changes in standard of living among population groups

Population Studies CenterUniversity of Michigan

Institute for Social Research

Report 08-654 Revised July 2009

Population Studies Center

Barbara A. Anderson and Mosidi S. Nhlapo

Changes in Standard of Living among Population Groups in South Africa: 1998-2006

Research Report

Changes in Standard of Living Among Population Groups

in South Africa: 1998-2006

Barbara A. Anderson Mosidi S. Nhlapo

(barba@umich.edu) (mosidin@statssa.gov.za) University of Michigan Statistics South Africa

Population Studies Center Research Report 08-654

September 2008 (Revised July 2009)

Acknowledgments: This research was supported by Statistics South Africa. Some of the research upon which this paper is based was supported by the United States National Institute of Child Health and Human Development Grant HD41028 to the Population Studies Center, University of Michigan. Heston Phillips, Miriam Babita, John Romani, Marie Wentzel, Johan van Zyl and Mike de Klerk provided helpful comments.

Changes in Standard of Living among Population Groups in South Africa: 1998-2006 2

Abstract This paper examines changes in standard of living in South Africa 1998-2006 overall and for population groups. Aspects of material standard of living (type of housing, source of drinking water, type of sanitation, source of lighting, source of cooking and access to a telephone) are combined to define four standard of living groups. The data sources are large (18,000-30,000 households) nationally representative surveys. The percent of urban African households in the best standard of living group (with a middle class standard of living), increased from 14% to 22% between 1998 and 2006, indicating emergence of an urban African middle class. About 40% of Coloured households, 75% of Asian households and 85% of White households had a middle class standard of living throughout the period. The percent of rural Africans in the worst standard of living group (no sanitation or a bucket toilet, drinking water from a dam, pond or stream) declined from 10% to 6%. In addition, the percent of all households, as well as the percent of urban African and rural African households, in the worst category of each of the standard of living indicators (no sanitation or bucket toilet, drinking water from stream, pond or dam, more than 16 minutes from a telephone, candles as source of lighting) declined substantially between 1998 and 2006, showing substantial poverty alleviation. There is no evidence of deterioration in the material standard of living of White households. The percent of Coloured, African, rural African and urban African households who enjoyed a decent standard of living – clean drinking water and decent sanitation (flush or chemical toilet or VIP) – also increased. Especially striking is the increase from 13% to 21% of rural African households with a decent standard of living. Almost all White households and Asian households had a decent standard of living at all dates. Also, the share of all households with a middle class standard of living comprised by African households increased from 21% to 40%, indicating the increasing need for manufacturers and advertisers to direct their attention to African consumers. There is variability in standard of living by province. Gauteng and Western Cape have the highest standard of living, and rural African households in Eastern Cape and KwaZulu Natal have the worst standard of living. However, the standard of living of rural African households in Eastern Cape and KwaZulu Natal improved between 1998 and 2006, although the situation for rural Africans in these two provinces remains much worse than elsewhere in South Africa. These improvements in standard of living were driven by the increased availability of clean drinking water, a decline in the use of bucket toilets in urban areas, and construction of pit latrines in rural areas. The increasing number of VIP’s in rural African communities has contributed substantially to improvement in the standard of living of rural African households.

This paper examines changes 1998-2006 in the material standard of living of households in South Africa. The main aspects of standard of living examined are the source of drinking water and the type of sanitation used. Lighting source, cooking source, type of housing, and household possession of a telephone are also considered in defining sets of standard of living indicators. This paper looks at the standard of living of South African households as a whole, as well as of all African households, rural African households, urban African households, Coloured households, Asian households and White households.

The apartheid regime was designed to maximize the welfare of White South Africans. However, there were three other officially-recognized population groups, Asians, Coloured and Africans. Asians were mainly comprised of persons with origins in India, although Asians also included some people with origins in China and elsewhere in Asia. Coloured were a mixed race group originating 300 years ago from the descendents of Portuguese and Malay on the one hand and predominantly the members of one African ethnic group, the Khokhoi, on the other hand. Africans are members of Bantu-based African ethnic groups. Under apartheid, Africans were subject to the most onerous restrictions, including being barred from residence in cities until 1985, Coloured were subject to fewer restrictions and Asians to even fewer restrictions. The effects of the differences among these population groupings dating from the apartheid period are reflected in persistent differences in conditions of life of South Africans today.

From the founding of the new South Africa in 1994, there have been aims to improve the

standard of living of all South Africans, especially those in the worst circumstances, as well as to increase the overall rate of economic growth and foster the growth of a diverse, economically productive middle class. One concern related to movement toward the attainment of these goals has been high and often increasing inequality of income and other economic indicators. This high level of inequality has raised social justice concerns as well as fuelling the potential for violent expression of frustrated aspirations (c.f. Johnston and Bernstein, 2007).

Many studies have looked at the distribution of income in South Africa and have found increases in income inequality (c.f. Leibbrandt et al., 2006; Seekings and Nattrass, 2005; South Africa, Statistics South Africa, 2002). The Luxembourg Income Study (LIS) has examined the income distribution in many countries. Recently, the LIS has expanded the set of countries in which comparable measures of the income distribution and income inequality are calculated. However, except for Taiwan and Mexico, all of the LIS countries are in the more developed region of the world (Luxembourg Income Study, 2007; Atkinson, 2004). There are some limitations in the approach of studies such as the LIS for a country such as South Africa, with a substantial rural population many of whose households are lacking even basic sanitation. In this kind of situation, examination of basic aspects of standard of living, especially related to source of drinking water and type of sanitation, assumes a relevance that it would not have in Europe or North America. Whether a household has no sanitation or has a pit toilet can more effectively differentiate the level of living experienced by poor households than can differences in reported income.

Households in South Africa have a variety of income sources, including wages and

salaries, remittances from members who are temporary labour migrants and government grants. The diversity of sources of income of many households presents challenges for LIS-type studies. Even when income data are reported, there are still difficulties. In the 2001 South Africa Census, which asked about personal income, for 10% of individuals the income item was missing, and when reported income of persons is aggregated into household income, 25% of households ended up with zero reported total income (Ardlington et al., 2006: 823).

The standard of living that a household experiences is an important aspect of quality of life and plays a major role in the health and productivity of household members. Whether a household has clean drinking water or not is important for the health of household members. Also even households with clean drinking water differ in their quality of life related to drinking water; whether the drinking water is from a tap in the residence or from a public tap does not matter for household members’ health, but it has an impact on the household’s quality of life. Obtaining drinking water from a public tap can be safe and healthful, but a household whose drinking water is from a public tap definitely is not experiencing a middle class standard of living.

This analysis has three specific goals. One is to identify households that have a middle class standard of living. These households are the best candidates for participation in the modern sector of the South African economy, both as producers and as consumers. An increase in the proportion of households in this group is directly related to South African economic growth and the prospects for economic growth in the near future. The growth of such a middle class has also been linked to social, political and economic openness and stability (Moaddel, 1995: 290; Neupert, 1981).

A second goal is to identify those households that have an extremely low standard of

living. Such households live in absolute poverty, with such poor quality drinking water and sanitation that there is a substantial threat to the health of household members. Reduction of the percent of households in this group is a direct indicator of the extent of poverty alleviation.

A third goal is to identify those households that have a reasonable and safe standard of

living regarding the source of drinking water and type of sanitation. These households have clean drinking water (water from a tap or a water tanker) and decent sanitation (a flush or chemical toilet or a VIP). The source of drinking water and type of sanitation might not be the most desirable or convenient, i.e., a household might not have a water tap in the dwelling or a flush toilet in the dwelling, but the water source and sanitation source are safe. In 2006, 99% of urban African households had clean drinking water, but only 40% had a water tap in the dwelling, for most other households the water tap was on site (in the yard). It has been a goal of the South African government that all households attain a decent standard of living.

With these three goals in mind, we combine aspects of standard of living to identify four

groups. The top group has characteristics that allow members of those households to experience a middle class standard of living. The bottom group has an extremely poor standard of living and could be characterized as living in absolute poverty.

Among the questions we address are:

1) Has the proportion of African households with a middle class standard of living grown over time? How has this growth differed by province of South Africa?

2) Has the standard of living of White households deteriorated?

3) How has the situation of the worst-off portion of the population, especially rural African households, changed? Has the proportion of the population in the worst standard of living category or with the worst source of drinking water or sanitation increased, decreased or remained unchanged? How has this situation differed by province of South Africa?

4) How has the proportion of households with a decent standard of living changed over time? Has there been a substantial increase in the proportion of households which have a clean source of drinking water and a safe and healthy type of sanitation, a situation that contributes both to household health and to self-respect?

There have been many approaches to defining middle class households. Some have

used education or occupation as defining characteristics, and others have used household possessions, items purchased or income (Black Diamond 1, 2007; International Centre for Policy Studies, 2002; Johnston, 2004; Senauer and Goetz, 2003).

We do not use household possessions to identify households with a middle class

standard of living partially because, although some household possessions are asked about in the data sources used, the list is not consistent across surveys taken in different years. Also, whether a household possesses an item, such as a television, reflects not only the purchasing power of that household but also lifestyle preferences that could have little to do with whether the household is middle class.

There is no general agreement about what the criteria should be for defining a

household as middle class (Johnston, 2004). However, all of the definitions aim at identifying households with a secure living situation without serious concerns about shelter, water and sanitation, and further identifying households that have the potential to contribute to economic growth and to participate in modern markets. Our definition of the middle class, using standard of living indicators, incorporates aspects of a secure and desirable living situation.

Data Sources

The analysis in this paper is based on the 1998-1999 October Household Surveys, the 2000-2001 Labour Force Surveys and the 2002-2006 General Household Surveys. These are national representative surveys covering 18,000-30,000 households.1

The October Household Surveys and the General Household Surveys are broad social and demographic surveys. The Labour Force Surveys focus more specifically on economic activity and labour force participation, although some information about aspects of standard of living is collected

We use the 2000 Labour Force Survey because the October Household Surveys were last administered in 1999, and the General Household Surveys began in 2002. The 2001 Labour Force Survey was used rather than the 2001 South African Census because in the area of sanitation, the 2001 Census recorded whether the household used a flush toilet but did not include the location of the flush toilet, such as whether it was in the dwelling, on site or off site. Whether a flush toilet is in the dwelling is an important factor in determining whether a household has a middle class standard of living.

Whenever the questionnaires for surveys or censuses are constructed, there are

reasons to retain earlier questions with exactly and same wording and reasons to modify or omit

1 The 1995-1997 October Household Surveys are valuable data sources but have specific shortcomings which limit their comparability with later surveys. These issues are discussed further in the Data Appendix.

earlier questions. We chose data sources which included the questions that we needed to classify households into standard of living groups in an identical or near-identical form.2

These surveys collect data that refer to persons within each household as well as characteristics of entire households. This paper is based on data that refer to a household as a whole, such as the household’s main source of drinking water. The population group of the household is identified as the population group of the person designated as the head of the household. Throughout this paper, the percent distributions refer to distributions of households rather than to distributions of persons. In 2004, for example, 76.6% of all households were headed by an African, but 79.2% of all persons were Africans.

In 2002, 17% of all South African households and 38% of rural African households

included at least one member who was gone from the household as a temporary labour migrant (Posel and Casale, 2006: 353). These temporary labour migrants often were an important source of household income; in 2002 76% of rural African households with a member away as a temporary labour migrant received remittances (Posel and Casale, 2006: 354). Also in 2002, in 36% of rural African households with a migrant member, remittances were the main source of income (Posel and Casale, 2006: 358).

After 1998, the surveys used in this study only collected information about labour force

activity and income for household members who slept at the household at least four nights in the previous week, making the economic activity and income of household members who are temporarily absent impossible to determine.3

The lack of information about household members who are temporary labour migrants is another reason for concentrating on aspects of standard of living rather than on the income of household members to indicate household wellbeing.

Other scholars have used the October Household Surveys and the General Household Surveys to look at changes in standard of living, in poverty or in overall social welfare. Leibbrandt et al. (2006) looked at the distribution of several indicators of standard of living, such as housing type, water source, sanitation, use of electricity for lighting, and use of electricity for cooking, across population groups and across provinces, using data from the 1996 and 2001 South African censuses. They examined the urban population as a whole and the rural population as a whole. An earlier Statistics South Africa (2001) publication presented indicators of the standard of living, including source of water and sanitation, 1995-1999, for South African households as a whole. Romani and Anderson (2002) looked at some aspects of standard of living 1994-1999 for non-Africans, rural Africans, and urban Africans, using October Household Surveys. Bhorat, Naidoo, and van der Westhuizen (2006) examined non-income welfare 1993-2004, including access to services and assets, using the 1993 SALDRU Survey, the 1999 October Household Survey, and the 2004 General Household Survey.

2 There is a small change across surveys in the question about type of housing, with living in a retirement community added as a response category in later surveys. We think it is not problematic that we classify those who live in a retirement community as residing in modern housing. The Data Appendix discusses a problem with the responses to the sanitation question in the 1998 October Household Survey and how we address that issue. 3 See Posel (2003) for a discussion of the kinds of data collected on temporary labour migrants in South African national surveys and censuses.

Population Groups

The number of households overall and by population group in the 2006 General

Household Survey is shown in Table 1, taking 2006 as a recent example of the situation in a given year. Table 1 shows the percent distribution of households by population group in the survey and the percent of households by population group in the weighted sample. The sample is weighted in order to take into account the difference between the distribution in the sample and the distribution in the actual South African population in 2006. Weighted data, using the household weight, are used in all analyses in this paper.4

Table 1 shows the actual number of cases by population group in the 2006 General Household Survey as well as the unweighted and weighted distribution by population group. We show estimates in this paper for each population group. However, in most of the surveys, there were only 400-500 Asian households included. This means that some instability in the results for Asian households should be expected.

Table 1. Unweighted and weighted distribution of households by population group of the

household head in the 2006 General Household Survey

Group Number of Households

Percent Distribution in Sample

Percent by Population Group in Sample Weighted by Household Weight

African 21,721 77.6 77.5 Coloured 3,430 12.2 7.6 Asian 509 1.8 2.5 White 2,316 8.3 12.3 Unspecified 26 .1 Total 28,002 100.0 100.0 Figure 1 shows the percent of households in each of the four population groups and of

all South African households who resided in an urban place in each year 1998-2006. The overall percent urban increased modestly from 60% in 1998 to 65% in 2006. In every year, at least 80% of Coloured households and at least 90% of White households and of Asian households were urban.

About half of African households resided in an urban place at each date. There was a

change from a little less than half of all African households residing in urban places in 1998 to a little less than 60% of African households residing in urban places in 2006. The overwhelmingly urban residence of non-African households raises the percent of households in urban places in 2006 from 57% for all African households to 65% for all households, countering the common impression that data for all South African households are similar to data for all African households.

In the unweighted data for the 2006 General Household Survey, there were 10,413

urban African households and 11,308 rural African households surveyed. As shown in Figure 1, African households are about evenly divided between rural and urban location and, as we shall see, African rural and urban households differ markedly in almost all household characteristics.

4 When confidence intervals are calculated and tests of statistical significance are done, the household weights are scaled so that the weighted total number of households in the survey in the given year is the same as the actual number of households in the survey in that year.

Figure 1. Percent living in urban places

As shown in Figure 2, at every date over 92% of all rural households had an African

household head. Thus, almost all rural households in South Africa are African households.

93.9 92.9 92.9 92.9 93.3 93.1 92.9 96.095.8

1998 1999 2000 2001 2002 2003 2004 2005 2006

Figure 2. African households as a percent of all rural households

There were only 623 rural Coloured households, 23 rural Asian households and 312 rural White households in the unweighted data for the 2006 General Household Survey. There was a similarly small number of rural Coloured, rural Asian and rural White households in the surveys used for other years. We examine Coloured, Asian and White households without

division into those in urban areas and those in rural areas, both due to the relatively small number of households for these groups, especially for Asians, and also due to the overwhelmingly urban location of those population groups. We examine results for all African households, as well as for rural African households and urban African households separately.

In the rest of the paper, first we look at changes over time for population groups (African

households, rural African households, urban African households, Coloured households, Asian households and White households) in each separate aspect of standard of living that we consider: type of housing, access to a telephone, lighting source, cooking source, source of drinking water and type of sanitation. After that, we combine aspects of standard of living to define four standard of living groups. We look at the change over time in the percent of South African households and of the population groups in the four standard of living categories. We then examine some aspects of the changing standard of living distribution by province of South Africa and for population groups within provinces. After that, we look at the changing population group composition of each of the standard of living groups. For 2006, we look briefly at the total monthly expenditures of households in the middle class standard of living category. Finally, we reflect on evidence of poverty alleviation in South Africa between 1998 and 2006.

Type of Housing

A household’s type of housing matters for the household’s well-being. Housing types differ in their sturdiness and protection from the elements. Formal housing generally performs better in these areas than other types of housing.

A household is considered to live in formal housing if it lives in any of the following

situations: in a formal structure on a separate stand, in a flat in a block of flats, in a townhouse or semi-detached house. In some surveys living in a unit in a retirement village was listed as a separate category. When a retirement unit was listed as an option, it also was considered to be formal housing.

Figure 3 shows the percent of households residing in formal housing over time. In every

year White households or Asian households have the highest percent in formal housing, followed by Coloured households, with a lower percent for African households.

There is very little change over time in the percent of households in formal housing,

overall or by population group, except for a slight decline in the percent recorded as in formal housing after 2004. We do not think that the reported dip in the percent living in formal housing in 2005 and to a lesser extent in 2006, reflects a real change. Rather, we think that it is the result of a combination of random variability and perhaps some variation in interviewer training. The main reason we are looking at whether a household resides in formal housing is to use it as a part of the definition of standard of living groups. We think that for a household to have a middle class standard of living it needs to reside in formal housing.

Figure 3. Percent residing in formal housing

Telephone in Household

Access to a telephone provides a means of communication for household members for both personal and business purposes. It is better to have a public telephone available close to the home than far away, but only with a telephone in the home or in the possession of a household member is it possible to receive calls easily, without prearranging the time a call will be received.

Figure 4 shows the percent of households which had a landline phone or in which a

household member had a cell phone. The increase over time is impressive. The percent with a landline or with a household member with a cell phone more than doubled for South African households as a whole between 1998 and 2006, more than quadrupled for all Africans, and increased by almost twentyfold for rural Africans, yielding an annual rate of increase 1998-2006 for rural Africans of 37%. The only groups for which there was not a large increase were White households and Asian households; over 85% of White households and over 81% of Asian households had a landline or a cell phone in the household at all dates. After 1998, the percent of Asian households with a cell phone or a landline was noticeably lower than for White households.

Figure 4. Percent with a landline in dwelling or a household member with a cell phone

This increase in phones in African households is overwhelmingly due to the spread of

cell phones. Figure 5 shows the percent of households that had a landline telephone in the dwelling.5

In 2006, 8% of all African households had a landline phone, which means that in 58% of African households, although there was not a landline, a household member had a cell phone.

Before 1999 cell phones were rare. In 1998, 95% of all households that had either a landline or a cell phone had a landline; in 2000 only 69% of households with a landline or a cell phone had a landline. The spread of cell phones and the resulting improvement in telephone access in rural areas was the result of the success of commercial cell phone companies.

There are several reasons for the popularity of cell phones. Cell phones are workable in

areas in which a landline is not available, such as in many rural areas. This is probably one reason for the predominance of cell phones for African households, especially for rural African households. Also a cell phone can be purchased just by having the necessary money to pay the cost. Even without the credit necessary for a cell phone contract, pay-as you-go schemes allow purchase of cell phone time without presenting evidence of the purchaser’s financial situation. Even when a household wanted, could afford a landline, and lived in an area where landlines were available, there were widespread reports of long delays in landline installation (c.f. “Chaos as new magistrate’s courts left without phones,” 2003). Also, in many rural areas, landline service was not available. In 2006, while 60% of rural African households had a landline or a cell phone, less than 3% of rural African households had a landline. At no date did more than 5% of rural African households or more than 26% of urban African households have a landline. 5 No data are shown in Figure 5 for 1999 because in that year the Labour Force Survey only asked whether the given household had a landline or a cell phone and did not ask about landlines and cell phones separately.

Figure 5. Percent with a landline telephone in dwelling

Note that in Figure 5 the percent of households with a landline declined for all groups,

including White households and Asian households. Between 1998 and 2006, possession of a landline in the dwelling declined at an annual rate of 7% for all South African households and declined at an annual rate of 2% for White households. For rural African households, the presence of a landline increased slightly between 1998 and 2000 and then declined to below its 1998 level. Thus, some households that had a landline at one time let that landline go once at least one household member had a cell phone, whether due to the greater convenience of cell phones, the lower operating cost of cell phones or some other reason. The South African press has reported that many households have given up their landline phones due to the greater convenience and lower cost of cell phones. It has been reported that a major reason why some households have retained their landlines has been to gain fast internet access (“Telkom customers hang up,” 2007).

Figure 6 shows that the percent of households with the least access to a telephone

dropped substantially over time. Between 1998 and 2006, the percent of households that needed to travel more than 15-16 minutes to reach an accessible telephone declined by more than 51% for all South African households and by more than 57% for rural African households.6

This represents a substantial improvement in telephone access. The proliferation of cell phone points (commercial establishments where time on a cell phone can be purchased), including in rural areas, has almost certainly played a major role in this improved access.

6 The categories in the 1998 OHS and the 2006 GHS were slightly different. Thus, being 16+ minutes from a phone in 1998 is compared with being 15+ minutes from a phone in 2006.

All South Africans Rural Africans

1998 2006

Figure 6. Percent of all South Africans and rural Africans more than 15-16 minutes from an accessible telephone: 1998 and 2006

The Main Lighting Source – Electricity and Candles Electricity is the most desirable lighting source. With a secure connection and in the

absence of power outages, it is always available. Also, unlike sources such as candles, wood or kerosene, it does not add to air pollution in the vicinity of the household. Candles are the least desirable lighting source. In this section we look at the increase in use of electricity for lighting and the decrease in the use of candles for lighting in South Africa since 1998.

There has been a major effort to extend electrification to an ever increasing proportion of

South African households. In 1994, the South African government set a target to have 2.5 million households newly connected to an electricity source by 2000. The South African government also has targeted 2012 as the year by which all South African homes will have access to electricity (ESKOM, 2007).

Figure 7 shows the percent of households in which electricity is the main lighting source. It is clear that the percent of households with electricity as the main lighting source has risen considerably for all groups except for White households and for Asian households, in which over 98% of households used electricity as the main lighting source at all dates. For all African households, the value for 2006 is 138% that of 1998. For rural African households, the percent of households with electricity as the main lighting source in 2006 was 188% of the 1998 value. Since 2003, ESKOM made an increased effort to extend electrification to households in deep rural areas (ESKOM, 2007). By 2005, the percent of Coloured households with electricity as the main lighting source was nearly as high as for White households or Asian households.

South Africa recommended provision of free basic electricity to people in 2001. The

intention of this recommendation was to help poor households afford electricity (DME, 2007). This program has operated through municipalities and has not been implemented in all locales.

Figure 7. Percent with electricity as the main lighting source

Table 3 shows among households connected to MAINS electricity, the percent who stated

that they had free electricity in 2004. Free MAINS electricity has been available to a higher percent of Coloured households than to any other group, including urban African households.

Table 3. Among households connected to MAINS electricity in 2004, the percent who said they received free electricity

Africans Rural Africans

Urban Africans

Coloured Asian White All South Africans

22.7% 13.2% 29.9% 50.5% 21.8% 26.0% 25.7%

The greater provision of free electricity to Coloured households than to other population groups is related to the greater provision of free MAINS electricity in Western Cape than in other provinces. In 2004, 21% of all South African households reported that they received free MAINS electricity. The highest provincial percent was for Western Cape at 60%, and the next highest was for Free State at 35%.

We next look at what has happened over time in the prevalence of the use of the least desirable lighting source, candles. Figure 8 shows the percent of households over time that used candles as the main lighting source.

Figure 8. Percent of households using candles as the main lighting source

Candles are most prevalent at all dates as the source of lighting among rural African households, for whom their use in 2006 was 64% of the 1998 value. However, even in 2006, over ¼ of rural African households still used candles as the main lighting source. There was some fluctuation in reported candle use among urban African households and Coloured households. However, at no date did the percent of urban African households using candles exceed 15% nor of Coloured households using candles exceed 11%.

Electricity or Gas as a Cooking Source

Electricity and gas are considered desirable cooking sources. With a reliable source, electricity and gas are dependable and non-polluting (in the case of electricity) or pollute at a very low level (in the case of gas).

Figure 9 shows the percent of households in which electricity or gas was the main

cooking source. Gas comprised a small portion of this use; in 2006, over 96% of the households for whom electricity or gas was the main cooking source used electricity as the main cooking source.

At all dates, the vast majority of White households and of Asian households (over 97%)

used electricity or gas as the main cooking source. Also, Coloured households increasingly used electricity or gas as the main cooking source, reaching 92% by 2005.

For Africans, the picture in Figure 9 is different from that in Figure 7. For all African households, between 1998 and 2006 use of electricity for lighting increased by 38%, but use of electricity or gas for cooking increased by 35%. For rural African households, the increase between 1998 and 2006 in electricity for lighting was 88%, but for use of electricity or gas for cooking was 68%.

Figure 9. Percent with electricity or gas as main cooking source

Table 4 shows for 2006, among households in which electricity was the main lighting

source, the percent for whom electricity or gas was the main cooking source. Almost all Coloured households, Asian households and White households that used electricity for lighting also used electricity or gas as the main cooking source (more than 96%). In contrast, among African households that used electricity as the main lighting source, only 73% used electricity or gas as the main cooking source, and among rural African households, only 50% of those that used electricity as the main lighting source used electricity or gas as the main cooking source.

Table 4. Among households with electricity as the main lighting source, the percent who use

electricity or gas as the main cooking source, 2006 General Household Survey Africans Rural Africans Urban Africans Coloured Asian White 73.0% 50.1% 87.2% 96.4% 99.7% 99.8%

There are three likely reasons for the gap between use of electricity for lighting and use of electricity or gas for cooking among African households. First, many African households use heat from the stove (using coal or wood) for heating in the winter, even if the household owns an electric or gas stove. Second, some African households with electricity do not own an electric or gas stove. Third, beyond a basic allotment in areas in which there is free electricity available, households pay for electricity.

In the formulation of the free basic electricity program, it is recommended that 50kWh

per month be made available to each participating household. This amount of electricity was estimated to be sufficient for “…basic lighting, small black and white TV, small radio, basic ironing and basic water boiling through an electric kettle for grid-connected consumers” (DME, 2007). Although a relatively small amount of electricity is used by lights, cooking requires much more electricity.

Among African households that used electricity as the main lighting source and who did not use electricity or gas as the main cooking source, 44% used wood for cooking, 39% used paraffin, 12% used coal and 5% used some other source of fuel (animal dung or other). Wood and coal stoves generate a substantial amount of heat, but paraffin stoves do not. Paraffin stoves were probably used due to economy considerations rather than for heating.

Interestingly, in 2005 among African households with electricity as the main lighting

source, but who did not use electricity or gas as the main cooking source, 42% reported they owned a gas or electric stove.7

For these households, the gas or electric stove is used sometimes, but not as the main cooking source, due to cost of electricity or gas, the need for the cooking source to supply heat in the winter or for both reasons. If all African households with electricity as the main lighting source who also owned an electric or gas stove stated that gas or electricity was their main cooking source, then in 2005 62% of African households would have reported they used electricity or gas as their main cooking source rather than 53%.

Clearly if a household owned a gas or electric stove, then gas or electricity was used for cooking at least part of the time. If more than one source is used for cooking, then it is somewhat subjective which source is identified as the main source. The fieldwork for the General Household Surveys occurs in July, in the middle of the winter. Households that use an electric or gas stove in summer but a wood or coal stove in winter might be more likely to state that they do not use electricity or gas as the main source of fuel for cooking when answering this question in the winter than they would if they answered this question in the summer.

Main Source of Drinking Water

Clean drinking water is important for overall health and plays a substantial role in infant and child health and survival (Anderson et al., 2002; Fewtrell et al., 2005; Ross et al., 1988). Persons with compromised immune systems, such as those with AIDS, are especially vulnerable to water-borne infections, even infections which are not typically serious for healthy individuals (Kgalushi, Smits and Eales, 2004; Laurent, 2005: 6).

The South African Constitution states that all South Africans have the right to a healthy

environment (Constitution of South Africa, Chap.2, Sec. 24), an important component of which is access to clean drinking water. Extending the availability of clean drinking water to all South Africans has been a major government goal (DWAF, 1994: 1; 2003a: 1). In addition, inequity among population groups in the quality of the sources of drinking water, especially in the availability of piped water, has been a concern from the founding of the Department of Water Affairs and Forestry (DWAF 1994: 3-4).

7 No question about possession of a gas or electric stove was asked in the 2006 GHS.

Part of the stated mission of the Department of Water Affairs and Forestry (DWAF) is “…ensuring that water services are provided to all South Africans in an efficient, cost-effective and sustainable manner” (DWAF, 2007). As of 2003, the goal was that all South Africans would have access to clean drinking water by 2008 (DWAF, 2003a: 6). Improving the source of drinking water available to households also has been a focus of the Millennium Development Goals, and, according to the United Nations, the source of drinking water is an integral part of defining whether a household lives in absolute poverty (UN, 2000: paragraph 18, chapter 2).

Table 5 shows a classification into five categories of source of drinking water, using

items available in the surveys. Tap water and water from a water tanker are considered clean. Of course, the water that comes from a tap is not necessarily clean, and the water from a well or a stream could be clean. The deaths of 123 children in Eastern Cape from diarrhoea from households that had access to “clean” water in the first half of 2008 is a reminder that just because water is from a tap does not guarantee it is safe, whether the problem stems from improper or inadequate chlorination of the water supply, from unsanitary storage of the water in the dwelling or from some other cause (“Killing water,” 2008).

Table 5. Classification of types of drinking water source according to the categories

in the surveys used Tap in Dwelling Water tap in dwelling Tap on Site Water tap on site, such as in the yard Other Clean water Neighbour’s tap, public tap, water tanker Borehole Well Borehole, well, rainwater tank Stream Dam Pond Stream, dam, pond, river, pool, stagnant water, other

The classification employed here is that which has typically been used in South Africa to

classify drinking water sources as clean or not. This classification is also consistent with the DWAF view of the water ladder, which involves a target first of providing households with clean drinking water and then moving to provide an increasing proportion of households with a more desirable source of drinking water, such as a tap on site (DWAF, 2003a).

The perceptions by South Africans of the cleanliness of various water sources are

consistent with this classification. Table 6 shows for 2005 the percent of South African households who view their main source of drinking water as: (1) safe to drink, (2) clear (no colour or mud), (3) good in taste, and (4) free from bad smells, according to the five types of drinking water source used in this study.

Table 6. Perceptions of characteristics of main source of drinking water according to category of drinking water source, 2005

Tap in Dwelling

Tap on Site

Other Clean Water

Borehole Well

Stream Dam Other

% water safe to drink 98% 97% 95% 73% 30% % water clear 97% 96% 94% 75% 36% % water good in taste 97% 96% 91% 67% 36% % water free from bad smells 97% 96% 93% 73% 39%

The sources of water considered clean in this paper are consistently more likely to be viewed as clean by survey respondents than are those sources viewed as not clean. In fact, the ordering of sources of drinking water in Table 5 corresponds completely with the ordering of the extent to which people perceive water as safe to drink, clear, good in taste and free of bad smells shown in Table 6.

Figure 10 shows the distribution of sources of drinking water for all South African

households by year 1998-2006. In Figures 10-16 the three categories of clean water are indicated by red patterns, and the two not clean categories of sources of drinking water are indicated by light solid green and by dark solid green. For each year, in each graph, the percent of households with a given source of drinking water is shown, and all sources of drinking water add to 100%.

In Figure 10, it is clear that use of the worst source of drinking water (stream dam other)

has become less common over time, declining from 12% in 1998 to 6% in 2006. Also the percent of all South African households which used a clean source of drinking water has increased, from 83% in 1998 to 89% in 2006. This was mainly due to an increase in the percent of households with a tap on site, which increased by 30% between 1998 and 2006.

Figure 10. Drinking water source for all South African households

Figure 11 shows that for African households there has been a decline in the percent of

households with the worst source of drinking water (from 16% to 7%) and an increase in the percent with clean water (from 78% to 87%).

Figure 11. Drinking water source for African households

Figures 12 and 13 show the source of drinking water for rural African households and for

urban African households, respectively. Figure 12 shows a decline in the percent with the worst water source from 29% in 1998 to 17% in 2006 – a decline of 43%. The decline in the worst water situation was due to an increase of 13% in other clean water (mainly a public tap) and to an increase of 25% in tap on site. At all dates less than 6 percent of rural African households had a tap in dwelling. Thus, although there was minimal improvement in the availability of a tap in dwelling – a component of a middle class standard of living – there was a substantial improvement in the availability of a healthful drinking water source.

Figure 12. Drinking water source for rural African households

As shown in Figure 13, almost all urban African households had clean drinking water at all

dates. However, between 1998 and 2006, there was almost no change in the percent those who had a tap in the dwelling. In 2006, although 84% of urban African households had a tap either in their dwelling or on site, more than half of those households had the tap on site (in the yard).

Figure 13. Drinking water source for urban African households

Figure 14 shows the situation for Coloured households. Almost all Coloured households had

clean water at all dates, and there was little change in the percent with a water tap in the dwelling, being 70-77% in every year. It is notable that the sources of drinking water were considerably better for Coloured households than for urban African households – in 2006, 76% of Coloured households but only 40% of urban African households had a water tap in their dwelling.

Figure 14. Drinking water source for Coloured households

Figure 15 shows the sources of drinking water for Asian households, and Figure 16

shows similar information for White households. At every date, over 89% of Asian households and over 95% of White households had a tap in the dwelling.

Figure 15. Drinking water source for Asian households

Figure 16. Drinking water source for White households

Figure 17 summarizes the percent of households with a tap in dwelling over time by group. There was little change for any group.

Figure 17. Percent of households with water tap in dwelling

With 78% of South African households headed by an African in 2006, it is common to

interpret results for all South African households as indicative of the situation for African households. However, if the situation for non-African households is vastly different than that for African households, this generalization does not hold. As shown in Figure 17, in 2006, 39% of all South African households had a tap in dwelling, while this was true for only 25% of all African households. This difference is because the percent with a tap in dwelling was much higher for non-African households – 76% for Coloured households, 90% for Asian households and 94% for White households.

Figure 18 shows the percent of households that had a tap either in the dwelling or on the

site. This percent has increased over time for every group except for White households and Asian households, almost all of which had a tap in the dwelling or on the site at every date. Although a tap on site is less convenient that a tap in the dwelling, it is much more convenient than using a public tap, which involves going farther to obtain water and sharing that tap with other households.

Figure 19 shows the percent across time with clean drinking water by group. The percent of all South African households with clean drinking water increased from 83% in 1998 to 89% in 2006. The percent of rural African households with clean water increased from 59% in 1998 to 70% in 2006, an annual rate of increase of 2.2%. However, in 2006, more than 30% of rural African households still did not have clean drinking water. If the rate of increase in the percent of rural households obtaining clean drinking water that held in 1998-2006 persisted, then all rural African households would have clean drinking water in 2023. Almost all White households, Asian households, Coloured households and urban African households had clean drinking water at all dates.

Figure 18. Percent of households with water tap in dwelling or on site

Figure 19. Percent of households with clean drinking water

Figure 20 shows the percent of households over time in the worst water situation, those whose drinking water source is a stream, dam or other source. The scale has been changed from 0-100% in Figures 18 and 19 to 0-30% in Figure 20 in order to make the changes over

time in Figure 20 clearer. Less than 2% of urban African, Coloured, White and Asian households used the worst category of drinking water in any year 1998-2006, while in every year the percent of rural African households with the worst source of drinking water exceeded 16%. However, the decrease by more than 40% in the percent of rural African households with the worst drinking water source is impressive and signifies a substantial improvement in the quality of life and the health prospects for rural Africans.

Figure 20. Percent of households with drinking water from stream dam or other source (worst

drinking water category)

Close to 100% of urban African, Coloured, White and Asian households had clean

drinking water in every year since 1998, and by 2006 70% of rural African households had clean drinking water. However, in terms of a desirable standard of living, all households would aspire to having a tap in the dwelling. No substantial progress was made in the percent of households with a tap in dwelling between 1998 and 2006 for any group considered, and there was substantial room for improvement in this area for every group except for White households and Asian households.

Type of Sanitation

Good sanitation is important for health and plays an especially important role in infant and child health and survival (Anderson et al., 2002; Burger and Esrey, 1995; Habicht, DaVanzo, and Butz, 1988; Lee, Rosenzweig, and Pitt, 1997). Contamination of drinking water due to poor disposal of human waste is a major cause of the spread of water-borne infectious diseases (Carr, 2001). The importance of adequate sanitation and the risks when sanitation is not adequate were made clear from the 2001 cholera outbreak in KwaZulu-Natal (DWAF, 2001).

In addition to access to clean drinking water, another part of assuring that all South Africans live in a healthy environment is provision of adequate sanitation. Improving sanitation has also long been a South African governmental goal (DWAF, 1994: 6, 9).

There has been South African governmental interest in increasing the number of VIP’s (ventilated improved pit toilets). VIP’s are substantially better than ordinary pit toilets, having a lower odour level and providing greater protection from insects. They have been characterized by DWAF as an “appropriate and adequate basic level of sanitation service” (DWAF, 1994: 15). It was reaffirmed in 2003 (DWAF, 2003a: 46) that a VIP was an acceptable basic type of sanitation and Hanekom (2005) reiterated this view.

Austin (1996) argues that when properly maintained, VIP’s are as safe for public health

as more complex and expensive sewage systems. VIP’s need to be emptied every five years (DWAF, 2002), and although this schedule for emptying the VIP is not extremely frequent, there can be problems if they are not emptied on schedule and subsequently overflow. Overflowing of waste counters many of the advantages of VIP’s over unventilated pit latrines. When a VIP is constructed on extremely steep terrain emptying can be very difficult (Hanekom, 2005). However, these are issues of proper maintenance and appropriate construction for the facility to operate appropriately, similar to the need to treat water regularly with chlorine or other chemicals in order to assure that the drinking water that comes from taps connected to a drinking water system actually is clean.

Bucket toilets remain in use in urban African households. However the South African

government has long considered bucket toilets unacceptable, and there have been efforts to eliminate their use (DWAF, 1994: 15). Based on the 2006 General Household Survey, in that year about 314 thousand households still used a bucket toilet, and about 850 thousand households had no access to sanitary facilities, either on their residential site or off that site.

Table 7 shows the classification of sanitation used in this paper according to the

categories of sanitation available in the surveys used. Having access to a flush or chemical toilet or a VIP is a decent sanitation situation in terms of health effects and keeping water free of contamination by human waste.

Table 7. Classification of type of sanitation according to the categories in the surveys used

Flush in Dwelling Flush toilet in dwelling, whether connected to sewer system or to septic tank

Flush on Site Flush toilet on site, whether connected to sewer system or to septic tank

Other Flush or Chemical

Flush toilet off site, or chemical toilet, whether the chemical toilet is in dwelling, on site or off site

VIP (Ventilated Improved Pit Toilet)

A pit latrine with special features that reduce odour and insects in the latrine, whether on or off site

Pit Toilet Unventilated pit latrine whether on site or off site Bucket Other None Bucket latrine whether on or off site, none

Figure 21 shows the percent distribution of type of sanitation for all South Africans over time.8

8 The distribution of all pit toilets between VIP’s and other pit toilets for 1998 is estimated based on data on type of sanitation for 1999. See the Data Appendix for an explanation.

In the figures the four categories comprising a decent sanitation situation (use of a flush or chemical toilet or a VIP) are indicated by red patterns, and the two other sanitation categories, other pit toilet and bucket other none, are indicated by light and by dark solid green.

As shown in Figure 21, there is a decline of 43% between 1998 and 2006 in the percent of households using the worst type of sanitation, bucket other none, an annual rate of decline of 6%. The percent of households using a VIP increased from 6% to 10%, almost doubling.

Bucket Other None 15.9 13.9 14.3 13.3 12.6 11.1 11.0 10.2 9.0

Other Pit Toilet 26.0 26.2 25.4 26.3 26.5 25.5 24.3 23.6 24.1

VIP 4.2 4.2 4.2 4.5 4.8 5.5 7.5 7.6 7.8

Other Flush or Chemical 1.8 1.6 1.2 1.1 0.3 1.1 1.2 1.9 1.7

Flush on Site 15.6 18.9 19.2 19.0 19.7 19.9 19.1 19.5 20.0

Flush in Dw elling 36.5 35.3 35.6 35.9 36.1 36.9 37.0 37.2 37.5

1998 1999 2000 2001 2002 2003 2004 2005 2006

Figure 21. Type of sanitation for all South African households

Other Pit Toilet 35.3 34.8 33.2 34.1 33.9 32.6 31.3 30.2 30.7

VIP 5.5 5.4 5.4 5.7 5.9 7.0 9.5 9.6 9.8

Flush on Site 19.4 23.4 23.5 23.2 23.8 23.8 23.1 22.6 23.2

1998 1999 2000 2001 2002 2003 2004 2005 2006

Figure 22. Type of sanitation for African households

Figure 22 shows the sanitation situation for all African households, Figure 23 for rural African households, and Figure 24 for urban African households. Less than 5% of rural African households had a flush toilet in the dwelling in any year. For all African households there was a substantial decline in the percent in the worst sanitation category and an increase in the percent with decent sanitation. There was also a substantial increase in the percent of African households with a flush toilet in the dwelling.

Other Pit Toilet 55.6 56.9 53.9 55.6 56.7 56.3 52.4 53.5 57.2

VIP 7.3 7.5 7.5 8.4 9.2 11.4 15.4 16.0 17.1

Flush on Site 4.2 6.6 7.9 8.1 6.3 7.5 7.7 4.8 4.3

1998 1999 2000 2001 2002 2003 2004 2005 2006

Figure 23. Type of sanitation for rural African households

Other Pit Toilet 13.8 13.1 11.9 12.4 12.1 10.8 10.6 12.5 10.9

VIP 3.6 3.4 3.2 3.0 2.8 3.0 3.8 4.7 4.3

Flush on Site 35.6 39.9 39.6 38.6 40.5 38.7 38.1 36.3 37.3

1998 1999 2000 2001 2002 2003 2004 2005 2006

Figure 24. Type of sanitation for urban African households

For rural African households there was little increase in the percent with a flush toilet in the dwelling or in the use of any other category of flush or chemical toilet. Throughout the period, the percent of rural African households with any kind of flush or chemical toilet was in the range 8-13%. In 1998, 8% of rural African households used some kind of flush or chemical toilet, and in 2006 9% of rural African households used some kind of flush or chemical toilet. Thus, flush or chemical toilets have barely been in use among rural African households, and the prospects for this changing substantially in the near future seem small.

Between 1998 and 2006, there was an increase of 19% in the percent of urban African

households with a flush toilet in the dwelling. However, even in 2006, urban African households were about as likely to have a flush toilet on site (in the yard) as in the dwelling. The vast majority of urban African households use some kind of flush or chemical toilet, going from 71% in 1998 to 78% in 2006.

The VIP has been seen as a way to improve sanitation in situations where sewage or septic

tank systems are not available or are too costly. Figures 23 and 24 show that there was a substantial increase in the use of VIP’s in rural African households and some increase in urban African households. Figure 25 shows the percent of urban African households and of rural African households with VIP’s 1999-2006.9

It is clear that the bulk of VIP installations were in rural areas.

The cost of a basic VIP starts at R600 and requires about R60 per year to be budgeted for the costs of emptying if the emptying occurs once every five years. Thus, the increase in VIP’s has required a substantial financial investment, mainly from various parts of government. A variety of government departments have cooperated in VIP construction, including the Department of Water Affairs and Forestry, the Department of Local and Provincial Government and the Department of Science and Technology. There have been programs to construct VIP’s for household use (c.f. Tshikhudo, 2008) as well as for facilities such as schools (c.f. Mvula Trust, 2006).

Urban African 3.4 3.2 3.0 2.8 3.0 3.8 4.7 4.3

Rural African 7.5 7.5 8.4 9.2 11.4 15.4 16.0 17.1

1999 2000 2001 2002 2003 2004 2005 2006

Figure 25. Percent of all rural African and of urban African households with a

Ventilated Improved Pit Toilet (VIP)

9 Figure 25 shows data on VIP’s beginning in 1999 rather than beginning in 1998. There seems to be a problem with reporting of VIP’s in 1998. For 1998, for African households, 40.8% of all households are reported as having a pit toilet (whether a VIP or not), which is consistent with the 1999 results. However, for 1998, 14.5% of African households are reported as having a VIP and 26.3% are reported as having a non-VIP pit toilet. The report of 14.5% of African households having a VIP in 1998 seems impossible; this figure is more than 50% higher than the percent of African households with a VIP in 2006.

Figure 26 shows the percent of rural African households that used a VIP, a non-

ventilated pit toilet or a bucket toilet or no sanitation. The percentages for each of these categories are stacked so that the height of each bar represents the total percent of rural African households with any of these three types of sanitation. The percent of all rural households in the three sanitation categories varied little between 1999 and 2006, staying in the range of 87-91% for all rural African households. However, there was substantial change in the roles that these three categories played. The decline over time in the percent of rural African households with no sanitation or a bucket toilet was almost completely countered by the increasing percent of households with a VIP. The near stability of the non-VIP pit toilet category was the result of some households moving out of the unventilated pit toilet category into the VIP category and other households moving out of the bucket or none category into the unventilated pit toilet category.

Figure 26. Percent of rural African households with a VIP, a non-ventilated pit toilet or a

bucket toilet or no sanitation Figure 27 shows similar information for urban African households. Urban African

households also experienced a reduction in the percent in the bucket or none category and some increase in the VIP category, although less dramatically than for rural African households. However the percent of all urban African households in the three sanitation categories generally declined, since the percent with a flush toilet on site or in the dwelling increased from 71% in 1999 to 75% in 2006, as shown in Figure 24.

By 2006, the percent of rural African households in the category “Bucket Other None”

was 58% of the 1998 value, and the percent of urban African households in the category “Bucket Other None” was 61% of the 1998 value. Thus, many African households moved out of the worst sanitation category between 1998 and 2006.

Figure 27. Percent of urban African households with a VIP, a non-ventilated pit toilet or a bucket

toilet or no sanitation The nature of the “Bucket Other None” sanitation category differed between rural African

households and urban African households. Figure 28 shows for 1998 and Figure 29 shows for 2006 the composition of this category over time for rural African households, urban African households, and all South African households.10

7.93.5

3.528.0

Rural Africans Urban Africans All South Africans

NoneBucket

Figure 28. The composition of type of sanitation within the worst sanitation category, 1998

10 The category “Other” sanitation is not shown in Figures 28 and 29. There were no households recorded in the “Other” category in 2006, and in 1998, the “Other” category was recorded for .5% of rural African households, and for .3% of urban African households.

4.3 3.0

15.68.3

Rural Africans Urban Africans All South Africans

NoneBucket

Figure 29. The composition of type of sanitation within the worst sanitation category, 2006

We see in Figures 28 and 29 that almost no rural African households used a bucket

toilet. The situation for rural African households was that 28% in 1998 and 16% in 2006 had no sanitation facilities. The decline in the worst sanitation category is mainly accounted for by a shift into the unventilated pit toilet category. Apparently between 1998 and 2006 a substantial number of rural African households dug a pit toilet. This reflected household and community initiative rather than the effects of government programs or private commercial enterprises.

Bucket toilets were a more common type of sanitation for urban African households. The

percent of urban African households with a bucket toilet as the main type of sanitation declined from 8% in 1998 to 4% in 2006. In both 1998 and 2006 about twice as many urban African households used a bucket toilet as had no sanitation facilities.

Table 8. Frequency of removal of bucket toilets used as a household’s main type of sanitation: 1998 and 2006

1998 2006 Once a week or more often 89% 85% About once a fortnight 7% 8% About once a month 2% 4% Less often than once a month 2% 3% Total 100% 100%

The decline in the percent of households that use bucket toilets is a very good thing.

However, there are also questions about the characteristics of the bucket toilets that continue to be used. An important aspect of a bucket toilet is how frequently the waste is removed. A study in Port Elizabeth indicated that much of the dissatisfaction with bucket toilets stems not just from their general undesirability but also from the infrequency with which they are emptied (Thomas et al., 1999: xiii). For all South African households, Table 8 shows that there was no improvement in the frequency with which the waste was removed between 1998 and 2006 among households in which a bucket toilet was the main type of sanitation.

In 2003, a target was set to eliminate all bucket toilets by 2006 (DWAF, 2003: 6). Although there has been substantial reduction in bucket toilets, they had not been eliminated by 2006. One challenge to the eradication of bucket toilets is the large number of migrants from rural to urban places who often take up residence in poor informal and often illegal settlements, resulting in an increase in the number of people with no sanitation or using bucket toilets (City of Cape Town, 2001: 14-15). The issue of sanitation and other basic aspects of standard of living for recent migrants to cities, some of whom are illegally occupying land, is a substantial problem. Some municipal authorities have taken the position that municipalities have no obligation to supply basic services to those who are illegally occupying land (City of Cape Town, 2001: 14).

It is certainly understandable that municipalities do not wish to expend scarce resources

on services in illegal settlements. However, if there is contamination of the water supply that leads to a health problem, such as diarrhoea or cholera, the problem is as real whether the source was inadequate sanitation from a legal settlement or from an illegal settlement. One advantage of population surveys, such as the General Household Surveys, is that they aim to survey the entire population, whether they are residing in a locale legally or not. This is important in order to understand the standard of living of the entire South African population, regardless of how they came to reside in a particular location.

Figure 30 shows the sanitation situation for Coloured households. As for urban African

households, there was a large decline in the percent of Coloured households in the worst category of sanitation; the percent in 2006 was 30% of the value in 1998. The overall sanitation situation for Coloured households was much better than that of urban African households; in 2006 70% of Coloured households had a flush toilet in the dwelling, while this was true for only 38% of urban African households.

Other Pit Toilet 3.0 3.3 4.6 4.1 3.9 4.0 3.5 2.6 2.5

VIP 1.7 1.8 2.0 2.0 2.1 1.6 2.4 1.8 1.7

Flush on Site 12.4 13.9 15.6 14.8 15.3 17.0 16.7 21.4 21.6

1998 1999 2000 2001 2002 2003 2004 2005 2006

Figure 30. Type of sanitation for Coloured households

Other Pit Toilet 0.0 1.1 1.7 0.7 1.0 0.5 0.7 2.5 3.4

VIP 0.2 0.8 0.1 0.6 0.4 0.0 0.2 0.4 0.8

Flush on Site 5.1 4.2 4.9 2.4 2.6 2.8 2.4 3.1 3.8

1998 1999 2000 2001 2002 2003 2004 2005 2006

Figure 31. Type of sanitation for Asian households

Figure 31 shows the sanitation situation for Asian households and Figure 32 for White

households. Over 91% of Asian households and over 97% of White households had a flush toilet in the dwelling at every date.

Figure 33 shows the percent of households with a flush toilet in the dwelling for each of

the groups considered. The large differences among groups are quite clear. As in Figure 17, which examined the presence of a water tap in the dwelling, the percent with a flush toilet in the dwelling for all South African households is substantially higher than for all African households at every date (38% versus 23% for 2006). This is because the percent of White households, Asian households and Coloured households with a flush toilet in the dwelling is much higher than for African households.

Other Pit Toilet 0.0 0.1 0.1 0.0 0.1 0.1 0.0 0.1 0.1

VIP 0.0 0.0 0.1 0.0 0.0 0.0 0.0 0.0 0.1

Flush on Site 2.2 1.8 1.1 0.7 0.4 0.7 0.4 2.1 2.1

Flush in Dwelling 97.8 97.9 98.5 99.3 99.4 99.0 99.5 97.7 97.2

1998 1999 2000 2001 2002 2003 2004 2005 2006

Figure 32. Type of sanitation for White households

Figure 33. Percent of households with flush toilet in dwelling

Between 1998 and 2006, there was virtually no change in the percent of all South African households with a flush toilet in the dwelling. There was some increase for urban African households, from 32% in 1998 to 38% in 2006. For all African households there was an increase from 17% in 1998 to 23% in 2006. There was a small change for rural African households and no change for Coloured households. Almost all White households and Asian households had a flush toilet in the dwelling at all dates.

Figure 34 shows the percent of households with decent sanitation – a flush toilet, chemical toilet or VIP, regardless of its location. This percent increased between 1998 and 2006 for all groups except White households and Asian households, for whom it was close to 100% at all dates.

Figure 34. Percent of households with a flush toilet, chemical toilet or VIP

Having a flush toilet, chemical toilet or VIP, regardless of the location, likely provides a

situation in which the water supply is not contaminated by human waste, and the household benefits from the resulting health effects. However, all households aspire to having a flush toilet in their dwelling. Urban African households (and thus African households as a whole) are the only group that increased in the percent with a flush toilet in the dwelling. Other notable changes in sanitation are a decline in the percent of both urban African and rural African households with the worst sanitation and an increase in VIP’s, especially for rural African households.

Figure 35 looks at the percent of households using the worst type of sanitation, bucket,

stream or other. The scale has been changed from 0-100% to 0-30% in order to make the changes over time in Figure 35 clearer.

In 2006, the worst type of drinking water, shown in Figure 20, affected 17% of rural

African households. As shown in Figure 35, the worst type of sanitation affected 17% of rural Africans. Thus for rural Africans as a whole, very poor sanitation and very poor quality drinking

water were problems of a comparable magnitude. The percent of urban African households with the worst source of drinking water was less than 1% in 2006, while in 2006 over 11% of urban African households used the worst type of sanitation. Thus, for urban Africans extremely poor sanitation presented a more widespread problem in 2006 than did extremely poor quality drinking water.

Figure 35. Percent of households with a bucket other or none (worst type of sanitation)

Standard of Living Groups

We divide households into four groups by standard of living. These four groups are described in Table 9. A basic division of households by standard of living is into: (1) those that have clean drinking water and decent sanitation and (2) those that do not. A household is considered to have clean drinking water if the household’s main source of drinking water is a water tap (whether in the dwelling, on site or not on site, such as a public tap) or a water tanker – one of the top three categories of source of drinking water. A household is considered to have decent sanitation if it used a flush or chemical toilet (whether in the dwelling, on site or not on site) or a VIP (ventilated improved pit toilet) – one of the top four categories of sanitation.

The top two groups described in Table 9 have both clean drinking water and decent

sanitation, according to our definition. The two groups that do not have a decent standard of living do not have both clean drinking water and decent sanitation. Members of the top two groups do not have substantial threats to their health from their drinking water or sanitation facilities, while members of the bottom two groups are vulnerable to substantial health risks from their drinking water and/or sanitation. It has been a goal of the South African government to achieve a situation in which all households have a decent standard of living – clean drinking water and a flush or chemical toilet or a VIP.

We call the top group middle class because it has a middle class standard of living. It shares with the second group clean drinking water and decent sanitation, but households in this top group also have several other characteristics that are necessary for a household to have a middle class standard of living. Households in the top group reside in formal housing. Also, they have both a water tap and a flush toilet in the dwelling. In addition, electricity is the main source of lighting and electricity or gas is the main source of cooking. There is a landline phone in the dwelling or a household member has a cell phone.

Whether a household has a decent standard of living is related to service delivery. It is

an aim of government departments to try to reach the point when all households have clean drinking water and a flush or chemical toilet or a VIP. If a household has both of these characteristics, then we define that household as having a decent standard of living.

Whether a household has a middle class standard of living is not mainly an issue of

service delivery. Although households aspire to a middle class standard of living, the difference between a decent standard of living (the second best standard of living group, labelled middle low) and a middle class standard of living is not an issue of healthy conditions but rather is an issue of quality of life. Households are often willing to devote a substantial amount of their resources to attain a middle class rather than simply a decent standard of living.

Table 9. Definition of four standard of living (SOL) groups

Basic Requirements SOL Group Additional Requirements Decent Standard of

Living Must have both: 1) Clean drinking water (water from a tap or water tanker) and 2) Sanitation from a flush toilet, chemical toilet or a VIP

Middle Class All of the following must be true: Lives in formal housing Water tap in dwelling Flush toilet in dwelling Electricity is main light source Electricity or gas is main cooking source Has a landline phone or a household member has a cell phone

Middle Low Does not fit all of the requirements of the middle group, for example might have the water tap on site, such as in the yard, or might have a VIP

Not a Decent Standard of Living

Not both clean drinking water and sanitation from a flush toilet, chemical toilet or a VIP

Low Middle Low Does not fit all the requirements of the low group, for example might have an unimproved pit latrine, or might have a rainwater tank for drinking water

Low All of the following must be true: Does not live in formal housing Drinking water source is a stream, dam or other source Uses a bucket toilet, other or none Main cooking source is not electricity or gas

The bottom group shares with the third group the fact that it does not have both clean

drinking water and decent sanitation. In fact the bottom group has very poor quality drinking water and very poor sanitation. For households in the bottom group the main source of drinking water is a stream, dam, or other source – the bottom category of source of drinking water. In addition, the household’s source of sanitation is a bucket toilet or no toilet – the bottom category

of type of sanitation. Also, for households in the bottom group, the main source of cooking fuel is neither electricity nor gas, and the household does not live in formal housing.

Figure 36 shows the division of all South African households into the four standard of

living groups over time. The middle class group has changed very little – from 24% to 26%. The low (worst) group has declined – from 4% to 2%. The two intermediate groups – middle low and low middle low -- have changed very little, with some shift from the low middle low category to the middle low category. The shift from the low middle low SOL category to the middle low SOL category is important, because this constitutes a change from a household not having a decent standard of living to the household having a decent standard of living.

Figure 36. Distribution among SOL groups for all South African households

Figure 37 shows the distribution among the four standard of living categories for all

African households, Figure 38 for rural African households, and Figure 39 for urban African households. For all Africans, the percent of households which have a middle class standard of living almost doubled over time – an increase of 93%. The percent in the worst standard of living category (Low SOL) declined by 2006 to 47% of the 1998 value. There has also been an increase in the percent in the middle low category, from 36% to 43%.

The percent of rural African households in the low SOL category declined by 41%, from

10% to 6%. The vast majority of rural African households were in the low middle low category. This category declined somewhat over time, while the percent in the middle low category increased. This shift into the middle low SOL category was mainly due to the construction of VIP’s discussed earlier. The percent of rural African households with a middle class SOL more than quadrupled, but only from about 0.5% to about 2%.

Figure 37. Distribution among SOL groups for African households

Figure 38. Distribution among SOL groups for rural African households

Figure 39. Distribution among SOL groups for urban African households

The percent of urban African households with a middle class standard of living increased

by 59% between 1998 and 2006. Also, because of the increase in middle class households, the percent of urban African households with a decent standard of living increased over time. Almost no urban African households (less than 1%) were in the low SOL category at any date.

Figure 40 shows the standard of living distribution for Coloured households. Coloured households were about twice as likely to have a middle class standard of living as urban African households at every date. The percent of Coloured households in the low middle low category (second from the worst) in 2006 was 44% of its 1998 value. Almost no Coloured households were in the low category at any date. The one-year increase in the estimated percent of Coloured households in the middle class for 2002 (53% in 2002, compared with 39% in 2001 and 47% in 2003) is likely due to a misclassification of some households with a flush toilet on site as having a flush toilet in the dwelling. See Figure 30 for sanitation categories for Coloured households.

Figure 41 shows the standard of living distribution of Asian households and Figure 42

show similar information for White households. There is no discernible trend for White households, and it is unclear whether there is any trend for Asian households, given the small number of Asian cases. At every date at least 69% of Asian households and at least 80% of White households had a middle class standard of living.

Figure 40. Distribution among SOL groups for Coloured households

Figure 41. Distribution among SOL groups for Asian households

Figure 42. Distribution among SOL groups for White households

Standard of Living Patterns by Province

South Africa is a large country, and consideration of levels and trends in standard of

living would be deficient without some examination of standard of living by geographic location, specifically by province. In this section we compare the standard of living situation in provinces for 1998 and 2006. The results for 1998 use an estimation of the presence of VIP’s based on the relation of VIP’s to other factors in 1999.11

Table 10 shows the distribution by population group for each province in 2006. There is substantial variability across provinces. In only five provinces do 2% or more of all households have a Coloured household head. In fact, in the households included in the 2006 General Household Survey only 3 households in Limpopo, 12 households in Mpumalanga and 48 households in North West Province had a Coloured household head. Only in KwaZulu Natal and Gauteng did more than 2% of households have an Asian head. In the tables by province we only show the results for the African population and for the White population due to the small number of provinces in which there were a substantial number of Coloured or Asian households in the surveys.

11 See the Data Appendix for further discussion of the estimation of households with VIP’s in 1998. Between 1998 and 2006 the boundaries of every province changed except for Western Cape and Free State. The possible effects of these boundary changes are not taken into account in this analysis.

Table 10. Distribution of population groups by province, 2006

African Coloured Asian White Total Western Cape 30.2 44.7 .3 24.8 100.0 Eastern Cape 85.9 6.2 .2 7.7 100.0 Northern Cape 44.3 41.8 .9 13.1 100.0 Free State 83.6 4.1 .1 12.2 100.0 KwaZulu Natal 80.0 2.0 8.8 9.3 100.0 North West 87.7 1.8 .6 10.0 100.0 Gauteng 77.0 2.7 2.1 18.2 100.0 Mpumalanga 91.2 .9 .4 7.5 100.0 Limpopo 95.9 .1 .5 3.5 100.0

Western Cape and to a lesser extent Northern Cape stand out in the relatively small proportion of households that are African -- less than 50% for each province. These two provinces also contain the bulk of Coloured households in South Africa. Of the 3429 Coloured households in the 2006 GHS, 2739 (80%) were in these two provinces. On the other hand, in Limpopo and Mpumalanga over 90% of households were African. Western Cape and Gauteng are also unusual since in those provinces over 18% of households were White in 2006.

0 20 40 60 80 100

Western Cape

Eastern Cape

Northern Cape

Free State

KwaZulu Natal

North West

Gauteng

Mpumalanga

Limpopo

1998 MiddleClass SOL

Figure 43. Percent of households with a middle class standard of living by province, 1998 and 2006

Figure 43 shows the percent of households with a middle class standard of living by

province in 1998 and 2006. This percent increased between 1998 and 2006 in every province except Western Cape, but Western Cape had the highest percent with a middle class standard of living at both dates, followed by Gauteng.

It is useful to keep in mind the differences among provinces in the percent of African

households that reside in urban areas. As shown in Figure 44, provinces differ greatly in the percent of all African households that reside in urban places, ranging from 14% in Limpopo to 96% in Gauteng and 97% in Western Cape.

0 20 40 60 80 100

Western Cape

Eastern Cape

Northern Cape

Free State

KwaZulu Natal

North West

Gauteng

Mpumalanga

Limpopo

Figure 44. Percent of all African households that reside in urban places by province, 2006

0 5 10 15 20 25 30

Western Cape

Eastern Cape

Northern Cape

Free State

KwaZulu Natal

North West

Gauteng

Mpumalanga

Limpopo

Figure 45. Percent of urban African households with middle class standard of living by province,

1998 and 2006

We noted that virtually all middle class African households are in urban areas. Figure 45 shows the percent of urban African households that had a middle class standard of living in 1998 and 2006. Figure 45 shows an increase in the percent of urban African households with a middle class standard of living in every province except for Western Cape. However, while for all households, in 2006 Western Cape had the highest percent with a middle class standard of living, for urban Africans in 2006, Western Cape had the lowest percent with a middle class standard of living. Also, in every province except for Western Cape, the percent of urban Africans with a middle class standard of living increased between 1998 and 2006.

Perhaps the percent of urban African households with a middle class standard of living in Western Cape declined between 1998 and 2006 due to a large number of poor African migrants to Western Cape, especially from Eastern Cape, between 1998 and 2006. One surprising feature of Figure 45 is the low variability in the percent of urban African households with a middle class standard of living across provinces in 2006. All the values for 2006 are in the range 15%-24%.

Table 10 showed that Coloured households were most frequently found in Western

Cape and in Northern Cape, but that 2.7 percent of all household heads were Coloured in Gauteng. In 2006, 55% of Coloured households in Western Cape had a middle class standard of living, but 66% of Coloured households in Gauteng had a middle class standard of living. Gauteng is not a traditional area of settlement for Coloured households. Perhaps Coloured households only migrated to Gauteng if their economic prospects there were quite good, thus resulting in the higher percent of Coloured households in Gauteng than in Western Cape with a middle class standard of living.

The literature on migration indicates that the probability of migrating often has a J-

shaped or U-shaped relation to education or economic situation (Lee, 1966; Zodgekar and Seetharam, 1972). Those who have a very good education (or standard of living) and those with a very poor education (or standard of living) are more likely to migrate than those with intermediate characteristics. Those with good characteristics are not likely to migrate unless their prospects at the destination are very good, that is, they think they will likely benefit substantially from the move. This might be the case if a family member already has an assured good job before the move. Those with a very poor situation are likely to migrate because they think they will benefit from the move, but they may not have as assured a situation at the destination. Also, although they may benefit from the move, their characteristics might not look very good in comparison with longer-term residents at the destination. In 1998, Coloured migration to Gauteng was likely much less common than it was by 2006. This could be why Coloured households in Gauteng were more likely to have a middle class standard of living than those in Western Cape by 2006.

In 2006, KwaZulu Natal and Gauteng were the only provinces in which Asians

comprised more than 2% of all households, constituting 8.8% of households in KwaZulu Natal and 2.1% of households in Gauteng. In 2006, 66% of Asian households in KwaZulu Natal and 77% of households in Gauteng had a middle class standard of living. Perhaps Asians only migrated to Gauteng if they knew they had very good economic prospects there.

Figure 46 shows the percent of the households in each province in 1998 and in 2006

which had a decent standard of living. In both 1998 and 2006, the percent of households with a decent standard of living was the highest in Western Cape and in Gauteng and the percent with a decent standard of living was the lowest in Eastern Cape and in Limpopo. The percent with a decent standard of living was considerably higher in KwaZulu Natal than in Eastern Cape or Limpopo. KwaZulu Natal ranked fifth among provinces in this percent.

For 1992-96, Gauteng and Western Cape were the most popular destinations for

interprovincial migrants (Kok et al., 2003: 37), and according to a 2001-02 HSRC survey, Western Cape, Gauteng, and Kwazulu Natal were the most popular destinations among those planning to become interprovincial migrants (Wentzel, Viljoen and Kok, 2006: 193-194). The popularity of Gauteng and Western Cape as migration destinations is likely related to the high percent of households with a decent standard of living in those provinces.

Figure 46. Percent of households with a decent standard of living by province, 1998 and 2006

Figure 47. Percent of African households with a decent standard of living by province, 1998 and 2006

Figure 47 shows the percent of African households with a decent standard of living by

province. In 2006, over 97% of White households had a decent standard of living. Thus the results by province for White households are not shown. For African households the percent with a decent standard of living increased in every province between 1998 and 2006. African households in Limpopo had the lowest percent with a decent standard of living, with African households in Eastern Cape faring somewhat better. A higher percent of African households in KwaZulu Natal enjoyed a decent standard of living.

Figure 48. Percent of urban African households with a decent standard of living by province,

1998 and 2006

Figure 48 shows the percent of urban African households with a decent standard of

living in 1998 and 2006, and Figure 49 shows similar information for rural African households. The percent of urban African households with a decent standard of living increased between 1998 and 2006 in every province except the Northwest, in which there was a small decrease.

Figure 49. Percent of rural African households with a decent standard of living by province,

1998 and 2006 For rural African households the proportion with a decent standard of living increased in

every province except Gauteng, in which there was a small decrease. Rural Africans in Gauteng had the second highest percent with a decent standard of living, with the highest percent for rural African households in Western Cape. However, the percent of rural Africans with a decent standard of living in very low (below 30%) in five provinces: Eastern Cape, Limpopo, KwaZulu Natal, North West and Mpumalanga. Clearly in several provinces there is a great need for improvement in the standard of living for rural Africans.

It is interesting to compare the percent of rural African households and urban African

households with a decent standard of living for 2006 in Figures 48 and 49, focussing on Western Cape. Only in Western Cape does a higher percent of rural African households than urban African households have a decent standard of living in 2006. This is likely another result of a high level of migration of poor Africans from Eastern Cape to Western Cape. These poor migrants settled in urban parts of Western Cape. Members of rural African households in Western Cape likely had lived in that province for a longer time than the urban African migrants.

Figure 50 shows the percent of households by province that are in the worst SOL group

in 1998 and 2006. In many provinces this percent was close to zero at both dates. In every province in which at least 1% of the population was in the worst standard of living group in 1998, this percent declined between 1998 and 2006. Eastern Cape and KwaZulu Natal stand out in Figure 50 for having by far the largest percent in the worst standard of living group among all provinces at both dates. However, this percentage declined substantially between 1998 and 2006 in both provinces.

Figure 50. Percent of all households in low SOL group by province, 1998 and 2006

Figure 51 shows the percent of rural African households with a low standard of living (the

worst standard of living). This percent is by far the largest in Eastern Cape, followed by KwaZulu Natal, although the percent in both provinces declined substantially between 1998 and 2006.

Figure 51. Percent of rural African households in low SOL group by province, 1998 and 2006

It was clear in the earlier examination of type of sanitation that the construction of VIP’s has played a substantial role in improving the standard of living of rural African households. We look next at the changing role of VIP’s across provinces.

Figure 52 shows the percent of rural African households with a VIP in 1999 and in 2006. The increase in this percentage is an indicator of the intensity of the VIP construction program in the rural area of a given province. The percent with a VIP more than tripled in Eastern Cape, Free State and KwaZulu Natal.

Figure 52. The percent of rural African households with a VIP, 1999 and 2006 Figure 53 looks at VIP’s in a different way. It shows the percent of all pit toilets (whether

ventilated or not) which were VIP’s. This figure indicates how far a province has yet to go to replace all unventilated pit toilets with VIP’s.

Figure 53. The percent of all pit toilets in rural African households which are VIP’s, 1999 and 2006

Looking at Figure 52, even in 1999 over 10% of rural African households had a VIP in Northwest Province and in Mpumalanga. There has been a very vigorous program of VIP construction between 1999 and 2006 in KwaZulu Natal and in Free State. Looking at Figure 53, Western Cape and then Northern Cape have made the most progress in replacing unventilated pit toilets with VIP’s, whereas Gauteng and Limpopo have the farthest to go in replacement of unventilated pit toilets with VIP’s.

Of course the aim is not just to replace unventilated pit toilets with VIP’s but also to

reach a situation where no households use bucket toilets or have no sanitary facilities. Figure 54 shows among rural African households in 1999 the percent with no sanitation, with unventilated pit toilets and with VIP’s, and Figure 55 shows similar information for 2006. The difference between the top of each column and the 100% line indicates the percent of rural African households in the given province that had some type of flush or chemical toilet. At both dates, over 60% of rural African households in Western Cape and in Gauteng had some type of flush or chemical toilet, and over 60% of rural African households in Northern Cape had a flush or chemical toilet or a VIP. In 1999, over 80% of rural African households in Eastern Cape, Limpopo and KwaZulu Natal had no sanitation or had an unventilated pit toilet. By 2006, in KwaZulu Natal, this had dropped to 69% both because of construction of VIP’s and due to a reduction in the percent of rural African households with no sanitation.

We noted in Figure 53 that Gauteng and Limpopo were the two provinces in which in

2006 the smallest proportion of all pit toilets were VIP’s. However, Figure 55 makes clear that the situation in these two provinces is vastly different, since in 2006 in Gauteng 16% of rural African households had an unventilated pit toilet, while this was true for 71% of rural African households in Limpopo.

Figure 54. Percent of rural African households by province with no sanitation, a pit toilet or a VIP, 1999

Figure 55. Percent of rural African households by province with no sanitation, a pit toilet or a VIP, 2006

Figure 56 shows by province the relation between the percent of urban African households with a middle class standard of living and the percent of rural African households with a low standard of living in 1998, and Figure 57 shows similar information for 2006. If provinces that had relatively poor rural African populations also had relatively poor urban African populations, then there should be a negative relation between the values on the two axes. If provinces in which the rural African population was poor, the urban African population was relatively well off, there should be a positive relation. Figures 56 and 57 have the same scales on both axes in order to make changes over time easier to see.

0 5 10 15 20 25

% Urban African HH Middle Class SOL

Figure 56. Percent of urban African households with a middle class SOL plotted against the

percent of rural African households with a low SOL by province, 1998

0 5 10 15 20 25

% Urban African HH Middle Class SOL

Figure 57. Percent of urban African households with a middle class SOL plotted against the percent of rural African households with a low SOL by province, 2006

As we saw in Figures 45 and 51, we see in Figures 56 and 57 that between 1998 and

2006 the percent of urban Africans with a middle class standard of living generally increased (as indicated by a shift to the right between Figure 56 and Figure 57), and the percent of rural Africans with a low standard of living generally declined (as indicated by a downward shift between Figure 56 and Figure 57). However the situations in Eastern Cape and in KwaZulu Natal deserve further note. In those two provinces, between 1998 and 2006 the percent of urban African households with a middle class standard of living increased to 2.6 and 1.9 times the 1998 value respectively, while the percent of rural African households with a low standard of living declined to .60 and .55 of the 1998 value respectively.

The large increase in the percent of urban African households in the middle class and

the relatively smaller decline in the percent of rural African households in the low standard of living category led to an increasing disparity between the standard of living of urban Africans and of rural Africans, especially in Eastern Cape. In 1998, 68% of all Africans in the lowest SOL category lived in Eastern Cape and 24% lived in KwaZulu Natal. In 2006, 70% of all African households in the lowest SOL category lived in Eastern Cape, and 24% of all African households in the lowest SOL category lived in KwaZulu Natal. Thus, in 1998 only 8% of all Africans in the lowest SOL category lived somewhere other than Eastern Cape or KwaZulu Natal, and in 2006 only 6% lived outside of those two provinces. In 1998, 7% of all African households with a middle class standard of living, and in 2006, 10% of all African households with a middle class standard of living resided in Eastern Cape.

Next we look in somewhat more detail at changes in standard of living between 1998 and 2006 in Eastern Cape, KwaZulu Natal, Western Cape and Gauteng. Eastern Cape and Kwazulu Natal are distinguished by the high proportion of the rural African population in the low standard of living category in 1998 and 2006 and by the huge increase in the percent of the urban African population with a middle class standard of living. Western Cape and Gauteng had the highest percent of all households with a middle class standard of living, but were not high in the percent of their urban African populations with a middle class standard of living in 2006. Understanding the dynamics in these provinces would be informative in discerning the underlying processes in these changes in standard of living.

Figures 58 and 59 show for rural Africans in Eastern Cape and rural Africans in KwaZulu Natal, respectively, the percent of households in the low standard of living category along with the percent with the worst category of drinking water and the percent with the worst category of sanitation. Although having formal housing or using electricity or gas as the main cooking source can also move a household out of the low standard of living category, drinking water source and sanitation are the main characteristics determining whether a household falls into the low standard of living group. The same scale is shown in Figures 58 and 59.

In Eastern Cape all the indicators examined in Figure 58 improved (decreased) from

1998-2000, stagnated through 2004 and continued to improve 2004-2006. In KwaZulu Natal (Figure 59), despite some fluctuations, all the indicators tended to improve throughout the time period. It is not clear what led to the different patterns in Eastern Cape and in KwaZulu Natal, but in each province the downward trajectories of the percent of the population in the worst water category and in the worst sanitation category were similar.

Low SOL 31.3 26.9 28.8 29.9 32.1 27.1 29.3 23.8 20.1

Stream Dam Pond 66.2 59.1 57.5 60.1 60.1 59.3 56.2 52.6 45.3

Bucket None 50.5 44.8 48.7 48.9 50.1 44.7 45.1 43.8 37.4

1998 1999 2000 2001 2002 2003 2004 2005 2006

Figure 58. Percent of rural African households in Eastern Cape, with a lowest SOL, in the worst

water category and in the worst sanitation category

Low SOL 11.4 16.5 11.7 10.0 8.2 7.5 6.1 6.6 6.2

Stream Dam Pond 42.0 38.6 40.7 30.9 31.3 27.7 24.4 25.9 23.0

Bucket None 30.3 28.7 27.0 24.8 22.1 17.9 17.2 17.3 14.7

1998 1999 2000 2001 2002 2003 2004 2005 2006

Figure 59. Percent of rural African households in KwaZulu Natal, with a low SOL category, in the

worst water category and in the worst sanitation category

Figures 60-63 show the percent of urban African households with a middle class standard of living, with a tap in dwelling, and with a flush toilet in the dwelling for Eastern Cape, KwaZulu Natal, Western Cape and Gauteng. The same scale is used in Figures 60-63.

In Eastern Cape and KwaZulu Natal, all of the indicators improve fairly steadily throughout the

period. For urban Africans in both of these provinces the rate of improvement in obtaining a tap in the dwelling is rapid in comparison to the virtually constant level for all urban African households, shown in Figure 17. Although Figure 33 shows some increase in the percent of all urban African households with a flush toilet in the dwelling, the rate of improvement in this is also much more rapid in these two provinces than for all urban African households.

Middle Class SOL 8.9 12.4 15.2 17.3 17.1 17.9 17.3 21.7 23.5

Tap in Dwelling 30.6 37.7 34.1 38.2 39.7 38.2 36.4 38.3 43.8

1998 1999 2000 2001 2002 2003 2004 2005 2006

Figure 60. Percent of urban African households in Eastern Cape with a middle class SOL, in the

best water category and in the best sanitation category

Middle Class SOL 12.4 13.5 17.7 18.3 25.2 18.8 21.5 24.7 23.6

Tap in Dwelling 34.1 36.7 44.7 40.3 46.3 43.7 41.6 43.9 38.7

1998 1999 2000 2001 2002 2003 2004 2005 2006

Figure 61. Percent of urban African households in KwaZulu Natal with a middle class SOL, in the

The picture for Western Cape shown in Figure 62 is very different. Over time the percent of urban African households with a middle class standard of living, as well as the percent of house-holds with the other two indicators fluctuated greatly and declined after 2002. Although some of the fluctuation could result from the small number of urban African households in the surveys from Western Cape, it is clear that something different happened in Western Cape than in Eastern Cape and Kwazulu Natal.

Figure 63 shows a different situation for urban Africans in Gauteng. The indicators for

Gauteng do not fluctuate as much as they did for Western Cape in Figure 62, but they also indicate deterioration after 2004.

Middle Class SOL 15.9 8.3 13.2 19.4 23.4 19.5 20.1 14.4 15.5

Tap in Dwelling 42.4 31.8 34.7 35.2 43.2 41.9 41.4 37.1 35.4

1998 1999 2000 2001 2002 2003 2004 2005 2006

Figure 62. Percent of urban African households in Western Cape with a middle class SOL, in the

Middle Class SOL 14.3 14.6 16.1 17.2 19.6 23.7 27.0 23.4 22.3

Tap in Dwelling 46.1 42.1 42.5 43.8 44.1 46.7 46.6 41.8 41.8

1998 1999 2000 2001 2002 2003 2004 2005 2006

Figure 63. Percent of urban African households in Gauteng with a middle class SOL, in the best

water category and in the best sanitation category It is likely that interprovincial migration is the main reason for the differences between

the trajectory for urban African middle class households in Eastern Cape and KwaZulu Natal on the one hand and Western Cape and Gauteng on the other hand. Western Cape and Gauteng are the major destinations for interprovincial migrants. A strong migration stream for Africans flows from Eastern Cape to urban Western Cape. In urban Western Cape, with a small proportion of all households being African, newly arrived poor African migrants affect the characteristics of urban African households in urban Western Cape as a whole. This leads to deterioration in the characteristics of urban African households as a whole in Western Cape, even if every individual urban African household in Western Cape improved its standard of living over time. This phenomenon, in a less dramatic fashion, is likely the explanation for the patterns seen for Gauteng in Figure 63. Although Gauteng is a major migration destination, a much larger percent of the households in Gauteng than in Western Cape are African (77% versus 30%). Thus, with a larger longer-term resident African population in Gauteng than in Western Cape, the effect of poor recent African migrants on the characteristics of all urban African households is less for Gauteng than for Western Cape.

For both 1993 and 1995, Leibbrandt and Wollard (1999) concluded that Eastern Cape was the poorest province and Gauteng and Western Cape were the provinces with the least poverty. For 1996 and 2001 Leibbrandt et al. (2006) also concluded that Eastern Cape and Limpopo were the poorest provinces and Gauteng and Western Cape were the provinces with the lowest level of poverty. Naidoo, Leibbrandt and Dorrington (2007) found and Leibbrandt and Mlatsheni (2007) report that migrants from rural Eastern Cape to Cape Town generally did not fare well in their income-producing efforts. This is similar to the finding of van der Berg et al. (2002) that male African migrants from Eastern Cape to Western Cape were not very successful. However, even though migrants from Eastern Cape may have fared poorly in

Western Cape compared to Western Cape natives or longer term residents, this does not mean that the actual situation of these migrants did not improve over time. This is especially likely in light of the poor standard of living situation for rural African households in Eastern Cape shown in Figure 51. The migration of poor Africans from Eastern Cape to Western Cape represents the lower socioeconomic status tail of the U-shaped migration propensity discussed earlier (Lee, 1966; Zodgekar and Seetharam, 1972).

Population Group Composition of Standard of Living Groups

In Figures 37-42 we looked at the percent distribution of population groups among the

four standard of living categories 1998-2006. Another interesting question is: What is the percent distribution of population groups among the four SOL categories, and how has this distribution changed over time? This question is relevant to the role of different population groups in the market, and the answer affects the considerations of manufacturers and advertisers.

Figures 64 and 65 answer this question. They show the percent distribution of population

groups within each of the four SOL categories for 1998 and 2006. The two worst categories (the low middle low category and the low category) are occupied almost totally by African households at both dates. Note that Figures 64 and 65 show the percent distribution by population group within each SOL group. Although 99.3% of all the households in the low standard of living category had an African head of household in 2006, recall from Figure 37 that in 2006, only 2.7% of all households in which the head was African were in the low SOL group.

Figure 64. Percent distribution of households by population group within four SOL groups: 1998

Figure 65. Percent distribution of households by population group within four SOL groups: 2006

The most striking result in Figures 64 and 65 is the increase in the percent of

households with a middle class standard of living comprised by African households between 1998 and 2006, an increase from 21% to 40%. Since the percent of White households which had a middle class standard of living changed trivially between 1998 and 2006 (81% to 82% in Figure 42) and of Coloured households increased between 1998 and 2006 (43% to 49% in Figure 40), this increase in the percent of households with a middle class standard of living which were headed by an African is due to an improvement in the standard of living of African households rather than due to a deterioration in the standard of living of Coloured households or of White households.12

It is also interesting that in 2006 more households with a middle class standard of living had an African head than had a White head. This is consistent with the conclusion by Seekings and Nattrass (2005:306) that between 1995 and 2000 the percent of the top income decile obtained by those who were African or Coloured increased from 22% to 34-40%.

These findings do not mean that African households were as likely to have a middle

class standard of living as non-African households. As shown in Figure 64, in 1998 African households were only .3 times as likely have a middle class standard of living as their overall population percent would have implied, while Coloured households were 1.8 times, Asian households were 2.6 times and White households were 3.2 times as likely to have a middle class standard of living as would be implied by their population percent. As shown in Figure 65, by 2006, African households were .51 times as likely (one-half as likely) to have a middle class 12 The percent of Asian households with a middle class standard of living is 75% both in 1998 and 2005, while it is 69% in 2006. The reported decline between 2005 and 2006 is likely related to fluctuations between surveys and the small number of Asian households in each survey.

standard of living as their population percentage would have implied. The values for non-African households in 2006 were almost unchanged from 1998, 1.9 for Coloured households, 2.6 for Asian households and 3.2 for White households.

However, this change in the population group composition of households with a middle

class standard of living is not meaningless. The increase between 1998 and 2006 in the percent of all middle class households with an African head reflects increasing purchasing power in the market of Africans and an increased motivation for those manufacturing and selling goods aimed at attracting disposable income to direct product development and advertising at African consumers. The awareness of business and financial interests of the increasing portion of disposable income controlled by African households is clear in a Standard Bank (2005) paper.

1998-2000 7.5 0.5 14.6 40.9 75.0 83.3 23.4

2001-2003 10.7 1.1 19.8 46.1 78.7 84.0 24.6

2004-2006 12.8 1.4 22.2 47.6 74.6 85.3 26.2

All Africans

Rural Africans

Urban Africans

Figure 66. Percent of households with a middle class standard of living (top SOL group)

Figure 66 shows the percent of households with a middle class standard of living within

each group. To smooth out year to year variations, means for three-year periods (1998-2000, 2001-2003, and 2004-2006) are shown. The increase over time for Coloured households and for urban African households is clear. There was little change in the percent of White households or of Asian households with a middle class standard of living. These results indicate that the increase in the share of middle class households comprised by African households (Figures 64 and 65) was not due to any worsening in the situation of any non-African group.

The percent of rural African households with a middle class standard of living remained

very small (1.4% in 2006). One might ask whether urban residence was a necessary precondition for attaining a middle class standard of living. Some main components of a middle class standard of living, a tap in the dwelling and a flush toilet in the dwelling, are certainly far easier to achieve in an urban area, where often one can be connected to municipal water and sewage systems.

To achieve these aspects of a middle class standard of living in a rural area is possible,

but it requires substantial expenditure of resources. That a middle class standard of living is possible in a rural area is clear from the situation of rural White households. In 2006, 5% of

White households lived in a rural area. Among rural White households in 2006, 63% had monthly expenditures of more than R2500, and 37% had monthly expenditures of more than R5000. Among the rural White households, 73% had a tap in the dwelling, 88% had a flush toilet in the dwelling and 62% had a middle class standard of living. A well can be connected to a water system with taps in the dwelling, and a flush toilet can be connected to a holding tank and a septic field, but this is more expensive and requires more maintenance than would be necessary in an urban setting with connections to municipal systems.

>R2500 per month >R5000 per month

Urban Africans Coloured Asian White All South Africans

Figure 67. Percent of households with a middle class SOL with monthly expenditure greater than R2500 and greater than R5000, 2006

Despite the increase in the percent of urban African households with a middle class

standard of living shown in Figure 66, not all middle class households experience the same economic situation. Figure 67 shows the percent of middle class households in 2006 whose monthly household expenditures exceeded R2500 and the percent whose monthly household expenditure exceeded R5000.13

This is not shown for rural Africans, for whom less than 2% of households had a middle class standard of living in 2006.

The monthly household expenditure of middle class White households is much more likely to be above R2500 and to be above R5000 than for Asian middle class households, middle class Coloured households or middle class urban African households. Among middle class households, Asian households tend to have the second highest monthly expenditures, followed by Coloured households.

Figure 68 shows the percent within each group who have a decent standard of living. As

in Figure 66, the values shown are averages over a three-year period. These households have both decent water and decent sanitation, something to which all South African households 13 The Labour Force Survey 2001 and the General Household Surveys in 2002 and later asked a question about total household expenditures in the previous month. This was coded in eight categories, with less than R400 as the lowest category and R10,000+ as the highest category. This is a rough indicator of monthly household income or monthly household consumption.

aspire. This percent increased for every group except White households and Asian households, for which it was over 96% in every period. Even in 2004-2006, only slightly more than half (54%) of all African households had both decent sanitation and a decent source of drinking water, and 79% of rural African households still did not have both clean drinking water and decent sanitation.

Figure 68. Percent of households with a decent standard of living (Top two SOL groups)

As seen in Figure 68, urban African households were about four times as likely as rural

African households to have a decent standard of living. However, urban and rural African households that achieved a decent standard of living did so in very different ways. Among urban African households with a decent standard of living, 46% had a flush toilet in their dwelling and 46% had a flush toilet on the site, but not in their dwelling. Among rural African households with a decent standard of living, 60% had a VIP. Among urban African households with a decent standard of living, 48% had a tap in their dwelling, and 46% had a tap on site but not in their dwelling. Among rural African households with a decent standard of living, 45% had a tap on site but not in the dwelling, and 30% used a public tap.

Households with a decent standard of living are not all equally likely to be above some

expenditure threshold. Figure 69 shows the percent of households in the top two SOL groups which had monthly expenditures of R800 or more in 2006. The overall picture is similar to that in Figure 67. Within a given standard of living group, White households and Asian households are the most likely to have monthly expenditures above a particular value, with Coloured households somewhat less likely and with urban African households even less likely. Forty-seven percent of African households with a decent standard of living had monthly expenditures of less than R800.

Figure 69. Percent of households with a decent standard of living (Top two SOL groups) with monthly expenditures greater than R800, 2006

Forty-two percent of rural African households with a decent standard of living have

monthly expenditure of over R800, but this is true for 55% of urban African households in the top two SOL groups. Note from Figure 68 that in 2004-2006, while 82% of urban African households are in the top two SOL groups, this is only true for 21% of rural African households.

1998-2000 5.4 10.0 0.5 0.2 0.1 0.0 4.0

2001-2003 4.3 8.7 0.2 0.1 0.0 0.0 3.4

2004-2006 3.3 7.0 0.1 0.1 0.0 0.0 2.6

All Africans

Rural Africans

Urban Africans

Figure 70. Percent of households in Low (worst) SOL group

Figure 70 shows the percent of households by group over time in the low (worst)

standard of living category, again as averages over three-year periods. There has been a steady decline in the percent of rural Africans in the worst group, but even in 2004-2006, 7% of rural African households remain in a dire standard of living situation. Also, in 2006, 39% of the rural African households in the low SOL group had monthly expenditures of less than R400.

Virtually all of those in the low SOL group were rural African households. In fact, in 1998, 94.8% of those in the low SOL group were rural African households, and in 2004, 98.8% of those in the low SOL group were rural African households. It is useful to examine somewhat further the characteristics of those rural African households which are in the low SOL group. We will look at rural African households in the low OL group in comparison to all rural African households in 2004.

In 2004, 62% of those rural African households in the low SOL group are headed by a

female, in contrast to 45% of rural African households that are not in the low SOL group. This high level of female-headedness seems mainly related to a disproportionate number of these female heads being widows – 53% of female heads of households in the low SOL group compared to 42% of other rural African female-headed households. Thus, this is not a situation where the husband or significant other is absent due to labour migration. It could be that female-headed households without an adult working for pay elsewhere are especially disadvantaged in their efforts to move out of the situation of absolute poverty in which the low SOL group lives.

Concluding Comments

Since 1998, some aspects of the lives of South Africans have definitely improved. An increasing percent of all households have: (1) access to clean water, (2) access to a telephone, either through a landline in the dwelling or through a cell phone possessed by a household member, and (3) electricity as the major source of light. A decreasing percent of South African households have no sanitation or use a bucket toilet for sanitation. Also, an increasing percent of urban African households and of Coloured households have arguably entered the middle class as indicated by standard of living measures. Also the percent of all groups that have a decent standard of living has increased over time. The overall percent of South African households with a middle class standard of living, though, has changed little, remaining at about 25%.

White households and Asian households experienced a very good standard of living throughout the time period considered, with little evidence of any deterioration in this over time. Some might expect that the situation of Coloured households and of urban African households would be similar. For the most part, they do not look similar. Rather the situation of Coloured households is far better than that of urban African households.

Let us look at how some indicators of a very poor standard of living have changed for all South African households (Figure 71), for rural African households (Figure 72), and for urban African households (Figure 73).

Note that the scale in these three figures differs (0%-40% in Figure 71, 0%-80% in

Figure 72, and 0-20% in Figure 73). The scale was changed among the figures so that variations within each figure would be clear.

Figure 71. Percent of South African households with a poor standard of living: 1998 and 2006

Figure 72. Percent of rural African households with a poor standard of living: 1998 and 2006

Figure 73. Percent of urban African households with a poor standard of living: 1998 and 2006 For all South African households, rural African households, and urban African

households, there was a substantial decline between 1998 and 2006 in the percent of households in the worst situation in each of the areas considered in these figures. This indicates true poverty alleviation. Whether the pace of this poverty alleviation should have been more rapid is a matter for policy debate. Some aspects of the improvements indicated in Figures 71-73 are mainly the result of South African governmental actions, while other improvements probably have little to do with governmental action. The South African government deserves credit for encouraging rural electrification. But improvement in telephone availability is mainly the result of the commercial effort of cell phone companies. The South African government has worked to improve the quality of the drinking water supply, and has worked to eliminate bucket toilets, but the reduction in the percent of rural African with no sanitation probably is the result of households or village groups digging holes for pit latrines.

The 2005 report on South Africa’s progress toward its Millennium Development Goals

expresses and sets goals concerning the situation of slumdwellers (South Africa, 2005: 8, 52, 56). This referred to urban African households, especially concerning their sanitation situation. Although much could be improved in the sanitation situation of urban Africans, it seems clear that the sanitation situation and other aspects of the lives of rural Africans are in more need of improvement than urban Africans. In urban areas, with denser settlement, improvements in sanitation and other aspects of life are likely easier and less costly on a per household basis than for more sparsely settled and remote rural households. However, it seems clear that the greatest need for improvement in the standard of living and alleviation of severe poverty remains with rural African households. Providing a poor rural household with a pit latrine in place of a having no sanitation facilities does not increase that household’s purchasing power and does not directly contribute to South African economic growth, but it does improve the standard of living, convenience of daily life and likely health of that rural household. The increase in the number of VIP’s in rural areas has substantially improved the standard of living of rural African households, but there remains much to be done to extend VIP sanitation to those who currently use an unventilated pit latrine, a bucket toilet or have no sanitation at all.

The situation across provinces also varies, although there is relatively little variation in the percent of urban African households with a middle class standard of living. Eastern Cape and KwaZulu Natal, despite improvements over time, stand out as places where a disproportionate percent of rural African households continue to live in absolute poverty. Hopefully further improvements will reduce the disparity between these two provinces and the rest of rural South Africa.

References Anderson, Barbara A. and Phillips, Heston E. 2006. Trends in the percentage of children who

are orphaned in South Africa: 1995-2005. Report No. 03-90-06. Pretoria: Statistics South Africa.

Anderson, Barbara A., Romani, John H., Phillips, Heston E., and van Zyl, Johan A., 2002. “Environment, access to health care, and other factors affecting infant and child survival among the African and Coloured populations of South Africa, 1989-94,” Population and Environment, 23: 349-364.

Ardlington, Cally, Lam, David, Leibbrandt, Murray, and Welch, Matthew. 2006. “The sensitivity to key data imputations of recent estimates of income poverty and inequality in South Africa,” Economic Modelling, 23: 822-835.

Atkinson, A. B. 2004. “The Luxembourg Income Study (LIS): Past, present and future,” Socio-Economic Review, 2: 165-190.

Austin, L. M. 1996. “STED systems in South Africa,” a paper presented at the 22nd Water, Engineering and Development Centre (WEDC) Conference, New Delhi. Available at www.lboro.ac.uk/departments/cv/wedc/papers/22/groupg/austin.pdf Accessed 22 June 2008

Bhorat, Haroon, Naidoo, Pranushka, and van der Westhuizen, Carlene. 2006. "Shifts in non-income welfare in South Africa,” Working Paper 06/108 May, Development Policy Research Unit, University of Cape Town.

Bird, Kate, Hulme, David, Moore, Karen, and Shepherd, Andrew. 2002. Chronic poverty and remote rural areas. Chronic Poverty Research Centre Working Paper No. 13. School of Public Policy, University of Birmingham, England. Available at http://www.chronicpoverty.org/pdfs/13Bird_et_al.pdf Accessed 10 June 2007.

Black Diamond 1. 2007. website of the Unilever Institute of Strategic Marketing, University of Cape Town. Available at http://www.unileverinstitute.co.za/black_diamond.asp Accessed on 7 June 2007.

Burger, S. E., and Esrey, S. A. “Water and sanitation: Health and nutrition benefits to children,” in P. Pinstrup-Andersen, D. Pelletier and H. Alderman (Eds.), Child growth and nutrition in developing countries: Priorities for action, Ithaca: Cornell University Press. Pp. 153-175.

City of Cape Town, City of. 2001. Water services development plan: Chapter 3 Customer profile and service levels. Available at http://www.capetown.gov.za/en/Water/WaterservicesDevPlan/Documents/chapter3_Custom_Profile.pdf Accessed 29 June 2008.

Carr, Richard. 2001. “Chapter 5. Excreta-related infections and the role of sanitation in the control of transmission, in World Health Organization.” Water quality: Guidelines, standards and health. Lorna Fewtrell and Janie Bartram (Eds.), London: IWA Publishing, pp. 89-113. Available at http://whqlibdoc.who.int/publications/2001/94154533X_chap5.pdf Accessed 7 June 2007.

“Chaos as new magistrate’s courts left without phones.” 13 November 2003. The Herald, Available at http://www.epherald.co.za/herald/2003/11/13/news/n07_13112003.htm Accessed 3 July 2008.

Cross, Catherine. 2003. Migrant worker remittances and micro-finance in South Africa. A study for the ILO and TEBA Bank, Pretoria; Human Sciences Research Council. Available at http://www.ilo.org/public/english/employment/finance/download/cross.pdf Accessed 28 May 2008.

Department of Minerals and Energy (DME). 2007. Free basic electricity. Available at

http://www.dme.gov.za/energy/elect_fbe.stm Accessed 5 June 2007. Department of Water Affairs and Forestry (DWAF). 1994. Water supply and sanitation policy

White Paper. Water – an indivisible national asset. November. Available at http://www.dwaf.gov.za/Documents/Policies/WSSP.pdf Accessed 8 June 2007.

Department of Water Affairs and Forestry (DWAF). 2001. Minister Kasril’s budget speech, 15 May 2001. Available at http://www.info.gov.za/speeches/2001/010515245p1002.htm Accessed 18 June 2007.

Department of Water Affairs and Forestry (DWAF). 2002. Sanitation for a healthy nation: The policy on basic household sanitation made easy. February. Pretoria: Government Printers. Available at http://www.dwaf.gov.za/dir_ws/content/lids/PDF/Technical.pdf Accessed 25 June 2008.

Department of Water Affairs and Forestry (DWAF). 2003a. Strategic framework for water services: Water is life, sanitation is dignity. September. Available at http://www.dwaf.gov.za/Documents/Policies/Strategic%20Framework%20approved.pdf Accessed 8 June 2007.

Department of Water Affairs and Forestry (DWAF). 2003b. Statement by Minister Ronnie Kasrils on the celebration of the 9 millionth person to receive safe water since 1994. http://www.dwaf.gov.za/Communications/PressReleases/2003/9%20Mil%20Media%20Release.doc Accessed 8 June 2007

Department of Water Affairs and Forestry (DWAF). 2007. Vision. Available at http://www.dwaf.gov.za/misson.asp Accessed 8 June 2007.

Ellison, Kirk W., Runyon, Richard P., and Haber, Audrey. 1990. Fundamentals of social statistics, New York: McGraw-Hill.

ESKOM. 2007. Electrification: ESKOM’s electrification programme in South Africa. Available at http://www.eskom.co.za/live/content.php?Item_ID=2786&Revision=en/0 Accessed 8 June 2007.

Fewtrell, Lorna, Kaufmann, Rachel B., Kay, David, Enanoria, Wayne, Haller, Laurence, and Colford, John M., Jr. 2005.”Water, sanitation, and hygiene interventions to reduce diarrhoea in less developed countries: a systematic review and meta-analysis,” The Lancet Infectious Diseases, 5: 42-52.

Habicht, J-P, DaVanzo, J., and Butz, W. P. 1988. “Mother’s milk and sewage: Their interactive effects on infant mortality,” Pediatrics, 81: 456-461.

International Centre for Policy Studies. 2002. The report on preliminary research “Ukraine’s middle class,” October, Kyiv, Ukraine. Available at http://www.icps.com.ua/doc/middle_class_report_e.pdf Accessed 31 May 2007

Hanekom, Derek. 2005. “Keynote address at the opening of the Third International Conference on Ecological Sanitation by the Deputy Minister of Science and Technology, Derek Hanekom.” 23 May. Available at http://www.info.gov.za/speeches/index.htm Accessed 20 June 2008.

Johnston, Alastair Iain. 2004. “Chinese middle class attitudes toward international affairs: Nascent liberalization?” China Quarterly, 179: 603-628.

Johnston, Sandy, and Bernstein, Ann. 2007. Voices of anger: Protest and conflict in two municipalities. Report to the Conflict and Governance Facility. April, Johannesburg: Centre for Development and Enterprise.

Kgalushi, Rudzani, Smite, Stef, and Eales, Kathy. 2004, People living with HIV/AIDS in a context of rural poverty: the importance of water and sanitation services and hygiene education, Johannesburg: Mvula Trust and Delft: IRC International Water and Sanitation Centre. Available at http://www.irc.nl/page/10382 Accessed 23 June 2008.

“Killing waters,” 14 May 2008, TVSA, Available at http://www.tvsa.co.za/default.asp?blogname=special_assignment_episodes&articleid=8081 Accessed 23 June 2008.

Kok, Pieter, O’Donovan, Michael, Bouare, Oumar and van Zyl, Johan. 2003. Post-Apartheid patterns of internal migration in South Africa, Cape Town; HSRC Press.

Laurent, P. 2005. Household drinking water systems and their impact on people with weakened immunity, MSF-Holland, Public Health Department, February. Available at http://www.who.int/household_water/research/HWTS_impacts_on_weakened_immunity.pdf Accessed 7 June 2007.

Everett S. Lee. 1966. “A theory of migration,” Demography, 3: 47-57. Lee, Lung-fei, Rosenzweig, Mark R., and Pitt, Mark M. 1997. “The effects of improved nutrition,

sanitation, and water quality on child health in high-mortality populations,” Journal of Econometrics, 77: 209-235.

Leibbrandt, Murray, Poswell, Laura, Naidoo, Pranushka, and Welch, Matthew. 2006. “Chapter 3. Measuring recent changes in South African inequality and poverty using 1996 and 2001 census data,” in Haroon Bhorat and Ravi Kanbur, Eds. Poverty and policy in post-Apartheid South Africa. Cape Town: HSRC Press. Pp. 95-141.

Leibbrandt, Murray, and Woolard, Ingrid. 1999. “A comparison of poverty in South Africa’s nine provinces,” Development Southern Africa, 16: 37-54.

Leibbrandt, Murray, and Mlatsheni, Cecil. 2007. “Slim pickings at the Cape of little hope,” May 10, Available at http://www.capeinfo.com/vision/argus/Leibbrandt&Mlatsheni.php Accessed 27 May 2008.

Luxembourg Income Study (LIS). Website. Available at http://www.lisproject.org Accessed 27 May 2008.

Mansoor Moaddel. 1994. “Political conflict in the world economy: A cross-national analysis of modernization and world-system theories,” American Sociological Review, 59: 276-303

Mvula Trust. 2006. The rural schools sanitation programme. The Mvula Trust Case Study Series no 2. Available at www.mvula.co.za/redir/content/download/432/4121/file/Rural%20Schools%20Sanitation% Accessed 20 July 2008.

Naidoo, N. Leibbrandt, M., and Dorrington, R. 2007. “Magnitudes, personal characteristics and activities of Eastern Cape migrants: A comparison with other migrants and with non-migrants using data from the 1996 and 2001 censuses.” University of Cape Town. Unpublished mimeo.

Neupert, R.F. 1981. Fertility and middle class in Latin America. Doctoral dissertation, Department of Sociology, Brown University,

Oosthuizen, Morne, and Naidoo, Pranushka. 2004. “Internal migration to the Gauteng Province,” Working Paper 04/88 December, Development Policy Research Unit, University of Cape Town.

Posel, Dorrit. 2003. “The collection of national survey data in South Africa (1993-2001): rendering labour migration invisible,” Development Southern Africa, 20: 361-368.

Posel, Dorrit, and Casale, Daniela. 2006. “Chapter 10. Internal labour migration and household poverty in post-apartheid South Africa,” in Haroon Bhorat and Ravi Kanbur, Eds. Poverty and policy in post-Apartheid South Africa. Cape Town: HSRC Press. Pp. 351-365.

Romani, John H., and Anderson, Barbara A., 2002. “Development, health and the environment: Factors influencing infant and child survival in South Africa,” Occasional Paper Number 5, Research Programme in Integrated Rural and Regional Development, Human Sciences Research Council, Pretoria.

Ross, John A., Rich, Margaret, Molzen, Janet, and Pensak, Michael. 1988. Family planning and child survival in one hundred developing countries. New York: Center for Population and Family Health, Columbia University.

Seekings, Jeremy, and Nattrass, Nicoli. 2005. Class, race and inequality in South Africa. New Haven: Yale University Press.

Senauer, Benjamin, and Goetz, Linda. 2003. “The growing middle class in developing countries and the market for high-value food products,” a paper presented at the Workshop for Global Markets for High-Value Food, Economic Research Service, Washington, DC, February 14. Available at http://www.farmfoundation.org/documents/Ben-Senauerpaper2--10--3-13-03_000.pdf Accessed 29 May 2007.

South Africa. 2005. South Africa Millennium Development Goals Country Report 2005. See: http://www.undg.org/archive_docs/6584-South_Africa_MDG_Report.pdf Accessed 31 May

2007. South Africa, Presidency. 2007. Development indicators mid-term review. Pretoria: The

Presidency. Spirer, Herbert F., Spirer, Louise, and Jaffe, A. J. 1998. Misused statistics. (Second edition).

New York: Marcel Dekker. South Africa, Statistics South Africa. 2002. Earning and spending in South Africa: Selected

findings from the income and expenditure surveys of October 1995 and October 2000. Pretoria: Statistics South Africa.

Standard Bank. 2005. Changing colour of the South African consumer, Standard Bank Economics Series, September.

Statistics South Africa. 2001. South Africa in transition: Selected findings from the October household survey of 1999 and changes that have occurred between 1995 and 1999. Pretoria: Statistics South Africa.

Taylor, K. W, and Frideres, James. 1972. “Issues versus controversies: Substantive and statistical significance,” American Sociological Review, 37: 464-472.

“Telkom customers hang up,” 14 June 2007, The Times (Johannesburg), p. 13. Thomas, E. P., Seager, J. R., Viljoen, E., Potgieter, F., Rossouw, A., Tokota, B., McGranahan,

G., and Kjellen, M. 1999. Household environment and health in Port Elizabeth, South Africa, Urban Environment Series Report no. 6, Stockholm: Stockholm Environment Institute in Collaboration with South African Medical Research Council and Sida.

Tahikhudo, Elmon. 2008. “Toilets for rural communities,” Zoutnet 6 June. Available at www.zoutnet.co.za/details.asp?StoNum=6450 Accessed 6 July 2008.

United Nations. 2000. World summit for social development: Programme of action. August 2000. Available at http://www.un.org/esa/socdev/wssd/agreements/index.html Accessed 31 May 2007.

Van der Berg, Servaas, Burger, Rudolf, Leibbrandt, Murray, and Mlatsheni, Cecil. “Migration and the changing rural-urban interface in South Africa: What can we learn from census and survey data?” a paper presented at the DPRU/FES Conference on Labour Markets and Poverty in South Africa, Johannesburg, 22-24 October. Available at http://www.commerce.uct.ac.za/research_units/dpru/Conf2002pdf/Migration_and_the_changing_rural-urban.pdf Accessed 3 June 2008.

Weir-Smith, Gina, and Zama, S’bonsile. 2006. “Appendix B: GIS as a tool in migration research,” in Pieter Kok, Derik Gelderblom, John O. Oucho and Johan van Zyl, Eds., Migration in South and southern Africa: Dynamics and determinants, HSRC Press: Cape Town. Pp. 307-325.

Wentzel, Marie, Viljoen, Johan, and Kok, Pieter. 2006. “Migrant motivations and capacities in relation to key migration streams,” in Pieter Kok, Derik Gelderblom, John O. Oucho and Johan van Zyl, Eds., Migration in South and southern Africa: Dynamics and determinants, HSRC Press: Cape Town. Pp. 171-204.

Zodgekar, A.V., and K. S. Seetharam, K. S. 1972. “Interdivisional migration differentials by education for groups of selected SMSA’s, United States, 1960,” Demography, 9: 683-699.

Data Appendix

Decision not to use the data from the 1995-1997 October Household Surveys and a problem with the data from the 1998 October Household Survey

The 1995-1997 October Household Surveys had some deficiencies, compared with the

surveys conducted in 1998 and later. The African sample for the 1995 survey seems to have been drawn from disproportionately relatively well-off areas, leading to an estimated decline in the welfare of Africans between 1995 and 1996, even when this is not plausible. For example, for 1995 it was reported that 32% of all African households had a water tap in the dwelling – for each year 1996-2005, the reported percent of African households with a tap in the dwelling never exceeded 26%. Although we used recalculated weights based on the1996 South African Census for the 1995 data, the new weights do not totally adjust for the problem with the 1995 sample.

The 1996 October Household Survey was conducted in the same year as the South

African Census, putting a strain on Statistics South Africa resources, with some effects on the quality of that survey (c.f. Anderson and Phillips, 2006: 3).

In the 1997 October Household Survey, for Africans the sanitation situation reported for

Africans is inexplicably good, being substantially better than that reported for either 1996 or 1998. For example, it was reported that in 1997, 19% of rural African households had a flush toilet in the dwelling, while this was reported for 3% of rural African households for both 1996 and 1998. The implausibly better sanitation situation in 1997 compared to 1996 and 1998 also is clear in Romani and Anderson (2002).

There is a problem with the data for Ventilated Improved Pit Toilets (VIPS) for 1998. The

1998 October Household Survey reported that 14.5% of all African households had VIPs, although the percent reported for 1997 was 6.8 and for 1999 was 5.4. Even in 2006, the reported percent was 9.8. We do not know exactly what the problem was in 1998, but we suspect there was an interviewer training problem,

We did not want to discard the 1998 data. Thus, we estimate the proportion of all pit

latrines that are VIP’s and the proportion that are unventilated pit latrines using the relationships in the data for 1999. This approach probably somewhat overestimates the number of VIP’s in 1998, but the effect is to minimize the amount of estimated change over time. Specifically, we estimate for each group for 1999 (All South African households, African households, Coloured households, Asian households, White households, rural African households, urban African households, and subdivisions of these groups by province and rural/urban areas of provinces) the proportion of all pit toilets that are VIP’s according to whether or not the household had clean drinking water. For example, in 1999, for rural African households with clean drinking water, 13.7% of the pit toilets were VIP’s, while for rural African households which did not have clean drinking water, 7.9% of pit toilets were VIPs. These subgroup proportions of pit toilets that were VIPs were applied to the actual 1998 data to subdivide households with pit toilets between those with VIP’s and those with unventilated pit toilets. This procedure allowed the estimation of the proportion of households with VIP’s and also allowed estimation of the percent of groups in the second standard of living category and the percent in the third standard of living category. This was important for estimating the percent of households with a decent standard of living in 1998.

Testing for a statistically significant change over time or between groups in a given year

Whether there is a significant change between two dates in the percent of a group with a given characteristic, such as the percent of all South African households in formal housing, can be tested by constructing 95% confidence intervals around the estimates of the percent of households with a given characteristic in each of the years under consideration. Given the assumptions of the measure, the 95% confidence interval is the range within which there is a 95% probability that the true value of the percent lies. The width of the confidence interval depends on the magnitude of the percent and on the number of cases (households) in the group considered.14

If the 95% confidence intervals for the two years do not overlap, then it is reasonable to conclude that the percent of households with a given characteristic actually changed between the two years.

Figure 3 shows the percent of households residing in formal housing. For each of the seven groups considered in Figure 3 except for Asians there was a statistically significant change in the percent of households in formal housing between 1998 and 2004. There were only 492 Asian households in the 1998 OHS and 706 Asian households in the 2004 GHS. That is, for every group except Asians, the 95% confidence interval bracketing the percent for 1998 did not overlap with the 95% confidence interval for the percent in 2004. Three groups (all African households, urban African households, and White households) showed a significant increase, and the other two groups (rural African households and Coloured households) showed a significant decrease. The statistical significance of quite small changes in a percent is partly because the number of households in each group in each year is very large.

Just because the change in a percent between 1998 and 2004 is statistically significant,

this does not mean that it represents a substantively important change. It is difficult to imagine that a change from 57.9% of African households residing in formal housing in 1998 to 58.5% of African households residing in formal housing in 2004 made an important difference in the lives of members of African households. This change in six years represents an annual rate of increase of .2%, which is unlikely to be meaningful to anyone. This phenomenon when a statistically significant difference is not meaningful is sometimes discussed as the difference between substantive significance and statistical significance (Elifson, Runyon, and Haber, 1990: 336; Spirer, Spirer, and Jaffe, 1998: 143-145, 236-237; Taylor and Frideres, 1972).

Almost any apparent change over time in any graph presented in this paper is

statistically significant. However, we do not think that all of these changes are substantively important. Significance tests are extremely important to protect researchers against making unwarranted conclusions about a change over time or a difference between two groups when the number of cases analysed is fairly small. However, that is not the situation in the analyses in this paper, since typically a large number of households comprises each group considered.

In this paper, we do not mention every time a change over time is statistically significant.

Also, we will not discuss as important changes that do not also represent a statistically significant change over time or difference between groups in a given year.

14 To estimate the confidence interval, the household weights for a given survey are multiplied by a constant which results in the number of weighted cases for the given year equalling the number of surveyed cases for that year, but the relative weights of various households is the same as in the survey’s original household weights.

Variability in survey data

The data sources we are using are surveys that take place every year, covering a comparable population and asking comparable questions. These are important advantages of the data sources used.

However, these are not the only considerations that influence the comparability of survey

data over time. Sometimes, apart from the possibility of random fluctuations, there are seemingly small changes in the survey administration, such as interpretation of questions by interviewers, that lead to implausibly large changes in some indicator from one year to the next year. These kinds of things occur in surveys in many countries. As discussed, this was part of the reason that we begin the analysis with data from 1998 rather than from an earlier year.

We do not interpret large changes in one year when there is a rebound in the

subsequent year to a value close to the value in the year before the large change as real. We were able to substantially guard against this in choosing the first year for analysis, but we are more limited in determining whether all values for the last year in the analysis, 2006, has yielded reasonable data in all or almost all areas. We are somewhat cautious in fully accepting results for 2006 that are radically different from what would have been expected based on the results for 2004 and 2005. Fortunately, the results for 2006 appear to be reasonable.

changes in standard of living among population groups

Documents

Transcript of changes in standard of living among population groups

FOR SMALL GROUPS

Religion, standard of living and the quality of life

Non Living Resources

Living Faithfully in Your World - Google Groups

A Living Fossil

Representations of Lie groups and quantum groups, Pitman Re

On the Homology of Automorphism Groups of Free Groups

The biological standard of living (and its convergence) in Colombia, 1870-2003. A tropical success story

सुचना - Google Groups

WORK GROUPS LIST

The Living Church

living - religions

LIVING IN MANCHESTER

PROFITRAK's - Google Groups

Nested social groups within the social environment of a dementia care assisted living setting

Automorphism Groups of Finite p-Groups - arXiv

Active Living Guide

Untitled - Google Groups

RSPO Standard Operating Procedure for Standard Setting ...

Kahler groups, quasi-projective groups and 3-manifold groups