Changes in standard of living among population groups in ... · Statistics South Africa Changes in standard of living among population groups in South Africa, 1998–2006 (Report

Changes in standard of living among population groups in South Africa,

1998–2006

Statistics South Africa

Pali Lehohla Statistician-GeneralReport No. 03-03-02


Changes in standard of living among population groups in South Africa, 1998–2006 (Report No. 03-03-02)

i

Changes in standard of living among population groups in South Africa, 1998–2006 Published by Statistics South Africa, Private Bag X44, Pretoria 0001 © Statistics South Africa, 2009 Users may apply or process this data, provided Statistics South Africa (Stats SA) is acknowledged as the original source of the data; that it is specified that the application and/or analysis is the result of the user's independent processing of the data; and that neither the basic data nor any reprocessed version or application thereof may be sold or offered for sale in any form whatsoever without prior permission from Stats SA. Changes in standard of living among population groups in South Africa, 1998–2006, Statistics South Africa, Pretoria, 2009. [Report no 03-03-02 (2009)] ISBN 978-0-621-39029-2 A complete set of Stats SA publications is available at Stats SA Library and the following libraries:

National Library of South Africa, Pretoria Division National Library of South Africa, Cape Town Division Library of Parliament, Cape Town Bloemfontein Public Library Natal Society Library, Pietermaritzburg Johannesburg Public Library Eastern Cape Library Services, King William’s Town Central Regional Library, Polokwane Central Reference Library, Nelspruit Central Reference Collection, Kimberley Central Reference Library, Mmabatho

This report is available on the Stats SA website: www.statssa.gov.za Copies are obtainable from: Printing and Distribution, Statistics South Africa Tel: (012) 310 8044 (Free publications) Tel: (012) 310 619 / 310 8358 Email: [email protected] / [email protected] Fax: (012) 321 7381



ii

CONTENTS Executive Summary .................................................................................................................................................... 1 1. Introduction ....................................................................................................................................................... 2 2. Date Sources ..................................................................................................................................................... 4 3. Population Groups ............................................................................................................................................ 6 4. Household Access to Services ........................................................................................................................ 8 4.1 Type of Housing .................................................................................................................................................. 8 4.2 Telephone in the Household ............................................................................................................................... 9 4.3 The Main Lighting Source – Electricity and Candles ........................................................................................ 12 4.4 Electricity or Gas as a Cooking Source ............................................................................................................ 14 4.5 Main Source of Drinking Water ......................................................................................................................... 16 4.6 Type of Sanitation ............................................................................................................................................. 24 5. Standard of Living Groups ............................................................................................................................. 36 6. Standard of Living Patterns by Province ..................................................................................................... 42 7. Population Group Composition of Standard of Living Groups ................................................................. 57 8. Concluding Comments ................................................................................................................................... 64 References ................................................................................................................................................................. 67 Data Appendix ........................................................................................................................................................... 71 LIST OF TABLES Table 1: Unweighted and weighted distribution of households by population group, 2006 GHS .............................. 6 Table 2: Among households connected to MAINS electricity in 2004, those that received free basic electricity .... 13 Table 3: Among households with electricity as the main lighting source, the percentage using electricity or

gas as the main cooking source, 2006 GHS .............................................................................................. 15 Table 4: Classification of types of drinking water source according to the categories in the surveys used ............. 17 Table 5: Perceptions of characteristics of main source of drinking water according to category of drinking

water source, 2005 ..................................................................................................................................... 17 Table 6: Classification of type of sanitation according to the categories in the surveys .......................................... 25 Table 7: Frequency of removal of bucket toilets used as a household’s main type of sanitation: 1998 and

2006 ............................................................................................................................................................ 31 Table 8: Definition of four standard of living (SOL) groups ...................................................................................... 37 Table 9: Distribution of population groups by province, 2006 .................................................................................. 42



iii

LIST OF FIGURES

Figure 1: Percentage living in urban places ............................................................................................................. 7 Figure 2: African households as a percentage of all rural households .................................................................... 8 Figure 3: Percentage residing in formal housing ...................................................................................................... 9 Figure 4: Percentage with a landline in dwelling or a household member with a cellphone .................................. 10 Figure 5: Percentage with a landline telephone in dwelling ................................................................................... 10 Figure 6: Percentage of households more than 15-16 minutes from an accessible telephone: 1998 and

2006 ........................................................................................................................................................ 11 Figure 7: Households with electricity as the main lighting source .......................................................................... 13 Figure 8: Households using candles as the main lighting source .......................................................................... 14 Figure 9: Households with electricity or gas as main cooking source .................................................................... 15 Figure 10: Drinking water source for all South African households ......................................................................... 18 Figure 11: Drinking water source for African households ........................................................................................ 18 Figure 12: Drinking water source for rural African households ................................................................................ 19 Figure 13: Drinking water source for urban African households .............................................................................. 19 Figure 14: Drinking water source for Coloured households ..................................................................................... 20 Figure 15: Drinking water source for Asian households ........................................................................................... 21 Figure 16: Drinking water source for White households .......................................................................................... 21 Figure 17: Percentage of households with water taps in dwelling ........................................................................... 22 Figure 18: Households with water tap in dwelling or on site .................................................................................... 23 Figure 19: Households with clean drinking water ..................................................................................................... 23 Figure 20: Households with drinking water from stream dam or other source ......................................................... 24 Figure 21: Type of sanitation for all South African households ................................................................................ 26 Figure 22: Type of sanitation for African households ............................................................................................... 26 Figure 23: Type of sanitation for rural African households ....................................................................................... 27 Figure 24: Type of sanitation for urban African households ..................................................................................... 27 Figure 25: Rural and urban African households with a VIP ..................................................................................... 28 Figure 26: Rural African households with a VIP, a non-ventilated pit toilet, a bucket toilet or no sanitation ........... 29 Figure 27: Urban African households with a VIP, a non-ventilated pit toilet or a bucket toilet or no sanitation ....... 30 Figure 28: The composition of type of sanitation within the worst sanitation category, 1998 .................................. 30 Figure 29: The composition of type of sanitation within the worst sanitation category, 2006 .................................. 31 Figure 30: Type of sanitation for Coloured households ........................................................................................... 32 Figure 31: Type of sanitation for Asian households ................................................................................................. 33 Figure 32: Type of sanitation for White households ................................................................................................. 33 Figure 33: Households with flush toilet in dwelling ................................................................................................... 34 Figure 34: Households with a flush toilet, chemical toilet or VIP .............................................................................. 35 Figure 35: Percentage of households with a bucket toilet or none .......................................................................... 36 Figure 36: Distribution among SOL groups for all South African households .......................................................... 38 Figure 37: Distribution among SOL groups for African households ......................................................................... 38 Figure 38: Distribution among SOL groups for rural African households ................................................................. 39 Figure 39: Distribution among SOL groups for urban African households ............................................................... 39 Figure 40: Distribution among SOL groups for Coloured households...................................................................... 40 Figure 41: Distribution among SOL groups for Asian households ........................................................................... 41 Figure 42: Distribution among SOL groups for White households ........................................................................... 41 Figure 43: Households with a middle class standard of living by province, 1998 and 2006 .................................... 43 Figure 44: African households that reside in urban places by province, 2006 ......................................................... 43 Figure 45: Urban African households with a middle class standard of living by province, 1998 and 2006 ............. 44 Figure 46: Households with a decent standard of living by province, 1998 and 2006 ............................................. 45 Figure 47: African households with a decent standard of living by province, 1998 and 2006 ................................. 46 Figure 48: Urban African households with a decent standard of living by province, 1998 and 2006 ...................... 47



iv

Figure 49: Rural African households with a decent standard of living by province, 1998 and 2006 ........................ 47 Figure 50: The percentage of all households in low SOL group by province, 1998 and 2006 ................................ 48 Figure 51: The percentage of rural African households in low SOL group by province, 1998 and 2006 ................. 48 Figure 52: Rural African households with a VIP, 1999 and 2006 ............................................................................. 49 Figure 53: The percentage of all pit toilets in rural African households which are VIP’s, 1999 and 2006 ............... 50 Figure 54: Rural African households by province with no sanitation, a pit toilet or a VIP, 1999 .............................. 51 Figure 55: Rural African households by province with no sanitation, a pit toilet or a VIP, 2006 .............................. 51 Figure 56: Urban African households with a middle class SOL plotted against rural African households with a

low SOL by province, 1998 ..................................................................................................................... 52 Figure 57: Urban African households with a middle class SOL plotted against rural African households with a

low SOL by province, 2006 ..................................................................................................................... 52 Figure 58: Rural African households in Eastern Cape, with a lowest SOL, in the worst water category and

sanitation category .................................................................................................................................. 54 Figure 59: Rural African households in KwaZulu-Natal, with a low SOL, in the worst water category and

sanitation category .................................................................................................................................. 54 Figure 60: Urban African households in Eastern Cape with a middle class SOL and best water and sanitation

category ................................................................................................................................................... 55 Figure 61: Urban African households in KwaZulu-Natal with a middle class SOL, best water and sanitation

category ................................................................................................................................................... 55 Figure 62: Urban African households in Western Cape with middle class SOL, best water and sanitation

category ................................................................................................................................................... 56 Figure 63: Urban African households in Gauteng with a middle class SOL, in the best water category and in

the best sanitation category .................................................................................................................... 56 Figure 64: The distribution of households by population group within four SOL groups: 1998 ............................... 58 Figure 65: The distribution of households by population group within four SOL groups: 2006 ............................... 59 Figure 66: Percentage of households with a middle class standard of living (top SOL group) ................................ 60 Figure 67: Households with middle class SOL with monthly expenditure greater than R2500 and R5000,

2006 ........................................................................................................................................................ 61 Figure 68: Households with a decent standard of living (Top two SOL groups) ...................................................... 62 Figure 69: Households with a decent standard of living with monthly expenditures greater than R800, 2006 ....... 62 Figure 70: Percentage of households in Low (worst) SOL group ............................................................................ 63 Figure 71: South African households with a poor standard of living: 1998 and 2006 .............................................. 64 Figure 72: Rural African households with a poor standard of living: 1998 and 2006 ............................................... 65 Figure 73: Urban African households with a poor standard of living: 1998 and 2006 .............................................. 65



1

Executive Summary1

This paper examines changes in the standard of living in South Africa overall and for the four main

population groups, 1998-2006. 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 percentage of urban African households in the best standard of living group -middle class standard of

living- increased from 14% to 22% between 1998 and 2006, indicating an 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 percentage of rural African households 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 percentage of all households, 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

percentage 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 (Ventilated Improved pit latrine) in rural African communities has contributed

substantially to improvement in the standard of living of rural African households.

1 Acknowledgments: 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.



2

1. Introduction This paper examines changes in the material standard of living of households in South Africa 1998-2006.

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 access to 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.

Improving the standard of living of all South Africans, especially those in the worst circumstances, as well

as increasing the overall rate of economic growth and fostering the growth of a diverse, economically productive

middle class have been among the most important goals of the post-1994 government. 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 lack 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 African Population 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 healthy, but a household whose drinking water is from a public tap

definitely is not generally considered to be experiencing a middle class standard of living.



3

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 a middle class has also been linked to social and political 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 in the percentage 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 convenient or desirable, 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). Providing households with a decent standard of living has been

one of the key goals of government, including access to clean water and safe sanitation.

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 characterised 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?

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?

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, which contributes both to household health and to self-

respect?



4

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 owns 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.2 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 secure and

desirable living circumstances.

2. Date Sources

The analysis in this paper is based on the 1998-1999 October Household Surveys, (OHS) the 2000-2001,

Labour Force Surveys (LFS) and the 2002-2006 General Household Surveys (GHS). These are national

representative surveys covering 18,000-30,000 households.3 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 LFS because the OHS were last administered in 1999, and the GHS began in 2002. The

2001 Labour Force Survey was used rather than the 2001 South African Population 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 questionnaires for surveys or censuses are constructed, there are reasons to retain earlier

questions with exactly the same wording and reasons to modify or omit 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.4

2 It is known that consumption patterns of African and of White households in South Africa differ, even for households at the same fairly high income levels. There are various proposed explanations of this difference in consumption patterns, including cultural differences in consumption preferences and differences arising from the generally lower level of assets among African households than among White households with the same expenditure level (Nieftagodien and van der Berg, 2007).

3 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. 4 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.



5

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 only to households, 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 percentage distributions refer to distribution of

households rather than to distribution 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 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 income for household members

who slept at the household at least four nights in the previous week, making income of household members who

are temporarily absent impossible to determine.5 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

estimate of total household income to indicate household wellbeing.

Other scholars have used the OHS and the GHS 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 Population

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.

5 See Posel (2003) for a discussion of the kinds of data collected on temporary labour migrants in South African national surveys and censuses.



6

3. Population Groups

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

shown in Table 1, using 2006 as a recent example of the situation in a given year. Table 1 shows the percentage

distribution of households by population group in the survey and the percentage 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.6

Table 1: Unweighted and weighted distribution of households by population group, 2006 GHS

Group Number of households Percent distribution in sample

Percent by population group in sample weighted by

household weightAfrican 21 721 77,6 77,5Coloured 3 430 12,2 7,6Asian 509 1,8 2,5White 2 316 8,3 12,3Unspecified 26 0,1Total 28 002 100,0 100,0

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.

Figure 1 show the percentage of households in each of the four population groups and for all South African

households who resided in an urban area in each year 1998-2006. The overall percentage 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 White, Asian and Coloured

households raises the percentage of households in urban places in 2006 from 57% for all African households to

65% for all South African households, countering the common impression that data for all South African

households are similar to data for all African households.

6 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.



7

Figure 1: Percentage living in urban places

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 locations and as we shall see, African rural and urban households differ

markedly in almost all household characteristics.

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.

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 differentiating between those in urban areas and 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.



8

Figure 2: African households as a percentage of all rural households

93.9 92.9 92.9 92.9 93.3 93.1 92.9 96.095.8

0

20

40

60

80

100

1998 1999 2000 2001 2002 2003 2004 2005 2006

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 percentage 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. Furthermore, 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.

4. Household Access to Services 4.1 Type of Housing

The type of dwelling occupied by a household matters for its 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: 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 percentage of households residing in formal housing over time. In every year White

and Asian households have the highest percentage in formal housing, followed by Coloured households, with a

lower percentage for African households.



9

Figure 3: Percentage residing in formal housing

There is slight change over time in the percentage of households that reside in formal housing, overall or

by population group, except for a small decline in the percentage recorded as in formal housing after 2004. We do

not think that the reported decline in households 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 for 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 consider that for a household to have a middle

class standard of living it needs to reside in formal housing.

4.2 Telephone in the Household

Access to a telephone provides a means of communication for household members for both personal and

business purposes. It is more convenient 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 percentage of households which had a landline phone or in which a household member

had a cellphone. The increase over time is impressive. The percentage with a landline or with a household member

with a cellphone more than doubled for South African households as a whole between 1998 and 2006, more than

quadrupled for all Africans, and increased by almost twenty fold 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

and Asian households; over 85% of White households and over 81% of Asian households had a landline or a

cellphone in the household at all dates. After 1998, the percentage of Asian households with a cellphone or a

landline was noticeably lower than for White households.



10

Figure 4: Percentage with a landline in dwelling or a household member with a cellphone

The increase in phones in African households is overwhelmingly due to the spread of cellphones. Figure 5

shows the percentage of households that had a landline telephone in the dwelling.7 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 cellphone.

Figure 5: Percentage with a landline telephone in dwelling

7 No data are shown in Figure 5 for 1999 because in that year the October Household Survey only asked whether the given household had a landline or a cellphone and did not ask about landlines and cellphones separately.



11

In 1998, 95% of all households that had either a landline or a cellphone had a landline; in 2000 only 69% of

households with a landline or a cellphone had a landline. The spread of cellphones and the resulting improvement

in telephone access in rural areas was the result of the success of commercial cellphone companies.

There are several reasons for the popularity of cellphones. Firstly 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 cellphones for

African households, especially for rural African households. Also a cellphone can be purchased just by having the

necessary money to pay the cost. Even without the credit necessary for a cellphone contract, pay-as you-go

schemes allow purchase of cellphone 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 (The Herald, 2003). Also, in many rural areas, telephone

landline service was not available, in 2006, 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.

Note that in Figure 5 the percentage 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. An indication that, some households that had a landline at one time let that landline go once

at least one household member had a cellphone, whether due to the greater convenience of cellphones, the lower

operating cost of cellphones or some other reason. Press reports indicated that many households have terminated

their landline phone services due to the greater convenience and lower cost of cellphones and that a major reason

why some households have retained their landlines has been to gain fast internet access (The Times,14 June

2007).

Figure 6: Percentage of households more than 15-16 minutes from an accessible telephone: 1998 and 2006

32%

66%

16%

28%

0%

20%

40%

60%

80%

All South Africans Rural Africans

1998 2006



12

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

substantially over time. Between 1998 and 2006, the percentage 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.8 This represents a substantial improvement in telephone access. The

proliferation of cellphone points (commercial establishments where time on a cellphone can be purchased),

including in rural areas, has almost certainly played a major role in the improvement of access.

4.3 The Main Lighting Source – Electricity and Candles Electricity is one of the most efficient energy sources. With a secure connection and in the absence of

power outages, it is always available. Also, unlike sources such as candles, wood or paraffin, it does not add to air

pollution in the vicinity of the household. Candles are the most inconvenient and unsafe lighting source. In this

section we look at the trends in use of electricity for lighting and as well as the use of candles for lighting in South

Africa since 1998.

Major efforts were made through the Reconstruction and Development Program (RDP), driven by Eskom to

extend electrification to an ever increasing proportion of South African households. The government’s

Electrification Programme set a target to have 2.5 million households newly connected to an electricity source by

2000 and to ensure access to electricity for all South African homes by 2012 (ESKOM, 2007).

Figure 7 shows the percentage of households in which electricity is the main lighting source. It is clear that

the percentage of households with electricity as the main lighting source has risen considerably for all groups

except for White and 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 percentage 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 rural areas

(ESKOM, 2007). By 2005, for Coloured households the percentage with electricity as the main lighting source was

nearly as high as for White and Asian households.

In 2001 the government implemented the free basic electricity policy which allowed all households to

receive free 50kWh of electricity per month. This policy was intended to benefit mostly poor households that could

otherwise not afford electricity (DME, 2007).This program has operated through municipalities and has not been

implemented in all locales.

8 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.



13

Figure 7: Households with electricity as the main lighting source

Table 2 shows among households connected to MAINS electricity, the percentage with access to free

basic electricity in 2004. Free basic electricity has been available to a higher percentage of Coloured households

than to any other group, including urban African households.

Table 2: Among households connected to MAINS electricity in 2004, those that received free basic 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 basic electricity to Coloured households than to other population groups is

related to the greater provision of free electricity in Western Cape than in other provinces. In 2004, 21% of all South

African households reported that they received free basic electricity. The highest provincial percentage 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 convenient lighting

source, candles. Figure 8 shows the percentage of households over time that used candles as the main lighting

source.



14

Figure 8: 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 by 2006, over a 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 percentage of urban African

households using candles exceed 15% nor of Coloured households using candles exceed 11%.

4.4 Electricity or Gas as a Cooking Source

Electricity and gas are a healthy and dependable source of cooking and non-polluting (in the case of

electricity) or pollute at a very low level (in the case of gas) available to households.

Figure 9 shows the percentage 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 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%.



15

Figure 9: Households with electricity or gas as main cooking source

Table 3 shows for 2006, among households in which electricity was the main lighting source, the

percentage 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 3: Among households with electricity as the main lighting source, the percentage using electricity or gas as the main cooking source, 2006 GHS

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 allocation in areas in which

there is free electricity available, households pay for electricity.

Since only 50kWh of free basic electricity is allocated to participating households per month ,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. Bekker et al (2008) also

mention the perceived lack of affordability of electricity by low income households.



16

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 economic 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.9 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 winter than they would if they answered this question in summer.

4.5 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 is one of the major policies of the government

(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 of the Department of Water Affairs and

Forestry (DWAF 1994: 3-4).

Part of the stated mission of DWAF is “…ensuring that water services are provided to all households 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 forms part 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).

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



17

Table 4 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 (TVSA, 2008).

Table 4: 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 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 convenient 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 5 shows for 2005 the percentage 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 5: 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 order 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 5.

Figure 10 shows the distribution of sources of drinking water for all South African households by year. 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 percentage of households with a given source of drinking water is shown, and all sources of drinking water add

to 100%.



18

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 percentage 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 percentage 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 have been a decline in the percentage of households

with the worst source of drinking water (from 16% to 7%) and an increase in the percentage with clean water (from

78% to 87%).

Figure 11: Drinking water source for African households



19

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

respectively. Figure 12 shows a decline in the percentage 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% 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 healthy drinking water source.

Figure 12: Drinking water source for rural African households

Figure 13: Drinking water source for urban African households



20

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 percentage of 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 14 shows the situation for Coloured households. Almost all Coloured households had clean water at

all dates, and there was little change in the percentage 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.



21

Figure 15: Drinking water source for Asian households

Figure 16: Drinking water source for White households

Figure 17 summarises the percentage of households with a tap in dwelling over time by group. There was

little change for any group.



22

Figure 17: Percentage of households with water taps 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 the

other three population groups is vastly different than that for African households, this generalisation 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 percentage with a tap in dwelling was much

higher for households belonging to the three groups – 76% for Coloured households, 90% for Asian households

and 94% for White households.

Figure 18 shows the percentage of households that had a tap either in the dwelling or on the site. There

was on increase 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.



23

Figure 18: Households with water tap in dwelling or on site

Figure 19: Households with clean drinking water

Figure 19 shows the percentage across time with clean drinking water by group. The percentahe of all

South African households with clean drinking water increased from 83% in 1998 to 89% in 2006. The percentage 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 percentage 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.



24

Figure 20 shows the percentage of households over time in the worst water category, water from 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 percentage 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: Households with drinking water from stream dam or other source

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 percentage 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.

4.6 Type of Sanitation 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. Proper 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 associated with inadequate sanitation were made clear from the 2001 cholera

outbreak in KwaZulu-Natal (DWAF, 2001a).



25

The leading policy on both the provision of clean drinking water and adequate sanitation, the 2003

Strategic Framework for Water Services of the Department of Water Affairs and Forestry, set 2010 as the date

when all South African households will have access to basic sanitation facilities (DWAF: 2007).

DWAF considers the construction of VIPs (ventilated improved pit latrines) as one of the safer and more

affordable options to the provision of basic sanitation. At an estimated initial capital cost of around R1000 per

household, VIPs are substantially better than ordinary pit toilets, having a lower odour level and providing greater

protection from insects. DWAF characterise VIPs as an “appropriate and adequate basic level of sanitation

service” (DWAF, 1994: 15). It 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, VIPs are as safe for public health as more complex

and expensive sewage systems. VIPs need to be emptied every five years (DWAF, 2002), there can be problems if

they are not emptied on schedule and subsequently overflow. Overflowing of waste counters many of the

advantages of VIPs 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 construction and maintenance 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 some urban African households. However the government has long

considered bucket toilets unacceptable and major efforts have made to eliminate their use (DWAF, 1994: 15;

DWAF, 2008). 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 6 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 considered as decent

sanitation in terms of health effects and keeping water free of contamination by human waste.

Table 6: Classification of type of sanitation according to the categories in the surveys 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



26

Figure 21 shows the percentage distribution of type of sanitation for all South Africans over time.10 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 percentage of households

using the worst type of sanitation, bucket other none, an annual rate of decline of 6%. The percentage of

households using a VIP increased from 6% to 10%, almost doubling.

Figure 21: Type of sanitation for all South African households

0%

20%

40%

60%

80%

100%

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 22: Type of sanitation for African households

0%

20%

40%

60%

80%

100%


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

10 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.



27

Figure 22 shows the sanitation 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 percentage in the worst sanitation

category and an increase in the percent with decent sanitation. There was also a substantial increase in the

percentage of African households with a flush toilet in the dwelling.

Figure 23: Type of sanitation for rural African households

0%

20%

40%

60%

80%

100%


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 24: Type of sanitation for urban African households

0%

20%

40%

60%

80%

100%


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



28

For rural African households there was small increase in the percentage with a flush toilet in the dwelling

and almost no change in the use of any other category of flush or chemical toilet. Throughout the period, the

percentage 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 percentage 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, increasing 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 VIPs

in rural African households and some increase in urban African households. Figure 25 shows the percentage of

urban African and rural African households with VIPs 1999-2006.11 It is clear that the bulk of VIP installations were

in rural areas.

Figure 25: Rural and urban African households with a VIP

0

5

10

15

20

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

The cost of a basic VIP is less than R1000 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 VIPs has required a substantial

financial investment, mainly from various parts of government. A number 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. The VIP construction programs

included both households (c.f. Tshikhudo, 2008) as well as for facilities such as schools (c.f. Mvula Trust, 2006). 11 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 percentage of African households with a VIP in 2006.



29

Figure 26 shows the percentage 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 percentage of rural African households with any of these three types of sanitation. The

percentage of all rural households in the three sanitation categories varied slightly 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 percentage of rural African households with no

sanitation or a bucket toilet was almost completely countered by the increasing percentage 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 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 percentage of all urban African

households in the three sanitation categories generally declined, since the percentage 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 percentage of rural African households in the category “Bucket Other None” was 58% of the

1998 value, and the percentage 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 26: Rural African households with a VIP, a non-ventilated pit toilet, a bucket toilet or no sanitation



30

Figure 27: 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 for rural African households, urban African households, and all South African households.12

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.

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

7.93.5

12.1

0.4

3.528.0

0

5

10

15

20

25

30

Rural Africans Urban Africans All South Africans

NoneBucket

12 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.



31

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

4.3 3.0

15.68.3

1.1

2.8

0

5

10

15

20

25

30

Rural Africans Urban Africans All South Africans

NoneBucket

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

percentage 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.

The decline in the percentage of households that use bucket toilets is impressive. 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 frequency with which they are

emptied (Thomas et al., 1999: xiii). For all South African households, Table 7 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.

Table 7: 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%

In 2003, a target was set to eliminate all bucket toilets in formal settlements built after 1994 by December

2007 (DWAF, 2003: 6). Although there has been substantial reduction in bucket toilets, they had not been

eliminated by 2007 (Mail & Guardian; 2007). Another challenge is the continued use of bucket toilets in informal

settlements. A large number of migrants from rural to urban places often take up residence in poor informal and

often illegal settlements, which results 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).



32

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 an area 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 for Coloured households. As for urban African households, there was a

large decline in the percentage of Coloured households in the worst category of sanitation; the percentage in 2006

was 30% of the value in 1998. The overall sanitation 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.

Figure 30: Type of sanitation for Coloured households

0%

20%

40%

60%

80%

100%


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



33

Figure 31: Type of sanitation for Asian households

0%

20%

40%

60%

80%

100%


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 shows the sanitation 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 32: Type of sanitation for White households

0%

20%

40%

60%

80%

100%


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



34

Figure 33 shows the percentage 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 percentage 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

percentage of White households, Asian households and Coloured households with a flush toilet in the dwelling is

much higher than for African households.

Between 1998 and 2006, there was virtually no change in the percentage 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 33: Households with flush toilet in dwelling



35

Figure 34: Households with a flush toilet, chemical toilet or VIP

Figure 34 shows the percentage of households with decent sanitation – a flush toilet, chemical toilet or VIP,

regardless of its location. This percentage increased between 1998 and 2006 for all groups except White

households and Asian households, for whom it was close to 100% at all dates.

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

the possible contamination of the water supply 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 percentage with a flush toilet in

the dwelling. Other notable changes in sanitation are a decline in the percentage 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 percentage of households using the worst type of sanitation, bucket or none. The

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



36

Figure 35: Percentage of households with a bucket toilet or none

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

percentage 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.

5. Standard of Living Groups

We divide households into four standard of living groups (SOL). These four groups are described in Table

8. 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 – one

of the top four categories of sanitation.

The top two groups described in Table 8 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. One of the goals of the government is 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.



37

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 cellphone.

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

clean water and proper sanitation, 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 but not middle

class 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 8: 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 cellphone

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 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.



38

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 -- also changed slightly, 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: Distribution among SOL groups for African households



39

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 percentage

of households which have a middle class standard of living almost doubled over time – an increase of 93%. The

percentage 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 percentage in the middle low category, from 36% to 43%.

Figure 38: Distribution among SOL groups for rural African households

Figure 39: Distribution among SOL groups for urban African households



40

The percentage 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 middle low category increased. This shift into the middle low SOL category was

mainly due to the construction of VIP’s discussed earlier. The percentage of rural African households with a middle

class SOL more than quadrupled, but only from about 0.5% to about 2%.

The percentage 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 percentage 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: Distribution among SOL groups for Coloured households

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

percentage 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 percentage 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.



41

Figure 41: Distribution among SOL groups for Asian households

Figure 42: Distribution among SOL groups for White households



42

6. Standard of Living Patterns by Province

South Africa is a large country, any consideration of levels and trends in standard of living would be

deficient without some examination of standard of living by province. In this section we compare the standard of

living within provinces for 1998 and 2006. The results for 1998 use an estimation of the presence of VIPs based on

the relation of VIPs to other factors in 1999.13

Table 9 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 and Asian households in the surveys.

Table 9: Distribution of population groups by province, 2006 African Coloured Asian White Total

Western Cape 30,2 44,7 0,3 24,8 100

Eastern Cape 85,9 6,2 0,2 7,7 100

Northern Cape 44,3 41,8 0,9 13,1 100

Free State 83,6 4,1 0,1 12,2 100

KwaZulu-Natal 80,0 2,0 8,8 9,3 100

North West 87,7 1,8 0,6 10,0 100

Gauteng 77,0 2,7 2,1 18,2 100

Mpumalanga 91,2 0,9 0,4 7,5 100

Limpopo 95,9 0,1 0,5 3,5 100

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.

Figure 43 shows the percentage of households with a middle class standard of living by province in 1998

and 2006. This percentage increased between 1998 and 2006 in every province except Western Cape, but

Western Cape had the highest percentage with a middle class standard of living at both dates, followed by

Gauteng.

13 See the Data Appendix for further discussion of the estimation of households with VIPs 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.



43

Figure 43: Households with a middle class standard of living by province, 1998 and 2006

50.5

17.2

29.4

23.3

24.4

16.4

34.7

16.8

8.0

51.8

12.4

24.9

18.0

24.1

12.3

34.6

15.8

4.8

0 20 40 60 80 100

Western Cape

Eastern Cape

Northern Cape

Free State

KwaZulu Natal

North West

Gauteng

Mpumalanga

Limpopo

1998 MiddleClass SOL


It is useful to keep in mind the differences among provinces in the percentage of African households that

reside in urban areas. As shown in Figure 44, provinces differ greatly in the percentage of all African households

that reside in urban places, ranging from 14% in Limpopo to 96% in Gauteng and 97% in Western Cape.

Figure 44: African households that reside in urban places by province, 2006

97

39

83

78

51

38

96

44

14

0 20 40 60 80 100

Western Cape

Eastern Cape

Northern Cape

Free State

KwaZulu Natal

North West

Gauteng

Mpumalanga

Limpopo

We noted that virtually all middle class African households are in urban areas. Figure 45 shows the

percentage of urban African households that had a middle class standard of living in 1998 and 2006. Figure 45

shows an increase in percentage 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 percentage with a middle class standard of living, for urban Africans in 2006, Western Cape had the lowest

percentage with a middle class standard of living. Also, in every province except for Western Cape, the percentage

of urban Africans with a middle class standard of living increased between 1998 and 2006.



44

Figure 45: Urban African households with a middle class standard of living by province, 1998 and 2006

15.5

23.5

20.1

16.3

23.6

22.2

22.3

21.5

22.7

15.9

8.9

12.8

8.5

12.4

14.0

14.3

16.0

22.6

0 5 10 15 20 25 30

Western Cape

Eastern Cape

Northern Cape

Free State

KwaZulu Natal

North West

Gauteng

Mpumalanga

Limpopo



Perhaps the percentage 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 percentage 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 percentage 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 with 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. Equally,

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 prospect at the destination as those with favourable characteristics. 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.



45

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 also only migrated to Gauteng if they knew they had very good economic prospects there.

Figure 46 shows the percentage of the households in each province in 1998 and in 2006 which had a

decent standard of living. In both 1998 and 2006, the percentage of households with a decent standard of living

was the highest in Western Cape and in Gauteng and the percentage with a decent standard of living was the

lowest in Eastern Cape and in Limpopo. The percentage 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 regard.

For 1992-1996, Gauteng and Western Cape were the most popular destinations for interprovincial migrants

(Kok et al., 2003: 37), and according to a 2001-2002 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 percentage of households with a decent standard of living in those provinces.

Figure 46: Households with a decent standard of living by province, 1998 and 2006



46

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

Figure 47 shows the percentage 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 percentage with a decent standard of living increased in

every province between 1998 and 2006. African households in Limpopo had the lowest percentage with a decent

standard of living, with African households in Eastern Cape faring somewhat better. A higher percentage of African

households in KwaZulu-Natal enjoyed a decent standard of living.

Figure 48 shows the percentage 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 percentage 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.

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

percentage with a decent standard of living, with the highest being rural African households in Western Cape.

However, the percentage of rural Africans with a decent standard of living is 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.



47

It is interesting to compare the percentage 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 are a

higher percentage of rural African households than urban African households with 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 are more likely to settle 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 48: Urban African households with a decent standard of living by province, 1998 and 2006

Figure 49: Rural African households with a decent standard of living by province, 1998 and 2006



48

Figure 50 shows the percentage of households by province that was in the worst SOL group in 1998 and

2006. In most provinces this percentage 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 percentage declined between 1998 and 2006.

Eastern Cape and KwaZulu-Natal stand out in Figure 50 for having by far the largest percentage 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: The percentage of all households in low SOL group by province, 1998 and 2006

Figure 51: The percentage of rural African households in low SOL group by province, 1998 and 2006



49

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

standard of living group). This percentage is by far the largest in Eastern Cape, followed by KwaZulu-Natal,

although there was a substantial decline in both provinces between 1998 and 2006. There is also a slight increase

from 0.7 to 1.5 for Northern Cape.

It was clear in the earlier examination of type of sanitation that the construction of VIPs has played a

substantial role in improving the standard of living of rural African households. We look next at the changing role of

VIPs across provinces.

Figure 52 shows the percentage 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 percentage with a VIP more than tripled in Eastern Cape, Free State and KwaZulu-Natal.

Figure 52: Rural African households with a VIP, 1999 and 2006

Figure 53 looks at VIPs in a different way. It shows the percentage of all pit toilets (whether ventilated or

not) which were VIPs. This figure indicates how far a province has yet to go to replace all unventilated pit toilets

with VIPs.

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 VIPs, whereas Gauteng and Limpopo have the farthest to go

in replacement of unventilated pit toilets with VIPs.



50

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

Of course the aim is not just to replace unventilated pit toilets with VIPs 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 percentage with no sanitation, with unventilated pit toilets and with VIPs, 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 percentage 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 VIPs. However, Figure 55 makes clear that the situation in these two provinces is

vastly different, 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.



51

Figure 54: Rural African households by province with no sanitation, a pit toilet or a VIP, 1999

Figure 55: Rural African households by province with no sanitation, a pit toilet or a VIP, 2006

Figure 56 shows by province the relation between the percentage of urban African households with a

middle class standard of living and the percentage 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.



52

Figure 56: Urban African households with a middle class SOL plotted against rural African households with a low SOL by province, 1998

WC

EC

NCFS

KZ

NW

GP

MP LM

0

5

10

15

20

25

30

35

0 5 10 15 20 25

% Urban African HH Middle Class SOL

%Ru

ral A

frican

HH Low

SOL

Figure 57: Urban African households with a middle class SOL plotted against rural African households with a low SOL by province, 2006

WC

EC

NCFS

KZ

MP NW

GPLM

0

5

10

15

20

25

30

35

0 5 10 15 20 25

% Urban African HH Middle Class SOL

% Rural African

HH Low

SOL



53

As we saw in Figures 45 and 51, we see in Figures 56 and 57 that between 1998 and 2006 the percentage

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 percentage 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 percentage

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 percentage of rural African households with a low standard of living declined to 0.60 and

0.55% of the 1998 value respectively.

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

smaller decline in the percentage 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 African households, 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 percentage of the urban African population with a middle class standard of living. Western Cape

and Gauteng had the highest percentage of all households with a middle class standard of living, but were not high

in the percentage 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 percentage of households in the low standard of living category along with the percent with the

worst category of drinking water and the percentage 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.



54

Figure 58: Rural African households in Eastern Cape, with a lowest SOL, in the worst water category and sanitation category

0

20

40

60

80

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 59: Rural African households in KwaZulu-Natal, with a low SOL, in the worst water category and sanitation category

0

20

40

60

80

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

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 in the percentage of households in the

worst water category and in the worst sanitation category were similar.



55

Figures 60-63 show the percentage of urban African households with a middle class standard of living as

well as the percentage of urban African households with a tap in dwelling and the percentage 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.

Figure 60: Urban African households in Eastern Cape with a middle class SOL and best water and sanitation category

0

10

20

30

40

50

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 61: Urban African households in KwaZulu-Natal with a middle class SOL, best water and sanitation category

0

10

20

30

40

50

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 38.7Flush in Dwelling 28.7 30.3 38.1 38.5 39.4 40.2 38.2 42.1 47.2

1998 1999 2000 2001 2002 2003 2004 2005 2006

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 percentage of all urban African households with a flush toilet in the dwelling, the rate of

improvement is much more rapid in these two provinces than for all urban African households.



56

Figure 62: Urban African households in Western Cape with middle class SOL, best water and sanitation category

0

10

20

30

40

50

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 63: Urban African households in Gauteng with a middle class SOL, in the best water category and in the best sanitation category

0

10

20

30

40

50

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

The picture for Western Cape shown in Figure 62 is very different. Over time the percentage of urban

African households with a middle class standard of living, as well as the percentage of households 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.



57

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.

It is likely that inter-provincial 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 inter-provincial

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 percentage 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 fare 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 circumstances of 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).

7. Population Group Composition of Standard of Living Groups

In Figures 37-42 we looked at the percentage distribution of population groups among the four standard of

living categories, 1998-2006. Another interesting question is: What is the percentage distribution of population

groups within 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.



58

Figures 64 and 65 answer this question. These show the percentage 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 percentage 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.

The most striking result in Figures 64 and 65 is the increase in the percentage 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 percentage 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 percentage 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.14

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

14 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.



59

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

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 percentage 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 0.3 times as likely have a

middle class standard of living as their overall population percentage 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 percentage. As shown in Figure 65,

by 2006, African households were 0.51 times as likely (one-half as likely) to have a middle class standard of living

as their population percentage would have implied. The values for other groups 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 percentage 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 Standard Bank (2005).



60

Figure 66: Percentage of households with a middle class standard of living (top SOL group)

0

20

40

60

80

100

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

Coloured Asian White All South Africans

Figure 66 shows the percentage 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 percentage of White and 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 percentage 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 rural areas. 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. In

2006 84% of rural White households had a flush toilet in the dwelling connected to a septic tank.



61

Despite the increase in the percentage of urban African households with a middle class standard of living

shown in Figure 66, not all middle class households experience the same economic conditions. Figure 67 shows

the percentage of middle class households in 2006 whose monthly household expenditures exceeded R2500 and

the percentage whose monthly household expenditure exceeded R5000.15 This is not shown for rural Africans, for

whom less than 2% of households had a middle class standard of living in 2006.

Figure 67: Households with middle class SOL with monthly expenditure greater than R2500 and R5000, 2006

32

14

53

23

56

27

73

4552

28

0

20

40

60

80

100

>R2500 per month >R5000 per month

Urban Africans Coloured Asian White All South Africans

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 percentage 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 households aspire. There was an increase 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.

15 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.



62

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 while 46% had a flush toilet on the site. 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.

Figure 68: Households with a decent standard of living (Top two SOL groups)

Figure 69: Households with a decent standard of living with monthly expenditures greater than R800, 2006



63

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

threshold. Figure 69 shows the percentage 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.

Forty-two percent of rural African households with a decent standard of living had monthly expenditure of

over R800, but this was 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.

Figure 70 shows the percentage 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 percentage 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.

Figure 70: Percentage of households in Low (worst) SOL group

0

2

4

6

8

10

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

Coloured Asian White All South Africans

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 SOL group in comparison to all

rural African households in 2004.



64

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.

8. Concluding Comments

Since 1998, some aspects of the lives of South Africans have definitely improved. An increasing

percentage of all households have: (1) access to clean water, (2) access to a telephone, either through a landline

in the dwelling or through a cellphone possessed by a household member, and (3) electricity as the major source of

light. A declining percentage of South African households have no sanitation or use a bucket toilet for sanitation.

Also, an increasing percentage of urban African households and of Coloured households have arguably entered

the middle class as indicated by standard of living measures. Also the percentage of that have a decent standard of

living has increased over time all groups. The overall percentage of South African households with a middle class

standard of living, though, has changed trivially, 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 over time. Some might expect that the situation of

Coloured households and of urban African households would be similar. For the most part, they are not 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: South African households with a poor standard of living: 1998 and 2006



65

Figure 72: Rural African households with a poor standard of living: 1998 and 2006

Figure 73: 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 percentage 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 government effort and intervention, while other improvements probably

have little to do with government 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 cellphone

companies. The government has worked to improve the quality of the drinking water supply, and has worked to

eliminate bucket toilets, but the reduction in the percentage of rural African with no sanitation probably is the result

of households or rural community groups digging holes for pit latrines.



66

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. Although much could be improved about access to better

sanitation for urban Africans, it seems clear that the sanitation and other basic aspects of the lives of rural Africans

are in more dire 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 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 VIPs 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 relative little variation in the percentage 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 disproportionately higher percentage 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.



67

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 Bekker Bernard, Eberhard Anton, Gaunt Trevor, and Marquard Andrew. 2008. “South Africa’s rapid electrification programme: Policy, institutional, planning, financing and technical innovations” Energy Policy 36 (2008) 3115-3127. 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 hhttp://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 andservicelevels.Availableat 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. The Herald, 13 November 2003 “Chaos as new magistrates courts left without phones.” 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.



68

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. Department of Water Affairs and Forestry (DWAF). 2008. The national sanitation bucket replacement programme: Lessons learnt. Available at http://www.wsscc.org/fileadmin/files/pdf/For_country_pages/South_Africa/WIN%20SA%20Sanitation%20Bucket%20Replacement%20March%202008.pdf Accessed 10 September 2009. 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. TVSA, 14 May 2008 “Killing waters,”, 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.



69

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. Nieftagodien, Sihaam, and van der Berg, Servaas. 2007. “Consumption patterns and the black middle class: The role of asets,” Stellenbosch Working Paper Series No. WP04/2006. Available at http://www.ekon.sun.ac.za/wpapers/2007/wp022007 Accessed June 11, 2009. 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.



70

South Africa. 2005. South Africa Millennium Development Goals Country Report 2005. 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.



71

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 percentage 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 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 percentage reported for

1997 was 6.8 and 5.4 for 1999. Even in 2006, the reported percentage 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 VIPs

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 minimise 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 VIPs, 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 percentage of groups in the second

standard of living category and the percentage in the third standard of living category. This was important for

estimating the percentage of households with a decent standard of living in 1998.



72

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 percentage of a group with a given

characteristic, such as the percentage of all South African households in formal housing, can be tested by

constructing 95% confidence intervals around the estimates of the percentage 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 percentage lies.

The width of the confidence interval depends on the magnitude of the percentage and on the number of cases

(households) in the group considered.16 If the 95% confidence intervals for the two years do not overlap, then it is

reasonable to conclude that the percentage of households with a given characteristic actually changed between the

two years.

Figure 3 shows the percentage 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 percentage 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 percentage 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 percentage is partly because the number of households in each

group in each year is very large.

Just because the change in a percentage 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 0.2%, 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 comprise 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.

16 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.



73

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 problems 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 earlier

years.

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.