Of all government services, housing is the one most difficult to provide. It cuts across all spheres of government (national, provincial and local). Its provision is very closely tied to the provision of basic services (water, electricity and sanitation). Most crucially, stripped of government targets and social goals, housing affects people on an emotional and psychological level in the most profound way.

Even the definitions of what constitutes ‘housing’, ‘settlements’ and ‘informal’ are highly contested (and sparked robust debate throughout the day). The packed crowd at the workshop was a mix of government officials, researchers, consultants and number nerds, all clearly passionate about housing.

Defining ‘informal settlements’ really was one of the goals of the workshop according to its official title: South African Informal Settlement Definition, Profiling and Trends: Comparison of new Census Data & Local Level Settlements Enumeration Data. The first half of the day focused on official housing data, particularly for informal housing, and the lessons that could be learned from the numbers. This data was compared with the department’s own research into informal settlements.

Research into the data (and the organisation of the workshop) was done by the Housing Development Agency (HDA), a public development agency within the DHS. The HDA began in 2009 and has two tasks. One is to fast-track the acquisition and release of state, private and communally-owned land for human settlement developments. The other is to provide a range of housing services including project management, planning and support services.

The HDA worked with Eighty20 on the analysis of official housing data and the primary data collection and analysis of informal settlements in Limpopo. The two presented the findings of a very extensive piece of research, rich in numbers and insights.

The official data analysed include the 2001 and 2011 censuses, the 2007 community survey (CS), income and expenditure surveys (IESs) and general household surveys (GHSs), all produced by StatsSA. Other data sources include Eskom (for measuring electricity provision to informal settlements). The primary data produced by the DHS was collected on-site from seven settlements in Limpopo.

The data from StatsSA are disaggregated to the household level, but the full data set for the 2011 census hasn’t been released yet. This isn’t the only limitation to the data. The IES and GHS data can’t be disaggregated below the enumerated area (EA) level. EAs are fairly big areas, and few are complete formal or informal, making them poor proxies for analysis.

Many of the questionnaires for the surveys are not specifically designed for informal settlements. There are longitudinal issues (questions change over time, settlements can change dramatically over time). In general, new settlements and settlement growth occur far too quickly to be properly captured by the official statistics.

EAs are one potential proxy. The ‘dwelling types’ field is another, but the different options might not accurately cover an ‘informal settlement’.

There doesn’t seem to be a strict definition of an ‘informal settlement’ anyway. Most of the official definitions highlight the type of dwelling and the land tenure. The StatsSA definition includes references to an ‘unplanned settlement…which has not been approved’.

In general, other official definitions normally have some association with at least one of the following themes: illegality, informality, vulnerability or ‘some form of social stress’. The City of Johannesburg’s definition contains some vivid snippets: ‘…impoverished group of households…illegally or without approval…created a shanty town…out of scrap metal…’

The City of Tshwane’s definition is more technical, speaking about ‘…one shack or more constructed on land without consent of the owner’. The problem is that even the definition of what constitutes a ‘shack’ can be debated.

The Census data differentiates between ‘shacks in back yards’ (SIBYs) and ‘shacks not in back yards’ (SNIBYs). The first type of shack is defined as a formal dwelling, since it is built on formally proclaimed land, the second type is informal.

The researchers used certain EAs as proxies for informal dwellings and combined these with the data on SNIBYs to measure trends in informal dwellings. This approach was a finesse of the challenges thrown up by the definitions and the data. It also led to one or two confusing outcomes: the data show a decline in the number of households living in informal EAs but an increase in the number of SNIBYs.

For all of the challenges that the StatsSA data sets pose, there are useful insights found in them. It is known that the growth in household numbers is significantly higher than growth in the population. There has been a dramatic increase in the proportion of households that contain just one person, and this phenomenon is driving the overall trend of declining household sizes.

Here’s where things become complicated: according to official definition, a ‘household’ is defined as a person or people ‘sharing financial resources’ whether the members live under the same roof most of the time or some of the time. A person living by himself, working as a migrant and sending remittances home could qualify as a member of two different households.

Migration trends show that 19% of the population moved between 2001 and 2011: two-thirds of these moved within their province, one-fifth moved across provinces and one-tenth immigrated from outside of South Africa.

Once again, the official data don’t analyse migration patterns down to the municipal level, nor do they focus on the type of migrants, i.e. why people choose to migrate. This type of information could be very important to municipalities and other government departments. If most economic migrants live by themselves, send remittances to their rural homes and do not intend to buy property where they live, municipalities might offer different services to them.

There’s a move towards formal dwellings. In 2011, 3.5 million more households were living in formal dwellings compared with 2001. This has been coupled with a significant decline in ‘traditional dwellings’.

There’s also been a fall, over the same period, in the proportion of people that own their own dwellings. This fall is seen across all types of dwellings. The shift has been into the rental market, and the growth in formal dwelling rental is noticeable in the big metro municipalities. There has been a noticeable contraction in mortgage lending between 2007 and 2011 and the rental market will remain important over the medium term.

As much as 8% of all South Africans live in informal residential EAs. Their migration patterns are similar to patterns in other living areas. The majority of people move within their province, although most of the migration in the Western Cape into informal EAs is from the Eastern Cape. The second-largest source of immigrants for most provinces is outside of South Africa.

Female-headed households tend to be larger: nuclear families and one-person households are headed by men; extended families tend to be led by women.

The proportion of SNIBYs that are one-person households sending maintenance/ remittances home is higher than the national average. These households are very connected to other households due to the financial support they provide.

About 984,000 children live in informal EAs, 495,000 are between 7 and 17. Overall school attendance rates for younger children in informal EAs are similar to the overall average, but there is a noticeable drop-off in attendance as children get older.

Labour force participation rates in informal residential EAs are high, but unemployment rates are also high. The participation rates in informal EAs are even higher than in formal EAs. Questions about the length of residence by people in informal EAs and whether they return ‘home’ can’t be answered yet, but will hopefully be addressed in the future. A lower proportion of employed adults in informal EAs is employed in the formal sector compared to national average.

More data on the sources of employment (sector and specific employer) in informal areas was a common request during the workshop, but this is not available with the current official data sets.

Data on monthly incomes is highly problematic. Income is understated (in common with most income analysis across the world) and there are big discrepancies between the census and IES data. Per capita income can provide more nuance on the wellbeing of a household. A total of 9% of households have a per capital income of under R5 per day according to the data, but it is not clear whether these numbers measure imputed income i.e. unpaid household labour including subsistence farming.

During the question-and-answer time for this portion of the workshop the general dissatisfaction with the official statistics became clear. Many of the audience members asked questions that pinpointed the limitations of the surveys. These questions covered households’ willingness to pay for basic services, details of income transfers from one household to another and the workings of the informal housing market (i.e. the unofficial, unregistered transfer of RDP houses).

To be fair, these are notoriously difficult questions to answer: willingness-to-pay studies are contentious, income is unreported and under-reported, and the concept of property ownership becomes increasingly fluid when sales are conducted under the radar or with the agency of traditional authorities.

The analysis of the official data was important for at least two reasons. Firstly, there are some important trends that can be teased out of the numbers, particularly the strong growth of single-person households involved in economic activities. Secondly, the department has comprehensively identified the questions it would like to answer, and this could provide valuable feedback to StatsSA for its future design of surveys.

The primary data-gathering by the DHS differed from the StatsSA data in many ways. It specifically targeted informal settlements and was focused only on the themes of housing and development. The survey was administered by community-based enumerators using smart phones (Android-operated).

The full questionnaire includes household-level and individual questions. In addition to gathering data, the enumerators logged GPS coordinates for each survey and took photos of each structure surveyed. These pictures were used as corroborating evidence (for example, data on household incomes could be compared with pictures of housing structures).

The data was used to create a detailed database and a baseline profile (on the household and individual level) of informal settlements. Information on the use of structures (as homes, businesses, mixed use and other) was also gleaned from the responses and photographs.

A comparison of the census data with the collected data drives home the point that the EA classification is not a reliable indicator. For example, in one Limpopo informal settlement there are three EAs that describe a mix of formal and informal settlements, but the collected data suggests the whole area is informal. The number of households in the EAs is also at odds with the households counted in the DHS survey. Some of the discrepancies can be due to the difference in definitions.

The Limpopo study highlights the inadequacy of the dwelling-based definitions. The question was asked again by the presenters: what is a shack? The slide show of various dwellings in Limpopo showed that ‘shacks’ all look very different and some of them don’t fit the official definitions of ‘informal settlements’.

The department was able to gather detailed information on the back of the three-year relationship that the HDA has built up with the Limpopo provincial government. The survey covers seven informal settlements spread over five municipalities, mainly in the southern half of the province:

–       Smash Block (Thabazimbi). This settlement contains more than 5,000 households. Most of the residents used to stay in mining compounds in the area, and most households are the one-person, economic migrants described above. The structures are relatively smaller than in the other settlements.

–       Extension 6 Jacaranda (Modimolle). This settlement is five years old. It is more formal and less congested than many other settlements. The average size of stands is between 300 and 400 square metres and formal township establishment has already commenced.

–       Motetema (Elias Motsoaledi). The settlement is over five years old. Most of the households consist of families. Many residents are building their own houses andnot relying on government services,.

–       Mohlakaneng  (Polokwane, in Seshego).

–       Praktiseer Extension 2 / Paktiseer Extension 3/ Tubatse A  (all in Greater Tubatse). Tubatse A is a very old settlement. It has second-generation inhabitants. Extension 2 is part of the overflow of Tubatse A, and Extension 3 has continued this process. The settlement has an established social fabric and networks.

In addition to administering questionnaires, the department conducted focus group discussions and field work on issues relating to poverty and settlement patterns.

The settlements differ in size, ranging from 191 households to 5,380 households. About 2% to 6% of all occupied structures are used for business purposes, while vacant structures comprise between 11% and 30% of all structure.

Many structures have been around for over a decade, and the age of the structures and the settlements informs the nature of official upgrades, e.g. servicing of sites versus demolition of structures and building of RDP houses. Many of the dwellings have larger grounds than many formal properties.

In Extension 6 Jacaranda, Mohlakaneng and Smash Block the structures are largely made of tin or metal sheets. In the other four settlements the majority are more formal and permanent with walls made of brick and cement. The majority of households (except in Smash Block) say they own their dwelling, although few have proof of ownership. A total of 47% of people in Smash Block pay someone to rent the dwelling.

In Tubatse A the majority claim proof of ownership, but this doesn’t only mean possession of a title deed. Ownership can be defined as permission from the municipality in the form of a letter, or a deed of grant. When discussing an incremental upgrading process by the authorities, the official policy is to consider alternate forms of ownership other than a title deed.

The vast majority of those surveyed agree with the statement that ‘this place is our home’ and very few plan to move within a year or two. The majority of respondents claim that they are willing to make changes and upgrades to the property.

Pit latrines are the most common form of sanitation facility, and even in established settlements the vast majority are using VIPs. This speaks to the need for the servicing of land and the slow pace at which this happens (largely because official databases aren’t updated quickly enough);

The main source of drinking water varies widely. Tankers dominate in Jacaranda, while in Praktiseer Extension 3 & Mohlakaneng most people use public taps. Almost all households in Motetema have piped water. The majority of households do not have refuse removal services and have their own dump (or just dump anywhere). Some in Tubatse and Mohlakaneng have their refuse removal by contracted community members.

Households in Jacaranda, Mohlakaneng, Smash Block do not have electricity. In other settlements most households have access through pre-paid meters supplied by Eskom. (There is a sense from the researchers that Eskom doesn’t want to service privately-owned land with dubious tenure).

Almost every individual has a blanket, cell phone, a bed. Almost all in Mohlakaneng and Smash Block have paraffin stoves. Most in Praktiseer Extensions 2 & 3, Tubatse and Motetema have TVs, fridges, and electric stoves. Most households in Praktiseer Extension 2 & Tubatse have a VCR/DVD, M-Net/DSTV, microwave oven. These settlements also have the highest ownership of motor vehicles (20/21%). Very few households have a motor vehicle.

Most respondents cited bad roads, poor street lighting, air pollution, unauthorised rubbish dumps, crime, and noise as significant problems. Nevertheless, most are satisfied with the current position of their house.

Household sizes vary quite widely: Mohlakaneng and Smash Block have small average household sizes (1.6 and 1.4 respectively) compared to other settlements. The majority of households in Jacaranda, Praktiseer 2 & 3, and Tubatse have more than five people.

Smash Block has 54% of people who are salaried employees. Informal and domestic work is common in Jacaranda. Very few people work for themselves, own businesses or volunteer. Most people are unemployed. Most of the employed work for formal companies. Few people work for unregistered companies, although the percentage is quite high in Jacaranda, as is the percentage of residents involved in domestic work. In a high percentage (35% to 50%) of all households nobody is employed (Smash Block is the exception).

Most households in the settlements earn less than R1,400. Seventeen percent of households in Mohlakaneng earn no money. Over 30% of households in Smash Block earn between R5,000 and R7,500 per month. Most households earn income from grants, and rental income may be under-reported.

The use of formal financial products is low in some settlements. Most people have bank accounts, most belong to a burial society (but not in Smash Block). Most belong to a savings group or stokvel (but again, not in Smash Block or Jacaranda, and only a third belong in Mohlakaneng). Most do not pay instalments on goods, about 20% pay furniture stores and clothing stores. Between 28% (in Praktiseer 2, Smash Block and Tubatse) and 78% (Jacaranda) qualify for housing finance.

There are plans to extend the scope of the questionnaire to include more detailed questions about employers (sector, names of employers), location of landlords (inside or outside of the settlement) and generally more focus on spatial analysis, of both the settlements themselves and the location of each household within the settlement.

Even the initial research by the department confirms a few long-held truths. One is that there is no ‘one-size-fits-all’ approach to housing development. Some settlements have sprung up almost as dormitory towns (like Smash Block). Others have grown over generations and house families and communities.

There was a broad consensus within the workshop that new approaches were needed by government and that there needed to be a shift away from supply-side solutions (i.e. building ever more houses) towards solutions that focus on greater negotiation with communities.

The need to change the approach to ‘informality’ was also emphasised. Rather than focusing on the ‘formalisation’ or ‘eradication’ of settlements, authorities should emphasise the ‘upgrading’ of settlements. To some it might look like an exercise in semantics but to those in the field it represents a fundamental shift in approaches. It would also meet more in situ upgrades and less demolition of shacks and structures.

This will be harder for government. The world over, governments have always favoured building more houses. House-building is a measurable activity with tangible outcomes. It’s much harder to measure the progress made through hours of consultation with communities, also harder to take credit for successes based on negotiation with all parties.

That is probably too cynical a reading of the future. For one, the financial pressures of building more and more houses will eventually trump the desire to continue the business-as-usual building of houses. For another, the department is expending much effort in using modern spatial tools to design better settlements and to tackle the problem intelligently. It looks like it is moving away from building boxes on the hillside and towards the design of sustainable communities. That is welcome news to the rest of us. DM

Photo: Johannesburg, South Africa, 28 February 2013. (Greg Nicolson)