Definitions and concepts

Unemployment is a phenomenon of job seeking arising out of joblessness. An unemployed person is one who has attained the minimum age of employment (e.g. 15years), and in a reference week was ‘without work’, ‘available for work’ and ‘actively seeking work’ (ILO, 1982). This definition implies that a jobless individual who is available for work but fails to make the effort to seek work for various reasons cannot be classified as unemployed and can only be referred to as a discouraged worker. In addition, a full-time employed person who decides to seek (additional) work with another employer can only be termed as moonlighting (Reference Baah-Boateng, Adjei and OduroBaah-Boateng et al., 2013). The definition of unemployment based on job seeking, however, has come under scrutiny particularly in relation to its applicability to Africa. Indeed, it is increasingly seen as inadequate to characterise low-income countries’ labour markets (Reference Cling, Gubert and NordmanCling et al., 2007; Reference Fares, Claudio and PeterFares et al., 2006; World Bank, 2006). In Africa, a considerable number of jobless people may be available for work but fail to look for work for various reasons including the perception of no job or, in some cases, seasonality of jobs, creating a high discouraged worker effect (Reference Baah-BoatengBaah-Boateng, 2015).

In times of recession and associated unemployment problems, youth as a demographic group tends to be hugely affected. Reference CurtainCurtain (2001) refers to youth as the demographic group who are at a separate stage in the life cycle between childhood and adulthood. In demographic analysis, the Commonwealth defines youth to cover 15–29years compared with the African Youth Charter’s classification of youth as those aged 18–35years (African Union (AU) Commission, 2006). In Ghana, Kenya and Tanzania, the African Union (AU) definition of 15–35years for youth is adopted for policy purposes compared with 12–30years in Nigeria, while South Africa’s National Youth Policy puts the youth into 14–35years category (see Reference MkandawireMkandawire, 2000).

Theoretical perspectives on unemployment

The economic literature explains unemployment from different theoretical perspectives. Neoclassical economists view unemployment as voluntary because the labour market is deemed to be always in equilibrium based on the assumption of flexible wages and perfect market information. An example of voluntary unemployment within the neoclassical framework is structural unemployment created by a mismatch between demand for labour and the skills of jobseekers, made obsolete by slow response to technological advancements. Frictional unemployment is seen as being caused by lack of labour market information, increasing the time that jobseekers take to locate and move into a new job and the time and resources that employers take to identify and recruit suitable workers to fill vacancies – failure to obtain the relevant information too is seen as having a voluntary component (Reference PhelpsPhelps, 1970; Reference StiglerStigler, 1962).

Any intervention in the labour market that distorts wage flexibility causes involuntary unemployment. Involuntary unemployment arises if individuals cannot obtain work even if they are prepared to accept lower real wages than similar qualified workers who are currently employed (Reference ShackletonShackleton, 1985). At the microeconomic level, an enactment of minimum wage legislation that forces firms to pay wages above the ‘market-clearing wage’ is seen in neoclassical theory as having implications for involuntary unemployment, even though some argue that minimum wages have relatively minor employment and unemployment effects (Reference Neumark and WascherNeumark and Wascher, 1995). Involuntary unemployment is also seen as arising from the actions of firms to pay higher wages above the equilibrium wage as an incentive to increase the efficiency of employees to avert workers’ shirking behaviour, reduce labour turnover, avoid adverse selection of job applicants or as a gift of exchange for high productivity from workers under the efficiency wage model (Reference AkerlofAkerlof, 1970, Reference Akerlof1982; Reference SalopSalop, 1979; Reference Shapiro and StiglitzShapiro and Stiglitz, 1984; Reference StiglitzStiglitz, 1974). The phenomenon can also emanate from the behaviour of employees of firms to resist competition from jobseekers to prevent underbidding of their wages through lack of cooperation or harassment of newly engaged workers within the insider–outsider framework (see Reference Lindbeck and SnowerLindbeck and Snower, 1988). Thus, unemployment is thought, within this framework, to arise if firms take into account only the interests of the employees (insiders) without regard to the interest of jobseekers (outsiders) in the firms’ wage determination process (Reference Bentolila, Juan and JimenoBentolila et al., 2011).

From a macroeconomic perspective under the Keynesian framework, by contrast, unemployment could occur as a result of deficiencies in aggregate demand over certain periods in the business cycle such that the jobs created are not enough for everyone who wants to work (Reference KeynesKeynes, 1936). Contrary to the neoclassical argument that attributes involuntary unemployment to institutional factors, this type of demand-deficient unemployment is involuntary since the unemployed are constrained by limited job availability. A further factor, beyond individual control, is seasonal unemployment caused by predictable seasonal variation in demand – additionally, climatic conditions also create involuntary unemployment.

These theories of unemployment were formulated in more developed economies with wage-employment dominated labour markets and clear employer–employee relationships. They have only limited relevance in Africa, dominated by self-employment. The efficiency wage theory, insider–outsider model and labour market intervention through minimum wage imposition are applicable to formal wage employment, which accounts for quite a small proportion of employment in many African countries. It is not a market-clearing wage, but a high degree of informality, dominated by vulnerable employment comprising own account and contributing family jobs, that accounts for the apparently low unemployment rates in many African countries (Reference Baah-BoatengBaah-Boateng, 2015). Countries such as South Africa, Namibia, Tunisia, Algeria and Morocco that have high levels of wage employment and bigger formal sectors more closely resembling developed labour market structures have to combat high rate of unemployment. Thus, while these theories may be applicable to the formal sector dominated labour markets in the developed world, they can only be used to explain joblessness in the formal sector, which constitutes a small segment of labour markets in Africa.

Drivers of youth unemployment: The literature

Until recently, empirical research on youth unemployment in Africa was somewhat scanty, owing to the lack of regular availability of labourical literature on the subject of unemployment has generally established a relatively higher unemployment rate among youth than adults (see, for example, AfDB et al., 2012; Reference Baah-BoatengBaah-Boateng, 2013, Reference Baah-Boateng2015; Reference Johnson, Layard, Ashenfelter and LayardJohnson and Layard, 1993; Reference Sackey and OseiSackey and Osei, 2006; United Nations Economic Commission for Africa (UNECA), 2005). In 2009, South Africa’s youth unemployment rate was 48% compared to 19% for adults, while the youth unemployment rate in Egypt was more than six times higher than the rate for adults in 2007 (AfDB et al., 2012). Using age and age squared in estimating unemployment, Reference Blackaby, Leslie and MurphyBlackaby et al. (1999) show how this may reflect job shopping by younger workers and the lack of job opportunities as workers approach retirement age.

An assessment of the literature on the drivers of youth unemployment can be undertaken from both microeconomic and macroeconomic perspectives. At the microeconomic level, the demographic characteristics of youth including sex, race, education and skills and individual location have implications for their chances of securing a job. It is widely acknowledged that youth unemployment rates are higher in urban than in rural areas, while gender differences in the youth unemployment rate are mixed (AfDB et al., 2012; Reference Baah-BoatengBaah-Boateng, 2014). Human capital endowment measured by education, skills and work experience is said to be the key determinant of the labour market success of individuals – better educated individuals with more professional experience in the world of work are more likely to be successful in finding jobs (Reference BeckerBecker, 1975; Reference MincerMincer, 1974). However, some studies have observed higher unemployment rates among individuals with at least a secondary school education, relative to the less educated, in developing countries (e.g. AfDB et al., 2012; Reference Aryeetey, Baah-Boateng, Ackah, Hino and RanisAryeetey et al., 2014; Reference Baah-BoatengBaah-Boateng, 2013, Reference Baah-Boateng2014, Reference Baah-Boateng2015; Reference Dickens and LangDickens and Lang, 1995). These authors attribute the phenomenon to the large informal sector’s unattractiveness to the educated. This coupled with the slow growth of the formal sector leaves them with virtually no option than to battle with unemployment. The role of individual’s family background has also been observed as a determinant of youth unemployment. Essentially, the education status or occupation of family members in the labour market can affect an individual’s employment prospects, as a result of the professional networks that may arise from the jobs and acquaintances of family members (Reference Verhaeghe, Li and Van de PutteVaerhaeghe et al., 2012; Reference Zhang and ZhaoZhang and Zhao, 2011).

From the macroeconomic angle, youth unemployment can be explained from the overall characteristics of the labour market in terms of the interaction between demand for and supply of youth labour. The direction of the effect of aggregate demand deficiency on youth unemployment is similar to its effect on overall unemployment, even though the degree of impact differs. Indeed, as a result of the limited labour market endowment such as lower skills and work experience, limited networking and the concentration of precarious employment contracts among the young, youth become more vulnerable to layoffs in times of economic recession (Reference Baah-BoatengBaah-Boateng, 2013; Reference Bell and BlanchflowerBell and Blanchflower, 2011). The tendency for most advertised jobs to require substantial work experience implies that most young people are locked up in an ‘experience trap’ through limited opportunities for work experience. Indeed, young people are highly vulnerable to aggregate demand fluctuations (Reference Choudhry, Marelli and SignorelliChoudhry et al., 2012) suggesting the importance of demand-side causes of youth unemployment.

Empirically, Reference Aryeetey, Baah-Boateng, Ackah, Hino and RanisAryeetey et al. (2014) have established a strong demand-deficient effect of youth unemployment based on a model that shows that in times of economic recession, there is a higher probability of unemployment among young jobseekers who hold strong preferences for specific jobs, instead of looking for any job, and among jobseekers pursuing full-time jobs as against part-time work. Using macroeconomic variables, Reference AnyanwuAnyanwu (2013) established a significantly positive effect of domestic investment and real gross domestic product (GDP) on youth employment in Africa, confirming the importance of demand factors on youth unemployment. However, he found negative and significant effects of government consumption and real GDP per capita on youth employment. He thus argues for the promotion of government expenditure effectiveness and reform of the fiscal systems by all levels of government to address youth employment challenges.

From the supply side of the macroeconomic argument, the rising volume of the inflow of young people into the labour market relative to the number of jobs available is a major source of youth unemployment (Reference Perugini and SignorelliPerugini and Signorelli, 2010). The ‘youth bulge’ explanation of youth unemployment has been stretched to account for the incidence of civil war in Africa: Reference Collier, Hoeffler and RohnerCollier et al. (2009) argue the urgent need to address the impact of the high and increasing youth population as a means to avert civil wars in Africa. There is however no clear consensus as to whether the ‘youth bulge’ argument could be isolated as a driver of youth unemployment without looking at aggregate demand-side issues. Some studies (e.g. Reference Aryeetey, Baah-Boateng, Ackah, Hino and RanisAryeetey et al., 2014; Reference O’HigginsO’Higgins, 2012) found youth unemployment to be strongly linked to deficiency in aggregate demand. In their study, Reference Korenman and NeumarkKorenman and Neumark (1997) affirm that the size of the youth population has no statistically significant effect on youth unemployment when controlling for demand factors.

Wage levels determined by the demand and supply interaction also have implications for youth unemployment. Increase in wages from institutional perspectives through minimum wage legislation, trade unions’ high-wage advocacy or firms’ own actions (efficiency wage and insider–outsider model) have the tendency of causing rising youth unemployment. The effect of high wages on firms’ cost of production and profits could influence their decision to reduce the employment of young people relative to their adult counterparts if the skills of the two groups are considered to be substitutes. Wage expectations of young people in their job-seeking decisions also affect their success in securing jobs. Thus, according to Reference Aryeetey, Baah-Boateng, Ackah, Hino and RanisAryeetey et al. (2014) and Reference Baah-BoatengBaah-Boateng (2013), a high reservation wage has the effect of increasing youth and overall unemployment in Ghana.

Model specification and data sources

In this analysis, the ILO definition of unemployment based on job seeking out of joblessness and readiness to work is adopted, and the target group is the youth aged 15–24years based on the United Nations (UN) definition. Different countries and organisations define youth in different ways. The adoption of the UN definition of 15–24years for the study seeks to ensure consistency and uniformity across countries against the backdrop of varied definition and measurement of youth as demographic group.

Unemployment is a consequence of disparities between demand for, and supply of, labour. From a Keynesian economics point of view, unemployment arises from aggregate demand deficiencies over certain periods in the business cycle, creating a shortage of jobs for an increasing number of jobseekers (Reference KeynesKeynes, 1936). Several empirical studies have identified limited job creation opportunities through slow growth of the employment-friendly sectors of agriculture and manufacturing (Reference Aryeetey, Baah-Boateng, Ackah, Hino and RanisAryeetey et al., 2014; Reference Baah-BoatengBaah-Boateng, 2013). On the supply side, rapid growth of the population and low quality of education relative to skills requirements in the labour market also help explain the employment generation challenges of many African countries (AfDB et al., 2012; Reference Sackey and OseiSackey and Osei, 2006).

Following Reference Osberg, Apostle and ClairmontOsberg et al. (1986), we specify the incidence of youth unemployment from demand and supply perspectives as

(1) ${\mathrm{YUR}}_i=f\left(S_i\right)+g\left(D_j\right)$

where YUR _i is the youth unemployment rate, S_i refers to the incentives faced by young people to become or remain employed and D_j measures the incentives to firms to employ young people. In a panel framework involving a sample of African countries over a period of time and accounting for labour market conditions and other controls, youth unemployment in Africa is formulated as

(2) ${\mathrm{YUR}}_{it}=S_{it}{\beta }_k+D_{it}{\delta }_k+X_{it}{\theta }_k+U_i+{\epsilon }_{it}$

where YUR is youth unemployment rate, S denotes a vector of supply factors, D represents a vector of demand factors while X is a vector of labour market–related and other controls, U is a vector of time invariant variables that vary only over countries and ε is random disturbance term. The subscript i represents countries and t denotes time of 3-year average (i.e. 2000–2002; 2003–2005; 2006–2008; 2009–2011).

The supply-side variables in the model are youth as a share of total population to capture the ‘youth bulge’ effect on youth unemployment and gross enrolment rate as a proxy for education. The growth rate of real GDP and the share of agriculture and manufacturing in GDP (as a measure of the job creation effect of growth) are used to explain the demand side of youth unemployment. Labour market–related variables such as employment-to-population ratio are used to account for a country’s job creation ability (see ILO, 2009), and the vulnerable employment rate is used to link the degree of informality to the incidence of youth unemployment (Reference Baah-BoatengBaah-Boateng, 2015). The effect of the reservation wage on youth unemployment could not be addressed in the study due to the non-availability of data on wages across countries on the continent. Other control variables include degree of urbanisation to measure urban–rural differences of youth unemployment, female–male ratio of youth labour force participation rate to capture the effect of gender, the country’s total population as a proxy for country size and country’s per capita GDP to account for country’s level of development. The model for estimation is specified as

(3) $\begin{array}{c}{\mathrm{YUR}}_{it}={\beta }_0+{\beta }_1{\mathrm{yp}}_{it}+{\beta }_2{\mathrm{ed}}_{it}+{\beta }_3{\mathrm{gr}}_{it}+{\beta }_4{\mathrm{qgr}}_{it}+{\beta }_5{\mathrm{emp}}_{it}+\\ {\beta }_5{\mathrm{vul}}_{it}+{\beta }_6{\mathrm{urb}}_{it}+{\beta }_7{\mathrm{fem}}_{it}+{\beta }_8{\mathrm{size}}_{it}+{\beta }_9{\mathrm{dev}}_{it}+U_i+{\epsilon }_{it}\end{array}$

where the description of the variables in the model is as reported in Table 3. The apparent problem of endogeneity between youth and employment-to-population ratio is avoided, because YUR is calculated as the share of the labour force aged 15-24, whereas emp is the ratio of total employment-to-working age population aged 15+. Data for estimation are drawn from two main sources – KILM, 8th edition (ILO, 2014b), and WDI, World Bank (2015). Data on youth unemployment, youth population, employment-to-population ratio, vulnerable employment, female–male labour force participation rate and total population were sourced from KILM, while data on youth education, GDP growth rates, quality of growth, urban population and per capita GDP were derived from WDI

Table 3. Description of variables and expected signs of parameters.

GDP: gross domestic product.

Descriptive statistics for the variables in the estimated models are shown in Appendix 1 Table 5 and indicate that the panel identifier, country, does not vary within the panels or it is time invariant. A total of 41 of 54 African countries where data for all the variables could be obtained were covered in the analysis and are listed in Appendix 1 Table 6.

Estimation strategy and analysis of results

The estimation process begins with a test of the appropriate estimation technique (fixed effect vs. random effect). The fixed effect approach is necessary in analysing the effect of variables that vary over time by controlling for all time-invariant differences between countries. In contrast, the rationale behind the random effect technique is that the variation across countries is assumed to be random and uncorrelated with the regressors included in the model. In choosing the suitable method of estimation, the Hausman test for fixed versus random effect was carried out and the results (Appendix 1 Table 7) indicate that a random effect technique is appropriate in estimating equation (3) based on probability greater than that represented by the chi-square value of 0.8034. Mindful of the potential multicollinearity problem with the inclusion of per capita GDP to proxy level of development and vulnerable employment rate (which also largely reflects the level of a country’s labour market) were introduced separately in the estimation but no significant change was noticed.

Table 4 presents the estimation results for a youth unemployment panel regression model based on the random effect approach. The result of a Breusch and Pagan Lagrangian multiplier test of random effect (see Appendix 1 Table 8) confirms the appropriateness of the random effect estimation technique. In all, the model performed quite well with overall R ² of about 0.796 to suggest that about 80% of variation in youth unemployment rate was related to the regressors included in the model. The Wald chi-square and associated probability value imply that the regressors in the model significantly and jointly explain youth unemployment in Africa.

Table 4. Random-effects GLS regression of 41 countries over 3-year average period of 2000–2011: Dependent Variable: Youth unemployment rate.

GLS: generalized least squares; GDP: gross domestic product.

*** p<0.01, **0.01<p<0.05, *0.05<p<0.10.

On the supply side, the youth population share was found to significantly explain youth unemployment in Africa. A 1% point increase in the share of youth population in total population implied a 0.72% point rise in the youth unemployment rate, indicating that the ‘youth bulge’ is a major driver of youth unemployment in Africa. The study, however, found no significant effect of education on youth unemployment in Africa.

On the demand side, the quality of economic growth, measured by the share of agriculture and manufacturing value added in total national output, was found to have a significantly negative effect on youth unemployment, while real GDP growth showed no significant effect on youth unemployment. Thus, an average increase by 1% point in the share of agriculture and manufacturing in total national output by 1% point was found to have a reducing effect of 0.12% points on the youth unemployment rate. This may explain why youth unemployment remains a challenge on the continent against the backdrop of strong growth performance. The recent remarkable growth in Africa has been driven largely by extraction of natural resources such as oil, gold and diamonds (among others), which are known to have weak employment-generating effect. Indeed, high growth driven by mining and oil extraction in countries like Nigeria, Gabon, Angola, Ghana, Botswana, Zambia and so on at the expense of agriculture and manufacturing does not promote the creation of sufficient job opportunities to meet the increasing youth population. Reference Baah-BoatengBaah-Boateng (2013) argues that the high economic growth driven largely by low labour absorption economic activities of mining and oil extraction in particular as against the slow growth of high employment elasticity sectors of agriculture and manufacturing underscores the strong demand-deficient unemployment in Ghana.

As expected, the ability of countries to generate employment measured by employment-to-population ratio contributes significantly to a decline in youth unemployment. Thus, increasing employment-to-population ratio does not only imply declining inactivity in the labour market but also has the effect of reducing youth unemployment and vice versa. In addition, however, the structure of the labour market has implications for youth unemployment in Africa such that a high rate of vulnerable employment has also contributes to a decrease in youth unemployment and vice versa. Thus, a 1-percentage point higher vulnerable employment rate suggests a 0.25 percentage point lower youth unemployment rate (Table 4). Indeed, in countries where the structure of the labour market provides a platform for informality (dominated by vulnerable employment) to flourish with no social benefit for the unemployed, youth tend to seek refuge in the informal sector for survival rather than remaining unemployed. On the other hand, in a well-structured and regulated labour market with some social protection programmes for the unemployed and limited or no avenues for informality to develop, youth have no choice but to register as unemployed to enjoy social benefits, resulting in reported high open unemployment (AfDB, 2012; Reference Baah-BoatengBaah-Boateng, 2015). Indeed, in countries such as Benin, Burkina Faso, Malawi and Rwanda where informality is estimated to be over 90%, unemployment rates are very low at less than 5%, while in South Africa, Namibia, Botswana, Tunisia and Algeria, where informal sector constitutes less than 25%, youth unemployment rates are high at over 25%.

The empirical results also contribute to the evidence for higher urban youth unemployment rates relative to rural rates (Reference AnyanwuAnyanwu, 2013; Reference Baah-BoatengBaah-Boateng, 2013; Reference Dickens and LangDickens and Lang, 1995; Reference Sackey and OseiSackey and Osei, 2006). Thus, the significantly positive coefficient of the urban dummy variable suggests that a 1-percentage point rise in the degree of urbanisation measured by urban share in total population has a 0.2percentage point increasing effect on youth unemployment rates. The attraction of urban life vis-à-vis low income from rural agriculture draws many people – particularly youth – into the cities in search for better-paid jobs that are however often not readily available. This phenomenon helps account for the higher incidence of youth unemployment in urban than rural areas (Reference Baah-BoatengBaah-Boateng, 2013).

The gender dimension of youth unemployment in Africa appears to suggest stronger labour market challenges for young females than young males. The ratio of female-to-male labour force participation rates indicates that greater participation of females relative to males has an increasing effect on youth unemployment in Africa. People participate in the labour market as either employed or unemployed, and this observation implies that a higher proportion of young females than young males participate in the labour market as unemployed. The size of a country is observed to have a decreasing effect on youth unemployment even though the statistical significance of the coefficient is weak. The level of a country’s development measured by per capita real GDP is observed to have no statistically significant effect on youth unemployment in Africa, contrary to findings by Reference AnyanwuAnyanwu (2013) that per capita real GDP has negative and statistically significant effects on youth employment, indicating a positive effect of per capita GDP on youth unemployment in Africa.