Does location matter? The spatial equity implications of the Integrated Housing Development Program in Addis Ababa, Ethiopia

ABSTRACT

As Africa is rapidly urbanizing, there is perhaps no other urban policy area more pressing than the elimination of slums along with the development of equitable access to affordable housing for low-income residents, which requires immediate attention. Ethiopia has taken on this challenge through its Integrated Housing Development Program (IHDP), a policy that incorporates slum clearance and resettlement through a government-led development of condominium housing targeting those displaced and other low- and middle-income households. However, construction of most condominiums has taken place at large-scale conglomerations on periphery of the city considered peri-urban. These sites are far from the city center where slum clearance took place and where most social and economic activity continue to be concentrated. Utilizing a survey instrument, this study examines the mediating effect of IHDP residents’ perception of access to public, urban facilities, and transportation condition, two dimensions of spatial equity. The analysis found that those living at peri-urban IHDP sites perceive lower levels of spatial equity compared to urban IHDP condominium residents. Moreover, peri-urban residents’ perception of spatial equity is a statistically significant mediating factor on the relationship between their location and lower levels of overall satisfaction living at their IHDP condominiums.

KEYWORDS:

• Spatial equity
• slum clearance-relocation
• housing and development
• rapid urbanization

Introduction

The current strain of rapid urbanization in Ethiopia is most felt in the capital city Addis Ababa, with a population in 2000 of 2.4 million that grew to 4.8 million in 2019 and is projected to grow to 7.4 million by 2030 (World Population Review, 2022). Like all African regions experiencing rapid urbanization, Addis Ababa faces persistent issues and emerging urban challenges due to its increased population. Perhaps a significant challenge is addressing conditions of slums and providing access to affordable quality housing for the city’s growing residents. To ameliorate the persistence of slums and provide affordable housing for low and middle-income households, the Ethiopian Ministry of Urban Development and Construction (MUDC) launched the government-led Integrated Housing Development Program (IHDP) in 2005. With the target of reducing slum areas by 50%, the government cleared over 392 hectares of inner-city land deemed slums for redevelopment between the years 2008 and 2015, affecting over 25,000 households, of which 75% were previous government owned Kebele rental housing tenants (Soressa & Hassen, 2018). To address the relocation of those displaced by slum clearance and the overall shortage of affordable housing, the IHDP has built and transferred 175,000 IHDP condominium units as of July 2017 (Ezega, 2017).

The production and spatial distribution of IHDP condominium sites over time has taken place in two specific phases. The first primarily constructed pockets of condominium blocks in the central part of the city at or near the same location where clearance of existing housing took place. The second phase has condominium sites that are large in scale (number of units) and have been built or are planned to be built on the outskirts of the city on former farmland considered peri-urban. This shift has put forth implications of spatial equity related to access to public, urban facilities (i.e. schools, parks, government offices, hospitals and clinics) and transportation for those that have been relocated due to slum clearance and others that are residing at peri-urban IHDP sites.

The program has been the subject of inquiry by international organizations that have published reports that mostly provide overviews of the policy (UN-Habitat, 2017; World Bank, 2016), while few studies have been conducted using small qualitative samples to address households affected by slum clearance (French & Hegab, 2011; Soressa & Hassen, 2018). Anecdotal evidence has also trickled out from local media reporting on complaints from residents living at peri-urban sites, including their transportation woes and access to public, urban facilities such as schools, government offices, hospitals, and other urban amenities (Agizew, 2015; Bogale, 2017; Daftari, 2015). Yet, no scientific study has unpacked the spatial equity dimensions of access to public, urban facilities and transportation conditions living at peri-urban IHDP sites as of yet.

Utilizing a survey instrument, this study fills this gap by conducting original empirical research on the perceptions of IHDP condominium residents in Addis Ababa. More precisely, it examines: 1) the relationship between location and residents’ perception of spatial equity living at urban and peri-urban IHDP sites, and 2) how perception of spatial equity mediates the relationship between residents’ location (urban/peri-urban) and overall satisfaction living at IHDP condominiums.

Background

Slums in Addis Ababa

UN-Habitat (2016) defines slums as urban areas that lack one or more of the following: access to improved water; access to improved sanitation; sufficient living area/overcrowding; structural quality/durability of dwellings, and security of tenure. Using the UN-Habitat slum definition, it was estimated that in 2011, 80% of Addis Ababa’s housing stock was in a slum, with government-owned Kebele rental units comprising 70% of that stock (UN-Habitat, 2011). In addition to the conditions of slums in the city, a housing deficit of between 900,000 and one million units existed in Addis Ababa in 2010 (French & Hegab, 2011).

Slums in Addis Ababa today can be categorized in two typologies. The first typology consists of government-owned, low-income rental kebele housing units – most with no access to water, electricity, and drainage systems – located in the old core of the city with tenants having some legal status (Alemayehu & Stark, 2018). A second typology are made up of informal housing units constructed of low-quality materials such as corrugated metal and wood that are built on illegally occupied land and/or that do not conform to the land use or zoning systems and other regulations set by the government (Mathema, 2014). Households residing in what are deemed informal settlements do not have any legal status to their squatter housing. Hence, unlike most slums in the developing world that are found in conglomerations at the edge of cities, slums in Addis Ababa are unique in that they exist in pockets situated within middle-and high-income neighborhoods within the center of the city. Moreover, slum housing now makes up much of the urban fabric and housing arrangements (UN-Habitat, 2017).

Slum clearance and resettlement

Slum clearance and resettlement has a long controversial history, which began after World War II in the West (Collins & Shester, 2013; Teaford, 2000; Yelling, 2000). In the United States, urban renewal projects involved slum clearance, aided by Title I of the Housing Act of 1949, which aimed to revitalize American central cities by facilitating federal subsidies for locally planned redevelopment projects (Forest, 1985; Hassan, 2012). The primary goal of urban renewal was to improve the living conditions of the poor, low-income, and working populations of inner cities. Driven by ideals of modernization and urban renewal, many city neighborhoods deemed slums were bulldozed over to make space for new, modern high-rise housing and office blocks, roads, highways, and park development.

Sociologists were the first to sound the alarm about the negative consequences of slum clearance and relocation of communities in the West. In their seminal study of a slum clearance and relocation program in Bethnal Green of East London in England, Young and Willmott (1957) found that the majority of those interviewed would have preferred to stay in their original neighborhood, and they felt a loss of attachment from their long-term residences, families, and social networks. After residents had been relocated some 20 miles away from their original neighborhood of Bethnal Green, Young and Willmott (1957) reported further that ‘people’s relatives are no longer neighbors sharing the intimacies of daily life … their new neighbors are strangers’, and that those interviewed ‘frequently complained of the unfriendliness of the place’ (p. 147). Several other studies also highlighted the destruction of communities, and the socioeconomic dislocation that the 1950–60s slum clearance programs unwittingly forced on poor and working-class populations in England (Jennings, 1962; Paris, 1979; Thomas, 1986). These sentiments were also profoundly presented across the pond by Jane Jacobs (1969), who addressed urban renewal projects taking place in New York City.

This is not to say that slum clearance did not exist as a solution to undesired urban settlements both prior to and post-WWII in developing countries. Native and traditional housing and sanitation practices drastically differed from ‘modern’ practices of Western urban forms and planned settlements. In Lagos, Nigeria, the British colonial administrators cleared large areas of existing indigenous settlements under the guise of health concerns to make way for modern housing developments for colonial settlers as early as 1910–20s (Bigon, 2008). In South Africa, under the Apartheid regime, informal settlements deemed ‘slums’ and ‘squatter camps’ were systematically bulldozed to make way for ‘planned and controlled, racially and socio-economically segregated residential areas’ (Huchzermeyer, 2011, p. 8).

As the developing world becomes more urbanized, slum clearance has become the ‘silent companion to urban growth’ (Cernea, 1993, p. 3). Slum settlements in urban centers occupy valuable land in growing cities, making them a target for redevelopment by governments and developers alike. Given the lack of socioeconomic and political power among slum dwellers, many slum clearance and resettlement programs in developing countries have an authoritarian approach, leaving little room for the community to participate or negotiate before or after such programs. Arandel and Wetterberg (2013) define an authoritarian approach as one that ‘prioritizes technocratic engineering and urban planning goals over the social and economic concerns of residents or of the slum as a community’, with projects ‘considered successful if they meet or exceed technical requirements, such as requisite quality and quantity of housing completed within the time and funds allocated’ (p. 141). Arandel and Wetterberg (2013) rightfully note that their definition of an authoritarian approach is an oversimplification, but despite some governments’ inclinations to allow resident participation and consultation in projects, studies show that authoritarian practices largely prevail.

An early example of the authoritarian tendencies of slum clearance in Africa is the demolition of residents’ houses in Maroko, Lagos in the 1980s. The government of Nigeria began an urban renewal program in 1982 through a top-down approach in Maroko that involved no initial study of the location or its residents prior to the clearance and had little to no engagement with residents (Sule, 1990). Many residents were rent-paying dwellers who were not informed by their landlords of the eviction notice. In the end, fewer than 7% of residents were relocated, with the majority left homeless or ending up in new slum developments, from which they would eventually be ejected again (Agbola & Jinadu, 1997; Sule, 1990). Similarly, studies of the Villes Sans Bidonvilles (VSB) program in Morocco show that little attention was paid to the spatial impact of resettlement and to residents’ participation in the program (World Bank, 2006).

The literature further suggests that housing provision on the outskirts of a city for those affected by urban renewal and slum clearance has several unintended consequences. First, it prevents the most vulnerable and poor from making a living due to lack of sufficient transportation and linkages to jobs, which are concentrated in the city center (Alemayehu et al., 2018). Second, it affects families because of the distance to schools for their children, as well as to clinics, hospitals, and other public urban amenities (Opoko et al., 2015). Third, it involves a transfer of poverty to the fringes of society, making it more concentrated and pronounced without necessarily solving the issue itself. Finally, there is the problem of the reemergence of new slums in city center areas (where there were none), or the expansion of existing slums due to households returning from the outskirts of the city for better access to jobs and schools and to meet other socioeconomic needs that were not available (World Bank, 1998).

Dimensions of spatial equity

Urban policy scholars have put the focus on geography in the past several decades, pushing it towards the forefront of urban research within the context of social justice theory. Spatial justice theory derives from conceptualization of social justice into geographical space (Harvey, 1973). In the urban context, it is concerned with how resources ranging from housing, public services, and mobility are distributed spatially, and who benefits and why from the spatial organization and development of urban spaces. In Seeking Spatial Justice, Eduard Soja (2010) posits that the ‘theorization of spatial justice is the omnipresence of geographically uneven development and its associated spatial inequalities’ (p. 71). Soja further states:

Relating to the theorization of (in)justice, this means that whatever we do will very rarely, if ever, be distributed perfectly evenly or randomly over space. Our actions and activities will tend more or less to be nodal, focused around particular centers or agglomerations, and this centering or nodality will generate unevenly distributed advantages and disadvantages depending on location and accessibility with respect to the center or node. These fundamental or ontological features of human spatial organization give rise to more complex and unjust empirical geographies. (Soja, 2010, p. 72)

Soja’s theorization of spatial justice and equity is heavily influenced by Henri Lefebvre’s The Right to the City, and specifically a concept that Lefebvre defines as:

The right to the city should modify, concretise and make more practical the rights of the citizen as an urban dweller (citadin) and user of multiple services. It would affirm, on the one hand, the right of users to make known their ideas on the space and time of their activities in the urban area; it would also cover the right to the use of the centre, a privileged place, instead of being dispersed and stuck into ghettos (for workers, immigrants, the ‘marginal’ and even for the ‘privileged’ (Lefebvre, 1996, p. 34).

Many have used the works of Lefebvre and Soja to examine developing countries governements’ current perception of the city, which has raised the question of who has the right to the city as cities of the South (re)develop, renew, and modernize (Erdi-Lelandais, 2014; Parnell & Pieterse, 2014). Particularly related to this study are scholars and researchers that examine policies of slum clearnace and the resettlement of existing dwellers to fringes of the city, dislocating them from social and economic opportunities (Davis, 2013; Huchzermeyer, 2011; Koonings & Kruijt, 2009). Spatial equity also has horizontal spatial dimentions that measure equal distribution of public, urban facilities among residents regardless of their locations or socioeconomic conditions (Bennett, 1983). Spatial equity, therefore, takes on dimensions of both access to the city from the transportation perspective, including access to multiple modes of affordable and fair transportation, and access to public, urban facilities. Urban facilities include public schools, government offices, playgrounds, parks, hospitals, and clinics (Ashik et al., 2019).

Research questions

Informed by the literature, and more specifically by the framework of spatial equity, three research questions have been developed. While the primary focus is on residents’ perceptions of spatial equity, the overarching inquiry is to see how perceptions differ between residents living at inner-city and peri-urban IHDP sites.

RQ₁: Are residents living in the inner city more satisfied than residents living at peri-urban sites?

RQ₂: Do residents living in inner-city IHDP sites perceive higher levels of spatial equity gained than residents living at peri-urban IHDP sites?

RQ₃: Does perception of spatial equity play a mediating role in the relationship between IHDP location and satisfaction level?

Research hypotheses

H₁: Residents living in inner-city IHDP sites show higher overall satisfaction than residents living at peri-urban sites.

H₂: The perception of residents living in inner-city IHDP sites shows higher levels of spatial equity gained by living in their condominiums than those living at peri-urban sites.

H₃: Residents’ perceptions of spatial equity gained mediates the relationship between the location of the IHDP site and their overall satisfaction with their IHDP condominiums.

Data and methods

Data description

The study uses a survey instrument for data analysis. The secondary data was obtained from Clairvoyant Marketing Agency (CMA) based in Addis Ababa, Ethiopia, which administered the collection of data at four inner-city and six peri-urban IHDP locations between September and November 2020. A total of 1,018 cases – 298 inner city and 720 peri-urban – were collected by Urban Planning graduate students at the Ethiopian Institute of Architecture, Building Construction and City Development (EIABC) during the collection period.

The target population consists of current tenants of IHDP condominium units in Addis Ababa. Samples were drawn through a four-stage probability, stratified-quota, random sampling method to achieve N = 1,020. A sample size of 1,020 has been determined to be representative of the 175,000 IHDP units at a four percent margin of error and a 95% confidence level. At this level, the outcome of the study will confidently represent 43%–51% of the population. Additionally, a finite population correction has been applied to the sample size formula.

Surveys were conducted with primary adults or adults over the age of 21 who were also household members using a closed-end survey questionnaire through face-to-face interviews at randomly selected sites and units. Prior to the collection of data, a pretest of the questionnaire was conducted to identify and ameliorate possible weaknesses or biases in the survey questions. Since internet access is uncommon in Addis Ababa, physically conducting the survey was key in acquiring sufficient data to generalize from the findings. Data collection conforms to keeping participants’ anonymity by not using names or any other identifiable information on the questionnaire. Sites are labeled as either inner-city or peri-urban, and actual condominium site names are labeled by their legally designated IHDP names. However, building numbers and unit numbers are coded, which keeps the exact household unit and interviewee information discrete while allowing analysis of data of different IHDP locations between and within inner-city and peri-urban sites.

Dependent and independent variables

Dependent Variable (DV): Level of overall satisfaction with IHDP condominium

The DV will be operationalized using a Likert-type survey question that directly asks household members about their overall satisfaction with their current IHDP housing, using the scale presented in Table 1:

Table 1. Dependent variable.

Dependent Variable	Very satisfied	Fairly satisfied	Neither satisfied or unsatisfied	Not satisfied	Very unsatisfied
Overall, how satisfied are you with living at this condominium?	5	4	3	2	1

Independent Variable (IV): The IV is operationalized as a binary variable using MUDC’s designation of IHDP condominium sites being located within the central city or peri-urban areas: (0) Inner city (1) Peri-urban

Mediating variables (MVs)

Mediating variables (MVs) are variables that lie along the pathway between an independent and dependent variable, explaining all or part of the effect of the IV on the DV (Hayes, 2013). Informed by the literature and theoretical framework, two mediation models have been developed to measure the extent to which IHDP has delivered spatially equitable outcome to its residents. Perceived Spatial Equity is operationalized through two separate dimensions on perception of spatial equity related to 1) access to public urban facilities and 2) public transportation condition. First, a parallel-mediation model that utilizes four mediators related to the dimension of access to public urban facilities is used, which include schools (1^st – 12^th grade), hospitals and clinics, government offices, and parks and open space. These urban facilities have been identified as an important indicator for measurement of spatial equity in the literature (Ashik et al., 2019). A second single-mediation model is utilized to analyze the role of residents’ perceptions on public transportation conditions and how those perceptions exert their influence on the relationship between the location of IHDP site (IV) and overall satisfaction (DV).

Control variables

Three control variables are used that have a significant relationship between the mediator, independent, and dependent variables (Opoko et al., 2015). It is important to note that several key control variables have been excluded due to cultural factors. First, household income has been excluded because it is not a culturally acceptable question to ask in Ethiopia and could discourage individuals from participating further in a survey (Alemayehu & Stark, 2018). However, a control variable on auto ownership has been included as a proxy for household income. This proxy variable has been used in other studies conducted in developing nations (Ali & Dadush, 2012; Milanovic & Yitzhaki, 2002). Moreover, it is appropriate because car ownership is very expensive in Ethiopia, due to the cost of purchasing one and the high cost of gas in the country. Second, it is not culturally acceptable to ask one’s age in Ethiopia; therefore, this control variable has also been excluded. Third, due to the ethnic unrest in the country at the time of the collection of data, a question about ethnicity was also left out of the survey. The control variables that will be used and how they will be operationalized are as follows:

• Gender is operationalized as the sex of the surveyed individual – (0) Male (1) Female

• Tenure status is operationalized as the tenure type that the current occupant has with the unit – (0) Owner Occupied (1) Renter Occupied

• Auto ownership is operationalized using ownership of at least one automobile by the household – (0) Owns (2) Does not Own

Method of analysis and models

The primary method of analysis in this study utilizes Andrew F. Hayes’ mediation model using PROCESS macro version 3 in SPSS statistical software. Mediation analysis is used to investigate how an independent variable or antecedent variable (X) exerts its influence on an outcome variable (Y) through one or more intervening or mediating variables (M). This study utilizes mediation analysis to explore how location of IHDP site (X) influences overall satisfaction (Y) through measures of spatial equity mediators (M_j) – that is, the dimensions of spatial equity variables – operating in parallel or as a single mediator. To be precise, the study will use three parallel-mediation models and one simple mediator model to investigate the research questions.

Access to public, urban facilities model

The conceptual and statistical model with perceptions of access to urban facilities – a dimension of spatial equity investigating access to schools, hospitals and clinics, government offices, and parks and open space – is presented in Figure 1 illustrating that five paths are possible with X influencing Y: directly through path c’ and indirectly via M₁ through paths a₁ and b_1, via M₂ through paths a₂ and b_s, and so on.

Figure 1. Conceptual and statistical model with four mediators (access to schools, hospitals and clinics, government offices, and parks and open space) as items that encompass public, urban facilities as a dimension of spatial equity.

In statistical terms, these paths are represented with five equations, where a parallel multiple mediators with k = 4 consequent variables to estimate all the effects of X on Y.

(a) $M_1={i_M}_1+a_{1}X+e_{M1}$(a)

(b) $M_2={i_M}_2+a_{2}X+e_{M2}$(b)

(c) $M_3={i_M}_3+a_{3}X+e_{M3}$(c)

(d) $M_4={i_M}_4+a_{4}X+e_{M4}$(d)

(e) $Y=i_y+cX+b_1{M}_1+b_2{M}_2+b_3{M}_3+b_4{M}_4+e_y$(e)

In equations (a), (b), (c), and (d), a₁, a₂,a₃, and a₄ estimate the amount by which two cases that differ by one unit on X are estimated to differ on M₁, M₂, M₃, and M_4, respectively, where i denotes cases, i_M and i_Y are regression constants, and e_Mi and e_Yi are errors in estimation of M and Y. For example, in the model presented in Figure 1, equation (a) estimates the effect of location of an IHDP site’s (X) influence on a resident’s perception of access to schools (M₁) through Path a₁. Similarly, equations (b), (c), and (d) estimate the effect of an IHDP location’s influence on perceptions of access to hospitals and clinics (M₂), government offices (M₃), and parks and open space through Paths a₂, a₃, and a_4, respectively. In equation (e), b₁ estimates the effect of M₁ on Y, controlling for X and keeping all other mediator variables constant.

In a four-mediator model, the indirect effects of each mediator indirectly affect the relationship between X and Y as a product of the two paths linking X to Y through that single mediator. In other words, the specific indirect effect of resident’s perception of access to schools (M₁) in Figure 1 is a product of the regression coefficients corresponding to the two paths a1 and b₁, and when multiplied together, they yield the specific indirect effect of location of IHDP site (X) on overall satisfaction (Y) through perception of access to schools (M₁).

When all specific indirect effects are added together, they yield the total indirect effect of X on Y though all mediators in the model. Therefore, for a parallel multiple mediator model with four mediators represented by equations (a) through (e), the total indirect effect of X on Y is a₁b₁ + a₂b₂ + a₃b₃ + a₄b₄. Finally, the sum of the direct (c’) and indirect effects with four mediators is the total effect (c) and is represented in equation (f):

(f) $\frac{Totaleffect\left(c\right)=Indirecteffects+Directeffect(c^{\mathrm{\prime }})}{c=a_1{b}_1+a_2{b}_2+a_3{b}_3+a_4{b}_4+c^{\mathrm{\prime }}}$(f)

Transportation conditions model

Figure 2 represents a conceptual and statistical single-mediation model with perception of transportation condition as the mediator (M), location of IHDP site as the independent variable (X), and overall satisfaction as the dependent variable (Y). It also shows all paths that are possible in a mediation model with a single mediator. The statistical equation with a single-mediator model is written as:

Figure 2. A conceptual and statistical model with the single mediator of perception of transportation condition as a dimension of spatial equity exerting its influence on the relationship between IHDP site (X) and overall satisfaction (Y) with all possible paths.

(g) $M=i_1+aX+{e_M}_{\hspace{0.17em}}$(g)

(h) $Y=i_2+cX+bm+e_y$(h)

The equation represents a system in which at least one antecedent X variable (location of IHDP site) is proposed as influencing an outcome Y (overall satisfaction) through a single intervening variable (perception of transportation condition). There are two specific pathways by which X is influencing Y: (a) one pathway is the direct pathway that leads from X to Y without passing through M, and (b) the second pathway from X to Y is the indirect effect of X on Y through M. In the equations (a) and (b), i₁ and i₂ are regression intercepts, e_M and e_Y are errors in the estimation of M (transportation condition) and Y (overall satisfaction), respectively, and a, b, and c’ are regression coefficients given to the predictor variable (IHDP location) in the model in the estimation of the consequents.

Results

This section presents the results from the mediation models in two parts. The first part presents the results for the parallel mediation analysis of access to public, urban facilities. The second part presents the results for the single mediation analysis of transportation conditions.

Access to public, urban facilities result

Four variables are used to analyze perception of spatial equity from the dimension of access to public, urban facilities and the variables’ mediating effect on the relationship between location of IHDP (IV) site and overall satisfaction (DV). All four mediating variables were measured through Likert-scale questions on respondents’ perception of access to public, urban facilities that included schools (1^st – 12^th grade), hospitals and clinics, government offices, and parks and open space. Three control variables are also included in the model: gender (male or female), car ownership (own or do not own), and tenure status (rent or own). Table 2 displays the descriptive statistics, frequency, mean, and standard deviation for the variables used in the model.

Table 2. Distribution of mediating urban public facilities variables that influence the relationship between the independent variable (X) IHDP site and dependent variable (Y) overall satisfaction for n = 298 (urban) and n = 720 (peri-urban) for a total of N = 1,018. Not included in the table are the three control variables (gender, car ownership, and tenure status).

How do you rate access to the following?
	Very Good	Good	Neither Good nor Bad	Bad	Very Bad	N	Mean	SD
Schools (1st – 12th Grade)
Urban	28 (9.4%)	240 (80.5%)	22(7.4%)	8 (2.7%)	0 (0%)	N = 298	3.97	0.524
Peri-urban	13 (1.8%)	384 (53.3%)	83 (11.5%)	217 (30.1%)	23 (3.2%)	N = 720	3.2	0.997
Hospitals and Clinics
Urban	112 (37.6%)	157 (52.7%)	14 (4.7%)	15 (5.0%)	0 (0%)	N = 298	4.23	0.758
Peri-urban	77 (10.7%)	285 (39.6%)	88 (12.2%)	220 (30.6%)	50 (6.9%)	N = 720	3.17	1.175
Government Offices
Urban	74 (24.8%)	159 (53.4%)	33(11.1%)	32(10.7%)	0 (0%)	N = 298	3.92	0.886
Peri-urban	41 (5.7%)	228 (31.7%)	116(16.1%)	208 (28.9%)	127 (17.6%)	N = 720	2.79	1.223
Parks and Open Space
Urban	89 (29.9%)	142 (47.7%)	24 (8.1%)	40 (13.4%)	3 (1.0%)	N = 298	3.92	1.002
Peri-urban	77 (10.7%)	281 (39.0%)	42 (5.8%)	256 (35.6%)	64 (8.9%)	N = 720	3.07	1.235

Preliminary data screening found that no missing values are present for all variables with N = 1,018 (urban n = 298, peri-urban n = 720). A histogram for the four mediating variables further found that all variables were relatively normally distributed. One early observation of frequencies is that 30.1% of peri-urban residents rated their access to schools as ‘bad’, compared to only 2.7% at urban IHDP sites. Similarly, 30.6% of peri-urban residents also rated access to hospitals and clinics as ‘bad’, compared to a mere 5.0% of residents at urban sites.

Results of effect size are reported at the 95% confidence interval (CI) level with p < .05 values and are presented in Figure 3 and Table 3. For Path a₁ in the model, the location of an IHDP site is found to be a statistically significant predictor of perceptions of access to schools, b = −0.763, t (df = 1,013) = −12.400, p < .05. This indicates that those living at peri-urban sites are estimated to differ, on average, −0.763 units lower in their perception of access to schools at their IHDP sites than urban residents. When controlling for gender (b = 0.062, t (df = 1,013) = 1.104, p = 0.27), car ownership (b = 0.021, t (df = 1,013) = 0.326, p = 0.75), and tenure status (b = −0.133, t (df = 1,013) = −2.394, p < 0.05), the analysis found that only tenure status of condominium was a statistically significant confounding factor, which indicates that those who rent view their condition of access to schools less favorably than those that own their condominium units.

Figure 3. Parallel-mediation model showing the direct (c’) and indirect paths by which perception about access to schools, hospitals and clinics, government offices, and parks and open space as items that encompass public, urban facilities as a dimension of spatial equity influence overall satisfaction (N = 1,018). Indirect, direct, and total effects, along with b coefficients for Paths a₁, a₂, a₃, a₄, b₁, b₂ b₃, and b₄ are reported, along with their respective 95% confidence intervals. Statistical control variables are not represented in the model for simplicity.

Table 3. Analysis results for Paths a₁ through a₄ and b₁ through b₄ with N = 1018. That is, the relationship between X and mediator variables M₁ through M₄ with control variables, and the relationship between the mediator variables (M_j) and the outcome variable ‘overall satisfaction’ (Y). Effect size is reported at 95% confidence interval (CI) level with statistically significant p < .05* and p < .001** values.

Influence of IHDP site (X) on Mediators Mj with Control Variables N = 1,018
Paths	M1: Schools a1	M2: Hospitals and Clinics a2	M3: Government Offices a3	M4: Parks and Open Space a4
IHDP Location (X)	b = −0.763 p < .05**	b = −1.048 p < .05**	b = −1.115 p < .05**	b = −0.835 p < .05**
	95% CI [−0.8838, −0.6423)	95% CI [−1.1932, −0.9024]	95% CI [−1.2697, −0.9606]	95% CI [−0.9953, −0.670]
Control Variables
Gender (Male, Female)	b = 0.062 p = 0.27 95% CI	b = 0.187 p < 0.05* 95% CI	b = 0.088 p = 0.22 95% CI	b = 0.074 p = 0.32 95% CI
	[−0.0479, 0.1712}	[0.0548, 0.3186]	[−0.0519, 0.2285]	[−0.0710, 0.2196]
Car Ownership (Own, Don’t Own)	b = 0.021 p = 0.74 95% CI [−0.1034, 0.1445]	b = −0.043 p = 0.57 95% CI [−0.1926, 0.1057]	b = −0.125 p = 0.12 95% CI [−0.2833, 0.0339]	b = −0.077 p = 0.36 95% CI [−0.2411, 0.0875]
Housing Tenure (Own, Rent)	b = −0.133 p < 0.05* 95% CI [−0.2423, −0.0240]	b = −0.179 p < 0.05* 95% CI [−0.3100, −0.0472]	b = −0.174 p < 0.05* 95% CI [−0.3141, −0.0347]	b = −0.079 p = 0.29 95% CI [−0.2232, 0.0662]
Influence of Mediators (Mj) on (Y) ‘Overall Satisfaction’
	M1: Schools b1	M2: Hospitals and Clinics b2	M3: Government Offices b3	M4: Parks and Open Space b4
	b = 0.360 p = 0.23	b = 0.119 p < 0.05**	b = 0.107 p < 0.05**	b = 0.010 p = 0.64
	95% CI [−0.0221. 0.0937]	95% CI [0.0707, 0.1680]	95% CI [0.0650, −0.0895]	95% CI [−0.0305, 0.0497]

An IHDP site predicting perceptions of access to hospitals and clinics (Path a₂) was statistically significant, b = −1.048, t (1,013) = −14.142, p < .05. The analysis finds that those living in peri-urban sites perceive access to hospitals and clinics, on average, −1.048 units lower than urban residents. Similarly, the location of an IHDP site was a statistically significant predictor of residents’ perception of access to government offices (Path a₃), b = −1.115, t (1,013) = −14.156, p < .05, with those living at peri-urban sites reporting negatively on their access to government offices compared to their urban counterparts. Lastly, an IHDP site is also statistically significant in predicting perception of access to parks and open space (a₄), b = −0.835, t (1,013) = −10.232, p < .05, with peri-urban residents reporting negative views on their access to parks and open space compared to urban residents.

These findings suggest that hospitals and clinics are the most significant public, urban facilities that peri-urban residents reported as the least accessible at their locations. Moreover, car ownership as a control variable used as a proxy for household income was not a significant confounding factor to all mediator variables, suggesting that a household’s income status is not a significant factor in the analysis. As previously mentioned, gender, when controlled for, was a statistically significant factor for a single mediator. The findings indicate that female respondents rated their access to hospitals and clinics at a statistically significant higher level than their male counterparts, b = 0.187, t (1,013) = 2.778, p < .05. Meanwhile, tenure status was found to be a statistically significant confounding variable to all mediators except for perception of access to parks and open space, with renters perceiving access to all the other public urban facilities at a much lower level than owners.

Although an IHDP location is a statistically significant predictor for all mediators, the mediators’ influence in predicting overall satisfaction varied when an IHDP site was absent in the analysis. For all respondents, overall satisfaction was positively predicted only by access to hospitals and clinics (Path b₂), b = 0.119, t (1,009) = −4.816, p < .05, and access to government offices (Path b₃), b = 0.107, t (1,013) = 4.990, p < .05. Statistically, this indicates that for every increase in perception of access to hospitals and clinics and government offices, overall satisfaction increases by 0.119 and 0.107, respectively. However, neither access to schools, b = 0.036, t (1,013) = 1.214, p = 0.23, nor access to parks and open space (Path b₄), b = 0.010, t (1,013) = 0.470, p = 0.64 were a statistically significant predictor of overall satisfaction.

When leaving all mediators out of the equation and looking at the direct influence control variables have on overall satisfaction (DV), only whether a resident owns or rents their condominium, b = −0.263, t (1,013) = 1.214, p < .001, was statistically significant. The two other control variables of gender, b = 0.888, t (1,013) = 1.865, p = 0.06, and ownership of car b = −0.090, t (1,013) = −1.708, p = 0.09 had no significant influence on residents’ overall satisfaction. This finding indicates that renters report less overall satisfaction than owners.

To sum up, of all the mediators in the parallel-mediation model, access to hospitals and clinics exerts its effect the most indirectly between IHDP location and overall satisfaction, b = −0.125, closely followed by access to government offices, b = −0.120. We also find that the direct effect of IHDP location on overall satisfaction (c’) is not statistically significant, b = −0.076, SE = 0.061, p = 0.21, while the total effect (c) of the model is statistically significant, b = −0.356, SE = 0.054, p < 0.05. Thus, only when the mediators are exerting their influence do we find that IHDP location to be a statistically significant predictor of overall satisfaction in this particular parallel-mediation model. The total effect of X on Y and all four specific indirect effects for each mediator are represented in Table 4.

Table 4. Parallel-mediation analysis results of the total (c), direct (c’), and indirect effects for location of IHDP site (X) on overall satisfaction (Y) via the mediating variables related to access to public, urban facilities as a dimension of spatial equity.

Indirect, Total, and Direct Effects
Indirect effect via M1: Schools a1b1	Indirect effect via M2: Hospitals and Clinics a2b2	Indirect effect via M3: Government Offices a3b3	Indirect effect via M4: Parks and Open Space a4b4
b = −0.027	b = −0.125	b = −0.120	b = −0.008
95% CI [−0.0893. 0.0219]	95% CI [−0.2201, −0.0907]	95% CI [−0.2105, −0.0895]	95% CI [−0.0561, 0.0361]
	Total effect c	Direct effect c’
	b = −0.356 p < .05** 95% CI [−0.4613, −0.2513]	b = −0.076 p = 0.21 95% CI [−0.1955, 0.0427]

Transportation condition result

This section analyzes the mediating effect of perception of public transportation conditions (M), a dimension of spatial equity, on the relationship between IHDP site (X) and overall satisfaction (Y). The method for analysis is a simple mediation model with one mediator, which is a single question asking survey respondents how they would rate the transportation conditions at their current IHDP location compared to their previous housing location measured on a Likert-style scale of ‘much better’, ‘better’, ‘same’, ‘worse’, or ‘much worse’. The variable yielded no missing values, N = 1018, SD = 1.033, and a histogram further showed a close-to-normal curve of distribution. Descriptive statistics of the data presented in Table 5 show the frequency for urban (n = 298) and peri-urban (n = 720) IHDP sites for N = 1,018. It should be noted that 19.1% of respondent at urban IHDP sites reported that their transportation condition is ‘much better’, while only 3.3% of peri-urban residents felt the same. Moreover, 8.9% of peri-urban residents rated their condition as ‘much worse’, compared to only 0.3% for urban residents.

Table 5. Frequency for respondents on transportation condition at urban and peri-urban IHDP sites. Not included in the table are the three control variables (gender, car ownership, and tenure status) .

How do you rate the transportation condition at this location
	Much Better	Better	Same	Worse	Much Worse	N	Mean	SD
Urban	57 (19.1%)	189 (63.4%)	39 (13.1%)	12 (4.0%)	1 (0.3%)	N = 298	3.97	0.717
Peri-Urban	24 (3.3%)	289 (40.1%)	110 (15.3%)	233 (32.4%)	64 (8.9%)	N = 720	2.97	1.102

The mediation model analysis findings at 95% CI are presented in Table 6 and Figure 4. First, the analysis found that Path a is statistically significant, b = −0.973, t (1,013) = −13.990, p < .05. The negative effect size indicates that peri-urban residents perceive transportation conditions (M) at their location, on average, −0.973 units lower than those living at urban IHDP sites. When controlling for gender, this influence was not a factor, but both car ownership, b = −0.211, t (1,013) = −2.951, p < .05, and housing tenure status, b = −0.127, t (1,013) = −2.951, p < .05, were significant factors. It is not surprising that respondents who do not own a car perceived transportation condition, on average, −0.211 units lower than those who have at least one member of the household owning a car.

Figure 4. A single-mediation model showing the direct (c’) and indirect paths by which perception of transportation condition as a dimension of spatial equity influences overall satisfaction (N = 1,018). Indirect, direct, and total effects, along with b coefficients for Paths a and b, are reported at their respective 95% confidence interval level. Statistical control variables are not represented in the model for simplicity.

Table 6. Analysis results for Paths a and b with control variables (N = 1,018). Table also presents the total (c), direct (c’), and indirect effects for location of IHDP site (X) on overall satisfaction (Y) via the mediating variable of perceived transportation condition as a dimension of spatial equity. Effect size is reported at 95% confidence interval (CI) level with statistically significant p < .05* and p < .001** values.

Influence of IHDP site (X) on Mediators M, and Indirect, Total and Direct Effects for N = 1,018
Paths	M: Transportation Condition a	M: Transportation Condition b	Total Effect c	Direct Effect c’
	b = −0.973 p < 0.05**	b = 0.083 p < .05*	b = −0.356 p < 0.05**	b = −0.276 p < 0.05**
	95% CI [−1.1096, −0. 8366]	95% CI [0.0356, 0.1299]	95% CI [−0.4613, −0.2513]	95% CI [−0.3899, −0.1617]
Covariates
Gender (Male, Female)	b = 0.081 p = 0.20 95% CI	b = 0.116 p < 0.05* 95% CI	b = 0.122 p < 0.05* 95% CI	Indirect Effect of M on Y Indirect effect via M: Transportation Condition b = −0.081 95% CI [−0.1678. −0.0357]
	[−0.0431, 0. 2046]	[0.0206, 0.2103]	[0.0269, 0.2174]
Car Ownership (Own, Don’t Own)	b = −0.211 p < 0.05* 95% CI [−0.3507, −0.0705]	b = – 0.092 p = 0.10 95% CI [−0.1991, 0.0161]	b = −0.109 p = 0.05 95% CI [−0.2167, 0.0012]
Housing Tenure (Own, Rent)	b = −0.127 p < 0.05* 95% CI [−0.2423, −0.0240]	b = −0.298 p < 0.05** 95% CI [−0.3926, −0.2034]	b = −0.309 p < 0.05** 95% CI [−0.4034, −0.2137]

Perception of transportation condition influencing overall satisfaction (Path b) when the location of an IHDP site was not a factor was also statistically significant, b = 0.083, t (1,013) = −3.442, p < .05. That is, for every unit increase in perception of transportation conditions, overall satisfaction increases by 0.083 units. It should also be noted that when controlling for gender, it was not statistically significant for Path a, but it was found to be a significant confounding variable for Path b, b = 0.116, t (1,013) = 2.390, p < .05. Additionally, while controlling for car ownership was not statistically significant, housing tenure was, b = −0.298, t (1,013) = −6.183, p < .05, with renters showing negative perceptions of transportation conditions compared to owners that leads to their lower overall satisfaction compared to owners.

The mediation analysis of location of IHDP site indicates that it has a significant, negative influence on overall satisfaction indirectly through perception of transportation condition, b = −0.081, SE = 0.034, p < .05. In other words, living in a peri-urban IHDP site negatively affected residents’ perceptions of transportation condition, and in turn their perception of the condition of transportation negatively influenced their overall satisfaction living in their IHDP condominiums. The results also show that there is a statistically significant, negative total effect (c) −0.356. Thus, we can conclude that although IHDP location is found to be a statistically significant predictor of overall satisfaction (c’), b = −0.276, SE = 0.060, p < 0.05, with a negative effect size indicating that those living in peri-urban sites, on average, perceive transportation condition −0.276 units lower of overall satisfaction than urban residents, the model also suggests that perception of condition of transportation is a statistically significant factor that has a partial indirect effect that exerts its influence negatively between location (IV) and overall satisfaction (DV). Hence, we can accept our hypotheses that spatial equity has a mediating effect on the relationship of IHDP location and overall satisfaction, with those living at peri-urban sites showing lower levels of spatial equity, and that this influences their lower levels of overall satisfaction compared to urban IHDP residents.

Discussion

In investigating the mediating role of perceived access to public, urban facilities, and transportation conditions – two dimensions of spatial equity of interest in this study – both models were found to be statistically significant in influencing the relationship between IHDP site and overall satisfaction. In the first model, the analysis looked to unpack which public urban facilities had significant roles as mediators. When simply analyzing the four mediators as dependent variables, the analysis found that peri-urban residents perceived access to schools, hospitals and clinics, government offices, and parks and open space negatively compared to those living at urban IHDP sites. Particularly, peri-urban residents perceive their access to hospitals and clinics and government offices to be worse at their locations.

Peri-urban residents’ lower levels of perception of access to government offices can be related to the fact that most government offices are located in the inner-core of the city. It should be noted, almost all legal processes for individuals in Ethiopia require complicated bureaucratic juggling (Gebre, 2020). Typically, one must go from one government office to another, and then yet another to complete the processing of a single document, most of which is still done through manual paper forms. What can be done over the internet in a few minutes by most in the West may take a city resident an entire day or more in Addis Ababa. It will be an overwhelming transformative investment for the city to transition to a paperless system for its administrative functions, but in the meantime, it should consider how to make some basic services available at satellite offices situated closer to peri-urban areas with dense populations.

When controlling for gender, we find that it is only statistically significant in relation to access to hospitals and clinics, with women having a more positive perception than men. The possible reason for this outcome maybe that women were overrepresented in the data, particularly at urban IHDP locations where they made up 60.1% of respondents. In all other cases, however, gender was not a statistically significant factor. Neither was car ownership, which was used as a proxy for household income. However, in this instance it is also related to mobility, which affects perception of access to public, urban facilities, even though it was not a statistically significant factor.

In addition, we find that renters view their access to schools, hospitals and clinics, and government offices negatively compared to owners. Tenure status was only a statistically significant factor for access to parks and open space. Renters’ negative perceptions of transportation condition were also found to be statistically significant. We also found that both perception of access to public, urban facilities and transportation condition exerted their influence negatively between location of IHDP site and overall satisfaction, and their effect sizes were statistically significant in all cases. Moreover, in the analysis of access to public, urban facilities, we find that those who held positive perceptions of their access to hospitals and clinics and government offices had higher levels of overall satisfaction, even when IHDP location was held constant.

It is not an understatement to say that access to public transportation underpins many if not all the dimensions of spatial equity. While not a provision that was included in the development of IHDP locations, access to public transportation has become a major factor in determining many other related equity issues for peri-urban residents. Addis Ababa has addressed public transportation conditions in the city, most significantly by constructing the first light-rail system in Sub-Saharan Africa, and it is currently expanding its city bus system and. However, the majority of city dwellers living in peri-urban areas are not connected to these transportation modes, and depend on the most prominent form of public transportation – minivans nicknamed ‘blue donkeys’ that hold small number of passengers (Miller, 2018).

Addis Ababa’s minivans hold up to twelve passengers at a time and cover most of the city, transporting daily travelers to areas that both the light rail and busses do not serve. Considering that peri-urban IHDP sites are large in scale and contain several hundred condominium units, it is not unexpected that residents living at these sites were less satisfied with their transportation condition than urban residents. It is common to see long lines of people waiting for the minivans as early as 5:00 a.m. to make it to work by 8:00 a.m., something that would take a car owner thirty minutes. In fact, research has shown that IHDP owners have been forced to either sell or rent their condominium units, primarily due to existing transportation conditions (Kassahun & Bishu, n.d.).

Some studies have examined the relationship between life satisfaction and whether one is a homeowner or a renter in the West (Helderman et al., 2004; Kuroki, 2019; Shanley, 1979). In the United States, home ownership is at the core of the American dream, but it has not been attainable for all segments of society (Barnes & Jaret, 2003). The correlation between satisfaction and housing tenure has consistently led to conflicting results in the literature. On the one hand, studies have found that homeowners tend to be more satisfied than renters, but as expense of location increases (i.e. property taxes), homeowners show lower levels of satisfaction (Kuroki, 2019). In most studies, however, homeowners are generally more satisfied than renters, and they prefer paying their mortgage to paying rent, even outside the United States (Park & Seo, 2020). Homeowners attribute their happiness to the improved lifestyle that is afforded to them, but most importantly because they believe that homeownership builds emotional and financial equity.

Conclusion

To summarize, housing is a human rights issue that governments must address so there are equitable outcomes for all citizens. At the policy level, the right to adequate housing does not require a country to build housing for its entire population, however, it must not discriminate, paying special attention to the most vulnerable and marginalized guaranteeing adequate housing for everyone (United Nations, 2014). It is also a misconception that a country does not have the right to engage in development or modernization projects that may result in displacement. For growing cities, certain areas may need to be considered for clearance, upgrading or for general redevelopment needs. In the case of Addis Ababa, most cleared areas have been redeveloped for large-scale new road systems, commercial corridors, business districts, and transportation systems. Nevertheless, international law does stipulate that displacement should not be done without consultation with those affected, and there should be solutions that minimize the scale of disruptions caused by displacement (United Nations, 2014).

It is important to note most peri-urban sites surveyed had only been built, delivered, and occupied within two to five years of the collection of data for this study, and many of the spatial conditions investigated may differ over periods of time. To an extent, this is a limitation to this study, as it captures a moment in time of a policy that is continuing to expand in the city. For instance, the city is currently developing and expanding its public transportation systems, which may drastically decrease the spatial conditions of peri-urban residents in the coming years. Hence, it will be important to see how residents’ perceptions change over time by duplicating the study two, five, or ten years from now.

However, as this study has shown, the current location of condominiums has spatial equity implications, as the perceptions of the residents themselves demonstrated. The disproportionate effects of lower levels of spatial equity on peri-urban residents should be of concern and the distribution of urban public facilities and transportation services needs to be addressed alongside condominium construction at peri-urban sites. One policy recommendation is that the city administration conduct a thorough review of the policy from the perspective of those living at all areas of IHDP locations on their needs related to equity and use findings to inform and address those needs and apply them to existing and future development of IHDP condominiums in the city. Through the literature review and study, we find that very little seems to have been done by the city government in conducting research or monitoring the progress of IHDP condominiums and/or residents once they have been delivered to recipients. What little does exist in the literature does not address the conditions through the lens of equity. The IHDP should put in place a systematic review of the policy facilitated by on the ground survey and interview of residents, as well as other geospatial assessment methodologies regularly. This should in turn inform the adjustment of the policy accordingly towards a just and equitable outcome for all residents at all locations.

Disclosure statement

No potential conflict of interest was reported by the author(s).

Correction Statement

This article has been corrected with minor changes. These changes do not impact the academic content of the article.