Section 4

Contents

Introduction
Making the two surveys comparable
Income comparisons
Expenditure comparisons
Measures of income inequality
Summary

Comparing the surveys of 1990 and 1995

Introduction

In this section, we compare the findings of the two most recent income and expenditure surveys, namely the survey of household expenditure of 1990, as reported by the Central Statistical Service in 1992, and the income and expenditure survey of 1995, on which this report is based. We examine similarities and differences in 1990 and 1995 with regard to the following:

Making the two surveys comparable

The reader will recall from Section 1 of this report that the areas in which the 1990 and the 1995 surveys were conducted differed from each other. In 1990, the survey was conducted amongst households in the 12 main urban areas of South Africa, while in 1995, it was conducted in all parts of the country. The two studies are therefore not directly comparable.

In order to make the two studies comparable, the following steps were taken:

While these comparisons do indicate trends, the findings for 1995 should be treated with caution, since the sample size in the 12 main urban areas tended to be rather small, particularly for coloured and Indian households.

Income comparisons

In the 12 main urban areas of the country, we firstly look at average annual household incomes in 1990, inflated to 1995 values, and compare these to the 1995 values. We also study average annual incomes in each quintile for African, coloured, Indian and white households, and the proportion of households in each quintile by race.

Average annual household incomes in 1990 compared with 1995

Table 7 describes the average annual household income in each population group in the 12 main urban areas of the country, adjusted for inflation, in 1990 and 1995. It shows that, on average, the annual incomes of African, coloured and Indian households living in these 12 areas rose substantially, while the incomes of white households decreased slightly.

It is also noteworthy that the average annual incomes in the 12 main urban areas are generally significantly higher than those in other parts of the country, as indicated in Chapter 2. For example, in October 1995, in all parts of the country, the average annual household income was R41 000, compared with more than double this amount (R83 000) in the 12 main urban areas.

Table 7: Distribution of household income by population group in 1990 and 1995 in the 12 main urban areas of the country

Year

Income in Rands (000)

 

African

Coloured

Indian

White

All groups

1990:

Mean

12

22

26

69

39

1990 inflated to 1995 values:

Mean

20

38

45

117

67

1995:

Mean

48

64

87

113

83


Average annual income in each quintile in 1990 and 1995 by race

Annual household incomes by quintile, with each quintile containing approximately 20% of households, were recalculated for the 12 main urban areas for both 1990 and 1995. This was necessary for the 1990 survey, since the data set had previously been divided into three income categories, not quintiles, for analysis. In 1995, it was also necessary to recalculate quintiles, since in this report national income quintiles were used for analysis, rather than those for the 12 main urban areas. As we have seen, annual average household incomes are substantially higher in the 12 main urban areas than they are in the rest of the country.

The new quintile groups for 1990 and 1995 are indicated in Table 8. The bottom quintile represents those households with very low, and the top quintile those with very high, incomes for the 12 main urban areas for each year.

Table 8: Income quintiles for the 12 main urban areas in 1990 and 1995

Quintiles

1990

1995

Quintile 1 (top)

Quintile 2

Quintile 3

Quintile 4

Quintile 5 (bottom)

R46 800 or more

R22 800-46 799

R12 220-22 799

R6 900-12 219

R6 899 or less

R118 800 or more

R69 360-118 799

R39 600-69 359

R18 360-39 599

R18 359 or less

 

Table 9 shows the average annual income in each quintile by race, using for all races the quintile breaks as defined in Table 8. When comparing the poorest households in 1990 with the poorest households in 1995 (quintile 1), it is noteworthy that, for the main urban areas, average annual income of the poorest 20% had substantially improved. It had gone up from approximately R7 000 in 1990 to R12 000 in 1995 for all races (1990 figures adjusted for inflation).

Table 9: Average annual income in each quintile for 1990 and 1995 by race

Year

Income in Rands (000)

 

African

Coloured

Indian

White

All groups

1990 mean income:

Quintile 1 (top)

Quintile 2

Quintile 3

Quintile 4

Quintile 5 (bottom)



68

30

16

9

4



68

33

17

9

5



78

32

17

9

5



97

35

18

10

5



95

34

17

9

4

1990 mean income inflated

to 1995 values:

Quintile 1 (top)

Quintile 2

Quintile 3

Quintile 4

Quintile 5 (bottom)



116

52

28

16

7



115

56

29

16

8



133

55

29

17

9



165

60

30

17

8



162

58

29

16

7

1995 mean income:

Quintile 1 (top)

Quintile 2

Quintile 3

Quintile 4

Quintile 5 (bottom)



236

88

52

27

12



159

98

53

26

13



170

89

50

33

13



198

93

52

28

13



201

92

52

27

12

 

The average annual income of African households in the top quintile had increased very substantially in 1995, compared with 1990, but only 6% of all African households were in this top category in 1995, compared to 33% of all white households. The relatively small sample size of African households in this top income category means, however, that these results should be treated with caution.

Percentage of households in each quintile in 1990 and 1995 by race of household head

Figure 18 indicates the proportion of households in each income quintile by the race of the head of household in 1990 and 1995 in the 12 main urban areas. It shows that, while there is an overall improvement in average household incomes among African-headed households, as well as in average household incomes in each quintile, inequalities between Africans seem to be increasing. For example, in 1990, 34% of African households were in the bottom income quintile. This proportion increased to 38% in 1995. In 1990, there were only 2% of African households in the top quintile, but this proportion increased to 6% in 1995.

A similar pattern can be seen among Indian households, although this pattern should be treated with caution, because of the small number of Indian households in the 12-area sample of 1995. For example, in 1990, 8% of Indian households were found in the bottom income quintile, while in 1995, this proportion had increased to 15%. At the other extreme, 17% of Indian households were in the top quintile in 1990, compared with as many as 27% in 1995.

Figure 18: Proportion of households in each income quintile by race of household head

 

When looking at the income distribution by quintiles among whites in the 12 main urban areas, proportionately fewer households were found in the top income quintile (33%) in 1995 compared to 1990 (51%), and proportionately more in the lower three quintiles. There is therefore a pattern of increasing equality in incomes in comparison with the other race groups.

Expenditure comparisons

In Table 10, we indicate the proportion of total expenditure of the average African, coloured, Indian and white household on different items of expenditure in 1990 and 1995.[In 1990, expenditure on income tax and insurance was given as one category. Therefore, the 1995 categories were made comparable with those used in 1990, and expenditure on insurance and income tax was combined.]

This table shows that the main items of expenditure in both 1990 and in 1995 were food, housing, income tax, insurance, savings and investment and transport.

Table 10: Percentage of total annual expenditure per item by race in 1990 and 1995

Item of expenditure

African

Coloured

Indian

White

 

1990

1995

1990

1995

1990

1995

1990

1995

 

%

%

%

%

%

%

%

%


Food

Drinks

Tobacco

Clothing

Footwear

Housing

Fuel and power

Furniture and equipment

Household operation

Domestic workers

Medical services and requirements

Transport

Communication

Recreation, sport, etc.

Reading matter

Education

Personal care

Restaurants, bars etc.

Holidays

Income tax, insurance, savings, investments

Other

Total

22

3

2

8

2

6

3

4

3

<1

2

5

1

1

1

2

3

1

1

20


10

100*

20

3

1

5

2

15

<1

6

2

1

4

12

2

2

1

1

3

1

<1

17


2

100*

27

1

2

4

1

14

4

3

2

1

3

6

2

2

1

1

3

1

1

19


2

100*

21

2

1

6

2

18

<1

5

2

2

4

6

2

2

<1

1

3

1

<1

21


1

100*

23

1

1

4

1

15

4

4

2

1

4

7

2

2

1

1

2

1

1

22


1

100*

16

1

<1

4

1

17

<1

3

2

1

5

13

4

3

<1

2

3

1

<1

21


2

100*

12

1

1

3

<1

21

2

2

1

1

3

7

1

3

<1

1

2

1

2

34


2

100*

12

1

1

2

1

20

<1

2

1

2

5

13

2

3

<1

2

2

1

<1

28


2

100*

* Due to rounding totals do not always add up to exactly 100%

 

Measures of income inequality

We now compare incomes in the 12 main urban areas of South Africa in 1990 and 1995, using Lorenz curves and Gini coefficients.

Figure 19 shows that, in the 12 main urban areas, incomes have become less unequal in 1995, compared to 1990. This applies particularly to those in the middle and upper income ranges.

It therefore seems as if the income, and hence life circumstances, of those households in the middle income range is improving more rapidly, while for the poorest in the main urban areas, life circumstances and income are improving less rapidly.

The danger in this pattern is the creation of a large under-class of marginalised people in the main urban areas of the country, at the same time as the life circumstances of those in middle-income range improve. Careful monitoring of the life circumstances of the poorest people over time is therefore necessary for any future policy formulation and implementation.

Figure 19: Lorenz curve: households in the 12 main urban areas, 1990 and 1995

 


Table 11 compares the Gini coefficients in 1990 and 1995 among all households in the 12 main urban areas, and by race.

Table 11: Gini coefficients in the 12 main urban areas in 1990 and in 1995

Variable

Gini coefficient, 1990

Gini coefficient, 1995

All households

0,63

0,55

Population group:

African

Coloured

Indian

White



0,35

0,37

0,29

0,50



0,51

0,42

0,46

0,44

Summary

On average, it seems as if an increase in income among African, coloured and Indian households in the 12 main urban areas of the country is associated with spending proportionately less of the total household expenditure on food, and proportionately more on income tax, insurance, transport and housing. A decrease in income, on average, and a move towards proportionately fewer households being found among the top 20% of earners among whites, is associated with spending proportionately less on income tax, insurance, savings and investments.

Income disparities between households in middle and upper income brackets is decreasing in the 12 main urban areas, but the poorest households continue to have access to an extremely small proportion of all household income. Among African, coloured and Indian households, the gap between poor and more affluent is growing, while it is decreasing among white households.