The development status of women in South Africa: Patterns and progress in the provinces

Abstract

The development status of women in South Africa declined between 1996 and 2001. This study examined whether the decline was pervasive throughout the nine provinces, and whether any development patterns were apparent among the provinces. The findings revealed that women had lost ground across the board, both in absolute terms and relative to men. They also showed that certain ranking patterns prevailed among the nine provinces and, notably, that women in provinces with pre-existing poor development statuses may also be predisposed to a poor development status relative to men. This double burden may translate into a perpetuation of poverty for women who find themselves in these provinces.

Keywords:

• gender and development
• development progress
• Human Development Index
• South Africa

1. Introduction

An earlier review found that, contrary to expectations, the development status of women in South Africa at an aggregate, national level had deteriorated between the years 1996 and 2001, in both absolute and relative terms (Booysen-Wolthers et al., 2006). This was contrary to expectations. Since factors contributing to this trend, and policies for dealing with it, are likely to have strong provincial or even local dimensions, this study aimed to measure the development progress of women in the individual provinces. In particular, we aimed to ascertain whether the same decline was evident in each of the nine provinces between 1996 and 2001 as it was nationally, and whether there were any systematic differences between the provinces in the absolute and relative development status of women.

For many decades the measurement of human development and the inclusion of gender distinctions therein have been a challenge. The Gender Development Index (GDI), the Gender Empowerment Measure (GEM) (United Nations Development Programme [UNDP], 1995), the Relative Status of Women (Dijkstra & Hanmer, 2000), the Gender Equality Index (Wieringa, 1999) and the Threshold Measure of Women's Status (Kishor & Neitzel, 1997) are all measures attempting to capture the gender dimension of human development.

It is important first to note the implications of the term ‘gender’. It differs from the term ‘sex’ in that ‘sex’ refers to a biological trait while ‘gender’ is widely understood as a socially constructed value dictating norms and behaviour. ‘Gender’ came to the fore with the Gender and Development approach to development, where the social relationship between men and women is considered key to development. ‘Gender’ has, however, morphed into a catch-all term that loosely implies distinction by sex, with the relationship implications that it entails. The paper follows this convention and uses the term ‘gender’ in its wider context.

An indicator to highlight differences in the pace of human development of men and women is hard to come by. To fill this gap, a gender-specific Human Development Index (HDI) was created that provided a reasonable measure with which to investigate gender differences in human development (Booysen-Wolthers et al., 2006). This measure is used in this paper to evaluate the development status of women in the provinces.

It has been difficult to make provincial longitudinal comparisons in South Africa owing to the lack of comparable data following the change in provincial boundaries in 1994. Moreover, information on poverty and human deprivation was not officially collected on a national scale prior to 1994 (Hirschowitz et al., 2000). Julian May's was one of the first studies (1998 and subsequently 2000) to reconcile the data from different eras to provide an official provincial baseline picture of poverty in South Africa. Further assessments of poverty in the provinces appear in the work of Whiteford et al. (1995) (poverty and income inequality analyses), Whiteford & Van Seventer (1999) (showing the extent of provincial poverty), Leibbrandt & Woolard (1999), Alderman et al. (2000) (providing a detailed poverty map of South Africa) and Bhorat et al. (2001) (geographically situating the poor by using three poverty measures). Statistics South Africa (Hirschowitz et al., 2000) contributes to the non-metric poverty measures with its Household Infrastructure Index and Household Circumstances Index, which can be used to compare the developmental status of the provinces.

The UNDP's South African Human Development Report (2000) provides a wide-ranging picture of the dimensions of poverty and human development in South Africa, cutting across races and provinces. Gender inequality is a key feature of this report, which includes GDI and GEM comparisons between the provinces for 1996. The subsequent UNDP report on South Africa (2003) offers longitudinal trends in provincial human development and poverty, and adds to these a measure of deprivation, the Service Deprivation Index. Although a gender dimension is included in this index, in the form of a comparison between male-headed and female-headed households and the GEM (reported only at a national level), the GDI is not included. In this report, gender inequality no longer seems to be regarded as a serious theme, thereby thwarting provincial longitudinal UNDP comparisons in this regard.

Despite the progress in provincial comparative analyses, none of the existing studies concentrates solely on the development status of women within the provinces, a gap this paper aims to fill.

2. Methodology of analyses

In this paper we follow the methodology developed in a previous paper (Booysen-Wolthers et al., 2006:609–10), but adjust it for provincial level analysis, which gives us two perspectives on the development status of women:

1. the size of the scrutinised group in relation to the total gender-specific population, which we call the predominance, for example the portion of the total female population that is considered economically active, and

2. the size of the scrutinised female group relative to its male counterpart, which we term the gender parity ratio. This refers to the ratio of girls to boys in, for example, primary school enrolment. The absolute difference between the two incidence groups is called the gender gap and this, although neither technically identical nor expressed in the same manner, is often used interchangeably with the term ‘gender parity ratio’ to describe the between-gender (i.e. female to male) relative position of women.

Applying these perspectives to unemployment, for example, provides depictions such as are shown in Figure 1.

Figure 1. The dimensions of unemployment status

Figure 1(a) shows the predominance of unemployment; that is, the portion of the female population (F) that is unemployed (F_u) (i.e. predominance = F_u / F), and similarly the portion of the male population (M) that is unemployed (M_u). Figure 1(b) represents the gender parity ratio of unemployment. This expresses the unemployment status quo in terms of the ‘gender gap’; that is, how far do women lead or lag behind men in terms of unemployment numbers (gender gap = F_u/ M_u). Our example in Figure 1(b) represents a hypothetical society in which for every one unemployed man (M_u) there are 1.2 unemployed women (F_u).

To determine whether there are systematic differences in human development between the provinces, we must focus our attention on the dynamics within each gender group. In addition to between-gender (i.e. female to male) comparisons, we therefore also consider within-gender (i.e. female group to female group) comparisons.

For each of the measures (i.e. predominance and gender parity ratio), both the absolute level and changes in the level between 1996 and 2001 can be compared across provinces. This can be done for the gender-specific HDI as a whole and for each of the seven components on which the HDI is based.

We start out on an aggregate level by reviewing the provincial patterns and trends in the gender-specific HDI. This is then augmented by reviewing the disaggregated individual components of the HDI (or their closest proxies). This not only produces a richer analysis; it also ensures that the limitations of aggregate measures are avoided, particularly in view of the potentially distorting effect of HIV/AIDS on such indicators. It must be pointed out that attempting to offer reasons for differences in provincial patterns is beyond the scope of this paper.

3. Aggregate trends: The human development indices of the provinces

The HDI endeavours to capture some of the complexities of human development by focusing on three areas related to the capabilities of people to improve their lives: health and longevity, attainment of knowledge, and the ability to have a decent standard of living (UNDP, 1995:131). The indicators used in this composite index are life expectancy at birth, literacy, the combined primary, secondary and tertiary education gross enrolment ratio (GER), and Gross Domestic Product per capita (PPP US$).

The gender-specific HDI used in this paper provides separate measures of human development for women and men, unlike what is referred to in this paper as the standard HDI measure, which is not adjusted for gender differences. To enable valid comparison with the gender-specific measure, the HDI used in this paper was calculated using own-source data. Small differences between our figures and the UNDP's published HDI figures are thus to be expected.

3.1 The standard HDI

South Africa's national standard HDI peaked at 0.733 (UNDP calculation) in the mid-1990s but subsequently showed a sharp decline, with the HDI reaching 0.687 in 2001 (UNDP, 2003:282). Note that by focusing on the period 1996–2001 we are exploring gender differences in human development on a downward phase of the national HDI; the results of analysing an upward phase or one done over multiple cycles might be different.

The standard HDIs (authors' own calculations) for the provinces in 1996 and 2001 are shown in Figure 2. Limpopo, lying well below the national average, had the lowest standard HDI of all the provinces in both years, Gauteng had the highest in 1996, and in 2001 the Western Cape had the highest of all the provinces. What is of interest, though, is that the HDI values of all nine provinces deteriorated between 1996 and 2001. KwaZulu-Natal's was the biggest percentage drop (–9.6 per cent), followed by Mpumalanga's (–8.5 per cent).

Figure 2. Provincial standard HDI for 1996 and 2001

3.2 Gender-specific HDI

The national gender-specific HDI for both women (i.e. female-specific – FHDI) and men (i.e. male-specific – MHDI) declined between 1996 and 2001, with the figure for women showing a sharper decline (from 0.75 to 0.69) than that for men (from 0.77 to 0.73) (Booysen-Wolthers et al., 2006). The performance of the provinces in this regard is thus of particular interest for ascertaining whether this trend was shared by all provinces.

Figures 3 and 4 show the gender-specific HDI measures for each province. Limpopo had the lowest development status level for both women and men for 1996 and 2001, falling well below the national averages for these years. Gauteng had the highest in 1996 for both women and men, but had slipped behind the performance of the Western Cape by 2001. The nine provinces all showed a drop in both the FHDI and MHDI levels between 1996 and 2001. KwaZulu-Natal had the most pronounced gender-specific decreases of all the provinces, with the FHDI dropping by 11.4 per cent and the MHDI by 8 per cent.

Figure 3. Provincial FHDI for 1996 and 2001

Figure 4. Provincial MHDI for 1996 and 2001

3.2.1 Within-gender trends

On an absolute, within-gender level the provinces differed substantially: the FHDI of the best-performing was well above that of the worst-performing province. This trend is confirmed by comparing the three best-performing with the three worst-performing provinces, which were separated by a margin of 13 per cent in 1996 and 18 per cent in 2001. Provinces with relatively poor FHDI values were thus falling further behind those with healthier ones.

3.2.2 Between-gender trends

The development status of women, as measured by the gender-specific HDI, was inferior to that of men in all nine of the provinces, in both 1996 and 2001. Furthermore, the gender gap, as shown in Figure 5, widened in all provinces during this period. Note that the display of figures up to four decimal points is to enable distinction between levels that often lie within close range of each other, and should not be interpreted as a claim to great accuracy.

Figure 5. Provincial gender gaps per province for 1996 and 2001

The Eastern Cape is somewhat of an exception as it was the only province where the FHDI exceeded the MHDI in 1996. This situation nevertheless reversed itself from one unfavourable to men in 1996 to one unfavourable to women in 2001.The gender parity ratio of the Eastern Cape deteriorated by almost 4 per cent (Table 1), constituting the biggest change among the provinces between the two years. The average gender gap increased during this time, with the national gender parity ratio deteriorating by 2.8 per cent between the two years.

Table 1: Changes in provincial gender-specific HDI gender parity ratios between 1996 and 2001

	EC	FS	GT	KZN	LP	MP	NC	NW	WC	RSA
Gender parity ratio changes	−3.97%	−3.04%	−2.56%	−3.65%	−2.09%	−3.50%	−1.47%	−2.22%	−1.02%	−2.79%

Examining the gender gap from a different point of view, we see that the three provinces with the worst gender parity ratios showed a gender gap deterioration exceeding that of the three with the best ratios. This implies that women in provinces with already poor gender performances were losing more ground, relative to men, than women in provinces with comparatively more favourable HDI gender parity ratios.

3.2.3 Aggregate findings

In terms of the pervasiveness of the national decline in female development status – on an aggregate level – it was found that:

• the FHDI declined in all provinces between 1996 and 2001, and

• despite a decline also in the MHDI during this period, all the provinces displayed a deterioration in the development position of women relative to men, as measured by the gender-specific HDI.

The national decline in female development status between 1996 and 2001 was therefore mirrored in the provinces.

In terms of the provincial patterns, on an aggregate level we find that:

• Limpopo fared the worst in terms of female development, both on an absolute level for both years and on a relative level in 1996.

• Gauteng was the leader in terms of female development status in 1996, but the Western Cape moved into the lead in 2001.

• The Eastern Cape had the best relative female development status in 1996, but ceded the top position to the Western Cape in 2001.

• There was a marked difference in the absolute FHDI levels of the provinces, with the interprovincial performance gap increasing between 1996 and 2001.

• In some provinces women were losing ground not only relative to men, but also faster relative to women in other provinces.

• KwaZulu-Natal emerged as the province losing most ground in female human development in absolute terms.

• Women in the Eastern Cape lost most ground relative to men during this period.

4. Disaggregate trends: Provincial patterns in the components of the HDI

In this section we disaggregate the aggregate, composite measure to reveal the national female development decline and the provincial patterns at a disaggregated level. To do this we review the data for the individual development areas on which the HDI is broadly based: life expectancy, attainment of knowledge and standard of living. For the FHDI, the components comprise the index of female life expectancy at birth to approximate the longevity of women, and the female literacy rate, the female primary school gross enrolment ratio and the female secondary school gross enrolment ratio (as per the HDI standard method) to indicate female attainment of knowledge. The female economically active population (EAP), female-headed households with zero income and female unemployment reflect female standard of living.1 The latter two indicators are deviations from the HDI standard method.

Some of the above indicators require explanation:

• The HDI calls for the use of functional literacy as a proxy for literacy. We nevertheless used the census category ‘percentage of population aged 15+ with no education’, subtracted from 100, as an indicator. Although gender stratified functional literacy data are available from 1999, the use of a 1999 figure to calculate a 1996 index was deemed too inaccurate, given the pace of changes in South Africa during the latter part of the 1990s.

• The data problems with tertiary education found on a national level (Booysen-Wolthers et al., 2006) apply also on a provincial level. The data on primary and secondary school enrolments must furthermore be interpreted with care. Although GER ratios exceeding 100 – as are prevalent among the provinces – are conceivable, the incentive for schools to inflate enrolment figures does pose a risk of inaccuracy in data sourced from educational institutions, as the data in this paper were. The conclusions drawn from this data are therefore somewhat tenuous. (Alternative enrolment data sources were considered but these proved unsatisfactory for use in this study.)

• Given the definition of the EAP, inclusion of both the EAP and unemployment as part of the analysis may look like duplication. Unemployment nevertheless is included as an added indicator of the (in)ability to secure a degree of material well-being.

4.1 Absolute levels of predominance

The degree to which the low level of human development permeated the female population of the individual provinces (i.e. the predominance – which may be viewed as a female human development rate) prompts a number of observations. This section focuses on the within-gender, absolute development differences between the provinces.

Tables 2 and 3 show the absolute female development rates for the provinces, for each HDI component area, highlighting the provinces that fared the best and worst.

Table 2: Provincial female development status in 1996, per HDI and component areas

Sources: Authors' calculations; Strauss et al., 1996; ASSA, 2002; DoE, 2003; StatsSA, 1998, 2003; UNDP, 2003; Global Insight, 2004.

Table 3: Provincial female development status in 2001, per HDI and component areas

Sources: Authors' calculations; Strauss et al., 1996; ASSA, 2002; DoE, 2003; StatsSA, 1998, 2003; UNDP, 2003; Global Insight, 2004.

Note that the EAP seems to capture only formal-sector female economic activity, or at the very least grossly undercounts the informal sector. The omission of the vast area of informal female economic activity from this analysis or the undercounting of it (due to the notorious difficulty in obtaining reliable data in this sector, let alone gender-stratified data) skews the results, as the EAP data only partly reflect the changes experienced by women during this period.

It also has to be pointed out that the accuracy of data on female-headed households with zero income for the 1996 and 2001 censuses has been questioned, the consensus being that both of these overestimated the number of households with no income. This probable overestimation, as well as the innate flaws of an analysis based on the gender of household heads (see Budlender, 2003), should be borne in mind when interpreting the results of this section.

Systematic differences between the provinces are obvious in some component areas, and less so in others. The scales do nevertheless tilt towards the women of Limpopo as having, across the board, the poorest absolute female development status of all the provinces in 1996 and 2001. This is measured by the performance of this province in the seven individual HDI component areas. Women in the Western Cape, on the other hand, had the best female development status, if measured in the same manner. Limpopo, the Eastern Cape and KwaZulu-Natal most frequently appeared in the bottom one-third group of worst-performing provinces, while the Western Cape, Gauteng and the Free State most frequently appeared in the best-performing one-third group.

The following are some further points of interest:

• KwaZulu-Natal had the lowest female life expectancy of the whole country, in both 1996 and 2001. In 1996 women in the Western Cape, with the highest provincial female life expectancy, could have expected to live 7.1 years longer than the women in KwaZulu-Natal. By 2001 they could have expected to live an astounding 15.3 years longer owing to the fall in KwaZulu-Natal's life expectancy, reflecting the devastating effect HIV/AIDS appears to have had on this province.

• An unexpected result is that Limpopo – another province with usually poor development results – had the best female secondary school GER values of all the provinces, for both years. This is particularly mystifying given this province's poor primary school GER values compared with those of other provinces. The fact that Limpopo furthermore had the highest female illiteracy rates of all the provinces for both years adds another element of doubt to a positive interpretation of the high GER value. We would argue that Limpopo's leading position in female secondary school GER is indicative of the poor overall level of female secondary GER values of the rest of the country, rather than a particular female development achievement on the part of Limpopo. Given the poor performance of this province in other areas of female development, any development indicator that uses Limpopo's performance as a benchmark must cause concern.

• Despite being the province with the worst female secondary school GER in both years, the Northern Cape was the only province showing an improvement in this component area between 1996 and 2001.

• Gauteng, not surprisingly, had the highest female EAP (as a percentage of the total female population) in both years, while Limpopo had the lowest. The definition of the EAP seems to refer only to those employed in the formal sector (Booysen-Wolthers et al., 2006:618). The informal sector, which is a mainstay of many women, especially in the rural economies, is therefore not (fully) captured in this measure. Almost 90 per cent of Limpopo's population live in the rural areas (Statistics South Africa [StatsSA], 1998, 2003), which makes an accurate assessment of their true EAP difficult until such time as non-formal sector activities can be more accurately reflected in the statistics.

• The Eastern Cape and Limpopo swapped places as the province with the highest proportion of female-headed households living in poverty. What is perhaps more striking here is the significant increase in the numbers across the board: there was a 57 per cent mean increase in the portion of female-headed households living in poverty between 1996 and 2001.

• In the case of female unemployment, the Eastern Cape also swapped with Limpopo for the weakest place position, which explains the female-headed household poverty swap described above. The worsening development position of women across all provinces is also particularly apparent in this component area.

4.2 Changes in predominance between 1996 and 2001

To ascertain whether the decline in the national absolute female development status was mirrored throughout the provinces, we turn to Table 4. In five of the seven component areas, the provinces – with one exception, Northern Cape with secondary school GER – showed a decline in the absolute development status of women between 1996 and 2001, thus following the national trend.

Table 4: Percentage changes between 1996 and 2001 in the absolute level of the provincial female development status, per HDI component area

^aThe differing bases of the percentage changes must be borne in mind when comparing the results.

To find systematic provincial differences in the pace at which the development status of women changed between 1996 and 2001, we focus first on the overall trend. In general, the provinces essentially moved concurrently when it came to changes in component areas between 1996 and 2001. The only exceptions were the Northern Cape (female secondary school GER) and the Eastern Cape, the Free State and KwaZulu-Natal (illiteracy).

Turning to patterns within this trend, we see that status changes seem more random and no one province stands out as showing a consistent decline or improvement across the board between these two years.

As Table 4 shows, the Eastern Cape had the sharpest drop in female primary school GER values between the two years, yet still remained the top performing province in this component area. The same can be said of Limpopo and female secondary school GER.

To find whether a higher propensity to change could be detected among provinces with initially bad (or good) development performance levels, we examined the changes experienced by these provinces between 1996 and 2001, to test whether the adage ‘once bad, always bad’ applies.

Table 5 shows the three provinces within each HDI component area with the best and worst female development levels in 1996, together with their absolute changes between the two years. For the period 1996 to 2001, an initially high development status led to relatively bigger changes in female development status than was the case with lower development status, in four of the seven HDI component areas.

Table 5: Percentage changes in the three provinces with the 1996 best and worst female development levels, per HDI component area

	Female life expectancy	Female illiteracy	Female primary school GER	Female secondary school GER	Female EAP	Female-headed households with zero income	Female unemployment
Best absolute status 1996	WC, LP, GT	WC, GT, FS	EC, KZN, MP	LP, MP, FS	GT, WC, FS	WC, NC, GT	WC, GT, NC
Change in index points 1996 − 2001	−8.15%	+0.04%	− 20.33%	− 15.01%	+14.16%	+ 70.86%	+ 25.47%
Worst absolute status 1996	KZN, EC, MP	KZN, MP, LP	NC, WC, GT	NC, WC, KZN	LP, EC, KZN	LP, EC, MP	LP, EC, NW
Change in index points 1996 − 2001	− 14.40%	− 3.44%	−6.14%	−3.3%	+ 18.73%	+47.15%	+8.59%
Note: Figures in bold denote the bigger change of the two groups, in each development area.

Viewing provincial performance gaps from another perspective, we compare the interprovincial performance gap of 1996 with that of 2001 to see whether the initial provincial differences in female development levels were maintained. Table 6 shows the results.

Table 6: Ratios of the three best-performing to the three worst-performing provinces' development levels

	Female life expectancy	Female illiteracy	Female primary school GER	Female secondary school GER	Female EAP	Female-headed households with zero income	Female unemployment	Mean result
1996	1.0928	2.9620	1.2481	1.3369	1.6401	1.9740	1.6478	1.7002
2001	1.2077	2.8592	1.0995	1.2112	1.5769	1.7055	1.4381	1.5854
% change	+10.51%	−3.47%	−11.91%	−9.4%	−3.85%	−13.60%	−12.73%	−6.75%
Performance gap	Widened	Narrowed	Narrowed	Narrowed	Narrowed	Narrowed	Narrowed	Narrowed

Note that in the cases of female illiteracy, female-headed households with zero income, and female unemployment, a smaller absolute number implies a better performance. For comparison's sake, the ratios in these instances were therefore calculated as worst to best performing levels.

In six of the seven component areas, the performance gap between the three top and the three bottom provinces narrowed. It was only in the area of life expectancy that the performance gap widened. The net effect was a mean narrowing of the interprovincial performance gap by 6.75 per cent between 1996 and 2001. This, however, is not necessarily good news, since in four of the areas the narrowing in the gap resulted from a loss in female development status in the best-performing provinces, which exceeded that of the worst-performing provinces. Only in the cases of female illiteracy and female EAP did the narrowing represent real improvement: the poorest performing group was catching up with the best-performing group.

4.3 Absolute levels of gender parity

A comparison of gender parity ratios, as shown in Tables 7 and 8, reveals that Limpopo once again has to be singled out. Measured by its performance in the HDI component areas, it had the worst development rate of women relative to men in both years (although less so in 2001). At the other end of the scale, the Eastern Cape had the best gender parity ratios in most of the seven component areas in 1996 and 2001.

Table 7: Gender parity ratios in 1996, per gender-specific HDI and component area

Source: Authors' calculations based on Strauss et al., 1996; ASSA, 2002; DoE, 2003; StatsSA, 1998, 2003; UNDP, 2003; Global Insight, 2004.

Table 8: Gender parity ratios in 2001, per gender-specific HDI and component area

Mpumalanga and Limpopo furthermore seemed disproportionately often to find themselves in the bottom one-third as far as component area gender ratio accomplishments are concerned. On the other hand, the Western and Eastern Cape most often featured in the top one-third, where the development status of women most closely approximated that of men. Except for the areas of primary school enrolments and unemployment, the interprovincial gender parity ratio patterns were, by and large, maintained between 1996 and 2001.

There are fewer surprises here than with the absolute level, yet a few findings do deserve mention:

• Women in the Eastern Cape – one of the less-developed provinces – had the highest margin of life years over their male counterparts, higher than any other province, while the Western Cape (1996) and Gauteng (2001) – two of the more developed provinces – had the smallest margins.

• Despite the Eastern Cape having the best relative unemployment rates of all the provinces, 11 per cent more women than men were nevertheless unemployed in 1996, and almost 14 per cent more in 2001. This can therefore not be celebrated as an absolute achievement on the part of women.

• The high EAP gender parity of Limpopo is, at first glance, a positive relative development for women. However, if this is viewed in the context of this province having had the worst Gross Domestic Product per capita of all the provinces in both 1996 and 2001 (UNDP, 2003:281), as well as the worst absolute female EAP of all the provinces, this merely indicates a very low overall economic activity rate in this province. The combination of the worst absolute EAP, worst relative poverty rate in households and worst relative unemployment rates (second weakest in 2001), certainly solicits concern for the women of Limpopo.

4.4 Gender parity changes between 1996 and 2001

The pervasiveness of the national decline in the relative development status of women was evident from the fact that this status declined in six of the seven component areas for most of the provinces during this period (see Table 9). Only in the area of EAP did the development status of women improve, relative to that of men – albeit only marginally.

Table 9: Absolute changes in the gender parity ratios between 1996 and 2001, per gender-specific HDI and component area

^aThe distinction of most and least change is based on size of change regardless of direction of change.

Turning to provincial patterns, we found that, in general, the provinces all had the same direction of change across all component areas between the two years. The only exceptions to this were the Western Cape (primary and secondary school enrolments), and Mpumalanga and the North West (household poverty).

It is once again difficult to find distinctive systematic differences in the provincial changes, since the existing patterns are not sufficiently conclusive to permit precise categorisation of the provinces.

To find whether an initial poor gender parity ratio predisposes a province to more change over time, we turn to Table 10. It shows that the converse seems to be true in that provinces with the best gender parity ratios in 1996 experienced bigger changes between 1996 and 2001 than those with the worst gender parity ratios. With the exception of EAP, all these changes were to the detriment of women.

Table 10: Changes in the provinces with the best and worst gender parity ratios in 1996, per HDI component area

	Gender-specific HDI	Life expectancy	Illiteracy	Primary school enrolments	Secondary school enrolments	EAP	Households with zero income	Unemployment
Best gender parity ratio 1996	EC, GT, FS	EC, FS, KZN	GT, WC, NW	EC, GT, NC	EC, MP, NW	LP, EC, KZN	GT, NC, WC	EC, KZN, WC
Absolute change 1996 − 2001	− 0.0316	− 0.0406	0.0694	− 0.0290	− 0.0987	0.0030	0.0751	0.0169
Worst gender parity ratio 1996	KZN, MP, LP	WC, GT, NC	LP, KZN, EC	LP, FS, MP	NC, GT, WC	NC, GT, NW	LP, MP, EC	LP, FS, MP
Absolute change 1996 − 2001	−0.0297	−0.0194	0.1073	−0.0122	−0.0092	0.0016	0.0333	0.0108
Note: Values in bold denote the bigger change of the two vertically read values.

A scrutiny of the interprovincial gender parity performance gaps (Table 11) reveals that the interprovincial performance gap (three best to three worst) narrowed by 2.47 per cent between the two years, inching its way towards more equal gender parity ratios across the provinces. (As before, note that in the cases of female illiteracy, female-headed households with zero income and female unemployment, a smaller absolute number implies a better performance. These ratios were therefore calculated as the worst to best performing levels.)

Table 11: Ratios of the three best-performing to three worst-performing provinces' gender parity ratios

	Life expectancy	Female illiteracy	Female primary school enrolments	Female secondary school enrolments	Female EAP	Female-headed households with zero income	Female unemployment	Mean result
1996	1.0340	1.6997	1.0446	1.1326	1.2271	1.8767	1.2767	1.3273
2001	1.0199	1.6895	1.0333	1.0912	1.2285	1.7321	1.2676	1.2946
Change	−1.36%	−0.60%	−1.09%	−3.65%	0.11%	−7.70%	−0.72%	−2.47%
Performance gap	Narrowed	Narrowed	Narrowed	Narrowed	Widened	Narrowed	Narrowed	Narrowed

4.5 Comparing the absolute with the relative status of women

It has to be noted that a good gender parity ratio does not necessarily translate into a good development status for women: a good gender parity ratio might simply mean that both men and women have a poor development status. Tables 12 and 13 combine information on the best and the worst provincial performances, in terms of both the absolute and the relative female development positions for 1996 and 2001, so as to show whether the individual provinces tended to perform the same in both fields.

Table 12: Provinces with best and worst female development status in 1996, per HDI and component area

Table 13: Provinces with the best and the worst female development status in 2001, per HDI and component area

Tables 12 and 13 indicate that certain provinces had the distinction of performing worst (or best) of all the provinces in both fields. This was particularly prevalent among the worst performers, indicating that a poor female development rate has a high probability of being translated into a poor gender parity ratio. The opposite is also true, but less convincingly so. This trend was also noticeably more apparent in 1996 than in 2001.

5. Concluding findings

This paper set out, first, to find whether the national decline in the development status of women was pervasive throughout all provinces during this period. The analysis confirms that the development status of women had indeed declined both in within-gender and between-gender terms across all the nine provinces between the years 1996 and 2001.

At the aggregate gender-specific HDI level:

• the FHDI declined across the board, with a mean (weighted by female population share) provincial decline of 8.52 per cent, and

• although the development status of men also declined during this period, the mean provincial gender gap widened by 2.85 per cent between 1996 and 2001.

As measured on a disaggregated level (by the component areas of the HDI):

• the absolute development status of women declined in all of the provinces, with only one exception, in five of the seven component areas reviewed, and

• the relative development status of women declined in two thirds of the provinces, in six of the seven component areas examined.

The second aim of this paper was to see whether there were any systematic differences in the level of, and changes in, the development status of women between the provinces for the years 1996 and 2001. In terms of level of female development status, it was found that, on an aggregate level:

• there was a distinct difference between the provinces in terms of their absolute and relative female development levels,

• Limpopo had the worst absolute development status in both years and the worst relative female development status of all the provinces in 1996,

• Gauteng was the leader in absolute female development in 1996, and the Western Cape in 2001, and

• the Eastern Cape had the best relative female development status in 1996, with the Western Cape taking top position in 2001.

On a disaggregated level, it was found that:

• here too, distinct provincial differences prevailed in terms both of absolute and relative female development levels,

• of all the provinces in both years, the women of Limpopo had the worst development status not only relative to other women but also relative to men,

• the women of the Western Cape on the other hand, had the best absolute development status in 1996 and 2001, while the Eastern Cape held the best relative position in both years, and

• it would appear that a poor female development status predisposes women to poor development gender parity ratios, while a high absolute development rate seems to be linked to a comparatively less bleak development rate relative to that of men.

In terms of provincial patterns of changes in the female development status of women between 1996 and 2001 we found that, on an aggregate level:

• the interprovincial performance gaps widened slightly in terms of both absolute and relative female development status, and

• women in KwaZulu-Natal lost most development ground in absolute terms, whereas women in the Eastern Cape lost most in relative terms during this period.

On a disaggregate level, it is apparent that:

• the 1996 interprovincial female development status ranking was for the most part upheld – on both an absolute and a relative level,

• with very few exceptions, the provinces changed concurrently across the component areas,

• no clear provincial change pattern manifested itself, either within-gender or between-gender,

• the mean interprovincial performance gaps narrowed on both absolute and relative levels, and

• high absolute female development rates and favourable gender parity ratios (i.e. high relative development status) seem to bring about more marked changes in their respective fields than is the case with less favourable rates and ratios.

We can therefore conclude that systematic differences between the provinces in the development status of women are detectable. The patterns appear to be aggregate, or disaggregate, level specific in the case of changes between 1996 and 2001.

All of this clearly does not bode well for women who find themselves in provinces with low development status, since it implies that they will in all probability continue to bear a double burden, namely a poor development rate compared with other women and a particularly poor relative development rate compared with men. It is conceivable that the poor are destined to remain poor, especially should they be women, and should they find themselves in a province with a prior poor female development status.

This raises questions as to why women's development status continues to lag behind that of men. It also begs contemplation on the suitability of indicators to reflect and gauge the intricacies of a gendered development process. These are all matters that need to be addressed if we are to find sustainable solutions to the development backlog of women in all of South Africa's nine provinces.

Notes

¹See Booysen-Wolthers et al. (2006:609) regarding the omission of gendered wage differences from the analysis.