The role of natural capital in sustaining livelihoods in remote mountainous regions: The case of Upper Svaneti, Republic of Georgia

Ecological Economics 117 (2015) 22–31

Contents lists available at ScienceDirect

Ecological Economics journal homepage: www.elsevier.com/locate/ecolecon

Analysis

The role of natural capital in sustaining livelihoods in remote mountainous regions: The case of Upper Svaneti, Republic of Georgia Robin J. Kemkes ⁎ Department of Economics, University of Massachusetts Amherst, Thompson Hall, Amherst, MA 01003, USA

a r t i c l e

i n f o

Article history: Received 14 January 2014 Received in revised form 24 April 2015 Accepted 9 May 2015 Available online 25 June 2015 Keywords: Common pool resources Rural households Natural capital Property rights Post-Soviet

a b s t r a c t In the Greater Caucasus of the Republic of Georgia, proponents of a new ski tourism zone and long-term timber concessions claim that new wage opportunities will benefit households. These developments will also limit access to common-pool resources (CPRs). This study uses the sustainable livelihoods framework to identify the conditions under which a development strategy will improve livelihood outcomes in the region. Analysis of original household survey data, in-depth interviews, and field observation reveals that households depend on CPRs for a range of market and non-market benefits. Low-income households depend on CPRs for up to 60% of their total income. OLS regression estimates show that households in villages farthest from market centers have a higher income dependence on CPRs and are more likely to participate in forest use activities. A majority of households report that there are few available substitutes. To improve livelihood outcomes, a development strategy should secure access to market benefits from CPRs, or wage income must increase in proportion to lost CPR income and affordable substitutes must be provided. Access to non-substitutable components of CPRs must be secured, and the distribution of changes in access to natural capital and new wage opportunities must be accounted for. © 2015 Elsevier B.V. All rights reserved.

1. Introduction Across the world, large-scale land acquisitions by transnational and national interests to secure increasingly scarce food and natural resources result in the dispossession of land, water, forests and other common properties from local communities (White et al., 2012). Land deals — through sale, lease or coercion — typically take place in areas where property rights are insecure (Borras et al., 2011). Rural regions and their environments are increasingly viewed as a resource to be exploited, and local communities often lack the power to resist (Marsden, 2009). In the post-Soviet context, natural resources are privatized and rural areas are commercially developed in an effort to attract investment and to create employment opportunities. However, these enclosures have the adverse effect of limiting access for locals to resources that are important for their livelihoods. A multitude of recent studies, including a meta-analysis covering 51 cases across 17 developing countries in East Africa, Southern Africa, Asia and Latin America (Vedeld et al., 2007) and a global survey across 24 developing countries (CIFOR: Center for International Forestry Research, 2011), have established that flows of income — primarily fuel wood, wild foods and fodder — harvested from common pool resources (CPRs) such as forests, meadows and pastures, make up a significant share of total income for rural households. On average, forest ⁎ Northland College, 1411 Ellis Avenue South, Ashland, Wisconsin 54806, USA. E-mail address: [email protected].

http://dx.doi.org/10.1016/j.ecolecon.2015.05.002 0921-8009/© 2015 Elsevier B.V. All rights reserved.

environmental income comprises about 20 to 22% of total household income (Vedeld et al., 2007). This overwhelming evidence of dependence demonstrates potential synergies between the sustainable management of resources and poverty alleviation (Sunderlin et al., 2005) or at the very least, the role that maintaining forests plays in preventing a rise in poverty (Wunder, 2001). At the same time, 86% of forests and wooded areas across the world are formally owned by central governments (Agrawal et al., 2008). Therefore, the degree to which the state protects, manages use, restricts access or extracts raw materials from CPRs also affects rural households that depend on CPR income for their livelihoods. On the one hand, the state can support CPR use by local communities, while on the other hand, it can undermine livelihoods by restricting access or by allowing or promoting unsustainable use of CPRs. The forests of the former Soviet Republic of Georgia are of global importance. They make up part of Conservation International's Caucasus biodiversity hotspot, and many areas, inaccessible deep in the mountains, are rare cases of intact temperate zone forests (Forest Law Enforcement and Governance (FLEG), 2010). These forests serve important environmental functions such as providing habitat for endangered species, including the West Caucasian tur (Capra caucasica), and mitigating global climate change by sequestering and storing carbon. Deforestation accounts for 10 to 15% of annual greenhouse gas (GHG) emissions (Van der Werf et al., 2009). The cultural heritage and traditional property regimes of Upper Svaneti, a district in the Greater Caucasus of Georgia (see Fig. 1), are

R.J. Kemkes / Ecological Economics 117 (2015) 22–31

23

Fig. 1. Upper Svaneti district in the Republic of Georgia. Modified by author from De Waal (2011).

also important in their own right. Georgia is home to three UNESCO World Heritage sites, including the village of Ushguli in Upper Svaneti, which has a preserved landscape of medieval-type stone houses and towers. For centuries, the region has been settled by the Svans who, even during feudal times, maintained their independence. Farms were collectivized during the Soviet period, but villages maintained historical property boundaries by oral recordkeeping. After the collapse of the Soviet Union, meadows and cropland that were part of collective kolkhoz farms were redistributed by villagers according to traditional village and family ownership. Although the forests and interspersed pastures are currently managed as part of the state Forest Fund (the total stock of forest assets in the country) by the Ministry of Energy and Natural Resources, informal traditional boundaries still exist on these lands, too. In Upper Svaneti, local state-employed forest rangers identify trees to be harvested within informal, traditional village forest boundaries, and within villages, households adhere to traditional forest boundaries by not trespassing in or harvesting fuel wood or nontimber forest products (NTFPs) from non-family forests. Two state-led developments are now jeopardizing household access to forests, pastures and meadows in Upper Svaneti. First, the construction of a commercial ski tourism zone is creating conflict over cropland and meadows held under traditional ownership and threaten environmental degradation, such as deforestation and erosion. Second, the revision of

the state Forest Code to allow 49-year concessions to forests — including their underground water and mineral resources — to be awarded to private companies will limit access to forests and interspersed pastures for locals and damage ecosystem functions if enforcement of management plans does not improve. State officials cite increased job opportunities in the tourism and timber industries as benefits for the communities in the district, suggesting that opportunities for wage income from the planned development are a sufficient substitute for lost CPR income and non-market benefits. Before these developments were initiated by the state, it had designated a large portion of the Upper Svaneti district as a planned Protected Area, which would have taken into consideration traditional use zones and placed restrictions on long-term concessions and resorts. Georgia has an extensive system of 64 Protected Areas developed through a project initiated by the World Bank on a grant from the Global Environment Facility (GEF) in 2002. The district of Upper Svaneti was included in this project, but since then, the Ministry of Environment and Natural Resources has been restructured and charged with a new mission of capitalizing on Georgia's forest resources and scenic beauty rather than protecting them, as part of a new development strategy to attract investment. The purpose of this study is to identify the conditions under which a rural development strategy will sustain or improve livelihoods for

Fig. 2. Sustainable livelihoods framework. Adapted from Carney (1998).

24

R.J. Kemkes / Ecological Economics 117 (2015) 22–31

households in Upper Svaneti. Over the summer of 2011, I gathered data across the district through a mixed-methods approach using a sustainable livelihoods framework. The sustainable livelihoods framework was developed as a guide to analyze household assets and strategies in order to reveal the trade-offs associated with diverging development paths (Bebbington, 1999; Carney, 1998; Scoones, 1998) and to focus on understanding the complex, local realities affecting development outcomes (Scoones, 2009). The framework includes information about stocks of capital — built, financial, natural, social and human. Households employ these types of capital to achieve development outcomes such as increased income, improved well-being, increased food security, more sustainable use of natural resources and social inclusion. Institutions and policies mediate household access and use of resources (See Fig. 2). Natural capital, one of the key components of the livelihoods framework, is a stock that produces a flow of services that satisfy human needs directly or indirectly (Ekins et al., 2003). The ecosystem functions that produce these services fall under four categories: regulation functions that produce services such as climate stability and water purification; production functions that provide goods such as food and raw materials; habitat functions that provide wild plants and refuge to animals; and opportunities for recreation, aesthetic enjoyment, cultural information and spiritual inspiration (De Groot et al., 2002). Some of the components or flows of services from natural capital are nonsubstitutable in that they contribute to livelihood development outcomes in a way that cannot be replicated by any other type of capital (Ekins et al., 2003). The following analysis demonstrates the level of benefits provided by forests, pastures and meadows as identified and reported by households in the district. Further research evaluating regulation functions that provide indirect benefits such as erosion control and avalanche protection that support subsistence agriculture, for example, would provide an even more comprehensive assessment of the importance of natural capital to livelihoods. Households in Upper Svaneti report benefiting from flows of goods and services produced across the four types of natural capital functions: fuel wood, crops and fodder for which local market prices can be derived; spring water, NTFPs and wild game provided through regulation and habitat functions, which are not sold in local markets; and leisure and the provision of sites for cultural and religious practices, which do not hold market value. Because maintaining CPRs in regions where households depend on them for their livelihoods is vital for preventing a rise in poverty, assessing household income dependence on forests, pastures and meadows is an important first step in assessing how different development paths might affect households in the district. Across studies, measures of CPR dependence have been interpreted as the degree to which households are destabilized by limited access to natural capital or the degree to which other sources of income would have to increase to maintain livelihoods if access is limited or quality is degraded (see Narain et al. (2008) for a discussion of interpretations). Thus, household income dependence is a measure of how much income the new tourism and timber industries would have to provide to sustain the current standard of living in Upper Svaneti to compensate for limited access to CPRs, at current prices and access to markets. In the analysis I also employ two ordinary least squares (OLS) regression models with village fixed effects to assess household determinants of CPR income dependence and participation in non-market forest use activities. I find that low income households and households in villages farthest from the planned ski tourism zone are most income dependent and more likely to participate in forest use activities. These results provide insight into how different development paths might affect the level of inequality across the district. I first provide a recent history of economic development in Georgia, and I describe the current state of forest governance and property rights in Upper Svaneti. I then elaborate on my data collection methods and income calculation. Next, I report the results of the analysis, including summary statistics, household income dependence on CPRs and

participation in forest use activities, and household determinants of income dependence and forest use. I conclude with a discussion of the degree of substitutability of the services derived from CPRs in Upper Svaneti and the conditions under which rural development would sustain or improve livelihoods in the district. 2. Georgia in Transition 2.1. Development Indicators Although the states of the former Soviet Union exhibit differing levels of development and political, environmental and social characteristics, the Caucasus republics are still struggling for economic stability (Agyeman and Ogneva-Himmelberger, 2009). In Georgia, GDP grew by 7% in 2011 when this study was conducted, despite the global economic downturn, a big improvement since 2009 when rates were negative following the 2008 war in South Ossetia. However, Georgia is still a lower-middle income country and remains poorer than most of its post-Soviet neighbors. Gross National Income (GNI) per capita was US$2860 in 2011, compared to US$3360 in Armenia, US$5290 in Azerbaijan and US$3120 in Ukraine (World Bank, 2011b). Liberalization policies, such as deregulation and privatization of public lands, have been promoted in an effort to expand the economy. Georgia is now ranked in the top twenty of the World Bank's Ease of Doing Business Index (World Bank, 2011a). Since the Rose Revolution in 2003, the government has adopted what has been termed the “Singapore” model, under which the primary objective of policy decisions is to attract foreign direct investment (De Waal, 2011). This policy agenda has led, over the past decade, to Georgia becoming more “monopolized” than “marketized” (Jones, 2012, p. 183), and it has not addressed the pressing problems of rural poverty or unemployment (De Waal, 2011). The 2008 National Human Development Report found that broad economic reforms implemented in 2004 have resulted in strong GDP growth, but they still have not reduced poverty or extreme poverty (Welton et al., 2008). Relying on market liberalization to spur growth can degrade natural resources upon which rural communities depend, thereby exacerbating the very problem of poverty that increased economic growth claims to address. The 2009 World Bank Poverty Assessment Report described rural poverty as Georgia's most serious socioeconomic concern (World Bank, 2009). Rural households account for 47% of the country's population and 60% of those living under the poverty line (CARE International in the Caucasus, 2010; World Bank, 2011b). Moreover, wealth inequality has been increasing between urban and rural areas (Agyeman and Ogneva-Himmelberger, 2009; De Waal, 2011), contributing to an increasing power imbalance, which has strengthened the ability of state and corporate actors to impose unwanted development on local communities. Inequality in the distribution of power drives resource degradation (Boyce, 1994). Adding to the precarious nature of livelihoods in rural Georgia, two decades into the transition to a market economy, property rights in rural areas still lack definition and enforceable title. 2.2. Forest Governance in Upper Svaneti Forests are estimated to cover 40% of Georgia's total land area (Food and Agriculture Organization (FAO) of the United Nations Forestry Department, 2010), although no recent inventory has been conducted. Forests, including the land, water and mineral resources contained within them, make up the state Forest Fund. There is a large share of mature forests, 38% of the total forested area, inaccessible in high mountainous areas. The Upper Svaneti district is located in the northwestern-most corner of Georgia in the Greater Caucasus, contiguous with the breakaway region of Abkhazia. The main tree species in the district are Oriental beech (Fagus orientalis), Norway spruce (Picea abies), Nordmann fir (Abies nordmanniana) and Georgian oak (Quercus iberica). The district

R.J. Kemkes / Ecological Economics 117 (2015) 22–31

is also home to 212 species of flora endemic to the Caucasus and fauna included in the International Union for the Conservation of Nature (IUCN) Red List of Threatened Species for Georgia, including two goatantelope species — the West Caucasian tur and the chamois (Rupicapra rupicapra) — the brown bear (Ursus arctos), the Golden Eagle (Aquila chrysaetos) and several species of owl. During the Soviet period, forests were classified as national forests or collective, kolkhoz, farm forests. National forests were exploited by the state for timber harvesting and kolkhoz forests were used to supply local populations with fuel wood and to provide supporting ecosystem functions for agricultural production, such as riparian buffer zones. During the socio-economic crisis of the early transition phase in the 1990s, many former kolkhoz forests were easily degraded due to their accessibility near roads and villages. Eventually, jurisdiction over all former Soviet national and kolkhoz forests was transferred to the Georgian State. Alongside the nominal state management of forests in Upper Svaneti, there exists a sense of village rights, and within village forests, there exists family ownership. Traditional forest boundaries are recognized but not strongly enforced (villager interviews, summer 2011). This overlapping governance structure is typical of forests in developing countries (Agrawal et al., 2008). Under the current regime, households continue to access, use and depend on forests as they have for millennia, while the state retains the right to exclude use and thereby affect livelihoods. For almost all households in Upper Svaneti, firewood is the primary fuel source for cooking and heating. Although use is governed by the state through licenses to harvest fuel wood and timber, forests are in effect an open access resource. Each ranger in the district is responsible for a very large area, approximately 7000 ha, and rangers lack the necessary resources to properly monitor use (personal communication with forest ranger, summer 2011). Even so, according to a local forest ranger, the supply of coniferous trees is large enough to support the fuel needs of the local population sustainably. Villagers can easily obtain a license to harvest fuel wood from the government offices in the district capital of Mestia, whereas a license to harvest timber requires the approval of an application by administrators in the regional Forest Division outside the district. Local people find it almost impossible to obtain a license to harvest timber even for small home and fence repairs. Although it is difficult for the state to exclude timber extraction due to limited monitoring capacity, villagers typically do not harvest timber illegally, as the penalty is high, and sawmills will not process timber without a license. In 2008, the first 20-year timber concessions were auctioned to private companies, including a license to a joint Georgian–Chinese venture in the lower portion of Upper Svaneti. The first report on the performance of the timber companies found that citizens were not involved in the decision-making process, the terms and conditions of the licenses did not comply with the law, and no measures against deforestation were taken (Macharashvili, 2009). 2.3. Contested Property in Upper Svaneti The Svans are an ethnic and linguistic subgroup within Georgia who have inhabited the district of Upper Svaneti for millennia. Throughout history, land ownership has been an important element of Svan culture and identity. Prior to Georgia's annexation into the Russian Empire, territorial communities called khevis were formed according to geographical, historical, and agrarian conditions. These communities were bound by kinship and common interests, such as protecting common property and providing welfare for community members (Topchishvili, 2005). An elected community leader, the makhvshi, regulated use of meadows, pastures and forests, and the redistribution and surveying of private cropland according to patrilineal inheritance rules. Village and extended family councils resolved disputes. Soviet property regimes and modes of production in rural mountainous regions of Georgia were superimposed on traditional community

25

boundaries and kin relations. There were few adjustments to social life under collectivization in these communities because their contribution to national production was relatively insignificant compared to larger kolkhoz farms (Dragadze, 1976). In Upper Svaneti, property was collectivized but village boundaries remained and communities continued to resolve disputes. Although traditional governance structures and social relations in the villages of Upper Svaneti have survived the Soviet era to varying degrees, today they are susceptible to further disruption. After the fall of the Soviet Union, crop and meadowlands in Upper Svaneti were redistributed according to traditional family ownership, but legal registration of crop and meadow plots has never taken place in high mountainous regions of Georgia (Engel et al., 2006). Cropland is typically located just outside the villages, and households maintain traditional private ownership. However, very few households hold secure, formal title to these properties. No market exists for cropland. Most households also have a few meadow plots from which they harvest fodder. Some meadows are located adjacent to crop areas, while others are found at higher elevations. A majority of meadows are fenced, with only one or two households, usually from the same family, harvesting fodder from each meadow. Households do not hold formal title to meadows, but because they exclude use by other households, they have effectively established traditional private ownership. However, households again do not hold title to meadows adjacent to cropland, and meadows at higher elevations are interspersed throughout the state Forest Fund so that formal jurisdiction is unclear. Pastures are also dispersed throughout the forests, but they are not fenced and are used collectively by villagers, thus, they are common pool resources — animals graze freely for three months out of the year. Given the region's steep terrain, there is very little room for agricultural expansion in Upper Svaneti. Only 6.7% of the territory is agricultural land, of which 7% is arable, 9% is meadow and 84% is pasture (Engel et al., 2006). Forests, steep slopes, and village dwellings along the river valley make up the remaining area. With the beginning of the construction of the ski tourism zone, which the Ministry of Economy and Sustainable Development plans to fully execute by 2015, meadowland has become valuable development property. Households have attempted to register land with little success. Several disputes over meadowland have erupted between the state developers and locals since the initiation of the tourism zone. A majority of respondents in the survey, over 80%, indicate they are “somewhat worried” or “very worried” over limited access to meadows, pastures and forests over the next five years. 3. Field Site and Data 3.1. Household Activities in Upper Svaneti The inhabitants of Upper Svaneti are concentrated in sixteen clusters of small villages scattered throughout the Enguri river valley, including the administrative capital of Mestia. The district forms half of the Samagrelo-Upper Svaneti region. The capital of the region, Zugdidi, is located in Samagrelo, just over the southwestern border of Upper Svaneti. According to voter registries, the total population of the district is 8468 people, but the actual population is likely to be lower due to emigration and seasonal habitation. Village populations range from 105 to 1750, and there are approximately 2822 households in the district. Inhabitants speak both the Georgian language and the local dialect, Svan, an ancient oral antecedent of the Georgian language, spoken by only 50,000 people worldwide (Rayfield, 2012). Most households engage in subsistence crop production and animal husbandry to some degree. Cattle breeding is the main livestock activity. Oxen are used for plowing fields and pulling sleds of fodder down from meadows at high elevation and fuel wood from forests. Fuel wood is the primary source of energy for heating and cooking for the majority of households. Fodder is harvested from meadows two times over a summer season

26

R.J. Kemkes / Ecological Economics 117 (2015) 22–31

Fig. 3. Villages in Upper Svaneti where households were surveyed.

and animals are allowed to graze freely in pastures for three months out of the year. Jam and honey production hold high value for households that have fruit trees or beehives on their property. Some households sell crops, jam or honey at the market in the regional capital of Zugdidi. The main road leading to Zugdidi was partially renovated at the time this study was conducted. A small portion of the population in Upper Svaneti is made up of displaced persons from the 1992–93 war in neighboring Abkhazia. Government employment in Mestia is the primary source of wage employment, while some households have members employed in part-time, limited-term road construction. Households do not sell farm labor, however neighbors and family members typically join together for harvesting. 3.2. Data Collection Using the sustainable livelihoods framework, I conducted a survey of 250 households in nine villages across the Upper Svaneti district to collect information about stocks of assets, flows of income and subsistence goods, forest use, and perceptions of the importance of forests to livelihoods (see Fig. 3). The head of the household was self-selected in each home, and was usually the eldest male. Females identified themselves as the head of the household if no adult male resided there. Households were selected using a two-stage sampling procedure. Villages were purposively selected to maximize variability according to distance from the main road, distance from Mestia (the district administrative capital), distance from Zugdidi (the regional capital), elevation and village size.1 Households within the villages were selected through a systematic random walk procedure. Approximately 9% of the households in the district were sampled. Final sample size was 248 households after removing two observations that lacked sufficient data for analysis. I also conducted in-depth interviews with a state forestry official, the head district official in Mestia, two village officials, and two forest

rangers in Upper Svaneti about the planned development, forest use and forest management practices. I also obtained information about non-market and cultural values of forests through in-depth interviews with five villagers and through observation while living in the region over a period of two months. In-depth interviews and observation inform the interpretation of the results and concluding discussion. 3.3. Income Calculation Income from CPRs is ideally calculated as the natural rent realized through consumption or sale of goods harvested. However, rural households in developing countries earn little profit, and the opportunity cost of labor for harvesting is low so that shadow prices are difficult to calculate. Alternatively, it is accepted in the literature to calculate income from common pool resources as value-added, which is the gross benefit net intermediate inputs, not including labor (Sjaastad et al., 2005). Common pool resource income in this study is calculated as total income from forest harvesting activities and grazing in pastures.2 Total threatened income is calculated as CPR income plus the value of fodder harvested from meadowlands under traditional private ownership. Threatened income is derived from the resources — forests, pastures and meadows — that villagers are at risk of losing due to the changes in the Forest Code and the development of the ski tourism zone. Although income from CPRs and agricultural and livestock activities are reported separately for the purpose of this study, it is worth noting their interdependence in several respects: for example, honey production included in agricultural income is dependent on bees being able to access healthy forests. Local market prices were used for goods included in the calculation. Wage employment income and home enterprise income were reported directly.3 Use of forests for non-timber forest product (NTFP) collection is reported, but these values are not included in the income calculation 2

1

It was not possible to randomly select households using voter precinct rolls because addresses were not always accurate and some homes in the region are abandoned. I used a purposive sampling procedure to capture a variety of responses with limited resources. Because the villages were purposively selected, it is unknown whether they are generalizable to the entire Upper Svaneti district.

See the Appendix A for a full account of the income calculation. Missing values for wage employment were imputed by regression estimates based on relevant household variables such as wage employment type, number of hours and days worked, age of employed member and education of adults in the household. Two observations were dropped from the final data set because values for missing variables could not reasonably be imputed. 3

R.J. Kemkes / Ecological Economics 117 (2015) 22–31

because there are no local markets for NTFPs. Similarly, information about use of the forests for hunting was gathered, however due to the sensitivity of potential illegal activities, detailed information about quantities of harvest were not included in the survey and values are not included in the income calculation. The use of the forest for collecting water is reported but values are not included in the income calculation because no market values exist. For these reasons, CPR income levels presented in the results are underestimates of actual CPR use. While recognizing the importance of private assets and longrun, permanent income in household livelihoods (see Narain et al., 2008), the following analysis is based on current income. A measure of current income is relevant in regions where property rights are threatened, households have few private assets and accumulation opportunities are limited. 4. Results: Summary Statistics 4.1. Household Characteristics Summary statistics and units for the variables used in the analysis are given in Table 1. The vast majority of households in the sample, 96%, claim ownership of crop and meadowland. The average area of land claimed for growing crops and harvesting fodder is 2171 square meters (0.22 ha). Approximately 22% of households in the sample are headed by a female. None of the female heads of household is married. As a result of patrilineal inheritance tradition, females head half of the landless households. The mean level of education for the household head is about 12 years, which is much higher than education levels found in studies conducted in India and Africa (Kamanga et al., 2009; Narain et al., 2008). However, this result is in line with average education levels for similar populations in other post-socialist countries (Rocco et al., 2011). The mean annual total income, including both monetary and subsistence income, for the sample is US$3767 per adult equivalent unit (Aeu). Non-monetary subsistence income makes up 71% of the average household's total income. The average income for landless households is significantly lower than for households that claim land, indicating the importance of land tenure in household livelihoods. Landless households do obtain income from threatened resources, including fuel wood and grazing if they own livestock. 4.2. Diversified Strategies Households engage in diversified livelihood strategies. Eighty-eight percent of households earn income from more than one source. Rural households typically engage in diversified strategies to mitigate risk and to buffer against income shocks (Ellis, 2000). Livestock income, on average, is the greatest source of income for households in the sample, followed by wage employment. However, almost half of the households do not engage in any type of wage employment. Most wage employment opportunities are in the government sector in the administrative capital of Mestia, and 39% of households reported at least one member employed in a government position. These jobs are not available to all households, as they require political connections and access to Mestia. Agricultural production and harvesting from threatened resources are also important income-generating activities. Less than one percent of households hold annual savings. This lack of financial assets underscores the importance of diverse income sources, including subsistence activities, for buffering households against unpredictable conditions. 4.3. Income and Asset Distribution Table 2 shows the configuration of household income, assets and characteristics by income quartile. The Gini coefficient for the sample is 0.34. Inequality in total income in Upper Svaneti is primarily due to variations in livestock, wage and agricultural income. The mean income

27

per Aeu for the top quartile is more than five times the mean income of the lowest quartile. The lowest income quartile lives on an average of US$3.59 per person per day, including both monetary and subsistence income. Income from agriculture, livestock, wage employment and threatened resources all increase monotonically across income quartiles. The mean income from CPRs also increases from the lowest to the highest quartile; this result is consistent with previous studies showing that absolute levels of CPR income increase with total income. Increasing income from CPRs across income groups is primarily driven by income obtained from grazing activities, which is, of course, highly correlated with the number of livestock owned by the household. The number of livestock a household can support is at least partially dependent on the amount of meadowland a household can claim under traditional private ownership. Therefore, households with larger traditional land holdings earn higher levels of income from common pool grazing than do land-poor households. All but one of the households in the sample use fuel wood as their primary source for heat, and three-quarters of households in the sample use the forest for grazing livestock. Eighty-five percent of the respondents indicated that forest access is “very important” for their household's livelihood. The second income quartile had the highest share of households reporting forest access to be vital to their livelihoods. Not only do households depend on CPRs in general for current consumption and as a safety net, each resource is critical for survival. A majority of the households in the sample, 85%, consider purchasing fuel wood or finding a substitute to be “somewhat difficult” or “very difficult” if access to the Forest Fund was restricted. Seventy-three percent of respondents found it “very difficult” to obtain a license for timber; only one household in the sample reported harvesting timber. This barrier to access forces households to live under unsafe conditions and also requires that they allocate already limited cash income toward small home or fence repairs. Such institutional constraints also limit the opportunity for villagers to participate in Table 1 Household characteristics. Units

Mean

SD

Income variables Total income (Inc) Income from threatened resources Income from fuel wood Income from fodder Income from timber Income from grazing Income from non-threatened sources Income from wage employment Income from home enterprisec Income from remittances Income from government transfers Income from livestock Income from agriculture

US$/Aeuab US$/Aeu US$/Aeu US$/Aeu US$/Aeu US$/Aeu US$/Aeu US$/Aeu US$/Aeu US$/Aeu US$/Aeu US$/Aeu US$/Aeu

3767 572 92.02 160.21 2.41 317.36 1917 739.76 47.15 29.28 263.68 1563.24 533.12

2631 397 74.80 158.55 37.94 258.36 1424 985.85 158.76 81.90 291.74 1425.28 1241.36

Asset variables Land claimed by household (Land) Savings (1 = household has savings) Loans (1 = household has loans) Education of household head (Ed) Age of household head (Age) Female household head (1 = female) (Fem) Vehicles owned (Veh) Household size (Size) Moved to area since 1990 (1 = yes) (Ref) Married (1 = married)

m2/Aeu Dummy Dummy Years Years Dummy Number Aeu Dummy Dummy

2171 0.004 0.17 11.75 59.13 0.22 0.25 2.56 0.09 0.65

2587 0.06 0.38 2.55 15.05 0.42 0.55 0.98 0.28 0.48

N = 248 a

Aeu (adult equivalent units) calculated using the new OECD scale, see Deaton (1997) for details. b Income reported in US dollars using the August 2011 exchange rate 1 GEL = .6 US$. c Does not include agricultural processing enterprises in order to avoid doublecounting.

28

R.J. Kemkes / Ecological Economics 117 (2015) 22–31

Table 2 Mean values for income sources, asset holdings and household characteristics by income quartile.

Table 3 Dependence on threatened resources by income quartile as percentage of total household income.

Income quartiles Lowest 25% Income Agriculture Livestock Threatened resources Fodder from meadows CPRs Fuel wood Timber Grazing in pastures Household enterprisea Wage employment Government transfers Remittances Total income Assets Cropland (m2/Aeu) Meadowland (m2/Aeu) Total landb (m2/Aeu) Vehicles (#) Own house (1 = yes) Livestock (US$/Aeu) Nut and fruit trees (#) Beehives (#) Savings (1 = yes) Loans (1 = yes) Household characteristics Household head education Female head of household Household head age Household size (Aeu) Number of adults (N18) Number of children (b18) Moved to area since 1990 Married Landless

98 448 241 62 179 74 0 105 36 164 260 39 1311

Income quartiles

25–50%

50–75%

Top 25%

203 1234 488 123 365 95 10 261 40 560 229 21 2788

510 1567 644 197 447 107 0 340 29 988 267 22 4043

1322 3003 915 259 657 93 0 564 84 1248 299 35 6929

355 428 868 0.05 0.97 248 13 0 0 0.13

519 927 1596 0.26 0.94 556 14 0.03 0 0.18

806 1392 2337 0.24 0.97 727 20 0.19 0 0.18

1019 2830 3881 0.45 0.97 1142 22 2.3 0.02 0.21

11.7 0.24 58 2.5 3.16 1.21 0.15 0.66 0.13

11.9 0.24 56 2.8 3.53 1.29 0.08 0.60 0.03

11.8 0.19 61 2.5 3.47 0.66 0.05 0.66 0

11.6 0.21 61 2.5 3.34 0.74 0.08 0.69 0

N = 248 a

Does not include income from agricultural processing home enterprises to avoid double-counting. b Total land includes cropland, meadowland, land rented to others and fallow land under traditional private ownership.

Threatened income Fodder CPRa Fuel wooda Timber Grazinga N = 248 a

a

Lowest 25%

25–50%

50–75%

Top 25%

Total

17.5 4.2 13.3 6.2 0 7.1

17.6 4.4 13.2 3.4 0.3 9.4

16.0 4.9 11.1 2.6 0 8.4

13.9 4.1 9.8 1.5 0 8.3

16.2 4.4 11.8 3.5 0.0 8.3

Prob N F for ANOVA test across income quartiles at α = .10.

resources are degraded or access is restricted. CPR income (not including meadowland under traditional private ownership) makes up 12% of total income. This measure is lower than results from studies conducted in other parts of the world because it does not include values for wild foods. It may also be lower because harvesting and collecting opportunities are limited by short productive seasons and harsh winters. Threatened income comprises a larger share of total income for households in the lower two income quartiles than in the top income quartiles, with the top quartile depending on threatened income the least (see Table 3). Although the share is almost equal for the lower two quartiles, the range for the lowest quartile reaches as high as 60% of total income, while the second quartile range extends to a maximum of 40% share of total income. Degradation would have the greatest impact on those already struggling to make ends meet. Declining dependence on CPRs as income rises implies that CPR income serves as a safety net for households that have the ability to earn substantial income from other sources, while playing a more important role in current consumption for those with fewer alternative income sources. Although dependence declines, the safety net function is still critical for mitigating shocks, smoothing seasonal employment and maintaining diverse livelihood strategies for higher income households. Conversely, if access to CPRs was limited, landless households and households with smaller traditional land holdings would be at a disadvantage because they would not be able to fall back on private stocks of fodder or meadows for grazing. 5.2. Forest use Activities

high-return timber activities. Income from fuel wood increases from the lowest quartile through the third income quartile but falls in the top income quartile. Households in the top income quartile may have more cash available to purchase fuel wood in local markets. Infrastructure for reliable electricity remains limited, even for wealthier households. It is notable that when asked to select their preferred approach to development in Upper Svaneti, 60% of respondents reported a preference for small-scale tourism, such as locally owned guesthouses and shops; 23% preferred that things stay the way they always have been with only improvements to infrastructure, such as roads and bridges; only 8% preferred that outside investors bring jobs to the region; the remaining households did not report a preference. 5. Results: Income Dependence and Participation in Forest use Activities 5.1. Dependence on Threatened Income Income from threatened resources makes up, on average, 16% of total income for households in the sample (see Table 3). This measure represents the degree to which livelihoods will be destabilized if

Other forest use activities are also important for supporting livelihoods in the region (see Table 4). NTFPs such as berries, mushrooms, herbs and nuts are gathered by 21% of the households. These products not only supplement agricultural food production, they also serve an important cultural function as gathering them reinforces a connection with family and village forests, and they are key ingredients in regionspecific food dishes. And unlike timber harvesting, the collection of NTFPs is not restricted by the state. Interestingly, fewer households in the lowest income quartile gather NTFPs than in the highest income Table 4 Percentage of households reporting participation in forest use activities across income quartiles. Forest use

Fuel wood Timber NTFPs Grazing Hunting Water Leisure N = 248

Income quartiles Lowest 25%

25–50%

50–75%

Top 25%

Total

90% 0% 16% 53% 50% 26% 26%

97% 5% 24% 85% 70% 44% 44%

87% 0% 19% 84% 55% 29% 39%

81% 5% 24% 79% 53% 42% 42%

89% .02% 21% 75% 56% 35% 36%

R.J. Kemkes / Ecological Economics 117 (2015) 22–31

29

Table 5 OLS regression model of proportion of income from threatened resources. Independent variables

Dependent variable Threatened income dependence (Dep)

Constant Total annual income/Aeu (Inc) Total land/Aeu (Land) Number of vehicles (Veh) Household head female (Fem) Household head age (Age) Household size (Aeu) (Size) Refugee (Ref) Village residence (Vill) Khaishi Chuberi Nakra Lakhamula Becho Latali Mestia Ipari Ushguli N = 248

19.910*** (0.00) −6.418e−4*** (0.01)

20.387*** (0.00) −8.908e−4*** (0.00) 2.355e−4** (0.04)

−6.494e−2 (0.98) −6.722e−2** (0.04) 0.108 (0.84) −3.154*** (0.00)

6.667e−2 (0.97) −5.900e−2* (0.09) −3.855e−1 (0.56) −2.780*** (0.00)

20.237*** (0.00) −7.712e−4*** (0.00) 2.541e−4** (0.02) −2.276*** (0.00) −1.107e−1 (0.96) −6.364e−2* (0.06) −1.037e−1 (0.87) −2.954*** (0.00)

1.499e−1 (0.82) −1.782e−1 (0.76) 8.659*** (0.00) 1.364*** (0.01) 7.400*** (0.00) 6.557*** (0.00) Baseline 10.814*** (0.00) 3.666*** (0.00) R2 = 0.215

−3.664e−1 (0.64) −2.559e−1 (0.69) 7.538*** (0.00) 1.263** (0.03) 7.494*** (0.00) 6.965*** (0.00)

−5.526e−1 (0.48) −6.881e−1 (0.31) 7.096*** (0.00) 1.241** (0.03) 7.343*** (0.00) 6.452*** (0.00)

9.623*** (0.00) 3.062*** (0.00) R2 = 0.239

9.369*** (0.00) 2.173*** (0.00) R2 = 0.256

Variable significance: *α = .10; **α = .05; ***α = .01; p-values in parentheses. Cluster robust standard errors are used to account for the two-stage sampling procedure.

quartile. This result is likely a reflection of traditional family boundaries within forests, which are still recognized, although not strongly enforced. It is possible that the seasonality of NTFPs in the region due to harsh winters limits the amount of time and labor available for gathering when harvesting crops and fodder are a priority in the summer months. Furthermore, female household members typically gather NTFPs, and because the majority of respondents are males, NTFP gathering may be under-reported in the sample. In most villages water is channeled down from springs located underground at higher elevations. Field observation revealed that this is the primary source of water for drinking and household use in most villages. It is possible that some respondents did not perceive using this water resource as a type of forest use per se, despite its importance, so the result should be considered a lower estimate of the provision of clean water by forests. Over one-third of households reported using the forest for leisure. Furthermore, each village has over a dozen small churches throughout the forest that are used for annual ceremonies and rituals. In interviews, villagers described Lamproba, a procession of male relatives carrying birch bark torches to honor the dead. Villagers, when entering the forest at high altitudes, place a hazelnut branch in the crook of a tree to honor Saint Giorgi and the sacred forest. 6. Results: Determinants of Dependence and Forest use 6.1. Determinants of Dependence on Income from Threatened Resources In order to assess which households are most vulnerable to falling deeper into poverty as a result of losing access to forests, I estimate an OLS regression model with village fixed effects of household characteristics on the share of threatened resource income in total income (Dep) as Dep j ¼ β1 þ β2 Inc þ β3 Land þ β4 Veh þ β5 Fem þ β6 Age þ β7 Size ð1Þ þβ8 Ref þ β9 j Vill þ ε j ¼ village where Inc is total annual household income per Aeu, Land is the total area of land claimed by the household per Aeu, Veh is the number of vehicles owned by the household, Fem is a female-headed household, Age is the age of the household head, Size is the number of Aeu in the household, and Ref is whether or not the household moved to the district since 1990. Vill is the village in which the household resides. Because the variables Inc, Land and Veh are likely correlated, I first estimate a model

including Inc and all other household variables but without Land or Veh, then a model including Inc and Land, and finally a model including all three variables. The results of the three models (see Table 5) consistently show that dependence on threatened resources falls as total income increases, which is congruent with the previous ANOVA test across income quartiles reported in Table 3. Interestingly, villages located farther away from the market centers of Mestia and the regional capital of Zugdidi, and at a greater distance from the main road, have a higher dependence on threatened resource income. Villages with higher measures of dependence include Ipari, Becho, Nakra and Latali. Households in these villages are particularly vulnerable to the negative impact of forest concessions as these areas are also the most desirable for commercial timber harvesting because the forests have maintained their quality throughout the Soviet period and early transition phase. 6.2. Probability of Participating in Forest use Activities Additionally, I estimate a linear probability model4 with village fixed effects of household characteristics on the probability that a household participates in a forest use activity as the following: For j ¼ β1 þ β2 Inc þ β3 Land þ β4 Veh þ β5 Fem þ β6 Age þ β7 Size ð2Þ þβ8 Ref þ β9 j Vill þ ε j ¼ village where the independent variables are the same as in Eq. (1) and the dependent variable For is whether or not a household participates in gathering NTFPs or hunting. Recall that these activities are not captured in the threatened resource income measure that was used as the dependent variable in Table 5. These results (reported in Table 6), therefore, shed light on a second dimension of CPR dependence. Again, because the variables Inc, Land and Veh are likely correlated, I estimate three models in order to identify the effects of each variable. The results show that, even when excluding Land and Veh from the model, Inc does not have a significant effect on whether or not a household participates in gathering forest products. This implies that households across the income spectrum are similarly dependent on these resources. 4 Less than 9% of the values predicted by the linear probability model fall outside the 0– 1 interval of the dependent variable.

30

R.J. Kemkes / Ecological Economics 117 (2015) 22–31

Table 6 Linear probability model of participation in forest use activities.a Independent variables

Dependent variable Engage in forest use activities (For)

Constant Total annual income/Aeu (Inc) Total land/Aeu (Land) Number of vehicles (Veh) Household head female (Fem) Household head age (Age) Household size (Aeu) (Size) Refugee (Ref) Village residence (Vill) Khaishi Chuberi Nakra Lakhamula Becho Latali Mestia Ipari Ushguli N = 248

6.725e−1*** (0.00) −1.46e−5* (0.09)

6.808e−1*** (0.00) −1.89e−5** (0.03) 4.13e−6 (0.15)

9.367e−2 (0.15) −1.305e−3 (0.25) 9.048e−2*** (0.00) −2.764e−1*** (0.00)

9.598e−2 (0.14) −1.161e−3 (0.30) 8.181e−2*** (0.00) −2.698e−1*** (0.00)

6.809e−1*** (0.00) −1.89e−5** (0.04) 4.13e−6 (0.14) 2.267e−4 (1.00) 9.600e−2 (0.15) −1.160e−3 (0.28) 8.178e−2*** (0.00) −2.698e−1*** (0.00)

−4.734e−1*** (0.00) −5.181e−1*** (0.00) 1.071e−1*** (0.00) −7.495e−1*** (0.00) 9.329e−2*** (0.00) 5.336e−2*** (0.00) Baseline 7.468e−2*** (0.00) −1 −2.376e *** (0.00)

−4.824e−1*** (0.00) −5.195e−1*** (0.00) 8.747e−2*** (0.00) −7.513e−1*** (0.00) 9.493e−2*** (0.00) 6.051e−2*** (0.00)

−4.824e−1*** (0.00) −5.195e−1*** (0.00) 8.752e−2*** (0.01) −7.513e−1*** (0.00) 9.494e−2*** (0.00) 6.056e−2*** (0.00)

5.379e−2** (0.01) −2.482e−1*** (0.00)

5.382e−2** (0.01) −2.481e−1*** (0.00)

Variable significance: *α = .10; **α = .05; ***α = .01; p-values in parentheses. Cluster robust standard errors are used to account for the two-stage sampling procedure. a Forest use activities defined as gathering NTFPs or hunting.

Not only do households located farther away from market centers and the main road have a higher dependence on threatened resources, they also have a higher probability of participating in a forest use activity. For example, a household in Becho is almost 9 percentage points more likely to participate in a forest use activity than a household in Mestia. Households with female heads are more likely than maleheaded households to report engaging in forest use activities, as are larger households, which have a greater supply of labor for gathering. 7. Discussion Because household dependence on CPR income can be interpreted as the degree to which other sources of income would have to increase to maintain livelihoods if access is limited or quality is degraded, new employment opportunities in Upper Svaneti would have to offer, on average, 16% of current income levels. In addition, consider, as Bebbington (1999) has discussed in his conception of the livelihoods framework, that some assets are relatively more important than others for different people in different parts of the world. Specifically, in remote mountainous regions where travel out of the area is often difficult and time consuming, particularly during winter months, purchasing substitute goods outside the region or importing goods into the region is costly. Therefore, households would need to be compensated further to maintain or improve current economic conditions. Although dependence on CPR income declines for households in the top income quartile, they still depend on fuel wood for their livelihoods — all but one household reported that fuel wood is their primary source for heating and cooking. Rather than gathering fuel wood themselves, households in the top income quartile can purchase fuel wood harvested by other households in the district at local market prices. The results also show that a majority of households reported that it would be difficult to purchase or find substitutes for CPR goods. Therefore, until substitutes are readily available at comparable prices, current CPR income may not be entirely offset by a concordant increase in wage income. In addition, some of the components, or flows of goods and services, from natural capital are non-substitutable in that they contribute to livelihoods in a way that cannot be replaced by any other type of capital (Ekins et al., 2003). Although households in the top income quartile depend on CPR income for a smaller percentage of their total income, they still rely on forests, pastures and meadows as a safety net to buffer against shocks.

Retaining the ability to produce food for consumption and to gather forest products provides food security in a time of rising global prices and increasingly volatile markets. Diverse livelihoods are important for maintaining resiliency to chance events in a region with a severe climate and unpredictable economic and political institutions. Thus, under current conditions, the safety net function is a nonsubstitutable component provided by natural capital in the region. Furthermore, if forests were degraded, as they have been under the concession granted in 2008, they would no longer provide water purification, and the flow of spring water used by households across the district would diminish, negatively impacting livelihoods. Until infrastructure is developed to provide clean water to households across the district, spring water remains a non-substitutable good. Similarly, habitat functions provided by forests and meadows supply NTFPs and wild game that serve as a safety net for households and hold cultural significance. Finally, the recreational and spiritual uses of forests in Upper Svaneti cannot be substituted for by wage income. Among the seven sustainability principles outlined by Ekins et al. (2003) is the preservation of landscapes of special human or ecological significance, because of their rarity, aesthetic quality or cultural or spiritual associations. Recall that Upper Svaneti had been designated as a planned Protected Area and one of the villages, Ushguli, is a UNESCO World Heritage site — the cultural landscape of Upper Svaneti is an example of a resource that serves a critical national and global function. Finally, because households in the lowest income quartile depend on CPR income for up to 60% of their total income, and households in villages farthest from market centers are most dependent with limited access to roads and market centers, the distribution of changes in access to natural capital and new wage income opportunities must also be considered to prevent a rise in inequality. An increase in inequality may affect livelihood outcomes. In some non-OECD countries, wealth inequality has been found to be negatively correlated with happiness and well-being (Graham and Felton, 2005). 8. Conclusion In this study, I analyze original survey data collected through a sustainable livelihoods framework and information obtained through in-depth interviews to determine the current level of income derived

R.J. Kemkes / Ecological Economics 117 (2015) 22–31

from CPRs and participation in forest use activities that provide nonmarket goods, as reported by households. The results provide a basis for assessing the conditions under which a rural development strategy will lead to improved development outcomes for households as defined through the livelihoods framework, such as an increase in income, well-being, reduced vulnerability and improved food security. The three most important considerations for a rural development strategy to sustain or improve livelihoods in Upper Svaneti are the following: 1) if access to CPRs is restricted or if natural capital is degraded, wage income must increase in proportion to lost CPR income and households must be provided with affordable substitutes, 2) access to the nonsubstitutable components of CPRs, including the safety net function, must be secured, and 3) the distribution of access to natural capital and new wage opportunities must be considered to prevent a rise in inequality that would reduce well-being. The inhabitants of Upper Svaneti have shaped their landscape through traditional agricultural practices, gathering and harvesting forest products and participating in rituals. A rural development strategy that sustains household livelihoods is critical to the survival of this unique cultural landscape of global significance. Acknowledgments American Councils Title VIII Combined Research and Language Training Program funded the fieldwork for this study. The organization had no involvement in the study design; in the collection, analysis and interpretation of data; in the writing of the report; or in the decision to submit the article for publication. I would like to acknowledge the fortitude and professionalism of the team of enumerators from the Caucasus Research Resource Center who conducted the household survey and Nino Gudashvili and Nino Mtvarelishvili who provided research assistance and translation. Thank you to Dr. James Boyce, Dr. Michael Ash, Dr. Sylvia Brandt, Dr. Juan Camilo Cardenas, Dr. William Keeton and Dr. Ceren Soylu for their helpful comments and suggestions on earlier drafts of this paper. Appendix A. Supplementary data Supplementary data to this article can be found online at http://dx. doi.org/10.1016/j.ecolecon.2015.05.002. References Agrawal, A., Chhatre, A., Hardin, R., 2008. Changing governance of the world's forests. Science 320 (5882), 1460–1462. http://dx.doi.org/10.1126/science.1155369. Agyeman, J., Ogneva-Himmelberger, Y., 2009. Environmental Justice and Sustainability in the Former Soviet Union. MIT Press. Bebbington, A., 1999. Capitals and capabilities: a framework for analyzing peasant viability, rural livelihoods and poverty. World Dev. 27 (12), 2021–2044. http://dx. doi.org/10.1016/S0305-750X(99)00104-7. Borras, S.M., Hall, R., Scoones, I., White, B., Wolford, W., 2011. Towards a better understanding of global land grabbing: an editorial introduction. J. Peasant Stud. 38 (2), 209–216. http://dx.doi.org/10.1080/03066150.2011.559005. Boyce, J.K., 1994. Inequality as a cause of environmental degradation. Ecol. Econ. 11 (3), 169–178. http://dx.doi.org/10.1016/0921-8009(94)90198-8. CARE International in the Caucasus, 2010. Pathways out of Poverty: Annual Report 2010 (Retrieved from Retrieved from www.care-caucasus.org.ge). Carney, D., 1998. Sustainable rural livelihoods: what contribution can we make? Papers Presented at the Department for International Development's Natural Resources Advisers' Conference, July 1998. Department for International Development (DFID) (p. vii + 213 pp.)

31

CIFOR: Center for International Forestry Research, 2011. New Global Study Shows High Reliance on Forests Among Rural Poor. Poverty and Environment Network, London (June 15). Deaton, A., 1997. The analysis of household surveys: a microeconometric approach to development policy. World Bank Publications. De Groot, R.S., Wilson, M.A., Boumans, R.M.J., 2002. A typology for the classification, description and valuation of ecosystem functions, goods and services. Ecol. Econ. 41 (3), 393–408. http://dx.doi.org/10.1016/S0921-8009(02)00089-7. De Waal, T., 2011. Georgia's Choices: Charting a Future in Uncertain Times. Carnegie Endowment for International Peace. Dragadze, T., 1976. Family Life in Georgia. New Society. Ekins, P., Simon, S., Deutsch, L., Folke, C., De Groot, R., 2003. A framework for the practical application of the concepts of critical natural capital and strong sustainability. Ecol. Econ. 44 (2–3), 165–185. http://dx.doi.org/10.1016/S0921-8009(02)00272-0. Ellis, F., 2000. The determinants of rural livelihood diversification in developing countries. J. Agric. Econ. 51 (2), 289–302. http://dx.doi.org/10.1111/j.1477-9552.2000.tb01229.x. Engel, E., von der Behrens, H., Frieden, D., Möhring, K., Schaaff, C., Tepper, P., …, Gigauri, G., 2006. A Case Study on Zemo Svaneti, Georgia. Food and Agriculture Organization (FAO) of the United Nations Forestry Department, 2010. Global Forest Resources Assessment 2010 Country Report Georgia (Rome). Forest Law Enforcement and Governance (FLEG), 2010. Review of the Current and Proposed Institutional Changes in Georgia With Reference to the Impact on Forest Law Enforcement and Governance. Graham, C., Felton, A., 2005. Inequality and happiness: insights from Latin America. J. Econ. Inequal. 4 (1), 107–122. http://dx.doi.org/10.1007/s10888-005-9009-1. Jones, S.F., 2012. Georgia: A Political History Since Independence. I. B. Tauris. Kamanga, P., Vedeld, P., Sjaastad, E., 2009. Forest incomes and rural livelihoods in Chiradzulu District, Malawi. Ecol. Econ. 68 (3), 613–624. http://dx.doi.org/10.1016/j. ecolecon.2008.08.018. Macharashvili, I., 2009. Forestry Sector Reform in Georgia (Policy Brief.). Caucasus Institute for Peace, Democracy and Development. Marsden, T., 2009. Mobilities, vulnerabilities and sustainabilities: exploring pathways from denial to sustainable rural development. Sociol. Rural. 49 (2), 113–131. http:// dx.doi.org/10.1111/j.1467-9523.2009.00479.x. Narain, U., Gupta, S., van 't Veld, K., 2008. Poverty and resource dependence in rural India. Ecol. Econ. 66 (1), 161–176. http://dx.doi.org/10.1016/j.ecolecon.2007.08.021. Rayfield, D., 2012. Edge of Empires: A History of Georgia. Reaktion books. Rocco, L., Brunello, G., Crivellaro, E., 2011. Lost in Transition? The Returns to Education Acquired Under Communism 15 Years After the Fall of the Berlin Wall (Working Paper No. 17). AlmaLaurea Inter-University Consortium (Retrieved from http:// ideas.repec.org/p/laa/wpaper/17.html). Scoones, I., 1998. Sustainable Rural Livelihoods: A Framework for Analysis vol. 72. Institute of Development Studies Brighton. Scoones, I., 2009. Livelihoods perspectives and rural development. J. Peasant Stud. 36 (1), 171–196. http://dx.doi.org/10.1080/03066150902820503. Sjaastad, E., Angelsen, A., Vedeld, P., Bojö, J., 2005. What is environmental income? Ecol. Econ. 55 (1), 37–46. http://dx.doi.org/10.1016/j.ecolecon.2005.05.006. Sunderlin, W.D., Angelsen, A., Belcher, B., Burgers, P., Nasi, R., Santoso, L., Wunder, S., 2005. Livelihoods, forests, and conservation in developing countries: an overview. World Dev. 33 (9), 1383–1402. http://dx.doi.org/10.1016/j.worlddev.2004.10.004. Topchishvili, R., 2005. Svaneti and its Inhabitants, Ethno-historical Studies in History of Georgia Mountain Regions. Van der Werf, G.R., Morton, D.C., DeFries, R.S., Olivier, J.G.J., Kasibhatla, P.S., Jackson, R.B., …, Randerson, J.T., 2009. CO2 emissions from forest loss. Nat. Geosci. 2 (11), 737–738. http://dx.doi.org/10.1038/ngeo671. Vedeld, P., Angelsen, A., Bojö, J., Sjaastad, E., Kobugabe Berg, G., 2007. Forest environmental incomes and the rural poor. For. Policy Econ. 9 (7), 869–879. http://dx.doi.org/10. 1016/j.forpol.2006.05.008. Welton, G., Zurabishvili, T., Nozadze, N., 2008. Georgia Human Development Report 2008: The Reforms and Beyond. United Nations Development Program (UNDP), Tbilisi, Georgia. White, B., Borras Jr., S.M., Hall, R., Scoones, I., Wolford, W., 2012. The new enclosures: critical perspectives on corporate land deals. J. Peasant Stud. 39 (3-4), 619–647. http://dx.doi.org/10.1080/03066150.2012.691879. World Bank, 2009. Georgia - Poverty assessment. World Bank, Washington, DC. http:// documents.worldbank.org/curated/en/2009/04/10503390/georgia-povertyassessment. World Bank, 2011a. World Bank, Doing Business Project. Retrieved from. http://www. doingbusiness.org. World Bank, 2011b. World Development Indicators. Retrieved from. http://data. worldbank.org/indicator. Wunder, S., 2001. Poverty alleviation and tropical forests—what scope for synergies? World Dev. 29 (11), 1817–1833. http://dx.doi.org/10.1016/S0305-750X(01)00070-5.