Trees, Forests and People 12 (2023) 100398

Contents lists available at ScienceDirect

Trees, Forests and People

journal homepage: www.sciencedirect.com/journal/trees-forests-and-people

Economic valuation of urban forest using contingent valuation method: The

case of Hawassa city, Ethiopia
Girmaye Zegeye a, Solomon Erifo a, Getu Addis b, Girma Gezimu Gebre c, d, *
a
Department of Natural Resource Economics and Policy, Wondo Genet College of Forestry and Natural Resource, Hawassa University, Ethiopia
b
Departments of Rural Development and Agricultural Extension, College of Agriculture, Hawassa University, Ethiopia
c
Departments of Agribusiness and Value Chain Management, College of Agriculture, Hawassa University, Ethiopia
d
The Japan Society for the Promotion of Science (JSPS) Postdoctoral Research Fellowship Program, Ritsumeikan University, Kyoto 603-8577, Japan

A R T I C L E I N F O A B S T R A C T

Keywords: Urban forests are a significant and essential, but frequently ignored, component of urban landscapes. In many
Urban forest cases, only the resource’s commercial value is taken into account when developing policies. To this end, the
Urban forest park contingent valuation method (CVM) was applied to evaluate the economic value of urban forest parks (UFPs) in
Contingent valuation method
Hawassa City. This paper is estimating households’ maximum willingness to pay (WTP) for urban forest parks
Willingness to pay
(UFPs) and identifying factors affecting households’ (HH’s) WTP for conservation and development UFPs using
double bounded (DD) CVM. The data was analyzed using a bivariate probit and probit model. Using the double
bounded and open ended (OE) strategy choices, it was found that the communities’ joint WTP was worth
23,209,329.6 Ethiopia Birr (ETB) and 14,761,008 ETB, respectively. The cost of conservation and development
of UFP was found to be 1,696,000 ETB. According to the research, age, sex, marital status, family size, income,
ownership status, education level, and bid amount have significant impact on the project’s WTP. In general, the
study showed that residents might gain approximately (23,209,329.6 - 1,696,000ETB) 21,513,329.6 ETB as a
welfare enhancement by comparing the cost of conservation and development of UFPs with their benefits;
consequently, such projects would benefit them more. As a result, the government should place greater emphasis
on its UFP policies and projects. Besides, specific socioeconomic aspects should be considered when developing a
project or policy for the conservation and development of UFPs in urban areas.

1. Introduction shelterbelts along roads and railways, and so on. The environment in
urban regions is different from that in rural areas in terms of meteoro­
United Nations created 17 Sustainable Development Goals (SDGs) as logical and edaphic conditions, habitat surface and connectivity, plant
a result of extensive discussions among its member states to come to an composition and structure, pollution, and intense human activity
agreement on the world we want by 2030. One of the most notable goals (Aronson et al., 2017; Dobrosavljevi et al., 2020). By 2050, urban areas
of the 17th SDGs is SDG 15, which aims to "protect, restore, and promote are expected to accommodate 68% of the world’s population (UN,
sustainable use of terrestrial ecosystems, sustainably manage forests, 2018). However, for both areas, forests provide income, energy, and
avoid desertification, halt and reverse land degradation, and halt food, which are essentially for their survival. Finally, environmental
biodiversity loss" (Glass and Newig, 2019). Forests are among the most deterioration and shifting land use are putting increasing pressure on
species-rich environments on the planet. Because of escalating urbani­ forests in cities and rural areas (Adekunle et al., 2008; Gashu and
zation, rising human population, and escalating demand for food and Gebregziabher, 2019; Assefa et al., 2021; Piana et al., 2021).
natural resources, their area is rapidly shrinking. There are many simi­ Forests are especially important in urban areas because they improve
larities between the urban and rural forest environments. For instance, human health while lowering greenhouse gas (GHG) emissions and air
there is a significant demand for forest goods and services because of pollution (Adekunle et al., 2008; Gashu and Gebregziabher, 2019;
high population pressure in both locations. However, in urban areas, Assefa et al., 2021). In addition, forests provide inhabitants with benefits
woody vegetation is also present in the form of tree rows, parks, gardens, such as carbon dioxide sequestration, nutrient recycling, and

* Corresponding author.
E-mail address: grima.gezimu@gmail.com (G.G. Gebre).

https://doi.org/10.1016/j.tfp.2023.100398

Available online 2 June 2023

2666-7193/© 2023 The Author(s). Published by Elsevier B.V. This is an open access article under the CC BY license (http://creativecommons.org/licenses/by/4.0/).
G. Zegeye et al. Trees, Forests and People 12 (2023) 100398

biodiversity preservation (Gashu and Gebregziabher, 2019). The other necessitates rapid action if urban woodlands continue to provide envi­
service functions of forests in urban areas that play a crucial role in ronmental services, particularly in metropolitan settings. The continu­
urban dwellers include pleasant landscapes, quiet, and potential recre­ ation of forest resource products and services should be a focus of
ational activities (Adekunle et al., 2008; Gashu and Gebregziabher, concern (Belkayali et al., 2010; Tavárez and Elbakdze, 2021).
2019; Assefa et al., 2021; Tavárez and Elbakdze, 2021). Consequently, The various benefits of urban forests, both marketed (especially
urban dwellers are beginning to understand and articulate the impor­ those goods that are directly harvested from forests and woodland re­
tance of forests as part of the city’s infrastructure, quality of life, and sources by millions of people around the world, such as fuel wood,
extensive recreational use (Adekunle et al., 2008; Abebe et al., 2019; medicinal products, and wild foods, which escape trade and market
Gashu and Gebregziabher, 2019; Assefa et al., 2021; Tavárez and exchange (Small et al., 2017)) and non-marketed in developing nations,
Elbakdze, 2021). However, environmental deterioration has brought are often ignored. However, various development options for the area
about significant problems that have hurt the capacity of urban forests to may be attractive to politicians and bureaucrats because of the imme­
provide these goods and services. The challenges of environmental diate return from development projects (Tavárez and Elbakdze, 2021).
deterioration in metropolitan settings will become even more urgent in Since these advantages are said to have high social value through their
the following decades because population pressure is an unavoidable contribution to individual wellbeing, the absence of a market does not
fact (Tyrvainen et al., 2005; Adekunle et al., 2008; Belkayali et al., always indicate the absence of value (Yirssaw, 2021).
2010). In this regard, a study by Wieditz and Penney (2007) demon­ It has been unable to accurately gage both the volume of resources
strated that environmental concerns make it increasingly difficult for and the contribution of UFPs’ goods and services to Ethiopian society’s
urban trees to survive and mature, and to offer protection from the ef­ welfare. Owing to rural-to-urban migration, a high rate of natural pop­
fects of hot weather, powerful storms, and other effects of climate ulation growth, and rising per capita income, Ethiopia’s metropolitan
change. areas have experienced rapid population expansion in recent years.
This study attempts to understand the economic importance of urban Consequently, an increased demand for crucial infrastructure and gov­
forests using the contingent valuation method (CVM). Ecosystem ser­ ernment services is anticipated (Assefa et al., 2021; Abebe et al., 2019;
vices most likely to be priced in markets are those with consumptive, Gashu and Gebregziabher, 2019). Consequently, green spaces are under
direct use values, while non-consumptive use values are rarely valued. increasing stress. Therefore, there is now less green space in Ethiopian
The stated preference method is best known for capturing the value of cities and the pressure from environmental damage and changing land
both consumptive and non-consumptive values. CVM is the best-known use is mounting. By determining their total economic value (TEV) ,1 this
example of the stated preference method. The stated preference method study attempts to quantify the advantages of urban forest resources in
uses a direct approach to elicit WTP, using choice modeling and CVM to Hawassa. Therefore, it is advantageous to articulate and clarify the issue
evaluate environmental changes (Small et al., 2017). The CVM seeks to of reducing ambiguity among policymakers and bureaucrats.
discover how people would value certain environmental changes by Hawassa City is one of the communities in a nation with the fastest
creating a hypothetical market by directly questioning a sample of the economic growth. It is understood that growing urbanization will
population (Haab et al., 2020). Urban forests are an important and in­ improve societal welfare when accompanied by the construction of
tegral, yet often overlooked, portion of urban environments (Yirssaw, buildings, roads, and other infrastructure; however, Hawassa’s rapid
2021). In many instances, only the commercial value of this resource is urbanization and high demand for several development projects have
considered in the policy development. This situation is more prominent resulted in an unwarranted loss of urban forest resources, endangering
in developing nations (Tavárez and Elbakdze, 2021). their future (Hawassa City Municipality (HCM), 2015). Additionally,
There is also a lack of economic literature on the economic value of forests are not effectively managed, and their economic benefits have
urban forest ecosystems in developing countries, and the economic not been fully realized. As the TEV of these resources cannot be effec­
benefits of these resources are still being assessed (Dumenu, 2013; tively communicated, society’s well-being cannot be assured (Ning and
Chaudhry et al., 2013; Small et al., 2017; Tavárez and Elbakdze, 2021; Lee, 2019; Tyrvainen et al., 2005; Nowak, 2020). For instance, the
Jamean and Abas, 2023). Policymakers frequently consider the re­ construction of roads, condominiums, buildings, and other development
source’s direct economic value (Hanemann, 2006; Ikeuchi et al., 2013; work that is carried out in the city takes place in areas left for greening in
Tavárez and Elbakdze, 2021; Zhu and Shin, 2021). They argue that the main plan and other areas that used to be covered by indigenous
politicians and bureaucrats may be drawn to various regional develop­ trees (Gashu and Gebre-Egziabher, 2019).
ment alternatives because of the quick returns from development pro­ For the purpose of this study, authors took the following definition of
jects. However, the numerous advantages of urban forests in developing urban forest as a standard: It includes trees in yards, along streets and
countries, both as commodities that are sold and those that are not, utility corridors, in protected areas, and in watersheds. This extends to
eventually receive less attention. Urban forest resources are also dis­ individual trees, street trees, tree-lined parks, and even the vegetation
regarded from conventional economic viewpoints when evaluating re­
sources because of market failures. Therefore, unmarketed urban forest
services are rarely used (Belkayali et al., 2010). Therefore, authorities’ 1
Both use and non-use values are included in the TEV framework. There are
decisions on development projects, such as the construction of schools, three types of use values: direct use, indirect use, and option values. Existence
shopping centers, housing apartments, and other structures in open and value and bequest value are examples of non-use values (TEEB, 2010). Tavárez
green spaces, would be biased if the advantages of urban forest areas and Elbakdze (2021) use the general technique of allocating TEV of natural
were not correctly assessed (Tavárez and Elbakdze, 2021). resources to urban forests. For this study, TEV of urban forests was divided into
Therefore, it is crucial to have the motivation to create a strategy for six categories: purely non-consumptive values (aesthetic values, recreational
using natural resources responsibly based on the individual choice value, shading services, and so forth); indirect values (protection functions and
values of realized price corrections (Belkayali et al., 2010). Significant climate mitigation); and consumptive values (food, timber, mushrooms, medi­
research has been conducted on these issues, highlighting the impor­ cine, and so forth); option, bequest, and existence were ranked fourth, fifth, and
tance of technical and green space management sciences. Today, eco­ sixth, respectively. A direct use value is easiest to account for because it is
typically a component of formal marketplaces, according to TEEB (2010) and
nomic valuation studies have revealed many previously unattainable
Tavárez and Elbakdze (2021). Since non-use values are subject to the greatest
insights (understating the economic values of nonuse value) into the
degree of uncertainty, non-use values (NUVs) are more difficult to define than
benefits and expenses of urban green spaces and their management use values (UVs). Furthermore, as we move from consumptive UV to existence
(Tyrvainen et al., 2005). The challenge for economic research is to value, the tangibility of the goods and services to measure and value declines.
determine how to identify these values and compare them with those of Therefore, it was anticipated that such a conclusion would be found for option,
other human endeavors that jeopardize urban forest survival. This bequest, and existence values (see Appendices 4a and 4b).

2
G. Zegeye et al. Trees, Forests and People 12 (2023) 100398

and soil that grow beside and beneath the trees (Lafortezza et al., 2017; often neglected as valuable environmental assets. Another excellent
Konijnendijk van den Bosch et al., 2017). Therefore, this study examines example in this area is Davies et al. (2023) study on Southapton’s
how households view urban forest goods and services and associated willingness-to-pay for the provision of urban ecosystem services under
environmental issues, as well as the economic value of urban forests and both objective and subjective uncertainty.
the variables affecting households’ WTP for conservation and develop­
ment of urban forest parks (UFPs). 3. Material and methods

2. Literature review 3.1. Geographical description of the study area

The conventional measures of economic output, national income, Hawassa city was established in 1960. The name of the city comes
and wealth underestimate the values and contribution of natural re­ from Lake Hawassa, which is to the west of the settlement, and Hawassa
sources to economic welfare. This is particularly true for forests and means "large water body" in "Sidama language" The city has been the
woodland resources, which are collected from open access and escape seat of the former SNNPRS and the Sidama Zone since 1991, and it
trade and market exchange. Technical difficulties remain unresolved served as the capital of the previous Sidamo province beginning in 1968.
when measuring the non-market value of intangible goods and services Currently, the city is the capital city of the newly formed Sidama Region.
of forests and woodland resources (Van de Ven, 2018; Sajise et al., Hawassa is one of the nation’s fastest-growing cities (HCM, 2011). With
2021). The total economic value (TEV) framework is used to assess the direct accountability to regional governments, Hawassa has become a
economic value of ecosystem services. It consists of use and non-use self-governing city. Addis Ababa is 275 km (approximately 170.88 mi)
values, with direct use values being the easiest to account for and south of the city. The elevation of the city ranges from 1692 to 1742 m
non-use values being the most difficult to measure and having the above sea level. With an average yearly temperature of 19.5 0C, the
greatest uncertainty (Nijnik and Miller, 2017; Sajise et al., 2021). climate is categorized as sub-humid tropical. The city is bordered to the
The valuation of non-market environmental resources is conducted west by Lake Hawassa, to the east by Wondo Genet Woreda, to the south
in two ways. The first method, which is based on individual preferences, by Shebedino Woreda, and to the north by Oromia (Fig. 2). Its co­
is rigorously based on either observed behavior (revealed preference, ordinates are 38◦ 24′ to 38◦ 40′ east longitude and 06◦ 49′ to 07◦ 10′
RP) towards some marketed good with a connection to the non- north latitude. The city administration (municipality), eight sub-city
marketed good of interest or stated preferences (SP) in surveys with administrations (Kifle Ketema), and 32 kebele administrations are the
respect to the non-marketed good and is rooted in welfare economics. three levels of governance that constitute the city (HCM, 2011).
The second method is based on the preferences of policymakers, scien­ The city was home to 316,734 people as per the 2007 G.C. census,
tific experts, or specific interest groups. These methods include implicit including 162,928 men and 153,806 women. The population of the
valuation, the Delphi method, and multi-criteria analysis (Christian city’s urban regions was 201,020, while its rural areas had 115,714
et al., 2018; Johnston et al., 2017; Sajise et al., 2021). However, only the residents. Urban areas make up 63.5 percent of the population in the city
second method is discussed in the next subsection, as stated preferences administration, while rural areas make up 36.5 percent. The city’s
(SP) will be presented. According to Ben-Akiva et al. (2019), SP methods growth rate of 4.02 (4.8 in the city and 2.8 in the countryside) was used
use a direct approach to elicit WTP, such as choice modeling and CVM. to forecast or approximate current data (CSA, 2007). A total of 528,706
Choice modeling does not ask questions directly, instead it ask people to people, including 314,090 men and 296,505 women, will reside in the
rank alternatives. The CVM seeks to discover how people would value city in 2021.
certain environmental changes by directly questioning a sample of the The following summarizes the conditions of urban green infrastruc­
population concerned. The valuation task is to determine how much ture in Hawassa. Out of a total of 4158.5 ha, the Hawassa city admin­
better or worse an individual is as a result of a change in an environ­ istration area has 4.1% forest cover, or 169.5 ha, and 18.5%, or 1912.5
mental resource. The total value of goods can be calculated by multi­ ha, of various types of urban green infrastructure. In Hawassa City,
plying the average WTP or WTA by the population size. forests and various types of urban green infrastructure make up
A particular advantage of CVM over other valuation techniques is approximately 22% of the entire urban area (Hawassa City Municipal­
that it is specifically designed to identify non-use values. By exclusively ity, 2015). According to Benti et al. (2021), Ethiopian urban planning
avoiding economic pricing models, CVM aids in determining the value of standards are 40%, 30%, and 30% for built-up spaces, roads, and green
environmental resources and non-market commodities. Despite criti­ infrastructure, respectively. This standard stipulates that approximately
cism, this is the only trustworthy paradigm that supports environmental 70% of the urban land should be allocated for physical planning. The
evaluation. The approach is also used for the valuation of wilderness as result show that the total coverage of green infrastructure in Hawassa is
well as common environmental facilities such as forests, common lands, below average. This shows that there are still plenty of efforts to meet
national parks, and common water bodies (Nautiyal and Goel, 2021). these criteria. While roundabouts have the least coverage in Hawassa,
Because the CVM at first glance seems very easy to use, it is vulnerable to gardens and yards account for the majority of the various types of urban
misuse, which can easily undermine confidence in the method. There­ green infrastructure. Forest parks make up approximately 18.5% of the
fore, the construction and framing of a CV survey should therefore be various types of green infrastructure (Gashu and Gebre-Egziabher,
carefully examined before the results of it can be used for policy pur­ 2019). In addition to the concepts of this study, both primary and sec­
poses. It has many criticisms concerning its inconsistency with rational ondary patches of forest and man-made forests in Hawassa are consid­
choice, the implausibility of responses, the absence of a meaningful ered urban forest parks.
budget constraint, and the extent of information provision and accep­ Any tree, even those in and around the urban setting, is considered
tance by the market. Some authors have used experiments to compare part of the urban forest. Urban forests include trees found in parks,
the stated values with the actual values for a range of goods. On tech­ squares, institutions, along rivers, and around lakes. In addition, urban
nical grounds alone, several potential biases may arise that could un­ woods offer vital ecological and commercial benefits to cities and people
dermine the validity of information gathered using the CVM (Christian who live within. For this study, the researchers used the definition of
et al., 2018; Johnston et al., 2017; Sajise et al., 2021). urban forests given by Lafortezza et al. (2017) and van den Bosch et al.
A notable example of WTP in an urban setting is the study conducted (2017). According to them, the trees and shrubs in a city constitute an
by Yirssaw (2021) to assess the total economic value of urban green urban forest. It includes trees in the yards, along streets and utility
spaces protected in Bahir Dar city. He used the contingent value tech­ corridors, protected areas, and watersheds. This extends to individual
nique (CVM) to evaluate people’s willingness to pay (WTP) to preserve trees, street trees, tree-lined parks, and even vegetation and soil that
and enhance urban green spaces, which are in danger of extinction and grow beside and beneath the trees. Researchers are required to provide

3
G. Zegeye et al. Trees, Forests and People 12 (2023) 100398

more detailed definitions of urban forests in relation to this particular precision, 95 percent confidence level, and 10 percent contingency were
study in Hawassa City. As a result, the "Tikur Wuha" River, other sea­ used. Therefore, the sample size of the study was 318 using the Cochran
sonal stream tributaries to Lake Hawassa, and "Amora Gedel" Parks are method. The formula used in this study is as follows:
all considered part of the urban forest in the city for the purposes of this (z)2
study. In addition, trees can be found throughout Hawassa city. Urban n= p(1 − p) + 10% contingency
e
trees in the research area include those that are planted inside homes
and between roads, as well as those that are planted along streets, in Therefore, the sample size was computed as follow:
various villages, and on pedestrian roads. ( )2
1.96
According to Loomis et al. (2000) and Awad and Holländer (2010), n= 0.25(1 − 0.25) + 10% contingency
.05
accurate benefit estimates using the CVM require detailed descriptions
of the resources to be valued. Consequently, urban forest products and = 289 + 10% contingency (10% * 289 = 29)
services were specified and presented to respondents in both current and = 289 + 29
projected amounts. Finally, for the purpose of valuing urban forest parks Total Sample Size = 318
in Hawassa city, the researcher divide urban forest goods and services Although the sample size was 318, we added 2 more individuals to
into a sixth category (Fig. 1). boost the results; therefore, a total of 320 individuals were included in
this study.

3.2. Data and sampling procedures

3.3. Value elicitation format
Primary data collection was the main source of the study. Addi­
In this study, the WTP of the HHs for the conservation and devel­
tionally, secondary data from the Hawassa city municipality and the
opment of UFPs in Hawassa was elicited using the double dichotomous
natural resource units of the chosen sub-cities were used for the sample
(DD) technique. This strategy decreases non-responses and removes
frames and socioeconomic data. Various offices were consulted in order
outliers because DD ensures that the respondent tells the truth. The
to obtain additional information.
National Oceanic and Atmospheric Administration (NOAA, 1993) panel
Multistage sampling procedure was used to select sample house­
approved the method, which is also preferable to a single dichotomous
holds. First, out of eight sub-cities, three were randomly selected. These
(SD) because it elicits more details about each respondent’s WTP.
include Menaharia, Bahil Aderash, and Adiss Ketema. Second, three
Open-end (OE) and SD elicitation techniques were employed to improve
kebeles (Harer, Piassa, and Philadelphia) were selected from three
the reliability of the findings and widen the scope of cross-checking the
selected sub-cities, one from each. Selection of Kebeles were based on
veracity of the respondents. The beginning and upper bids were selected
economic factors. For instance, Menaheria (Piassa) is a high-income
during the pilot survey to enable the exact positioning of the extremities
residential area; Bahl Adarash (Harer) a middle-income residential
on the cumulative probability distribution. The lowest offer was chosen
area; and Adiss Ketema (Philadelphia) a low-income residential area.
because it was acceptable. Owing to these limitations, ETB was set to 20.
Finally, 320 HHs were selected for the survey using simple random
The upper offer was chosen because it was unacceptable. Comparable
sampling.
results were found for the upper bid amount, indicating that most re­
It is crucial to choose a suitable sample size for household surveys.
spondents disapproved of it. Two hundred ETB is the agreed-upon price.
Three crucial factors were considered when determining the sample size
Based on this, the first five offers of 20, 40, 80, 120, and 200 ETB were
for this study: 1) Error margin, 2) confidence level, and 3) percentage of
selected.
the sample that selected a particular response to a survey question.
Additionally, a prediction of the expected proportion of successes must
be made based on intuition or prior information because the size of the 3.4. Questionnaire design
proportion influences random sampling error (Cochran, 1977). Hussey
and Hussey (1997) stated that confidence intervals greater than 90% Several important informants carefully developed and tested the
and error bounds less than 10% are acceptable. To accommodate those questionnaire, bid values, payment methods, and CVM questions. On
who cannot be reached, many researchers have regularly increased the April 25, 2021, a conference was held with important informants. Ten
sample size by 10%. A sample’s likelihood of selecting a certain response key informants were selected, and factors such as age, gender (five males
to a survey question was p= 0.25 and q= 1-p= 0.75; as a result, 5 percent and five females), educational background (university, college, junior

Fig. 1. The Total Economic Value of Urban Forests.

4
G. Zegeye et al. Trees, Forests and People 12 (2023) 100398

Fig. 2. Map of Hawassa City Administration (HCA, 2021).

and senior high school, illiterate groups), and occupation (taxi driver, respondents were asked why the WTP in the OE was less than the
technician, government employer, non government employer, owner of pre-agreed amount in the DD choice. It was believed that doing so would
small businesses, stay-at-home parents, etc.) were all considered (20 to make it possible to compare the data obtained using various elicitation
50 years old). In addition, a pretest trial was conducted before techniques and to divide a person’s total WTP into various advantages
completing the official survey. Ten pre-test interviews were conducted and services offered by urban tree resources (see Appendix 4b). The
over the weeks May 4–8, 2021. Four male and eight female pre-test socioeconomic factors and demographic details of HH are discussed in
participants, whose ages ranged from 23 to 60 years, were chosen at the last section.
randomly.
The survey was conducted between May 22 and June 15, 2021. To
analyze HHs’ WTP and identify the factors that influence HHs’ WTP for 3.5. Data analysis
the conservation and development of UFPs, a CV survey was conducted
in Hawassa. In addition to facts like payment situation, contingent Descriptive statistics and an econometric model were used to eval­
choice question with debriefing questions, and socioeconomic statistics, uate data from the CV study. Using descriptive statistical methods, such
the questionnaire also asked questions about views and ethics. Twelve as frequencies, means, modes, and percentages, variables of interest,
enumerators were engaged and trained in the survey, which was con­ such as age, education, income, sex, and WTP values, were examined.
ducted by the researcher. The CVM questionnaire is challenging to SPSS version 20 was used to analyze the descriptive data. Probit and
administer, and throughout the training, consideration was given to bivariate probit models were used to determine the variables influencing
enumerators’ educational backgrounds. Before asking CV questions, HHs’ WTP and to compute the mean and aggregate HHs’ WTP for the
respondents were given a set of information cards that provided a clear conservation and development of UFPs in Hawassa. STATA version 10
explanation of what an urban forest was as well as lists of its six values was used to investigate several regressions, including the probit and
and benefits. This helped them better understand the situation. bivariate probit models.
The questionnaire consists of five components. The study region is
described in detail in the first section. Respondents’ attitudes toward 3.5.1. The probit models
and perceptions of environmental issues in general, as well as defores­ The equation below represents the probit model used to identify the
tation problems and the city’s tree resources, are covered in the second crucial factors that significantly affect a person’s maximum willingness
segment. The third section of the survey included basic inquiries about to pay for the establishment and preservation of urban forest parks in the
beliefs regarding the use of urban woodlands. The respondents’ rela­ research area. As a dichotomous variable, the response (dependent)
tionship with urban forest resources is also presented in this section. Part variable in this study can have two values: 1 if the respondents are
four addresses the valuation dilemma. The scenario attempted to pro­ willing to pay (a "yes" response), and 0 if they are not (a no response).
vide the respondent with as much information as possible about the Qualitative response models should be used to estimate this type of as­
hypothetical market. Principal factors were incorporated into the sce­ sociation. Probit and logit models are related. Both logit and probit are
nario, as recommended by Christian et al. (2018), Johnston et al. (2017), quantile functions, or the inverse of the cumulative distribution function
and Sajise et al. (2021). WTP inquiries are the main topic in this (CDF) of a probability distribution, because they are sigmoid functions
segment. The OE question was utilized after the DD option question, and with domains between 0 and 1. The quantile function of the logistic
distribution is logit, whereas that of the normal distribution is probit.

5
G. Zegeye et al. Trees, Forests and People 12 (2023) 100398

There is no compelling reason to choose between the logit and probit YY = 1 for a yes-yes answer and 0 otherwise, NY =1 for a no-yes
models. Simply explained, people prefer logit to probit because of its answer, and 0 otherwise.
mathematical ease; this shouldn’t be a problem now that academics
have created computer models (Klieštika et al., 2014). Because the And the jth contribution to the bivariate probit Likelihood function
normal distribution function (probit model) closely approximates the becomes.
cumulative normal distribution, it was selected for this investigation.
The normal distribution serves as the probit link function, and there is a Lj (μ / t) = Φε1ε2 (d1j ((t1-μ1)/σ1), d2j ((t2-μ2)/σ2), d1j d2j ρ).
multivariate normal distribution as well. This implies that (if you use a Where: Φε1ε2 = Standardized bivariate normal distribution function
probit), you can alter the covariance matrix while simultaneously esti­ with zero means
mating a number of response variables.
Conservation and development of UFPs appears as a variable in y1j = 1 if the response to the first question is yes, and 0 otherwise
people’s utility functions: y2j = 1 if the response to the second question is yes, 0 otherwise
d1j = 2y1j-1, and d2j = 2y2j-1
U = U (X, Z, F)
ρ = Correlation Coefficients
Where F represents the goods and services that individuals obtain from σ = Standard deviation of the error
urban forest parks; Z represents other environmental resources that a
person is concerned about; and X represents a vector of market-valued The mean WTP (μ) was calculated using the following formula of
goods and services. UFP conservation and development are advanta­ Johnston et al. (2019);
geous in this regard because they immediately influence utilities.
μ = − α/β (1)
Johnston et al. (2019) proposed the following model:

Yi* = ß1 + ß2×2i + ß3×3i + …. + ßkXki + ui Where α is a coefficient for the constant term and ß is a coefficient for the
{ bid that the HH was asked to pay.
I = 1 if yi ≥ ti
Y∗i is observable, but Yi
I = 0 if yi < ti
3.6. Description of variables and working hypotheses
Pr. (Yi = 1) = Pr. (Yi ≥ 0) = Pr. (ui > - ß1 - ß2×2i - ß3×3i - … -ßkXki) = F (ß1
The identification of prospective explanatory factors and their con­
+ ß2×2i + ß3×3i+…+ßkXki)
nections to the dependent variable are required when the analytical
Where Yi = ith respondent’s true unobserved point for the environmental processes and their prerequisites are understood. The primary factors
resource. anticipated to affect HH WTP for the conservation and development of
UFPs are described below.
ß = a coefficient for X
ti = the offered threshold, assigned arbitrarily to the ith respondent
Y = f (Eb, INb, Ab,Sb,YRb, EMb, MSb, CHb, FZb, OWb, BVb)
I = discrete response of respondent to the WTP question (1=yes,
Y = HHs’ WTP for the conservation and development of urban forest
0= no)
parks
ui =unobservable random component distributed N (0, σ)
Eb = Education level of b individual
Xi = vector of explanatory variables /observable attributes of the
INb = Income of b individual
respondent
Ab = Age of b individual
F = the cumulative distribution function of ui
Sb = Sex of b individual
YRb = Years residency of b individual
The standard normal distribution is the probability density function
EMb = Employment status of b individual
(pdf) of the error term ui ~ N (0,σ). A likelihood ratio test was used to
MSb = Marital Status of b individual
evaluate whether the link between the respondents’ socioeconomic
CHb = Do b individual has Child or Not
factors (a proxy for their underlying utility function) and their stated
FZb = Family Size of b individual
WTP remained constant across the WTP questions.
OWb = Houses Ownership status of b individual
LR = − 2 (ln L 0 - ln L1) BVb = Initial Bid Price given to b individual

where L0 represents the maximum likelihoods estimate derived from the

4. Results and discussion
unconstrained model and L1 represents the maximum likelihood esti­
mate derived from the restricted model.
This chapter presents descriptive statistics and Econometric model
analysis of the results of this study.
3.5.2. The bivariate probit models
The bivariate normal probability density functions are the most
common bivariate distributions used by statisticians. They permit a non- 4.1. Descriptive statistics results
zero correlation, and a regular logistic distribution is not permitted
(Johnston et al., 2019). A bivariate probit model was used to determine 4.1.1. Socio-Economic characteristics
the mean WTP for the DD option in this experiment. The model looks In this section, the socioeconomic characteristics of the respondents
like this (Johnston et al., 2019). The probability function’s jth contri­ as well as their willingness and non-willingness to pay are presented and
bution is given as interpreted. Specifically, some of the respondents’ characteristics
regarding willingness to pay are presented. Tables 4.1a and b summa­
Lj ( μ/ t) = Pr (μ1 + ε1j > t1, μ2 + ε2j < t2) YN * Pr (μ1 + ε1j > t1, μ2 + ε2j > t2) rized the socioeconomic variables considered in this study. With an
YY
* Pr (μ1 + ε1j < t1, μ2 + ε2j < t2) NN* Pr (μ1 + ε1j < t1, μ2 + ε2j > t2) NY average age of 38.9 years, the respondents’ ages varied from 17 to 75
years (Table 1a). Regarding respondents’ gender (Table 1b), there were
This formulation is referred to as the bivariate discrete choice model
slightly more male respondents (50.9%) than female respondents
Where: μ = mean value for WTP
(49.5%). Nonetheless, approximately 66.8% of the survey participants
were married. The percentage of single and widowed respondents was

6
G. Zegeye et al. Trees, Forests and People 12 (2023) 100398

Table 1a the results were consistent with expectations), protest rates (the per­
Descriptive Statistics of Continuous Socioeconomic Variables (N = 289). centage of responses that are protest bids; if it surpasses 40%, it raises
Min Max Not Willing Mean Std. issues), and content validity, a great deal of work has been conducted.
Willing to pay Deviation To obtain a meaningful aggregate of benefits, four key issues involving
to pay Mean sample design and execution must be considered, according to Christian
Mean
et al. (2018), Johnston et al. (2017), and Sajise et al. (2021). These issues
HHINCOM 500 8000 1344 3409 2695.85 1.848 included population choice, sampling frame, sample non-response, and
monthly sample selection bias. Face-to-face interviews and random sampling
income (in
ETB)
from a list of HH were conducted. By eliminating "protest zeros," "un­
EDULVL (in 0 12+8 7.5 13.4 10.95 2.747 realistic high values," and other unwanted replies from the final sample,
No) effective final total response rates could be attained. Therefore,
YEARS (in 1 40 21.21 20.04 20.35 1.413 zero-protest responses are not included in the analysis (Sylvie, 2010;
year)
Johnston et al., 2017; Darra-Sila, 2019; and Sajise et al., 2021). As a
AGE (in year) 17 75 47.2 36.5 38.9 1.353
FMLSIZE (in 1 8 3.9 2.2 2.62 2.019 result, none of the aforementioned biases were anticipated in the
No) analysis.
In both the OE and DC/SD surveys, the respondents were asked
Source: Survey data (2021).
whether they would be willing to donate to a community fund for the
conservation and development of UFPs in their city before the WTP
Table 1b question. An inquiry such as this was primarily used to validate a zero
Descriptive Statistics of Discrete Socioeconomic Variables (N = 289). WTP response. It was realized that asking participants to directly indi­
cate a WTP total would cause them to feel pressured to respond with
Variables Frequency Percentage
something other than zero if this was not included in the inquiry. A total
EMPLOY Employed 174 60.2
of 320 HHs were interviewed (although the sample size was 318, the
Self-Employed 75 26.0
Unemployed 26 9 researcher added 2 more individuals to boost the result). In the sense of
Retired Persons 14 4.8 refusing to participate in the appraisal process, there was hardly any
MARSTATUS Married 193 66.8% evidence of "protest bidding." Although too small, their justification for
Single 55 19.0 such a response was questioned. The majority claimed that they were
Widowed 28 9.7
Divorced 13 4.5
too busy with other tasks throughout the survey period to have sufficient
HOWNER Own house 92 31.8 time for interviews. However, only 12 respondents (4% of the 320 re­
Relatives 17 5.9 spondents) justified this decline. in total of 308 volunteers were con­
Rented 115 39.8 tacted. Approximately 3.6 percent (11) were eliminated because of
‘Yekebele’ and other 65 22.4
incomplete responses (Table 2). Thus, 297 interviews were conducted.
HHINTEREST little concerned 2 0.7
concerned 69 23.9 11 people responded "no" to the "will/won’t" pay question. All the 11
very concerned 130 44.0 respondents were asked to provide an explanation of their responses.
very much concerned 88 30.4 Based on their justifications for saying "won’t pay," it had been planned
SEX Male 147 50.9% to divide them into two groups. The first category, which consisted
Female 142 49.1%
primarily of non-payers, belonged to free riders. They stated that
CHILDREN Have Not 61 21.1%
Have 228 78.9 although the area was valuable, others were anticipated to pay enough
HHAWARE No 77 26.6 for it, so they would not have to (almost 73.4 percent of non-payers (8 of
Yes 212 73.4 them), equivalent to 2.7 percent of the total 297 sample). The second
PARTICIPATE No 109 37.6
category of non-payers, which was based on income and prior re­
Yes 180 62.4
sponsibilities, included approximately 27% of non-payers, or 1% of the
Source: Survey data (2021). 297 samples (Table 1). Although there are no income restrictions in this
situation, the free-riding effect indicates that WTP computation is biased
19% and 9.7%, respectively (Table 1b). The remaining 4.5% were downward. More importantly, the presence of a larger group of extreme
divorced heads of HH. The average education level is 10.95 and ranges free riders may indicate the presence of more respondents who, despite
from 0 to 12+8, as shown in Table 1a, making the survey less chal­ asserting a non-zero sum, are not extreme free riders. This shows that the
lenging and, to some extent, requiring less work. According to Table 1a, valuation scenario design’s free rider incentive is extremely low, which
the respondents have an average HH size of 2.62, which is a small raises the validity of the study. Finally, a response rate of 90.3% (289 out
number for HH and goes from 1 to 8. The respondents’ average monthly of 320 respondents) was obtained (after deducting 8 free-riders, 11
income is 2695.85 birr, with a range of 500 to 8000 birr (Table 1a). As uncompleted and not clear responses, and 12 protest bids) for the two
mentioned in Table 1b, the respondents’ primary sources of income survey versions. Generally, the non-response rates, including incomplete
(60.2%) came from their work as employees of governmental bodies,
non-profit organizations, and for-profit businesses. Revenue from re­
spondents’ own sources, such as businesses, accounted for about 26% of Table 2
their total income (urban agriculture, retail business, etc.). 4.8% or so of Sample sizes and response rate.
the respondents’ sources of income come from their retirement ac­ Freq. %
counts. The remaining 9% of respondents depend on gifts, remittances, Sample Size 320 100
and their counterparts for their income instead of having their own. You Sample Response 320 100
can see in Appendix 5 the definition of the hypothesized explanatory Sample Non-Response 0 0
Free-Rider 8 2.5
variables included in the model.
Protest Bids 12 3.8
Incomplete/Do Not Know 11 3.4
4.1.2. Response rate Valid WTP Responses 289 90.3
Based on the suggestions of Christian et al. (2018), Johnston et al. Zero WTP 3 0.94
(2017), and Sajise et al. (2021), validity testing was conducted for OE, Responses Rate (WTP>0) 286 89.4

SD, and DD. To demonstrate construct validity (i.e., assessing whether Source: Survey data (2021).

7
G. Zegeye et al. Trees, Forests and People 12 (2023) 100398

and not cleared questionnaires, were below 10%, which is less likely to operational conditions" that they offer as guidance is the use of
affect the validity of the study (see Table 2). well-known environmental commodities. Given that the main objective
The response rate for the two survey versions was 90.3 percent after of this study is to understand the WTP of dwellings for hypothetically
8 free-riders, 11 illegible and confusing responses, and 12 protest bids. planned conservation and development of UFPs, the above evidence
The non-response rates were often less than 10%, which had little clearly demonstrates the success of the study in general and especially
bearing on the validity of the study (Table 2). The high response rate has an impact on the WTP of the project. As a result, it is essential to
suggests substantial interest in UFPs. This finding explicitly refutes the decide which environmental products should be included in a hypo­
assertion made by some detractors that the prevalence of protest bids thetical market.
invalidates CVM studies (Christian et al., 2018; Johnston et al., 2017; Urban forest products and services were specified and presented to
and Sajise et al., 2021). Nonetheless, the scenario presented in this study the respondents in both current and projected amounts. According to
was favorable. Additionally, it is far lower than the NOAA panel’s sug­ Loomis et al. (2000) and Awad and Holländer (2010), accurate benefit
gested level, which is highly unlikely to be lower than 20%. According to estimates using CVM require detailed descriptions of the resources to be
Christian et al. (2018), Johnston et al. (2017), and Sajise et al. (2021), valued. Some differences exist in the research areas’ preferences for
item non-responses of 20% to 30% are typical in CV studies. The pay­ value.
ment method used was community funding. The question formulation Based on their individual experiences, the respondents were asked to
and payment method were both successful, as seen by the item rate these six characteristics and the advantages of urban forests in terms
non-response rate of only 10%. There are two types of payment vehicles, of relevance and value. The mean value of the scores that respondents
mandatory and voluntary. Mandatory payment vehicle procedures such assigned to the six benefits of UFP ranged from 1 (least valued) to 6
as taxes and admission fees are frequently skipped to remove payment (most valued). With an overall mean of 4.87 from the respondents, non-
vehicle bias and lower the number of protest bids (Christian et al., 2018; consumptive UV was selected as the top value (Table 4). Urban and
Johnston et al., 2017; and Sajise et al., 2021). Therefore, a voluntary peri‑urban woodlands’ primary values have minimal financial value,
payment vehicle (community fund) was used in this study. In fact, according to research like that by Tyrvainen and Vaananen (1998).
non-response rates were less than 10%, which had little bearing on the These are referred to as non-consumptive UVs and have advantages such
validity of the study (Table 2). Hence, using a voluntary payment vehicle as good setting, pure air, tranquility, and leisure activities. However, the
resulted in a successful outcome. Choice format was the main strategy consumption of UVs (2nd by getting 4.18) is lower in urban areas than in
for eliciting residents’ WTP for urban forests in the city. To increase the rural areas. Additionally, although the NUVs (option, bequest, and ex­
reliability of the results and broaden the scope of cross-checking re­ istence) may not be as important in urban settings as they are in natural
spondents’ honesty, we attempted to employ OE and SD payment wooded areas, they still need to be considered. Additionally, TEEB
vehicles. (2010) asserts that, when humans transition from consumptive to exis­
tential values, the tangible nature of goods and services decreases. As a
4.1.3. Household perception and attitudes and related environmental result, the respondents in the study city place a high value on
problem non-consumptive uses, indicating that urban forests are highly appre­
Table 3 shows that 73.4 percent of the respondents thought that ciated for the intensive recreational use of wooded green spaces and
urban forests had no negative effects on the neighborhood. Thus, the urban forests, which is also a common feature in other urban forest
respondents had a positive opinion of urban woodlands. The distribution studies (Tyrvainen and Vaananen, 1998).
of their yes responses to the willing to pay question also differed, at 79.7
percent and 55.8 percent, respectively. About 62.4 percent of the re­
spondents indicated that they would be open to participating in 4.2. Econometric model results
community-based environmental campaigns against environmental or­
ganizations, people, and actions, as well as campaigns against stricter 4.2.1. Analysis of determinants of households willingness to pay
environmental legislation. Additionally, those who were prepared to
participate in environmental activities answered the willing to pay 4.2.1.1. Multi-collinearity and heteroscedasticity diagnosis. Before using
questions more frequently (81.7%) than their peers (Table 3). the probit and bivariate probit models, the hypothesized explanatory
During the focus group discussion and pre-test of the survey’s variables were checked for multicollinearity and heteroscedasticity. For
viability, six natural resource issues were identified as major obstacles to all explanatory factors, heteroscedasticity was examined and found to be
the city. Respondents rated these six environmental issues as high, statistically insignificant (see Appendix 1a). The degree of interdepen­
medium, or low severity. The most frequently mentioned environmental dence between the explanatory variables was validated, and multi-
problem among the respondents was deforestation. The total of 57.4% of collinearity among the explanatory factors was assessed. Conse­
HHs gave the issue a "high" rating. According to most of the respondents, quently, the multicollinearity of the data did not provide any significant
deforestation is a significant environmental issue. Urban forest products challenges (see Appendices 1b and 1c). As a result, the model was
and services are also necessary for daily tasks; therefore, they are expanded to include hypothesized continuous and dummy variables.
already familiar with them. In this regard, the "reference operating
conditions" were produced by Cummings et al. (1986), Bateman and 4.2.1.2. Results and discussion of the probit model. The results are sum­
Turner (1993), and Awad and Holländer (2010). One of the "reference marized in Table 5. The value − 2ln (LR/Lu) in the table serves as a

Table 3 Table 4
Attitudes of HHs towards Urban Forest Uses. Benefits and services of urban forest resources and their respective ranks given
by the respondents.
Frequency % Respondent’s responses
for the first bid value Values and Benefits No min max Mean Std. Deviation Rank
Willing to Not Willing
Non-Consumptive UV 289 1 6 4.87 1.523 1st
pay to pay
Indirect UV 289 1 6 4.26 1.229 2nd
UFs caused negative Yes 77 26.6 55.8% 44.2% Consumptive UV 289 1 6 4.18 1.095 3rd
effects in the city No 212 73.4 79.7 20.3 Option UV 289 1 6 2.98 1.863 4th
WTP in Environmental No 109 37.6 60.6 39.4 Bequest UV 289 1 6 2.44 1.162 5th
Activities: Yes 180 62.4 81.7 18.3 Existence UV 289 1 6 2.24 1.342 6th

Source: Survey data (2021). Source: Survey data (2021).

8
G. Zegeye et al. Trees, Forests and People 12 (2023) 100398

Table 5 negotiating and trading powers. This can also make them more con­
Probit estimate for the single bounded dichotomous choice model for the cerned about the transferability of resources to future generations.
project. Numerous valuation studies pertaining to nature conservation, such as
Variables Coefficient z-value P>|z| Marginal effect the one conducted by Chaudhry and Tewari (2006), have found a
AGE − 0.5331919 − 2.66 0.008*** − 0.0342129
comparable revenue effect. A greater WTP was found in those with
SEX − 1.110008 − 2.36 0.018** − 0.0712249 higher salaries, suggesting that higher wages are likely to contribute
EDULVL .5223988 5.17 0.000*** .0335203 more.
HHINCOM .5434179 3.29 0.001*** .0346658 At the 10% level, marital status had a significant impact on WTP for
MARSTATUS .2882212 1.77 0.076* .018494
the conservation and development of UFPs (Table 5). This finding sug­
CHILDREN .54.34179 1.02 0.309 .0476758
FMLYSIZE − 0.2156607 − 1.72 0.075* − 0.0138381 gests that having marital status (=4) increases the likelihood of saying
YEARS .010666 0.09 0.929 .0006844 yes more than having any other type of marital status. This might be
HOWNER − 0.4614443 − 2.72 0.006*** − 0.0296091 because married couples are more established than single (=3) in­
EMPLOY − 0.0977192 0.59 0.559 − 0.0296091 dividuals, who may feel a stronger connection to the city and parks.
HHAWARE − 0.1135757 − 0.33 0.740 − 0.0069352
HHINTEREST .1215566 0.56 0.577 .0077998
Additionally, widowed (=2) and divorced (=1) respondents experienced
PARTICIPATION − 0.127533 − 0.47 0.641 − 0.0081833 significant social unrest, which adversely affected their WTP for the
BIDAMOUNT − 0.0364305 − 6.53 0.000*** − 0.0023376 projects. For this reason, marital status was found to significantly affect
_cons 4.307668 2.31 0.021** the WTP for the financial benefits of urban forests.
Number of observations = 289
Family size significantly and negatively affected WTP for the pres­
Log likelihood function = − 41.313858
− 2ln (LR/LU) = 65.20 ervation and growth of UFPs (significant at the 10 percent level) (see
Prob > chi2 = 0.0000 Table 5). However, for each additional HH family member, the likeli­
Pseudo R2 = 0.7534 hood of HH WTP decreased by 1.4 percent. This can be because addi­
***, ** and* indicate statistically significant at 1%, 5%, and 10% probability tional expenses become necessary as the family expands in size. Having
levels, respectively. more family members means that HH expenses increase. Residential
Source: Survey data (2021). property ownership had a considerable impact on the WTP of the con­
servation and development UFPs (at 1 percent, first responders are more
rigorous test of significance. According to the null hypothesis, the con­ likely to remain in the community if they own a property there, which
stant parameters as well as those of all explanatory variables are all zero. increases the WTP for the project because those who stay longer more
The tabular chi-square for the probit model with 15 degrees of freedom beneficial, particularly for the optional value of urban forests). The
and 95% confidence level was 65.20, thus rejecting the hypothesis. initial bid price is negative and significant at the 1 percent level. For
Additionally, there was a connection between WTP and socioeconomic each extra bid value, the likelihood of voting "yes" for WTP decreases by
circumstances (Table 5). 0.23 percent. Thus, it is obvious that as the bid price increases, the
Respondents’ age had a negative relationship with the dependent probability of a "yes" response decreases.
variable (significant at the 1 percent level). As a result, the respondents’
WTP sharply declined with age. Therefore, younger respondents indi­ 4.2.2. Willingness to pay for the open-end, single and double dichotomous
cated a significant preference for urban amenities and leisure services choice
and were more likely to donate larger amounts to the initiative. This Because they were believed to be the most immediately acceptable
outcome is consistence with that reported by Chaudhry and Tewari and meaningful for respondents, category values such as £10, £20, and
(2006). Younger people were more willing to contribute than older £30 were used in this study. Technically correct numbers such as £3, £7,
adults. Respondents’ gender (male = 1; female = 2), which had a and £12 are likely to cause uncertainty and ambiguity among re­
negative relationship with the dependent variable and was significant at spondents, according to Bateman et al. (1995).
the 5% level, is another pertinent sociodemographic characteristic. This
indicates that, compared to male respondents, female respondents were 4.2.2.1. Average willingness to pay of open-ended question. After the DD
able to contribute less WTP to the project. One explanation is that question, a separate OE question is posed. All respondents to the DD
women typically have fewer financial resources than men. It is antici­ questions provided a response close to the amount they had promised to
pated that male respondents will be more financially independent, pay. This might be because the respondents revealed their "actual" WTP
which could significantly affect their WTP. This finding is consistent because of the survey’s effectiveness in achieving its stated purpose.
with those of Adekunle (2008) and Carlsson et al. (2004), who found According to Hanley and Barbier (2009), calculating the mean or me­
that men have higher WTPs than women. dian WTP for OE reactions is a straightforward process. The mean WTP
The higher the respondents’ formal education level, the more ready is the appropriate number to use in cost-benefit analysis, even though
they were to contribute to the project. By contrast, respondents’ edu­ the authors frequently concentrate on the median WTP because it is less
cation level had a significant and favorable effect on HHs’ WTP for the affected by extreme values and is helpful from the perspective of polit­
project (significant at the 1% level) (see table 5). This might be a result ical consensus. Additionally, WTP confidence intervals must be pro­
of education raising the social and environmental consciousness. vided. Table 5a demonstrates that there are no appreciable differences in
Consequently, resident’s capacity to acquire, use, and analyze knowl­ the mean WTP between the Philadelphia and Piassa kebeles. The
edge increases with their level of education. Additionally, Adekunle average WTP for Harare Kebele is lower than the other two. However,
(2008), Chaudhry and Tewari (2006), and Carlsson et al. (2004) found this difference was not consequential (Table 6a).
that this variable has a favorable and significant impact on WTP for The average WTP for each initial bid was determined using survey
urban forest resources. Therefore, HH income significantly and favor­ data. The mean WTP increased with the original bid, as shown in
ably influences WTP for UFP conservation and development (significant
at the 1 percent level). The likelihood of voting for WTP increased by 3.4
Table 6a
percent as HHs’ incomes moved from one income bracket to the next.
Average Willingness to Pay in terms of cash for the three kebeles (ETB).
There are two possible explanations for this observation. First, because
luxury goods and services have greater social and cultural value than Mean WTP Kebeles
Harer Philadelphia Piassa Average
economic value, people with higher income are more inclined to enjoy
126.67 150.74 146.24 141
them. Second, having more money gives respondents greater
Source: Survey data (2021).

9
G. Zegeye et al. Trees, Forests and People 12 (2023) 100398

Table 5b. In this instance, a starting-point bias exists. However, the first Table 6c
offer of 80 had a mean value of 141.9, which is the same as the overall Truncation effects for Open End Willingness to Pay study a.
mean value for the OE scenario (141 ETB). This shows that the cumu­ No. of upper tail 0 1 3 14 29 43 58
lative probability distribution of the bid amount is assigned symmetri­ truncated
cally. According to Christian et al. (2018), Johnston et al. (2017), and % of upper tail 0% .35% 1% 5% 10% 15% 20%
truncated
Sajise et al. (2021), two important types of starting point bias are the
anchoring effect and yes-or-no-saying prejudice. Starting point bias oc­ N 289 288 286 275 260 246 231
curs when the interviewer’s opening bid affects the respondent’s WTP Mean 140.9 139.7 137.3 126.8 115.1 107.2 99.9
Std. Error of 6.09 5.98 5.79 5.08 4.32 4.0 3.8
level. This study also demonstrated a yes-saying bias in this area. It is Mean
difficult to draw a connection between this "yes-saying" and "warm Median 100 100 100 100 100 100 100
glow"; however, the statistical effects do not prove a connection. For Std. Deviation 103.5 101.5 97.9 84.2 69.7 63.0 57.4
example, a skewed distribution can reflect the socioeconomic and Maximum 500 500 400 400 300 250 200
preferential characteristics of a sample. Moreover, according to Chien Source: Survey data (2021).
et al. (2005), even DD cannot be free from doom; it is in fact among the a
All rows, except the upper three, are measured in ETB.
methods that dominate CV surveys of non-market goods and services
that address strategic bias. It is regarded as providing an easier response 4.2.2.2. Willingness to pay for different benefits of the project. The re­
than open-ended questions, and its follow-up questions, which produce spondents’ WTP allocation for the various goods and services they
double-bound questions, improve the efficiency of questionnaires. received from the proposed project in the municipality was ascertained.
Bishop (2018) argued that because of the warm glowing embedding The respondents’ mean WTP for the advantages and services they
effect, the CV method could be invalid; therefore, it needs to be treated received from the initiative varied (Table 7).
in the CV study. Additionally, the "warm glow" effect may be limited by The indirect and consumptive UVs received 29.1 ETB and 28.4 ETB,
the referendum structure. Because the survey design particularly respectively, whereas the non-consumptive UVs obtained the highest
addressed this problem, the effect of "yes-saying" was negligible. The average WTP value of 33.5 ETB for the project (Table 7). This result
impact of "no-saying" is present, but it is small in comparison to "yes-­ shows that the NUVs (bequest value and existence value), which have a
saying’s" impact. However, when respondents adhere to the in­ value of 21.4 percent, are less valuable than the UVs (non-consumptive,
terviewer’s initial bid, an anchoring effect is present in the early bids. In indirect, consumptive, and option UVs) of the project (78.6%). This
addition to, the researchers discovered that 289 observations were could be because as respondents moved from UVs to NUVs, the tangi­
classified into five groups; with approximately the same number of in­ bility of the services and products offered by forest parks decreased.
dividuals’ in each group (Table 6b). Furthermore, it is anticipated to have a high value, particularly for non-
Further studies on the economic valuation of urban forest parks consumptive UV, because the majority of respondents were young.
should be conducted. In addition, in the discussion on `yes-saying’ and
`warm glow’ influences WTP in the results, relevant studies should be 4.2.2.3. Mean willingness to pay for single and double bound dichotomous
referred to in this section, and this discussion should be better explained. question. The bivariate probit model was used to model the responses
Table 6c shows the details of the univariate statistics of the OE WTP collectively, adhering to the basic methodology provided by Cameron
for a variety of upper tail truncation points. As the table portrays, the and Quiggin (1994), because obtaining the follow-up question is
omission of just the single highest bid (0.35% of the OE sample) caused contingent on the respondent’s response to the first question.
the OE mean WTP to fall to 139.68 ETB (a reduction of about 1%). A total of 578 suitable responses were gathered from the first and
Similarly, truncating the top 1% of the bids caused a reduction in the second bid summaries (see Appendix 2a). Rho (ρ) is positively and
mean of nearly 3%. However, when the percentage that was truncated significantly (at the 99 percent confidence level) different from zero in
increased, the percent reduction from the mean WTP of OE increased the bivariate probit estimates for the project (Tables 8a and 8b). This
reasonably (i.e., 5% truncating led to a reduction of 10% of the mean; indicates that there is a positive correlation between the two responses
10% truncating led to a reduction of 18.4% of the mean; etc.). and that it differs from zero. For example, the project LR test (χ2 =
Table 6c shows that, alongside evidence of starting point bias, other 68.38) indicated a substantial correlation between the two disturbances
respondents in the OE sample appeared to exhibit strategic overbidding. (table 7b). The estimated correlation of 0.3826 was statistically signif­
Such statistical effects are not conclusive proof of strategic overbidding, icant and was far from zero (p < 0.0079). As a result, it was possible to
such as starting point bias; a skewed distribution may simply reflect the jointly determine the first (table 8a) and second (table 8b) responses.
socioeconomic and preference characteristics of the sample. Addition­ The random component of WTP for the first question is not perfectly
ally, upon inspection, it was discovered that, given their capacity to pay, connected to the random component of the follow-up question, as the
the amounts stated by those at the upper tail of the bid distribution correlation coefficient is less than one.
appeared feasible. However, there is little evidence for a degree of Using these factors and the calculation, the SD probit estimate’s
strategic overstatement by a small number of respondents in the OE mean WTP for the project was 170.6 ETB (1). The project’s DD bivariate
experiment. Generally speaking, as Tables 6b and 6c show, it can be probit estimate mean WTP for the initial bid (bid-one) and follow-up bid
observed that the presence of starting point bias and strategic over­ amounts ranged from 172.9 ETB to 221.7 ETB (bid-two), respectively. In
bidding was conspicuously low.

Table 6b Table 7
Willingness to Pay for the different initial bids. Mean Willingness to Pay for the different benefits and services of urban forest.
Initial bids WTP Type of benefits and services Mean WTP (ETB) %
Min Max Mean Frequency
Consumptive UV 28.40 20.14
20 10 250 101.5 57 Non-Consumptive UV 33.50 23.80
40 20 400 117.72 57 Indirect UV 29.10 20.64
80 20 400 141.90 58 Option UV 19.80 14.00
120 20 400 151.03 58 Bequest UV 15.70 11.13
200 20 500 189.52 59 Existence UV 14.50 10.30
Total 289 Total 141.0 100.0

Source: Survey data (2021). Source: Survey data (2021).

10
G. Zegeye et al. Trees, Forests and People 12 (2023) 100398

Table 8a Table 9
Parameter estimates of probit for the project. Aggregate benefit measures by site.
Variable Estimate Standard Error Sites Population Average Number HH’s Total HHs’
size HH Size of HHs Mean WTP (ETB)
Bid-one − 0.0094229 0.0013536
WTP
Constant 1.607284 0.172218
(ETB)
Log- likelihood − 140.98607
− 2ln (LR/Lu) 53.08 Hawassa 201,020a 3.76 53,463 221.7 11,852,747.1
City ETB ETB
393,627b 3.76 104,688 221.7 23,209,329.6
ETB ETB
Table 8b
Parameter estimates of bivariate probit for the project. Source: Computed from the survey data (2021).
a
population size for urban areas in the city administration of in Hawassa
Variable Estimate Standard Error Town 2007.
b
Bid-one − 0.0090572 0.0012352 population size for urban areas in the city administration in Hawassa Town
Constant 1.56598 0.1522848 2021.
Bid-two − 0.004179 0.0008412
Constant 0 0.9264863 0.1311518
ρ*** 0.3826682 0.1261699 benefit to the communities of 23,209,329.6 ETB. If we apply the SD WTP
Log- likelihood − 318.35779 estimate and the OE WTP estimate, these benefits would become
− 2ln (LR/Lu) 68.38 17,859,772.8 ETB and 14,761,008 ETB, respectively. When compared to
Source: Survey data (2021). the mean WTP quantity of the OE choice format, the mean WTP of the
dichotomous choice forms (both SD = 170.6 and DD = 221.7) was
shown to be significantly larger (141 ETB). The findings of Johnston
general, these data are significantly lower than the typical WTP amount
et al. (2017) and Sajise et al. (2021) likewise support this result. Since
from the OE question, which is 141 ETB (both SD and DD). In addition,
they are concerned, studies conducted in industrialized nations have
the mean WTP of the SD probit estimate is lower than that of the DD
shown that the OE format has resulted in underestimated estimates of
estimate.
the consumer surplus, in contrast to the dichotomous choice format.
When compared to the mean WTP quantity of the OE choice format,
Additional research has confirmed this finding (e.g., Johnston et al.,
the mean WTP of the dichotomous choice forms (both SD = 170.6 and
2017; Sajise et al., 2021). OE WTP questions provide lower estimates
DD = 221.7) was significantly larger (141 ETB). The dichotomous
than the dichotomous choice format, according to research by Johnston
choice format, one of the possible variations of the closed-ended format
et al. (2017) and Sajise et al. (2021).
questionnaire used in CVM studies, in which respondents were given
To obtain a legitimate aggregate of benefits, Johnston et al. (2017)
only two alternative answers to choose from, produced more believable
and Sajise et al. (2021) stated that four crucial considerations regarding
results. This result also supports the work of Johnston et al. (2017) and
sample design and execution must be made: population choice, sampling
Sajise et al. (2021), according to which research in developed countries
frame, sample non-response, and sample selection bias. This study
has shown that the OE format has produced underestimated consumer
employed a face-to-face interview technique with a random sample
surplus values. The OE WTP questions provided lower estimates than the
strategy using a list of HH. By eliminating "protest zeros," "unrealistic
dichotomous options format.
high values," and other unwanted replies from the final sample, effective
final total response rates could be attained. To avoid using zero protest
4.2.2.4. Aggregate WTP and potential revenue from the proposed project
replies in this study, they were not used (Sylvie, 2010; Johnston et al.,
4.2.2.4.1. Mean and aggregate WTP. The researcher must choose
2017; Darra-Sila, 2019; and Sajise et al., 2021). Thus, none of the
which estimates to use to generate the WTP measure if the bivariate
aforementioned biases were anticipated throughout the analysis.
probit model is estimated on a DD choice question with a follow-up and
Hanley and Barbier (2009) claimed that these three factors deter­
the parameter indicates that either the mean, variance, or both differ
mine aggregation decisions. The first step was the appropriate popula­
between the initial bid-price and follow-up (Johnston et al., 2019). The
tion selection. The second problem is changing the sample mean to the
total number of YY (11.42%) and NN (49.53%) responses accounted for
population mean. The total number of HHs in the population, N, can be
60.55 percent of the total responses for the project (see Appendix 6);
multiplied by the sample mean. The third problem is the selection of the
therefore, we looked at the data to select a suitable WTP from the two
period over which benefits should be averaged. This study attempts to
bivariate estimates. This implies that the second bid amount is closer to
address these three crucial difficulties by aggregating the mean WTP for
the unobserved real value of the person. For example, let the first
residential units for the advantages of urban forests in Hawassa.
random bid for an individual be 80 ETB, and let the respondent accept
Consequently, we attempt to determine the mean WTP for the project
the first bid. This indicates that the second bid amount is closer to the
and the total advantages of utilizing the second estimate of the DD
unobserved real value of the person. For instance, allow the response to
bivariate model (WTP is worth 23,209,329.6 ETB). If we apply the SD
accept the first random bid of 80 ETB for a particular person. Re­
WTP estimate and the OE WTP estimate, these benefits would become
spondents may have accepted the second proposal at 160 ETB. There­
17,859,772.8 ETB and 14,761,008 ETB, respectively. Thus, the real WTP
fore, the respondents’ actual WTP was greater than or equivalent to 160
may vary between 23,209,329.6 ETB and 14,761,008 ETB. In Hawassa,
ETB. Consequently, the second bid is a more accurate estimate than the
the average cost of establishing and maintaining a forest park for 10
first bid is. The same applies to NN response. Both the first and second
years was 1696,000 ETB (see Appendix 3). Consequently, considering
bid amounts have an equal chance of being closer approximations of the
the lower (OE) and higher (DD) WTP estimations, the aggregate project
genuine value, even for the remaining 39.45% of NY and YN responses.
value are 8.7–13.7 times higher than the management expenses (OE and
Accordingly, on average, the second bid number was 80.3 percent better
DD). Consequently, people’s well-being may improve throughout the
approximation of the genuine WTP value than the first bid number.
duration of the project. To this end, the positive economic value of urban
The total benefits to inhabitants are summarized in Table 9 using the
forests in Hawassa was achieved through this study. Studies carried out
second estimate of the DD bivariate model to obtain the mean WTP for
in different areas have shown that forests in urban areas bring benefits in
the project. The city’s yearly rate of urban population increase is 4.8
terms of money. This is in accordance with the results of the present
(CSA, 2007). Therefore, the population size of the city administration of
study. For instance, the economic benefits of urban forests include:
Hawassa’s urban areas was calculated in 2021 using a growth rate of 4.8.
providing ecosystem services evaluated at $330 million per year for
Table 9 shows that the proposed project will provide a total financial

11
G. Zegeye et al. Trees, Forests and People 12 (2023) 100398

Halifax, Montreal, Vancovuer, and Toronto (Alexander and DePratto, The first and most significant contribution the quality of life of local
2014); increasing the value of urban forests by between $1.88 and residents to the conservation of urban forests is raising awareness among
$12.70 for every dollar spent on maintaining them, depending on the the general public, academics, decision-makers, and both government
city (Alexander and McDonald, 2014). In fact, the total WTP value and non-government groups engaged in the development and mainte­
should not be used alone to measure the total financial benefit (or total nance of urban forests. The fact that this research successfully charac­
economic value) of communities. Other ecosystem services provided by terizes and defines urban forests and their values in Hawassa is another
UFP also contribute to the TEV. For instance, urban forests increase significant contribution it makes to the field of economic valuation of
property values (Escobedo et al., 2015), and help create attractive urban forests. A community fund was also successful in increasing the
business districts and improve visitors’ perceptions of them (Wolf, locals’ willingness to pay for the contribution of the urban forest in the
2003). study area, in addition to the research study itself. Lastly, variables
4.2.2.4.2. Establishment cost of urban forest park. The complete influencing HHs’ willingness to pay (WTP) for the preservation and
establishment cost of the park was considered because only the infor­ development of urban forest parks were understood, and as a result, they
mation required for the conservation and development of urban forest would be taken into account throughout the planning and execution of
parks was accessible. The local markets, Hawassa City Municipality, and urban forest projects in urban compounds. For that purpose, our study
Wondo Genet College of Forestry and Natural Resources provided in­ has shown the beneficial economic value of urban forests in Hawassa
formation on setting expenses from the nursery to planting (Appen­ City.
dix 3). It would cost between 50 and 100 ETB to start a tree plantation.
Additionally, it was found that an average of 2500 trees could survive on 5. Conclusion and recommendation
one hectare. If the maximum price of 10 ETB per tree is considered,
starting a tree plantation will cost 250,000 ETB. The overall monthly The findings indicated that approximately 87% of the respondents
costs for the manager/forester (diploma holder), support staff members, were willing to pay in terms of money, and approximately 13% of them
operating expenses, security, and other expenses were 1446,000 ETB. were willing to provide unpaid labor for the creation and conservation of
Consequently, 1696,000 ETB was the predicted total cost (see urban forest parks. Most respondents were willing to pay for the pro­
Appendix 3). posed project, while only approximately 10% submitted protest bids.
4.2.2.4.3. Estimating potential revenue from the project. Using the The high response rate in the study area suggested that urban forest
second estimate of the DD bivariate model, it was calculated that resi­ parks are highly valued. This result vehemently refutes the claim made
dents may experience an increase in welfare (23,209,329.6 - 1696,000) by certain observers that the preponderance of protest bids makes CVM
21,513,329.6 ETB. This welfare increase would result from the cost of studies invalid. However, the situation described in this study was
establishing an urban forest park being weighed against its value. Ac­ positive. According to the results of the descriptive statistics, re­
cording to the lowest benefit estimate, the development of urban forest spondents had a favorable attitude toward the preservation and growth
parks would benefit society by at least (14,761,008–1696,000), or of the environment in general and urban forests in particular. The
13,065,008 ETB. This estimate is based on the WTP from the OE outcome also shows how eager residents are to engage in environmental
question. activities and how worried they are about potential losses and the
Cities can benefit from forest protection because deforestation has a endangerment of these natural resources. Thus, this has a favorable
significant impact. Consequently, a collection of influential cities pro­ impact on WTP. These findings indicate that urban forest parks are
motes forest preservation and restoration worldwide (Anderson et al., highly valued in the research area. Deforestation was the most important
2021). According to Anderson et al. (2021), cities promote forest pro­ environmental issue experienced by residents in the study areas across
tection in four creative ways. First, developing urban forests in urban all sample kebeles. The results of the descriptive statistics show that non-
settings, such as Sierra Leone launched the Freetown Tree Town effort in consumptive UV is the highest advantage that society considers to be
Freetown with the goal of planting and growing a million trees by 2022, gained from UFPs, followed by indirect and consumptive UVs, among
focusing on low-income regions, riverbanks, and areas at risk of land­ the benefits and services derived from urban forest resources. Accord­
slides. In collaboration with neighborhood community-based organiza­ ingly, the respondents’ high regard for non-consumptive uses in the
tions, Kochi (one of India’s fastest-growing cities) cultivates trees in study city suggests that there is a high appreciation for intense recrea­
metropolitan coastal areas to help shield the metropolis from high heat tional use, a beautiful landscape, clean air, and a place where the resi­
and flooding. The second is to safeguard upstream forests for water se­ dents can find peace and quiet from wooded green spaces and urban
curity. Many cities have come to the realization that they need to forests, which is also a general feature in other urban forest studies.
maintain their most priceless natural resources, especially drinking The probit model was used to assess the probability that HHs would
water, outside their city limits. 33 of the 105 largest cities in the world be willing to pay for the conservation and development of UFPs. The
directly depend on forests for their water sources; including their ca­ projects were significantly influenced by the socioeconomic and de­
pacity to control and clean water supplies (a forested watershed pro­ mographic characteristics of the respondents, including age, sex, marital
tected through a Water Fund with Cali, Colombia, is an example of such status, family size, income, ownership status, and educational attain­
an experience). The third is Taking Political Action for Forests Near and ment across sites. Additionally, one of the significant aspects that
Far. Knowing that forests located miles outside their borders have an affected WTP for this project was the bid cost. The mean WTP of the OE
impact on the quality of life of local residents, several forward-thinking scenario was calculated using descriptive statistics and was 141 ETB.
city officials have voiced their concern for regional or global defores­ The mean WTP of the SD scenario, which is 170.6 ETB, was calculated
tation. These leaders have begun to use their political, cultural, and using a probit estimate, on the other hand. The mean WTP of the project
economic influence as their most potent weapons (a patch of forest for the DD scenario was calculated using the bivariate probit model, and
cleared for farming in the Brazilian Amazon Rainforest). Using buying the range for the initial bid (bid one) and the follow-up bid amount was
power to stop deforestation is the final option. Agricultural products 172.9 ETB to 221.7 ETB (bid two). 221.7 ETB was chosen as the mean
such as wood pulp, palm oil, cattle, and soy are responsible for almost WTP for the DD bivariate probit estimate because the second bid amount
80% of deforestation. Cities, that consume a large portion of these was an 80.3 percent better estimate of the genuine WTP value than the
goods, have the capacity to influence production by switching to less first bid amount. Around 23,209,329.6 ETB in total benefits were pro­
destructive methods, discouraging deforestation-causing industries, and jected for the project (based on the DD mean WTP estimate). 14,761,008
promoting goods that support sustainable forest management. For ETB was the anticipated minimum aggregate benefit (based on the OE
instance, a group of influential people in New York City is investigating mean WTP estimate). Thus, 1696,000 ETB were spent on building the
how the city might sustain trees through infrastructure design. conservation and development of UFP. Comparing the expense of

12
G. Zegeye et al. Trees, Forests and People 12 (2023) 100398

establishing UFP with its benefits, the total net benefit predicted from forests’ roles in urban energy demand, and others in order to understand
the project was 21,513,329.6 ETB using the DD mean WTP estimate. the overall decision regarding the economic valuation of urban forests
Using the lowest benefit estimate, or WTP estimate, from the OE ques­ for further study.
tion, residents may experience a welfare boost of approximately
13,065,008 ETB. As a result, Hawassa’s efforts to preserve and promote Funding
UFPs have been successful and have improved the welfare of the com­
munity there. Authors not received fund for this paper
These findings could be used to make the following broad policy
recommendations: First, socioeconomic characteristics (such as age, sex, Ethical approval
marital status, family size, income, ownership status, and education
level of the respondents) must be considered while creating an urban Not Applicable
forest project or policy to ensure its successful implementation. In
addition, legislators and urban planners should support educational and Consent to participate
awareness-raising initiatives to improve WTP. Second, it is possible that
appropriate urban forest management techniques and payment methods Not Applicable
that consider urban residents’ financial capacity will improve their
involvement in the project. In addition, CVM might be a useful method Consent to publish
for determining people’s WTP for maintaining or enhancing the pro­
tective and esthetic qualities of urban forests. In addition, CVM might be Not Applicable
a useful method for mobilizing and significantly increasing financial
resources for the conservation and development of urban forest parks. Authors’ contributions
To this end, according to Glass and Newig (2019), UN SDG 15 is
accompanied by 10 targets, including one to mobilize and significantly The authors collected the data and performed preliminary analysis
increase financial resources from all sources to conserve and sustainably and interpretation. All the authors have read and approved the final
use biodiversity and ecosystems. Other targets include integrating manuscript.
ecosystem and biodiversity values into national and local planning,
development processes, and strategies and accounts for reducing Availability of data and materials
poverty. Third, the findings of this study can be applied to city planning
to assess the environmental effects of various land use options. The loss If this is demanding, the datasets that are used and analyzed anon­
of urban forest park values should be considered in the analysis when ymously during the current study are available from the corresponding
choosing the optimum land use alternative from the perspective of so­ author upon reasonable request.
ciety because the major question in urban settings is more frequently
where to construct than whether to build at all. Finally, additional Consent to publish
research incorporating other urban forest resources, such as rows of
street trees found in compounds, dispersedly distributed trees, forested Not applicable
areas with cycling or walking routes, and narrow belts of forest vege­
tation between homes and roads, would be more understandable and Declaration of Competing Interest
beneficial.
Further research need to address issues such as environmental en­ The authors declare that they have no competing interests.
gineering problems, issues related to erosion control, noise and air
pollution abatement, wastewater management and protection, envi­ Data availability
ronmental health, the role of urban forests in job creation, environ­
mental quality in real estate investment and recreational activity, urban Data will be made available on request.

Appendix

Appendix 1a: Breusch-Pagan/Cook-Weisberg Test for Heteroscedasticity Ho: Constant Variance

Variable chi2 df p

Sex of the respondent (1 for male, 2 otherwise) 20.82 1 0.0000

Whether they have children (1 for have children, 0 otherwise) 1.32 1 0.2508
Family size 1.67 1 0.1966
Length of stay in the city 9.26 1 0.0023
Ownership (house) status of respondents 4.47 1 0.0345
Condition of employment 10.21 1 0.0014
Prior awareness of respondents (1 for yes, 2 otherwise) 18.44 1 0.0000
Respondent’s worry about the loss of the resources 13.64 1 0.0002
Attitudes towards environment (1 for yes, 2 otherwise) 11.05 1 0.0009
Bid value 31.88 1 0.0000
Education level of the respondent 50.46 1 0.0000
Total monthly income of the respondent 44.61 1 0.0000
Marital status of the respondents 1.57 1 0.2096
Age of the respondent 0.20 1 0.6572
simultaneous 104.31 14 0.0000

The test statistic is ½ ESS ~ x2(p - 1) ~ x2(p) under H0 asymptotically. We reject if it is too large.

13
G. Zegeye et al. Trees, Forests and People 12 (2023) 100398

Source: Computed from the survey data (2021)

Appendix 1b: VIF of the Continuous Explanatory Variables used in the study

Variables Collinearity Statistics

Tolerance VIF

HOWNER .617 1.620

MARSTATUS .807 1.240
YEARS .737 1.357
HHINTEREST .802 1.247
EDULVL .337 2.964
HHINCOM .430 2.325
BIDAMOUNT .916 1.091
AGE .325 3.079
EMPLOY .728 1.374
FMLYSIZE .320 3.123

Source: Survey data (2021)

Appendix 1c Contingency Coefficients for Dummy Variables

A B C D

HHAWARE (A) 1 0.237 0.231 − 0.005

SEX (B) 1 0.094 − 0.118
PARTICIPATION (C) 1 − 0.052
CHILDREN (D) 1

Source: Survey data (2021)

Appendix 2a: Responses for double bounded dichotomous questions

Acronym Description Mean Stan. dev.

t1 Exogenous threshold for the first question 92.80 64.32

t2 Endogenous threshold for second question 130.70 108.42
I1 Discrete response of the first question (1=yes, 0=no) 0.7336 0.4430
I2 Discrete response for second question (1=yes, 0=no) 0.6436 0.4800

Source: Survey data (2021)

Appendix 2b: Joint Frequencies of Discrete Responses

I1=1 and I2=1 0.4913

I1=1 and I2=0 0.2422
I1=0 and I2=0 0.1142
I1=0 and I2=1 0.1522

Source: Survey data (2021)

Appendix 3: Establishment Cost of Urban Forest Park

The costs of establishing a plantation (10 ha) include the gathering of seeds, nursery activities, and materials, site preparation, pitting and
repeating, planting, and replanting, as well as the transporting of seedlings, weeding, and culture. The undertaking will take ten years. Costs for
management and guarding came from the same sources. Since the government announced a land lease at a rate of 600 ETB per ha per year for forestry
investment, the cost included the land rent at that rate. Unskilled work had a wage ratio of ETB 50 per day. Additionally, the national minimum wage
for unskilled laborers is depicted in this image. Seedlings can be purchased for between 3 and 10 birr.

Cost item Unit Cost/month Propose area (ha) Cost/tree Tree/ha Total cost/year

Plantation establishment cost (Nursery planting) No 10 10 ETB/tree 2500 250,000

Opportunity cost of the land (Land rent) Ha 600/ha/year * 10 year 10 6000
Guarding No 2 * 1500 * 12 * 10 year 360,000
Manager/ Forester (Diploma Holder) No 4000 * 12 * 10 year 480,000
Assistance worker No 2 * 1500 * 12 * 10 year 360,000
Running cost 10,000×12 120,000
Other expenses 10,000×12 120,000
Total 1471,000

14
G. Zegeye et al. Trees, Forests and People 12 (2023) 100398

Appendix 4: Total Economic Values (TEVs) Attributed to Environmental Assets

Appendix 5: Definition of Hypothesized Explanatory Variables Included in the Model

Variable Description of Variable Type of Unit of Measurement Expected

Variable Effect

HHINCOM Average monthly income of a HH. Continuous Ethiopian Birr +

HOWNER This is about house ownership status of the HH. 1= own house; 2= government house (‘yekebele bet’); 3= relative –
house and 4= living in rent house.
EMPLOY It indicates the employment status of a respondent. Categorical 1= professional 2= businessmen 3= retired persons, 4= otherwise. –
AGE Age of a respondent Continuous Years –
SEX Sex of the respondent. Dummy 1= male 2= female –
CHILDREN It is whether the HH having children or not. Dummy 1 = HH has children and 0 otherwise +
FMLSIZE Size of the family in a HH Continuous Number of persons in the HH –
EDULVL It is about literacy level of the respondent. Continuous Measured in number and put in to ranges +
MARSTATUS Marital status of the respondent Categorical 1= divorced/separated, 2= widowed, 3= single & 4= married +
YEARS Length of years the HH head staying in the city. Continuous Years +
HHAWARE It is about the prior awareness of the respondent about Dummy 1 = HH has awareness 2 = otherwise –
the urban forestry,
HHINTEREST It is about HH’s Concern on the possibility of losses of Categorical 5= very much concerned, 4= very concerned, 3= concerned 2= little +
Urban Forest Service’s. concerned, 1= otherwise.
PARTICIPATE HH willingness to participate in environmental Dummy 1 = participate 2 = otherwise –
activities
BIDAMOUNT This variable is bid amount assigned to the respondent categorical Ethiopian Birr. (20, 40, 80, 120 & 200ETB) –
exogenously.

Appendex 6. Distribution of “Yes” and “No” Answers to 1st and 2nd Bids (578 Usable Answers)

15
G. Zegeye et al. Trees, Forests and People 12 (2023) 100398

Total response for the 1st question

First bid Second bid No Yes
response for the 2nd Total No response for the 1st question (N) response for the 2nd Total Yes response for the 1st question (Y)
question question
No (NN) Yes (NY) No (YN) Yes (YY)

20 10 0 1 1
40 6 50 56
40 20 0 4 4
80 16 37 53
80 40 5 14 19
160 14 25 39
120 60 12 9 21
240 19 18 37
200 100 16 16 32
400 15 12 27
Total Frequency NN = 33 YY = 44 N = 77 YN = 70 YY = 142 Y = 212
% 11.42 15.22 26.64 24.22 49.13 73.36

Source: Survey data (2012)

References Cummings, R.G., Brookshire, D.S., Schulze, W.D. (Eds.), 1986. Valuing Environmental
goods: a State of the Arts Assessment of the Contingent Valuation Method. Roweman
and Allanheld, Totowa, NJ.
Abebe, S.T., Dagnew, A.B., Zeleke, V.G., Eshetu, G.Z., Cirella, G.T., 2019. Willingness to
Darra-Sila, N., 2019. Willingness to Pay for Improved Conservation of Water Catchment
Pay for Watershed Management. Resources.
in Hai District. Mzumbe University, TANZANIA.
Adekunle, M.F., Momoh, S., Agbaje, B.M., 2008. Valuing Urban Forests: the Application
Davies, H.J., Wu, H., Schaafsma, M., 2023. Willingness-to-pay for urban ecosystem
of Contingent Valuation Methods. Ethiopian J. Environ. Stud. Manag. 1 (2).
services provision under objective and subjective uncertainty. Resource and Energy
Alexander, C. & DePratto, B. (2014). The Value of Urban Forests in Cities across Canada.
Economics. Elsevier.
Special Report – TD Economics.
Dobrosavljević, J., Marković, C., Marjanović, M., Milanović, S., 2020. Pedunculate Oak
Alexander, C. & McDonald, C. (2014). Urban Forests: the Value of Trees in the City of
Leaf Miners’ Community: urban vs. Rural Habitat. For. Ecol. Manage.
Toronto. Special Report: TD Economics.
Dumenu, W.K., 2013. What are we missing? Economic value of an urban forest in Ghana.
Anderson, J., Pool, J.R., Haniy, S.U., Evers, H., Narayanan, P., 2021. Cities Are Surprising
Ecosyst. Serv. 5, 137–142. September 2013Elsevier.
Leaders in Forest Conservation. World Resource Institution.
Escobedo, F.J., Adams, D.C., Timilsina, N., 2015. Urban forest structure effects on
Aronson, M.F.J., Lepczyk, C.A., Evans, K.L., Goddard, M.A., Lerman, S.B., MacIvor, J.S.,
property value. Ecosyst. Serv. 12, 209–217.
Nilon, C.H., Vargo, T., 2017. Biodiversity in the city: key challenges for urban green
Gashu, K., Gebregziabher, T., 2019. Public assessment of green infrastructure benefits
space management. Front. Ecol. Environ.
and associated influencing factors in two Ethiopian cities: bahir Dar and Hawassa.
Assefa, W.W., Eneyew, B.G., Wondie, A., 2021. The impacts of land-use and land-cover
BMC Ecol. Open Access.
change on wetland ecosystem service values in peri-urban and urban area of Bahir
Glass, L.M., Newig, J., 2019. Governance for achieving the Sustainable Development
Dar City, Upper Blue Nile Basin, and North western Ethiopia. Ecol. Process.
Goals: how important are participation, policy coherence, reflexivity, adaptation and
Awad, I., Holländer, R., 2010. Applying Contingent Valuation Method to Measure the
democratic institutions? Earth Syst. Governance J. 2.
Total Economic Value of Domestic Water Services: a Case Study in Ramallah
Haab, T., Lewis, L.Y., Whitehead, J., 2020. State of the Art of Contingent Valuation.
Governorate, Palestine. Eur. J. Econ., Finance Admin. Sci. ISSN 1450-2275 Issue 20
Environ. Sci. https://doi.org/10.1093/acrefore/9780199389414.013.450.
(2010) © Euro-Journals, Inc. 2010 http://www.eurojournals.com.
Hanemann, W.M., 2006. The economic conception of water. In: Water Crisis: Myth or
Bateman, I.J., Turner, R.K., 1993. Valuation of environment, methods, and techniques:
Reality, 61. Taylor and Francis, London, UK, pp. 74–76, 200614.
the contingent valuation method. In: Kerry Turner, R. (Ed.), Sustainable
Hanley, N., Barbier, E.B., 2009. Pricing Nature: Cost–Benefit Analysis and Environmental
Environmental Economics and management: Principles and Practice. Belhaven Press,
Policy. Edward Elgar Publishing Limited, UK.
London, pp. 120–191.
Hawassa city Municipality, 2015. Annual report, 2015. Municipality Office, Hawassa.
Bateman, I.J., Langford, I.H., Turner, R.K., Willis, K.G., Garrod, G.D., 1995. Elicitation
Hawassa City Municipality (HCM), (2011). Hawassa city administration socio- economic
and truncation effects in contingent valuation studies. Ecol. Econ. 12, 161–179.
profiles; By Hawassa city administration finance and economic development department.
Belkayali, N., Atan, M., Talay, L., Akpinar, N., 2010. Determination of economic value of
Hawassa, Ethiopia.
Göreme Historical National Park via contingent valuation method. Acad. J. Sci. Res.
Hussey, J., Hussey, R., 1997. Business Research: a Practical Guide For Undergraduate
Essays 5 (9), 934–940.
and Postgraduate Student. Macmillan, London.
Ben-Akiva, M., McFadden, D., Train, K., 2019. Foundations of Stated Preference
Ikeuchi, A., Tsuji, K., Yoshikane, F., Ikeuchi, U., 2013. Double-bounded dichotomous choice
Elicitation: consumer Behavior and Choice-based Conjoint Analysis. Foundat. Trends
CVM for public library services in Japan: proc. Soc. Behav. Sci. 2013, 73, 205–208.
R Econom. 10 (1–2), 1–144. https://doi.org/10.1561/0800000036.
Jamean, E.S., Abas, A., 2023. Article Valuation of Visitor Perception of Urban Forest
Benti, S., Terefe, H., Collo-Concha, D., 2021. Challenges and prospects to sustain natural
Ecosystem Services in Kuala Lumpur. Land (Basel) 12, 572. https://doi.org/
and working landscapes in the urban areas in Ethiopia. Current Research in
10.3390/land12030572, 2023.
Environmental Sustainability. Elsevier.
Johnston, R., Boyle, K., Adamowicz, W., Bennett, J., Brouwer, R., Cameron, T.A.,
Bishop, R.C., 2018. Warm Glow, Good Feelings, and Contingent Valuation. J. Agric.
Hanemann, W.M., Hanley, N., Ryan, M., Scarpa, R., Tourangeau, R., Vossler, C.,
Resour. Econ. 43 (3), 307–320. September 2018(14 pages); Published By: Western
2017. Contemporary Guidance for Stated Preference Studies. J. Assoc. Environ.
Agricultural Economics Association.
Resour. Econ. 4 (2), 319–405.
Carlsson, F., Köhlin, G., Mekonnen, A., 2004. Working Papers in Economics no. 151.
Johnston, C., McDonald, J., & Quist, K., (2019). A Generalized Ordered Probit Model:
Department of Economics Gothenburg University, Sweden.
communications in Statistics - Theory and Methods to link to this article: https://doi.
Central Statistical Agency (CSA), (2007). Population and Housing Census of Ethiopia Result
org/10.1080/03610926.2019.1565780.
at Country Level: Addis Ababa, Ethiopia.
Klieštika, T., Kočišováa, K., Mišanková, M., 2014. Logit and Probit Model Used For
Chaudhry, P., Sharma, M.P., Singh, G., Bansal, A., 2013. Valuation of Urban
Prediction of Financial Health of Company. Elsevier. Science Direct.
Environmental Amenities in Developing Countries: a Case Study from Chandigarh,
Konijnendijk van den Bosch, C.C., Ferrini, F., Fini, A., Ferrini, F., & Fini, A. (2017).
India. Glob. J. Sci. Front. Res. Environ. Earth Sci. 13 (2). Version 1.0 Year 2013 Type:
Introduction. In Routledge Handbook of Urban Forestry.
Double Blind Peer Reviewed International Research Journal Publisher: Global
Lafortezza, R., Pauleit, S., Hansen, R., Sanesi, G., & Davies, C. (2017). Strategic Green
Journals Inc. (USA) Online ISSN: 2249-4626 & Print ISSN: 0975-5896.
Infrastructure Planning and Urban Forestry. In Francesco Ferrini, Cecil C. Konijnendijk
Chaudhry, P., Tewari, V.P., 2006. A comparison between TCM and CVM in assessing the
van den Bosch, & Alessio Fini (Eds.). Routledge Handbook of Urban Forestry. (pp.
recreational use value of urban forestry. Int. For. Rev. 8 (4), 439–448.
179–193). London, UK and New York, NY.
Chien, Y.L., Huang, C., Shaw, D., 2005. A general model of starting point bias in double-
Loomis, J., Kent, P., Strange, L., Fausch, K., Covich, A., 2000. Measuring the total
bounded dichotomous contingent valuation surveys. J. Environ. Econ. Manage. 50
economic value of restoring ecosystem services in an impaired river basin: results
(2), 362–377. September 2005.
from a contingent valuation survey. Ecol. Econ. 33, 103–117.
Christian, A., Vossler, and Holladay, J.S., (2018). Alternative Value Elicitation Formats in
National Oceanic and Atmospheric Administration (NOAA), 1993. Report of the NOAA
Contingent Valuation: Mechanism Design and Convergent Validity: © 2018. This
Panel on Contingent Valuation. Fed. Regist. 58, 4601–4614.
manuscript version is made available under the CC-BY-NC-ND 4.0 license http://cre
Nautiyal, H., Goel, V., 2021. Sustainability assessment: metrics and methods. Methods
ativecommons.org/licenses/by-nc-nd/4.0/.
Sustain. Sci. 2021.
Cochran, W.G., 1977. Sampling Techniques, 3rd ed. John Wiley & Sons, New York.

16
G. Zegeye et al. Trees, Forests and People 12 (2023) 100398

Nijnik, M., Miller, D., 2017. Valuation of ecosystem services: paradox or Pandora’s box Tyrvainen, L., Pauleit, S., Seeland, K., de Vries, S., In: Konijnendijk, C.C., Nilsson, K.,
for decision-makers? One Ecosystem: Ecosystem and Sustainability Data Journal. Randrup, T.B., and Schipperijn, J., (eds.) (2005) Urban Forests and Trees. A reference
Ning, P., Lee, S.H., 2019. Estimating the Young Generation’s Willingness to Pay (WTP) book, Springer-Verlag, Berlin Heidelberg, Germany.
for PM2.5 Control in Daegu, Korea, and Beijing, China. Sustainability. Tyrvainen, L., Vaananen, H., 1998. The Economic Value of Urban Forest amenities: an
Nowak, D.J., 2020. Understanding i-Tree: Summary of Programs and Methods. Forest Application of the Contingent Valuation method: Landscape and Urban Planning.
Service; US Departments of Agriculture. University of Joensuu, Faculty of Forestry, Joensuu, Finland, pp. 105–118.
Piana, M.R., Pregitzer, C.C., Hallett, R.A., 2021. Advancing management of urban United Nations, 2018. World urbanization prospects: the 2018 revision. Econ. Soc. Aff.
forested natural areas: toward an urban silviculture? Front. Ecol. Environ. https://population.un.org/wup/Publications/Files/WUP2018-KeyFacts.pdf.
Sajise, A.J., Samson, J.N., Quiao, L., Sibal, J., Raitzer, D.A., Harder, D., 2021. Contingent Van de Ven, P., 2018. Measuring Economic Welfare: a Practical Agenda for the Present
Valuation of Nonmarket Benefits in Project Economic Analysis a Guide to Good and the Future. In: Paper prepared for the 6th IMF Statistical Forum on “Measuring
Practice. ASIAN DEVELOPMENT BANK. Economic Welfare in the Digital Age: What and How?” November 19 –20, 2018.
Small, N., Munday, M., Durance, I., 2017. The challenge of valuing ecosystem services Washington. IMF Headquarters.
that have no material benefits. Global Environmental Challenge. Elsevier. Wieditz, I., and Penney, J., (2007). Climate Change Adaptation Options for Toronto’s Urban
Sylvie, N., 2010. Assessment of Farmers’ Willingness to Pay for the Protection of Forest: Clean Air Partnership, Canada, and Toronto.
Nyabarongo River System, Rwanda. Student Paper. University of Nairobi, p. 98. Wolf, K.L., 2003. Public response to the urban forest in inner-city business districts.
Tavárez, H., Elbakdze, L., 2021. Urban forests valuation and environmental disposition. J. Arboricult. 29 (3), 117–126.
The Case of Puerto Rico. Forest Policy and Economics. Yirssaw, T., 2021. The Economic Valuation of Urban Green Parks: the Application of
The Economics of Ecosystems and Biodiversity (TEEB), (2010). Mainstreaming the Contingent Valuation Method. J.Econ. Sustain. Dev.
Economics of Nature: a synthesis of the approach, conclusions, and recommendations Zhu, Z.X., Shin, H.J., 2021. Financial Analysis for Improving River Water Quality
of TEEB. through Introduction of Organic Agriculture. Sustainability.

17