A sectoral analysis of output elasticity of employment in South Africa

that, while South Africa’s formal agriculture is no longer labour-intensive, due to agricultural mechanisation, economic policy consciousness in the informal sector, including agri-entrepreneurship, is necessary to create inclusive mass employment in South Africa. Contribution: This study delves into the informal sector, which has been frequently overlooked as a potential solution to South Africa’s unemployment crisis.


Introduction
The consistent rise in South Africa's unemployment rate since 2008 (see Figure 1) is cause for concern, as the scenario appears to be deviating from the National Development Plan (NDP)'s goal of reducing unemployment to at most 6% by 2030.The seasonally adjusted unemployment rate in the third quarter of 2021 (34.69%) is almost six times higher than the 2030 targeted maximum threshold.Based on Okun's law, one can assume that South Africa's economy is underperforming.However, the country's output, as measured by the real Gross Domestic Product (GDP), has been steadily positive, save for the COVID-19 structural break in 2020, from which the economy is gradually recovering (see Figure 1).Okun (1962) proposed the inverse output -unemployment nexus and, as a result, concluded that economic growth is a prerequisite for job creation.This proposition has been the basis for most job-creating economic growth empirics in developing countries.
In South Africa Okun's proposition ceased to hold in the early 1990s due to structural changes (Mkhize 2019).This has been the case for the majority of African countries, regardless of the distortion period (see, for example, Ajilore & Yinusa 2011;Dada 2018;Léautier & Hanson 2013).Thus, for the past decades, the growth pattern in the majority of African countries has been described as 'jobless growth' (Léautier & Hanson 2013), and it seems that South Africa is no exception (Figure 1).Jobless growth tends to be a threat to inclusive growth, resulting in poverty, inequality and social unrest (Dada 2018).South Africa is in a state of disarray as a result of these challenges, which necessitate immediate policy intervention.According to the World Bank Group's April 2020 poverty and equity statistics, approximately 55.5% of South Africa's population lives below the national upper-bound poverty line, while the income distribution, as measured by the Gini index, stood at 63.3%.Following the severe civil unrest that the country experienced in July 2021, these statistics are likely to have worsened.
For the past three decades, the South African economy has undergone several fundamental structural reforms to boost economic growth and address the country's ongoing tripeptide of challenges (unemployment, poverty and inequality).Educational reforms to promote human capital development through education, industrialisation, and various socioeconomic transformation initiatives are among these reforms.This period has, therefore, seen an influx of secondary and tertiary graduates into the job market, yet various industries were experiencing structural shifts from labour-intensive to capital-intensive.Such structural changes often result in a surplus in the labour force due to the accumulation of new entrants and job losses (Altman 2006).In line with Altman (2006), South Africa's Quarterly Labour Force Survey (QLFS) statistics show that job losers and new entrants increased by approximately 50.4% and 38.4%, respectively, from the first quarter of 2008 to the first quarter of 2019.While industrialisation and human capital development through education are commendable efforts, Léautier and Hanson (2013) caution African countries not to neglect the agriculture sector.Africa remains an 'agriculture-based society and, will need to look to the agriculture sector to generate the bulk of needed employment in the coming years ' Léautier and Hanson (2013:1).Similarly, Dada (2018) asserts that agriculture-led economic growth is inclusive, and thus has a high potential for creating mass employment in Africa.
The period has also seen a rise in employment in the informal sector.According to QLFS data, informal employment increased from 15.3% in 2013 to 18.3% in 2019, while formal employment decreased from 71.2% to 68.5% over the same period.It is assumed that the unemployed youth population living below the poverty line turned to the informal sector for employment.This is consistent with Blaauw's (2017) assertion that in a predominantly poor society, informality is the primary means of earning income and overcoming poverty.As of 2018, one in every six employed South African citizens worked in the informal sector, which contributed approximately 18% of the country's GDP (Fourie 2018).This brief on the informal sector exhibits the sector's potential to support the NDP's 2030 agenda of creating 90% of jobs by small, medium-and micro-enterprises (SMMEs).However, the sector has remained on the periphery of economic analysis and policy consciousness.

Employment-output Nexus
Employment and output relationships can vary, depending on a variety of factors, including output per worker, time frames, labour and goods market adjustments, and sectoral economic activities (see, for example, Landmann 2004; Sahin, Tansel & Berument 2013).In the short run, output growth may not lead to an increased labour demand.Rather, it may only result in increased working hours because firms may avoid the costs of recruiting and training new workers by paying overtime to the existing workers, holding other things constant.However, constant output growth may overwhelm the existing workforce, forcing firms to increase employment capacity in the long run.
Using Australian workers, Okun (1962) proposed that a country's output growth must be approximately 3% above the nominal rate in a year to achieve a 1% point increase in employment.In several studies, however, it is argued that, while Okun's coefficient is a useful benchmark in the formulation of an employment-output macroeconomic policy, it varies with the economy and time frame under consideration, among other factors, that influence the employment intensity of output growth (see, for example, Karim & Aomar 2016;Okun 1970).On the other hand, while industrialisation and other structural reforms may lead to increased productivity and thus output growth both in the short and long run, they also force firms to downsize their labour force, weakening or even eliminating the positive employment-output relationship.For instance, Upender (2006) found that the employment-output relationship in the Indian agricultural sector switched from positive during the pre-reform period (1983)(1984)(1985)(1986)(1987)(1988)(1989)(1990)(1991) to negative in the post-reform period (1992)(1993)(1994)(1995)(1996)(1997)(1998)(1999)(2000).Thus, the fourth industrial revolution may result in jobless growth as it promotes the use of machinery in place of the considerable quantity of human labour, decreasing the demand for labour, however, with a positive impact on output.
Furthermore, as Sahin et al. (2013) point out, various movements occur in the labour and goods markets as a result of the market's specific macroeconomic factors and institutional settings.In general, disequilibrium between the two markets tends to be a result of labour market rigidities.
Firstly, the effects of the output shocks phase out faster in the short run than those of labour market shocks.Secondly, closing the long-run employment gap may take longer than closing the long-run output gap.Thirdly, changes in the labour market are smaller than changes in the goods market, and changes in output are accompanied by smaller changes in employment.As a result, the economy's employment creation puzzle may be structural, and the supply side's ability to generate employment may be weaker in the short run.However, as the interaction between the forces of the goods and labour markets develops, they tend to reach equilibrium in the long run.
The relationship between employment and output can also be explained in terms of the economy's sectoral economic activities (see, for example, Mkhize 2019; Ajilore & Yinusa 2011).The heterogeneity in the sectoral employment intensity of output growth, according to Sahin et al. (2013), is due to differences in cyclical behaviour, wages, technological intensity, capital-labour substitution effects, and employment multipliers.This implies that looking through a sectoral lens is vital when considering the issue of employment intensity of output growth.However, empirical evidence on the subject is still lacking, particularly in sub-Saharan Africa, where unemployment is rampant.In Botswana, Ajilore & Yinusa (2011)  Considering the educational attainment of the current unemployed population in South Africa (Figure 2), the informal sector may be a significant source of employment.
Figure 2 shows that 60% of the unemployed population have not attained secondary education.Furthermore, due to the country's educational reforms, there has been an influx of tertiary education graduates in the labour market, resulting in a decrease in the demand for secondary and matric education graduates, which account for 33% of South Africa's unemployed population.This implies that 93% of the unemployed population in South Africa can hardly be absorbed in the formal sector (Figure 2).In several studies it is suggested that when there is an excess of labour in the market, informality may serve as a cheap alternative to formal employment (

Empirical strategy
Following Mkhize (2019), Ajilore & Yinusa (2011), and Upender (2006), this study utilises the double-log linear labour demand function (Equation 2) derived from the constant elasticity of substitution (CES) production function (Equation 1) through solving the marginal product of labour: Where: As in the Cobb-Douglas production function, y is the output, K is the capital input, L is the labour input, and A is the productivity parameter.The subscript γ represents the extent of substitution between capital and labour inputs and is related to substitution elasticity δ = 1/(1 + γ) (Romeo 2020).
The subscripts ∂ and Ø proxy for the returns to scale and distribution parameters, respectively.Equation 1 can be decomposed to the double-log linear employment function specified: For the purpose of this study, y is the sectorial gross value added (GVA) at factor cost in real terms, L is the employed population in a specific sector, X is a set of regressors that are justified in the literature as vital in explaining labour demand, ε is the disturbance term, and t represents time.These regressors include sector-specific nominal wages (seasonally adjusted), the user cost of capital, and the inflation rate.The estimated coefficients of Equation 2 are interpreted as elasticities because the model is specified in logs.

Estimation technique
The study adopted the Engle-Granger two-step testing procedure (Engle & Granger 1987) to test the null hypothesis of jobless growth assuming evidence that it exists if there is no co-integration between employment (L) and output (y).
Engle-Granger's two-step testing procedure proceeds in two linked steps.The first generates the residuals from the static co-integration equation.The generated Ordinary Least Squares (OLS) residuals are a measure of disequilibrium.A co-integration test is thus a test of whether the OLS residuals have a unit root: The residuals have a unit root.
Accepting H 0 implies that there is no co-integration; thus, confirming jobless growth in that particular sector; otherwise, Okun's law applies.If co-integration holds, the static cointegration equation is said to be super-consistent, implying that the standard OLS diagnostic test may not be required because the only thing that matters is residual stationarity.A valid error correction term (ECT[−1]) exists when a set of variables is co-integrated.Accordingly, the second step is to estimate the error correction model by fitting a regression of first-differenced residuals on lagged residuals using the generated OLS residuals from the first step.To indicate the presence of a long-run causal relationship, the ECT(−1) must be significantly negative.

Estimated results and discussion
The findings are presented in the Engle-Granger two-step testing procedure, but we begin with descriptive statistics (Figure 3).

Real GVA at factor cost (R millions) Employed populaƟon Year
Table 2 shows that the null hypothesis of no co-integration could not be rejected on total agriculture and formal agriculture models, confirming jobless growth.Similar findings were documented in Turkey by Sahin et al. (2013).They found that the Turkish agriculture sector did not demonstrate an employment-output link both in the short run and long run during the period , reflecting characteristics of jobless growth.This finding, as highlighted in descriptive statistics, confirms that South Africa's agriculture sector predominantly depends on commercial agriculture.Commercial agriculture has become capitalintensive as a result of agricultural mechanisation; therefore, the evidence of jobless growth in the sector is not surprising.
On the other hand, we rejected the null hypothesis in both the informal agriculture and informal non-agriculture models, demonstrating Okun's law features in the sectors.This implies that there was a long-run relationship between the informal sector's employment and output for the period .

Employment intensity of output growth results
The findings on the employment intensity of output growth in South Africa's agriculture and informal sectors are shown in Table 3 below.Following the results in Table 2, we estimated Equation 2 in error correction form for both the informal agriculture and informal non-agriculture models to determine the adjusting speed to equilibrium, while the total agriculture and formal agriculture models were estimated within a shortrun perspective.The ECT(−1) denotes the rate at which labour demand returns to equilibrium, following a change in output and other explanatory variables in the model.The coefficients employment elasticity of output growth, pointing to the causal relationship between employment and output.
Table 3 shows that the employment-output relationship is statistically insignificant in both total agriculture and formal agriculture models, even in the short run.Regarding the co-integrated models, we found negative and highly significant ECT(-1) values in both informal agriculture (-0.64) and informal non-agriculture (-0.86) models, implying a disequilibrium adjusting rate of 64% and 86% within a year, respectively.However, the informal agriculture sector's employment intensity of output growth is statistically insignificant.This is not surprising considering that informal agriculture in South Africa has been neglected since the 1990s (see Figure 3). 1 The findings of the non-agriculture informal sector model show a fairly elastic (1.35%) employment intensity of output growth over the study period, with a statistical significance at 1%, ceteris paribus.If we review our findings together with those of Mkhize (2019), we derive that the non-agriculture informal sector is the South Africa's second most labourintensive sector after the finance and business services sector (1.56%).
All control variables entered the respective models as expected.The estimated coefficient of cost of capital in the informal agriculture sector model is highly significant and negative, implying that the high cost of capital discourages potential agri-entrepreneurs and the growth of existing ones.Anecdotal evidence suggests that South Africa's agrientrepreneurs (both start-ups and existing) face capital constraints to finance raw materials, input, and operational costs.Although some raw materials can be substituted with labour, such as labour instead of tractors, productivity will be low.Conversely, the same variable is significantly positive in the non-agriculture informal sector model.This implies that non-agriculture SMMEs can substitute capital with labour when capital costs are high.This is especially true in South Africa, where the service industry, particularly trade, dominates the informal sector (Rogan & Skinner 2017).Thus, an increase in the cost of capital reduces capital demand while increasing labour demand.The estimated nonagriculture informal sector's wage coefficient is positive and highly significant, indicating that entrepreneurs are drawn to high-paying informal economic activities, and their wages according to Zhou and Pindiriri (2015), depend on output.Inflation raises operating costs, particularly for SMMEs, and thus discourages entrepreneurship.
The long-run regressions in As shown in Table 4, no null hypothesis could be rejected for any of the tests, indicating that the long-run estimates in Table 3 are robust and correctly specified.
1.However, Okun's law validity in the sector suggests that it has the potential to significantly contribute to the country's employment in the long run.

Conclusion
Given the ILO's assertion about the inverse relationship between education level and informality, discerning the employment intensity of output growth in the informal sector is vital, especially in developing countries where education attainment levels are relatively low.For South Africa in particular, QLFS statistics show that from 2008 to 2019, an average of 93% of the unemployed population did not have tertiary education qualifications, implying that the informal sector can be a significant source of employment.However, the sector has remained on the periphery of economic analysis and policy consciousness.Furthermore, literature perceives agriculture as an inclusive labour-intensive sector, and thus a source of mass employment in developing economies.Thus, this study examined the responsiveness of sectoral employment to changes in the sectoral output of South Africa's agriculture and informal sectors for the period (1993-2018), using the Engle-Granger two-step testing procedure on the doublelog linear labour demand function.We found that employment and output are only co-integrated in the informal sector (both agriculture and non-agriculture), confirming Okun's law features.However, the employment elasticity of the informal agriculture sector's output growth was statistically insignificant over the study period.This finding our statistics, which show that small-scale farming is neglected in South Africa, despite its potential to generate self-employment.The findings of the nonagriculture informal sector exhibited a highly significant and fairly elastic (1.35%) employment intensity of output growth over the study period, with an equilibrium adjustment rate of 86% within a year, ceteris paribus.This finding leads to the conclusion that the non-agriculture informal sector is South Africa's second most labourintensive sector, after the finance and business services sector.In the formal agriculture sector, the employmentoutput relationship appears to have been eliminated by agriculture mechanisation which promoted jobless growth in the sector.These findings suggest that South Africa's formal agriculture is no longer labour-intensive as a result of agricultural mechanisation.However, in line with sustainable development goal (SDG) 8.3, economic policy consciousness about the informal sector, including agrientrepreneurship, is necessary to create inclusive mass employment, which is an essential missing piece in the puzzle to curb unemployment in South Africa.Given the availability of data for a significantly longer period (prior 1993 or alternatively quarterly data), future studies can employ robust time-varying techniques for the same study.

FIGURE 1 :
FIGURE 1: South Africa's unemployment -Gross Domestic Product relationship for the period Q1 2008 to Q3 2021.

FIGURE 3 :
FIGURE 3: Employment and gross value added in the South Africa's Agriculture and Informal sectors.

Unemployment rate (%, seasonally adjusted) Period: Q1 2008 to Q3 2021 Real GDP (R millions, seasonally adjusted)
Average proportion of South Africa's unemployed population according to the highest level of education for the period Q1 2008 to Q1 2019.
output growth in the informal sector, yet according to the International Labour Organisation (ILO)'s press release on 30 April 2018, 85.8% of employment in Africa is informal, with informal agriculture accounting for the largest proportion.The press release also emphasised the significance of education levels in informality in developing economies: Source: StatsSA, 2019, Quarterly Labour Force Survey (QLFS) dataset, Stats SA, PretoriaFIGURE 2: Mkhize's (2019)ier & Hanson 2013)sipanodya 2020)Pasipanodya 2020).Agriculture, on the other hand, is perceived as a labour-intensive and inclusive source of mass employment, typically in Africa (see, for example,Dada 2018;Léautier & Hanson 2013).Thus, the current study builds onMkhize's (2019)empirical work to provide an alternative empirical analysis of the sectoral employment intensity of output growth in South Africa, with a particular emphasis on agriculture and informal sectors.The analysis is carried out from the theoretical standpoint of Okun's law.

TABLE 1 :
Variables description and data sources.In terms of data, the study utilised readily available sectoral time-series annual data from 1993 to 2018.The study periodis constrained by the availability of data.Table1describes variables and data sources.
Notes: The critical values for the Engle-Granger co-integration test on regression residuals at 1%, 5%, and 10% are -2.62,-1.95, and 1.61, respectively.The null hypothesis of no co-integration is rejected if the | test statistic's value| > |the critical value at 5%|.ADF, Augmented Dickey-Fuller.

TABLE 3 :
Labour demand elasticities in South Africa's agriculture and informal sectors.