Economic Research Southern Africa (ERSA) is a research programme funded by the National

Treasury of South Africa. The views expressed are those of the author(s) and do not necessarily represent those of the funder, ERSA or the author’s affiliated

institution(s). ERSA shall not be liable to any person for inaccurate information or opinions contained herein.

The South African labour market,

1995-2013

Lyle Festus, Atoko Kasongo, Mariana Moses and Derek Yu

ERSA working paper 493

February 2015

The South African labour market, 1995-2013

Lyle Festus,∗ Atoko Kasongo,† Mariana Moses‡ and Derek Yu§

February 2, 2015

Abstract

This paper investigates the changes in the South African labour marketin the post-apartheid period in 1995-2013 by updating the work by Oost-huizen (2006) and Yu (2008). The three main data sources used are theOctober Household Survey of 1995, the Labour Force Survey of September2004 and the Quarterly Labour Force Survey of 2013 Quarter 4. It wasfound that while unemployment has risen over the period, employment hasalso increased. Nonetheless, the extent of employment increase was notrapid enough to absorb all net entrants in to the labour force, resulting inincreasing unemployment, or an employment absorption rate of below 100per cent. Unemployment continues to be concentrated in specific demo-graphically and geographically defined groups, most notably blacks, thepoorly educated and the youngsters residing in Gauteng. Unemploymentis a chronic problem for the youth in particular, as nearly three quar-ters of them never worked before. Finally, the employment absorptionrate was the highest in some less developed provinces like Northern Cape,Mpumalanga and Limpopo, thereby suggesting the possible success of thegovernment’s efforts to promote the development in the poorer provinces.

JEL Classification: J40Keywords: South Africa, labour market trends, labour force, employ-

ment, unemployment

1 Introduction

The South African labour market has in the past played a very significant rolein the country’s economic development. In the pre-democratic regime, priorto 1994, it was used as a mechanism to segregate society. This was achievedthrough various legislations which segmented the labour market along raciallines, to the disadvantage of non-whites. For example, the Bantu EducationAct of 1953 ensured that non-whites received a sub-par quality of education

∗Postgraduate student, Department of Economics, University of the Western Cape†Lecturer, Department of Economics, University of the Western Cape‡Lecturer, Department of Economics, University of the Western Cape§Senior Lecturer, Department of Economics, University of the Western Cape & Part-time

Researcher, Development Policy Research Unit, University of Cape Town

1

relative to their white counterparts. This in turn limited their potential forachievement within the labour market and consequently limited their ability toimprove their standard of living. Furthermore, legislation such as the Group Ar-eas Act of 1950 and the Black Labour Act of 1964 established segregated areasof residence in urban areas, in the case of the former, as well as strict limitationson the type and conditions of employment available to Blacks for the latter. Fur-thermore, the Industrial Conciliation Act of 1924 allowed the establishment ofIndustrial Councils (ICs), which were permanent collective bargaining institu-tions. These ICs did not recognize African trade unions and prevented Africanemployees from taking part in the collective bargaining process, resulting in thediscrimination of non-white workers for the benefit of their white counterparts.In addition, the Mines and Works Act Number 12 of 1911 was the first in aseries of laws which effectively limited the occupations available to non-whiteworkers. This was achieved by reserving certain highly-paid and highly-skilledjobs for whites only. All of these laws contributed to the exploitation of non-white workers in the apartheid-era labour market. However, it can also be saidthat the South African labour market contributed toward the abolishment ofapartheid, as certain labour market entities, in particular trade unions withtheir vast membership, were vital to the inception of the democratic process inSouth Africa.

During the post-apartheid period, there are radical changes in the SouthAfrican labour market in terms of its legislation. Among these include the Ba-sic Conditions of Employment Act of 1997 which stipulates the minimum wagesapplicable to certain sectors, as well as specifying minimum working conditionsfor labourers and outlining some of their rights. In addition, the Labour Rela-tions Act of 1995 outlines processes regarding collective bargaining in the labourmarket and the resolution of labour disputes. Furthermore, the EmploymentEquity Act of 1998 encourages Affirmative Action, which boils down to the needfor employing more non-white workers in order to reduce societal inequalities.The effect of these new laws have had on the labour market, given South Africa’shistory, is of particular importance. Moreover, because labour market incomeremains the main income source for poverty reduction, its reform is essential toaddressing inequality and raising the standard of living in South Africa.

Currently, the South African labour market continues to play a pivotal rolein society. However, the reasons for its prominence have changed somewhat.To this end, the persistent unemployment problems plaguing the labour markethave become a focal point for South African government and society. The afore-mentioned persistent unemployment stems from a range of issues including thelow quantity and quality of education of the previously disadvantaged groups(e.g. non-whites). In this regard, recent surveys have shown that South Africanstudents on average are among the worst-performing groups when compared totheir peers globally. In particular, the Trends in Mathematics and Science Study(TIMSS) 2011 results showed South African was the second worst-performingcountry in Mathematics and the worst-performing country in Science, out of the63 participating countries. On the other hand, the Southern and Eastern AfricaConsortium for Monitoring and Educational Quality (SACMEQ) 2007 reports

2

have shown that the average reading and mathematics performance of Grade6 pupils in South Africa (495) was close but marginally below the SACMEQoverall average (512) in reading and (510) in mathematics. In both readingand mathematics, the proportion of learners achieving at the higher SACMEQlevels of competence was significantly low. Hence, the South African educationsystem is producing a continued stream of insufficiently educated new work seek-ers. This creates a supply of workers who may remain unemployable. Moreover,there has been an emergence of large numbers of unemployed youth in SouthAfrica. This phenomenon is attributed to, amongst others, their lack of experi-ence, which reduces their employability and contributes toward unemployment.This factor is arguably related to the aforementioned low quality and quantityof education in South Africa.

In addition, the unemployment problem the South African labour marketexhibits also occurs from more direct labour market issues such as employmentand wage rigidity. For the former, the labour legislation introduced post-1994,while affording worker’s rights, also served to limit the ability of employersto adjust their consumption of labour. This inhibits the ability of the labourmarket to function efficiently and thereby increases unemployment. In the caseof the latter, wage rigidity due to legislated minimum wages contributes toincreased levels of unemployment. This occurs when minimum wages are setabove market clearing levels, which limits the ability of employers adjust theirconsumption of labour efficiently. On this point, both Kingdon and Knight(1999) as well as Bhorat, Kanbur, and Mayet (2012) state that the sector inwhich the minimum wage is established plays a large role in whether or not ishas a negative effect on employment or the number of hours worked.

It is also argued that some labour market participants have unrealisticallyhigh reservation wages, which prevents them from being employed. However,Kingdon and Knight (2004) point out that there are no reliable data on reser-vation wages and therefore it is not easy to estimate the proportion individualswho are unemployed because of this problem. Furthermore, skills mismatchis a serious factor contributing to the unemployment problem, and graduateunemployment is an example. This happens because the graduates producedby the education system are either not demanded by the labour market or arealready abundant in supply, based on their area of study. This in turn cre-ates unemployable graduates. On this, Bhorat (2004) states that “institutionsof higher education are ostensibly not matching their curriculum design effec-tively enough with the labour demand needs of employers”. In addition, Pauwet al. (2006) attribute the graduate unemployment problem to a skills deficitissue. The aforementioned graduate unemployment problem is also indicativeof a serious structural break in the South African labour market.

Another serious problem facing the labour market is South Africa’s informalsector, which also plays a role in contributing to unemployment. As the informalsector is not only small in relation to the formal sector, always hovering ataround 2 million employed people, but there are also various barriers of entry tothis sector, thereby making it difficult for those retrenched or unable to find workin the formal sector to obtain employment in informal sector, thereby worsening

3

the unemployment problem. To this end, Kingdon and Knight (2004) statethat “the informal sector is not generally a free-entry sector in South Africa,and that there may be barriers which prevent many of the unemployed fromentering much of this sector”. Burger and Woolard (2005) more specifically citea lack of infrastructure and inadequate access to credit markets as barriers toentering the informal sector in South Africa.

Looking at other factors accounting for the unemployment problem, Oost-huizen (2006) states that “economic growth has been unable to provide thenecessary employment opportunities required by population growth and risinglabour force participation rates, resulting in a rapidly rising rate of unemploy-ment”. Yu (2008) also found rapid increases in both the labour force and labourforce participation rates in the late 1990s, which could contribute to the increaselevels of unemployment, as the pace of job creation is not rapid enough to absorball net labour entrants.

In the past, many studies were conducted on the South African labour mar-ket performance post-apartheid. However, most of these papers focused onthe on the first decade after democracy. The general findings of these studieswere that employment growth in South Africa was not sufficient to absorb allnew entrants to the labour market (e.g., Oosthuizen 2006; Kingdon and Knight2007; Yu 2008; Hodge 2009). Furthermore, economic growth and subsequentemployment growth was not rapid enough to reduce unemployment (Burger andWoolard 2005; Oosthuizen 2006). However, as Bhorat (2009) and Hodge (2009)discussed, the rising unemployment figures post-apartheid are due in small partto substantial labour supply increases. In addition, as Yu (2008) found, themost disadvantaged segments of the population i.e. those who have low levelsof education, those who reside in relatively poorer provinces and blacks weremost likely to be unemployed.

Since 2005, the South African labour market and associated legislation hasagain changed, with the objective of addressing many of the issues that are nowvisible from labour market data. To this end, policies such as the revised BasicConditions of Employment Act were introduced, with the aim of which beingreducing the labour market rigidities while still providing certain basic rightsto the workers. In addition, 2006 saw the implementation of the Acceleratedand Shared Growth Initiative South Africa (ASGISA) which outlined SouthAfrica’s developmental framework and identified key areas and requirements foreconomic growth. One of ASGISA’s main goals was to reduce the unemploy-ment rate under the narrow definition to 15 per cent by 2014. This plan wassubsequently replaced by the New Growth Path (NGP) in 2010 which, amongother things, readjusted the timeline for growth targets outlined in ASGISA tobe achieved. Importantly, the NGP also set a new target to reduce the unem-ployment rate to 15 per cent, to be achieved by 2020. More recently, the NGPwas replaced by the National Development Plan (NDP) which was introducedin 2012. The labour market goal of the NDP is to reduce the unemploymentrate to 6 per cent by 2030. With regard to other labour market legislation intro-duced since 2005, the Employment Tax Incentives Bill or Youth Wage Subsidywas legislated in 2013 and was officially implemented on 1 January 2014. It pro-

4

vides a fiscal incentive for employers to hire more youth workers, with the hopeof creating employment and providing the youth with essential experience andskills. As a result of these changes to both legislation and strategy regardingthe South African labour market and in order to evaluate the success of labourmarket policy, the need for an updated labour market study is justified.

Hence, the main objective of this paper is to provide an 18-year labourmarket review for the period 1995 to 2013, just before the abovementionedEmployment Tax Incentives Bill was launched. The data used for this papercomes from the following sources. From the year 1995 to 1999, the OctoberHousehold Surveys (OHSs), taking place annually, were used. From the year2000 to 2007, the Labour Force Surveys (LFSs), which took place twice a year,were used. From the year 2008 to 2013, the Quarterly Labour Force Surveys(QLFSs), taking place four times a year, were used. Furthermore, OHS andLFS data were weighted with Census 2001 weights, while the QLFS data wereweighted with Census 2011 weights. Therefore, we expect some unsubstantiatedfluctuations of the data between the year 2007 and 2008 as a result adoption ofthe different weighting method.

The remainder of the paper is structured as follows. Section 2 provides anoverview of the South African labour from a broad perspective, while Section 3examines labour market trends specifically between OHS 1995, LFS 2004 andQLFS 2013. This is followed by a multivariate econometric analysis in Section4 based on the three periods mentioned above. Section 5 concludes the studyand provides some policy recommendations. Also, for the rest of the paper,OHSs will be referred to as OHS 1995, OHS 1996, etc., while the LFSs will bereferred to as LFS 2000a (for the first round of LFS in 2000), LFS 2000b (secondround in 2000), LFS 2001a, LFS 2001b, and so forth. Finally, the QLFSs willbe referred to as QLFS 2008Q1 (for the QLFS conducted in the first quarter of2008), QLFS 2008Q2 (second quarter of 2008), and so forth.

2 The South African labour market: Long-term

trends between 1995 and 2013

Figure 1 presents the size of the labour force and the labour participation ratefrom 1995 to 2013.1 The labour force under the narrow definition increasedrapidly during the OHS period and increased steadily during the LFS period.It peaked at nearly 19 million in the first quarter of 2009, before a downwardtrend was observed due to the impact of the global recession. An upward trendwas observed again since 2011 and the LF number reached an all-time high of20.0 million in QLFS 2013Q4. A similar trend could be observed for the labourforce number under the broad definition, and this number reached 22.2 millionin the last quarter of 2013.

The labour force participation rate (LFPR) under the narrow definition in-

1Table A.1 in the appendix shows the number of working-age population, LF, employed,unemployed, as well as LFPRs and unemployment rates in all surveys under study.

5

creased rapidly in the OHSs, before showing a downward trend in general from2000 to 2004. An upward trend was observed between 2005 and 2008, withthe LFPR peaking at 59.4 per cent in QLFS 2008Q2, before it declined due tothe impact of recession. The LFPR has been hovering between 55 and 57 percents since QLFS 2009Q3. The LFPR under the broad definition also increasedrapidly in the OHSs, before fluctuating unsteadily between 2000 and 2007. Thisrate has been stabilized at the 61-63 per cent range since 2008.

Note that the abrupt decline of the broad LFPR during the changeover ofLFS and QLFS is attributed to the change in the methodology to distinguishthe discouraged workseekers (Yu 2009 & 2013). In particular, Yu (2013: 707)highlights the fact that the discouraged workseekers are identified more strictlyin the QLFSs, as the respondents’ answers to the question “what was the mainreason why you did not try to find work or start a business in the last fourweeks?” is considered (this question was not involved at all in the OHS andLFS methodology). Only if the respondents’ answers to the question are ‘nojobs available in the area’, ‘unable to find work requiring his/her skills’ or ‘losthope of finding any kind of work’, then they could be classified as discouragedworkseekers. Hence, this causes the number of discouraged workseekers andsubsequently the broad unemployed (which is the sum of narrow unemployedand discouraged workseekers) to drastically decrease between LFS 2007b andQLFS 2008Q1.2

The number of employed is represented in Figure 2.3 In general, an upwardtrend was observed between OHS 1996 and QLFS 2008Q4, although the abruptincrease of about one million between LFS 2007b and QLFS 2008Q1 is attributedto the change of weighting method by Stats SA, as discussed in Section 1.4

Between QLFS 2008Q4 and QLFS 2009Q4 there was decrease in employment ofnearly a million due to recession. Employment showed an upward trend againsince QLFS 2010Q2, and reached an all-time high of 15.2 million in the lastquarter of 2013.

Figure 3 focuses on the unemployment aggregates. First, the number ofunemployed under the narrow definition showed an upward trend between OHS

2Yu (2013: 713-714) attempted to apply the QLFS labour market status derivation method-ology to identify the discouraged workseekers in both LFSs and QLFSs. At the end, therewas still an abrupt decline (albeit less serious) of his number during the changeover of LFSand QLFS, and Yu (2013: 714) claimed that the possible reasons for the still relatively highernumber of discouraged workseekers in the LFSs could either be real, or due to the differencein the questionnaire structure between LFSs and QLFSs. These discussions are beyond thescope of this study.

3The last two columns of Table A.1 in the appendix show employment by formal andinformal sectors. However, Essop and Yu (2008) note that it is not possible to estimateinformal sector employment between OHS 1995 and OHS 1996, because employees were notasked to declared the sector of employment.

4When QLFS 2013Q4 results were released using Census 2011 weights for the first time,Stats SA also released the QLFS 2008Q1-2013Q3 labour market aggregates (which were origi-nally weighted with Census 2001 weights) with Census 2011 weights. This explains the abruptfluctuations of the aggregates between LFS 2007b (weighted with Census 2001 weights) andQLFS 2008Q1 (re-weighted with Census 2011 weights). Table A.2 in the appendix showsresults.

6

1995 and LFS 2003a. In fact, the number of employed during the 18-year periodunder study was the highest in LFS 2003a at 5.1 million. This number stabilizedat the 3.9-4.4 million range between LFS 2003b and QLFS 2009Q4, before aslight upward trend was observed during the 2010-2013 period. Looking at thenumber of unemployed under the broad definition, a similar trend was observedbetween OHS 1995 and LFS 2003b. In fact, this number also peaked at thelatter survey (8.3 million). A downward trend was observed in general betweenLFS 2003b and LFS 2007b. An abrupt decline of the number of broadly definedunemployed from 7.3 million to 5.6 million was due to the change in methodologyto capture the discouraged workseekers, who form part of the broadly definedunemployed (Yu 2009). This number increased steadily between 2008 and 2013.

As far as the unemployment rate under the narrow definition is concerned,it increased from 17.6 per cent in OHS 1995 to the peak level of 31.1 per centin LFS 2003a. This rate showed a downward trend and dropped to as lowas 21.5 per cent in the last quarter of 2008. Unfortunately, due to the globalrecession, an upward trend was observed in 2009, and this rate has been hoveringbetween 24 per cent and 25 per cent in 2010-2013. Finally, the broadly definedunemployment rate increased from 30.8 per cent in OHS 1995 to the highestlevel of 42.5 per cent in LFS 2003a, before a downward trend was observed until2007. After the abrupt decline from 35.6 per cent to 27.8 per cent during thechangeover of LFS and QLFS, this rate increased steadily to 32.0 percent inQLFS 2010Q1, before hovering in the 31.5-33.5 percent range.

As a result of the incomparability of the broad methodology between OHSs/LFSsand QLFSs, for the remainder of the paper, the labour market aggregates underthe narrow definition will be the focus of the analyses, unless stated otherwise.

3 The South African labour market: Snapshots

between 1995, 2004 and 2013

3.1 Labour force participation

The demographic composition of the labour force under the narrow definitionin OHS 1995, LFS 2003b and QLFS 2013Q4 is captured in Table 1. In 2013,the labour force number was more than 20 million, which reflects a cumula-tive growth of 8.5 million individuals since 1995. The Black racial group hasconsistently dominated the labour force accounting for 67.9 per cent in 1995,72.7 per cent in 2004 and rising to 76.0 per cent in 2013. Furthermore, whenconsidering the increase in the size of the labour force, the blacks accountedfor a share in excess of 85 per cent in all three surveys being assessed. Themale population accounted for the larger proportion of the labour force in allthree surveys, while the gender share of the increase in labour force was almostequitably split between the two groups.

Table 1 shows that the 24-34 years and 35-44 years cohorts accounted forthe greatest increase of labour force during the period under study. Also, the18-29 year olds, who are age-eligible for the Employment Tax Incentives Bill,

7

experienced a growth of labour force number of 40.6 per cent between 1995 and2004, and less rapid growth of 13.3 per cent between 2004 and 2013.

Considering the spread of the labour force provincially, the Western Capeand Gauteng provinces collectively account for 46.4 per cent of the share in 2013rising from 39.4 per cent in 2004, with more than 72 per cent of the increaserecorded between 2004 and 2013. KwaZulu-Natal, by contrast, while accountingfor a labour force in excess of 3 million individuals during 2013, experienced adecline in share of the increase from 18.2 per cent to 5.3 per cent during therespective 1995-2004 and 2004-2013 periods. The labour force size has morethan doubled in urban areas, rising from 7.5 million individuals in 1995 to 15.5million in 2013, reflected by an increase of slightly under 4 million per period.Consequently, the share of the labour force in rural areas has decreased onaverage by roughly 7 percentage points.

As for educational attainment, there was a decline in the share of labourforce members who have no schooling and incomplete primary levels, expressedby a cumulative fall of 10.7 percentage points. This implies that more yearsare spent on acquiring an education which positively impacts upon the labourforce share where participants with incomplete secondary schooling and Matricaccount for 62.6 per cent, 68.4 per cent and 72.8 per cent, respectively for theperiods concerned. Furthermore, the share of labour force with post-Matricqualifications, while remaining unchanged at 12 per cent in 1995 and 2004,have risen to slightly under 17 per cent in 2013. This can be attributed tothe increasing realisation that a diploma or degree aids in securing better jobopportunities.

In summary, labour force growth since the transition could be largely at-tributed to the Black, urban individuals residing in Gauteng, aged 25-44 yearsat the time of the survey, and having attained at least some secondary schooling.

Table 2 shows the LFPRs in the three surveys under investigation. TheLFPR of Whites and Coloureds have consistently exceeded 60 per cent, record-ing levels of 67.4 and 64.1 per cent respectively in 2013. As far as gender isconcerned, in 2013, the LFPR gap between male and females have slowly beenbridged with males still accounting for the highest rate of 63.4 per cent versusthe 50.5 per cent of females, possibly attributable to the difference in retirementage. An excess of 70 per cent of LFPR was recorded in 2013 for the three agegroup categories ranging between 25 and 54 years. While less than 42 per centof the 55-65 year old members are active participants, the LFPR has steadilyincreased over the two decades possibly due to fewer people choosing to retireearly. The decline in the LFPR of the youngest cohort (15-24 years), expressedat 25.5 per cent in 2013 from 28.2 per cent in 2004, is linked to higher levels ofeducation being attained and concomitantly years spent studying.

As to be expected, Gauteng and the Western Cape have garnered the highestLFPR levels rising from the region of 60 per cent in 1995 to slightly under 70per cent in 2013, a phenomenon that aligns with the rural-urban dynamics ofprovinces. By contrast, LFPRs were the lowest in Limpopo (40.4 per cent in2013) and the Eastern Cape (45.2 per cent in 2013). The LFPR has alwaysbeen higher for the urban dwellers in all three surveys under study. Finally,

8

the significance of education in securing and retaining employment is notedas labour force members with no schooling recorded the lowest LFPR in allthree surveys, with only slightly improved levels for participants with incompleteprimary and secondary schooling at just below 45 and 47 per cent respectivelyin 2013. As expected, the LFPR was the highest for the Degree and post-MatricCertificate/Diploma holders, at 88.8 and 85.5 per cents respectively in 2013.

In summary, at the end of 2013, higher LFPRs were recorded for predom-inantly White males, aged 25-44 years, residing in urban areas in Gauteng orWestern Cape, and attaining post-Matric qualifications.

3.2 Employment

The same three surveys would be used to examine the employment aggregates.First of all, there has been a general progressive increase in employment for theperiod 1995 to 2013, as shown in Table 3. The greatest increase was 3.5 millionjobs experienced between 2004 and 2013. The bulk of the increase in employ-ment between 1995 and 2013 took place amongst the blacks, with an increase ofabout 4.8 million jobs. Furthermore, in 2013 the share of blacks in employmentwas 73% while the whites only accounted for 13.1% of the total employment.The increase in black employment between 1995 and 2013 can be attributed tothe increase of educational attainment of the black workseekers and AffirmativeAction policies, which not only encouraged black people to participate in thelabour market but also increased their likelihood of employment.

Although employment increased in both genders, the female share of em-ployment substantially increased from 39.1% in 1995 to 43.9% in 2013. Thisis consistent with the increase in the female labour force participation as indiscussed in Section 3.1 above. The major increase in employment in the agecategory was for the age group 25-34 years and 35-44 years. Those betweenthe ages of 15-24 years only experienced a net increase of 200 000 jobs between1995 and 2013 and a decline in the share of employment from 11.8 per cent to8.7%. The two age cohorts above the age of 45 years had a combined share ofemployment from 23.6 per cent in 1995 to 28.8 per cent in 2013. This reveals anincrease in the proportion of jobs held by elderly people pointing to an ageingworkforce. In addition, the age group 18-29 years (who are eligible for the youthwage subsidy) experienced a decline in the share of employment from 27.7 percent in 1995 to 26.9 per cent in 2004 and finally to 23.9 per cent in 2013. Inaddition, the number of youth employment increased from 2.6 million in 1995 to3.1 million in 2004 and finally to 3.6 million in 2012. The extent of increase ofemployment for this youth age cohort is slow compared with the middle-age co-horts, and hence validates the rationale of the government intervention throughthe youth wage subsidy.

Gauteng, KwaZulu-Natal, Western Cape and Eastern Cape held the highestshare of the total employment. In particular, Gauteng accounted for the largestshare of employment expansion of more than 2 million between 1995 and 2013,or 38.5 per cent. Interestingly, Mpumalanga and Limpopo, the two relativelyless developed provinces, accounted for 10.0 per cent and 10.3 per cent of the

9

share of increase of employment between 1995 and 2013.As expected, the share of employed with no education declined (from 8.1

per cent in 1995 to 2.4 per cent in 2013). This is a reflection of the increasingdemand for skilled labour by the employers. Those with Matric recorded thehighest increase in the share of employment (more than 40 per cent). Thestructural shift towards the hiring of more skilled labour is further consolidatedby the increase in the proportion of employed accounted for by those with post-Matric qualifications (from 14.1 per cent in 1995 to 19.8 per cent in 2013).

As argued by Oosthuizen (2006: 17), merely comparing employment growthbetween two periods might not necessarily provide a clear picture of employ-ment performance of the South African labour market. Hence, Table 4 presentsthe target growth rate (TGR), actual growth rate (AGR) and employment ab-sorption rate (EAR) during selected periods.

TGR is the rate at which employment must grow to provide employmentto all the net entrants to the labour market between two periods of time (fromperiod X to period Y) which need not be consecutive.TGR =

LFY−LFX

EXwhere LF and E stand for then number of the labour

force and employed respectively (Oosthuizen 2006: 17). On the other hand,AGR is the growth rate of the number of employed from period X to period Y,(Oosthuizen 2006: 16), and is calculated as EY−EX

EX.

Finally, EAR measures the proportion of the net increase in the labourforce from period X to period Y that find employment during the same period.EAR =

EY−EX

LFY−LFX=

AGR

TGR. An EAR of 100 per cent implies that the full net

increase in the labour force between two periods were employed (Oosthuizen2006: 18)

Focusing on the 1995-2013 period, for all the net entrants into the labourforce to find jobs, employment would need to grow by 89.4 per cent between1995 and 2013. However, the actual employment growth rate was “only” 60.0per cent, thereby resulting in the EAR of 67.1 per cent. This implies thatemployment growth was not rapid enough to absorb all the net entrants to thelabour market between 1995 and 2013, as out of the 100 net entrants to thelabour force, only 67 of them were able to find employment.

Nonetheless, this aggregated view may obscure the varied experiences ofgroups defined by various demographic and location characteristics (Oosthuizen,2006: 18). Therefore, Table 4 also shows the TGRs, AGRs and EARs by race,gender, age cohort, province, area type and educational attainment. Once againfocusing on the 1995-2013 period and looking at race, the TGR was the highestfor blacks (120.4 percent), followed by coloureds (64.7 percent), while this ratewas the lowest for the whites (10.9 percent). Similarly, AGR was the highestfor blacks (80.8 percent), followed by coloureds (38.0 percent) and this rate wasthe lowest for whites (6.7 percent).

If one only interprets the AGR, it is possible to reach an incorrect conclusionby claiming Affirmative Action is highly successful as the AGR was the highestfor the previously most disadvantaged group (i.e. blacks). It is because theAGR (80.8 percent) was actually lower than the TGR (120.4 percent) for blacks.Hence, it means employment growth was not rapid enough to fully absorb the

10

net entrants into the labour force during the 18-year period for blacks. In fact,the last column of Table 4 shows that the EAR for blacks (67.1 percent) was“only” about 5 percentage points above that of whites (62.2 percent). Given thefact that the majority of the LF consists of blacks (see Table 1), it is importantfor the government to implement the necessary measures to boost the AGRand EAR of this group further, so as to drastically alleviate the persistentunemployment problem.

As far as gender is concerned, the TGR, AGR and EAR were higher forthe females between 1995 and 2013. With regard to the five age cohorts, itis interesting that all three rates increased across the elderly cohorts, with theEAR rising from 24.6 per cent for the 15-24 year-old to 89.8 per cent for the55-65 year-old. Looking those aged 18-29 years (the age-eligible cohort for theEmployment Tax Incentives Bill), the TAR was very high at 86.7 per cent, butthe AGR was only 37.7 per cent. As a result, only 43.5 per cent of the netentrants to the labour market could successfully find employment. This findingprovides another justification for the implementation of the Bill.

With regard to the provincial results, it is interesting that the TGR was inexcess of 100 per cent in three provinces (Gauteng, Mpumalanga and Limpopo).Also, these three provinces also recorded the highest AGR (83.2 per cent, 97.1per cent and 101.5 per cent respectively). It is quite surprising that the EARwas the highest in Limpopo (86.4 per cent), followed by KwaZulu-Natal (82.0per cent) and Western Cape (70.1 per cent). The positive finding in Limpopocould be a reflection of the government’s efforts to promote the development inthe poorer provinces.

Finally, all three rates were the highest for the Degree holders. In particular,the EAR was 92.7 per cent, which means that more than 9 net entrants tothe labour market with Bachelor Degree could successfully find employment,but such likelihood was only as low as 58.0 per cent for those with incompletesecondary education.

Table 5 illustrates employment according to the three broad skill categories.There was a consistent increase in the absolute number of employment in allthree categories, but such increase was the most rapid for the highly-skilledemployment category. Hence, the share of employment accounted for by thesepeople increased from 8.7 per cent in 1995 to 14.0 per cent in 2013. In addition,the share of semi-skilled employment declined by 2 percentage points over the18-year period. The bulk of the decline can be seen in the skilled agriculturaland fishery workers with a loss of 253 000 jobs between 2004 and 2013. Thisdecline can be attributed to the improvement in technology such that labour isreplaced by capital. Furthermore, the share of unskilled labour in employmentalso contracted by 4 percentage points, despite the fact that employment in do-mestic work and elementary occupations experienced an increase of 0.31 millionand nearly 1 million respectively between 1995 and 2013.

It can be seen from Table 6 that employment in the primary sector employ-ment contracted from 1.7 million in 1995 to 1.1 million in 2013. As of 2013 theprimary sector accounted for only 7.5 per cent of employment, compared withthe 17.9 share in 1995. In particular, employment decrease in the agriculture,

11

hunting, forestry and fishing industry amounted to more than 0.5 million be-tween 1995 and 2013. In contrast, secondary sector employment grew by slightlyabove 1 million from 1995 to 2013, with the bulk of the increase being accountedfor by the construction industry.

The tertiary sector experienced most of the employment growth holding ashare of employment at 72.1 per cent in 2013, rising from 61.0 per cent in 1995.The employment in financial, insurance and business services industry morethan trebled between 1995 and 2013 (rising from 0.6 million to 2.0 million), whileemployment in the wholesale and retail industry nearly doubled (increasing from1.7 million to 3.2 million).

Figure 4 compares the annual average growth of real gross value added(GVA) and employment by each broad industry category between 1995 and2013. It can be seen that finance and construction industries outperformedother industry categories with an annual employment growth rate of more than5 per cent and the real GVA growth rate of approximately 5 per cent. In con-trast, the two categories in the primary sector (agriculture and mining) experi-enced a decline in employment growth. This is evidence of the structural shift ofthe labour market towards an increased demand for highly-skilled workers. Fur-thermore, mining industry was the only industry category experiencing negativegrowth rate in both employment and real GVA. This could be attributed to notonly structural change of the economy, but also other problems experienced inthe mining sector, ranging from minimum wages, continuous strike activities,to stagnant productivity (Bhorat 2004, Burger and Woolard 2005, Oosthuizen2006).

Table 7 presents the ‘simple elasticity’ estimates that describe the relation-ship between output and employment, and is calculated as: average annualpercentage change of employment / average annual percentage change of realGDP.5 First of all, the table shows that the simple real GDP ‘elasticity’ of totalemployment increased from 0.74 (when comparing OHS 1995 with LFS 2004b)to 0.92 (when comparing LFS 2004b with QLFS 2013Q4). Furthermore, the‘elasticity’ of formal sector employment increased from 0.86 (1997-2004) to 1.17(2004-2013).

As far as formal and informal sector employment is concerned, Table A.3in the Appendix shows that the non-agricultural informal sector employmenthas been hovering around the 2.0-2.5 million ranges since 2005, while non-agricultural formal sector employment increased steadily from 8.0 million to10.8 million between 2005 and 2013. Kingdon and Knight (2005) explained thatthere are some constrains to entry into the informal sector such as lack of ac-cess to credit, lack of infrastructure and lack of training. The informal sector isalso characterised by low remuneration, weak job security, and lack of pensionfund as well as other benefit. These constrains impede people from joining theinformal sector and hence this resulted in an increase of unemployment.

Finally, Table A.2 shows that the number of self-employed has stabilized

5As mentioned by Oosthuizen (2006: 8), formal modelling is needed to control for differentvariables that could impact on the relationship between output and employment. Hence, thefigures presented in Table 7 are, strictly speaking, not output-employment elasticities.

12

at approximately 2.2 million in recent years, while the number of employeesshowed an upward trend in general, reaching a peak level of 13.1 million inQLFS 2013Q4.

To conclude, there has clearly been a structural change in the South Africanlabour market, as indicated by the more rapid increase of employment in thehighly-skilled occupations and tertiary-sector industries, for those with post-Matric qualifications.

3.3 Unemployment

This section analyses the demographic characteristics of the unemployed andthe unemployment rates between 1995 and 2013. Table 8 depicts the numberof unemployed estimated under the narrow definition. This number increasedby about 2.8 million over the two decades measuring slightly over 4.8 millionin 2013, from 2.0 million in 1995. Blacks accounted for the highest share ofunemployed individuals in all three surveys (around 85 per cent). Also, theblacks accounted for the largest share of the increase of unemployed (86.8 percent between 1995 and 2013). As far as gender is concerned, it is interesting thatin 1995 and 2004, the number of male unemployed was lower than the numberof female unemployed, but this no longer took place in 2014. This finding couldbe partly attributed to policy reforms such as the Affirmative Action.

At the provincial level, Gauteng accounted for the highest share of unem-ployment (33.7 per cent in 2013). It is also the province accounting for thehighest share of increase of unemployed between 1995 and 2013 (40.6 per cent).In conjunction, the share of unemployment in rural areas fell from 36.5 per centin 1995 to 21.8 per cent in 2013. This result could be explained by the migrationof workseekers from rural areas to urban areas.

Individuals with incomplete secondary schooling have consistently recordedthe highest share of unemployment levels (averaging 50 per cent) in each of theperiods in question. By contrast, the share of unemployed individuals with littleor no primary schooling decreased from 25.4 per cent in 1995 to 7.3 per cent in2013. The benefits that education affords are evident in the lower share of 7 percent that the unemployed with post-Matric qualifications hold, despite nearlydoubling from the 3.1 per cent recorded in 1995.

Approximately 1.9 million unemployed individuals were aged between 25 and34 years at the time of the survey, accounting for the largest share of growthover the two decades (39.3 per cent). For individuals aged 15-24 years, the shareof unemployed initially increased then decreased from 33.4 per cent (2004) to26.4 per cent (2013), coinciding with delayed entry into the labour market dueto further studies. During the same period, 35-44 year olds recorded the reverseincreasing to 22 per cent. The combined share of those aged between 45 and65 years account for just above 13 per cent of unemployment growth between1995 and 2013. Finally, the severity of youth unemployment among 18-29 yearolds is evidenced by its share of total unemployed in excess of 49 per cent inthe three surveys assessed. These findings once again justify the introductionof the Employment Tax Incentives Bill.

13

In summary, the unemployed in 2013 were more likely to be youngsters below30 years, Black males, residing in urban areas in the Gauteng province, withincomplete secondary schooling.

Table 9 presents the unemployment rates in the three surveys under study.It can be seen that the national unemployment rate increased by 8.6 percentagepoints during the period from 1995 to 2004, before declining slightly to 24.1 percent in 2013. These trends are repeated when considering the racial profiles ofBlacks and Indians, recording unemployment rates of about 27 and 12 per centrespectively in 2013, while the unemployment rates for the Coloureds and whitesincreased continuously across the surveys. Nonetheless, the unemployment rateremained the lowest for the whites in all three surveys. In addition, femalesrecorded higher unemployment rates in all three surveys, but the gap betweenthe female and male rates narrowed from 9.1 percentage points in 1995 to only3.9 percentage points in 2013.

In light of the provincial demarcations and urban-rural influences on un-employment, the highest provincial unemployment rates were recorded in theEastern Cape in 1995 and 2004 at 24.3 per cent and 29.6 per cent respectively,being surpassed by the Free State (33 per cent) in 2013. The latter increasedfrom the lowest recorded rate of 12.4 per cent in 1995, which may be attributedto the shortage of job opportunities in rural areas. The Western Cape is consis-tently cited among the lowest recorded rates of unemployment, reflecting levelsranging from approximately 14 to 21 per cent between 1995 and 2013. Limpopo(one of the poorest provinces) experienced the lowest unemployment rate of 16.9per cent in 2013, declining from 27.8 per cent in 2004. However, this result is notthat surprising when referring to Table 4, which shows that Limpopo had thehighest EAR between 1995 and 2013. The urban-rural dynamic has a minimaleffect on the margin of difference observed between the unemployment rates,2013 a case in point where 24.3 and 23.3 per cent were respectively recorded forurban and rural areas.

In 2013, individuals with incomplete secondary schooling and Matric expe-rienced the highest recorded unemployment rates at levels ranging between 26and 30 per cent, despite declining over the last decade. In conjunction with thedeclined unemployment levels, those having no and incomplete primary school-ing each account for unemployment rates of about 18 per cent in 2013, whileindividuals with post-Matric qualifications recorded the lowest levels of 6 percent (degree-holders) and 14 per cent (post-Matric certificate/diploma). Thesignificance of education in gaining access to better job opportunities withinthe labour market is clearly in evidence, based on the lowest unemploymentrate afforded to individuals holding degrees. Furthermore, the unemploymentrates declined across the elderly age cohorts in all three surveys. Specificallyfocusing on the 18-29 year olds, the unemployment rate increased from 29.1 percent in 1995 to 39.5 per cent in 2013.

Finally, with reference to Figure 5, there is a disconcerting upward trendin both the number and proportion of unemployed who never worked duringthe period under study. Furthermore, this proportion was extremely high at

14

71.4 per cent for the 15-24 year-old in QLFS 2013Q4.6 This finding stronglysuggests the struggle encountered by most of the unemployed, particularly theyouth, in their quest to find their first job. As a consequence, there is need forgovernment support in terms of active labour market policy (i.e. EmploymentTax Incentives Bill, Expanded Public Works Program, job-seeking transportsubsidy, etc.).

In summary, the highest unemployment rates for 2013 were observed amongBlack females, aged between 15 and 24 years, residing in Free State, with in-complete secondary schooling, and who never worked before.

4 A multivariate analysis of labour force partic-

ipation and employment

In order to interpret the results of the multivariate regressions performed on thedata, we will first outline the methodology used in order to provide comparabil-ity, transparency and to aid with understanding the results themselves. Thatsaid, the following methodology was used in the regression analysis: Firstly,probit regressions are run on the labour force participation likelihood of theworking-age population for OHS 1995, LFS 2004b and QLFS 2013Q4. Themarginal fixed effects (MFXs) for these regressions are then calculated. TheMFX measures the instantaneous rate of change of a variable. Put differently,the MFX provides a good approximation of the change in the dependent variablefor a 1 unit change in the independent variable. For our analysis purposes, thesesMFX’s indicate the percentage change in labour force participation likelihoodfor a particular variable.

Secondly, Heckprobit regressions run on employment likelihood of the labourforce, conditional on labour force participation. Since not everyone in theworking-age population joined the labour force and eventually found employ-ment, the results of a probit regression on employment likelihood of the working-age population would be biased due to sample selection. Hence, the most com-mon technique applied to address this problem is a two-step Heckprobit model.The first step is a probit analysis to identify the factors determining whethersomeone in the working-age population would join the LF or not. The equationallows the estimation of the inverse Mills ratio (i.e., lambda), which is in turnincluded in the employment probit (i.e., the second step), making the latterregression conditional on labour force participation. If the inverse Mills ratiovariable is statistically significant in this probit, it indicates that the labourforce indeed differ from their counterparts who decided not to participate in thelabour force, and the two-step Heckman approach is necessary. These Heckpro-bit regressions are run in the same three surveys as mentioned above.

Table 10 shows the results of the labour force participation probit regressionsfor the three surveys under study. The reference group for these regressions were

6The proportion of unemployed who never worked before in the other age cohorts in QLFS2013Q4 were as follows: 25-34 years — 42.2 per cent; 35-44 years — 17.2 per cent; 45-54 years— 6.4 per cent; 55-65 years — 4.7 per cent.

15

black males residing in the rural areas of the Eastern Cape, with no formal ed-ucation and aged between 15 and 24 years old. From these probit regressionresults we see the following findings: Firstly, gender plays a major role in em-ployment likelihood, as females remain significantly less likely to participate inthe labour force than males, across all three surveys. The trend in this fact isnot yet clear, as from OHS 1995 to LFS 2004b it decreases from females being16.34 per cent less likely than males, to 12.78 per cent less likely to participatein the labour force. However, QLFS 2013Q4 again shows an increase at 13.44per cent. Thus, no definitive conclusion with regard to gender equity in thelabour force can be inferred from the data.

With regard to race, Coloureds are significantly more likely to enter labourforce across all three surveys when compared to Blacks. However this effectappears to be diminishing as its decreases from 11.47 per cent in OHS 1995 to2.01 per cent in QLFS 2013Q4. On the other hand, the opposite occurs withWhites, as the labour force participation likelihood is negative and significantacross all three surveys, again when compared to Blacks. This trend seemsto become increasingly negative from -1.82 per cent in OHS 1995 to -7.52 percent in QLFS 2013Q4. Similarly, Indians show increasingly negative labourforce participation likelihood versus the reference Black category, but to a lesserdegree, while the OHS 1995 result is statistically insignificant. Therefore, racestill remains a significant factor when determining the labour force participationlikelihood of an individual.

When examining the age characteristics, we see that across all the surveysand for all age cohorts, except for those between the age of 55 and 65 years inOHS 1995 and LFS 2004b, labour force participation likelihood is significantlygreater than the reference group (15-24 years). In addition to this fact, whenexamining the endpoints (OHS 1995 and QLFS 2013Q4) each age category showsan increasing trend in labour market participation likelihood over the reference15 to 24 age category. Thus, the data could reflect an increasing trend indiscouraged workers among the 15 to 24 age category.

Furthermore, those residing in urban areas exhibit a significantly greaterprobability of entering the labour force, as opposed to their rurally residingcounterparts. This trend is also increasing across all three survey years, from2.53 per cent in OHS 1995 to 10.97 per cent in the QLFS 2013Q4 period. Thus,as one would expect, residing in an urban area plays a significant role in labourforce participation likelihood, most likely due to the increased employment op-portunities available in these areas.

When looking at the province of residence, the data shows three distinctcategories. Firstly, those residing in Western Cape, Free State, Gauteng andMpumalanga are significantly more likely to participate in the labour force thanthose residing in Eastern Cape, across all three surveys. The trend among theseprovinces, with the exception of Mpumalanga tends to be erratic as they declinein the LFS 2004b and proceed to increase in QLFS 2013Q4. The aforementionedMpumalanga province shows a continual increase in labour force participationlikelihood from 5.92 per cent in OHS 1995 to 10.00 per cent in QLF S2013Q4.Secondly, the Northern Cape, KwaZulu-Natal and North West provinces ex-

16

hibit no definite trend when compared to the Eastern Cape across the surveyperiods. In addition, they either have statistically less significant or entirelyinsignificant MFXs for any one particular survey period. Lastly, compared toEastern Cape residents, Limpopo residents are less likely to enter the labourforce in all surveys. The data therefore shows that the province of residenceplays a significant role in the labour force participation likelihood of an individ-ual. In particular, provinces known for being economic centres show positive,significant labour force participation likelihoods most likely due to the largernumber of employment opportunities on offer.

In terms of education, in general, as educational attainment increases there isa significantly greater likelihood of entering the labour force. One exception forthis rule occurs in OHS 1995 and is characterised by the unusual negative MFXvalue for the incomplete secondary education variable. The other is shown bythe statistically insignificant MFX values for the degree education spline in bothOHS 1995 and LFS 2004b. Consequently, the degree variable is only statisticallysignificant in the QLFS 2013Q4 survey. Thus, only during this survey period didan individual possessing a bachelor’s degree have a significantly better chance ofparticipating in the labour force, as opposed to someone having a post-Matriccertificate/diploma. This point is indicative of a structural change in the labourmarket or a possible increase in the demand for relatively more educated, skilledpeople in the labour force in 2013. The data therefore shows that increased levelsof education do significantly increase the labour force participation likelihoodof an individual.

With regard to the household head variable, those who are household headsare significantly more likely to enter labour force across all the survey periods.However, a diminishing trend is noticeable across the periods as well. The signif-icance of this variable can be attributed the need for the head of the householdto be the breadwinner in the family. In addition, the psychological factors suchas improved self-esteem may also play a role in the household head’s decisionto participate in the labour force. Thus, the head of the household is morelikely to participate in the labour force as there is an extra incentive to do so.Similarly, those who are married or are living with a partner are significantlymore likely to participate in the labour force. The reason for this may againbe psychological as the need to take care of their partner may provide an extraincentive to labour force participation.

The number of children in the household variable shows a significant andincreasingly negative trend over the survey periods, as the MFX decreases from-1.83 per cent in OHS 1995 to -2.66 per cent in QLFS 2013Q4. When takinginto account the fact that the reference group is male, this is somewhat counter-intuitive, as one would expect a female reference group to become more likelyto stay home, as the number of children increases in order to serve as primarycaregivers. However, legislative changes toward gender equality may be able toexplain why men may now also be more likely to stay home and take care ofchildren. On the other hand, a more practical reason for this would be thatthe greater the number of children in a household, the greater the amount ofincome that can be claimed through social grants. We would expect that this

17

substitution of labour market income for social grant income would be moreprevalent in lower income households, where the income derived from labourforce participation is similar to social grant income. Thus, the labour forceparticipation likelihood of individuals will decrease as the number of children inthe household increases for few reasons.

In contrast to the number of children in a household, the presence of moremale members in the household leads to a significantly greater likelihood ofentering the labour force in the LFS 2004b and QLFS 2013Q4 survey periods,while the OHS 1995 result is shown to be statistically insignificant. The reasonfor this emerging behaviour could once again be psychological, as an improvedself-esteem or even peer pressure may provide an extra incentive to these indi-viduals’ participation in the labour force. Similarly, the greater the number offemale members in the household, the greater the probability of an individualentering the labour force, and this finding is statistically significant in all threesurveys.

Finally, the probit regressions show that the presence of more elderly mem-bers in the household leads to a significantly lower likelihood of entering thelabour force across all three survey periods. In addition, the negative MFXin absolute terms is increasing over the survey periods. This could again bedue to the substitution of social grant income for labour market income. Thus,individuals in households with elderly members are less likely to participate inthe labour force as there is less incentive to do so.

Next, as detailed in the second part of the methodology outlined previously,Table 11 below shows the results of the labour force participation Heckprobitregressions for the three periods. The reference group for these regressions wereagain black males residing in the rural areas of the Eastern Cape, with no formaleducation and aged between 15 and 24 years old.

From these Heckprobit regression results we see the following results: Firstly,females are more likely than males to be employed, across all three survey pe-riods. This is in stark contrast to the results obtained on the first probit re-gression. However, the explanation for this trend is somewhat logical, as anincreased focus on gender equity in recent years has made it more attractivefor females to enter the labour force. This as both the wages and employmentopportunities afforded to females has shown more parity with their male coun-terparts. Thus, the employment likelihood of females has been positive acrossall survey periods, when compared to males.

With reference to the race variables, Whites are shown to be the group withthe greatest likelihood of being employed, followed by Indians and Coloureds.This again, is at odds with the declining employment likelihood for Whitesshown by the probit regression. The important inference that must be drawnfrom this point is that despite Affirmative Action (AA) and Black EconomicEmpowerment (BEE) initiatives being legislated, labour force trends still showpersistent disparities between Blacks and other races. On this point and withthe aid of the Oaxaca-Blinder decomposition, Burger and Jafta (2006) concludethat “affirmative action policies have therefore not been successful in its aimto redress the disadvantages in employment experienced by designated groups

18

[or] to ensure their equitable representation in all occupational categories andlevels in the workforce”. Therefore, as the data shows, Whites, Indians andColoureds are still significantly more likely than Blacks to be employed, despitethe fact that the MFXs in all three groups declined between LFS 2004b andQLFS 2013Q4.

When examining the age cohort variables, we see the following results, com-pared to the 15 to 24 year reference age category: those aged 35 to 44 years areonly significantly more likely to be employed in QLFS 2013Q4, while those 45to 54 years do so in LFS 2004b. In addition, more importantly, those aged 55to 65 years are most the likely to be employed across all three survey periods.This may be due to a lack of experience prevalent amongst younger members ofthe labour force, making older workers much more valuable to their respectivefields, or because older workers are less likely to resign as they are closer to theirretirement age and stand to lose more financially. Thus, for either reason, thosein the 55 to 65 age category are most likely to employed.

In the case of provincial variables, the results were mostly mixed. Theexception to this was Mpumalanga and the North West province which showedincreasingly negative labour force participation likelihoods when compared tothe reference Eastern Cape province. The aforementioned mixed results can beseen in the seemingly random and in some case, insignificant MFX values forthe Western Cape, Northern Cape, Free State, KwaZulu-Natal and Gauteng.On the other hand, the Limpopo province shows a positive and significant MFXvalue in comparison to the Eastern Cape and taking into account the previousprobit regression results. However, when one takes into account the fact thatthe EAR between 1995 and 2013 was the highest in Limpopo (Table 4) and thatunemployment rate in this province was the lowest in QLFS 2013Q4 (Table 9),the negative MFXs of this province in the Heckprobit regressions might not betoo surprising, and the results reflect the possible success of the government’sefforts to promote the less developed provinces.

In terms of the education splines in the Heckprobit regression, the resultsseem more logical than the probit regression results obtained earlier. To thisend, an individual possessing an education either up to or including Matric iseither just as likely, or less likely to find employment as someone with no formaleducation. Furthermore, both the Matric and Certificate or Diploma and Degreesplines show a statistically significant and positive MFX value. For the latterthis applies to just LFS 2004b and QLFS 2013Q4. However, this shows thatan individual possessing a degree or higher is significantly more likely to beemployed, meaning the demand trend for more highly educated workers hasbeen prevalent since much earlier than the probit regression in Table 10 implies.This also implies that in order to survive in the labour market an individualneeds to pursue educational attainment beyond Matric level. Thus, as one wouldexpect, the relatively more educated an individual is, the more likely they areto be employed, especially with regard to qualifications higher than Matric.

Lastly, the results show that lambda is statistically significant in the Heck-probit regressions. This is of vital importance as it justifies the need for thetwo-step Heckprobit regression. Furthermore, implies that if we were to run the

19

employment probit as a one-step probit, the sample would be biased and wouldlead to incorrect results an interpretation. Therefore, it is essential to run theemployment probit as a two-step Heckprobit regression, to control for labourforce participation and provide an accurate snapshot of employment likelihoodover the years.

5 Conclusion

This paper looked at the South African labour market trends for the past 18years (1995-2013). It found that the labour force and LFPR increased since theend of apartheid in 1994. This partly played a role in the persistent and highlevels of unemployment still seen today as some of these workers will remainforever unemployable due to various reasons (e.g. skills mismatch). On top ofthis, the TGR far exceeded the AGR for most demographic categories, whichsuggests that the extent of employment growth was not rapid enough to absorbthe net entrants into the labour force.

This fact in conjunction with barriers to entry in the informal sector andan overall shift towards relatively more skilled occupations and slow economicgrowth contributed to the unemployment figures seen currently. In addition,based on the results obtained from the multivariate analysis, the following wor-rying results can be seen: In terms of labour force participation likelihood, racestill plays a major role in labour force participation and consequently employ-ment likelihood. Given that equality is very important in the South Africancontext, this area must be revisited. In addition, the regression results showthat urban areas are increasingly not able to provide sufficient employment op-portunities to their inhabitants. On this point, provincial data confirms thatemployment creation is a significant problem even in the major provinces. In thisregard, measures to encourage employment creation outside of major cities needto be implemented. Furthermore, the results show that experienced older work-ers are still more likely to be employed. This has a two-fold implication for SouthAfrica. In the short-term, younger generations will have to deal with poverty,while in the long-term, problems surrounding generating economic growth evensocial unrest may arise. Thus, correcting this problem may avert future eco-nomic, social and political stability issues in South Africa.

In terms of the positive trends seen in the data, we see that over time thelabour force in South Africa has become relatively more educated on the back ofsecondary and tertiary employment growing significantly. Furthermore, formalsector employment has also shown significant increases over the survey periodwhich is indicative of an improved standard of living at least for some workers,since formal sector employment conditions are much better than informal sectorones. In addition, the Heckprobit results showed that, gender equality legislationin the workplace seems to have had an impact on the number of females beingemployed. Since developmental economic theory advocates that women arevital to improving the general standard of living in society, this trend shouldbe fostered. In addition, results obtained through the education splines show

20

that educational attainment post Matric is becoming increasingly important inthe context of the South African labour market. Given that studies have shownan inverse correlation between the level of education and social issues withina country, the flourishing demand for relatively more educated labour marketparticipants may prove beneficial. Thus, particular focus should be placed onintroducing an effective education system coupled with adequate resources, inorder to satiate this labour market demand trend. In conclusion, if the 6 percent unemployment rate as set out in the National Development Plan (NDP) isto be achieved, it is clear that difficult decisions regarding South African labourmarket legislation and policy, as outlined above, need to be addressed. At thesame time, any such decisions taken should ensure that future economic growthbenefits all facets of society so as to address unemployment holistically.

Based on the issues identified in this paper, some policy changes need tooccur in order to address certain problems in the labour market. Firstly, policiessuch as Affirmative Action (AA), Black Economic Empowerment (BEE) andBroad Based Black Economic Empowerment (BBBEE) need to be revisited,since Black individuals are still least likely to participate in the labour force.Moreover, Van der Berg (2011) points to a growing intra-racial inequality gapin terms of income, illustrating the need for policies aimed at benefiting allSouth African citizens. Perhaps, the emergence of the trend toward demandingrelatively more educated workers provides some guidance on where employmentequity policy should go. Specifically, as employing individuals on the basis ofeducational attainment alone can be beneficial by simply improving efficiency,which is essential for a free market system to be effective. However, given SouthAfrica’s history and the need for reparations this outcome is unlikely. Thus,when revising the policies at hand it is necessary to consider many factors, notjust economic ones and such a discussion would be beyond the scope of thispaper. What can be inferred from the data is that a multifaceted approach isrequired to solve South Africa’s labour market problems, as these stem problemsfrom various factors.

With this point in mind, education can play a pivotal role in turning aroundthe unemployment problem South Africa is dealing with. This view is also re-flected by Burger and Woolard (2005) and Van der Berg (2011). In order toachieve this goal and provide the solution to the unemployment problem, edu-cational policy needs to be decisive and have long-term benefits and economicgrowth goals in mind. The need for these reforms can be shown empiricallyas recent global studies have shown South African students to be of relativelypoor quality in relation to their peers. It is therefore clear that policy reformin education is required. Besides improving global competitiveness, providinga higher standard and level of education does have other benefits. This, as amore educated population is associated with lower levels of crime and generallyhigher standards of living as well. It may also go some way to improving socialcohesion and stability. Furthermore, educational reforms should also include anintegrated skills transfer aspect, performing a role similar to that of SectorialEducation and Training Authorities (SETAs). This would provide some mech-anism for transferring skills from the older generation to the younger ones, in

21

order to ensure that essential skills are not lost when the older generation leavesthe labour force. As with the previously mentioned employment equity legisla-tions, the specifics of the necessary reforms in education are beyond the scope ofthis paper. However, looking at the example set by neighbouring African coun-tries with arguably fewer resources, reforming South Africa’s educational policyis both possible and can be beneficial to economic growth and the creation ofemployment.

The next issue identified in the results is that of broad unemployment, whichseems to be prevalent in non-urban areas. To address this issue, Bhorat (2012)advocates a transport subsidy for unemployed youth, in addition to the alreadyintroduced youth wage subsidy. The aforementioned transport subsidy will gosome way to addressing the issue of economic opportunities being limited tourban areas and only benefiting those residing in or adjacent to urban areasas individual outside these areas will then also be able to access these employ-ment opportunities. These youth transport subsidies would also, by their natureprovide a certain amount of disincentive to urbanisation, which given its associ-ated problems, would provide both budgetary and social benefits. Furthermore,policies and incentives to decentralisation may be beneficial in the long run, butempirical and practical evidence, given the developmental state South Africa isin is non-existent.

Lastly, while not directly discernible from the results of the data analysis,labour unrest in South Africa needs to be addressed. This, as the effect oflabour unrest in recent years has become undisputable. Furthermore, this effecthas impacted on both private and public enterprise in South Africa. In thisregard, Bhorat (2012) associates highly regulated labour markets with relativelow-income countries. Furthermore, Kingdon and Knight (2007) cite legal andprocedural requirements as a source of problems for employers. In addition, Yu(2013) points to collective bargaining as exacerbating rather than alleviating theunemployment problem in South Africa. Thus, a shift toward a less regulatedlabour market environment may bring long-term employment benefits at thecost of short-term earnings. However, given the relative power of unions andlabour market entities in South Africa, this outcome is may be less than likely.

References

[1] Bhorat, H. (2004). Labour market challenges in the post-apartheid SouthAfrica. South African Journal of Economics 72(5): 940-977.

[2] Bhorat, H. (2009). Unemployment in South Africa: Descriptors and deter-minants. Proceedings of the IZA (Institute for the Study of Labour)/WorldBank Conference on Employment and Development.

[3] Bhorat, H. (2012). A nation in search of jobs: Six possible policy sug-gestions for employment creation in South Africa. DPRU Working Paper12/150. Cape Town: Development Policy Research Unit, University of CapeTown.

22

[4] Bhorat, H. Kanbur, R. & Mayet, N. (2012). The Impact of Sectoral Min-imum Wage Laws on Employment, Wages and Hours of Work in SouthAfrica. DPRU Working Paper 12/155. Cape Town: Development PolicyResearch Unit, University of Cape Town.

[5] Burger, R. & Jafta, R. (2006). Returns to race: Labour market discrim-ination in post-apartheid South Africa. Stellenbosch University WorkingPaper 4/2006. Stellenbosch: Department of Economics, University of Stel-lenbosch.

[6] Burger, R. & Woolard, I. (2005). The state of the labour market in SouthAfrica after the first decade of democracy. Journal of Vocational Educationand Training. 57(4): 453-476.

[7] Hodge, D. (2009). Growth, employment and unemployment in SouthAfrica. South African Journal of Economics. 77(4): 488-504.

[8] Kingdon, G.G & Knight, J. (1999). Unemployment and wages in SouthAfrica: A spatial approach. Centre for the Study of African Economies.Oxford: Institute of Economics and Statistics, University of Oxford.

[9] Kingdon, G.G. & Knight, J. (2004). Unemployment in South Africa: Thenature of the beast. World Development: 32(3): 391-408.

[10] Kingdon, G.G. & Knight, J. (2007). Unemployment in South Africa, 1995-2003: Causes, problems and policies. Journal of African Economies. 16(5):813-848.

[11] Oosthuizen, M. (2006). The post-apartheid labour market: 1995-2004.DPRU Working Paper 06/103. Cape Town: Development Policy ResearchUnit, University of Cape Town.

[12] Pauw, K. Bhorat, H. Goga, S. Ncube, L. & Van der Westhuizen, C. (2006).Graduate unemployment in the context of skills shortages, education andtraining: Findings from a firm survey. DPRU Working Paper 06/115. CapeTown: Development Policy Research Unit, University of Cape Town.

[13] Van der Berg, S. (2011). Current poverty and income distribution in thecontext of South African history. Economic History of Developing Regions.26(1): 120-140.

[14] Yu, D. (2008). The South African labour market 1995-2006. StellenboschEconomic Working Paper 05/2008. Stellenbosch: Stellenbosch University.

[15] Yu, D. (2009). The comparability of Labour Force Survey (LFS) and Quar-terly Labour Force Survey (QLFS). Stellenbosch Economic Working Paper08/2009. Stellenbosch: Stellenbosch University.

[16] Yu, D. (2013). Revisiting unemployment levels and trends in South Africasince transition. Development Southern Africa. 30(6): 701-723.

23

Figure 1: Labour force number and labour force participation rates, 1995-2013

Figure 2: Number of employed, 1995-2013

24

Figure 3: Number of unemployed and unemployment rates, 1995-2013

Figure 4: Annual percentage growth of employment versus annual percentage growth of real gross value added (2005 prices) by industry, 1995-2013

25

Figure 5: Number and proportion of unemployed who never worked before, selected surveys

26

Table 1: Labour force under the narrow definition: OHS 1995, LFS 2004b and QLFS 2013Q4

LF Number Share 1995 vs. 2004 2004 vs. 2013 1995 vs. 2013

OHS 1995

LFS 2004b

QLFS 2013Q4

OHS 1995

LFS 2004b

QLFS 2013Q4 Number Share Number Share Number Share

Total Total 11 527 589 15 761 080 20 022 751 100.0% 100.0% 100.0% 4 233 491 100.0% 4 261 671 100.0% 8 495 162 100.0%

Race

Black 7 829 299 11 453 770 15 217 215 67.9% 72.7% 76.0% 3 624 471 85.6% 3 763 445 88.3% 7 387 916 87.0%

Coloured 1 361 640 1 657 357 2 102 630 11.8% 10.5% 10.5% 295 717 7.0% 445 273 10.4% 740 990 8.7%

Indian 401 060 483 742 565 416 3.5% 3.1% 2.8% 82 682 2.0% 81 674 1.9% 164 356 1.9%

White 1 935 590 2 130 253 2 137 490 16.8% 13.5% 10.7% 194 663 4.6% 7 237 0.2% 201 900 2.4%

Gender Male 6 712 969 8 791 142 10 971 891 58.2% 55.8% 54.8% 2 078 173 49.1% 2 180 749 51.2% 4 258 922 50.1%

Female 4 814 620 6 961 048 9 050 860 41.8% 44.2% 45.2% 2 146 428 50.7% 2 089 812 49.0% 4 236 240 49.9%

Age

15-24 years 1 769 981 2 668 275 2 602 218 15.4% 16.9% 13.0% 898 294 21.2% -66 057 -1.6% 832 237 9.8%

25-34 years 4 096 707 5 615 409 6 789 997 35.5% 35.6% 33.9% 1 518 702 35.9% 1 174 588 27.6% 2 693 290 31.7%

35-44 years 3 224 181 3 824 276 5 684 393 28.0% 24.3% 28.4% 600 095 14.2% 1 860 117 43.6% 2 460 212 29.0%

45-54 years 1 739 518 2 572 292 3 470 778 15.1% 16.3% 17.3% 832 774 19.7% 898 486 21.1% 1 731 260 20.4%

55-65 years 697 202 1 080 828 1 475 365 6.0% 6.9% 7.4% 383 626 9.1% 394 537 9.3% 778 163 9.2%

18-29 years 3 716 161 5 435 531 6 000 263 32.2% 34.5% 30.0% 1 719 370 40.6% 564 732 13.3% 2 284 102 26.9%

Province

WC 1 569 818 2 075 382 2 829 872 13.6% 13.2% 14.1% 505 564 11.9% 754 490 17.7% 1 260 054 14.8%

EC 1 211 519 1 812 095 1 845 068 10.5% 11.5% 9.2% 600 576 14.2% 32 973 0.8% 633 549 7.5%

NC 267 445 301 685 439 290 2.3% 1.9% 2.2% 34 240 0.8% 137 605 3.2% 171 845 2.0%

FS 858 356 1 087 130 1 111 115 7.4% 6.9% 5.5% 228 774 5.4% 23 985 0.6% 252 759 3.0%

KZN 2 160 459 2 929 125 3 156 293 18.7% 18.6% 15.8% 768 666 18.2% 227 168 5.3% 995 834 11.7%

NW 904 977 1 158 148 1 193 925 7.9% 7.3% 6.0% 253 171 6.0% 35 777 0.8% 288 948 3.4%

GAU 3 129 719 4 129 582 6 459 675 27.1% 26.2% 32.3% 999 863 23.6% 2 330 093 54.7% 3 329 956 39.2%

MPU 699 106 1 047 320 1 579 344 6.1% 6.6% 7.9% 348 214 8.2% 532 024 12.5% 880 238 10.4%

LIM 726 190 1 220 613 1 408 169 6.3% 7.7% 7.0% 494 423 11.7% 187 556 4.4% 681 979 8.0%

Area type Urban 7 597 108 11 576 677 15 513 215 65.9% 73.5% 77.5% 3 979 569 94.0% 3 936 538 92.4% 7 916 107 93.2%

Rural 3 930 481 4 184 403 4 509 536 34.1% 26.5% 22.5% 253 922 6.0% 325 133 7.6% 579 055 6.8%

Education

None 911 537 845 697 452 599 7.9% 5.4% 2.3% -65 840 -1.6% -393 098 -9.2% -458 938 -5.4%

Incomplete primary 1 913 406 2 094 435 1 460 425 16.6% 13.3% 7.3% 181 029 4.3% -634 010 -14.9% -452 981 -5.3%

Incomplete secondary 4 681 861 6 438 994 8 209 848 40.6% 40.9% 41.0% 1 757 133 41.5% 1 770 854 41.6% 3 527 987 41.5%

Matric 2 532 532 4 340 699 6 366 755 22.0% 27.5% 31.8% 1 808 167 42.7% 2 026 056 47.5% 3 834 223 45.1%

Matric + Cert/Dip 939 439 1 111 537 1 830 142 8.1% 7.1% 9.1% 172 098 4.1% 718 605 16.9% 890 703 10.5%

Degree 456 321 776 868 1 522 666 4.0% 4.9% 7.6% 320 547 7.6% 745 798 17.5% 1 066 345 12.6%

Other/Not specified 92 493 152 850 180 316 0.8% 1.0% 0.9% 60 357 1.4% 27 466 0.6% 87 823 1.0%

27

Table 2: Labour force participation rates under the narrow definition: OHS 1995, LFS 2004b and QLFS 2013Q4

LFPR

OHS 1995 LFS 2004b QLFS 2013Q4

Total Total 47.7% 53.8% 56.8%

Race

Black 43.1% 50.6% 54.7%

Coloured 60.1% 61.8% 64.1%

Indian 56.3% 58.8% 58.4%

White 63.3% 68.8% 67.4%

Gender Male 58.2% 62.0% 63.4%

Female 38.0% 46.2% 50.5%

Age

15-24 years 21.7% 28.2% 25.5%

25-34 years 63.9% 71.3% 73.7%

35-44 years 69.6% 73.3% 77.8%

45-54 years 62.7% 66.3% 70.0%

55-65 years 31.7% 37.9% 41.7%

18-29 years 41.3% 50.3% 50.3%

Province

WC 61.7% 66.0% 68.1%

EC 35.8% 45.3% 45.2%

NC 52.9% 53.4% 58.0%

FS 52.3% 57.0% 59.7%

KZN 44.0% 49.5% 47.9%

NW 45.1% 48.3% 50.4%

GAU 61.9% 65.9% 69.9%

MPU 43.0% 54.0% 58.9%

LIM 28.8% 39.0% 40.4%

Area type Urban 56.4% 62.2% 64.9%

Rural 36.7% 39.3% 39.9%

Education

None 40.1% 41.4% 36.1%

Incomplete primary 46.8% 46.8% 44.8%

Incomplete secondary 40.0% 45.6% 46.8%

Matric 61.4% 69.8% 70.7%

Matric + Cert/Dip 78.1% 86.6% 85.5%

Degree 82.0% 86.7% 88.8%

Other/Not specified 41.6% 66.2% 60.0%

28

Table 3: Number of employed: OHS 1995, LFS 2004b and QLFS 2013Q4

Number of employed Share 1995 vs. 2004 2004 vs. 2013 1995 vs. 2013

OHS 1995

LFS 2004b

QLFS 2013Q4

OHS 1995

LFS 2004b

QLFS 2013Q4

Number Share Number Share Number Share

Total Total 9 499 347 11 630 196 15 195 491 100.0% 100.0% 100.0% 2 130 849 100.0% 3 565 295 100.0% 5 696 144 100.0%

Race

Black 6 136 137 7 866 030 11 095 766 64.6% 67.6% 73.0% 1 729 893 81.2% 3 229 736 90.6% 4 959 629 87.1%

Coloured 1 144 836 1 296 317 1 619 629 12.1% 11.1% 10.7% 151 481 7.1% 323 312 9.1% 474 793 8.3%

Indian 358 589 418 797 494 820 3.8% 3.6% 3.3% 60 208 2.8% 76 023 2.1% 136 231 2.4%

White 1 859 785 2 014 698 1 985 276 19.6% 17.3% 13.1% 154 913 7.3% -29 422 -0.8% 125 491 2.2%

Gender Male 5 789 311 6 764 751 8 519 684 60.9% 58.2% 56.1% 975 440 45.8% 1 754 933 49.2% 2 730 373 47.9%

Female 3 710 036 4 860 273 6 675 807 39.1% 41.8% 43.9% 1 150 237 54.0% 1 815 534 50.9% 2 965 771 52.1%

Age

15-24 years 1 124 324 1 287 063 1 329 385 11.8% 11.1% 8.7% 162 739 7.6% 42 322 1.2% 205 061 3.6%

25-34 years 3 275 749 3 944 374 4 869 104 34.5% 33.9% 32.0% 668 625 31.4% 924 730 25.9% 1 593 355 28.0%

35-44 years 2 858 183 3 129 906 4 622 237 30.1% 26.9% 30.4% 271 723 12.8% 1 492 331 41.9% 1 764 054 31.0%

45-54 years 1 586 764 2 266 227 3 021 899 16.7% 19.5% 19.9% 679 463 31.9% 755 672 21.2% 1 435 135 25.2%

55-65 years 654 327 1 002 626 1 352 866 6.9% 8.6% 8.9% 348 299 16.3% 350 240 9.8% 698 539 12.3%

18-29 years 2 633 843 3 128 797 3 628 091 27.7% 26.9% 23.9% 494 954 23.2% 499 294 14.0% 994 248 17.5%

Province

WC 1 353 355 1 689 152 2 236 564 14.2% 14.5% 14.7% 335 797 15.8% 547 412 15.4% 883 209 15.5%

EC 917 098 1 276 170 1 332 779 9.7% 11.0% 8.8% 359 072 16.9% 56 609 1.6% 415 681 7.3%

NC 212 901 227 910 330 045 2.2% 2.0% 2.2% 15 009 0.7% 102 135 2.9% 117 144 2.1%

FS 752 051 776 099 744 876 7.9% 6.7% 4.9% 24 048 1.1% -31 223 -0.9% -7 175 -0.1%

KZN 1 712 758 2 089 722 2 529 716 18.0% 18.0% 16.6% 376 964 17.7% 439 994 12.3% 816 958 14.3%

NW 749 330 833 907 868 444 7.9% 7.2% 5.7% 84 577 4.0% 34 537 1.0% 119 114 2.1%

GAU 2 637 048 3 067 735 4 831 651 27.8% 26.4% 31.8% 430 687 20.2% 1 763 916 49.5% 2 194 603 38.5%

MPU 583 856 787 662 1 150 933 6.1% 6.8% 7.6% 203 806 9.6% 363 271 10.2% 567 077 10.0%

LIM 580 950 881 839 1 170 483 6.1% 7.6% 7.7% 300 889 14.1% 288 644 8.1% 589 533 10.3%

Area type Urban 6 309 317 8 637 002 11 738 367 66.4% 74.3% 77.2% 2 327 685 109.2% 3 101 365 87.0% 5 429 050 95.3%

Rural 3 190 030 2 993 194 3 457 124 33.6% 25.7% 22.8% -196 836 -9.2% 463 930 13.0% 267 094 4.7%

Education

None 770 646 720 256 370 633 8.1% 6.2% 2.4% -50 390 -2.4% -349 623 -9.8% -400 013 -7.0%

Incomplete primary 1 538 685 1 564 795 1 189 140 16.2% 13.5% 7.8% 26 110 1.2% -375 655 -10.5% -349 545 -6.1%

Incomplete secondary 3 682 335 4 320 886 5 729 298 38.8% 37.2% 37.7% 638 551 30.0% 1 408 412 39.5% 2 046 963 35.9%

Matric 2 093 433 3 138 018 4 739 794 22.0% 27.0% 31.2% 1 044 585 49.0% 1 601 776 44.9% 2 646 361 46.5%

Matric + Cert/Dip 888 596 1 001 154 1 578 569 9.4% 8.6% 10.4% 112 558 5.3% 577 415 16.2% 689 973 12.1%

Degree 444 862 752 183 1 433 366 4.7% 6.5% 9.4% 307 321 14.4% 681 183 19.1% 988 504 17.4%

Other/Not specified 80 790 132 904 154 691 0.9% 1.1% 1.0% 52 114 2.4% 21 787 0.6% 73 901 1.3%

29

Table 4: Target growth rates, actual growth rates and employment absorption rates

OHS 1995 vs. LFS 2004b LFS 2004b vs. QLFS 2013Q4 OHS 1995 vs. QLFS 2013Q4

TGR AGR EAR TGR AGR EAR TGR AGR EAR

Total Total 44.6% 22.4% 50.3% 36.6% 22.4% 61.2% 89.4% 60.0% 67.1%

Race

Black 59.1% 28.2% 47.7% 47.8% 28.2% 58.9% 120.4% 80.8% 67.1%

Coloured 25.8% 13.2% 51.2% 34.3% 13.2% 38.5% 64.7% 41.5% 64.1%

Indian 23.1% 16.8% 72.8% 19.5% 16.8% 86.1% 45.8% 38.0% 82.9%

White 10.5% 8.3% 79.6% 0.4% 8.3% 2318.9% 10.9% 6.7% 62.2%

Gender Male 35.9% 16.8% 46.9% 32.2% 16.8% 52.3% 73.6% 47.2% 64.1%

Female 57.9% 31.0% 53.6% 43.0% 31.0% 72.1% 114.2% 79.9% 70.0%

Age

15-24 years 79.9% 14.5% 18.1% -5.1% 14.5% -282.0% 74.0% 18.2% 24.6%

25-34 years 46.4% 20.4% 44.0% 29.8% 20.4% 68.5% 82.2% 48.6% 59.2%

35-44 years 21.0% 9.5% 45.3% 59.4% 9.5% 16.0% 86.1% 61.7% 71.7%

45-54 years 52.5% 42.8% 81.6% 39.6% 42.8% 108.0% 109.1% 90.4% 82.9%

55-65 years 58.6% 53.2% 90.8% 39.4% 53.2% 135.3% 118.9% 106.8% 89.8%

18-29 years 65.3% 18.8% 28.8% 18.0% 18.8% 104.1% 86.7% 37.7% 43.5%

Province

WC 37.4% 24.8% 66.4% 44.7% 24.8% 55.5% 93.1% 65.3% 70.1%

EC 65.5% 39.2% 59.8% 2.6% 39.2% 1515.4% 69.1% 45.3% 65.6%

NC 16.1% 7.0% 43.8% 60.4% 7.0% 11.7% 80.7% 55.0% 68.2%

FS 30.4% 3.2% 10.5% 3.1% 3.2% 103.5% 33.6% -1.0% -2.8%

KZN 44.9% 22.0% 49.0% 10.9% 22.0% 202.5% 58.1% 47.7% 82.0%

NW 33.8% 11.3% 33.4% 4.3% 11.3% 263.1% 38.6% 15.9% 41.2%

GAU 37.9% 16.3% 43.1% 76.0% 16.3% 21.5% 126.3% 83.2% 65.9%

MPU 59.6% 34.9% 58.5% 67.5% 34.9% 51.7% 150.8% 97.1% 64.4%

LIM 85.1% 51.8% 60.9% 21.3% 51.8% 243.5% 117.4% 101.5% 86.4%

Area Urban 63.1% 36.9% 58.5% 45.6% 36.9% 80.9% 125.5% 86.0% 68.6%

Rural 8.0% -6.2% -77.5% 10.9% -6.2% -56.8% 18.2% 8.4% 46.1%

Education

None -8.5% -6.5% 76.5% -54.6% -6.5% 12.0% -59.6% -51.9% 87.2%

Incomplete primary 11.8% 1.7% 14.4% -40.5% 1.7% -4.2% -29.4% -22.7% 77.2%

Incomplete secondary 47.7% 17.3% 36.3% 41.0% 17.3% 42.3% 95.8% 55.6% 58.0%

Matric 86.4% 49.9% 57.8% 64.6% 49.9% 77.3% 183.2% 126.4% 69.0%

Matric + Cert/Dip 19.4% 12.7% 65.4% 71.8% 12.7% 17.6% 100.2% 77.6% 77.5%

Degree 72.1% 69.1% 95.9% 99.2% 69.1% 69.7% 239.7% 222.2% 92.7%

Table 5: Employment by broad occupation category: OHS 1995, LFS 2004b and QLFS 2013Q4

OHS 1995 LFS 2004b QLFS2013Q4

Number

Highly-skilled 825 097 1 366 808 2 127 286

Legislators, senior officials and managers 499 595 908 768 1 233 013

Professionals 325 502 458 040 894 273

Semi-skilled 5 611 697 6 744 465 8 739 481

Technicians and associate professionals 1 058 897 1 148 089 1 639 907

Clerks 1 133 818 1 168 175 1 625 488

Service workers and shop and market sales 1 080 787 1 451 746 2 298 237

Skilled agricultural and fishery worker 114 486 328 213 75 295

Craft and related trade workers 1 117 053 1 536 315 1 848 480

Plant and machinery operators and assemblers 1 106 656 1 111 927 1 252 074

Unskilled 3 044 666 3 496 091 4 328 724

Elementary occupations 2 349 250 2 616 024 3 309 300

Domestic workers 695 416 880 067 1 019 424

Others/Unspecified 17 887 22 832 0

Others/Unspecified 17 887 22 832 0

Share

Highly-skilled 8.7% 11.8% 14.0%

Semi-skilled 59.2% 58.1% 57.5%

Unskilled 32.1% 30.1% 28.5%

100.0% 100.0% 100.0%

30

Table 6: Employment by broad industry category

OHS 1995

LFS 2004b

QLFS 2013Q4

Primary sector 1 673 951 1 465 181 1 140 807

Agriculture, hunting, forestry and fishing 1 233 552 1 060 893 714 851

Mining and quarrying 440 399 404 288 425 956

Secondary sector 1 964 091 2 634 449 3 099 788

Manufacturing 1 434 815 1 712 449 1 768 479

Electricity, gas and water supply 84 432 99 266 126 918

Construction 444 844 822 734 1 204 391

Tertiary 5 690 739 7 504 906 10 952 113

Wholesale and retail 1 665 345 2 539 864 3 233 002

Transport, storage and communication 476 005 562 628 963 023

Financial, insurance and business services 579 879 1 146 395 2 039 180

Community/social/personal services 2 171 561 2 182 449 3 471 732

Private households 797 949 1 073 570 1 245 176

Other / Unspecified 170 566 25 660 2 783

Other / Unspecified 170 566 25 660 2 783

Share

Primary sector 17.9% 12.6% 7.5%

Secondary sector 21.1% 22.7% 20.4%

Tertiary 61.0% 64.7% 72.1%

100.0% 100.0% 100.0%

Table 7: Employment elasticity to economic growth

Employment elasticity to economic growth

OHS 1995 LFS 2004b QLFS 2013Q4

Employment 9 499 347 11 630 196 15 195 491

Real GDP (2005 prices, R million) 1 134 582 1 492 330 1 993 433

1995 vs. 2004 2004 vs. 2013 1995 vs. 2013

Annual percentage change of employment 2.3% 3.0% 2.6%

Annual percentage change of real GDP 3.1% 3.3% 3.2%

∆ Employment / ∆ real GDP 0.74 0.92 0.83

Formal sector employment elasticity to economic growth

OHS 1997 LFS 2004b QLFS 2013Q4

Formal sector employment 6 436 017 7 684 843 10 780 187

Real GDP (2005 prices, R million) 1 214 768 1 492 330 1 993 433

1997 vs. 2004 2004 vs. 2013 1997 vs. 2013

Annual percentage change of formal sector employment 2.6% 3.8% 3.3%

Annual percentage change of real GDP 3.0% 3.3% 3.2%

∆ Formal sector employment / ∆ real GDP 0.86 1.17 1.03

31

Table 8: Number of unemployed under the narrow definition

Number of unemployed Share 1995 vs. 2004 2004 vs. 2013 1995 vs. 2013

OHS 1995

LFS 2004b

QLFS 2013Q4

OHS 1995

LFS 2004b

QLFS 2013Q4

Number Share Number Share Number Share

Total Total 2 028 242 4 130 884 4 827 260 100.0% 100.0% 100.0% 2 102 642 100.0% 696 376 100.0% 2 799 018 100.0%

Race

Black 1 693 162 3 587 740 4 121 449 83.5% 86.9% 85.4% 1 894 578 90.1% 533 709 76.6% 2 428 287 86.8%

Coloured 216 804 361 040 483 001 10.7% 8.7% 10.0% 144 236 6.9% 121 961 17.5% 266 197 9.5%

Indian 42 471 64 945 70 596 2.1% 1.6% 1.5% 22 474 1.1% 5 651 0.8% 28 125 1.0%

White 75 805 115 555 152 214 3.7% 2.8% 3.2% 39 750 1.9% 36 659 5.3% 76 409 2.7%

Gender Male 923 658 2 026 391 2 452 207 45.5% 49.1% 50.8% 1 102 733 52.4% 425 816 61.1% 1 528 549 54.6%

Female 1 104 584 2 100 775 2 375 053 54.5% 50.9% 49.2% 996 191 47.4% 274 278 39.4% 1 270 469 45.4%

Age

15-24 years 645 657 1 381 212 1 272 833 31.8% 33.4% 26.4% 735 555 35.0% -108 379 -15.6% 627 176 22.4%

25-34 years 820 958 1 671 035 1 920 893 40.5% 40.5% 39.8% 850 077 40.4% 249 858 35.9% 1 099 935 39.3%

35-44 years 365 998 694 370 1 062 156 18.0% 16.8% 22.0% 328 372 15.6% 367 786 52.8% 696 158 24.9%

45-54 years 152 754 306 065 448 879 7.5% 7.4% 9.3% 153 311 7.3% 142 814 20.5% 296 125 10.6%

55-65 years 42 875 78 202 122 499 2.1% 1.9% 2.5% 35 327 1.7% 44 297 6.4% 79 624 2.8%

18-29 years 1 082 318 2 306 734 2 372 172 53.4% 55.8% 49.1% 1 224 416 58.2% 65 438 9.4% 1 289 854 46.1%

Province

WC 216 463 386 230 593 308 10.7% 9.3% 12.3% 169 767 8.1% 207 078 29.7% 376 845 13.5%

EC 294 421 535 925 512 289 14.5% 13.0% 10.6% 241 504 11.5% -23 636 -3.4% 217 868 7.8%

NC 54 544 73 775 109 245 2.7% 1.8% 2.3% 19 231 0.9% 35 470 5.1% 54 701 2.0%

FS 106 305 311 031 366 239 5.2% 7.5% 7.6% 204 726 9.7% 55 208 7.9% 259 934 9.3%

KZN 447 701 839 403 626 577 22.1% 20.3% 13.0% 391 702 18.6% -212 826 -30.6% 178 876 6.4%

NW 155 647 324 241 325 481 7.7% 7.8% 6.7% 168 594 8.0% 1 240 0.2% 169 834 6.1%

GAU 492 671 1 061 847 1 628 024 24.3% 25.7% 33.7% 569 176 27.1% 566 177 81.3% 1 135 353 40.6%

MPU 115 250 259 658 428 411 5.7% 6.3% 8.9% 144 408 6.9% 168 753 24.2% 313 161 11.2%

LIM 145 240 338 774 237 686 7.2% 8.2% 4.9% 193 534 9.2% -101 088 -14.5% 92 446 3.3%

Area Urban 1 287 791 2 939 675 3 774 848 63.5% 71.2% 78.2% 1 651 884 78.6% 835 173 119.9% 2 487 057 88.9%

Rural 740 451 1 191 209 1 052 412 36.5% 28.8% 21.8% 450 758 21.4% -138 797 -19.9% 311 961 11.1%

Education

None 140 891 125 441 81 966 6.9% 3.0% 1.7% -15 450 -0.7% -43 475 -6.2% -58 925 -2.1%

Incomplete primary 374 721 529 640 271 285 18.5% 12.8% 5.6% 154 919 7.4% -258 355 -37.1% -103 436 -3.7%

Incomplete secondary 999 526 2 118 108 2 480 550 49.3% 51.3% 51.4% 1 118 582 53.2% 362 442 52.0% 1 481 024 52.9%

Matric 439 099 1 202 681 1 626 961 21.6% 29.1% 33.7% 763 582 36.3% 424 280 60.9% 1 187 862 42.4%

Matric + Cert/Dip 50 843 110 383 251 573 2.5% 2.7% 5.2% 59 540 2.8% 141 190 20.3% 200 730 7.2%

Degree 11 459 24 685 89 300 0.6% 0.6% 1.8% 13 226 0.6% 64 615 9.3% 77 841 2.8%

Other/Not specified 11 703 19 946 25 625 0.6% 0.5% 0.5% 8 243 0.4% 5 679 0.8% 13 922 0.5%

32

Table 9: Unemployment rates under the narrow definition

Unemployment rate

OHS 1995 LFS 2004b QLFS 2013Q4

Total Total 17.6% 26.2% 24.1%

Race

Black 21.6% 31.3% 27.1%

Coloured 15.9% 21.8% 23.0%

Indian 10.6% 13.4% 12.5%

White 3.9% 5.4% 7.1%

Gender Male 13.8% 23.1% 22.3%

Female 22.9% 30.2% 26.2%

Age

15-24 years 36.5% 51.8% 48.9%

25-34 years 20.0% 29.8% 28.3%

35-44 years 11.4% 18.2% 18.7%

45-54 years 8.8% 11.9% 12.9%

55-65 years 6.1% 7.2% 8.3%

18-29 years 29.1% 42.4% 39.5%

Province

WC 13.8% 18.6% 21.0%

EC 24.3% 29.6% 27.8%

NC 20.4% 24.5% 24.9%

FS 12.4% 28.6% 33.0%

KZN 20.7% 28.7% 19.9%

NW 17.2% 28.0% 27.3%

GAU 15.7% 25.7% 25.2%

MPU 16.5% 24.8% 27.1%

LIM 20.0% 27.8% 16.9%

Area Urban 17.0% 25.4% 24.3%

Rural 18.8% 28.5% 23.3%

Education

None 15.5% 14.8% 18.1%

Incomplete primary 19.6% 25.3% 18.6%

Incomplete secondary 21.3% 32.9% 30.2%

Matric 17.3% 27.7% 25.6%

Matric + Cert/Dip 5.4% 9.9% 13.7%

Degree 2.5% 3.2% 5.9%

Other/Not specified 12.7% 13.0% 14.2%

33

Table 10: Probit regression on labour force participation (under the narrow definition) likelihood

OHS1995 LFS2004b QLFS2013Q4

MFX X-bar MFX X-bar MFX X-bar

Female -0.1634***

0.5394 -0.1278***

0.5397 -0.1344***

0.5357

Coloured 0.1147***

0.1360 0.0427***

0.1374 0.0201**

0.1198

Indian 0.0007 0.0389 -0.0675***

0.0205 -0.1053***

0.0213

White -0.0182**

0.1306 -0.0769***

0.0766 -0.0752***

0.0719

25-34 years 0.3487***

0.2425 0.3354***

0.2314 0.3952***

0.2339

35-44 years 0.3473***

0.1918 0.3337***

0.1869 0.3948***

0.1808

45-54 years 0.2332***

0.1302 0.2328***

0.1406 0.3292***

0.1556

55-65 years -0.0797***

0.1080 0.0042 0.1019 0.0929***

0.1269

Urban 0.0253***

0.5380 0.0474***

0.5594 0.1097***

0.6297

WC 0.1428***

0.1029 0.0678***

0.1123 0.1116***

0.1282

NC 0.0706***

0.0417 -0.0184* 0.0653 0.0473

*** 0.0550

FS 0.1036***

0.0955 0.0292***

0.0741 0.0525***

0.0862

KZN 0.0774***

0.1988 -0.0027 0.2573 -0.0244**

0.1718

NW 0.0611***

0.0804 -0.0426***

0.0867 0.0036 0.0748

GAU 0.1629***

0.1092 0.0659***

0.1094 0.1105***

0.1611

MPU 0.0592***

0.0963 0.0633***

0.0733 0.1000***

0.0951

LIM -0.0672***

0.0949 -0.0721***

0.0928 -0.0388***

0.1124

Education spline: None to incomplete primary 0.0052***

5.1369 0.0079***

5.1920 0.0040* 5.5549

Education spline: Incomplete secondary -0.0076***

2.6215 0.0128***

2.6840 0.0269***

3.4034

Education spline: Matric 0.2150***

0.2315 0.2009***

0.2370 0.1589***

0.3317

Education spline: Matric + Cert/Dip 0.1224***

0.0714 0.1803***

0.0585 0.1368***

0.0983

Education spline: Degree -0.0052 0.0229 -0.0337 0.0196 0.0481**

0.0426

Household head 0.2990***

0.3115 0.2187***

0.3556 0.1753***

0.3622

Married or living with a partner 0.0950***

0.4502 0.0878***

0.3787 0.0633***

0.3466

Number of children 0-15 years in the household -0.0183***

1.8791 -0.0235***

1.7772 -0.0266***

1.5321

Number of male members 16-59 years in the household -0.0021 1.5624 0.0094***

1.4016 0.0056**

1.2873

Number of female members 16-59 years in the households 0.0128***

1.8099 0.0120***

1.6413 0.0259***

1.5078

Number of elderly 60+ years in the household -0.0397***

0.3619 -0.0527***

0.3208 -0.0614***

0.3427

Number of observations 80 387 67 871 54 568

Observed probability 0.4765 0.5385 0.5683

Predicted probability at x-bar 0.4599 0.5042 0.5424

Chi-squared statistic 28 510 21 132 20 244

Prob > Chi-squared 0.0000 0.0000 0.0000

Pseudo R-squared 0.2569 0.2246 0.2683 *** Significant at 1% ** Significant at 5% * Significant at 1%

34

Table 11: Heckprobit regression on employment likelihood, conditional on labour force participation

OHS1995 LFS2004b QLFS2013Q4

MFX X-bar MFX X-bar MFX X-bar

Female 0.0419***

0.4306 0.0742***

0.4705 0.0283***

0.4842

Coloured 0.0204***

0.1668 0.0740***

0.1560 0.0482***

0.1387

Indian 0.0763***

0.0461 0.1677***

0.0237 0.1293***

0.0231

White 0.1166***

0.1742 0.2063***

0.1015 0.1685***

0.0946

25-34 years -0.1148***

0.3260 -0.1831***

0.3160 -0.0821***

0.3188

35-44 years -0.0667***

0.2829 -0.0931***

0.2643 0.0002***

0.2564

45-54 years -0.0093 0.1726 0.0214***

0.1745 0.0793 0.1986

55-65 years 0.1092***

0.0735 0.2035* 0.0718 0.1935

*** 0.0881

Urban -0.0583***

0.6091 -0.0919***

0.6464 -0.0802***

0.7388

WC 0.0119* 0.1350 0.0332

*** 0.1388 -0.0249

*** 0.1619

NC -0.0147 0.0469 0.0145***

0.0674 -0.0072* 0.0559

FS 0.0378***

0.1061 -0.0294 0.0809 -0.0716 0.0940

KZN 0.0030 0.1911 -0.0013***

0.2265 0.0801***

0.1435

NW 0.0299***

0.0772 -0.0003 0.0773 -0.0021***

0.0672

GAU -0.0035 0.1445 -0.0370 0.1424 -0.0609 0.2081

MPU 0.0276***

0.0891 -0.0187***

0.0787 -0.0411***

0.0965

LIM 0.0371***

0.0665 0.0404* 0.0714 0.0939

*** 0.0786

Education spline: None to incomplete primary -0.0040***

5.2012 -0.0136***

5.3131 -0.0044 5.7026

Education spline: Incomplete secondary 0.0018 2.8317 -0.0139***

3.0248 -0.0195***

3.8356

Education spline: Matric -0.0647***

0.3310 -0.1069***

0.3486 -0.0042 0.4652

Education spline: Matric + Cert/Dip 0.0674***

0.1222 0.0681***

0.1018 0.0471***

0.1594

Education spline: Degree 0.0234 0.0408 0.0872***

0.0345 0.0743***

0.0709

Lambda -0.3095***

0.6255 -0.5297***

0.6103 -0.3345***

0.5376

Number of observations 37 189 33 784 28 930

Observed probability 0.8241 0.7379 0.7589

Predicted probability at x-bar 0.8837 0.7817 0.7879

Chi-squared statistic 6 611 7 229 4 253

Prob > Chi-squared 0.0000 0.0000 0.0000

Pseudo R-squared 0.1971 0.1845 0.1314 *** Significant at 1% ** Significant at 5% * Significant at 1%

35

Appendix Table A.1: Labour market trends, 1995-2013

Survey 15-65yrs Labour force

Employed Unemployed LFPR Unemployment rate Sector

Narrow Broad Narrow Broad Discouraged Narrow Broad Narrow Broad Formal Informal

OHS1995 24 190 583 11 527 589 13 731 073 9 499 347 2 028 242 4 231 726 2 203 484 47.7% 56.8% 17.6% 30.8% Not available

OHS1996 24 909 065 11 190 599 13 532 623 8 966 307 2 224 292 4 566 316 2 342 024 44.9% 54.3% 19.9% 33.7%

OHS1997 25 506 089 11 544 385 14 295 597 9 093 647 2 450 738 5 201 950 2 751 212 45.3% 56.0% 21.2% 36.4% 6 436 017 1 043 347

OHS1998 25 665 233 12 528 080 14 996 600 9 370 130 3 157 950 5 626 470 2 468 520 48.8% 58.4% 25.2% 37.5% 6 508 097 1 077 141

OHS1999 26 246 545 13 509 926 16 231 269 10 356 143 3 153 783 5 875 126 2 721 343 51.5% 61.8% 23.3% 36.2% 6 796 008 1 571 646

LFS2000a 26 465 110 16 205 643 18 424 127 11 874 409 4 331 234 6 549 718 2 218 484 61.2% 69.6% 26.7% 35.5% 6 672 951 1 819 556

LFS2000b 27 836 456 16 381 316 18 596 239 12 224 406 4 156 910 6 371 833 2 214 923 58.8% 66.8% 25.4% 34.3% 7 077 307 2 026 065

LFS2001a 28 062 004 16 668 067 19 361 231 12 260 207 4 407 860 7 101 024 2 693 164 59.4% 69.0% 26.4% 36.7% 6 798 257 2 836 182

LFS2001b 28 084 327 15 817 377 18 807 980 11 167 541 4 649 836 7 640 439 2 990 603 56.3% 67.0% 29.4% 40.6% 7 019 158 1 964 763

LFS2002a 28 298 255 16 494 331 19 535 489 11 603 398 4 890 933 7 932 091 3 041 158 58.3% 69.0% 29.7% 40.6% 7 089 163 1 821 426

LFS2002b 28 495 088 16 214 594 19 404 685 11 283 924 4 930 670 8 120 761 3 190 091 56.9% 68.1% 30.4% 41.8% 7 173 080 1 778 542

LFS2003a 28 724 521 16 409 029 19 642 235 11 297 621 5 111 408 8 344 614 3 233 206 57.1% 68.4% 31.1% 42.5% 7 223 138 1 827 711

LFS2003b 28 906 230 15 840 687 19 609 716 11 411 351 4 429 336 8 198 365 3 769 029 54.8% 67.8% 28.0% 41.8% 7 364 616 1 901 131

LFS2004a 29 099 787 15 787 749 19 549 788 11 378 217 4 409 532 8 171 571 3 762 039 54.3% 67.2% 27.9% 41.8% 7 473 638 1 764 630

LFS2004b 29 270 821 15 761 080 19 704 344 11 630 196 4 130 884 8 074 148 3 943 264 53.8% 67.3% 26.2% 41.0% 7 684 843 1 944 236

LFS2005a 29 489 763 16 172 520 19 991 966 11 894 320 4 278 200 8 097 646 3 819 446 54.8% 67.8% 26.5% 40.5% 7 741 991 2 068 479

LFS2005b 29 663 379 16 770 161 20 078 497 12 287 798 4 482 363 7 790 699 3 308 336 56.5% 67.7% 26.7% 38.8% 7 979 587 2 459 690

LFS2006a 29 817 824 16 707 953 20 386 846 12 437 963 4 269 990 7 948 883 3 678 893 56.0% 68.4% 25.6% 39.0% 8 051 532 2 187 940

LFS2006b 29 972 571 17 173 402 20 386 338 12 787 285 4 386 117 7 599 053 3 212 936 57.3% 68.0% 25.5% 37.3% 8 376 441 2 376 338

LFS2007a 30 160 997 16 965 854 20 464 900 12 634 896 4 330 958 7 830 004 3 499 046 56.3% 67.9% 25.5% 38.3% 8 414 719 2 129 164

LFS2007b 30 387 402 17 194 198 20 632 876 13 293 327 3 900 871 7 339 549 3 438 678 56.6% 67.9% 22.7% 35.6% 9 034 135 2 083 855

QLFS2008Q1 31 700 031 18 819 077 20 021 290 14 450 646 4 368 431 5 570 644 1 202 213 59.4% 63.2% 23.2% 27.8% 9 935 320 2 438 758

QLFS2008Q2 31 859 272 18 871 451 19 971 536 14 604 053 4 267 398 5 367 483 1 100 085 59.2% 62.7% 22.6% 26.9% 10 073 656 2 451 627

QLFS2008Q3 31 987 108 18 859 224 19 951 529 14 561 398 4 297 826 5 390 131 1 092 305 59.0% 62.4% 22.8% 27.0% 10 121 062 2 281 786

QLFS2008Q4 32 141 290 18 830 810 20 019 823 14 784 916 4 045 894 5 234 907 1 189 013 58.6% 62.3% 21.5% 26.1% 10 232 739 2 369 114

QLFS2009Q1 32 293 255 18 994 790 20 227 098 14 631 692 4 363 098 5 595 406 1 232 308 58.8% 62.6% 23.0% 27.7% 10 168 230 2 291 205

QLFS2009Q2 32 452 436 18 713 390 20 248 100 14 374 908 4 338 482 5 873 192 1 534 710 57.7% 62.4% 23.2% 29.0% 10 087 548 2 247 096

QLFS2009Q3 32 590 099 18 315 304 19 960 259 13 841 980 4 473 324 6 118 279 1 644 955 56.2% 61.2% 24.4% 30.7% 9 794 028 2 110 248

QLFS2009Q4 32 733 898 18 408 899 20 133 886 13 982 850 4 426 049 6 151 036 1 724 987 56.2% 61.5% 24.0% 30.6% 9 853 327 2 250 218

QLFS2010Q1 32 917 597 18 430 474 20 319 214 13 820 568 4 609 906 6 498 646 1 888 740 56.0% 61.7% 25.0% 32.0% 9 711 038 2 153 671

QLFS2010Q2 33 062 618 18 452 797 20 411 247 13 834 144 4 618 653 6 577 103 1 958 450 55.8% 61.7% 25.0% 32.2% 9 624 375 2 301 425

QLFS2010Q3 33 246 836 18 321 525 20 399 505 13 668 819 4 652 706 6 730 686 2 077 980 55.1% 61.4% 25.4% 33.0% 9 496 044 2 281 915

QLFS2010Q4 33 365 615 18 281 358 20 456 705 13 915 884 4 365 474 6 540 821 2 175 347 54.8% 61.3% 23.9% 32.0% 9 729 983 2 322 247

QLFS2011Q1 33 508 825 18 512 827 20 754 386 13 917 447 4 595 380 6 836 939 2 241 559 55.2% 61.9% 24.8% 32.9% 9 792 343 2 281 844

QLFS2011Q2 33 672 970 18 712 021 20 924 463 13 933 454 4 778 567 6 991 009 2 212 442 55.6% 62.1% 25.5% 33.4% 9 779 940 2 308 797

QLFS2011Q3 33 818 983 18 827 682 21 038 831 14 131 609 4 696 073 6 907 222 2 211 149 55.7% 62.2% 24.9% 32.8% 10 006 894 2 266 346

QLFS2011Q4 33 974 114 18 813 971 21 152 923 14 349 931 4 464 040 6 802 992 2 338 952 55.4% 62.3% 23.7% 32.2% 10 220 128 2 234 273

QLFS2012Q1 34 128 626 19 063 136 21 441 264 14 297 605 4 765 531 7 143 659 2 378 128 55.9% 62.8% 25.0% 33.3% 10 126 825 2 216 096

QLFS2012Q2 34 301 187 19 065 829 21 426 195 14 348 370 4 717 459 7 077 825 2 360 366 55.6% 62.5% 24.7% 33.0% 10 202 637 2 211 243

QLFS2012Q3 34 456 238 19 481 358 21 695 856 14 583 192 4 898 166 7 112 664 2 214 498 56.5% 63.0% 25.1% 32.8% 10 319 781 2 333 346

QLFS2012Q4 34 616 851 19 249 776 21 550 431 14 541 707 4 708 069 7 008 724 2 300 655 55.6% 62.3% 24.5% 32.5% 10 277 582 2 352 891

QFLS2013Q1 34 756 987 19 430 812 21 830 242 14 569 906 4 860 906 7 260 336 2 399 430 55.9% 62.8% 25.0% 33.3% 10 246 079 2 337 302

QLFS2013Q2 34 908 625 19 677 242 22 100 329 14 706 731 4 970 511 7 393 598 2 423 087 56.4% 63.3% 25.3% 33.5% 10 383 033 2 364 705

QLFS2013Q3 35 077 845 19 939 574 22 234 844 15 061 904 4 877 670 7 172 940 2 295 270 56.8% 63.4% 24.5% 32.3% 10 722 499 2 332 517

QLFS2013Q4 35 231 309 20 022 751 22 221 218 15 195 491 4 827 260 7 025 727 2 198 467 56.8% 63.1% 24.1% 31.6% 10 780 187 2 455 277

36

Table A.2: Labour market aggregates under the narrow definition in QLFS 2008Q1-2013Q3, using the Census 2001 weights and 2011 weights

Using Census 2001 weights Using Census 2011 weights

Labour force Employed Unemployed Labour force Employed Unemployed

LFS2007b (using Census 2001 weights) 17 194 198 13 293 327 3 900 871 17 194 198 13 293 327 3 900 871

QLFS2008Q1 17 826 085 13 636 995 4 189 090 18 819 077 14 450 646 4 368 431

QLFS2008Q2 17 863 803 13 749 288 4 114 515 18 871 451 14 604 053 4 267 398

QLFS2008Q3 17 788 621 13 668 530 4 120 091 18 859 224 14 561 398 4 297 826

QLFS2008Q4 17 732 760 13 861 822 3 870 938 18 830 810 14 784 916 4 045 894

QLFS2009Q1 17 833 149 13 652 530 4 180 619 18 994 790 14 631 692 4 363 098

QLFS2009Q2 17 510 797 13 388 133 4 122 664 18 713 390 14 374 908 4 338 482

QLFS2009Q3 17 086 407 12 896 820 4 189 587 18 315 304 13 841 980 4 473 324

QLFS2009Q4 17 146 008 12 983 951 4 162 057 18 408 899 13 982 850 4 426 049

QLFS2010Q1 17 133 544 12 825 578 4 307 966 18 430 474 13 820 568 4 609 906

QLFS2010Q2 17 075 327 12 766 534 4 308 793 18 452 797 13 834 144 4 618 653

QLFS2010Q3 17 392 161 12 998 660 4 393 501 18 321 525 13 668 819 4 652 706

QLFS2010Q4 17 285 752 13 151 352 4 134 400 18 281 358 13 915 884 4 365 474

QLFS2011Q1 17 497 435 13 135 337 4 362 098 18 512 827 13 917 447 4 595 380

QLFS2011Q2 17 673 452 13 138 658 4 534 794 18 712 021 13 933 454 4 778 567

QLFS2011Q3 17 772 580 13 333 561 4 439 019 18 827 682 14 131 609 4 696 073

QLFS2011Q4 17 755 077 13 514 108 4 240 969 18 813 971 14 349 931 4 464 040

QLFS2012Q1 17 959 039 13 436 149 4 522 890 19 063 136 14 297 605 4 765 531

QLFS2012Q2 17 933 020 13 466 866 4 466 154 19 065 829 14 348 370 4 717 459

QLFS2012Q3 18 332 020 13 667 835 4 664 185 19 481 358 14 583 192 4 898 166

QLFS2012Q4 18 097 084 13 597 355 4 499 729 19 249 776 14 541 707 4 708 069

QFLS2013Q1 18 234 352 13 634 013 4 600 339 19 430 812 14 569 906 4 860 906

QLFS2013Q2 18 460 637 13 738 998 4 721 639 19 677 242 14 706 731 4 970 511

QLFS2013Q3 18 664 637 14 057 925 4 606 712 19 939 574 15 061 904 4 877 670

37

Table A.3: Formal and informal sector employment, and number of employees and self-employed, 1995-2013

Survey

Employment status Employment by sector

Employees Self-employed Unspecified Domestic workers

Informal (non-agricultural)

Formal (non-agricultural)

Informal (agricultural)

Formal (agricultural)

Don't know

Not specified

Formal employment as % total employment

OHS1995 8 123 412 1 375 935 0 Not available

OHS1996 8 313 240 611 045 42 022

OHS1997 8 167 479 926 168 0 828 254 1 043 347 6 436 017 187 486 525 618 0 72 925 76.6%

OHS1998 8 339 925 1 025 748 4 457 747 281 1 077 141 6 508 097 202 082 725 474 0 110 055 77.2%

OHS1999 8 844 574 1 505 706 5 863 812 465 1 571 646 6 796 008 284 336 798 905 0 92 783 73.3%

LFS2000a 8 787 145 3 073 630 13 634 1 002 719 1 819 556 6 672 951 1 507 625 756 510 86 472 28 576 62.6%

LFS2000b 9 370 733 2 825 474 28 199 941 463 2 026 065 7 077 307 1 074 413 766 917 108 318 229 923 64.2%

LFS2001a 9 024 720 3 218 407 17 080 844 135 2 836 182 6 798 257 742 404 784 712 214 235 40 282 61.9%

LFS2001b 9 011 975 2 144 102 11 464 881 168 1 964 763 7 019 158 382 241 764 521 127 023 28 667 69.7%

LFS2002a 9 081 627 2 508 940 12 831 875 172 1 821 426 7 089 163 862 747 864 576 74 868 15 446 68.5%

LFS2002b 9 081 716 2 190 994 11 214 843 019 1 778 542 7 173 080 550 068 851 897 61 643 25 675 71.1%

LFS2003a 9 194 238 2 099 251 4 132 885 322 1 827 711 7 223 138 443 426 841 440 57 332 19 252 71.4%

LFS2003b 9 276 158 2 131 304 3 889 894 626 1 901 131 7 364 616 365 378 831 526 36 403 17 671 71.8%

LFS2004a 9 356 332 2 018 613 3 272 845 965 1 764 630 7 473 638 340 515 912 831 25 704 14 934 73.7%

LFS2004b 9 414 391 2 206 814 8 991 880 067 1 944 236 7 684 843 425 083 624 358 52 970 18 639 71.4%

LFS2005a 9 535 624 2 340 253 18 443 848 914 2 068 479 7 741 991 513 022 647 448 27 756 46 710 70.5%

LFS2005b 9 846 100 2 422 542 19 156 858 199 2 459 690 7 979 587 337 884 578 059 33 783 40 596 69.6%

LFS2006a 9 771 856 2 658 832 7 275 849 085 2 187 940 8 051 532 702 881 605 795 14 098 26 632 69.6%

LFS2006b 10 184 406 2 592 531 10 348 884 898 2 376 338 8 376 441 472 697 605 129 46 935 24 847 70.2%

LFS2007a 10 253 063 2 365 182 16 651 935 642 2 129 164 8 414 719 459 509 602 942 52 537 40 383 71.4%

LFS2007b 10 936 220 2 322 623 34 484 1 024 039 2 083 855 9 034 135 368 256 666 533 47 251 69 258 73.0%

QLFS2008Q1 12 214 626 2 236 020 0 1 234 278 2 438 758 9 935 320 165 943 676 347 0 0 73.4%

QLFS2008Q2 12 320 340 2 283 713 0 1 257 121 2 451 627 10 073 656 124 201 697 448 0 0 73.8%

QLFS2008Q3 12 287 147 2 274 251 0 1 348 845 2 281 786 10 121 062 116 232 693 473 0 0 74.3%

QLFS2008Q4 12 462 637 2 322 279 0 1 376 368 2 369 114 10 232 739 128 083 678 612 0 0 73.8%

QLFS2009Q1 12 322 116 2 309 576 0 1 393 306 2 291 205 10 168 230 132 558 646 393 0 0 73.9%

QLFS2009Q2 12 142 870 2 232 038 0 1 287 201 2 247 096 10 087 548 99 732 653 331 0 0 74.7%

QLFS2009Q3 11 840 232 2 001 748 0 1 255 181 2 110 248 9 794 028 77 876 604 647 0 0 75.1%

QLFS2009Q4 11 837 762 2 145 088 0 1 232 144 2 250 218 9 853 327 108 987 538 174 0 0 74.3%

QLFS2010Q1 11 737 057 2 083 511 0 1 272 107 2 153 671 9 711 038 91 303 592 449 0 0 74.6%

QLFS2010Q2 11 688 146 2 145 998 0 1 253 274 2 301 425 9 624 375 88 978 566 092 0 0 73.7%

QLFS2010Q3 11 509 780 2 159 039 0 1 216 336 2 281 915 9 496 044 88 634 585 890 0 0 73.8%

QLFS2010Q4 11 698 410 2 217 474 0 1 214 133 2 322 247 9 729 983 93 919 555 602 0 0 73.9%

QLFS2011Q1 11 730 987 2 186 460 0 1 215 173 2 281 844 9 792 343 98 535 529 552 0 0 74.2%

QLFS2011Q2 11 743 862 2 189 592 0 1 218 949 2 308 797 9 779 940 87 590 538 178 0 0 74.1%

QLFS2011Q3 11 959 646 2 171 963 0 1 205 165 2 266 346 10 006 894 84 513 568 691 0 0 74.8%

QLFS2011Q4 12 193 700 2 156 231 0 1 224 363 2 234 273 10 220 128 86 640 584 527 0 0 75.3%

QLFS2012Q1 12 158 459 2 139 146 0 1 260 774 2 216 096 10 126 825 89 354 604 556 0 0 75.1%

QLFS2012Q2 12 181 817 2 166 553 0 1 258 693 2 211 243 10 202 637 89 690 586 107 0 0 75.2%

QLFS2012Q3 12 328 466 2 254 726 0 1 230 036 2 333 346 10 319 781 86 844 613 185 0 0 75.0%

QLFS2012Q4 12 321 392 2 220 315 0 1 191 798 2 352 891 10 277 582 84 709 634 727 0 0 75.0%

QFLS2013Q1 12 331 144 2 238 762 0 1 221 452 2 337 302 10 246 079 86 996 678 077 0 0 75.0%

QLFS2013Q2 12 494 517 2 212 214 0 1 216 960 2 364 705 10 383 033 95 614 646 419 0 0 75.0%

QLFS2013Q3 13 003 373 2 058 531 0 1 265 543 2 332 517 10 722 499 112 218 629 127 0 0 75.4%

QLFS2013Q4 13 047 005 2 148 486 0 1 245 176 2 455 277 10 780 187 100 699 614 152 0 0 75.0%

38