were led by the regional government. A comparison between the Imperial period, the socialist period, and the survey year (2017) elucidated that, although the livestock population was highest in 2017, the vegetation in the survey catchment was the highest in 2017 [77].

The second prominent feature of the Tigray participatory approach is the integration of an indigenous hamlet-based labor quota system (the baito system; Tigrinya) into natural resource management programs. Haregeweyn et al. [79] investigated a campaign-based watershed management program that was implemented in a catchment (2343 ha). In Tigray, this program was launched in 2004 as a forerunner of the national-level program. A team comprising hamlet representatives, extension workers (development agents in Ethiopia), and agricultural officers makes management and collective work plans. An actual earthwork was implemented using the baito labor quota system. A continuous program implementation from 2004 to 2009 reduced runoff and soil loss from the catchment in 2009 to 27% and 89% of the 2004 level, respectively. Gullies in the catchment were almost rehabilitated.

Tigray People’s Liberation Front (TPLF), the ruling party of the regional government, originally advocated a people’s participatory approach. In 1991, TPLF took up the indigenous baito system for natural resource management programs and small-scale infrastructure development programs, including for roads, schools, and clinics [64]. Although FFW is sometimes paid for work participants, all adult men and women contribute 20–27 working days per year without payment under the baito [33]. In the two hamlets surveyed by Girmay [33] in 2003, 53% of the collective work was for cropland SWC, 18% was for hillside SWC, and 29% was for road maintenance, gully treatment, and communal pond maintenance. The baito has a bylaw that punishes villagers who do not participate in the collective work without any legitimate reasons. This bylaw, called sirit (Tigrinya), is codified in every hamlet.

The third feature of the Tigray participatory approach is a unique local administration system that connects the district (woreda), village (tabia), and hamlet (qushet), called the baito administration system. The cadres of the village and hamlet levels are elected, and the village representatives become a member of the district baito administration. Disputes between hamlets and conflict in a catchment between upper and lower streams are resolved at an upper-level baito administration. The primary factors of the successful campaign-based watershed management program in Tigray were a democratic local administration system that planned a micro-watershed management program and the baito labor quota system that implemented the work [79].

Kumasi and Asenso-Okyere [80] surveyed villagers’ perceptions of the baito system from twenty hamlets in three districts ¹⁵. The hamlet dwellers were connected directly or indirectly to each other through the network of information on collective work. Those who were at the center of the information exchange were (i) the Ethiopian Orthodox priest, (ii) the development group leader, and (iii) the baite leader at the hamlet level [80]. The construction of social capital through the institutionalization of the indigenous hamlet-based labor quota system and incorporation of the hamlet into the local administration system is part of successful CPR management in Tigray [82] ¹⁶.

3. Materials and Methods

3.1. Preliminary Survey

Boset district (1500–2200 m a.s.l.; annual rainfall, ~800 mm) comprises 33 administrative villages (kebele) with a population of ~175 thousand, an area of ~1.5 thousand km², and a population density of 119 person km⁻² (2014). The central town of the district, Welenchiti, is on the national road connecting the capital, Addis Ababa, the main town of the Oromia region, Adama, and Djibouti. Sedentary Amhara and Oromo are juxtaposed in the district. A village is composed of village territorial divisions and organizations, such as (i) the former village before merging, (ii) the iddir, and (iii) the hamlet (gott)¹⁷ [31].

Related to the first objective of this study, i.e., to elucidate the village organizations managing collective natural and life resources, a preliminary survey was conducted targeting nine villages in Tebo and Geldia seasonal river catchments (Figure 2). The uppermost part of the catchments is in the Tebbo Gersa mountainous area on the southeastern edge of the central Ethiopian Highlands (true highland area; 2100–2200 m a.s.l.). The remaining parts of the catchments contain representative variations of the semi-arid Ethiopian Rift Valley, such that it can be classified into (i) the north-western uplands (Rift margin; 1800–2100 m a.s.l.) and the southern lowland (Valley floors; 1500–1800 m a.s.l.) in terms of representative vegetation reflecting rainfall¹⁸ and geographical characteristics [84]; and (ii) mid-altitude dry (1500–1600 m a.s.l.) and mid-altitude moist sub-areas (1700–2000 m a.s.l.) in terms of the major maize production zones in Ethiopia [85] and major crops grown¹⁹ [86]. (iii) In terms of the distance from the individual village to the district market town, Welenchiti, the minimum is ~4 km (Donkori), and the maximum is ~15 km (Koka Gifawasen). Benin and Pender [47] assessed the determinants of two dependent variables: (i) whether restricted grazing areas existed in the village and (ii) whether violations of grazing restrictions occurred in the Amhara region. They used the following independent variables: (iii) average elevation of the sample village, (iv) annual average rainfall of the village, and (v) average distance from the village to the district market town. They found that (iii) did not have a significant impact on (i) or (ii), but (iv) had a significantly positive impact on (i) and (ii) (both p < 1%), and (v) had a significantly negative impact on (i) (p < 10%) and positive impact on (ii) (p < 5%) [47]. Thus, regarding geographical characteristics, rainfall, the major crops grown, and the distance to the district market town, this survey area was representative of the semi-arid Ethiopian Rift Valley portion around the Adama and Boset districts.

Semi-structured interviews were conducted to ascertain the cases of natural and life resource management and collective work since 1991 (during the EPDRE regime) in 2013. In Boset, PAs, iddirs, and youth and women’s associations are the main stakeholders in natural resource management and small-scale infrastructure construction programs [31,32]. A snowballing sampling strategy was used [87], initially targeting the cadres of these organizations and associations, and adding informants later if necessary. Interviews were also conducted with district administration personnel and international NGOs that had an office in Welenchiti. This preliminary survey found that village organizations and groups managing natural and life resources and small-scale infrastructure construction in Boset are mainly peasant associations (PAs), iddirs, youth and women’s associations, and user groups (

Table 2. Village organizations managing natural and life resources and controlling collective work (activity-wise) ¹.

Activities	Village Organization
	PA	Iddir	PA + Iddir	User Group	Y/W Association 2	Total
Pond (drinking/daily life use) construction	3 (3)	3 (2)	0	7 (2)	0	13 (7)
Pond (drinking/daily life use) maintenance	1 (1)	7 (4)	1 (1)	16 (4)	0	25 (10)
Pond (animal use) maintenance	0	1 (1)	0	0	0	1 (1)
Road repair (partially gully treatment)	10 (8)	2 (2)	6 (2)	2 (2)	0	20 (14)
DTW (drinking/daily life use) construction 3	1 (1)	4 (4)	6 (2)	1 (1)	0	12 (8)
DTW (drinking/daily life use) maintenance 3	1 (1)	0	1 (1)	0	2 (2)	4 (4)
Tebo River cleaning and maintenance 4	0	1 (1)	0	0	0	1 (1)
School reforestation	3 (3)	1 (1)	4 (1)	0	2 (2)	10 (7)
Church reforestation	0	9 (5)	0	0	0	9 (5)
Cropland SWC (soil/stone bunds) 5	3 (3)	0	0	0	0	3 (3)
Bare land and hillside SWC (bench terrace, etc.) 5	5 (5)	1 (1)	0	0	0	6 (6)
Bare land exclosure	0	0	0	0	2 (2)	2 (2)
Hillside management (exclosure/reforestation)	1 (1)	0	0	2 (2)	9 (5)	12 (8)
Grazing lands management	7 (7)	0	0	0	0	7 (7)
Gully treatment and maintenance	2 (2)	1 (1)	6 (2)	0	0	9 (5)
Total	37	30	24	28	15	134

¹ Figures in the table are numbers of organizations, and figures in parentheses are numbers of villages. ² Y/W association = youth and women’s associations. ³ DTW = deep-tube well. ⁴ In the villages, where a deep-tube well is not set, water in the Tebo River is a drinking and daily life water source. ⁵ SWC = soil and water conservation.

). In Table 2, PA + iddir indicates the activities collaborated on by a PA and iddirs. In many collaborative cases, after the district administration instructed or advised a PA about a development program, the PA discussed it with iddirs²⁰. Then, the iddirs with funds and a labor quota system implemented it [31]. The district cooperative office urged every village to establish youth and women’s associations composed of landless and deprived household heads in 1999. It tried to give them income-generation opportunities while staying in their villages²¹. User groups are small groups organized by local dwellers who live near the CPRs and use them collectively, but the district administration does not recognize them.

In the nine sample villages, some natural and life resources were managed by consistent organizations regardless of the sample village (Table 2), for example, (i) most of the cropland SWC (soil and stone bunds) and bare land SWC (bench terrace with trenches, etc.) measures were managed by PAs; (ii) most hillside and bare land exclosure and reforestation areas were managed by user groups or youth or women’s associations; (iii) collective grazing land management other than hillside exclosure by PAs; (iv) reforestation in the church compound was managed by iddirs; and (v) gully treatment and maintenance were managed by PAs and/or iddirs. In contrast, the village organizations managing other natural and life resources, such as communal pond construction and maintenance, road repair, deep-tube well construction and maintenance, and school reforestation, were inconsistent (Table 2).

3.2. Study Area and Methodology

To fulfill the first objective of the study, the following two types of Ethiopian studies were replicated. First, whether collective resource management was conducted or not (binary dependent variables) was predicted by the properties of each study village (population, size, distance to markets, etc.), mainly using binary probit models [25,47,70]. The second study analyzed the factors that affected each household of the village community participating in collective work using several regression analyses [33]. Girmay [33] considered the frequency and motivation of participation in collective work for resource management to be non-unanimous because a community comprised various villagers with diversified socioeconomic backgrounds. He surveyed two hamlets in Tigray (the total numbers of households were 75 and 78), performing a multiple regression analysis to predict non-compliance with the local labor quota systems (the baito system) for collective resource management (a ratio of under-contribution to expected maximum contribution; dependent variable). The independent variables were each community member’s socioeconomic attributes (household head’s age, sex, education, farmland holding, livestock ownership, labor endowments, etc.).

Considering the slow pace of communal forest and grassland conservation in the Boset district, this study examined all the natural and life resource management, in contrast to the previous studies in Ethiopia, which specialized in analyzing only collective woodlots or grazing lands management. Targeting all the activities shown in Table 2, first, multivariate analyses were performed to examine which factors strongly affected whether natural and life resources were conducted in the village. Second, the nine target villages were categorized into several clusters using principal component analysis and hierarchical cluster analysis to ease the interpretation of the regression analyses.

Among the natural and life resource management measures adopted in the study area (Table 2), pond construction and maintenance, road repair, deep-tube well construction and maintenance, Tebo River cleaning and maintenance, and gully treatment and maintenance required relatively large-scale collective earthwork involving from several dozen to more than one hundred workers over several days. Iddirs are the only village organizations with coercive power to force household heads to participate in the collective work through a by-law and penalty [31]. Thus, 21 iddirs encompassing the nine villages in Tebo and Geldia River catchments were targeted, of which 13 iddirs (62%) have experienced large-scale collective earthwork since 1991. Whether the iddir undertook this type of earthwork was assigned as the dependent variable (earthwork) for the multivariate analysis (

Table 3. Variables used for canonical discriminant analysis and descriptive statistics (n = 21).

Variables	Definition	Descriptive Statistics
		Mean 1	S.D.	Min.	Max.
Dependent variable:
Earthwork	Whether experienced collective earthwork since 1991? (dummy: no 0; yes 1)	0.62 ns	0.50	0	1
Independent variables:
Distance	Distance from Welenchiti (km)	8.90 *	4.35	2.82	17.00
Gotts	Hamlet (gott) numbers composing the iddir	2.3 **	1.3	1	4
IddirsNo	Iddir numbers in the village	3.1 **	1.4	1	5
Water	Number of communal water sources maintained by the iddir	0.4 ***	0.6	0	2

¹ The t-tests between the means of the two groups (earthwork = 0 and = 1): * p < 0.1, ** p < 0.05, *** p < 0.01, ^ns not significant.

). Referring to the previous case studies in Ethiopia [25,47,70], independent variables were selected (Table 3). These variables were independent of each other (independent samples t-test). Considering the relatively small sample size (n = 21) and the t-tests between the means of the two groups (earthwork = 0 and = 1), only distance, gotts, iddirsno, and water were selected as independent variables, and a canonical discriminant analysis was performed. The minimum size recommended for a discriminant analysis is five observations per independent variable [88].

To investigate the indigenous labor quota system in the semi-arid Ethiopian Rift Valley, i.e., the iddir system, this study conducted two surveys in 2013. First, targeting the 21 iddirs encompassing the nine villages in the Tebo and Geldia River catchments, interviews were conducted mainly with iddir cadres. Second, Warka and Odalega iddirs in Merko Odalega village implemented collective work for natural and life resource management after 1991, and participant lists were well organized. Warka iddir maintained the road connecting Warka gott and Welenchiti for four days during February and March 2012. It also repaired the road broken by the gully expansion for 12 days between June and September 2012. Odalega iddir weeded a communal pond and dug a waterway to the pond for three days during March and April 2012. It also repaired the road connecting Odalega gott and Welenchiti, which was broken by gully expansion, for ten days between May and August 2012. The iddirs planned and implemented all collective work; the administration or NGOs provided no FFW. Work participation lists kept in two iddirs were referred to. A structured interview was conducted targeting all household heads of the two iddirs (115 households for Warka and 122 for Odalega) to ascertain their socioeconomic attributes.

This study designated the ratio of under-contribution to expected labor contribution as the dependent variable (non-compliance;

Table 4. Variables used for the multiple regression analyses and descriptive statistics (n = 115, Warka iddir; and n = 122, Odalega iddir).

Variables	Definition	The Means (S.D.)
		Warka	Odalega
Dependent variable:
Non-compliance	The ratio of under-contribution to expected labor contribution	0.72 (0.17)	0.80 (0.21)
Independent variable:
Iddir	Iddir dummy: Warka 0, Odalega 1	---	---
Age	Household (HH) head age	0.76 (0.05)	0.78 (0.07)
Gender	HH head gender dummy: male 0, female 1	0.2 (0.4)	0.3 (0.5)
Education	Years of school education (the HH head)	1.5 (2.3)	0.3 (1.1)
Non-Agri	Total days in a week HH members working for non-agricultural jobs	0.3 (1.1)	0.2 (0.7)
Dependent	Dependency ratio (consumer units/producer units)	3.5 (2.1)	3.7 (2.5)
Female labor	Numbers of female HH members, ≦15 and < 55	1.1 (0.5)	1.1 (0.7)
Male labor	Numbers of male HH members, ≦15 and < 65), including regular agricultural labor	1.3 (0.9)	1.1 (1.1)
Land	Cropland holdings (ha) 1	3.0 (2.4)	2.2 (2.1)
Livestock	Tropical livestock unit of livestock owned	3.4 (3.8)	3.6 (6.9)
Cadre	Is the HH head an iddir cadre? (dummy: no 0, yes 1)	0.05 (0.22)	0.05 (0.22)

¹ Surveyed by the district agricultural office in 2007 for land registration.

). Eleven independent variables were selected referring to the previous case studies in Ethiopia (Table 4). First, the Pearson correlation coefficients between each pair of the dependent and independent variables were obtained. Second, regression analyses were performed with two models: one was the linear least squares model, and the other was the two-limit Tobit model considering the histogram of the non-compliance ratio was censored data (Figure 3). Third, categorical principal component analysis was performed to ease the interpretation of the regression analyses.

4. Results

4.1. Natural and Life Resource Management in the Boset District

In the Boset district, an international NGO (CARE International) implemented a natural resource management program from 1984 to 2001 across the district with FFW payment, containing (i) cropland SWC, (ii) hillside and bare land SWC, and (iii) hillside exclosure and reforestation (Figure 1). Cropland and degraded land (hillsides and bare lands) SWC is small-scale collective work involving between about ten and several dozen laborers. Thus, a PA directly selected laborers from the poor living near the site.

A free-grazing system locally called godantu (Oromo) is prevalent in Boset. Area exclosure is necessary to protect vegetation on communal hillsides from village livestock and large herds of camel brought from the neighboring district by nomads. However, hillside exclosure and reforestation have been implemented slowly in Boset. The district agricultural office assigned a PA to assume hillside management (hillside SWC, exclosure, and reforestation). Since 1992, PAs in Boset reforested degraded hillsides in 1992, 1993, 2000, and 2001. However, because no hillside was exclosed, the survival rate of planted seedlings was virtually zero. The agricultural office criticized the PA, saying PA members were interested in cropland SWC but not in hillside conservation (interviews with several district office personnel). These interview findings are consistent with the change by the district administration from a hillside conservation policy to a pro-poor one since the beginning of the 2000s. They instructed the PAs to entrust hillside management and reforestation to the youth and women’s associations. In some PAs, the associations began hillside exclosure and reforestation in 2000. Assumed to be responsible for the hillside management and reforestation, the associations were allowed to possess grass and trees. They hired a guard to prevent non-association members from cutting grass and trees and to prevent livestock from entering the exclosure area. They cut naturally grown grasses in November and sold them at the market.

Since 2003, the hillside management programs were implemented across the district with cropland and bare land SWC, which was financially supported by an international NGO (World Vision) until 2005. FFW payment was supplied for all the work. Between 2000 and 2002, the youth and women’s association members were motivated only by exclusively possessing natural resources on the hillsides. Since 2003, they have been also motivated by FFW payment. From 2000 to 2002, 13 hillsides in Boset were exclosed by youth or women’s associations and were reforested without FFW payment. In 2004, 18 hillsides were exclosed and reforested with FFW. For a tree seedling plantation, a small-scale water-harvesting structure was dug. Besides youth or women’s association members, a PA appointed laborers, numbering between about ten and several dozen, from the poor. A comparison of 2002 (n = 13 hillsides) and 2004 (n = 18) hillside reforestations showed that, when FFW was provided, the total number of seedlings planted per reforestation site was significantly higher, i.e., about three times greater than that without FFW (

Table 5. Comparison of reforestation between FFW payment in 2002 and without FFW in 2004 ¹.

	No FFW (n = 13)	With FFW (n = 18)
Number of seedlings planted per reforestation site	5581 ***	18467 ***
Number of seedlings per hillside area (ha−1)	619 *	813 *
Participants per reforestation site	232 ***	935 ***
Participants per hillside area (person ha−1)	20 ns	23 ns
Number of seedlings planted per participant	44 ***	101 ***
Working days per reforestation site	9.1 ns	8.5 ns

1 Figures are the means. * p < 0.1, *** p < 0.01, ^ns not significant. Source: [89] (p. 154, Table 4.6).

). In proportion to that, the total number of participants per site was nearly four times higher, and the number of seedlings planted per participant was more than two times higher (1% significant level).

However, in 2007, only three hillsides were continuously exclosed in Boset²². On the other hillsides in Boset, they suspended the hillside exclosure and reforestation in 2006. This accounted for (i) the low seedling survival rates (seven cases)²³; (ii) the difficulties of mutual decision-making between the youth or women’s association members because the association members became more interested in other lucrative activities such as sand collection and selling (seven cases); and (iii) the inability to promptly receive tree seedlings from the district nursery center (one case).

Youth and women’s associations usually did not have bylaws. They also did not have any rigid membership rules. District cooperative offices originally selected the association members from landless and deprived household heads and registered them. In 2005, the Merko Odalega youth association had 133 members; the number of household heads was 98 (74%), and that of dependents was 35 (26%;

Table 6. Ages and livestock ownership of Merko Odalega youth and women’s associations’ members (2005).

	Age		Livestock Ownership 1
	Household Heads	Dependents	Household Heads	Dependents
Youth association	29.3 ± 4.6 (n = 98)	22.5 ± 4.4 (n = 35)	2.2 ± 1.8 (n = 98)	5.5 ± 3.1 (n = 35)
Women’s association	45.5 ± 12.0 (n = 16)	34.3 ± 9.0 (n = 49)	2.1 ± 2.7 (n = 16)	5.0 ± 3.0 (n = 49)
Merko Odalega	No data		Mean = 2.9 (n = 352) 2

¹ Units are tropical livestock units (TLUs). TLUs of the dependents represented those of the household heads the dependents belonged to. ² Based on the all-household survey in 2007 [31]. Source: [89] (p. 156, Table 4.8 was modified by adding the mean value of Merko Odalega).

). Merko Odalega women’s association had 65 members; household heads numbered 16 (25%) and dependents numbered 49 (75%). Thus, it was against the original rule; 26–74% of the members were not household heads. The mean numbers of tropical livestock units (TLUs) of the dependent members in the youth and women’s associations were 5.5 and 5.0, respectively, which was much higher than the Merko Odalega mean value of 2.9 (Table 6; data taken in 2007). This indicated that (i) a portion of the household head members were not necessarily from the poor or deprived, and (ii) the dependent members were rather from affluent households participating in the daily activities of the associations.

A national campaign-based watershed management program began in the Boset district in 2012. Boset district was also selected as a project area for all Africa-level natural resource management programs, namely, the drylands development program (DRYDEV, 2013–2018). However, as Assefa [32] found, the effectiveness of the hillside conservation was unclear, and conflicts between the associations and farmers were frequently observed.

Of six hamlets (gotts) comprising Merko Odalega, three hamlet dwellers (Merko, Goro, and Sala) used Merko hillsides for life resource collection and livestock grazing. The three hamlet dwellers showed significantly higher knowledge of and influence on the decision-making about the issue (“Merko hillside Exclosure (2003–2009)”) [31]. This evidence indicated that the district agricultural officers’ observation that “PA members were not interested in hillside conservation” was not necessarily true; instead, it is more likely that Merko Odalega villagers showed resistance against the hillside exclosure policy (three hamlet dwellers using the hillsides) and indifference (three other hamlet dwellers not using the hillsides).

In Tigray and Amhara regions, communal grazing lands are managed by hamlet-based organizations; however, no such organization has been observed in the semi-arid Ethiopian Rift Valley, including the Boset district. First, livestock density in the semi-arid Ethiopian Rift Valley (Boset) was higher than in the semi-arid northern Highlands (Tigray). In 2002, cattle and tropical livestock densities in the east Shewa zone, where the Boset district is located, were 169 heads km⁻² and 468 TLU km⁻², which are much higher than those of the four zones in the Tigray region at the time, i.e., 39–64 heads km⁻² and 165–238 TLU km⁻², respectively (calculated by the author using several statistics). Second, villagers in Boset want to avoid friction with karayu (Oromo nomads) living in the Fantale district, which is the neighboring district to the east. Many farmers in Boset contract with a specific karayu and take their livestock out to Fantale for transhumance in the rainy season. Instead, karayu come to Boset with camel herds for free grazing in the dry season after crops are harvested. Holding a rifle, they walk around villages, and camels often damage live fences (euphorbia leucodendron) surrounding homesteads. Management of grazing lands other than hillsides is exclusively assumed by PAs (Table 2), which includes dispute management with karayu nomads.

4.2. Determinants of Natural and Life Resource Management

4.2.1. Canonical Discriminant Analysis

The canonical discriminant analysis correctly classified 91% of the originally grouped cases. The variables water (number of communal water sources maintained by the iddir), iddirsno (number of iddirs in the village), and distance (distance from Welenchiti) significantly contributed to the group discrimination in this order as positive coefficients (

Table 7. Canonical discriminant analysis.

	Independent Variables
	Distance	Gotts	IddirsNo	Water
Canonical discriminant function coefficients	0.159 **	−0.543 **	0.683 **	2.358 ***
Standardized canonical discriminant function coefficients	0.70	−0.66	0.99	1.18

** p < 0.05, *** p < 0.01.

). Whether an iddir has maintained common water resources (communal ponds or river watering places) since 1991 was the most critical factor determining whether the iddir members experienced working on a large-scale collective earthwork. When a village was composed of more iddirs, and when an iddir was farther from Welenchiti, the iddir more actively undertook collective work. The variable gotts (hamlet numbers forming the iddir) significantly negatively contributed to group discrimination, and the contribution was the lowest (Table 7). When an iddir was composed of more hamlets, the iddir tended not to implement large-scale collective earthwork.

Results showed that the greater the communal water sources in an iddir, the more the iddir tended to undertake large-scale collective earthwork. This finding can be backed up by the following interviews.

Table 8. Maintenance of communal ponds ¹.

	Collective Maintenance Work
	Fencing	Pond Cleaning	Pond Dredging	Canal Maintenance	Hiring Guards
PAs	0	0	0	0	1
Iddirs	8	8	6	1	8
User groups	16	16	0	0	4

¹ Multiple answers were allowed.

shows various collective maintenance works undertaken by each association or group out of the twenty-five maintenance cases for communal ponds (drinking and daily life use; PA, one; iddir, eight; and user group, sixteen; Table 2). Among the collective work shown in the table, the large-scale collective earthworks involving forced participants over several days are pond dredging and canal maintenance. Only six iddirs undertook these types of collective work, and no PAs or user groups undertook them.

The distance from Welenchiti was selected as a less important variable. In the three iddirs in Koka Gifawasen, one of the farthest villages from Welenchiti (Figure 2), the iddirs jointly maintain the Tebo River watering place and feeding roads. On the other hand, in other farthest iddirs in Senkel Kesel and Buta Dalecha, where communal ponds were constructed by user groups, large-scale collective earthwork was not implemented actively. In Warka and Odalega iddirs (Merko Odalega) and Donkori iddir (Donkori), close to Welenchiti, the iddirs maintained communal ponds until the late 1990s, while still actively implementing large-scale collective earthwork (of these three iddirs, deep-tube wells were constructed in the late 1990s in Warka and Donkori; after that, Donkori iddir stopped pond maintenance, whereas Warka iddir still maintained a communal pond for livestock use). The reason why a village composed of many iddirs actively undertook large-scale collective earthwork was not clear. The mean number of hamlets (gotts) comprising an iddir was 2.3 (Table 3). The fact that an iddir is unlikely to actively undertake large-scale collective earthwork when the iddir has more hamlets can indicate the difficulty of mutual consensus building between hamlets.

4.2.2. Principal Component Analysis

Table 9. Village organizations managing natural and life resources and controlling collective work (by village).

Villages	Number of Cases (% of the Total Cases in the Village)
	PA	Iddir	PA + Iddir	User Group	Y/W Association	Total
Senkel Kesel	1 (4)	2 (9)	2 (9)	12 (71)	0	17 (100)
Koka Gifawasen	3 (30)	6 (60)	0	1 (10)	0	10 (100)
Chemeri Jawis	0	7 (29)	17 (71)	0	0	24 (100)
Merko Odalega	6 (33)	6 (33)	0	0	6 (33)	18 (100)
Bekakuto Nume	4 (31)	0	5 (38)	0	4 (31)	13 (100)
Buta Dalecha	6 (26)	4 (21)	0	9 (47)	1 (5)	20 (100)
Buta Bedhaso	5 (38)	2 (15)	0	5 (38)	1 (8)	13 (100)
Donkori	6 (67)	1 (11)	0	0	2 (22)	9 (100)
Buta Wegere	6 (60)	2 (20)	0	1 (10)	1 (10)	10 (100)

shows the village organizations or groups and the natural and life resources they managed, and the controlled collective work in the nine villages, by village. Using the nine datasets of the sample village in the table, principal component analysis was performed to reduce four dimensions (four old variables; in this case, four village organizations—PA, iddir, user group, and youth and women’s associations) to fewer dimensions (new variables, i.e., principal components) that still contained most of the information in the initial datasets. To avoid complexity, PA + iddir was integrated into iddir because actual work in the field was implemented by iddir. Next, considering the nine datasets of the sample village as variables, hierarchical cluster analysis was conducted to classify the sample villages into clusters.

The principal component analysis found that factor 1 and factor 2 explained 51% and 33%, respectively (84% in total), of the total variance (

Table 10. Principal component analysis for the organizations managing natural and life resources ¹.

Factor	Eigenvalues	% of Variance	Structure Matrix
			PA	Iddir	User Group	Y/W Asso.2
Factor 1	2.026	50.6	0.87	−0.96	0.15	0.37
Factor 2	1.335	33.4	0.36	0.15	−0.91	0.74

¹ Bold figures (absolute values of the factor loadings > 0.4) in the structure matrix indicate that the variable highly contributes to interpreting the factor. ² Youth and women’s associations.

). The structure matrix indicated factor 1 represented collective work managed by PAs and iddirs, associations or organizations having between one and two hundred household members. PAs were associated with small-scale collective FFW programs, such as cropland, hillside, and bare land SWC (Table 2). Iddirs were involved in religious and cultural occasions, such as church and school reforestation. Besides those, PAs and iddirs managed large-scale collective earthwork for small-scale infrastructure construction and maintenance, such as school construction [31], road repair (partially gully treatment and maintenance), deep-tube well construction, and pond construction and maintenance. These types of collective work required from several dozens to more than one hundred workers over several days. From 1976 to the late 1990s, the government and donors urged the construction of deep-tube wells in the Tebo and Geldia River catchments. In the areas where deep-tube wells were built, villagers obtained water for drinking, daily use, and livestock use. Alternatively, villagers living far from the deep-tube wells had to rely on communal ponds or the Tebo River for such water sources. Relatively small-scale communal ponds were constructed and maintained by ten to dozens of households living close to an organizing user group. In contrast, large-scale communal ponds with more than several tens of household users were constructed and maintained by PAs and/or iddirs. Thus, factor 1 can represent the organizations that govern large-scale collective earthwork.

Factor 2 represented collective work managed by user groups and youth and women’s associations, and functional groups of up to tens of household members (Table 10). Many user groups were involved in pond construction and maintenance, whereas youth and women’s associations were mainly engaged in hillside management (Table 2). Some user groups, such as the water-harvesting user group in Buta Bedhaso, exclosed the hillside and reforested it to alleviate sedimentation into their communal pond.

Figure 4 (left) shows a scattergram of factor scores on which the nine villages are positioned. Factor 1 indicates that, the more a village is placed on the right side, the more the collective work organized by the PA was active. Conversely, the more it is on the left side, the more the work organized by iddirs was active, according to the signs of the factor loadings (factor 1) in Table 10. Similarly, factor 2 indicates that, the more a village is positioned toward the top, the more the collective work organized by youth and women’s associations was active. Conversely, the closer it is to the bottom, the more the work organized by user groups was active, according to the signs of the factor loadings (factor 2). The nine villages were divided into six clusters based on the dendrogram linked with distance 1 in the hierarchical cluster analysis (Figure 4, right). Referring to the positions of the six clusters in Figure 4 (left), we can call (i) cluster A (Merko Odalega and Bekakuto Nume), PA-youth/women’s associations’ group; (ii) cluster B (Buta Bedhaso, Donkori, and Buta Wegere), PA group; (iii) cluster C (Buta Dalecha), PA-user group’s group; (iv) cluster D (Senkel Kesel), user group’s group; (v) cluster E (Koka Gifawasen), PA-iddir group; and (vi) cluster F (Chemeri Jawis), iddir group. More interesting is the relationship between the geographical positions of the sample villages and the hierarchical cluster analysis. The villages classified into clusters A and B are located in the middle to lower reaches of the Tebo and Geldia Rivers. Youth and women’s associations were established, being instructed by the district cooperative office. Thus, clusters A and B, where PAs and youth/women’s associations actively worked, could be considered the villages in which the district administration strongly intervened. In contrast, clusters C, D, E, and F, where iddirs and user groups worked actively, could be considered the villages where villagers worked on their initiatives.

Rather than geographical locations such as a village in the upper or lower reaches of the catchments, it can be more meaningful to assess the spatial relationship between Welenchiti (the district market town), located in the downstream reaches of the Tebo and Geldia River catchments, and a village. Not only are all the district administrators located in Welenchiti, but also international NGOs’ offices. The village close to Welenchiti can readily obtain the administrative services and information these institutions provide. Conversely, a village far from Welenchiti is also far from the administrative services; villagers had to organize themselves or use indigenous organizations to fulfill basic human needs. The total number of natural and life resource management cases since 1991 (Table 9) averaged 12.2 for clusters A and B, where the administration-oriented organizations took the initiative. In contrast, the average was 17.5 for clusters C, D, E, and F, where the villagers’ self-organization worked, exceeding the former.

4.3. Village Organizations Controlling Collective Work

4.3.1. Iddir Labor Quota System

A difference in labor quota systems between Tigray (the baito system) and the study area (the iddir system) is that all the adult members should participate in the collective work in the baito system [33]. In contrast, at least one adult household member should participate in the iddir system [31]. Of the twenty-one iddirs in the nine study villages, seventeen iddirs (81%) used the iddir system (

Table 11. Basic information of the iddir labor quota system (n = 17).

Items	Details
Total collective working days in a year (person-day)	(i) 0–5 days: 10 (59%), (ii) 5–10 days: 4 (18%), (iii) 10–15 days: 2 (12%), (iv) more than 20 days: 2 (12%)
Penalty for non-compliance households	(i) Force more work: 2 (12%), (ii) let pay them a fine: 13 [1 birr/day: 2 (12%), 5 birr/day: 7 (41%), 10 birr/day: 3 (18%), 20 birr/day: 1 (6%)], (iii) providing no church service: 1 (6%), (iv) no specific penalty: 1 (6%)
Number of exemption households	(i) 0–5 households: 11 (65%), (ii) 5–10 households: 5 (29%), (iii) more than 10 households: 1 (6%)
How cooperative are villagers?	(i) Very cooperative: 14 (82%), (ii) cooperative: 2 (12%), (iii) not so cooperative: 1 (6%)

). Besides collective work for natural and life resource management, these 17 iddirs implemented collective work for small-scale infrastructure construction and maintenance.

In every iddir, 0–10 households were exempted from collective work participation (Table 11). These households were, first, single female household heads due to bereavement, divorce, migrant workers, etc. (Warka, four households; Odalega, twenty-one households); second, deprived youth household heads who had not yet inherited property such as farmland from their parents (Warka, one; Odalega, thirteen); third, elderly or sick male household heads who did not have any other male adult household members (Warka, 0; Odalega, six). Including these households who were exempted from work participation (non-compliance ratio = 1.0), the mean non-compliance ratios were 0.72 for Warka and 0.80 for Odalega; once in 4 to 5 working days, they generally participated in collective work.

4.3.2. Multiple Regression Analyses and Categorical Principal Component Analysis

The independent variables significantly affecting the non-compliance ratio (dependent variable) were almost the same according to the two regression analyses, namely, the linear least squares model and the two-limit Tobit model (

Table 12. Multiple regression analyses (the linear least squares and two-limit Tobit models; n = 237).

Independent Variables	The Linear Least Squares		The Two-Limit Tobit
	Coefficients	t-Value	Coefficients	Z-Value
Iddir	0.0647 ***	2.45	0.0980 ***	3.15
Age	0.4076 **	2.46	0.5384 **	2.17
Gender	0.0374	1.37	0.0655 *	1.87
Education	0.0087	1.13	0.0136	1.43
Non-Agri	−0.0233 *	−2.05	−0.0331 **	−2.47
Dependent	−0.0161 **	−2.19	−0.0182 **	−2.03
Female labor	0.0114	0.50	0.0041	0.14
Male labor	−0.0413 ***	−2.66	−0.0592 ***	−3.31
Land	−0.0110 *	−1.72	−0.0152 **	−1.99
Livestock	0.0022	1.21	0.0029	1.36
Cadre	−0.1564 ***	−2.80	−0.1711 ***	−2.78

* p < 0.1, ** p < 0.05, *** p < 0.01.

). The categorical principal component analysis showed that factor 1, which had about one-quarter of the total information, had significant factor loadings (absolute values) of dependency ratio (dependent), number of male household members (male labor), cropland holdings (land), and livestock ownership (livestock; Table 12). Every pair of these four variables had significant (1% level) correlations (Pearson correlation coefficients, r = 0.39–0.56). These households have large cropland holdings and own high livestock levels; thus, they need more male or regular labor. They are affluent and have a high dependency ratio based on a higher agricultural income. Variable iddir in factor 1 had a negative signal, slightly representing Warka iddir. All of these four variables, dependent, male labor, land, and livestock, also had significantly (1% level) negative correlations (r = −0.41–−0.17) with the non-compliance ratio (non-compliance). Factor 1 represented affluent households, many of whom were in Warka iddir. They actively participated in collective work. The multiple regression analyses supported these principal component analyses (Table 12); other than livestock, which was the non-significant variable, the higher the coefficients of the other significant variables, dependent, male labor, and land, the lesser the non-compliance ratio (non-compliance).

Factor 2 had attributes of greater household head age (age was positive and had a high absolute value) and low education background (education was negative and had a high absolute value;

Table 13. Categorical principal component analysis.

Factor	Eigenvalues	% of Variance	Structure Matrix (Variables)
			Non-Compliance	Iddir	Age	Gender	Education	Non-Agri	Dependent	Female Labor	Male Labor	Land	Livestock	Cadre
Factor 1	3.062	25.5	−0.62	−0.22	0.05	−0.43	0.36	0.15	0.77	0.43	0.73	0.68	0.63	0.33
Factor 2	1.806	15.0	0.21	0.45	0.64	0.31	−0.61	−0.39	0.14	0.33	0.00	0.39	0.35	−0.36
Factor 3	1.169	9.7	0.08	−0.26	0.56	0.24	0.39	0.08	−0.34	−0.40	−0.17	0.37	0.21	0.28
Factor 4	1.067	8.9	−0.00	0.71	0.16	−0.20	−0.02	0.03	0.14	−0.30	0.03	-0.24	0.05	0.58

). Iddir, which was the variable subsequently having a higher factor loading, had a plus sign, representing Odalega iddir. Non-compliance was positive. Factor 2 can represent deprived youth household heads and elderly or sick male household heads, many of whom were in Odalega iddir, being exempted from participation in collective work. The multiple regression analyses showed that iddir and age had significant positive relationships with non-compliance, supporting the principal component analysis. Factor 3 had attributes of elderly widow household heads (age and gender were positive and had high absolute values) and low household labor force (female and male labor). Non-compliance was positive. Factor 3 can represent widow household heads being exempted from participation in collective work. When factors 2 and 3 are combined, they account for about one-quarter of the information and represent households exempted from collective work.

Factor 4 can have an attribute of integrating two variables, iddir and cadre (iddir cadre household head; Table 13). Cadre and the non-compliance ratio (non-compliance) had a significantly (1% level) negative correlation (r = −0.26); iddir cadres had low non-compliance ratios. The mean non-compliance ratios of four iddir cadres, each of the Warka and Odalega, were 0.44 and 0.29, respectively. The multiple regression analyses supported these; the cadre significantly (1% level) negatively affected the non-compliance ratio (Table 12). Iddir in factor 4 was negative (Table 13). Considering these, factor 4 can represent iddir cadres, especially Odalega iddir cadres, who actively participate in collective work.

5. Discussion

5.1. Labor Quota System in the Semi-Arid Ethiopian Rift Valley

Although the government and donor agencies have sporadically implemented natural resource management programs in the semi-arid Ethiopian Rift Valley, villagers have self-organized and carried out natural and life resource management and collective work. In both the semi-arid Ethiopia regions of Tigray and Boset, representative hamlet-level indigenous organizations commonly build villagers’ consensus and mobilize them for collective work using labor quota systems, baito and iddir, respectively. In addition, iddirs in the semi-arid Ethiopian Rift Valley have funds, sometimes providing a portion of them for small-scale infrastructure construction programs [31].

However, the two labor quota systems have differences in customary practices, such as villagers’ frequencies in work participation and the significant factors that increased the non-compliance ratios. In Tigray, villagers participated in collective work a mean of 35–51 persons × days per year under the baito system [33], which was considerably higher than the 5–20 persons × days per year under the iddir system (Table 11). Besides that, the mean of the non-compliance ratio under baito was 0.37, which was significantly lower than those under iddir, of 0.72 and 0.80.

In the two hamlets surveyed by Girmay [33], 13–26% of the survey households participated in all expected labor contributions, whereas 50–65% contributed to less than half of those expected; there was a large gap in the non-compliance ratio between households. The significant factors that increased the non-compliance ratio were, first, men and women adult numbers in the sample household. The diagram of household adult numbers (x-axis) and non-compliance ratio (y-axis) presented a U-shaped curve; the greater the number of adult members in the household, the higher the tendency of the non-compliance ratio. The second significant factor that increased the non-compliance ratio was household livestock ownership. Communal grazing land management and exclosure reduce free-grazing land for livestock. Instead, someone in a household had to spend more time preparing livestock feeds (crop residues and cut-and-carried grasses). Girmay [33] considered that the present baito system, under which every adult participates in the work, imposed a substantial burden on large family households and households with a high livestock holding (mainly wealthy households). In contrast, the iddir system requires only one adult from a household to participate in the work, and the wealthy family positively participates in the collective work.

The baito has a bylaw that punishes non-compliant villagers; however, the bylaw was not strictly applied to all the non-compliant villagers [33]. Consistent non-compliant villagers are penalized in the baito and iddir systems, but minor non-compliance is usually overlooked. Overall, labor quota systems in both areas are welcomed by villagers ([80] and Table 11). The proportions of household heads who favored the baito system, regarding it as an effective form of natural resource management, were as high as 62% and 96% in the two hamlets in Tigray [33]. In Boset, about 80% of the study iddirs used the iddir systems; more than 90% replied that villagers were cooperative or very cooperative (Table 11). In the two iddirs studied (Warka and Odalega), iddir cadres and the affluent iddir members, which were both influential in the hamlet [31], positively contributed to the collective work. Economically wealthy households tend to be politically influential in the hamlet [31] and positively respond to expected labor contributions. This can motivate all hamlet members to work. At the same time, both systems institutionalize the consideration for the poor and socioeconomically vulnerable people in the form of work exemption.

In the Boset district, where the labor quota system is still at a conventional level in an area that the regional government does not institutionalize, five iddirs (24%) of the twenty-one study iddirs either did not use the iddir system or replied that villagers were not so cooperative (Table 11). When an iddir comprises many hamlets (gotts), or in the iddirs where communal pond construction and maintenance was implemented by user groups or youth/women associations in the past, the formation of a consensus on collective work initiated by the iddir may be complicated. Alternatively, in the iddir, where a deep-tube well has already been built, the maintenance of a communal pond tended to be neglected, lowering the incentive to participate in collective work. In the semi-arid Ethiopian Rift Valley, iddirs are the most appropriate village organization that governs villagers’ collective work for rural development; however, consideration should be given to deciding on the type of village organization that is the most appropriate after interviewing participants about past experiences of collective work and consensus-building processes in an individual iddir.

5.2. Local Administration System in the Semi-Arid Ethiopian Rift Valley

Comparing the participatory approaches of the Tigray region and Boset district (Oromia region), the most significant difference is in the relationship between district (woreda) administration, village organizations, and donor agencies. In Tigray, besides their customary labor contribution, FFW is supplied to the villagers participating in the same kind of collective work. The questionnaire survey of 400 households in 16 hamlets across Tigray found that 69% of the respondents experienced FFW for a mean of 89 days (18 days per year) for five years from 1994 to 1998 and received a mean of 262 kg of grains (52 kg per year) [94]. The baito administration decides who participates in the FFW among villagers with a low non-compliance ratio. Thus, the FFW incentivizes villagers to participate in baito collective work [33]. Under the PSNP project launched in 2005, the villagers who had low non-compliance ratios were given priority to participate in the FFW and HABP (Household Asset Building Program) [81]. HABP is a policy providing poor households with income-generation and micro-financial opportunities. Even if donor agencies change their aid priority from community development to an individual/household livelihood approach based on neoliberalism policy, the Tigray region, which has a unique local administration system, responds flexibly and institutionally maintains villagers’ motivation to participate in the baito collective work, thereby supporting Tigray’s past achievements of natural resource management and rural development. Among the baito collective work, the largest work volume is spent on stone bund construction for cropland SWC. Crop fields with stone bund terraces prevent 68% of soil loss [95], increasing crop yield in the semi-arid Highlands of Tigray (see Note 7). This is another advantage to sustaining the baito labor quota system.

Among the natural resource package programs implemented in the Boset district since the 1980s, (i) cropland SWC is controlled by PAs, (ii) hillside conservation is controlled by PAs or youth/women’s associations, and (iii) gully treatment is controlled by PAs + iddirs (Table 2). For (i) and (ii), donor agencies supplied participants with incentives such as FFW or opportunities for income-generation and microfinance programs. For (iii), donor agencies bore only material costs (e.g., gabions). Because the labor force required for (i) and (ii) is at the scale of up to several tens of workers, the collective work was implemented as a poverty reduction program. In contrast, (iii) needs large-scale collective earthwork, requiring from several dozen to over one hundred workers over several days. Any incentives to participate in the work, e.g., FFW, are not supplied. Practically, they need to rely on PAs and iddirs. Similarly, depending on the workload required and whether FFW is provided, iddirs participated, even in small-scale infrastructure construction programs [31]. Thus, in Boset, some incentives provided by donors were exclusively connected to poverty reduction programs (or the individual/household livelihood approach); the iddir system (or community development approach) has never been implemented.

5.3. Institutional Arrangement in Two Semi-Arid Ethiopia Regions

From an institutional arrangement point of view, the two semi-arid Ethiopia regions showed a large difference. In Tigray, natural resource management package programs are consistently implemented in the same CPR user organization, i.e., baito at the hamlet (qushet) level, which is the main decision-making body among villagers. Their governance activities, such as broader planning, coordination between baitos (qushet), monitoring, and conflict resolution, are controlled by the baito administration (district–village–hamlet) (

Table 14. Assessment of the institutional arrangements for managing natural and life resources and controlling collective work in two semi-arid Ethiopia regions (Boset and Tigray).

Scale of CPR Users 1	CPRs	User Organizations	Remarks	Design Principles 2
				1.	2.	3.	4.	5.	6.	7.	8.
Boset (semi-arid Ethiopian Rift Valley)
A dozen to tens	Small ponds. Hillside management. Gully treatment and road repair	User group	User groups and Y/W associations usually do/did not have bylaws. Y/W association did not have clear membership rules.	A	B	B	B	B	A	A	C
Tens to about two hundred	Bare land exclosure and hillside management	Y/W associations 3		B	B	B	C	B	C	C	C
Over about two hundred	Large ponds, deep-tube wells, and river watering places. Gully treatment. Church and school reforestation. Small-scale infrastructure	Iddir (gott)	The regional government does not institutionalize governance activities (coordination, monitoring, conflict resolution, etc.) between district–village–hamlet.	A	A	A	A	A	A	A	C
Tigray (northern semi-arid Ethiopian Highlands)
A dozen to over about two hundred	Water sources. Exclosure forests. Grazing land management. Gully treatment. Small-scale infrastructure	Baito (qushet)	Governance activities are coordinated by the baito administration (district–village–hamlet), which is institutionalized.	A	A	A	A	A	A	A	A

¹ Mean (± standard deviation) number of village (n = 9) households of the study area in the Tebo and Geldia River catchments was 397 (±178) in 2007 [96]. That of hamlets (n = 49) was 74 (±11) in the survey year (2013; by village, household numbers in the statistics were simply divided by the hamlet numbers in each village and calculated). That of iddir (n = 21) was 151 (±46; 2013; data obtained from interviews, mainly with iddir cadres). Thus, (i) CPR users numbering from a dozen to several tens can be within a hamlet scale, (ii) from several tens to about two hundred users can be approximately within an iddir scale, and (iii) more than two hundred users can be within a village or inter-iddir scale. ² Design principles were based on Table 1: 1. User and resource boundaries, 2. Appropriation and provision rules, 3. Collective-choice rules, 4. Monitoring, 5. Sanctions, 6. Conflict resolution, 7. Independence and long-term tenure, 8. Nested institutions. Sign A indicates that activities on sites generally fulfill each principle. Signs B and C indicate partially fulfilled and not fulfilled, respectively. ³ Y/W associations = youth and women’s associations.

). Institutional arrangements in the Tigray region can be said to be robust.

In Boset, CPR user organizations differ depending on the scale of CPR users and the nature of CPRs (Table 14). User groups were organized, particularly in the villages where iddirs were either absent or inactive, which assumed small-scale natural and life resource management and infrastructure maintenance (e.g., road repair). User groups usually have clear boundaries between CPR users and non-users. The youth and women’s associations were formal organizations and once assumed hillside management; however, the boundary between the users and non-users was unclear because they had loose membership rules. Neither user groups nor youth and women’s associations usually had bylaws. Their monitoring and sanctions were weak and they were institutionally fragile. In contrast, an iddir, which is a representative hamlet-level organization that used to be a CPR user organization governing an indigenous grazing land management system, can be said to be a robust CPR user organization.

However, a weak point in Boset is that the regional government does not institutionalize all the governance activities by connecting district administration, PA (village representative organization), and CPR user organization at the hamlet level. This insufficient nested layered structure [1] affects the robustness of all CPR user organizations in Boset. There are two requirements for the planning and implementation of sustainable participatory rural development programs in the semi-arid Ethiopian Rift Valley: (i) incorporation of hamlets (gotts), which govern villagers’ consensus building and have a labor quota system (the iddir system), and funds into the local administration system; and (ii) maintenance of villagers’ incentive to participate in collective work through the iddir labor quota system, which is institutionalized by the local administration to use several incentive programs provided by donor agencies.

6. Conclusions

It is safe to say that the basic human needs in Ethiopia, which is ecologically fragile, have depended on the performance of natural and life resource management programs since the 1970s. However, the programs involve many factors that hamper villagers’ initiatives to participate; for example, cropland SWC lowers farmers’ short-term benefit from crop cultivation in the humid Highlands; hillside exclosure constrains free livestock grazing and villagers’ access to forest resources; and gully treatment requires large-scale collective earthwork over several days. To ensure a project is successful, with a low cost and a broader extent, the government has to incentivize villagers and urge them to participate in natural resource management programs. In the Imperial period, hamlet dwellers were given the privilege of communal land use and access to natural and life resources under the Irist system. In the 1970s and 1980s, the socialist regime abolished the Irist system. The regime’s unclear policy on communal land management motivated the change of part of the communal land into croplands, and its mass-mobilization policy did not incentivize villagers, ending in failure.

Reflecting on the policy failure in the 1970s and 1980s, the Tigray regional government (TPLF) integrated hamlets (qushet), which had an indigenous grazing land management system (hizati) and a labor quota system (baito), into the local administration system. In addition, by combining several of the incentives provided by donor agencies and the labor quota system, the regional government institutionalized the incentive system to encourage villagers to increase their participation in collective work. These institutional arrangements are the primary factors that enable the Tigray region to realize relatively cheap, broader, yet sustainable small-scale infrastructure construction programs, i.e., rural development, and natural and life resource management. In contrast, the Amhara and Oromia regional governments either distributed communal lands to all households or granted the user-rights of the communal lands to formal groups; however, this was with weak entrepreneurship, so the vegetation was not able to be increased on communal lands and, as a result, neither soil loss nor runoff from catchments were able to be reduced.

Iddirs (or kires), hamlet-based village organizations, have a unique labor quota system and funds based on their bylaws. The only requirement for admission is that the household head lives in the hamlet; thus, it is a democratic organization that does not discriminate against the members by religion, gender, economic status, etc. The drawbacks of informal organizations can be modified by integrating hamlets into the administration system and codifying the bylaws of iddirs (kires). Institutional arrangements in the Tigray region fulfill the theory of sustainable CPR management. They can be a good model for rural development in the semi-arid Ethiopian Rift Valley and other parts of Ethiopia.