AN ANALYSIS ON THE ANTECEDENTS OF YOUNG EMPLOYEES’ EARNING POTENTIAL IN SOUTH AFRICA

Concerns surrounding the outcomes of young people in the labour market have become a global phenomenon. None more so has this been evident than in South Africa. While finding employment seems highly unlikely for most, circumstances surrounding the outcomes of those who are employed have likewise raised alarms. Understanding what contributes to decent employment in this regard holds important relevance not only in promoting the wellbeing of the cohort, but also advancing the knowledge on the requirements to utilise their local development potential. The purpose of the study was therefore to identify the antecedents of young South African employees’ earning potential. A quantitative approach and cross-sectional research design were employed, where secondary data collected through the most recent national labour market dynamics survey was used. The sample consisted of 27 493 young employees between the ages of 15 and 34 years. Descriptive statistics, cross-tabulations and a linear regression were utilised for the analysis. Results of the study showed the earnings potential of young South Africans to be intertwined with an array of aspects. In this regard, the remuneration prospects of young people depend on race, sector of employment and trade union membership. Outcomes were also highly gendered biased, while the impact of labour market inequalities including the lack of social capital and geographical characteristics likewise seems to affect their earnings ability. Recommendations in advancing the cohort’s labour market outcomes must, therefore, include enhanced public-private partnership formation through mandatory apprenticeship programmes. This must be supported by the relaxation of labour market regulations and a strategic focus that seeks to advance both soft and technical skills.


INTRODUCTION
The focus on quality and decent employment has gained unprecedented attention over that past few years, particularly for young people. Labour market dynamics facing the youth are influenced by a number of intertwined factors, which in most instances have proven to be structural in nature. This situation is not unique to South Africa, as the country has one of the highest youth unemployment rates in the world (Meyer, 2017;De Lannoy et al., 2018). Negative employment outcomes are mostly shaped by age, gender and race, with young people representing a larger share (Branson et al., 2019). The rise of non-standard employment as a result of instilling more flexibility in the labour market and reducing costs has in some way forestalled the fate of many young people with a large majority finding themselves in these kinds of employment; unstable and low paying. A growing number of young people are now being hired for a specific time and on temporary basis, reducing employment stability (Scherer, 2004). Such kinds of employment have brought about increased insecurity, low and unstable incomes, and ultimately poverty (Schmid, 2010).
This has raised some concerns as to which factors are significant in determining the earnings they receive. Classical theories of earnings determination have always argued that wage levels are established through the interaction of demand and supply in the labour market (Tanaka, 2014). On the other hand, some theories acknowledge the difference between workers, and therefore wages would according to these theories be determined by factors not related to a worker's productivity, but others such as institutions (Doeringer & Piore, 1985) and the productivity needed for a job (Thurow, 1975). The human capital approach contends earnings to differ primarily based on the attained physiognomies, with higher education and training as an indication of positive future earning potential (Becker, 1964).
Within the context of South Africa, studies surrounding the issue, particularly on the earnings potential of young people, have been scarce. Most have in fact focused primarily on the understanding of youth unemployment (Mlatsheni & Ranchhod 2017;Graham & Mlathseni 2015), while others (Banerjee et al., 2008;Ntlhola et al., 2019) have focused on how trade unionism affects wages. However, the knowledge revolving around the determinants of earnings young people receive, over and above the human capital argument, is still largely limited. The study in this regard sets out to investigate the factors that determine or rather influence young employees' earning potential in the South African labour market. From this perspective, it is important to not only understand why the cohort is struggling to find employment, but more equally so, the dynamics surrounding their access to quality and decent working conditions (Mankiw et al., 2018).

LITERATURE REVIEW
Earnings are thought to be affected by several factors such as human capital components, socio-family factors, employment and firm size, gender and family circumstance factors, mental attributes, wellbeing and appearance (Pytliková, 2013). These factors assume a pivotal role in the continuing existence of employees and their families (Bhattarai, 2017). This is particularly significant for developing countries where wage employment is extremely rare. In accordance with past research in the convention of Mincerian wages, various customary human capital controls, for example education, location and experience, are typically secured as elements deciding earnings (Mincer, 1974). An individual's earning potential reached its peak point when they have had roughly 30 years of work experience, which is by and large at around 45 years (Bhattarai, 2000).
Earnings differentials by sexual orientation in the labour market are regarded as one of the key issues (Atlonji & Blank, 1999). Even though there is no doubt that earning levels seem to increase with levels of education, certain earnings gaps are also evident between male and female employees (Nestiae, 2004). The sector of employment also seems to be one of the factors that are believed to have an effect on the earnings of workers. In fact, the segmented labour market theory proposes that workers in the informal sector are typically subject to lower earnings relative to their formal sector counterparts (Tansel & Kan, 2012).
The role of trade unions has sought to improve the earnings and working conditions of their members. Theoretically, unions are believed to generally reduce employment levels in the unionised sector by raising earnings above market-clearing levels by misusing their imposing firm model power (Freeman & Medoff, 1984). The high earnings in the unionised sector may cause an excess supply of labour, which puts downward pressure on the earnings of non-unionised workers. Trade unions are therefore thought to determine gains to the detriment of non-unionised workers and the unemployed (Reynolds, 2008). On the other hand, unions are believed to contribute towards the reduction of inequality through activities that advance the incomes of workers (Bryson, 2014).
According to Culpepper (2009), the geographical area is one of the essential factors that are used in determining the earnings of workers. Normally, firms gain by being in an area closer to an enormous assortment of intermediary inputs, which implies fewer costs (Amiti, 2005). Therefore, firms with the best market access can bear to pay an excess of 20 per cent higher worker earnings than those secluded from economic and market activity, i.e. rural areas (Amiti & Cameron, 2007). Young people staying in these areas are even at a detriment since they earn less income than their urban counterparts, irrespective of where the urban workers come from (Culliney, 2017). With regard to education, human capital assumes a crucial role in improving employment prospects, including future earnings (Green & Henseke, 2016). Van Der Berg et al. (2011) argue that the inequalities in the quality of education play an important role and can likewise contribute and sustain earnings disparities. Likewise, Hossain et al. (2015) find education, age (which is used as a proxy for work experience), gender and work environment causing earning differences.
Several studies have sought to understand the relationship between earnings and various personal, socio-economic factors. Gong and Van Soest (2002) confirmed the segmented labour market theory in Mexico and found that formal sector workers are better compensated for their jobs than their informal sector counterparts. Similar findings are reported by Baskaya and Hulagu (2011) in Turkey that formal workers in reality essentially earn more than informal workers notwithstanding controlling for observable qualities. Contrariwise, in Argentina, Pratap and Quintin (2006) found no differences among earnings of workers between the two sectors even after individual characteristics are controlled for. Comparable findings are likewise affirmed by Arias and Khamis (2008). Nahm et al. (2017) studied the relationship between hourly earnings and union membership in Australia and found that unionised workers earn more than those not part of a trade union. Specifically, the results showed a 12 per cent wage premium against non-unionised male workers and an 18 per cent wage premium for female unionised workers relative to non-unionised female workers. In South Africa, for example Ntuli and Kwenda (2014) found that unionised workers significantly earn higher wages than non-unionised workers, while other studies (Bhorat et al., 2012) observe the wage premium to be low, ranging between six and seven per cent when various elements such as firm size and business type among others are considered. The latest investigation by Ntlhola et al. (2019) presumes that unions have little earning compression impacts in South Africa.
Youth in rural regions contrasted with urban youth are not only found to experience poor work prospects in South Africa (Mlatsheni & Ranchhod, 2017), likewise they also earn lower earnings in rural Britain (Culliney, 2017). In terms of employment sector, in their investigation, Bargain and Kwenda (2011) found that earnings of workers in the informal sector come up short contrasted with their formal sector counterparts in Brazil, Mexico and South Africa, with the earnings' penalty considerably more significant in South Africa.

Research method and data
The primary objective of the study was to investigate factors that determine the earning potential of young South African employees. The study follows a crosssectional research design, and therefore a quantitative research approach was deemed fit and suitable. The paper used secondary data based on the 2017 labour market dynamics (LMD) dataset published by Statistics South Africa in June 2019. The dataset is constructed through the pooling of all data from the four quarterly labour force surveys (QLFS) to form an annual dataset (StatsSA, 2019). In conducting the surveys, StatsSA uses a master sampling technique in which the sample is drawn from primary sampling units (PSUs) that are equally divided into four subgroups and rotated for each quarter. The method ensures the sample to be representative across various aspects, including provincial level, metro and nonmetro areas as well as geography type (urban and informal). While the survey attributes the collection of data across all ages (15-64) of labour market participants, for the purpose of this study, the sample was restricted to youth in the 15 to 34 years age category who were employed either on a full-time or part-time basis. As a result, the sample totalled 27 493 young employees across all nine provinces within the country.

Data analysis and model specification
By analysing the data, descriptive statistics, cross-tabulations and a multivariate linear regression were utilised using the IBM SPSS Statistics Version 25 software. Where the latter is concerned, an ANOVA model was estimated, since some of the independent variables are of a qualitative nature (Gujarati, 2003). The model that was used was specified as follows: Where EP is the earning potential; a continuous variable represented by monthly wages, β0 is the constant term that will capture average earnings. βi is the constant term associated with Xi, which is age, βj is the constant term associated with Dj which are the categorical values that will be entered as dummies where the number of dummies will be n -1; where n is the number of categories. This means that the number of variables required is one less than the number of groups that are recorded (Field, 2018). Henceforth, where there are five categories, four dummies were created and the fifth dummy served as the baseline or reference point against which all other groups were compared. Keeping in line with the aforementioned, and the primary objective of the study, the model was formulated as follows: Here, monthly earnings served as the dependent variable, while a total of eight independent variables were included. The latter consisted of both continuous and categorical variables. Table 1 below elucidates on the coding of the explanatory variables including a description of the dummy variables that were used.  Prior to use, the model's adequacy was evaluated where correlation analysis was run with the aim of determining whether there was any presence of multicollinearity with the data used (Hair et al., 2010). In doing so, the study made use of various collinearity diagnostics, including the use of tolerance values and variance inflation factors (VIF).

RESULTS AND DISCUSSION
As the first step in the analysis, the study provides an overview of the demographic composition of the sample. Table 2 shows the frequencies and percentages of various selected demographic characteristics. Based on the results, the sample seemed to showcase a distribution slightly in favour of males (56.6%) compared to females (43.4%). Moreover, from a race perspective, the distribution shows a strong similarity with national estimates, where approximately 80 per cent of the sample were black, followed by 11.1 per cent who were coloured and 6.4 per cent of the sample who were white. These similarities were also expressed in the provincial distribution. In accordance with this attribute, most of the participants were based in Gauteng (27.7%), KwaZulu-Natal (15.9%) and the Western Cape (15.9), while only 3.6 and 5.4 per cent of the sample were situated in the Northern Cape and North West, respectively. Further results, as shown in Table 2, depict that approximately two thirds (67.5%) of the population were either living on their own or not married. This distribution most probably is representative of the age cohort of the sample. Based on the distribution according to education level, approximately three quarters (74.5%) attributed some form of secondary education, while less than seven per cent had no schooling or some form of primary education. Similarly, a large contingent indicated that they were employed in the formal sector (75.4) when compared to those employed in more informal sectors. These results tend to lend credence to the importance of having at least a secondary education in young peoples' pursuit in obtaining some form of stable employment (De Lannoy et al., 2018).
In accordance with elucidating the factors that possibly contribute to better employment opportunities, Table 3 shows the cross-tabulation results and chisquare statistics for different earnings categories. This is done with the purpose of identifying the presence of any noticeable differences for each respective socioeconomic characteristic.  Table 3 showcase that all selected attributes displayed low p-values (sig. = 0.000) and relatively high chi-square statistics inferring statistically significant differences in their distribution. For example, considering age, results suggest that each older age grouping had comparatively higher percentages of individuals in higher earnings categories. Similar to a gender perspective, slightly more males identified themselves to earn either between R5 000 and R18 000 (17%) or more than R18 000 (5.4%) when compared to females (15.8% & 5.0%). These findings also tend to resonate with those presented by Beukes et al. (2017).
Based on race, results from Table 3 showcase that more than half (56.3%) of the African/black individuals indicated to earn between R0 and R5 000, while Asian/Indians and white individuals had higher distribution in higher earning categories. In explaining these results, Graham and De Lannoy (2016) posit that even after the transition to democracy, Africans compared to other races have carried the brunt of the difficulties in the labour market partly due to historically low investments in education and low intergenerational transfers of human capital. In addition to this, the sample also showed noticeable differences based on geographical attributes. In fact, of those situated in urban populations, 18.3 per cent indicated that they earn between R5 000 and R18 000, and 6.7 per cent indicated to earn more than R18 000, while more rural provinces had comparatively lower distributions (15.1% & 4.2%) in these categories.
In accordance with the last two categories, distribution results for education seem to support the fact that demand in the youth labour market has shifted towards high skills (Mlatsheni, 2014). Results depicted in Table 3 show that 16.4 per cent of those attributing a tertiary level earned more than R18 000; this is significantly higher than those reported for secondary education (2.9%) as well as primary education (1.5%). Finally, given the cyclical attributes and sensitivity of demand for young work seekers, it was important to reflect on the nature of trade union membership and its dynamic with young peoples' earning potential. Results in this regard tend to suggest that membership for the youth have a positive influence on earnings potential, supporting the findings of Anand et al. (2016).
Here, results in Table 3 show that those with trade union membership had a 25 per cent lower representation in the R0 to R5 000 earnings category. However, distributions in the R5 000 to R18 000 (32.8%) and more than R18 000 (9.7%) categories were significantly higher compared to those who are not unionised.
Subsequent to the descriptive analysis, the study employed an ANOVA regression model. The purpose of the model is to assert a view on how the dependent variable varies across the different categories of the explanatory variables. In doing so, the model assists in identifying the factors that affect the earnings potential of young South African employees. Table 4 below reports the results of the utilised model. Here, the coefficient for age is positive, suggesting a positive association between age and monthly earnings. This implies that the older the young people are, the higher their monthly earnings would be. The p-value for the coefficient is estimated at 0.000, suggesting that the null hypothesis indicating the coefficient = 0 is rejected at the 1 per cent level of significance. Oluwajodu et al. (2015), in their study, likewise showed that younger workers lack experience and therefore preference for matured workers is higher. Branson et al. (2019) also found young people, particularly those between 20 and 25 years, to be more likely to be in low pay employment than individuals in the 55 to 64 age bracket. In addition to earning less, earnings growth for youth is more limited relative to other age groups (Ranchhod, 2013). The coefficient for gender was negative and statistically significant (p-value = 0.000), suggesting that female workers on average earn R1 613.32 less compared to their male counterparts. Similarly, Ryan (2013) also found that young females earn far less than their male counterparts do, irrespective of education status. Even when they are employed, female youth encounter significant adversities (Ingle & Mlatsheni, 2017;Branson et al., 2019). The coefficient for marital status, in particular for married youth, was positive and statistically significant with a pvalue of 0.000, therefore suggesting that compared to being single, being married is associated with more monthly income. The high income could be explained by the added social and monetary capital partners provide (de Jongh, 2017). In this regard, having access to additional networks increases the likelihood for individuals in accessing better employment opportunities.
Schooling also seems to play a notable role in the earning potential of young people. In the model, tertiary education was used as the reference point. The coefficients for no schooling, secondary and primary education were negative suggesting that compared to those with tertiary education, having lower levels of education is associated with lower levels of monthly earnings. At the end of the earnings spectrum, those without any form of schooling earn R9 213.05 less than those with tertiary education. This is followed by those with primary schooling who, on average, earn R8 423.91 less and secondary schooling with R7 349.37 less salary income a month. These findings are statistically significant at the 0.01 level of significance. Moleke (2010) argues that as the economy changes and industries grow, there is often a considerable impact on the demand for people with higher levels of education. The lower earnings and less stable employment for those with less than tertiary schooling are also influenced by the fact that many of these young people are from previously disadvantaged schools (Branson et al., 2019). Similar findings were also reported in Zimbabwe (Zhou, 2002) and India (Ahmed, 2016).
In as far as the race category is concerned, black/African was used as the reference point. Table 4 shows that white youth earn R7 712.24 more than black youth. The same is true for Asian/Indian youth who earn R4 119.63 more than their black counterparts. The p-value of 0.00 is statistically significant at the 0.01 level of significance. Race in this regard is a significant predictor of the earnings potential of young employees in South Africa. Even though the coefficient for coloured youth was positive, this was a non-significant finding. These findings are in line with those of Branson et al. (2019).
There is a positive relationship between earnings and youth employed in both the formal and informal sector. Results from this point of view suggest that, on average, young people in the sectors earn more than those working in the private household sector (the reference category). The p-values are statistically significant at the 0.01 and 0.05 level of significance, respectively. These findings are in line with those of Bargain and Kwenda (2011). In addition to these results, considering the geographical location, an earnings penalty is observed for those working in the rural provinces of South Africa, who on average earn R1 611.04 less compared to youth based in the urban provinces. Considering the relatively large amount of the sample situated in these areas, the findings support the reiterations of Graham and Mlatsheni (2015). From this point of view, being situated far from urbanised areas increases material costs of job search, consequently reducing the likelihood of finding decent employment opportunities.
Finally, results from Table 4 show a positive coefficient for trade union membership, which suggests that youth belonging to a worker's union in the sample earn R2 405.78 more compared to those who do not belong to one. This positive relationship is statistically significant at a 1% level of significance. It can, therefore, be concluded that union membership is a significant determinant of monthly earnings for young employees. A significant earnings effect from unionisation is also found by Kingdon et al. (2006). Nevertheless, the dynamic between unionisation and youth labour market success is somewhat complex. Demographic characteristics reported in Table 2 showed that only 21.5 per cent of the sample indicated to be unionised. Isaac (2018) likewise found similar patterns in Australia and according to these findings, the global decline in union membership has resulted in a fall in the bargaining power of workers, which, in turn, has contributed to low wages growth in recent years.

CONCLUSIONS AND RECOMMENDATIONS
The employment outcomes of South Africa's youth workforce have undoubtedly become one of the leading concerns facing economic role players, policymakers and national authorities alike. While a large contingent of young work-seekers is finding it increasingly difficult to secure employment, concerns surrounding the conditions of those who eventually find a job have likewise gained significant consideration. In light of this, the primary objective of the study was to investigate the antecedents of young employees' earning potential in South Africa. The findings in this regard carry significant implications in broadening the understanding of the complexity of the dynamics surrounding the youth labour market. Earnings and consequently access to decent employment opportunities in this regard are not only dependent on the level of education and work experience, but are likewise impacted by various structural factors. Among these, various geographical and job search barriers, skills mismatches as well as the lack of access to quality social capital are evident. Furthermore, even after two decades after the country's transition to democracy, racial and gender biases still seem to limit the cohorts' labour market outcomes.
Recommendations in advancing the outcomes of young people in the labour market must, therefore, include enhanced public-private partnership formation through the implementation of mandatory apprenticeships. Moreover, workintegrated learning (WIL) programmes must be proliferated from as early as secondary education curricula. Not only will these tactics contribute to the access of sustainable jobs, but will likewise enhance the cohorts' capacity to operate in the formal economy. In addition to these strategies, policymakers should aim to facilitate the relaxation of labour market regulations and the adoption of a strategic focus that seeks to advance both soft and technical skills, which all prove crucial for positive employment outcomes. While the study adds to the body of knowledge on the nature of the labour market specifically for young employees, it is not without its limitations. These primarily revolve around the cross-sectional nature of the inquiry as well as limited insight into specific local conditions. Future studies can, therefore, adopt more longitudinal designs as well as the comparison of different local regions in identifying the specific factors that affect young employees earning potential.