PROJECTING THE LABOUR SUPPLY TO 2024

PROJECTING THE LABOUR SUPPLY TO 2024 Charles Simkins Helen Suzman Professor of Political Economy School of Economic and Business Sciences University of the Witwatersrand May 2008 centre for poverty employment and growth innovative employment strategies

centre for poverty employment and growth HSRC Human Sciences Research Council May 2008 Acknowledgements We gratefully acknowledge the financial assistance of the Department of Trade and Industry Produced by: Prof Charles Simkins Contact: Dr Miriam Altman Executive Director: CPEG, HSRC E-mail: maltman@hsrc.ac.za Tel: +27 12 302 2402 2

Projecting the labour supply to 2024 Contents Tables... 3 Phase I report... 5 1. Terms of reference... 5 2. Population projections... 5 3. An analysis of labour force participation rates... 13 4. Labour supply projections... 16 Phase II report... 19 1. Terms of reference... 19 2. Household classification... 19 3. Determinants of labour force participation, September 2005... 21 4. Projecting the educational distribution to 2024... 27 5. Projecting the metro population to 2024... 28 6. The labour force projection: methodology and results... 29 Tables Phase I report Table 1 Population projection: 2007, 2014 and 2024... 7 Table 2 Labour force participation rates by population group, gender and age group 15-64, September 2005... 14 Table 3 Labour market statistics, 2001 to 2007... 15 Table 4 Aggregate labour supply, 2007-2024... 17 Phase II report Table 1 Estimates of individuals aged 15-64 in each household category, September 2005 Labour Force Survey... 20 Table 2A Odds ratios from logistic regression of labour force participation, September 2005... 22 Panel A Official definition... 22 Table 2B Odds ratios from logistic regression of labour force participation, September 2005... 24 Panel B Expanded definition... 24 Table 3 Labour force participation predicted by knowledge of age group... 25 Table 4 Percentage distribution of the 15-64 age group across educational categories... 28 Table 5 Projecting the metro population to 2024... 28 Table 6 Aggregate labour supply, 2007-2014... 30 Table 7 Population size and composition, September 2005 Labour Force Survey and Community Survey... 30 3

centre for poverty employment and growth HSRC Table 8 Aggregate employment and unemployment rate, 2007-2014... 31 Table 9 Distribution of individuals 15-64 by class... 31 4

Projecting the labour supply to 2024 Phase I report 1. Terms of reference 1. A population projection has been carried out from 2007 to 2024, using the SPECTRUM model by population group, gender and age group, based on: The population composition reported by the Community Survey 2007; Fertility as estimated by the Community Survey 2007 and projected to drop for Africans and Coloureds; Mortality as determined by background mortality and the effect of AIDS; and International migration, with some discussion of alternative projections. The projection has yielded estimates of the population by population group, gender and age group for each year between 2007 and 2024; this information is available in electronic format. Particular attention has been paid to projections for 2014 and 2024. 2. A tabulation of labour force participation rates (LFPRs) by population group, age and gender in the September Labour Force Surveys from 2000 to 2006 and projections on stated assumptions to 2024 have been carried out. Particular attention has been paid to projections for 2014 and 2024. Both broad and narrow labour force definitions were considered. 3. An estimate of labour supply to 2024 is given, obtained by multiplying LFPRs by population. Particular attention has been paid to projections for 2014 and 2024. 2. Population projections The population projections are built up using the SPECTRUM model with its AIM extension to model the spread of the AIDS epidemic. This model is used by Statistics South Africa in preparing its demographic forecasts. Population projections are built up from the following components: A start date for the projection and an end date, set at 2007 and 2024 respectively. 5

centre for poverty employment and growth HSRC An initial distribution of the population by population group, gender and fiveyear age group. This distribution is based on the 2001 Census and the 2007 Community Survey results 1 and is an extrapolation of the data to mid-2007. Fertility data. The Community Survey estimates the African Total Fertility Rate at 2.7, while the Coloured, Asian and White rates are estimated at 2.3, 1.4 and 1.4 respectively for 2007. Since the Asian and White rates are already low, they are projected to remain constant between 2007 and 2024. African and Coloured rates are projected to drop by 0.02 per year, so that they reach 2.36 and 1.96 respectively by 2024. Mortality data. The SPECTRUM/AIM model requires both the projection of background life expectancy (what life expectancy would be in the absence of AIDS) and parameters representing the spread of the AIDS epidemic. These data are based on the Actuarial Society s ASSA 2003 model. Two immigration scenarios are considered: a low projection in which net immigration rises linearly from 80,000 per year in 2007 to 114,000 per year in 2024, and a high projection based on double the stream in the low projection. All net immigration is assumed to be African. The implications of these assumptions are that, in the case of the low projection, the increase in the 15-64 age group accounted for by immigration will be 17.5% of the total increase in that group for the period 2007 to 2014. For the period 2014 to 2024, the corresponding proportion will be 29.8%. In the case of the high projection, the increase in the 15-64 age group accounted for by immigration will be 31.4% between 2007 and 2014 and 47.8% between 2014 and 2024. The Microsoft Excel file proj.xls, which accompanies this report, sets out both assumptions and results for each year between 2007 and 2024. Table 1 extracts information for 2007, 2014 and 2024 from this file. The main features of table 1 are: An estimated population of 48.7-million in 2007, rising to 54.7-million in 2024 under the low immigration assumption and to 56.5-million under the high immigration assumption. The population growth rate is 0.7% per annum over the whole period under the low immigration assumption and 0.9% per annum under the high immigration assumption. The proportion of the population in the 15-64 age group rises from 63.6% in 2007 to 67.7% in 2024 under the low immigration assumption and to 68.0% under the high immigration assumption. The population in the economically active age range will grow more rapidly than the population as a whole. The 15-64 growth rate will be 1.1% per annum in the low immigration case and 1.3% in the high immigration case. 1 The data are taken from Statistics South Africa, Community Survey 2007, Statistical Release P0301, 24 October 2007 (revised version). 6

Projecting the labour supply to 2024 Table 1 Population projection: 2007, 2014 and 2024 Africans (low immigration) 2007 2014 2024 Fertility Input TFR (total fertility rate) 2.70 2.56 2.36 Calculated TFR 2.70 2.56 2.36 GRR (gross reproduction rate) 1.34 1.27 1.17 NRR (net reproduction rate) 1.13 1.07 0.99 Mean age of childbearing 26.9 26.9 26.8 Child-woman ratio 0.41 0.37 0.35 Fertility table: UN sub-saharan Africa Mortality Male LE (life expectancy) 47.9 48.8 45.8 Female LE 52.8 52.0 43.6 Total LE 50.4 50.5 44.7 IMR (infant mortality rate) 81.9 86.3 85.5 U5MR (under 5 mortality rate) 125.7 121.7 120.1 Life table: UN East Asia Immigration (thousands) Male immigration 50 57 67 Female immigration 30 37 47 Total immigration 80 94 114 Vital rates CBR per 1,000 (crude birth rate) 24.3 23.8 21.2 CDR per 1,000 (crude death rate) 15.2 15.4 20.2 RNI percent (rate of natural increase) 0.92 0.84 0.10 GR percent 1.13 1.07 0.36 Doubling time 61.9 65.3 193.2 Annual births and deaths (thousands) Births 935.43 987.19 946.25 Deaths 582.69 638.7 900.06 Population (millions) Total population 38.44 41.49 44.56 Male population 18.51 20.17 22.18 Female population 19.92 21.31 22.38 Percent 0-4 11.16 10.59 9.72 Percent 5-14 22.09 20.06 19.02 Percent 15-49 53.75 56.19 57.13 Percent 15-64 62.09 65.36 67.21 Percent 65 and over 4.66 3.99 4.04 Percent females 15-49 52.86 54.98 55.24 Sex ratio 92.9 94.7 99.1 Dependency ratio 0.61 0.53 0.49 Median age 23 25 27 7

centre for poverty employment and growth HSRC Africans (high immigration) 2007 2014 2024 Fertility Input TFR 2.70 2.56 2.36 Calculated TFR 2.70 2.56 2.36 GRR 1.34 1.27 1.17 NRR 1.13 1.07 0.99 Mean age of childbearing 26.9 26.9 26.8 Child-woman ratio 0.41 0.37 0.35 Fertility table: UN sub-saharan Africa Mortality Male LE 47.9 48.8 45.8 Female LE 52.8 52.0 43.6 Total LE 50.4 50.5 44.7 IMR 81.9 86.3 85.5 U5MR 125.7 121.7 120.1 Life table: UN East Asia Immigration (thousands) Male immigration 100 114 134 Female immigration 60 74 94 Total immigration 160 188 228 Vital rates CBR per 1000 24.3 23.9 21.3 CDR per 1000 15.2 15.3 20.0 RNI percent 0.92 0.86 0.13 GR percent 1.33 1.31 0.62 Doubling time 52.3 53.3 113.2 Annual births and deaths (thousands) Births 940 1010 990 Deaths 580 650 930 Population (millions) Total population 38.44 42.15 46.42 Male population 18.51 20.58 23.26 Female population 19.92 21.58 23.16 Percent 0-4 11.16 10.59 9.72 Percent 5-14 22.09 19.79 18.72 Percent 15-49 53.75 56.60 57.80 Percent 15-64 62.09 65.69 67.67 Percent 65 and over 4.66 3.93 3.90 Percent females 15-49 52.86 55.28 55.73 Sex ratio 92.9 95.4 100.5 Dependency ratio 0.61 0.52 0.48 Median age 23 25 27 8

Projecting the labour supply to 2024 Coloureds 2007 2014 2024 Fertility Input TFR 2.30 2.16 1.96 Calculated TFR 2.30 2.16 1.96 GRR 1.12 1.05 0.96 NRR 0.98 0.92 0.83 Mean age of childbearing 27.5 27.4 27.2 Child-woman ratio 0.33 0.31 0.29 Fertility table: average Mortality Male LE 53.7 54.0 51.8 Female LE 58.5 58.1 53.2 Total LE 56.2 56.1 52.5 IMR 66.1 66.5 66.7 U5MR 93.5 90.8 91.2 Life table: UN East Asia Immigration Male immigration 0 0 0 Female immigration 0 0 0 Total immigration 0 0 0 Vital rates CBR per 1000 19.7 19.0 16.5 CDR per 1000 12.7 13.6 17.5 RNI percent 0.7 0.5-0.1 GR percent 0.7 0.5-0.1 Doubling time 99.0 130.3 0.0 Annual births and deaths (thousands) Births 86.60 87.10 77.28 Deaths 55.69 62.61 81.94 Population (millions) Total population 4.40 4.59 4.69 Male population 2.13 2.23 2.30 Female population 2.27 2.36 2.40 Percent 0-4 9.48 8.75 7.84 Percent 5-14 19.18 17.58 16.52 Percent 15-49 56.15 56.16 54.96 Percent 15-64 66.84 68.86 69.58 Percent 65 and over 4.51 4.82 6.06 Percent females 15-49 55.23 55.06 53.57 Sex ratio 93.85 94.39 95.82 Dependency ratio 0.50 0.45 0.44 Median age 26 28 30 9

centre for poverty employment and growth HSRC Asians 2007 2014 2024 Fertility Input TFR 1.40 1.40 1.40 Calculated TFR 1.40 1.40 1.40 GRR 0.68 0.68 1.40 NRR 0.65 0.65 1.33 Mean age of childbearing 27.2 27.2 27.2 Child-woman ratio 0.22 0.21 0.19 Fertility table: average Mortality Male LE 63.9 64.1 65.1 Female LE 69.3 69.0 65.1 Total LE 66.6 66.6 65.1 IMR 32.9 33.2 32.1 U5MR 42.8 41.5 39.8 Life table: UN East Asia Immigration Male immigration 0 0 0 Female immigration 0 0 0 Total immigration 0 0 0 Vital rates CBR per 1000 12.4 12.0 10.3 CDR per 1000 8.5 10.3 14.5 RNI percent 0.39 0.17-0.41 GR percent 0.39 0.17-0.41 Doubling time 178.6 404.9 0 Annual births and deaths (thousands) Births 15.56 15.35 13.02 Deaths 10.69 13.17 18.23 Population (millions) Total population 1.25 1.28 1.26 Male population 0.62 0.63 0.55 Female population 0.63 0.65 0.71 Percent 0-4 6.37 5.88 5.17 Percent 5-14 15.46 12.98 11.49 Percent 15-49 58.30 57.31 54.30 Percent 15-64 72.44 73.77 72.97 Percent 65 and over 5.73 7.36 10.37 Percent females 15-49 57.64 55.67 47.26 Sex ratio 98.12 97.43 77.94 Dependency ratio 0.38 0.36 0.37 Median age 30 33 38 10

Projecting the labour supply to 2024 Whites 2007 2014 2024 Fertility Input TFR 1.40 1.40 1.40 Calculated TFR 1.40 1.40 1.40 GRR 0.68 0.68 0.68 NRR 0.66 0.66 0.66 Mean age of childbearing 27.2 27.2 27.2 Child-woman ratio 0.19 0.19 0.20 Fertility table: average Mortality Male LE 67.3 67.2 65.5 Female LE 72.6 72.3 69.2 Total LE 70.0 69.8 67.4 IMR 24.6 25.1 25.5 U5MR 30.4 30.5 31.1 Life table: UN East Asia Immigration Male immigration 0 0 0 Female immigration 0 0 0 Total immigration 0 0 0 Vital rates CBR per 1000 9.7 9.9 8.8 CDR per 1000 13.0 14.3 17.6 RNI percent -0.32-0.45-0.88 GR percent -0.32-0.45-0.88 Doubling time 0 0 0 Annual births and deaths (thousands) Births 45.28 44.62 37.15 Deaths 60.33 64.86 74.60 Population (millions) Total population 4.65 4.52 4.23 Male population 2.27 2.20 2.05 Female population 2.38 2.32 2.19 Percent 0-4 4.97 4.82 4.49 Percent 5-14 11.73 10.06 10.07 Percent 15-49 51.70 50.32 46.34 Percent 15-64 70.81 70.96 68.59 Percent 65 and over 12.49 14.15 16.85 Percent females 15-49 50.89 49.07 44.30 Sex ratio 95.68 94.57 93.38 Dependency ratio 0.41 0.41 0.46 Median age 38 40 42 11

centre for poverty employment and growth HSRC 2007 2014 2024 All (low immigration) Immigration Male immigration 50.00 57.00 67.00 Female immigration 30.00 37.00 47.00 Total immigration 80.00 94.00 114.00 Vital rates CBR per 1000 22.2 21.9 19.6 CDR per 1000 14.6 15.0 19.6 RNI percent 0.77 0.68 0.00 Annual births and deaths (thousands) Births 1082.87 1134.26 1073.70 Deaths 709.40 779.34 1074.83 Population (millions) Total population 48.74 51.88 54.74 Male population 23.53 25.23 27.08 Female population 25.20 26.64 27.68 Percent 0-4 10.29 9.81 9.05 Percent 5-14 20.67 18.79 17.94 Percent 15-49 53.89 55.70 56.05 Percent 15-64 63.62 66.37 67.65 Percent 65 and over 5.42 5.03 5.35 Sex ratio 93.4 94.7 97.8 Population growth rate (%) 0.86 0.18 15-64 growth rate (%) 1.46 0.51 All (high immigration) Immigration Male immigration 100.00 114.00 134.00 Female immigration 60.00 74.00 94.00 Total immigration 160.00 188.00 228.00 Vital rates CBR per 1000 3.0 2.8 2.3 CDR per 1000 2.6 2.7 3.1 RNI percent 0.04 0.01-0.08 Annual births and deaths (thousands) Births 148.38 148.08 128.44 Deaths 127.29 141.29 175.70 Population (millions) Total population 48.74 52.54 56.60 Male population 23.53 25.64 28.16 Female population 25.20 26.91 28.46 Percent 0-4 10.29 9.82 9.07 Percent 5-14 20.67 18.59 17.73 Percent 15-49 53.89 56.04 56.63 Percent 15-64 63.62 66.62 68.02 Percent 65 and over 5.42 4.97 5.19 Sex ratio 93.4 95.3 98.9 Population growth rate (%) 1.04 0.39 15-64 growth rate (%) 1.57 0.47 12

Projecting the labour supply to 2024 3. An analysis of labour force participation rates Labour force participation rates are defined in South African official statistics as the employed plus the unemployed on the official measure, expressed as a percentage of the population between the ages of 15 and 64. LFPRs can be found for subgroups, with subgroups defined by population group, gender and age group often considered. Those people who are unemployed according to the expanded definition of unemployed but not according to the official definition are regarded as discouraged workers. Discouraged workers are not economically active according to the official definition, but they are according to the expanded definition. Both definitions are used in this study. It should be noted at the outset that there will be a change in the definition of employed persons in the new Quarterly Labour Force Survey, due to report for the first time on 28 August 2008. The main difference in the new definition is that a person must have worked at least an hour in market activities (done work for pay, helped unpaid in a household business or run his/her own business), whereas in the old definition a range of non-market activities, where the product was entirely consumed within the household, counted as employment as well. Additionally, there is a change in one of the criteria for unemployment. In the old definition, an unemployed person had to be available for and start work within two weeks of the interview. In the new definition, an unemployed person must have been available for work in the week before the interview. These changes should have a downward impact on the estimate of employment, but the impact on unemployment is not known. They will therefore have an unknown (but probably downward) impact on LFPRs. All analysis at this stage has to be based on the old Labour Force Survey and its definitions, but it must be appreciated that projections based on the old definitions may soon be regarded as obsolete. LFPRs by population, age group and gender, based on the September 2005 Labour Force Survey, are presented in table 2. 63.9% of all men between 15 and 64 are economically active on the official definition, compared to 50.3% of all women. On the expanded definition, the male rate rises to 72.2% and the female rate to 64.4%. The differences reflect lower LFPRs for women in all age groups. 13

centre for poverty employment and growth HSRC Table 2 Labour force participation rates by population group, gender and age group 15-64, September 2005 Official definition Male African Coloured Asian White All 15-19 10.6 30.0 22.8 12.4 12.5 20-24 52.8 75.4 67.5 73.1 56.1 25-29 77.9 87.0 88.6 93.2 80.0 30-34 85.1 90.2 96.3 91.5 86.6 35-39 86.2 87.8 93.3 94.3 87.7 40-44 81.9 89.0 89.9 91.3 84.2 45-49 78.7 84.5 87.5 94.4 82.2 50-54 71.8 79.3 78.1 86.1 75.4 55-59 64.9 63.2 74.5 76.9 67.2 60-64 44.6 22.3 74.9 43.5 43.6 All ages 60.6 72.8 77.1 77.9 63.9 Female African Coloured Asian White All 15-19 8.2 24.3 20.3 8.9 9.7 20-24 44.3 68.4 57.0 52.3 47.1 25-29 61.8 70.6 65.5 78.9 63.9 30-34 67.7 75.3 73.2 79.5 69.9 35-39 65.9 69.5 59.8 72.8 66.8 40-44 63.5 61.5 59.2 80.5 65.2 45-49 65.4 64.8 39.4 74.2 65.8 50-54 53.6 42.2 35.3 67.1 53.6 55-59 44.0 30.1 16.3 45.4 42.2 60-64 18.9 4.0 10.4 26.0 18.9 All ages 48.3 56.0 49.0 60.3 50.3 Expanded definition Male African Coloured Asian White All 15-19 15.8 35.5 27.4 13.2 17.4 20-24 69.4 84.1 72.5 77.4 71.1 25-29 91.7 96.7 91.0 93.6 92.2 30-34 95.3 95.0 96.3 93.9 95.1 35-39 94.1 90.7 64.5 87.2 94.2 40-44 90.0 90.8 91.6 91.6 90.3 45-49 85.7 87.6 89.8 96.1 87.7 50-54 79.5 82.4 80.8 86.8 81.2 55-59 70.4 67.1 74.5 78.8 71.7 60-64 48.9 25.3 74.9 43.5 46.7 All ages 70.3 78.0 79.4 79.5 72.2 14

Projecting the labour supply to 2024 Expanded definition Female African Coloured Asian White All 15-19 14.8 37.1 25.9 9.6 16.5 20-24 69.0 83.7 67.5 59.3 69.5 25-29 56.0 85.7 77.4 83.7 85.6 30-34 88.6 86.7 78.0 81.7 87.3 35-39 86.3 80.9 67.8 77.6 84.3 40-44 78.6 70.9 65.3 81.7 77.7 45-49 77.7 71.3 47.5 76.3 76.0 50-54 62.4 47.1 38.7 69.3 61.0 55-59 49.7 36.8 16.3 45.6 46.8 60-64 19.7 4.2 10.4 26.0 19.5 All ages 64.6 66.8 55.5 62.9 64.4 Table 2 should be interpreted in the light of table 3, which sets out labour market statistics for every September Labour Force Survey since 2001. Table 3 Labour market statistics, 2001 to 2007 September Thousands 2001 2002 2003 2004 2005 2006 Population 15-65 28,117 28,527 28,938 29,305 29,697 30,006 Employment 15-65 11,181 11,296 11,424 11,643 12,301 12,800 Unemployed 15-65 4,655 4,936 4,434 4,135 4,487 4,391 Labour force (narrow) 15,836 16,232 15,858 15,778 16,788 17,191 Unemployment rate (%) (narrow) 29.4 30.4 28.0 26.2 26.7 25.5 Participation rate (%) (narrow) 56.3 56.9 54.8 53.8 56.5 57.3 Labour absorption rate (%) 43.7 41.0 39.3 39.1 40.3 41.7 Discouraged workseekers 2,994 3,194 3,773 3,948 3,312 3,217 Labour force (expanded) 18,830 19,426 19,631 19,726 20,100 20,408 Unemployment rate (%) (expanded) 40.6 41.9 41.8 41.0 38.8 37.3 Participation rate (%) (expanded) 67.0 68.1 67.8 67.3 67.7 68.0 Tables 2 and 3 together show that: 1. The narrowly defined labour force grew from 15.8-million in 2001 to 17.2-million in 2006 and the expanded labour force grew from 18.8-million to 20.4-million, reflecting average annual growth rates of 1.7% and 1.6% respectively. 2. There is no clear trend in either the narrow or the expanded LFPR. 3. The narrow male LFPR for African men is considerably lower than for the three other population groups. Coloured male participation is intermediate, with Asian and White male rates similar. The gap narrows between African men and other men when expanded LFPRs are considered. There is little difference between narrow and expanded LFPRs for Asian and White men. 15

centre for poverty employment and growth HSRC 4. The pattern for women is different. The gap between the narrow and the expanded rate for Africans and Coloureds is smaller than for men. The Asian narrow and expanded participation rates are below the Coloured rates, reflecting a lower propensity to participate in the labour market after marriage. White narrow participation is slightly higher than Coloured participation, but the reverse is true for broad participation. 4. Labour supply projections These features suggest the following initial approach to labour supply projection: 1. A projection based on constant 2005 narrow and expanded participation rates by population group, gender and age group. 2. A projection based on some closing of the gap (say, by a third) between White and other narrow and expanded participation rates by gender and age by 2024 and linear interpolation for the years in between. Table 4 sets out the results of the projections under different assumptions. From 2007 to 2024, the narrowly defined labour force is projected to grow at an average annual rate of between 1.29% (on the no-convergence, low immigration assumptions) and 1.86% (on the one-third convergence, high immigration assumptions). The absolute increase in the labour force varies from 4.40-million to 6.53-million. The expanded labour force is projected to grow at an average annual rate of between 1.27% (from a higher base) and 1.66%. In this case, the absolute increase in the labour force varies from 5.03-million to 6.79-million. The Microsoft Excel file proj13.xls sets out the population group and gender composition of the projected labour force under the various assumptions. The percentage Africans rises from 72.3% in 2007 to between 78.0% and 79.1% on the narrow definition of labour force participation, and from 74.6% in 2007 to between 79.5% and 80.4% on the expanded definition. The percentage males rises from 55.7% to between 57.7% and 58.4% on the narrow definition and from 53.6% to between 55.6% and 56.3% on the expanded definition. The reason the proportion of men rises is because the proportion of men in the population is projected to rise, partly as a result of international migration. 16

Projecting the labour supply to 2024 Table 4 Aggregate labour supply, 2007-2024 Millions Low immigration Year Narrow Expanded Constant Convergence Constant Convergence 2007 18.13 18.13 21.05 21.05 2008 18.47 18.53 21.45 21.48 2009 18.84 18.96 21.87 21.93 2010 19.20 19.39 22.27 22.37 2011 19.55 19.81 22.67 22.80 2012 19.89 20.22 23.05 23.22 2013 20.21 20.62 23.42 23.63 2014 20.52 21.00 23.78 24.02 2015 20.82 21.37 24.12 24.40 2016 21.10 21.73 24.44 24.76 2017 21.36 22.07 24.75 25.10 2018 21.60 22.39 25.03 25.42 2019 21.82 27.69 25.28 25.72 2020 22.01 27.97 25.51 25.99 2021 22.18 23.22 25.70 26.22 2022 22.32 23.45 25.87 26.42 2023 22.44 23.65 25.99 26.59 2024 22.53 23.82 26.08 26.73 Growth % p.a. 2007-2014 1.78 2.12 1.76 1.90 2014-2024 0.94 1.27 0.93 1.07 2007-2024 1.29 1.62 1.27 1.42 High immigration Year Narrow Expanded Constant Convergence Constant Convergence 2007 18.13 18.13 21.05 21.05 2008 18.52 18.58 21.50 21.53 2009 18.93 19.05 21.97 22.04 2010 19.34 19.53 22.44 22.54 2011 19.74 20.00 22.90 23.03 2012 20.14 20.47 23.35 23.52 2013 20.52 20.93 23.79 23.99 2014 20.88 21.37 24.20 24.45 2015 21.24 21.80 24.62 24.90 2016 21.58 22.22 25.01 25.33 2017 21.90 22.63 25.39 25.75 2018 22.21 23.02 25.75 26.15 2019 22.49 23.39 26.08 26.52 2020 22.75 23.73 26.38 26.86 2021 22.98 24.06 26.64 27.17 2022 23.19 24.35 26.88 27.45 2023 23.37 24.62 27.07 27.69 2024 23.52 24.86 27.23 27.90 17

centre for poverty employment and growth HSRC Growth % p.a. 2007-2014 2.04 2.38 2.01 2.16 2014-2024 1.20 1.52 1.19 1.33 2007-2024 1.54 1.87 1.53 1.67 18

Projecting the labour supply to 2024 Phase II report 1. Terms of reference 1. A closer analysis of LFPRs between September 2000 and September 2006 has been carried out, with an emphasis on discovering the role that population group, gender, age group, education, metro/non-metro residence and household circumstance play in the decision to participate in the labour force. 2. Particular attention has been paid to the classification of households. A typology has been developed along the following lines: Well-educated households; Solidly employed lower middle class households; Solidly employed working class households; Households with little attachment to the labour market; and Households predominantly dependent on agriculture. 3. A projection has been made of the distribution of education among the population and of Senior Certificate, university and technikon output to 2024. This has been set against the background of outputs in other countries, as indicated in UNESCO statistics. 4. In the light of the above, refined projections of LFPRs and hence of labour supply have been produced. Phase II research partly depended on feedback received at one or two HSRC workshops. 2. Household classification The following household typology is proposed: Upper middle class households These contain at least one member with a degree who is employed, with an income of at least R11,000 per month in 2005 prices, or who has reached the age of 55. Households with more than 50% of employment in agriculture do not fall in this category. Lower middle class households These do not qualify for upper middle class status, but they contain at least two members with a Senior Certificate/NTC III or better who are employed, with an 19

centre for poverty employment and growth HSRC income of at least R4,500 per month in 2005 prices, or who have reached the age of 55. Households with more than 50% of employment in agriculture do not fall in this category. Working class households These do not qualify for either upper middle class or lower middle class status, but have at least one member in employment, or a member who has completed Grade 9 and reached the age of 55. Households with more than 50% of employment in agriculture do not fall in this category Agricultural households These are households who have at least one person employed, and the majority of employment is in agriculture. Underclass These are households with no person under the age of 55 in employment and no person aged 55 or over who has completed Grade 9. The September 2005 Labour Force Survey yields the following estimate of individuals age 15-64 in each category. Table 1 Estimates of individuals aged 15-64 in each household category, September 2005 Labour Force Survey Number Percent Upper middle class 1,202,141 4.1 Lower middle class 5,900,229 20.2 Working class 13,957,105 48.0 Agricultural 1,339,848 4.6 Underclass 6,739,140 23.1 Total 29,138,462 100.0 There are three difficulties with household class, which are defined as follows: Social class is a theoretical construct which on the whole changes slowly for an individual household. It should therefore not be completely defined on instantaneous characteristics, but instantaneous characteristics are all we have. Our definition will perform reasonably well in the aggregate, since errors will tend to cancel out, but will be less reliable at the individual household level. Social class is a problematic independent variable in the explanation of labour force participation, since it is partly constituted by employment and hence by labour force participation, complicating the relevant econometrics. 20

Projecting the labour supply to 2024 From a prediction point of view, this concept of class would require a prior prediction of labour force participation and hence it cannot be used to predict labour force participation. For these reasons, class is dropped as a variable predicting labour force participation. 3. Determinants of labour force participation, September 2005 Logistic regressions on labour force participation (both on the official and expanded definitions) have been run for people between the ages of 15 and 64, and for men and women separately. Five models have been used, which successively add independent variables to the regressions: Model I Five-year age groups only Model II Five-year age groups Four educational categories 2 Model III Five-year age groups Four educational categories Metro/non-metro residence Model IV Five-year age groups Four educational categories Metro/non-metro residence Population group Table 2A presents results for the official definition of labour force participation and table 2B presents results for the expanded definition. 2 At most, complete general education (up to and including Grade 9). Incomplete further education (Grades 10 and 11 and NTC I and II). Complete further education (Grade 12, NTC III, certificate or diploma with less than Grade 12). Higher education (Certificate or Diploma with Grade 12 and degrees). 21

centre for poverty employment and growth HSRC Table 2A Odds ratios from logistic regression of labour force participation, September 2005 Panel A Official definition Men Category Model I Model II Model III Model IV 15-19 1.00 1.00 1.00 1.00 20-24 8.95 7.20 7.28 7.48 25-29 27.88 21.66 21.72 22.45 30-34 45.33 34.94 34.80 35.47 35-39 49.98 39.18 39.47 39.89 40-44 37.17 31.65 31.14 31.23 45-49 32.20 28.12 28.26 28.26 50-54 21.48 19.38 19.10 18.90 55-59 14.36 13.04 12.48 12.39 60-64 5.39 4.64 4.42 4.37 At most general education 1.00 1.00 1.00 Incomplete further education 1.14 1.05 1.03 Complete further education 2.15 1.91 1.83 Higher education 3.20 2.79 2.65 Non-metro residence 1.00 1.00 Metro residence 1.78 1.71 African 1.00 Coloured 1.61 Asian 1.39 White 1.17 22

Projecting the labour supply to 2024 Women Category Model I Model II Model III Model IV 15-19 1.00 1.00 1.00 1.00 20-24 8.27 6.09 6.20 6.02 25-29 16.40 12.06 12.36 12.12 30-34 21.53 16.20 16.36 16.33 35-39 18.77 14.82 15.00 15.05 40-44 17.37 15.32 15.25 15.50 45-49 17.84 16.29 16.11 16.43 50-54 10.74 10.14 9.95 10.29 55-59 6.76 6.42 6.35 6.61 60-64 2.16 1.92 1.85 1.99 At most general education 1.00 1.00 1.00 Incomplete further education 1.27 1.20 1.24 Complete further education 2.43 2.17 2.42 Higher education 4.97 4.54 5.33 Non-metro residence 1.00 1.00 Metro residence 1.60 1.66 African 1.00 Coloured 1.16 Asian 0.50 White 0.71 23

centre for poverty employment and growth HSRC Table 2B Odds ratios from logistic regression of labour force participation, September 2005 Panel B Expanded definition Men Category Model I Model II Model III Model IV 15-19 1.00 1.00 1.00 1.00 20-24 11.62 9.91 10.01 9.92 25-29 55.88 46.33 46.33 45.96 30-34 92.77 75.76 75.18 74.56 35-39 77.01 63.88 63.88 63.56 40-44 44.21 39.09 38.35 38.21 45-49 33.81 30.40 30.31 30.48 50-54 20.43 18.77 18.43 18.57 55-59 12.00 10.98 10.52 10.65 60-64 4.14 3.72 3.56 3.65 At most general education 1.00 1.00 1.00 Incomplete further education 0.96 0.89 0.90 Complete further education 2.02 1.82 1.91 Higher education 1.84 1.63 1.83 Non-metro residence 1.00 1.00 Metro residence 1.60 1.59 African 1.00 Coloured 1.35 Asian 0.94 White 0.83 24

Projecting the labour supply to 2024 Women Category Model I Model II Model III Model IV 15-19 1.00 1.00 1.00 1.00 20-24 11.54 9.44 9.50 9.02 25-29 30.13 24.52 24.66 23.86 30-34 34.86 28.56 28.53 28.81 35-39 27.13 22.98 22.97 23.60 40-44 17.67 15.91 15.79 16.61 45-49 16.02 14.62 14.46 15.41 50-54 7.92 7.41 7.32 7.96 55-59 4.47 4.19 4.15 4.60 60-64 1.23 1.11 1.09 1.27 At most general education 1.00 1.00 1.00 Incomplete further education 1.02 0.99 1.08 Complete further education 1.95 1.85 2.44 Higher education 2.51 2.40 3.60 Non-metro residence 1.00 1.00 Metro residence 1.22 1.35 African 1.00 Coloured 0.93 Asian 0.31 White 0.42 Note: Non-reference category coefficients not significantly different from one at the 5% level are indicated in italics. All other coefficients are significantly different from one at the 5% level. The tables together indicate that knowledge of age group alone allows labour force participation to be predicted with a high degree of accuracy, especially given the additional contributions of the other variables. Table 3 Labour force participation predicted by knowledge of age group Labour force definition Gender Correct: Model I Best-performing model Correct: bestperforming model Official Male 69.6 IV 71.6 Official Female 69.9 II 70.9 Expanded Male 80.3 I 80.3 Expanded Female 80.0 II 80.3 It follows that changes in the age and gender composition of the population of working age are likely to have the greatest impact on the size of the labour force. Nonetheless, some of the other variables have a significant (if small) impact. 25

centre for poverty employment and growth HSRC Table 2A yields the following information: Men The odds ratio 3 for the 35-39 age group relative to the 15-19 age group is highest in three of the four models. In Model II it is 30-34, but the difference is probably not significant. As expected, the odds ratio rises with age, peaks and then drops off. The odds ratio for incomplete further education relative to none or general is not significantly different from one in two out of the three models in which education appears. However, the odds ratios are significantly greater than one for a complete further education and for higher education in all models in which education appears. Education at these levels promotes labour force participation. The odds ratio for metro residence relative to non-metro residence is significantly greater than one in both models in which it appears, so metro residence promotes labour force participation. The odds ratio for Coloureds and Asians relative to Africans are significantly greater than one, so membership of these groups in itself promotes labour force participation. Women The odds ratio for the 30-34 age group relative to the 15-19 age group is highest in all four models. The behaviour of the odds ratio by age group is as expected. The odds ratio for incomplete secondary education relative to none or general is significantly greater than one (unlike for men), but it is not large. If anything, the odds ratios for complete secondary education and higher education are higher than for men. The odds ratio for metro residence is very close to the value for men. The odds ratio pattern by population group is different from the male pattern. The odds ratio for Coloureds is again above one, but the value is lower than the male value. The odds ratio for Asian and White women is significantly below one, indicating that membership of these groups decreases the probability of labour force participation. Table 2B yields the following information: For men and women alike, the odds ratio for the 30-34 age group is the highest relative to the 15-19 age group. 3 Odds are the probability of something happening divided by the probability of it not happening. Odds ratios are the ratios of odds in two different states, for example, two different population groups or two different age groups. 26

Projecting the labour supply to 2024 For men and women alike, the odds ratio for incomplete further education relative to general or no education is not significantly different from one. In both cases, the odds ratios for complete further and higher education are significantly greater than one, but the higher education ratios are markedly lower than in the official labour force case. The odds ratios for metro residence relative to non-metro residence are significantly greater than one. In the case of women, the odds ratios are lower than in the official labour force case. For men, the odds ratio for Coloureds relative to Africans is significantly greater than one, but the ratios for Asians and Whites are not significantly different from one. For women, the ratio for Coloureds is not significantly different from one, while the ratios for Asians and Whites are lower. The regressions form the basis for projections in the light of changing population composition (including residence in metropolitan and other areas) and changing educational attainment. They can be used on simulated databases for future dates. The regression coefficients will be assumed constant, but this is a more (by a significant, if small, margin) sophisticated basis of projection than keeping LFPRs by population group, gender and age group constant. The next task is to project the educational distribution of the population and its distribution between metro and non-metro areas to the year 2024. 4. Projecting the educational distribution to 2024 The approach taken here is to derive education transition probabilities by comparing the September 2000 Labour Force Survey with the September 2005 Labour Force Survey. A four-parameter transition model is used. The algebra proceeds as follows: Suppose the probability of being promoted from Category 1 in a five-year period is p(1+k). The probability of proceeding from Category 1 to Category 2 is p, and the probability of proceeding from Category 1 to Category 3 is pk. The probability of being promoted from Category 2 to Category 3 is q, and the probability of being promoted from Category 3 to Category 4 is r. All other transitions are assumed to be zero. Since we have an observation of age group j (say 15-19) in 2000 and age group j+1 (say 20-24) in 2005, there is an opportunity to compare the predicted distribution in 2005 with the observed distribution. There are three independent equations in p, q, r and k (since the number in Category 4 is fixed when Categories 1, 2 and 3 are fixed). If one of the four parameters is arbitrarily fixed (say k at 0.5), the equations can be solved for p, q and r for each age group up to 25-29 in the starting year and each period. Age groups starting with 30-34 are assumed not to move any further between 27

centre for poverty employment and growth HSRC educational categories, except for moves between further and higher education, which can go on to 35-39. Table 4 sets out the projected distribution of the 15-64 population across educational categories. Table 4 Percentage distribution of the 15-64 age group across educational categories Year Male General Incomplete further Complete further Higher 2007 46.48 21.47 22.83 8.21 2009 45.17 21.87 24.69 8.27 2014 39.56 23.69 27.58 9.17 2019 35.08 25.04 30.01 9.87 2024 31.02 26.23 32.27 10.48 Female 2007 45.88 22.40 23.64 8.08 2009 44.46 22.66 24.63 8.25 2014 40.44 23.47 27.27 8.82 2019 39.61 24.20 29.71 9.18 2024 33.53 24.89 32.10 9.48 5. Projecting the metro population to 2024 The 2007 Community Survey indicates that the proportion of the population aged 15-64 living in metro areas was 38.32%. The growth rate of this population is projected at the maximum of 1.5 times the growth rate of the national 15-64 population at 1.75%. This yields the following projection: Table 5 Projecting the metro population to 2024 Year Percentage 15-64 15-64 population (millions) Metro Metro Non-metro 2007 38.32 11.88 19.12 2009 39.01 12.49 19.53 2014 40.76 14.03 20.40 2019 42.41 15.41 20.93 2024 45.39 16.81 20.22 28

Projecting the labour supply to 2024 6. The labour force projection: methodology and results Everything necessary for the main labour force projection has now been assembled. The September 2005 Labour Force Survey will be used as a basis for the projection. Projected Labour Force Surveys for 2009, 2014, 2019 and 2024 will be created by keeping everything in the 2005 Labour Force Survey the same, except the household weights. These weights will be adjusted by iterative proportional fitting of the weights to yield marginal totals by population group, gender, age group, educational category and metro/non-metro residence 4. Once the new weights are calculated, everything that can be tabulated from the September 2005 Labour Force Survey can be tabulated from the projected Labour Force Surveys. The results are set out in table 6. The projected labour force from the simulation is compared with constant LFPR and converging LFPR projections (low immigration) from Phase I of the report. A comparison yields the following results: 1. The rate of growth of the labour force in the simulation is closer to the constant LFPR projection than the converging LFPR projection for both the official and the expanded labour force. The reader is asked to recall the finding in section 3 that the changing educational and metropolitan mix, while exerting some upward pressure on LFPRs, does not make much difference to them. 2. Nonetheless, the simulated labour force projections are higher than the constant LFPR projections by some margin. Why is this? The reason is quite subtle. The population projection used in both Phase I and Phase II of the report is based on Community Survey results. The LFPRs used in Phase I of the report are tabulated directly from the 2005 September Labour Force Survey, benchmarked to what Statistics South Africa then thought was the size and composition of the population. However, if one were to back-project the population from the 2007 Community Survey, one would get a different population composition and size in 2005, and consequently different (and generally higher) LFPRs. Table 7 compares the two population projections. 4 Statistics South Africa benchmarks its surveys to demographic model estimates in this fashion, using the CALMAR programme. 29

centre for poverty employment and growth HSRC Table 6 Aggregate labour supply, 2007-2014 Official Expanded Converging Converging Year constant Simulation constant Simulation LFPR LFPR LFPR LFPR 2007 18.13 18.13 19.67 21.05 21.05 22.30 2009 18.64 18.96 20.45 21.87 21.93 23.16 2014 20.52 21.00 22.33 23.78 24.02 25.20 2019 21.82 22.69 23.82 25.28 25.72 26.77 2024 22.53 23.82 24.64 26.08 26.73 27.61 Annual growth rates (%) 2007-2009 1.40 2.26 1.96 1.93 2.07 1.91 2009-2014 1.94 2.06 1.77 1.69 1.84 1.70 2014-2019 1.24 1.56 1.30 1.23 1.38 1.22 2019-2024 0.64 0.98 0.68 0.63 0.77 0.62 2007-2024 1.29 1.62 1.33 1.27 1.42 1.26 Table 7 Population size and composition, September 2005 Labour Force Survey and Community Survey Size of 15-64 population ( 000s) Back-projected from 2007 Age composition (%) September 2005 LFS Community Survey 29,386 31,007 30.127 21.47 15-19 16.67 16.45 20-24 15.71 15.55 25-29 14.33 13.14 30-34 12.81 12.14 35-39 9.50 10.41 40-44 8.43 9.19 45-49 7.45 7.83 50-54 6.00 6.41 55-59 4.85 5.13 60-64 4.25 3.75 30

Projecting the labour supply to 2024 So, provided the Community Survey has improved our population estimate, the higher estimate of the labour force provided by the simulation is appropriate. Once the weights are recalculated, it is straightforward to project other variables. Table 8 sets out the employment projection, which can be regarded as a baseline around which alternatives can be considered. Table 8 Aggregate employment and unemployment rate, 2007-2014 Employment Unemployment rates (%) Official Expanded 2007 15,480,458 21.3 30.6 2009 16,092,059 21.3 30.5 2014 17.629,296 21.1 30.0 2019 18,876,723 20.7 29.5 2024 19,636,314 20.3 28.9 Annual growth rates (%) 2007-2009 1.96 2009-2014 1.84 2014-2019 1.38 2019-2024 0.79 2007-2024 1.41 Table 8 has employment growing quite rapidly from 2007 to 2014. There is greater room for improvement between 2014 and 2024. Equally, it is possible to project the class composition of the 15-64 population. Table 9 shows that South Africa is projected to become more middle class, with declining percentages of the population in the working class, in agriculture or belonging to the underclass. Stronger employment growth will accentuate the trend. Table 9 Distribution of individuals 15-64 by class Percent 2007 2009 2014 2019 2024 Upper middle 3.90 3.91 4.16 4.35 4.47 Lower middle 12.47 13.09 15.03 16.85 18.37 Working 53.56 53.31 52.42 51.62 51.02 Agricultural 3.14 3.14 3.03 2.91 2.88 Underclass 26.92 26.55 25.35 24.26 23.27 Total 100.00 100.00 100.00 100.00 100.00 31

centre for poverty employment and growth HSRC 32