Section 1

Contents

Background
Reasons for conducting an income and expenditure survey
The focus of this report

Introduction

Background

Political democracy in South Africa is, after many years of struggle, at last a reality. The new constitution (Act 108 of 1996) is founded on a set of values which embody non-racialism, non-sexism, respect for human dignity, equality, human rights and freedom for all. Explicit discrimination and denial of human rights, which formed the basis of the apartheid past, has been rejected by most South Africans.

Despite these recent advances in democracy, socio-economic deprivation and profound contrasts in life circumstances along racial, urban-rural and gender divides, persist. Although South Africa is a middle-level income country, comparable with Brazil, Chile, Malaysia, Poland, Thailand and Venezuela (World Bank/SALDRU, 1995), it is characterised by gross inequalities, partially the legacy of apartheid policies.

The government is committed to improving the life circumstances and quality of life of all South Africans, particularly those who were previously disadvantaged. To meet this challenge, and to plan and implement change, a variety of role-payers – government, the private sector, trade unions and other institutions of civil society – require accurate information on a range of aspects of South African life. The Central Statistical Service (CSS), with its vast numbers of data collections, is the most appropriate agency to provide such data.

This CSS report addresses the need for information of a particular type. It is a summary of the main findings of the October 1995 income and expenditure survey (IES), and describes the large differences in income distribution and expenditure patterns among South African households.[In this report, the term household refers to all people who live together for at least four days a week, who eat together and who share resources.]

Reasons for conducting an income and expenditure survey

There are numerous ways of collecting information on household income and expenditure. For example, people in selected households may be asked to keep receipts of all their purchases, or keep a diary of all expenditure over a specified time period. In addition to, or instead of these methods, a household survey can be conducted. Due to the relatively low level of literacy in South Africa, and the associated difficulty of record-keeping for many people, the CSS chose the route of utilising households for its October 1995 income and expenditure survey.

Through the IES, the CSS determined the proportion of expenditure in an average household, or in sub-groups of various types of households, that went towards purchasing each of a variety of goods and services, such as food, housing, transport and recreation. On the basis of this information, weights for each item of expenditure, based on household averages, or on other classification variables, were calculated.

Calculation of the CPI

The main purpose of the 1995 IES was to collect base-line information on household income and expenditure patterns for re-weighting the consumer price index (CPI).

In South Africa, the CPI is generally calculated in two stages.

Stage one
Firstly, information is collected from households in which questions are asked on:

Thereafter, the total expenditure of all households in the sample during the specified time period is raised to represent expenditure in all households in the country. From this new total, the CSS calculates the average annual expenditure per commodity or service, per household.

The CSS can also calculate the total annual expenditure, and average annual expenditure for each type of commodity or service, for various sub-groups of households – very low, low, middle, high and very high expenditure groups, for example. This can also be done for households in diverse geographic areas in different parts of the country, which can be broken down into metropolitan (metro), urban and rural areas.

In the past, the IES was conducted only among households in what were regarded as the 12 main urban areas of South Africa. [The 12 areas are the Cape Peninsula, Port Elizabeth-Uitenhage, East London, Kimberley, Bloemfontein, Free State Goldfields (Welkom-Virginia-Odendaalsrus), Durban-Pinetown, Pietermaritzburg, Pretoria-Centurion-Akasia, Witwatersrand, Vaal Triangle (Vereeniging-Van der Bijl Park-Sasolburg) and Klerksdorp-Stilfontein-Orkney.] Smaller towns and rural areas were excluded from the sample. But, in 1995, the whole country was included in the survey for the first time. This is discussed in a later section.

Stage two
In the second stage of calculating the CPI, the CSS collects the prices of all items of expenditure from different outlets.

In the past, the prices of goods and services were obtained in selected retail outlets in the same 12 main urban areas of the country where the household survey was conducted, but these outlets have now been extended, as discussed in the following section.[Two extra urban areas were added in 1994 for the collection of retail prices, even though no information was available on buying patterns in these areas, to ensure coverage of at least one retail outlet in all of the nine new provinces of South Africa. The new areas are Nelspruit, Witbank and Pietersburg, to cover Mpumalanga and the Northern Province.]

Changes in the calculation of the CPI, based on the 1995 IES

The CSS has recently introduced, and is continuing to initiate, a series of changes in the calculation of the CPI, in both stage one and stage two.

Stage one changes
The 1995 IES differed from previous household surveys of its kind in South Africa, since it was a countrywide survey covering metro, urban and rural areas, rather than a more limited sub-set of households in 12 major metro/urban areas of the country previously referred to. By extending the sample to include the whole country, a clearer indication of the life circumstances of all South Africans in all parts of the country can now be inferred.

Previously, only three income categories were used for the calculation of the CPI, with the lowest category including 78% of African households in the 12 main urban areas. In the 1995 IES, five approximately equal income groups (very low, low, middle, high and very high), each containing approximately 20% of households, and five expenditure groups, based on quintiles,[Quintiles divide a data set into five approximately equal groups, each group containing about 20% of the total number of households.] were derived. For reasons which will appear later in this report, income quintiles were used to describe differences in the distribution of income among various categories of households, for example households in urban versus households in rural areas; while expenditure quintiles were used to identify expenditure patterns among households falling into very low, low, middle, high and very high expenditure categories.

The effect of these changes in the 1995 IES sample, and the increase in the number of income categories, is that the country now has a clearer indication of the buying patterns of households ranging from the very poor to the very wealthy, living in metro, urban and rural areas.

Stage two changes
In the collection of information from retail outlets, the CSS now includes small towns. Since March 1997, it has published an inflation rate for small-town areas in the provinces, in addition to the major urban areas covered hitherto. This has involved a 50% increase in the number of price-questionnaires issued and processed.

The importance of calculating a rural CPI

The CSS cannot, at present, collect prices from outlets in rural areas: this type of collection is very expensive and the necessary funding is not available. However, if finance can be raised, the CSS plans to measure and publish a rural CPI. As a large proportion of South Africa’s households are situated in non-urban areas, this is of obvious importance. A rural CPI will enable decision-makers to obtain as complete a picture as possible of income and expenditure patterns, and the effects of inflation, in all parts of the country, rather than just in urban areas, as was previously the case.

This is of major importance: although households in non-urban areas may spend relatively little compared to those in urban areas, inflation may have a greater effect on the ability of rural households to survive where incomes do not keep up with inflation. More extensive information on spending patterns in rural areas will facilitate planning, programme development and poverty monitoring at all levels of government – national, provincial and local.

The focus of this report

In describing the findings of the 1995 IES, this report paints a picture of how income is distributed in South Africa by using the five income quintiles. It also examines expenditure patterns in households falling into very low, low, middle, high and very high expenditure groups.

The race and gender of the head of household,[The head of household is defined here as the person who is the main breadwinner in the household, or if the main breadwinner does not live in the household, for example, if he or she is a migrant worker, the person who assumes responsibility for decision-making in the household] and other variables such as province and the location of the household in an urban or non-urban milieu, are used as explanatory variables to describe income and expenditure patterns.

The research process

The questionnaire design
The 1995 IES questionnaire, in the same vein as the previous one, contains questions about all sources of household income. It also covers the purchase of a wide variety of products and services, including new items such as cellular telephones.

Drawing a sample
Two surveys, namely the CSS’s annual October household survey (OHS) and the IES were run concurrently during October 1995.

The fieldwork
Throughout South Africa, information was collected through face-to-face interviews in the 30 000 households which formed the sample. Field workers first administered the OHS questionnaire, and returned at a slightly later date to administer the questionnaire for the IES.

Data capture
Data capture of both the 1995 OHS and the IES took place at the head office of the CSS. Where possible, this process involved linking the information contained in the 1995 OHS with that contained in the IES.

Raising the sample to the population
Data collected on households were raised to the estimated number of households in the country in the various provinces, according to the proportions found in urban and non-urban areas in the 1991 census. All further discussions in this report are based on these raised figures.

Calculating new weights for the CPI
For the sample as a whole, weights were allocated for each item of expenditure according to the proportion of annual disbursements for that particular item by the average household. In addition, the same procedure was followed for households in each quintile.

Identifying income and expenditure quintiles
Two different sets of quintiles were obtained – those based on annual household income and those based on annual household expenditure.

To calculate income quintiles, information obtained on all sources of annual income for each household was used. This total annual income was divided, as closely as possible, into five groups or income categories, as indicated in Table 1. To calculate annual expenditure quintiles, the same procedure was used.

Table 1: Annual income and expenditure quintiles

 

Quintile 5

(bottom quintile)

Range

Quintile 4

Range

Quintile 3

Range

Quintile 2

Range

Quintile 1

(top quintile)

Range

Income

R400-6 868

R6 869-12 660

R12 691-23 940

R23 941-52 800

R52 801 +

Expenditure

R332-6 340

R6 341-11 589

R11 590-21 908

R21 909-49 497

R49 498 +

Undeclared income and expenditure in the process of identifying quintiles was dealt with in the following way:

Data analysis and report writing
After data processing, a series of tables and cross-tabulations were obtained. This summary report is based on those tables.

Raising factors and weights used for analysis of the 1995 IES
As already indicated, estimates using the 1991 census formed the basis for the calculation of raising factors and weights.

However, preliminary estimates based on the October 1996 population census have shown that the population of 37,9 million people in South Africa is smaller, and urbanisation more rapid, than was previously thought. These preliminary estimates are based on a limited set of variables from Census ‘96. For example, the CSS does not as yet know the number of households in the country, only the number of questionnaires that were completed during Census ‘96. Since this particular data set looks specifically at household incomes and expenditure, it is not at this stage possible to take the new 1996 census-based population estimates into account. The numbers and percentages in this report should, therefore, be regarded as indicative of patterns and trends, rather than as definitive numbers or proportions.