Understanding urban poverty in two high-density suburbs of Harare, Zimbabwe

ABSTRACT

Urban communities are heterogeneous and averages mask inequities and deprivations among poor and rich urban communities. This article examines the situation of households residing in two low-income, high-density suburbs of Harare, Zimbabwe. The aim of the research was to contextualise urban poverty by looking at selected urban communities and vertically analysing the patterns and determinants of poverty. A household survey was administered to 1000 households and qualitative data were collected through focus group discussions and key informant interviews. The study found high levels of income poverty and also found differences in poverty experiences between the two suburbs. The major proximate determinants of poverty were large family size; low education level of the household head; lack of income from permanent employment; low cash transfers; and short length of residence in the suburb. Increasing household income consumption can be addressed through scaling-up industries, which would result in more quality employment.

KEYWORDS:

• Urban poverty
• total consumption and food poverty lines
• high-density suburbs
• Harare

1. Introduction

Urban areas have historically been associated with economic development and relatively greater prosperity. Urban residents are consistently exhibited in national statistics to be better off compared with their rural peers. In Zimbabwe, 76% of rural households lived below the Total Consumption Poverty Line, compared with 38% of urban households (ZIMSTAT, 2013a). In addition, 23% of rural households live in extreme poverty, compared with 4% of urban households (ZIMSTAT, 2013a). However, inequality amongst urban residents in Zimbabwe is higher than those in rural areas. The Gini coefficient measures inequality in society and ranges from zero, perfect equality, to one, perfect inequality. In Zimbabwe, urban areas had a Gini Index of 0.39, compared with 0.37 for the rural areas (ZIMSTAT, 2013a). The heterogeneity in urban communities probably makes averages mask inequities and severe deprivations among poor and rich urban communities. In addition, while the depth of rural poverty is deep, with a poverty gap index of 0.36, the poverty gap index for urban areas is lower at 0.12 and many urban households live on the margins of the poverty line (ZIMSTAT, 2013a).

Urban and rural residents experience different forms of poverty and deprivation. There are specific problems that occur in urban areas which predispose vulnerable households to poverty. For example, while rural households rely on their own supply for consumption, food security for urban dwellers is about access rather than availability. Poor families often cannot afford to purchase adequate nutritious food, even when it is available in shops. This directly impacts on the development of children within the household, resulting in stunting, wasting and other forms of malnutrition.

The level of disaggregation of national studies is not sufficient for answering specific questions such as where the poor are located in the city and whether they have access to services. Average figures can be misleading and average urban poverty data also often fail to include those without formal addresses or permanent homes. The motivation for the urban poverty study was to understand more about the complexity of the urban situation. Since the poverty datum line study on consumption needs of urban families carried out by Cubitt and Riddell (1974), poverty studies in Zimbabwe have tended to concentrate on rural poverty (see for example Bird & Shepherd, 2003; Kinsey, 2010; Cavendish & Campbell, 2011).

This article explores the nature of urban poverty and draws on primary research undertaken by the authors consisting of a specialised study on aspects of multidimensional poverty. The study included a survey carried out in 1000 randomly sampled households in two high-density, low-income suburbs of Harare, Highfield (one of Harare's oldest suburbs) and Epworth (an informal settlement), in April and May 2013. The aim of the research was to contextualise urban poverty by looking at selected urban communities and vertically analysing the patterns and determinants of poverty (Manjengwa et al., 2014). The knowledge base resulting from this research can be used for promoting equitable and evidence-based interventions and public policies related to poverty in urban communities.

Two high-density suburbs, Highfield and Epworth, were selected for the Specialised Urban Poverty Study to gain an in-depth understanding of poverty in urban areas. The suburbs were selected through a consultative process with stakeholders, namely a Technical Working Group of key experts who were responsible for overseeing the design of the research. The urban poverty Technical Working Group consisted of representatives from government, including the Ministry of Public Service, Labour and Social Welfare, the Ministry of Finance and Economic Development and the Zimbabwe National Statistics Agency (ZIMSTAT), as well as non-governmental organisations, international development partners and academia.

Insights for site selection were obtained from experiences of members of the Technical Working Group as well as from literature and two previous studies, carried out by the Institute of Environmental Studies. These studies were the Moving Zimbabwe Forward Wellbeing and Poverty Survey carried out in 16 districts of Zimbabwe and the National Survey on Solid Waste Management, for the Environmental Management Agency, which used waste generated by a household as a proxy for mean monthly household income. In both studies, Epworth and Highfield of the Harare Metropolitan Province stood out as having relatively high levels of poverty. The Moving Zimbabwe Forward Wellbeing and Poverty Survey 2011 found that the incidence of household poverty in Epworth was 82%, similar to rural districts (Manjengwa et al., 2012). In addition, Highfield stood out as having a relatively high incidence of poverty in the high-density and low-density suburbs of the Harare sample.

Highfield, located in the southwest of the city and established on Highfield Farm in 1935, is the second-oldest high-density suburb in Harare after Mbare. Highfield is estimated to hold approximately 14 159 residential properties of sizes ranging between 200 and 400 m². Although Highfield was established as a low-income housing area during the colonial era, it has a section in Old Highfield where affluent black people were allowed to purchase stands and build houses of their choices in comparison with the other areas where the government had built low-cost basic housing. Today, many of the residential properties have multiple household occupancy.

Highfield is a typical well-planned urban suburb with a good road network, water and sewer systems. It has 11 primary and six secondary schools. There are a number of private clinics and surgeries in addition to the local authority's poly-clinic and satellite clinic.

Epworth was established as a Methodist Mission Station in 1890. There was a large influx of people during the late 1970s and early 1980s, with the population increasing from 20 000 in 1980 to 35 000 in 1987. The Methodist Church could not control the influx of people, and therefore transferred ownership of the farm to the Ministry of Local Government in 1983. The population of Epworth continued to rise and by 2012 had increased to 167 462 (ZIMSTAT, 2013b).

Epworth is within the Harare Metropolitan Province but is administered by a Local Board. It is the only informal settlement to have been tolerated by the Zimbabwean government in the post-independence period because some of its residents had lived there for a long time. Epworth had not been planned as an urban residential area, and therefore the rapid increase in population occurred on land without any water supply or sanitation facilities.

Epworth has five primary schools, two public secondary schools, one private college and four clinics. The road network, the school's recreational infrastructure and commercial infrastructure are not as well developed as in Highfield. In fact, there are areas with dwelling units without any established roads.

Although the two sample suburbs of the Specialised Urban Poverty Study have different histories and characteristics, they do not capture the entire varied and diverse urban situation in Harare. Nevertheless, the study does illustrate the complexity and diversity as well as levels of poverty experienced by the sample, within the urban situation.

Highfield and Epworth low-income residential suburbs were purposely sampled, after which multi-stage random sampling was used to select households in the two suburbs. The first stage in the multi-stage sampling was to determine the minimum sample size for each suburb using the Dobson Formula, thus making inferences about each suburb possible. The minimum sample size for each suburb was found to be 481, which we increased to 500 households. The total sample size for the survey was therefore 1000 households.

The second stage involved randomly selecting lists of 50 enumeration areas in each suburb from the ZIMSTAT Master Sample Database. Ten houses were then randomly selected using random number tables within each enumeration area. A further five houses were selected and reserved in each enumeration area for replacement purposes in case there were no household members to interview at any of the first 10 selected properties. On arriving at the selected properties, one household was randomly chosen for inclusion in the study in circumstances where there was multiple household occupancy.

The procedure was slightly modified in Epworth as it was difficult to identify enumeration area boundaries because most of the properties are informal and had no property numbers to use for randomisation. Five wards (Wards 1, 2, 4, 5 and 7) were selected with the help of the Epworth Local Board. The enumerators then divided each ward into 10 areas from which 10 houses were then randomly chosen.

Although there are many dimensions of poverty, the worst-case scenario is deprivation of food and essential or basic non-food items, respectively referred to as food and total consumption poverty. In Zimbabwe, poverty lines are set by ZIMSTAT, which are an expenditure level below which people or households are absolutely deprived. This article focuses on an income-based, consumption expenditure approach to poverty, although it recognises that the level of a family's income does not portray the whole picture of the well-being of the household.

2. Concepts of poverty

Poverty is one of the most serious challenges facing the world today, with more than 30% of the world's population estimated to be living in multidimensional poverty (UNDP, 2013). People in poverty are those who have benefited least from economic growth and development. Many of them live in remote rural areas or urban slums; have little access to productive assets; have low capabilities in terms of health, education and social capital; and suffer from chronic ill health or disabilities (CPRC, 2005).

In Shona, the concept of chronic poverty is captured in phrases such as nhamo yemadzinza (poverty passed down across generations) or nhamo inokandira mazai (poverty that lays eggs). Analytically many in Zimbabwe relate poverty to assets, which are often seen as indicators of wealth or an ability to avoid poverty. In many urban areas, employment, shelter and money define a household's wealth status (Manjengwa et al., 2012).

Poverty is multifaceted and there are many approaches to defining it. The most widely used are poverty lines, which measure deprivation of income (or the related expenditure and consumption) needed to meet basic needs for the maintenance of ‘physical efficiency’. The capabilities approach explains poverty in terms of what people are able to do and to become, rather than focusing on incomes (Sen, 1999). Poverty is then defined broadly as lack of capabilities rather than income and not its utility. This implies a focus on the welfare benefits of an income. For example, being educated, well-fed and free to exercise choice gives an individual a better living standard than being wealthy but in ill-health.

Widespread institutional exclusion and social exclusion on the basis of, for example, age, gender, ethnicity or disability represent formidable barriers for the efforts of the poorest to achieve security. The social exclusion approach conceptualises poverty as a state in which individuals are sidelined by societal structures from accessing resources (Ludi & Bird, 2007). This is taken further in the participatory approach, where poverty is defined as a state in which people have limited participation in the governance of their community (Chambers, 2006; de Campos Guimarães, 2009).

In contemporary poverty discourses this new generation of poverty definitions co-exists alongside the moneymetric understanding, and is seen as complementing rather than replacing the former (see Alkire & Foster, 2011).

3. The nature of urban poverty

The proportion of people worldwide living in urban areas passed the 50% mark for the first time in history in 2007 (UN-HABITAT, 2007). The pace of urbanisation in sub-Saharan Africa is twice the global average, making it the highest in the world. Urban populations in southern Africa are particularly rising rapidly, with South Africa and Botswana having urban populations of more than 60% (Crush & Frayne, 2010), and Zimbabwe at 33% (ZIMSTAT, 2013b). Unfortunately, rapid urbanisation in southern Africa is not associated with increased incomes and better standards of living as it is in some other developing regions (Ravillon et al., 2007).

The urban poor often live in overcrowded and unhygienic conditions, without sanitary facilities, clean water, solid waste collection or proper drainage because of the rapid and unplanned urbanisation (Tannerfeldt & Ljung, 2006). The high-density areas where the poor live are usually more prone to disease outbreaks and experience environmental hazard arising from density and exposure to multiple pollutants. For example, typhoid and cholera outbreaks have become predictable each rainy season in many of Harare's high-density suburbs in recent years (Joint Initiative, 2012).

Social fragmentation is another feature of urban communities due to lack of community and inter-household mechanisms for social safety nets. Many inhabitants of poor urban areas are lodgers or squatters, and tend to be transient, moving from area to area without putting down roots. Consequently their social capital is low. Lack of strong community values and social disintegration exacerbates crime and violence in urban areas, including rape, robbery, drug and alcohol abuse and prostitution. Another danger for urban dwellers is that of traffic accidents.

Understanding urban poverty presents a set of issues distinct from general poverty analysis and may require specific analysis (Baker & Schuler, 2004; Tannerfeldt & Ljung, 2006). Urban poverty differs from rural poverty, not only in its occurrence and depth but also in its nature, one of the most significant differences being that the urban poor depend on a cash income for survival. Urban economic contexts are cash based and the social fabric is not as strong compared with rural contexts (Joint Initiative, 2011). The urban poor face challenges such as high food prices, accommodation rent, cost of building materials, user fees for water and electricity, and associated debt. A study found that the average debt per urban household in Zimbabwe ranged from US$200 to 2500 (Joint Initiative, 2011). The situation is aggravated by high unemployment and under-employment and low economic activity rates. Since poor settlements are often on the periphery of the city, travel to work or facilities also incurs high costs. There are variations and significant inequalities between the better off and the poor within urban areas.

4. Poverty in Zimbabwe

There has been an increase in poverty in Zimbabwe since the onset of the economic structural adjustment in the early 1990s culminating in the economic crisis of 2004–8. National statistics did not capture the poverty situation during that period; however, by the time the crisis reached its peak in November 2008, it was estimated that up to 80% of the population was surviving on less than US%2 per day (UNDP, 2008; Chimhowu et al., 2010).

There was economic stabilisation and an upturn in the macro-economy in 2009 with the formation of the Government of National Unity, but benefits of this have been slow to reach people on the ground and poverty levels are still relatively high. Although poverty was far worse in rural areas than in urban areas of Zimbabwe, poverty in urban areas is increasing at a faster rate than in rural areas (Government of Zimbabwe, 2006). Urbanisation and development have not managed to eliminate poverty; on the contrary, in some cases it has even intensified.

The current prevalence of poverty in Zimbabwe has its roots in historical colonial policies consisting of a dualistic model of development where black men were taken to serve in urban industry and mines as cheap labour while women remained in rural areas (Chipika, 2007). Thus, poverty was systematically generated and sustained in the colonial development template in which men earned little in urban areas and women farmed below subsistence in rural areas.

The history of urban development in Zimbabwe is directly related to colonialism (Riddell, 1979). Racial inequality in urban areas was promoted by legislation such as the Industrial Conciliation Act of 1934 and its Amendment in 1959, which restricted black workers to mainly menial jobs. The 1974 study on consumption needs of urban families found wide discrepancies between black wages and the poverty datum line, with about 90% of black urban workers earning less than the poverty datum line for a family of six (Cubitt & Riddell, 1974). In addition to income poverty, basic services were lacking. The 1981 Commission of Inquiry into Incomes, Prices and Conditions of Service, chaired by Roger Riddell, noted that before independence in 1980 black townships, such as Harari (Mbare) and Highfield, were characterised by unclean water supplies and inadequate medical facilities (Government of Zimbabwe, 1981).

The relics of the colonial system continue to the present day despite attempts by government to redress the situation since independence in 1980. While the minimum wage, black affirmative action, indigenisation and income re-distribution policies in urban areas brought some improvements in the welfare of urban households, the original poverty template has generally remained intact.

Findings from previous studies on poverty in Zimbabwe point to the importance of income poverty as driving other dimensions of poverty. Participants in the Moving Zimbabwe Forward sample survey in 16 districts of Zimbabwe in 2011 identified that more money would solve most of their poverty-related problems, such as access to better healthcare, safe water, sanitation and quality education (Manjengwa et al., 2012). The Specialised Urban Poverty Study, which looked at multi-dimensions of poverty and deprivations, found that the level of consumption was more serious than the other dimensions and was the most significant factor in explaining deprivation intensity variation (Manjengwa et al., 2014). The study concluded that income poverty was the overwhelming problem affecting the well-being of the urban residents in the study (Manjengwa et al., 2014). Therefore, although we recognise that poverty is multidimensional and dynamic, the focus of this article is income-consumption poverty of a reference population.

5. Consumption expenditure poverty measurements for Highfield and Epworth suburbs

5.1 Total consumption and food poverty lines

Poverty lines are calculated by ZIMSTAT for each Province, using the prevailing prices of the different food and non-food items on the market that month. The Food Poverty Line, based on the national food basket of 17 food items, for the month March 2013 for Harare Metropolitan Province was US$35.00 per person (ZIMSTAT, 2013c). The food items in the national basket are considered adequate to supply 2100 kilocalories per person per day as specified by the World Health Organisation and consist of the most consumed foods, usually reflecting food preferences of the poor. A household whose individual members consumed food valued less than the poverty datum line (value) of US$35.00 in Epworth and Highfield was considered to be food poor, or in extreme poverty.

The value of total consumption expenditure for Harare Metropolitan Province for the same month, March 2013, was US$108.00 per person (ZIMSTAT, 2013c). Households whose individual members consumed food and non-food items valued below US$108.00 were considered to be poor because they failed to meet the minimum consumption expenditure requirements.

5.2 Household food and non-food consumption expenditure

The Specialised Urban Poverty Study used the already stated ZIMSTAT definitions to determine the money-metric measures for each household, using the data captured by the questionnaire on item by item (food and non-food) consumption.

Expenditures from own production, transfers and barter were valued using the respective suburb market prices. In cases where the household could not state a monetary value, calculations were done using the average value for the item in the suburb. Rentals for own accommodation were calculated using the average value per room within the respective enumeration area. Overall rent was then calculated by multiplying the number of rooms occupied by the average rentals per room in the area.

Each household's food consumption expenditures were computed for the month of March 2013, which was the reference period for the study (Equation 1). The total food consumption expenditure (FCE) for each household was divided by the respective number of household members to make the consumption expenditure comparable across households, as households had different sizes:(1)

where x_i is the individual household food consumption expenditure for the month and

n_i is the number of members in the household. The result is the per-capita food expenditure for each sample household.

The procedure assumes that food is equally distributed among household members, be they adults or children, and does not use weighting in recognition of the low calorific requirements for children because this is thought to be balanced by the additional micro-nutrients required by children (National Institute of Statistics, 2010; ZIMSTAT, 2013c).

Per-capita non-food consumption expenditure for each household was calculated using the same approach. The per-capita non-food consumption expenditure added to per-capita food expenditure gives per-capita total consumption expenditure for the month.

‘Poor’ households are classified as those whose consumption of food and non-food items falls below the Total Consumption Poverty Line but above the Food Poverty Line. Those whose total consumption falls below the Food Poverty Line are classified as ‘very poor’ or in extreme poverty (ZIMSTAT, 2011). The non-poor are households whose per-capita total consumption expenditure is above the Total Consumption Poverty Line.

5.3 Consumption expenditure poverty indices

Most studies, including this study, use Foster, Greer and Thorbecke (FGT) poverty indices to characterise ‘income’ poverty. The indices are P0, P1 and P2, also referred to as Head Count/poverty incidence, poverty gap/depth and poverty severity, respectively (Foster et al., 1984). These are defined as:(2)

where z is the poverty line, y_i is the income measure for the household, α is some form of sensitivity parameter, N is the number of people in the sample while H is the number of poor persons (whose incomes fall below the poverty line). When α = 0 this yields the headcount index. As α gets larger, the FGT measure gives more weight to the poorest.

P0 (Head Count or poverty incidence) is an index that attempts to give the breath or simply the incidence of poverty in the sample population. Conceptually the aim is to get individual consumption levels and find how many persons within the sample population have consumption levels below the minimum recommended. The P0 (α = 0) indicator is head count: the percentage of individuals estimated to be living in households with per-capita consumption below the poverty line for their province.

Reporting the Head Count index gives the number of people belonging to households in the different poverty categories. The Head Count shows how broad poverty is, although not necessarily how deep it is.

The P1 and P2 indicators are used to derive this information. The P1 (α = 1) indicator is the ‘poverty gap’ or depth. This is obtained by adding the difference between each individual's consumption and poverty line value. The total is divided by the poverty line value. One way to interpret P1 is that the value gives the per-capita cost of eradicating poverty, as a percentage of the poverty line, if money could be targeted perfectly. In practice, it is impossible to target the poor perfectly and issues such as administrative costs and incentive effects have to be taken into account if such a scheme is to be considered. The P1 measure gives an idea of the depth of poverty. However, it is limited because it is insensitive to how consumption is distributed among the poor.

To address this distribution, the P2 measure is used. The P2 (α = 2) indicator is the ‘squared poverty gap’ and gives an indication of poverty severity. The reason for squaring the shortfall is to give greater weight to those who are living far below the poverty datum line. Higher values of the indicator imply higher poverty. Using the index one can determine the sample with the poorest households that might need targeting.

5.3.1 Magnitude of poverty in Highfield and Epworth

The poverty incidence and indices in Highfield and Epworth are presented per overall sample and per suburb as well as per household, individual, child and gender of the household head (Table 1). Overall, 64% of households in the sample were below the Total Consumption Poverty Line; 55% were living in poverty and classified as ‘poor’, and almost 10% as ‘very poor’ or extremely poor (Table 1). This indicates that nearly one in every 10 households in the sample were very poor, and unable to meet their minimum food and non-food requirements.

Table 1. Poverty incidence and indices for the sample by suburb, individual and child.

Residence	Overall poverty (%) (includes the poor and very poor)	Poverty (%) (poor)	Extreme poverty (%) (very poor)	Poverty gap index	Poverty severity index
Households	64.4	54.9	9.5	0.27	0.14
Epworth	78.1	62.2	15.9	0.37	0.21
Highfield	50.8	47.6	3.2	0.17	0.08
All individuals	68.4	56.8	11.6	0.29	0.16
Epworth	81.8	62.1	19.7	0.4	0.24
Highfield	56.1	51.9	4.2	0.2	0.09
All children	72.0	59.2	12.8	0.31	0.17
Epworth	85.2	63.5	21.7	0.43	0.26
Highfield	59.0	55.0	4.0	0.16	0.09
Gender of household head
Male	60.4	51.6	8.8	0.27	0.14
Female	62.5	50.8	11.7	0.27	0.15

Source: Specialised Urban Poverty Study data.

Although not directly comparable due to methodological differences, the poverty levels in Epworth and Highfield obtained in the specialised study were relatively higher than indicated in national surveys. The Poverty Income Consumption and Expenditure Survey of 2011/12 found that the household poverty level for Harare was 35.7%; with extreme poverty at 3.3% (ZIMSTAT, 2013a).

Table 1 also shows that there was a high percentage (68%) of individuals who were poor, with 12% of them being extremely poor. Poorer households tend to have more members than non-poor households.

Although female-headed households had slightly lower levels of total consumption poverty than male-headed households, they had higher levels of extreme poverty (Table 1).

Generally, the sampled households and individuals in Epworth experienced more poverty and extreme poverty than those in Highfield (Table 1): 62% of households in Epworth were classified as ‘poor’ and 16% as ‘very poor’; whilst households in Highfield were comparatively better off, with 48% classified as poor and only 3% as very poor. Nearly 82% of individuals in Epworth and 56% of individuals in Highfield were living in poverty (Table 1).

As noted earlier, the poverty gap index is a measure of the intensity of poverty and gives the average depth of the poor relative to acceptable levels. The greater the gap, the deeper the poverty. The sampled urban households had a poverty depth of 0.27 (Table 1). This means that if poverty is assigned equally to the sample, each household would have a total consumption poverty of 27% below the poverty line of US$108 (as set by ZIMSTAT for March 2013). This has implications on the resources necessary for facilitating poverty reduction in the two urban suburbs. Theoretically, the government needs to spend 27% of the poverty line each month on each poor household member in the two suburbs to raise them above the poverty line.

Poverty in Epworth was twice as deep as that in Highfield, because these two suburbs displayed a poverty gap of 0.37 and 0.17, respectively (Table 1). This suggests that twice the amount of resources would have to be spent in Epworth compared with that in Highfield to raise poor households above the Total Consumption Poverty Line threshold.

Poverty depth measures relative deprivation of the poor, but it is neutral between different degrees of deprivation. Poverty severity takes into account inequality among the poor and is sensitive to the intensity of poverty. Poverty severity was higher in Epworth (0.21), compared with 0.08 in Highfield (Table 1). Correlation tests revealed that there was a strong relationship between poverty and location, as well as a significant difference between poverty experienced in the two suburbs.

6. What is driving poverty in Highfield and Epworth?

Different characteristics of well-being such as social, economic and demographic characteristics may be correlated to total consumption expenditure. The variables, referred to as factors, drive poverty either directly or indirectly by influencing the relative access of households and individuals to key components of well-being. Regression analysis can be carried out to better understand whether the correlations are genuinely explanatory. The urban poverty study used multiple linear regression analysis. Results from the regression show whether a factor has significant or genuine influence on per-capita consumption expenditure.

Approximately 35 household and household member characteristics were tested for their influence on per-capita total consumption expenditure using the regression procedure. Different factors which were correlated to total consumption expenditure were selected and then correlated amongst themselves to identify those that were strongly linearly related. Correlated factors explain almost the same variability in the total consumption expenditure if used in the model. Therefore, excluding one factor of such pairs removes redundancy.

The regression model used was of the form:(3)

where Y_i is household monthly consumption expenditure, β is a vector of coefficients, and X_i is a vector of all of the factors measuring household living standards, household head's level of education and other household characteristics. ?_i is an error factor. In this case, a regression coefficient is a conditional correlation that is the relation between total consumption per capita and, for instance, household size whilst holding other factors constant.

On demographic characteristics, factors such as household size, age of household head and gender of household head were considered. The age factor captures the ability to fend for the family. We also looked at social and economic factors such as educational attainment and assets. Level of education of the household head was also used where levels – namely none, Grade 7, Form 4, Form 6, diploma, graduate and postgraduate – were considered. Permanent employment was found to be the predictor of total consumption expenditure.

Regression models were run for each suburb and results show that poverty was largely driven by similar factors, but there were some specific factors peculiar to either Highfield or Epworth. Table 2 presents the top 10 out of 35 determinants of household monthly consumption expenditures, for both suburbs and individual suburbs.

Table 2. Determinants of household monthly consumption expenditures.

Determinant factor	Both suburbs coefficient	Epworth coefficient	Highfield coefficient
Log household size	−0.62**	−0.61**	−0.64**
Tenant	0.22**	0.16	0.22**
Head residency (five-plus years)	0.14**	0.11*	0.17**
Head no formal education	−0.32**	−0.34**	−0.30**
Head Grade7	−0.12**	−0.09	−0.14**
Head diploma	0.46**	−0.03	0.50**
Head graduate	0.35**	0.55	0.26*
Relative cash transfer	0.15**	0.07	0.17**
Permanent employment	0.21**	0.21**	0.22**
Rent income	0.26**	0.27**	0.26**
Constant	5.26	5.13	5.61

Source: Specialised Urban Poverty Study data.

Notes: *Significant at 10%; **Significant at 5%.

6.1 Poverty and household size

Household size significantly influenced poverty levels in both suburbs. Regression results show that a unit increase in household size causes approximately 62% decline in per-capita household consumption expenditure (Table 2) if other factors are held constant. By implication this means that the share of consumption expenditure per capita decreases with more household members. Large households are therefore more likely to fall into poverty compared with small households. Furthermore, survey results show that household size has a higher influence on per-capita household consumption expenditure compared with other household and individual characteristics included in the model.

Therefore, having more people in a household does not necessarily translate to more opportunities for the household to earn income. The survey found that only about 37% of households had at least a member permanently employed. Many industries have closed down and there is a serious deficit of decent jobs in urban areas of Zimbabwe. Section 6.4 indicates that temporary employment and business enterprises, which are predominantly petty trading, do not convey any significant advantage on the economic well-being of the household. Most of the informal employment activities are of a subsistence nature, such as selling mobile telephone airtime, sweets, cigarettes and vegetables. In addition, larger households tend to have more children as dependants. Larger households imply more mouths to feed.

6.2 Poverty and household head education

Other demographic characteristics, such as household head education, seem to also explain why household per-capita expenditure varies. Households with heads with no formal education or who have attained only Grade 7 of primary school tend to experience more poverty. According to regression results, the non-attendance of formal education by the household head increases the likelihood of a household falling into poverty, indicated by the negative coefficient in Table 2. On the other hand, the attainment of a diploma or degree by the household head increases the level of per-capita consumption expenditure. This may be because a higher educational level increases the likelihood of getting quality employment with better remuneration, which enables households to buy food and other household requirements. In addition, skills obtained through a diploma programme can be used for self-employment and informal artisanal income-generating activities, such as mechanics, carpentry, bricklaying and information and communication technology. In the scenario of scarcity of formal-sector jobs, education allows people to be innovative and to identify and exploit informal sector opportunities. For example, mobile phone and computer accessories and repairs can be accessed on street corners in commercial centres.

6.3 Poverty and length of residence

Households with heads who had been resident in the two suburbs for more than five years seemed to have better well-being (Table 2). Staying in an area for a prolonged period means one has knowledge of different livelihood opportunities and may utilise them to improve well-being.

6.4 Poverty and economic activity

Apart from demographic characteristics, household economic activities such as permanent employment and income from rentals positively influenced per-capita consumption expenditure of a household (Table 2). Permanent employment is a stable income source which enabled households to plan and access food and other commodities, and was the second important source of income after business enterprises. Results thus show that permanent employment improves household consumption expenditure.

Temporary employment tends to have no significant relationship with the well-being status of households. Results show that there is no relationship between those who had temporary employment and poverty (Table 3).

Table 3. Economic activities and household poverty category.

	Poverty category (%)
Household characteristic	Non poor	Poor	Very poor	Total households	p value
Income from permanent employment	46.3	33.3	20.8	36.8	<0.001
Income from temporary employment	26.4	28.3	33.3	28.0	0.402
Income from business enterprise (petty trade)	42.8	40.4	31.3	40.4	0.123
Access to cultivatable land	33.9	32.5	21.9	32.0	0.075
Owns a motor vehicle	19.2	5.5	3.3	10.2	<0.001

Source: Specialised Urban Poverty Study data.

Furthermore, having a business enterprise did not confer any particular advantage on a household's well-being (Table 3). Most businesses were actually small-scale subsistence activities and characterised as petty trade rather than profitable enterprises. The majority of businesses in the sample were vending items of low value such as vegetables, food and clothes.

Having income from rentals positively influenced per-capita consumption expenditure because rentals are sustainable sources of income, and as such significantly improve household consumption levels. A unit increase of income from rentals increased consumption expenditure by approximately 26% for the sample (Table 2).

An attempt was made to determine whether tenure status affected the level of household consumption expenditure. Tenants are households who officially rent a dwelling and can sublet part thereof. They therefore get income through subletting, which in the sample households significantly improved their ability to spend on food and other household essentials. Results show that being a tenant improved household consumption expenditure (Table 2).

Having access to land was associated with improved household's well-being. Families with access to land could grow food crops to complement and improve their families’ nutrition as well as sell surplus products. Ownership of a motor vehicle also had a strong relationship with poverty, with nearly 20% of the non-poor owning vehicles, compared with 5.5% of the poor and 3% of very poor households (Table 3).

6.5 Poverty and cash transfers

Cash transfers and receiving remittances are important coping strategies for poor households. The urban poverty study looked at receipt of cash transfers from the government, non-governmental organisations and relatives (Table 4). Generally, receipt of cash transfers was low for all households. A higher proportion of households in the different poverty categories got their cash transfers from relatives rather than from the government and non-governmental organisations.

Table 4. Receipt of cash transfers by poverty categories.

	Poverty category
Origin of cash transfer	Non poor	Poor	Very poor	Total households	p value
Government	3.1	2.4	0	2.4	0.217
Non-governmental organisation	1.1	1.4	1.0	1.3	0.887
Relatives	11.1	6.0	8.3	8.0	0.007

Source: Specialised Urban Poverty Study data.

Social capital, represented in the study by cash transfers from relatives, was thus a significant factor influencing household consumption expenditure levels. A unit increase in cash transfer from relatives led to a 15% increase in the level of consumption (Table 2). The influence of cash transfers on consumption expenditure was higher in Highfield compared with Epworth. Cash transfers from relatives were among the top five income sources in Highfield.

7. Conclusion

Although the urban situation is usually regarded as being better than the rural situation, the Specialised Urban Poverty Study in two suburbs of Harare has shown that the picture is more complex, and poverty levels are higher than expected. The Specialised Urban Poverty Study has disaggregated further than the national Poverty, Income, Consumption and Expenditure Survey and vertically analysed the patterns of poverty in selected urban communities. Two out of every three households in the sample were living below the Total Consumption Poverty Line.

The Specialised Urban Poverty Study revealed significant differences between the two selected suburbs. Poverty was more intense in Epworth than in Highfield. In the case of Epworth four out of five households lived in poverty, and of these approximately one out of five lived in extreme poverty. The depth and severity of poverty in Epworth were also relatively high and of concern, indicating that substantial amounts of money would be needed to bring all of the households out of poverty.

The results of the Specialised Urban Poverty Study confirm that despite all the efforts to improve the urban situation, nothing much has changed, the majority of the urban residents are poor and the original structural poverty template remains intact. Old suburbs such as Highfield and old informal settlements such as Epworth have always been poor and structural poverty has become the norm. The situation has been exacerbated by the economic decline, which resulted in manufacturing company closures, massive job losses in the formal sector as well as an influx of people from rural areas due to successive droughts and crop failures.

This study provided a better understanding of the nature and determinants of urban poverty, which can assist in providing solutions to address and improve the urban situation. The regression analysis found that poverty is driven by a number of demographic, economic and social factors, namely: large family size; low education level of the household head; lack of income from permanent employment; rentals and cash transfers; and short length of residence in the suburb.

Employment is a critical priority option in urban areas as indicated by better well-being for those with permanent employment. There is a need for a drive towards resuscitating industries to improve quality employment opportunities. Pro-poor macro-economic policies that create quality jobs should continue to be vigorously pursued. The informal sector should continuously be enhanced through grants, incentives and technology, and should be capacitated to have the requisite skills, with manpower training for technical and vocational skills. Relevant training and affordable loans are needed for start-up capital for business enterprises to improve their value and impact.

Another important issue is that of cash transfers from government and non-governmental organisations, which according to the urban poverty study appear to be negligible. Research in other countries has shown that giving cash to many poor families, rather than targeting just few extremely poor families, can have significant developmental impact (Hanlon et al., 2010). There is a need to rapidly scale up the coverage of the harmonised cash transfer programme to cover all districts in the country, as well as increase the value amount of transfer payments, especially to poor urban households, who have a myriad of cash expenses, as well as to widen the targeting to include all poor and vulnerable households, rather than just child-headed or widow-headed households, because of the extent and complexity of urban poverty.

Acknowledgements

This article draws on research undertaken by the authors during the Specialised Study on Urban Poverty in Highfield and Epworth High Density Suburbs, Harare Metropolitan Province, Zimbabwe, implemented by the Institute of Environmental Studies of the University of Zimbabwe, in collaboration with UNICEF, Zimbabwe Country Office and government Ministries, including the Ministry of Finance and Economic Development. The authors would like to acknowledge the support from UNICEF, Zimbabwe Country Office.

Disclosure statement

No potential conflict of interest was reported by the authors.

Additional information

Funding

This study was supported by UNICEF, Zimbabwe Country Office [Contract number 43135128].