The Basic Concepts Correlating mariculture and household income generation: a case of panggang island initiative

received 200 fing erlings as the project’s loan. After nine months, he successfully cultured the grouper and sold his harvest. Consequently, he have to pay the Rp2.6 million to CCMRS-IPB. If he paid the loan soon after he sold his production, then he will get extra 200 fingerlings for the next period. Now, he had up-graded his membership to “Silver” class and receives 400 fingerlings in total for his second year. Every time one is successful conducting his mariculture activity and repay the loan on time, the project will give “award” through an up-grade of his membership, e.g. from Biru to Silver or Silver to Gold. The advantage of this membership up-grade is to provide additional fingerlings as the incentive for work harder and honest. This project also gives dispensation for its member by giving a possibility to repay the loan in installment method especially for those who have any urgent matters, such as sickness of family member and children wedding. In an extreme case, one may cheat by confessing that his fish were dead or stolen in order to get dispensation to eliminate his responsibility to pay the loan. If he is convicted lying, then he will be punished by not getting any fingerlings on the next period. This irresponsible member usually will discontinue his membership. In this case, internal group monitoring is applied. Besides providing fingerling, CCMRS-IPB also provides loan in a form of mariculture production means, such as nets, drums and feed, on limited number. It distribution is based on sea farming committee’s decision. The mechanism of repayment method is the same with fingerling’s loan. In the fourth activity, connecting sea farming member to the market, CCMRS-IPB invited buyer and trader from Muara Angke, Jakarta. This is not really effective because most of members have their own buyer. In practice, the member will sold their production directly to the buyer or middlemen who came to Panggang Island. As an alternative, they will send their production directly to Muara Angke. The first option is preferable compared to the second option because it is less costly and easier. In the fifth activity, CCMRS-IPB provides field facilitators whom are responsible for giving assistancecounseling of any technical and non-technical issues and also managing the loan. Only if the member faced any serious problems, such as diseases in larger scale, the dedicated aquaculture expert from CCMRS-IPB would come and check to make any necessary treatment. Restocking is another activity which should be attached to sea farming project because it is the mean to reach its second objective, conserving the marine ecosystem through juvenile release. Formally, the agreement between local government, CCMRS-IPB, and sea farming members regulate that every member should contribute 5 percent from their production for restocking purpose. In practice, the regulation cannot be successfully implemented yet because of the lack of fingerling stock. Thus, no official restocking activity conducted by sea farming member up to now. There are two restocking activities in 2010 which used fund from the Ministry of Environment, Republic of Indonesia and CNOOC 10 . In summary, the project description of sea farming is presented on Table 5. 10 CNOOC China National Offshore Oil Corporation is Chinese offshore oil multinational company and its involvement in restocking activity is part of its CSR’s Corporate Social Responsibilty program. Table 5 Project Description of Sea Farming in Panggang Island Items Explanation Program Sea farming is a project to create sustainable shallow marine resource management system which uses maricultureas a base-activity. Financing Local government fund Kepulauan Seribu Administrative Regency, DKI Jakarta Province. Partnership Local government as the main sponsor and CCMRS-IPB as the project manager. Target Groups People who work in SSF such as fishermen and small-scale fish farmers. Other occupations are additional. Objectives Goals a. Improving local community’s welfare; and b. Conserving marine ecosystem. Activities a. Setting up the regulation, institution and infrastructure; b. Providing knowledge and skill-based training; c. Providing fingerling; d. Connecting sea farming member to the market; and e. Providing counseling particularly for handling any disease case. Source: CCMRS-IPB 2006, 2007 Research Operational Framework The increasing human population and the demand of fisheries products have pushed the increase of global fish supply. This market pressure has exploited the natural resources which caused overfishing and declining capture fisheries. To fulfill the gap between the demand and supply, aquaculture is one of the alternative solutions to this problem. There are many aquaculture projects that have been implemented in DCs and LDCs to fulfill the increasing demand of fisheries products, both for domestic and export market. Besides fulfilling the demand, most of these projects are mainly aimed to improve some of crucial development indicators such as poverty alleviation, food security and malnutrition, women empowerment, as well as environmental sustainability. In line with that, sea farming project in Panggang Island, Kepulauan Seribu was implemented with two main goals which are improving local community’s welfare and conserving marine ecosystem. This project was designed as co-management in fisheries management which involves local government, local community, and other stakeholders. In this type of fisheries management, all parties are sharing their responsibilities in managing the CPR. In this study, there are two main objectives which are identifying the determinants of household participation in the project activities and depict the projects contribution to poverty alleviation on household income based on quantitative analysis and descriptive analysis. Probit model was used to analyze which factors that are determine participation in the project. On the first model estimation, sea farming participation SFP served as the dependent variable and there are two groups of independent variables, i.e. household characteristics HC and household assets HA. The independent variables that categorized as household characteristics are age, education, occupation, household size, and membership in non-sea farming organization organization member; whilst independent variables that categorized as household assets ownership are television, mobile phone, and boat ownership. OLS regression was used to analyze the impacts of the project in improving local community welfare and alleviating poverty through household income generation. On the second model estimation, household income served as the dependent variable and there are three groups of independent variables, i.e. household characteristics HC, household assets HA, and sea farming SF. Independent variables that categorized as household characteristics are age, education, and membership in non-sea farming organization organization member; independent variables that categorized as household assets ownership are mobile phone and boat ownership; and sea farming participation as independent variable in sea farming group. In addition to the OLS regression, descriptive analysis was used to understand the benefits and constraints subjectively felt by the local community. Figure 7 illustrates the operational framework of this study. Figure 7 Research Operational Framework Increasing Demand of Fisheries Products Increasing Population GAP Alternative Solution: Aquaculture Co-management SEA FARMING SF Goal 1 of SF: Community Poverty Alleviation and Social Welfare Sustainable Fisheries Overfishing, Resource Depletion Market Pressure Goal 2 of SF: Marine Environment a. OLS Regression b. Descriptive Analysis Local Government Local Community CCMRS-IPB and Other Stakeholders Restocking Activity OPPORTUNITIES CHALLENGES HH Assets Ownership HA Variables: TV, mobile phone, and boat ownership HH Income Generation Declining Capture Fisheries Productivity The Determinants of SF Participation HH Characteristics HC Variables: age, education, occupation, HH size, and organization member Probit Model Mariculture Activity SF Impact SF Impact Improve Knowledge and Skill Mariculture and Business Increase Fish Stock Ecotourism Other Variables: HC: age, education, and organization member HA: mobile phone and boat ownership Remarks: Research Area Interaction Line Data Analysis Tools Research Objective Independent Variables SF Participation 3 RESEARCH METHOD Study Site, Data, and Survey Design The selection of the study site was purposeful rather than random. Panggang Island is selected as it is the only successful sea farming project in Indonesia since other projects failed to be implemented due to the lack of local government support. The location of Panggang Island is shown on Figure 8. It is located around 45 kilometers from Jakarta and it needs about 60 to 120 minutes by middle-speed boat. There are 13 islands in the Panggang Island village and only two islands are populated with 5,123 inhabitants BPS KAKS 2011. The two populated islands are Pramuka Island as the capital of KAKS, and Panggang Island as the most populated island in the region. This study is categorized as ex-post study because the observation was conducted after the event is completed. Both primary and secondary data were used in this study as described on Table 6. The primary data were based on a survey of rural coastal households in Panggang Island since all sea farming members live on this island. The author and enumerators administered the pretested structured questionnaire to sample household, which domiciled on three Rukun Warga RW and 21 Rukun Tetangga RT 11 in Panggang Island. The survey was conducted on August 2012 and a total of 82 households were interviewed. It consists of 39 sea farming members participants as the treatment group while the control group consists of 43 non-participants. However, there are only 77 households that are 11 RT and RW are two lowest zones under village level in Indonesia, but they do not included in the official division of Indonesian government administration. Figure 8 Map of Panggang Island, Kepulauan Seribu Kelurahan Pulau Panggang 2010 composed of 34 and 43 households for treatment and control groups respectively used in the analysis because of missing values for some variables. Two different approaches were used to select both groups. A stratified random sampling 12 technique was employed to select the treatment group as it used the list of sea farming members given by project’s field facilitator and it selected only the active members. Meanwhile, a convenience sampling 13 technique was employed to select the control group as they are chosen because they are available when the enumerators visited them to their house 14 . The data and information drawn from this survey were used to analyze the factors that influence sea farming project participation, project impacts to local community welfare, as well as benefits and constraints to the project. Table 6 Types and Sources of Data No Type of Data Detail Source 1 Primary Data a. Respondent characteristics e.g. age, gender, origin, education, occupation, household size b. Mariculture activity e.g. ownership, experience, commodity, technique, cost of production, access to credit c. Sea farming e.g. awareness, membership information, sea farming benefits and constraints d. Household assets ownership e.g. housing ownership, water supply, sanitation, electronics, and vehicles e. Household income and expenditure Interview with treatment and control groups 2 Secondary Data a. Demographic b. Socioeconomic condition c. Market potential d. Others a. CCMRS-IPB b. Local government regency and province c. Ministry of Marine Affairs and Fisheries d. BPS Central Agency of Statistics e. FAO f. Others Data Analysis The first two research questions were analyzed using statistical techniques that are described with more detail on the next two sub-sections. The raw data from the survey were processed using Microsoft Excel and Stata 11. In contrast with the first two research questions, benefits and constraints of sea farming for local community were analyzed by using descriptive analysis. This part is intended to explore household and local community’s points of view about sea farming project and its impacts. Along with the result of the first two quantitative analyses, this part is expected to provide some valuable insights for concluding and formulating useful policy implications related to sea farming. 12 Stratified random sampling is categorized as random or probability sampling technique. It selects research participants based on their membership in a particular subgroup or stratum VanderStoep and Johnston 2009. 13 Convenience sampling is categorized as non-random or purposive sampling technique. It selects people who are available or convenient for the study VanderStoep and Johnston 2009. 14 Four persons were involved in the field research and each of person interviewed ten persons in three RW for control group, while selected sea farming participants have been determined using the list from CCMRS- IPB’ field facilitator. It takes approximately 20-30 minutes, on average, to interview each respondent.

a. Model Estimation for Factors that Determine Individual Participation in

Sea Farming Factors affecting project participation were examined using probit probability unit regression model. Probit was used in several previous researches to determine factors affecting individual participation in project or adoption, e.g. Rahm and Huffman 1984; Nagubadi et al. 1996; Holloway et al. 2002; Ghadim et al. 2005; Kabunga 2011; and Amlaku et al. 2012. Probit was chosen because: 1. It specifies a functional relation between qualitative binary variable on the left hand side and various explanatory variables on the right hand side Feder et al. 1985. The qualitative binary variable is dichotomized between participating and not participating in the project. 2. It used small data sets only 77 households are included on the analysis. 3. If we compare with logit logistic regression, both models link functions yield very similar outputs when given the same inputs but differ in the tails Cameron and Trivedi 2010. Thus, the choice of probit versus logit depends largely on individual preferences IDRE-UCLA 2013. Below is the explanation about probit model. The equation and explanation about probit model in this study is taken from Dougherty 2001; Gujarati and Porter 2009. First, we define an unobservable utility index or “latent variable” which represents the preference of the i-th individual to participate and is expressed by: 3.1 Second, we relate the index with the actual decision to participate in the project. We denote if household participates in the project and otherwise. Then, we assume that there is a threshold level of the index, , such that if exceeds , the household decides to participate and vice versa. 3.2 The probability that is less than or equal to can be computed from the standardized normal cumulative distributive function CDF as follows: 3.3 In other words, as the standardized normal CDF gives the probability of the event occurring for any value of Z which is expressed as: 3.4 Then, we obtain estimates of the parameters by using maximum likelihood analysis. The marginal effect of is is computed as: 3.5 Since is the cumulative standardized normal distribution, its derivative, , is the standardized normal distribution and is formulized as: 3.6 The empirical model of sea farming participation is derived from utility maximization function as adapted from Rahm and Huffman’s adoption model Rahm and Huffman, 1984: where 3.7 The underlying utility function, , ranks the preference of i-th individual participation depends on and . is a vector of personal attributes and is a vector of management characteristics that associated with a particular program 15 . In this study, personal attributes is represented by household characteristics e.g. age, education, and occupation and household assets ownership e.g. television, mobile phone, and boat ownership. Household characteristics 16 are common factors in analyzing factors to determine project participationadoption Nagubadi et al. 1996; Kapanda et al. 2003; Amlaku et al. 2012. Thus the empirical model of sea farming described as a function below: 3.8 The dependent variable, sea farming participation SFP is dichotomous between 1 if respondent is participating in sea farming and 0 otherwise. The independent variables were organized into two groups, i.e. household characteristics HC and household assets HA. The definition of all independent variables along with its coefficients’ expected signs are presented on Table 7. Table 7 Definition of Independent Variables for Determinants of Sea Farming Participation No Variables Definition Expected Sign

A. Household Characteristics HC

1 Age Age of household head in years. Positive 2 Education Education of household head. 1= If household head attend more than nine years of education 0= Otherwise Positive 3 Occupation Primary occupation of household head. 1 = Fishermen 0 = Others Positive 4 Household size Size of the household. Positive 5 Organization member Social network of household head, whether he is a member of non-sea farming organization or not. 1 = Household head join non-sea farming organization 0 = Otherwise Positive

B. Household Assets Ownership HA

6 Television ownership before 2005 Number of television owned by the household. Positive 7 Mobile phone ownership before 2005 Number of mobile phone owned by the household. Positive 8 Boat ownership before 2005 Number of boat owned by the household. Negative 15 Detail explanation of the model can be seen on Rahm and Huffman 1984. 16 Some studies categorized age, education, occupation, etc as socio-demographic factors e.g. Amlaku et al. 2012. On the first group, the independent variables that classified as household characteristics are age age , education educt , occupation fishermen , household size hhsz , and organization member org_member . Other household characteristics such as gender and religion were not taken into account since there is no variability on those two variables. All respondents are men and muslims. Some of household characteristic variables such as educational status, occupation, and organization member were dummy variables. The author used dummy variables to simplify the categorization on each variable. The detail explanation on each variable is described below. Age of the household head is expected to have positive influence on the decision to participate in sea farming. The older the age of the household head, the more likely he decided to join the project as it gives alternative to do less fishing activity which is more risky and uncertain. Educational status of the household head provides a dummy measure of whether the household head attend less or more than nine years of education. In Indonesia, there is a program called “Wajib Belajar 9 Tahun” which means all citizens should obtain minimum nine years of formal education that combine six years in elementary school and three years in junior high school. This study assumed that if household head attended senior high school or university, and then he would have broader knowledge and skills compared to others who only attend nine years of education or less. Thus, it is hypothesized that educational status has positive influence to the participation in sea farming project as it is classified as 1 if the household head attended more than nine years of education. Occupation of the household head provides a dummy for household head primary occupation. It is categorized as fishermen equal to 1 and non-fishermen equal to 0. It is assumed that one who works as fishermen will be more likely to participate in sea farming project as it gives alternative to diversify source of income. Mariculture activity would be one way to mitigate risks in facing long monsoon when it is impossible for them to go fishing. Household size measures the number of people who live in each household. It is assumed that if household head has bigger household size, then the more likely he decided to participate in sea farming project as he have to obtain alternative source of income to fulfill the basic needs of other household members whom economically dependent. It is hypothesized that household size would positively correlate with the participation decision. Organization member provides dummy to capture whether the household head is actively participating in non-sea farming organization. It assumed that household head who actively participate in non-sea farming organization would have more information including about new project or technology. Thus, it is hypothesized that membership in non-sea farming organization has positive influence to the participation in sea farming project. On the second group, household assets, the authors took only television tv05 , boat bt05 , and mobile phone mp05 ownership as the independent variables in the model. These three continuous variables are measured before participation. The threshold is using the year of 2005 because it is the time when the project started. The other household assets’ variables, such as refrigerator, freezer, bicycle, and motorcycle, are excluded as the data are too small to be included in the model. Television ownership is a continuous variable that shows how many television a household possessed before 2005. Television has a lot of function especially as a source of knowledge and information. It is assumed that if household possessed television, they would have more access to knowledge and information. Thus, television ownership is hypothesized as positive to the decision of participation in sea farming project. Mobile phone ownership represents how many mobile phone a household possessed before 2005. It is hypothesized as positive to the decision to participate in sea farming project. It is because by having mobile phone, one’s ability to communicate with others is higher and its probability to obtain information is higher than the one who does not own any mobile phone. Boat ownership is selected in this model as boat is one of productive assets particularly in capture fisheries activity. The hypothesis for boat ownership is opposite to oc cupation’s hypothesis because boat ownership is unique. The size and capacity of the boat would significantly affect its capture ’s productivity. If a fisherman possesses boat, especially the bigger one, then it is more likely that he would be more confident to go fishing to a wider and further open sea in a longer period of time. He would also feel more secured in doing his fishing activity. Thus, he might not need another alternative source of income as he could get high profit from his capture activity.

b. Model Estimation for Sea Farming Impacts on Household Income

Impacts of sea farming on household income were examined by using OLS regressions. The model specification is given by: 3.9 The dependent variable is the total income which accumulates income from SSF capture andor mariculture and non-SSF income e.g. tourism, trade, self-employed, and labor work. Other incomes from direct transfer, food aid which stated on the questionnaire were removed. As mentioned on chapter 1, the total income is calculated on a period basis not yearly basis because the mariculture’s income is seasonal. In addition to that, one period refers to nine months of the total income. The independent variables are divided into three groups which are household characteristics HC, household assets ownership HA, and sea farming SF. The variables which related to mariculture activity e.g. mariculture ownership and mariculture experience are excluded from the model because there might be an issue of reverse causality to the total income. Table 8 shows the definition of independent variables for sea farming impacts on household income. Table 8 Definition of Independent Variables for Sea Farming Impacts on Household Income No Variables Definition Expected Sign

A. Household Characteristics HC

1 Age Age of household head in years. Negative 2 Education Education of household head. 1= If household head attend more than nine years of education 0= Otherwise Positive 3 Organization member Social network of household head, whether he is a member of non-sea farming organization or not. 1 = Household head join non-sea farming organization 0 = Otherwise Positive

B. Household Assets Ownership HA

4 Mobile phone ownership after 2010 Number of mobile phone owned by the household. Positive 5 Boat ownership after 2010 Number of boat owned by the household. Positive

C. Sea Farming SF

6 Sea farming participation Participation in sea farming project. 1 = Participants 0 = Non-participants Positive The independent variables on the first group are: age age , education educt , and organization member org_member . Age of the household head is expected to have negative relation to the total income. Productivity is decreasing when the household head become older, so the younger household head is assumed to be more productive and have higher income than the older one. Education provides a dummy measure indicating 1= for the household head whom attend more than nine years of education and 0= for otherwise. It is assumed that if household head attended higher education, then he would have a better chance to obtain higher income. Thus, it is hypothesized that educational status has positive relation to the total income. Organization member provides dummy indicating 1= for membership in non-sea farming organization and 0= for otherwise. It is hypothesized that membership in non-sea farming organization has positive relation to the total income. By joining organization, it assumed that the household head will build his network which could be beneficial for his economic and business activities. On household assets characteristics group, the independent variables are mobile phone mp10 and boat bt10 ownership after 2010. The year 2010 is used as the threshold in this model because the authors would analyze the project impacts after five years implementation. Mobile phone ownership is hypothesized as having positive relationship to total income as it could be the source of information to get alternative income. Boat ownership is hypothesized as having positive relation with total income because boat is productive assets in SSF. The only independent variable in the third group is sea farming participation sfp which is the main issue in this study. Sea farming participation provides dummy indicating 1= for participation in the project and 0= for otherwise. It is hypothesized to have positive relationship to household total income. By joining sea farming organization, it is expected that it could give an alternative source of income to local community particularly the participants.