Valuing wetland ecosystem in Ethiopia an empirical evidence with Choke mountain: Application of choice experiment

Ecosystem services have not market price. In Ethiopia, choke mountain wetland ecosystem has major contribution for providing water, regulating climate, and offering recreational. Valuing this natural resource helps to protect it from damage. The objective of the study is to value choke mountain wetland ecosystem by using choice experiment. To value it the attributes are biodiversity, recreational service, water availability, job opportunity, and monetary payment. Primary data was collected from 270 sample households. Both descriptive and econometrics analysis were conducted. The study found that households had good awareness about the choke mountain wetland ecosystem services. Mixed logit models were used for valuation, and all attributes are significant to affect the probability of choosing an alternative scenario. The study found that households gave higher value for biodiversity improvement than the rest and water availability is the least preferable attributes. The total WTP for the first and second alternatives was 7,133,034.216 and 7,031,211.146 respectively; it was high in the second alternative. Moreover, compensating surplus which reflect overall willingness to pay for a change from the status quo to alternative improvement scenarios were estimated. Estimated results for high improvement scenario and medium improvement scenario were ETB 2,119,457.028 and 1,955,828.606 respectively.


Introduction
Environmental resources are vital for human by providing recreational opportunities, improving health situation, facilitate production of goods and services. However, quantify their value is not easy. Wetland ecosystems, are one of environmental resources (Dagnew & Aboytu, 2015;Ramachandra et al,. 2004). It contribute up to 40% of annual global ecosystem services (Zedler & Kercher, 2005). It is estimated that wetlands cover at least 6% of the earth surface (Cools, Johnston, Hattermann, Douven, & Zsuffa, 2013) and 4% of Africa's land surface (Ibrahim & Minwyelet, 2018). Different types of wetlands cover around 2% of the total surface area of Ethiopia (Getinet, 2017;Ibrahim & Minwyelet, 2018).
Though wetlands area coverage in Ethiopia is trivial, they are productive and vital sources of water during dry season for both domestic and wild animals (Tadessse & Solomon, 2014). They provide, regulating cultural services to local people (Ibrahim & Minwyelet, 2018). They are spread in central highlands, rift valley and southwest borders of the country (Israel & Timar, 2018).
Valuing of wetland ecosystems can be performed through direct, indirect market, and survey-based valuation mechanism (Davidson et al., 2006). For this purpose, choice experiment is preferable because the ability of estimating use and non-use values of choice experiment of each attribute which are provided by mountain choke wetland ecosystems.
According to the study of Getnet (2012) the attribute of choke mountain wetland ecosystem is water availability. The level of attribute is based on number of rainy seasons. However, value of water availability is not estimated exactly. The other reason behind of the study is the mountain choke wetland ecosystem has a great potential to create job opportunity for the young people.
Farming based living style of the community is cause for degradation of mountain, valuing this attribute helps to find a way to conserve choke mountain wetland ecosystem. By considering the above situations, the study starts to value choke mountain wetland ecosystem using four qualitative and one quantitative attribute. Each of the attribute is leveled reasonably.

Theoretical framework of wetlands ecosystem services
Wetlands are transitional lands between and other water bodies (Ramachandra, Rajinikanth, & Ranjini, 2004).
Ecosystem can be defined as dynamic collections of plants, animals, and microorganisms interacting with each other and their abiotic environment. Wetland ecosystem services offer a wide variety of tangible and intangible benefits to large numbers of people (McCartney, Rebelo, Sellamuttu, & Silva, 2010).
The distinction between ecosystem functions and services matters. This is because from an economic point of view what is valued by people are the end products of the various ecological processes and functions, namely the ecosystem services which affect peoples' welfare (Groot et al., 2002). Based on neoclassical economic theory, market prices are usually an adequate reference for the value that society places on goods and services (Dora & Francisco, 2011), but with environmental goods and services, market imperfections distort their real prices or values, plus the value that individuals place on them cannot be readily observed (Barbier, Acreman, Knowler, & Bureau, 1997).
Mostly environmental services are unrecognized and undervalued because many of services are forms of public good, accruing outside monetary systems (McCartney, Rebelo, Sellamuttu, & Silva, 2010). Thus, valuation of environmental resources assists to determine how much people are willing to pay for environmental goods and services, and how much better or worse off because of changes in their supply (Emerton, 2016).
Choice experiment is a recent innovation in stated preference techniques. It combines the characteristics theory of value (Lancaster, 1966) and random utility theory (McFadden, 1974). CEM has distinct advantages over other stated preference techniques; it is able to value multiple attributes. The repeated choices in CE allow for internal validity yield information concerning the consistency of individual responses (Gebretsadik, 2014). It is used to reduce potential biases of contingent valuation method, several response difficulties such as strategic behavior, protest bids, and yeah saying (Getahun, 2010).

Empirical framework of valuing wetlands ecosystem services
There are different studies concerned with ecosystem valuation to value it. On the area of national park, Ali (2011)

Description of the study area
The study area covers the choke mountain range. The absolute location of the area extends between 10 • to 11 • N and 37 • 30ꞌ to 38 • 30ꞌ E, the highest peak is located at 10 • 42ꞌ N and 37 • 50ꞌ E, and it covers 17443 km 2 . The mountain is found in Ethiopian highlands with a peak elevation of more than 4000 m (Simane, Zaitchik, & Ozdogan, 2013). In this area, the wetland ecosystems range from sedge swamps to seasonally flooded grasslands which cover above 3386km 2 (Teferi et al., 2010).
The choke mountain wetlands ecosystem is source of several springs and rivers; it is main custodian of valuable water resources. Source: (Simane et al., 2013) The number of choice set, alternatives, blocks and largest attribute level determine required sample size in CE (Bekker-grob, 2015). In the study there was 9, 3, 2 choice sets, alternatives, and blocks respectively. The largest attribute level was 6. Therefore, the study can determine sample size based on equation given below: n = 500 * ℎ * * = 500 * 6 9 * 3 * 2 = 222 Where, 'n' is number of sample size, is number of blocks. The calculated result shows that at least 222 sample sizes are required to generate efficient result from the study. In the previous study, Getenet (2012) used 250-sample size to value the Choke mountain wetland ecosystem. The researcher restricted total sample size in to 270, and it was distributed for two kebeles proportionally. Identifying the attributes and attribute levels is the first step to undertake choice experiment survey design (Mesfin, 2010;Getenet, 2012;Latinopoulos, 2014;Birol et al., 2006). This identifies residents' opinions about current issues and problems related to the mountain, and five attributes (biodiversity, water availability for domestic supply and irrigation, recreational quality, job opportunity and annual cost for service improvements) are selected to describe choke mountain wetland ecosystem.  The CEM is a variant of conjoint analysis, which was primarily developed by Louviere and Hensher (1982), and Louviere and Woodworth (1983), and has its roots in Lancaster's characteristics theory of value, in random utility theory and in experimental design (Latinopoulos, 2014). According to random utility theory (Luce, 1959;McFadden, 1973), utility function for a choice alternative j held by a respondent i (Uij) is a function of the attributes of the alternatives presented to him/her (Zij): The assumption is that stochastic components are independently and identically distributed (IID) with a Gumbel distribution. This leads to the use of MNL or CLM model to determine probabilities of choosing g over h options (Hanley, Mourato and Wright, 2001).
Pr(Uig > Uih) = ∑ = , Vig≠Vih and, h ∈ …………………………………… (3) Where, Vi is indirect utility function, and is a scale parameter inversely related to the standard deviation of the error term means that the higher the scale parameter the lower the variance of the error term and hence the higher the model fits, and it is not separately identifiable in a single data set.
Utility may also depend on a set of individual socio-economic characteristics of respondent i.
The specification of the models in the study is expressed as follows: Vi= ASC+ b1BD+ b2 W + b3RQ + b4JO + b4Price…………………………… (5) Basic model Where: ASC = 0 for plan 3 (status quo option) and one for plan1 and plan2. In addition to this b1, b2, b3, b4and b5 are the coefficients associated with each of five attributes, i.e. improvement in biodiversity, availability of water for domestic supply and irrigation, recreation quality, job opportunity and annual cost for improvement respectively.

Random parameter /mixed logit model
The MNL and CLM models come with main problems. First, they assume that no correlation among the unobserved disturbance terms; it relies on the assumption of independence of irrelevant alternatives (IIA) that cannot be always realistic. Second, it fails to take the taste variation of individuals in to considerations, and third it violates axioms of stability and transitivity consumer's choices (Olila, Nyikal, & Otieno, 2015). However, RPLM (mixed logit model) is better since it takes in to account preference heterogeneity by allowing utility parameters to vary randomly and continuously over individuals and through interaction of the RPL model with individual's sociodemographic characteristics (Ibid). Therefore, IIA assumption is unnecessary in RPL model. The random utility function for the RPL model takes the following form: (7) Where, Uij is the total utility for respondent i Where, β are the estimated coefficients of the attributes in multinomial or random parameter logit model. The compensating surplus relating to a change in overall conditions can be also estimated by using the following formula: Compensating surplus (welfare change) = -(1/β monetary attribute) (V0-Vi) .…………….. (9) Where, V0 is the indirect utility at status quo, Vi is indirect utility with different alternative improvement scenarios with their specific levels of the attributes, and β is the estimated coefficient for monetary attribute.
Beyond welfare change, the study estimated marginal rate of substitution between non-monetary attributes. MRSxy represents the trade-offs between attributes. This can be calculated as the value of one attribute in terms of another attributes. MRSxy = where, indicates the marginal utility generated from an improvement of attribute "X" and indicates the marginal utility generated from an improvement of attribute "Y".

Results and discussion
The study was conducted by using 270 households as a sample size.

Source: Authors computation, 2020
The choke mountain wetland provides so many advantages for households by providing clean air and water, by conserving the soil fertility, by providing foods for their domestic animals, and as a means of farmland. In addition to this, it is very important for carbon sequestration. It also supplies water for the Nile dam (it constitutes around 9.5 percent from the total). Direct consumption, animal grazing, irrigation, farmland and soil conservation are the benefits of choke mountain wetland ecosystem. Direct consumption includes purposes of the area for drinking water, recreation, wood and hunting. The responses are summarized in table 4.3 below.

Source: Authors computation, 2020
According to the study, most of households in surrounding area used choke mountain wetland for animal grazing. The second purpose was direct consumption. It is clear that there are several rivers and springs originate from mountain. Therefore, it is used as a source of water for drinking. Its geographical location, animals and plants, spiritual and historical places have a power to attract individual's attention.

Source: Authors computation, 2020
Overgrazing was ranked as first factor for degradation of the area since it was reported as most frequent problem by 78.15 percent of respondents. The reason behind the result is too many cattle are using the mountain as a source of grass, and they feed it throughout the year. This resulted in over degradation of the area. The livelihoods of the farmers around the mountain highly depend on livestock rearing and farming. The second major problem was deforestation because farmers are using forests of mountain for construction of houses and fuel wood etc. The remaining factors, resettlements and agricultural land expansion problems, were ranked equally next to overgrazing and deforestation.

Source: Authors computation, 2020
The people who have enough awareness about the choke mountain wetland ecosystem services constitute around 34.44 percent. It covers 1/3 of the total respondent. The choke mountain wetland ecosystem service has a public good nature. Goods and services that have a public good nature are affected with a problem of the tragedy of the common.  The random parameter logit model can solve problem of violation of IIA assumption. Therefore, model is estimated in to account for IIA assumption and unobservable preference heterogeneity across respondents. shows that standard deviation of bd3, w3, rq2, jo2 and jo3 attribute levels statistically significant at 1 percent level of significance, and he remaining rq3 and w2 are statistically significant at 5 percent and 10 percent level of significance respectively.
However, mixed logit model cannot show source of heterogeneity. Therefore, estimating mixed logit model with interaction is necessary. The interaction of biodiversity and recreational quality with family size is positive and statistically significant at 1 percent level of significance. This implies that as number of family size increase, their willingness to choose improved level of biodiversity and recreational quality also increase.
This indicates that households consider payment for improvement of attributes; payment per each family member is low compared with benefit they gain as number of family size increase.
The interaction between improved level of water availability and age the respondent is positive and statistically significant at 10 percent level of significance. This shows that effort of individual to fetch water from springs or rivers and to use the river for irrigation without modern machine is decline as their age rise; therefore, they need improvement in water availability service.
The interaction of job opportunity with age of respondent and distance from mountain to house is negative at 1 percent and 5 percent level of significance. This implies that demand of individuals for creation of job opportunity declines when their age increase and their house far away the mountain.
The interaction between recreational quality and individual being married is negative. This tells that individuals who have gotten marriage have lower preference for improved recreational quality as compared with single, divorced, widowed and others. The rationality behind this result is that married individuals, particularly in rural area have not that much free time to refresh themselves.
Mostly they spend their free time with in a house or by visiting their relatives. In addition to this, the result shows that these households do not give attention for income generated from tourists.
The income levels of the household have also negative relationship with the improvement of recreational quality.
The other significant interaction is between water availability and occupation. The result shows that farmer is less likely to preferred improved water availability attribute as compared with merchants, public sector employees, private sector employees and others. This can be because of lack of fertile and enough farmland for irrigation. Furthermore, fetching water from rivers and springs is easy for farmers than others. As the interaction between job opportunity and occupation shows, farmers need job opportunity compared with others. An improvement of biodiversity helps to improve water supply, recreational services and creation of job opportunity. Therefore, it is able to be a backbone for improvement of other attributes.
However, it is not used for such purpose compared to its potential due lack of infrastructural development. Therefore, improving recreational quality helps to use full potential of site in area of ecotourism. In addition to this, improvement of job opportunity helps to increase income level of the households. The productivity of crop is low due to acidic nature of soil and cold weather condition of the area. Therefore, from the respondents' point of view improving job opportunity is better than improving water availability.
From the estimated result of marginal WTP, it is possible to derive total WTP by using probit  From the above table, value of biodiversity attribute expressed in terms of water availability attribute is highest relative to other trade-offs estimated in this study. This implies that households scarify more of water availability improvement to gain an additional improved level of biodiversity attribute at citrus paribus. The second highest trade-off is between recreational quality and water availability attribute. This also tells that households scarify more of water availability improvement to gain an additional improved level recreational quality, but they scarify more water availability improvement to get additional improved level of biodiversity than recreational quality improvement. The third highest trade-off is between improvement of job opportunity and improvement of water availability attributes. The trade-off between improvement of biodiversity and recreational quality is lower relative to the trade-off biodiversity with other attributes. This indicates that respondents scarify relatively less recreational quality improvement to gain an additional improved level of biodiversity.
Economic welfare measurement is calculated as the difference between the individuals' utilities that could be achieved under the status quo and changed scenario alternatives. To compute the welfare change, first, the values of the attributes in the status quo alternatives are substituted into the indirect utility function. Next, the values of the attributes in changed situation of scenarios are substituted into the indirect utility function, and then after the value of the alternative with a changed situation are subtracted from the value in the status quo alternative. Lastly multiplying this by the negative inverse of the coefficient of the monetary attribute gives welfare change (Ali, 2011 This implies implementation of improvement scenario one and improvement scenario two through policy intervention will generate ETB 2,119,457.028 and 1,955,828.606 welfare gain respectively.
Therefore, implementation of improvement scenario one is preferable to obtain more welfare change.

Conclusion
Choke mountain wetland ecosystem is found in Ethiopia, Amhara region East Gojam zone. It is the source of several springs and rivers; it is the main custodian of valuable water resources. It covers the 9.5 percent of Upper Blue Nile River that supplies water for millions of downstream Nile basin countries particularly Sudan and Egypt. Therefore, valuing this wetland ecosystem is useful for different purposes and at different scales in support of wetland wise use management and decision-making.
The objective of the study was to value of choke mountain wetland ecosystem. Choice experiment valuation method is preferable in the study because choice experiment valuation method has the ability to estimate both use and non-use values each attribute of wetland ecosystems. These attributes in the study are biodiversity, recreational quality, water availability, job opportunity and annual monetary payment. Choke mountain wetland ecosystem services compared to the current situation; they gave higher value for biodiversity improvement than the remaining attributes.
However, the monetary cost is negative and significant in mixed logit models. This indicates that as cost of improving mountain's wetland ecosystem services higher, probability of households to prefer improved alternative becomes lower.
The marginal WTP of water availability is lowest compared with other attributes because improvement of biodiversity increases the potential of the choke mountain wetland ecosystem to supply water, and improvement of recreational quality also satisfy demand for households to clean water. The total WTP for the first and second improvement alternatives was ETB 7,133,034.216 and 7,031,211.146 respectively; it was high in first improvement alternative. Furthermore, compensating surplus which reflect overall willingness to pay for change from status quo to alternative improvement scenarios were valued. Estimated results for improvement scenario one and two were ETB 2,119,457.028 and 1,955,828.606 respectively. This indicates that implementation of the improvement scenario one is preferable to obtain more welfare change. The study also estimated marginal rate of substitution between non-monetary attributes. The trade-off between biodiversity and water availability was dominated over the other trade-offs. The second highest trade-offs are between recreational quality and water availability followed by trade-off between job opportunity and water availability.