Unpacking the impacts of ‘participatory’ forestry policies: Evidence from Kenya

Forest Policy and Economics 69 (2016) 45–52

Contents lists available at ScienceDirect

Forest Policy and Economics journal homepage: www.elsevier.com/locate/forpol

Unpacking the impacts of ‘participatory’ forestry policies: Evidence from Kenya Jane Mutheu Mutune a,⁎, Jens Friss Lund b a b

Wangari Maathai Institute for Peace and Environmental Studies, 30197-00100, University of Nairobi, Kenya Department of Food and Resource Economics, University of Copenhagen, Denmark

a r t i c l e

i n f o

Article history: Received 11 March 2015 Received in revised form 4 March 2016 Accepted 9 March 2016 Available online xxxx Keywords: Community forest associations Forests Households Impact Livelihoods Participatory forest management

a b s t r a c t We evaluate the livelihoods of member and non-members of Community Forestry Associations under Kenya's participatory forest management (PFM) programme. We use propensity score matching of households based on recall based data from before implementation of PFM from 286 households and comparison of current incomes (2012), as well as review of records and interviews. Results reveal that members have higher total and forest-related incomes than non-members and indicate that impacts derive from labour and market opportunities supported by donor institutions, more than from differential access to forest products. In terms of governance the Kenya Forest Service largely remains in control of decision-making. Thus, PFM resembles Integrated Conservation and Development Project (ICDP) approaches. We conclude that current forest governance approaches in Kenya appear not to support participation in practice. Further, we conclude that impact evaluations must examine both outcomes and participatory forestry to provide meaningful policy evidence. © 2016 Elsevier B.V. All rights reserved.

1. Introduction Ten to twelve percent of the world’s natural forests are officially managed with some degree of popular participation, which is also the case in at least 21 sub-Saharan African countries promoting participatory approaches to natural resources management through participatory forest management (Sunderlin et al., 2008). In some of these cases, the changes in rights to manage forests seem to enable improved forest management (Thoms, 2008; Ribot et al., 2010; Takahashi and Todo, 2012; Lund et al., 2015), whereas the evidence base is more limited and provides a more mixed picture with regards to livelihood impacts (Sikor and Nguyen, 2007; Larson et al., 2007; Maharjan et al., 2009; Ameha et al., 2014). Furthermore, the existing evidence base on impacts of participatory forestry is geographically biased towards more studies from South Asia, notably Nepal and India (Lund et al., 2009; Bowler et al., 2012). This is problematic, given that the large differences in society and ecology, as well as the models of participatory forestry, between Asia and Africa inhibit the drawing of lessons from one context to the other. In Africa, there is an emerging literature on livelihood impacts from Tanzania (e.g. Shackleton et al., 2002; Kajembe et al., 2002; Lund and Treue, 2008; Vyamana, 2009; Pfliegner, 2011; Green and Lund, 2015; Scheba ⁎ Corresponding author. E-mail addresses: [email protected] (J.M. Mutune), [email protected] (J.F. Lund).

http://dx.doi.org/10.1016/j.forpol.2016.03.004 1389-9341/© 2016 Elsevier B.V. All rights reserved.

and Mustalahti, 2015), but the model of participatory forestry there differs from, among others, Kenya, as it is based on the village jurisdiction, as opposed to membership of an association. Studies on the livelihood impacts of participatory forestry face the same challenges as those confronting attempts at impact evaluation more generally. First, the policy that one seeks to evaluate must be empirically characterized. A number of studies have pointed to the fact that participatory forestry policies have failed to materialize on the ground as substantive changes in rights to resources for people in many of the areas where it has supposedly been implemented (Ribot et al., 2010). Impact evaluations therefore should empirically verify the extent to which forest governance processes on the ground resemble participatory forestry to avoid attributing outcomes to participation where participation does not exist. Second, the outcome must be empirically characterized. Studies of livelihood impacts of participatory forestry have used a great variety of indicators ranging from the share of community population that accesses a forest (Persha et al., 2011) to detailed evidence on incomes and assets (Ameha et al., 2014). Finally, impact evaluations must attend to the issue of attribution of the observed outcome to the policy as opposed to other processes. This is has been pointed to as the Achilles’ heel for impact evaluations in the context of participatory forestry in reviews (Lund et al., 2009). To attribute the outcome to the policy, impact evaluations must describe what would have happened to the units targeted by the policy in its absence - a situation termed the counterfactual. As this situation cannot be observed, impact evaluations seek to estimate what it would have been through various

46

J.M. Mutune, J.F. Lund / Forest Policy and Economics 69 (2016) 45–52

research designs ranging from the selection and observation of controls (Jumbe and Angelsen, 2006; Ameha et al., 2014) to the construction of a counterfactual through tracing processes of change in units targeted by the policy from before the policy was implemented (Lund et al., 2015). This study illustrates an impact evaluation of the livelihoods outcomes of PFM in Kenya. It does so by comparing members and nonmembers of community forestry associations (CFA) among communities living adjacent to the Eburu and Sururu Forest Reserves. Specifically, we examine the policy of PFM as it unfolds in practice on the ground and seek to evaluate its impacts through matching of CFA and non\\CFA member (NCFA) households based on re-call data to generate estimates of impacts on household income. In the following, we first describe the study area and methods used; then outline the key results followed by discussions and conclusions.

2. Study area PFM was introduced in Kenya following pressure mostly from civil society organizations as an approach to ensure sustainable forest management (Mugo et al., 2010). The first PFM site was at ArabukoSokoke Forest established in 1997 and today more than 100 CFAs exist around the major water towers of Kenya (Thenya et al., 2007). CFA membership is drawn from residents of forest adjacent communities and is, in practice, mostly formed from pre-existing community based organizations. Elected and vetted executive committees run the day to day activities of the CFA and set subsidiary bylaws that guide its activities and penalties in case of infringement by members or outsiders. The study was undertaken among forest adjacent communities to Sururu and Eburu forest of Eastern Mau Forest Reserve. The Eburu and Sururu forests are among nine forest blocks of Eastern Mau Forest Reserve and are found in Nakuru District, Rift Valley Province. The two blocks were selected purposively as they were the oldest of the nine forest blocks, i.e. we would expect livelihood impacts to have materialized. Sururu forest lies at 2400–2900 m above sea level. The forest covers an area of 20,461 hectares (ha) (UNEP and KFWG, 2006). The forest is divided into five smaller administrative units (beats) namely Gatimu, Station, Kanorero, Leporos and Mau Narok. The main tree species in the forest are Dombeya gotezeni, Pondo carpus, Olea Africana, Prunus africana and Juniperas procera. The forest is the upper catchment for Nderit River that supports the livelihoods of the forest adjacent communities and people in the Rift-Valley and Western Province. Community sensitization was initiated by KFS after enactment of the Forests Act, 2005. As a result, three CFAs, focusing on different parts of Sururu Forest, were formed in 2008. These include Mugameli, Masufa and Sururu-Nderit which brought together the various community based organizations and individual community members from villages adjacent to Sururu forest. Through the guidance of KFS, the three CFAs formed the umbrella Sururu CFA1 in 2008. At the time of the study, Sururu CFA had a forest management plan but no management agreement. Villages in Mau Narok, Sururu, Mpatipat and Leporos locations form membership to Sururu CFA. All households rely on the forest for fuel wood, building materials, fodder, beekeeping, seedling collection, medicinal herbs, water and organic fertilizer. Households accrue non-use values that include circumcision, spiritual, religious and aesthetic sites. Besides, households derive regulating services e.g. erosion, local climate, water, frost and diseases regulations important economic resilience and factors of production for agriculture. The Eburu Forest is located at the border of Narok District and Naivasha sub-district at an altitude range of 2400 to 2900 m above sea level. The forest covers an area of 8715 ha (KFS, 2008) and borders Ol Jorai and Ndabibi Agricultural Development Corporation farms in the North and South, respectively, and Loldia farm to the East (KFS, 2008). 1

The CFA is called Mau Sururu Likia CFA (MASULICOFA).

There are six beats within the forest namely Kiruya, Ngobobo, Songoloi, Morop, Jerusalem and Ole Sirwa. Based on impressions from forests walks during the study, the forest appears heavily utilized, and comprises mainly Acacia abyssinica trees and shrub vegetation, Dovyalis spp. The forest forms part of the catchment for Lakes Naivasha and Elementaita with several ground springs. The forest is source of Ndabibi River and other small streams that support the livelihoods of the surrounding communities. Starting in the early 2000s the communities worked together with Kenya Forest Working Group to set up a committee involving a number of stakeholders to oversee the management of the forest. This marked the beginning of a participatory approach to manage Eburu forest and eventually of Eburu CFA2 in 2006. The CFA has elected officials and a KFS approved management plan and agreement. Villages in Kiambogo, Ndunyu Buru, Ndabibi, Malewa and Eburu locations form membership to Eburu CFA. The households rely on the forest for wood fuel, beekeeping, medicinal herbs, building material, fodder, hunting and water from the natural steams for domestic use and geothermal power generation by Kenya Electricity Generating Company (KenGen) The forest is home to the endangered wild bongo and endemic for birds e.g. tangaza sun bird, Hartilup turraco.

3. Theoretical and empirical approach To unpack how the policy of PFM unfolds in practice, we draw on the ‘Actors, Powers, Accountability’ framework (Agrawal and Ribot, 1999). The underlying assumption of this framework is that decentralization policies must be assessed on the degree to which they result in transfers of meaningful powers to downwardly accountable bodies at the lower level in a jurisdictional hierarchy. Policies are therefore assessed by looking at who (actors) wield what powers over the resources targeted by the policy and the rules regulating how such actors can be held to account for their wielding of powers. In the context of PFM, we are particularly interested in the degree to which powers have been devolved to bodies that are accountable to forest adjacent communities, as these are typically the ostensible target group of such policies (Ribot et al., 2010). In this framework, powers refer to authority to make rules and decisions regarding forest management, as well as to implement, enforce and adjudicate said rules. In practice, this implies an attention to the degree to which the CFA can decide about the uses of the forest, including decisions that affect the benefits that forest use give rise to, such as marketing of commercially valuable forest products. This way of assessing what powers have been conferred on forest adjacent communities is in line with the general bodies of theory upon which participatory forestry policies are based, i.e. participation theory (e.g. Arnstein, 1969) and common pool resource theory (Ostrom, 1990). Accountability implies that the body receiving such powers can be held to account by people living in the forest adjacent communities. This assumes a form of governance where management responsibility is vested in an executive body at the community level that is kept to account through procedures of information sharing and elections, whereas participation and common pool resource theory often assumes a more direct participatory form of governance. Yet, as we will see, our case fits well with this framework. To examine these aspects of the policy empirically, we reviewed the Forests Act 2005 and CFA management plans and agreements and did in-depth interviews with key informants. These included: 4 KFS officials, 5 CFA officials, 4 CFA scouts, and 15 villagers, including 6 households engaged in firewood extraction, 5 households undertaking grazing and 4 households involved in charcoal production. The informants were purposely selected for their ability to inform the study objectives. The interviews were guided by interview guides specific for each main stakeholder group prepared in advance of the interviews.

2

The CFA ECOFA.

J.M. Mutune, J.F. Lund / Forest Policy and Economics 69 (2016) 45–52

Where the interviewee gave consent, the interviews were recorded, else detailed notes were taken. To evaluate livelihood impacts we took a quasi-experimental approach based on recall information about assets and income in 2003 (before) and 2012 (after) for a sample of NCFA (control) and CFA (treatment) households. Thus, the approach follows a ‘before-after-control impact’ (BACI) logic, yet with all data based on recall from interviews done in 2013/14. At the time of study, none of the CFAs had updated membership lists. Sample frames of all households differentiated by CFA membership was therefore created with the help of local informants. Then, NCFA and CFA households were randomly sampled in each of the villages. The study considered adult (above 20 years of age) household heads eligible to join CFA as the total population. Households that had decided to drop their CFA membership were considered NCFA households. A sample frame of household heads was developed with local informants. Based on criteria of proximity to forest, CFA membership and ethnicity, three and seven locations were selected for sampling in Eburu and Sururu forests, respectively. Each location was taken as a strata and sample size calculated using rigorous statistical methods (Booth et al., 2008). Each household head was allocated a unique number and research randomizer used to pick numbers randomly. The household head of selected households were interviewed. Samples of 59/48 CFA/ NCFA households from the Sururu area and 108/96 CFA/NCFA households from the Eburu area were drawn. However, due to resource constraints only 48/94 CFA/NCFA households from Sururu and 44/88 CFA/ NCFA households from Eburu were actually surveyed. Pre-tested household level questionnaires were administered through personal interviews by enumerators. Five enumerators were identified and trained for two days and, to minimize measurement errors, team interviews were conducted for three days, with the aim to generate common understanding of the questions and approaches to interviewing (Angelsen et al., 2011). The questionnaire was used to elicit household level data on cash and subsistence incomes in 2012 from various types of livelihood activities i.e. agriculture, livestock production, wage work, business, environment utilization, pensions and remittances. Incomes were reported in net income, i.e. the gross value minus costs of production but excluding households' own labour input (Cavendish, 2002). Following Claro et al. (2010) incomes are reported in adult equivalent units (AEU) to enable comparison between households of different sizes. During household interviews, BACI by scoring methods were used for participatory impact assessment to assess changes in incomes from diverse livelihood activities. Income sources were represented visually and respondents were asked to distribute counters (seeds) among these to illustrate relative proportion of household income derived from each livelihood activity in 2003 (before) and 2012 (after). Enumerators were to observe any changes in the scores from “before” and “after” then ask the respondents to explain the reasons for these differences. This kind of tool is particularly useful in measuring impact of an intervention where baseline data is weak or non-existing (Catley et al., 2007). A basic necessity survey (BNS) was used to obtain a locally relevant wellbeing ranking from the perspective of residents, as it was hypothesized that a wellbeing measure relative to the specific community would be likely to shape PFM related outcomes. The BNS forms the basis for an index of wealth for every household in the sample, relative to a locally-derived definition of poverty. Building on Davies and Smith (1998), we define basic necessities as assets or services that 60% or more of respondents agree are basic necessities that everyone in the community should be able to have and nobody should have to go without. Prior to the survey, a list of 25 assets and services was defined based on focus group discussions in villages not selected for the survey. The list was deliberately constructed to include some items no one would consider basic necessities. During the survey, respondents were asked which items they considered a basic necessity and which they owned currently.

47

The percentage of respondents who considered each item to be a basic necessity was then used as the weighting for that item in the calculation of the index (Davies and Smith, 1998). 3.1. Data analysis Average impact of PFM on rural livelihoods was calculated as the average effect of the treatment on the treated (ATT). ATT is defined as the difference between the expected value of the outcome by project participants while participating in the project and the expected value of outcome they would have received if they had not participated in the project. Following Ravallion (2006) and Smith and Todd (2005):


ATT ¼ E Y1 −Y0 │ P ¼ 1 ¼ E Y1 │ P ¼ 1 −E Y0 │ P ¼ 1

ð1Þ

where: ATT = average impact of treatment on treated; P = membership in CFA (P = 1 if members in CFA and P = 0 otherwise); Y1 = outcome of participant, and Y0 = outcome of the same household if it had not participated in CFA. However, we do not know what outcomes would have looked like if the household had not participated in CFA and we cannot observe outcomes of participation and non-participation for the same household hence missing data on the counterfactual. Rosenbaum and Rubin (1983) developed the propensity score matching approach that is most commonly used in quasi experimental methods to overcome the unobserved outcome E (Y0 │ P = 1). The remedy was to identify nonCFA member households (NCFA) and use them as counterfactual. Propensity score matching and difference-in-difference were employed for estimation of ATT. There are no clear guidelines for determining which variables to use propensity score matching (Caliendo and Kopeinig, 2008). In principle, variables used for matching should simultaneously affect the choice to participate in PFM and outcomes of PFM (Caliendo and Kopeinig, 2008). We selected the following: education, age, group membership, household size, and distance to forest, ownership of domestic animals, land size, and wealth class of household. Reasons behind the choice are that the capitals available to a household form the basis for livelihood strategies pursued by a particular household (Ellis, 2000). Households with similar sets of assets are more likely to follow similar livelihood portfolios ceteris paribus. Following the matching, we checked whether the distribution of variables was balanced or not. We did a t-test difference for the individual covariates besides a comparison of Pseudo R2, Likelihood ration (LR2) and P-value (Prob N χ2) for before and after matching in the probit model, thus check of matching quality. Patterns of household annual incomes and socioeconomic factors were also analyzed using t-tests. Hence, the difference in the livelihood outcomes between matched CFA and NCFA member households was inferred as the impact of the PFM implementation. The analysis was carried using STATA software 12, psmatch2 program developed by Leuven and Sianesi (2003). 4. Results 4.1. The policy in practice PFM in Sururu and Eburu has implied that the forests are managed by CFAs and Kenya Forest Service (KFS) officers in collaboration. The CFAs are membership-based organizations formed on the basis of community-based organizations (CBOs). People pay a one-off subscription fee to become members of a CBO, typically around KSH 300 (USD 3.5). The CBOs register their members in the CFA and pay a one-off subscription fee of KSH 5000 (USD 59) for the entire CBO. When in the CFA, households pay a one-off fee of Ksh 300 (USD 3.5) to join forest user groups' e.g. grazing, firewood and a monthly fee of KSH 100 (USD 1.2). At the time of study, the number of CFA members was 300 and 250 for Sururu and Eburu CFA respectively.

48

J.M. Mutune, J.F. Lund / Forest Policy and Economics 69 (2016) 45–52

Table 1 Basic socio economic attributes of households in Sururu and Eburu Forest areas in 2003 and 2012. Attributes of household head

Sururu

Eburu

CFA n = 48

NCFA n = 94

CFA n = 44

NCFA n = 88

Mean

Mean

Mean

Mean

Age in years (%), 2012 15–29 30–44 45–59 Over 60

6 39 37 18

5 39 42 14

6 28 31 35

7 34 32 27

Highest education level attained, (%), 2012 Primary school Secondary school Tertiary

58 38 4

64 30 6

52 38 10

61 34 5

Ethnicity (%), 2012 Kikuyu Kalenjin Maasai Ogiek Othersa Female headed households (%), 2012 Share of females in household (%), 2012 Household size, 2003

58 23 1 10 8 36 58 6.22

63 24 3 6 4 29 53 5.51

70 11 7 0 12 17 56 5.70

86 8 2 0 4 30 52 4.64

Farm and institutional attributes Own land size in acres AEU, 2003 Own land under trees in acres AEU, 2003 Distance to nearest forest edge in kilometres, 2012 Number of adult domestic animals, 2012 Group membership other than CFA in 2003 = 1, if yes Wealth score, 2003

0.45 0.11 1.70 12 0.56 83.23

0.49 0.14 1.74 9.6 0.40 67.53

0.81 0.23 1.59 14 0.59 84.90

0.87 0.25 1.78 13 0.35 79.90

a

Kamba, Luhyia, Kisii, Luo and Meru.

With a total population of adults estimated at 36,000 and 7119 for Sururu and Eburu, respectively, this means that only a minor fraction of the adult population are member of a CFA. The members of the respective user groups elect their own executives, i.e. chairperson, treasurer and secretary, who are in charge of the day to day management of activities. In turn, the user groups' executives elect the executives of the CFA who hold office for a period of three years. Yet, in practice KFS retains most decision-making powers, whereas CFAs have been allocated user rights to certain products. KFS retains the power to issue permits and licenses and decide on prices, amounts and specific procedures to be applied in extraction of forest products. Further, KFS retains all forest revenues. Finally, KFS retains the full forest resource ownership rights and the KFS Director can terminate the management agreement with the CFA or revoke a particular user right. The Forests Act 2005 empowers the CFAs to assist KFS in enforcing the provisions of the Act particularly in relation to illegal harvesting of forest products. In practice, the CFA board, constituted by chairpersons of the various community based organizations and CFA officials, has appointed CFA scouts to patrol the forest. The scouts directly report to the CFA chairperson, but also to KFS forest guards in the respective parts of the forest. The mandate of the scouts is limited to forest patrolling and reporting of forest crimes, whereas KFS has the powers to arrest, seize, detain and confiscate assets used in illegal harvesting of forest products. The scouts receive a monthly payment of 5000 Ksh financed by a grant from the African Wildlife Foundation and have received training by KFS financed by the Green Zone Development Project and African Wildlife Foundation. Key informants explained that any forest-related activity undertaken by the CFA, including donor-driven income-generating activities, are planned and executed through the KFS. Thus, in practice, the PFM policy has not resulted in a transfer of any meaningful powers in relation to forest management. CFA members are beneficiaries of specific, delimited user rights, over which the CFAs have little influence, and of donor-funded, forest-related, income-generating activities that are

planned and executed through the KFS. CFAs main ‘management’ role is that of paid labour provision for forest management works, as well as paid monitoring and enforcement practices, but also here their role

Table 2 Total net household income per AEU by CFA membership, 2012. Income sources

Cash income Crop Forest cash Firewood Charcoal Seedlings Honey Building materials Livestock Business (e.g. grocery) Remittances and Pension Casual labour Formal employment Total net cash income Forest income as % of total cash income Non-cash income Own farm produce Forest goods (e.g. firewood, herbs) Other sources- gifts and external support Total net household income Total forest income as % of total household income

Sururu

Eburu

CFA (n = 48)

NCFA (n = 94)

CFA (n = 44)

NCFA (n = 88)

10,539 9435 3748 0 1564 3400 723 10,344 3914 667 3852 3549 42,300 22

11,903 5777 3306 0 949 840 682 8309 2448 1713 6206 4973 41,329 13

14,775 6508 2800 1219 726 436 1327 12,059 2261 3576 2781 1345 43,305 15

14,101 4089 1072 1060 257 140 1560 12,884 1042 4106 6694 1085 44,001 9

11,586 6060

12,510 4804

13,252 4027

13,406 2800

379

118

438

171

60,325 25

58,761 18

61,022 17

60,378 11

Note: The values are in hundreds of Kenya Shilling Ksh (Ksh 85 ≈ USD 1.0 in 2012). The row ‘Forest cash income’ is the sum of the rows ‘firewood, ‘Charcoal’, ‘seeds and wildling collection, ‘building materials' and honey’.

J.M. Mutune, J.F. Lund / Forest Policy and Economics 69 (2016) 45–52

is also only one of supplying labour whereas the powers to make rules, enforce and adjudicate remain with the KFS. 4.2. The livelihood impacts

Table 4 Matching estimates on average impact of CFA and NCFA in Sururu and Eburu forest, per AEU. Total income, 2012

Summary statistics for key values are presented in Table 1. Both areas were dominated by young male headed households but Eburu had a higher average age of household heads. In both sites, CFA member households were larger and with slightly smaller farm areas per AEU than NCFA households. NCFA households had slightly larger on-farm areas allocated to trees than CFA households, particularly in Eburu. CFA households, on the other hand, had a much higher proportion of livestock owners. The basic necessity survey results show that in both sites, CFA members had higher relative wealth scores in 2003 than NCFA members. Table 2 illustrates that CFA and NCFA members in both sites engaged in diverse livelihood strategies including crop farming, livestock keeping, forest use, petty business, and casual labour. Households in Eburu had slightly higher crop incomes probably owing to their larger land holdings. The major crops in both sites included maize, Irish potatoes and vegetables. Livestock incomes from sheep, cattle and their respective products were the second most important source of household cash income. Households in Eburu had higher livestock cash income. In Sururu, NCFA households had lower livestock holdings in both 2003 and 2012 and correspondingly lower livestock income in 2012. Table 2 shows that forest income was important for both CFA and NCFA households, yet more so for CFA households owing mainly to higher cash incomes seedlings, honey and charcoal (Eburu only) and higher subsistence forest income (mainly firewood). The survey data reported no incomes from charcoal among households in Sururu. Charcoal burning is still considered illegal in the two sites. The charcoal incomes in Eburu could be higher than what appears from the data. KFS station managers reported cases of poaching particularly for charcoal and bamboo tree that formed the main building materials. Though the Forests Act encourages formation of charcoal producer organizations through which charcoal burning is legalised and regulated, in both of the sites these groups were non-existent. Key informants associated the low charcoal incomes by household members from fear of punitive rules placed by Forests Act 2005, which households were aware about. 4.3. Matching quality Table 3 shows the result from a probit model before and after matching. The low values of Pseudo R2 and LR chi2 values after matching indicate success of matching in both sites. After matching the variables in the probit model, for both sites, were not mutually significant with Prob N χ2 = 0.90 for Sururu and ProbN χ2 = 0.85 for Eburu. The implication is that there was no systematic difference in the distribution of the explanatory variables between the matched CFA and NCFA households. Appendix 1 shows that there are significant differences between the CFA and NCFA households in both sites for a number of key variables, including livestock ownership, poverty status, group membership, forest income, and household size. However, after matching, there are no significant differences between the two groups. Table 3 Households' matching quality indicators before and after matching, 2003.

49

Sururu n = 142 Household income Forest Tree seedling Beekeeping Firewood Fodder Building materials Eburu n = 132 Household income Forest Tree seedling Beekeeping Firewood Fodder Building materials Charcoal

Unmatched CFA

Difference

NCFA

Matched CFA

60,325 58,761

1564 (30)

Difference NCFA

60,325 54,260

6065* (72)

15,495 10,581 4914* (85) 15,495 11,043 4452* (109) 1564 949 615* (29) 1564 921 643* (31) 3400 840 2560** (121) 3400 1040 2360** (39) 8185 6528 1657 (105) 8185 6910 1275 (91) 1623 1582 41 (13) 1623 1494 129 (25) 723 682 41 (13) 723 678 45 (9)

61,022 60,378 10,535 726 436 4027 2800 1327 1219

644 (20)

61,022 55,602

6889 3646** (38) 10,535 257 469** (48) 726 140 296* (32) 436 2964 1063 (141) 4027 908 1892** (92) 2800 1560 −233 (23) 1327 1060

159 (12)

1219

5420 (103)

6480 4055** (45) 187 539** (83) 80 356* (25) 2463 1564 (58) 1690 1110 (72) 1080 247 (54) 980

239 (47)

Standard errors in parenthesis, obtained from 1000 bootstrap replications for matching methods. Total forest income row is a sum of incomes from tree seedlings, beekeeping, firewood, fodder and building materials. ⁎ Indicates significance at 5% level. ⁎⁎ Indicates significance at 1% level.

For matching to be undertaken any combination of characteristics observed in the treatment group should also be observed among the control group hence common support condition for CFA and NCFA households (Caliendo and Kopeinig, 2008). There was found to be an overlap of propensity scores for CFA and NCFA participants (Appendices 2 and 3). This implied that all CFA member households found suitable NCFA matches and thus the sample size was not reduced. 4.4. Impact of PFM on livelihoods Table 4 presents incomes of CFA and NCFA members before and after matching. Results show that the average impact of PFM on CFA as compared to NCFA households was positive. In both of the sites, when matched, CFA households had significantly higher forest, tree seedling, beekeeping, and firewood incomes than NCFA households. Also, in both sites, total household income was higher for CFA households, yet only significantly so in Sururu. We also analyse how impacts are distributed between wealth classes. We do this by dividing the sample into a poor and non-poor group, separated by a wealth score of 60%. Table 5 shows the absolute and relative (to total income) forest incomes for these two groups. Results are that the poor rely less on forests than the non-poor and that this difference grows from 2003 to 2012. To understand how this pattern is affected by CFA membership we further divide the two wealth groups into CFA and NCFA in Table 6. Results are that the forest reliance of CFA participants, poor as well as non-poor, has increased over the period 2003–2012, whereas that of NCFA households has decreased. Thus, CFA households are more forest

Table 5 Distribution of mean forest income by wealth class, per AEU.

Quality indicator

Sururu

Eburu

Before matching

After matching

Before matching

After matching

Attributes

Non-poor (n = 207)

Poor (n = 67)

p-Value

t-Value

Pseudo R2 LR2 Prob N χ2

0.26 107.70 0.000

0.01 5.23 0.90

0.26 90.25 0.000

0.01 3.70 0.85

Forest income, 2012 Relative forest income, 2012 (%) Relative forest income, 2003 (%)

10,911 24 22

5367 14 16

0.04 0.09 0.008

−2.02 −1.07 −2.687

50

J.M. Mutune, J.F. Lund / Forest Policy and Economics 69 (2016) 45–52

Table 6 Distribution of mean forest income by CFA membership and wealth class, per AEU. Attributes

Forest income in Ksh, 2012 Relative forest income, 2012 (%) Relative forest income, 2003 (%)

CFA

NCFA

Non-poor (n = 71)

Poor (n = 21)

p-Value

t-Value

Non-poor (n = 136)

Poor (n = 46)

p-Value

t-Value

12,759 29 24

6280 17 16

0.17 0.15 0.09

−1.36 −1.41 −1.69

6811 20 21

3627 12 16

0.17 0.01 0.390

−1.35 −1.63 −0.80

reliant than NCFA households in 2012. The decrease in forest reliance is particularly large for poor households. 5. Discussion The PFM policy in Kenya officially promotes participation by forest adjacent communities in forest management. This is articulated in the Forests Act, 2005 Part IV Sect. 45 that encourages communities to participate in management of forests adjacent to them and embraces PFM as a tool to actualize such community involvement. Specifically the Act requires CFAs to protect, conserve and manage forests together with KFS. Yet, the sharing of forest management powers between CFAs and KFS in Eburu and Sururu forests lies far from the ideals expressed in the policy. Our examination of the practice of forest management in Eburu and Sururu showed that KFS wields powers to make decisions on forest management, plans all management activities, wholly controls access to forest products, and enjoys all the proceeds resulting from forest products. Further, the forest scouts responsible for the enforcement are appointed by the CFA, but reports to KFS and are financed by the African Wildlife Foundation. The PFM regime observed in Eburu and Sururu is akin to deconcentration in the vocabulary of Agrawal and Ribot (1999) with the majority of powers vested in local offices of the central government agency, in this case Kenya Forest Service. Such a regime where representative (elected) local authorities exist, but have not been given any formal powers, defies the central assumptions of theories of participation and decentralization. As Ribot et al. (2010:36) note: “Local power without representation is not democratic decentralization. Nor is representative local authority without powers. Accordingly, we should not expect improved equity, efficiency or local enfranchisement from these two common institutional configurations (…)”. In terms of the outcomes of this policy, our results show that PFM has benefited CFA members through higher total household incomes driven mainly by higher incomes from nature based income generating activities e.g. seedlings, beekeeping and their respective trainings. This indicates that PFM is more about differential opportunities for engagement in income generating activities supported by NGOs and donor institutions, some of which are forest related, than about differential access to forest products. While there were no significant differences in income from firewood, fodder and timber among CFA and NCFA households, the data did indicate that poor NCFA households have seen their relative forest incomes reduced following the increased intensity of forest patrolling in the area. Whereas poor CFA households benefit from income generating activities supported by donors, poor NCFA households appear to have been hit particularly hard by increased forest policing and penalties. This finding resonates with other studies showing that poor community members are hit the hardest by increased control with and taxation of forest uses following implementation of participatory forestry (Lund and Treue, 2008; Vyamana, 2009; Kumar, 2002). Interviews confirmed these observations by revealing that CFA members had access to income generating activities such as dairy goat farming, fish farming, bee keeping, seedlings collection and production that were not available to households in the area before the implementation of PFM. In Sururu, the Green Zone Development Project and Kenya Commercial Bank Foundation had supplied members with 155 beehives and in Eburu Imarisha Naivasha had given Eburu CFA members

40 beehives. Further, beekeeping user group members had received training on apiary management supported by Community Development Trust Fund, Self Help Africa and Green Zone Development Project. Members of the Sururu CFA had also received training on honey value addition supported by Baraka Agricultural College. These activities, supported by a proliferation of different actors, had led to the introduction of new technologies, such as top bar, langstroth hives and smokers for harvesting of honey. Thus, the significant increase in annual beekeeping income is attributable to PFM because such benefits were derived by households upon active participation in the CFA. In both sites, tree seedlings incomes were higher among CFA than NCFA households, and highest among Sururu CFA members. The production of tree seedlings particularly for indigenous trees was reported to have expanded with the implementation of PFM and mainly due to donor interest. African Wildlife Foundation, Community Development Trust Fund, Yokohama University and Nakuru County Tree Nursery Association were supporting Sururu CFA members with rehabilitation of the forest by buying seedlings produced by the residents. In Eburu the Green Belt Movement was the only organisation involved in forest rehabilitation and its activities existed in the area before the advent of PFM. All of these institutions would buy seedlings from CFA members for Ksh 30, whereas NCFA households would be paid half of that and would only get a chance to sell seedlings when CFA members had exhausted their stock. CFA members in Sururu had higher tree seedling incomes than those of Eburu probably owing to the higher number of donors involved in funding forest rehabilitation. Transect walks in both forests confirmed the higher level of forest rehabilitation activity in the Sururu forests as compared to Eburu. Key informants reported that Sururu forest had been rehabilitated with more than 2 million seedlings all purchased by the aforementioned NGOs and donor organizations through KFS from tree nursery forest user groups. The CFA members provided most of the labour during rehabilitation and were paid a daily rate of Ksh 350 per person, a wage higher than the normal daily wage of Ksh 250 in the same locality. The title of this paper reads ‘Unpacking the impact of PFM’. We chose this title to reflect that impact analyses of PFM (or any other policy) must unpack not only outcomes, but also the policy itself. Our analysis who wields power over forest management in Eburu and Sururu illustrates that PFM in this context resembles deconcentration coupled with donor and NGO sponsored efforts in forest law enforcement and activities that improve the conditions for certain livelihood activities. This defies the notion of participation as involvement of forest adjacent communities in actual management of the forest. All the forest-related livelihood activities and forest management practices described above are heavily reliant on donor financial support and are planned and executed through the KFS, rather than by the CFAs themselves. Our case thus has more resemblance with an Integrated Conservation and Development approach whose principal objectives are conservation and social economic improvement but without resources ownership rights delegated to the community. Our assessment of the policy of PFM resonates with other recent studies from Kenya (Ongugo et al., 2008; Matiku et al., 2013; Mogoi et al., 2012; Chomba et al., 2015), which indicate that despite establishment of CFAs, community members involvement in forest management is limited to providing labour e.g. forest patrolling and rehabilitation but overall resource control rights including decision making remain vested in the KFS. Beyond Kenya, there is ample evidence that forest management approaches that are labeled

J.M. Mutune, J.F. Lund / Forest Policy and Economics 69 (2016) 45–52

‘participatory’, ‘community-based’, or collaborative are often highly centralized or captured by other powerful actors, leaving little discretionary powers to people living in and around the forests (Ribot, 2003; Ribot et al., 2010; Dressler et al., 2015; Schusser et al., 2015).

6. Conclusion With this study, we illustrate how impact evaluations must examine both policy outcomes and the policy itself to provide meaningful policy evidence. We do this through a study of participatory forestry as it unfolds in Eburu and Sururu forests in Kenya. We examine the policy of PFM as it unfolds in practice on the ground and seek to evaluate its impacts on the livelihoods of member and non-member households of Community Forestry Associations (CFA) formed under the participatory forest management programme in Kenya. Our results show that CFA member households had higher total household, forest, beekeeping and tree seedling incomes than nonCFA households. Overall livelihood impacts were driven more by differential forest-related labour and market opportunities supported by NGOs and donor institutions, than by differential access to forest products although there were indications that poor NCFA households experienced reduced relative forest incomes following increased intensity of forest patrolling. Our examination of the policy explains the observed outcomes, as the Kenya Forest Service remains in control of decision-making and access to forest resources. This is contrary to the intention of the participatory forestry policy and CFA members' expectations. Thus, CFA management in these sites resembles Integrated Conservation and Development Project (ICDP) approaches more than genuine participatory forestry. On this basis we argue that there is need to scrutinize current forest governance approaches in Kenya as they appear not to support participation in practice. Further, we conclude that impact evaluations must rigorously examine both outcomes and policy to provide meaningful evidence for policy makers. Supplementary data to this article can be found online at http://dx. doi.org/10.1016/j.forpol.2016.03.004.

Acknowledgements The research was undertaken under the project ‘Stabilizing Kenya by Solving Forest Related Conflicts’ (STAKE) financed by the Consultative Research Committee for Development Research under the Danish Ministry of Foreign Affairs (Danida), grant 11-036KU. We wish to thank the communities in Sururu and Eburu for their collaboration. We further acknowledge the assistance of the enumerators Joyline, Teresia, Virginia, Wycliffe and Samorai and the village guides Johana and Kirui, for their persistence and enthusiasm during the field research. We are grateful to the KFS station managers in Sururu and Eburu, Forester Kimeu and Kiita for sharing their interesting PFM experiences. We dedicate this paper to a colleague and scholar of the STAKE project, Bernard Muthee, who sadly passed away during this study.

References Agrawal, A., Ribot, J., 1999. Accountability in decentralization: a framework with South Asian and African cases. J. Dev. Areas 33, 473–502. Ameha, A., Nielsen, O.J., Larsen, H.O., 2014. Impacts of access and benefit sharing on livelihoods and forest: case of participatory forest management in Ethiopia. Ecol. Econ. 97, 162–171. Angelsen, A., Larsen, H.O., Lund, J.F., Smith-Hall, C., Wunder, S. (Eds.), 2011. Measuring Livelihoods and Environmental Dependence: Methods for Research and Fieldwork. Earthscan. Arnstein, S.R., 1969. Ladder of citizen participation. J. Am. Inst. Plan. 35, 216–224. Booth, W. C., G. G. Colomb and J. M. Williams (2008). The Craft of Research (Third edition). The University of Chicago Press, Chicago 60637. The University of Chicago Press, Ltd., London.

51

Bowler, D.E., Bu Buyung-Ali, L.M., Healey, J.R., Jones, J.P.G., Knight, T.M., Pullin, A.S., 2012. Does community forest management provide global environmental benefits and improve local welfare? Front. Ecol. Environ. 101, 29–36. Caliendo, M., Kopeinig, S., 2008. Some practical guidance for the implementation of propensity score matching. J. Econ. Surv. 22, 31–72. Catley, A., Burns, J., Abebe, D., Suji, O., 2007. Participatory Impact Assessment: A Guide for Practitioners. Feinstein International Centre. Cavendish, W., 2002. Quantitative methods for estimating the economic value of resource use to rural households. In: Campbell, B.M., Luckert, M.K. (Eds.), Uncovering the Hidden Harvest: Valuation Methods for Woodland and Forest Resources. Earthscan, London, UK (262 pp.). Chomba, S., Treue, T., Sinclair, F., 2015. The political economy of forest entitlements: can community based forest management reduce vulnerability at the forest margin? Forest Policy Econ. 58, 37–46. Claro, R.M., Levy, R.M., Bandoni, D.H., Mondini, L., 2010. Per Capita Versus Adultequivalent Estimates of Calorie Availability in Household Budget Surveys. Cad. Saúde Pública, Rio de Janeiro 26(11) pp. 2188–2195. Davies, R., Smith, W., 1998. The Basic Necessities Survey: The Experience of Action Aid Vietnam. Action Aid, Hanoi, Vietnam. Dressler, W., McDermott, M.H., Schusser, C., 2015. The politics of community forestry in a global age — a critical analysis. Forest Policy Econ. 58, 1–4. Ellis, F., 2000. A framework for livelihoods analysis. In: Ellis, F. (Ed.), Rural Livelihoods and Diversity in Developing Countries. Oxford University Press, Oxford. Green, K., Lund, J.F., 2015. The politics of expertise in participatory forestry: a case from Tanzania. Forest Policy Econ. 60, 27–34. Jumbe, C.B.L., Angelsen, A., 2006. Do the poor benefit from devolution policies? Evidence from Malawi's forest co-management program. Land Econ. 82, 562–582. Kajembe, G.C., Mbwambo, L., Katani, J.Z., Dugilo, N.M., 2002. Impacts of Decentralisation of Forest Management: Evidences from Tanzania. Arusha, Tanzania. Kenya Forest Service-KFS, 2008. Eburu Participatory Forest Management Plan 2008-2012. KFS Headquarters, Nairobi. Kumar, S., 2002. Does “participation” in common pool resource management help the poor? A social cost benefit analysis of joint forest management in Jharkhand, India. World Dev. 30 (5), 763–782. Larson, A.M., Pacheco, P., Toni, F., Vallejo, M., 2007. The effects of forestry decentralization on access to livelihood assets. J. Environ. Dev. 16 (3), 251–268. Leuven, E., Sianesi, B., 2003. PSMATCH2: stata module to perform full Mahalanobis and propensity scorematching, common support graphing, and covariate imbalance testing. Stat. Softw. Components S432001. Lund, J.F., Treue, T., 2008. Are we getting there? Evidence of decentralized forest management from the Tanzanian miombo woodlands. World Dev. 36, 2780–2800. Lund, J.F., Balooni, K., Casse, T., 2009. Change we can believe in? Reviewing studies on the conservation impact of popular participation in forest management. Conserv. Soc. 7, 71–82. Lund, J.F., Burgess, N.D., Chamshama, S., Dons, K., Isango, J., Kajembe, G., Meilby, H., Moyo, F., Ngaga, Y., Ngowi, S., Njana, M., Skeie, K., Theilade, I., Treue, T., 2015. Mixed methods approaches to evaluate conservation impact: evidence from decentralized forest management in Tanzania. Environ. Conserv. 42, 162–170. Maharjan, M.R., Dhakal, T.R., Thapa, S.K., Schreckenberg, K., Luttrell, C., 2009. Improving the benefits to the poor from community forestry in the Churia region of Nepal. Int. For. Rev. 11, 254–267. Matiku, Caleb, M., Callistus, O., 2013. The Impact of participatory forest management on local community livelihoods in the Arabuko Sokoke Forest, Kenya. Conserv. Soc. 11 (2), 112–129. Mogoi, J., Obonyo, E., Ongugo, P., Oeba, V., Mwangi, E., 2012. Communities, property rights and forest decentralisation in Kenya: early lessons from participatory forestry management. Conserv. Soc. 10, 82–94. Mugo, E., Nyandiga, C., Gachanja, M., 2010. Development of Forestry in Kenya (19002007): Challenges and Lessons Learnt. Kenya Forestry Working Group, Kenya. Ongugo, P.O., Mogoi, J.N., Obonyo, E., Oeba, V.O., 2008. Examining the roles of community forest associations in the decentralization process of Kenyan forests. Paper Presented to the IASC Conference 11-19th July 2008, England. Ostrom, E., 1990. Governing the Commons – The Evolution of Institutions for Collective Action, Political Economy of Institutions and Decisions. Cambridge University Press, UK. Persha L., Agrawal, A and Chhatre, A. (2011). Social and ecological synergy: local rulemaking, forest livelihoods, and biodiversity conservation. Science 331, 1606 DOI: http://dx.doi.org/10.1126/science.1199343 http://www.sciencemag.org/cgi/collection/ecology Pfliegner, K., 2011. The Impacts of Joint Forest Management on Forest Condition, Livelihoods and Governance: Case studies from Morogoro Region in Tanzania PhD Thesis University of East Anglia, School of International Development, UK. Ravallion, M., 2006. Evaluating Anti-Poverty Programs. Policy Research Working Paper 3625. World Bank, Development Economics Research Group, Washington DC. Ribot, J.C., 2003. Democratic decentralisation of natural resources: institutional choice and discretionary power transfers in sub-Saharan Africa. Public Adm. Dev. 23, 53–65. Ribot, J.C., Lund, J.F., Treue, 2010. Democratic decentralization in sub-Saharan Africa: its contribution to forest management, livelihoods and enfranchisement. Environ. Conserv. 37 (1), 35–44. Rosenbaum, P., Rubin, D., 1983. The central role of the propensity score observational studies for causal effects. Biometrika 70, 41–55. Scheba, A., Mustalahti, I., 2015. Rethinking ‘expert’ knowledge in community forest management in Tanzania. Forest Policy Econ. 60, 7–18. Schusser, C., Krott, M., Movuh, M.C.Y., Logmani, J., Devkota, R.R., Maryudi, A., Salla, M., Bach, N.D., 2015. Powerful stakeholders as drivers of community forestry — results of an international study. Forest Policy Econ. 58, 92–101.

52

J.M. Mutune, J.F. Lund / Forest Policy and Economics 69 (2016) 45–52

Shackleton, S., Camplbell, B., Wollenberg, E., Edmunds, D., 2002. Devolution and community-based natural resource management: creating space for local people to participate and benefit? ODI, Natural resource Perspectives No.76 Sikor, T., Nguyen, 2007. Why may forest devolution not benefit the rural poor? Forest entitlements in Vietnam's central highlands. World Dev. 35, 2010–2025. Smith, J.A., Todd, P.E., 2005. Does matching overcome LaLonde's critique of nonexperimental estimators? J. Econ. 125, 305–353. Sunderlin, W., Hatcher, J., Liddle, M., 2008. From Exclusion to Ownership? Challenges and Opportunities in Advancing Forest Tenure Reform. Rights and Resources Initiative, Washington, DC, USA. Takahashi, R., Todo, Y., 2012. Impact of community-based forest management on forest protection: evidence from an aid-funded project in Ethiopia. Environ. Manag. 50, 396–404.

Thenya, T., Nahama, E., Wandago, B., 2007. Participatory forest management experience in Kenya (1996-2006). Proceedings of the 1st National Participatory Forest Management Conference, 6-8 June 2007. Nairobi, Kenya. Thoms, C.A., 2008. Community control of resources and the challenge of improving local livelihoods: a critical examination of community forestry in Nepal. Geoforum 39, 1452–1465. United Nations Environmental Programme and Kenya Forest Working Group, 2006r. Eastern Mau and South West Mau Forest Reserves. Assessment and Way Forward. Kenya Forest Working Group. Vyamana, V.G., 2009. Participatory forest management in the Eastern Arc Mountains of Tanzania: who benefits? Int. For. Rev. 11, 239–253.