Cohort Size and Unemployment Rate: New Insights from Nigeria

Abstract

The debate on the nature of the relationship between cohort size and unemployment rate has been widely studied and generated a substantial body of literature in labor economics discourse. However, an in-depth reading of this literature suggests that, besides the fact that findings are mixed and do not provide conclusive evidences, one hardly ever comes across studies exclusively on African countries. Likewise, generalized studies across countries employing pooled data seem to dominate the literature. In light of these, the current study examines the nature of the said relationship, over the period 1970–2019, in Nigeria in a multivariate and dynamic framework. Employing Bounds testing procedure, the article finds that both the short-run and long-run impacts of cohort size on overall unemployment rate are positive and statistically significant. This suggests that aggregate unemployment rate tends to be higher when many young people supply labor. In view of these findings, the article recommends that government should collaborate with private sector to develop and implement functional microcredit schemes. Such schemes should be flexibly structured to avert institutional bottlenecks and enhance accountability and transparency in their management.

Keywords

Autoregressive distributed lag (ARDL)Shimer hypothesis cohort-crowding hypothesis rate of unemployment

Introduction

Over the last five decades, the nature of the relationship between the rate of unemployment and the size of a specifically defined age group has been intensely debated, extensively acknowledged, and discussed in development and labor economics discourse. On the theoretical front, there are two distinct schools of thoughts to this debate, namely the cohort-crowding hypothesis of Easterlin (1968) and hypothesis of Shimer (Garloff et al., 2013). On the one hand, the cohort-crowding theory, whose notion was grounded in the career phase model put forward by Welch (1979), maintains that, first, an increase in the proportion of a specifically defined age group, particularly the size of young persons in the working-age population, has repercussions on the employment outcomes of that particular group (Moffat & Roth, 2017; Roth, 2017). Second, since the aggregate unemployment rate is weighted average (Shimer, 2001) of the rates of various demographic groups, an increase in the percentage of those groups with above average unemployment rates will increase the aggregate unemployment rate (Reid & Smith, 1981). On the other hand, Shimer’s exposition, whose standpoint can be traced to the search and matching theory of unemployment advanced by Mortensen and Pissarides (1994), postulates that aggregate rate of unemployment tends to be lower (Ochsen, 2011) when many young people supply labor.

A considerable number of studies have assessed the macroeconomic and empirical implications of the two propositions. While some studies (Barwell, 2000; Biagi & Lucifora, 2008; Bloom et al., 1988; Flaim, 1990; Foote, 2007; Fuchs & Weyh, 2014; Garloff et al., 2013; Korenman & Neumark, 2000; Leveson, 1980; Moffat & Roth, 2017; Perry, 1970) found evidences consistent with the cohort-crowding hypothesis of Easterlin which underscored the potential negative relationship between cohort and its labor market prospects, in contrast, others (Abraham & Shimer, 2001; Freeman, 1979; Lam, 2014; Ochsen, 2009; Roth, 2017; Shimer, 2001; Skans, 2005) have also substantiated and lent credence to Shimer’s exposition which underlined the virtues that an upsurge in the young workers in the working-age population lessens not only the youth rate of unemployment but also the overall rate of unemployment. Likewise, a sizable number of studies (Nardone, 1987; Newhouse & Wolff, 2014; Ochsen, 2011; Organization for Economic Cooperation and Development [OECD], 1980; Russell, 1982; Zimmermann, 1991) found no clear relationship between cohort size and rate of unemployment. Basically, in accord with the theoretical discrepancies, studies have failed to suggest an overall dominance of one notion over the other.

Nearly all countries in the world have undergone or are currently undergoing growth in the proportion of youth labor force (aged 15–24 years), occasioned by different patterns in fertility, mortality, and migration. Worldwide, as of 2015, this subset of population, estimated at 1.2 billion, accounted for one out of every six persons globally (United Nations Department of Economic and Social Affairs [UNDESA], Population Division, 2015). By 2030, it is projected to have grown by 7% to nearly 1.3 billion. Albeit, as initially witnessed by developed countries, growth in the proportion of youth labor force is at present a worldwide phenomenon experienced in nearly all countries of the world. In the Oceania and Northern America, Caribbean and Latin America, and Europe where fertility rates have declined for decades, this subsection of population has stabilized in size and is anticipated to change little in the coming decades (UNDESA, 2015). In Asia, after rapid and sustained increase all through the latter half of the twentieth century, it is expected to decline from 718 million in 2015 to 711 million and 619 million in 2030 and 2060, respectively. Contrary to other regions of the world where the size of youth labor force has peaked, Africa’s young population has continued to grow rapidly. As noted by the 2017 revision of the United Nations World Population Prospects, this age group, numbering 226 million, as of 2015, constituted about two-fifths of Africa’s labor force and accounted for 19% of the global youth population. By 2030, it is anticipated to increase by 42% and continue to grow through remainder of the twenty-first century, more than double from current levels by 2055 (UNDESA, 2015).

In sub-Saharan Africa (SSA hereafter) alone, according to the Berlin Institute for Population and Development, this subset of population will grow by 225 million people by the middle of the century (Sippel et al., 2011), making SSA the “youngest” region in the world. Remarkably, about one-quarter of the youth surplus of SSA by 2050 will be accounted for by Nigeria alone (Sippel et al., 2011). In fact, by 2030, as reported by the Next Generation Nigeria Task Force convened by the British Council, the country will be one of the few economies in the world with a growing supply of young workers (Bloom, 2010). As a matter of fact, in the coming decades, while an upward trend will be observed in the global scarcity of young workers, Nigeria will remain a young country in most of the twenty-first century, with a median age of 21. With the current population of 200 million as of 2019, ranking 7th in the world, this age cohort, having risen from 35 million to 37 million between 2017 and 2018, is anticipated to grow to 51 million and 86 million by 2030 and 2050, respectively (Lam & Leibbrandt, 2013; Sippel et al., 2011). In this regard, some critical questions arise: What is the nature of the relationship between the size of this specific population age group and its labor markets prospects in Nigeria? Put differently, has increase in the share of young workers in the working-age population affected the aggregate rate of unemployment in Nigeria.

Although, as evidenced in the bulk of the studies cited above, the question of whether cohort size affects the labor markets prospects of its members (Moffat & Roth, 2017) has been extensively studied and generated a substantial body of literature. Nevertheless, an exhaustive reading of this literature suggests that, apart from the ambivalent nature of these findings, hardly any empirical studies (Odedokun & Round, 2001) have been reported solely for African countries, partly because of limited number of household data points for African countries. Also, the attention of these literatures has been heavily biased toward regional or cross-section/-country (sometimes, panel) studies and often failed to use long period data. In view of these, to provide an intuitive insight into the nature of the aforementioned relationship and related policy issues, and likewise shed more light on the contradiction-prone evidence, country-specific studies are needed.

Using annual time series data, for the period 1970–2019, this study employs autoregressive distributed lag (ARDL) bounds testing procedure to examine the subject in the context of Nigerian economy. The use of Nigerian data is supported not only by the high level of unemployment and poverty in Nigeria, but also because of the country’s recent renaissance to lessen unemployment as well as poverty, which makes country-specific evidence of immediate relevance to anti-unemployment/poverty policies. Albeit there is sizable literature assessing the macroeconomic implications of this specifically defined age group in terms of growth opportunities it presents, and the question of whether the size of this cohort affects the labor markets of its member has so far been left largely unaddressed. This current study, thus, fills this gap. Following the introduction, the remainder of the study is set out as follows. First, a brief review is presented of theoretical and empirical evidence. The next section focuses on methodology and data. This is followed by estimation techniques and empirical analysis. Findings from the analysis are summarized and are followed by conclusions and recommendations.

Literature Review

The nature of relationship between rate of unemployment and cohort size has been a subject of intense debate in development and labor economics discourse. Theoretically, there are two distinct notions to this debate. At one end of this argument are those who maintain that increase in proportion of young people in working-age population increases overall rate of unemployment, since unemployment rate is usually higher for younger workers (Ochsen, 2011). Essentially, according to the proponents of this view, the size of cohort is inversely related to its labor market opportunities (Garloff et al., 2013). At the other end of the debate are those who emphasize that a young labor force suggests a more flexible labor market (Kochar, 2007), with a larger percentage of the labor force willing to accept a new job. As such, firms will find areas with greater concentration of young workers more conducive to the opening of new firms (Kochar, 2007) and therefore will focus in these areas, generating additional jobs and reducing unemployment rates for older and young workers (Kochar, 2007). Basically, contrary to the first notion, the labor market entry of large cohort entails (Garloff et al., 2013) an increase in employment and a decrease in unemployment, respectively, for older and young workers. Numerous studies have examined the empirical implications of these propositions.

In his path-breaking work, The American Baby Boom in Historical Perspective, published in 1961, Easterlin posits that labor market fortunes of workers are negatively related to the size of their cohorts. Perry (1970) notes that an upsurge in the proportion of young people in working-age population increases overall rate of unemployment (Ochsen, 2011). Wachter et al. (1976) examine the effect of unemployment rates among prime-age males (25–54 years of age) and the proportion of individuals aged 16–24 (relative to the population of working age) on unemployment rates of female and male teenagers. Every result reveals a significant and robust positive relationship between cohort size and unemployment rate. Using US data, Freeman (1979) examines the effect of generational crowding on the labor market for young male workers. Results indicate that the foremost effect is on wages and not on employment.

Employing US data, Leveson (1980) regresses cohort size, minimum-wage, and adult unemployment rate variables on teenage unemployment rates, over the post-World War II period. In all the unemployment rate equations fitted separately to data for different sex/race groups, for the periods 1947–1979 and 1954–1979, the coefficient of cohort size is statistically significant and positive. Organization for Economic Cooperation and Development (OECD) (1980) assesses the effects of adult unemployment rates, cohort size, and a linear time trend on unemployment rates of both female and male teenagers at different ages for Canada, France, Australia, Finland, Italy, Germany, Japan, the USA, Sweden, and the UK. Results reveal that while cohort size has statistically insignificant and positive effects for Australia, France, Finland, and Sweden; in contrast, it has statistically significant and positive effects for Germany, Japan, the USA, the UK, Italy, and Canada.

Russell (1982) fits unemployment rates of age groups 20–24, 18–19, and 16–17 on a time trend, business cycle indicator, and the proportion of working-age population in the age group 16–24. Separate regressions were run for females and males over the period 1947–1980. Results indicate a strong positive relationship between unemployment and cohort when the business cycle indicator is controlled for. Focusing on the US labor market, Anderson (1982) estimates a number of disaggregated economic/demographic models. Results show that, relative to the unemployment rate of 25–54 year-old men, unemployment rate of young cohorts of both female and male is statistically significant and positively associated with cohort size variable. Using data from Israel, Ben-Porath (1985) investigates the effect of cohort size on unemployment and earnings and finds that it has positive and significant effect on employment rates.

Focusing on a number of industrialized countries (Canada, Australia, France, Sweden, Japan, the US and the UK), Bloom et al. (1988) differentiate between two alternative notions of the labor markets faced by young workers and find that cohort size tends to have a marked trade-off between relative employment effect and relative earnings effect, with cohort size reducing relative employment in some countries and relative earnings in others. Employing cointegration techniques to differentiate short- and long-run developments, Zimmermann (1992) investigates the effects of relative cohort age and relative cohort size on unemployment. While there is no sufficient evidence in the long run, in the short run, there is positive impact of relative cohort age and relative cohort size on unemployment. Using a panel data set for 15 countries, Korenman and Neumark (2000) isolate the effects of exogenous change in potential youth labor supply on youth unemployment and employment rates. Contrary to the conventional assumptions about demographic effects on labor markets, Shimer (2001) finds that unemployment rates and state‐level youth shares are negatively correlated. Foote (2007) updates Shimer’s analysis and shows that the astonishing correlation basically disappears when the end of the sample period is extended from 1996 to 2005.

Using a panel of OECD countries, Rodriguez-Palenzuela and Jimeno-Serrano (2002) regress the relevance of labor market institutions, macroeconomic shocks, and the relative size of youth population on unemployment rates and find that variations in size of youth populations occasioned by baby boom of 1950s as well as 1960s and the resulting decline of fertility are positively related to unemployment rates. Nordström Skans (2002) investigates the overall labor market performance effects of changes in age structure. Using Shimer’s (2001) method, the study finds that young workers benefit from being part of a large cohort. Biagi and Lucifora (2008) examine the impacts of education and demographic on unemployment rates, for the period 1975–2002, in Europe. Empirical results reveal that changes in population age structure are positively and significantly associated with unemployment rate of young workers. Analyzing the impacts of shifts in the age group size of the unemployed on the employment rate, results from Ochsen (2009), in most of the OECD countries, show that the unemployment rates for the age–group 55–65 are lower than the unemployment rates for the age groups 25–54 and 16–24. In another study, Garloff et al. (2013) analyze the impact of cohort on unemployment/employment in Western Germany and finds that small entry cohorts lessen the overall rate of unemployment and therefore improve the situation of job seekers. Also, Newhouse and Wolff (2014) investigate the relationship between cohort size and employment outcomes of workers at different ages. Using cross-country panel (data) of 83 developing countries, the article finds that a fall in cohort size is associated with improvement in employment outcomes for youth in middle-income countries, but there is scarce evidence that the improvements continue to adulthood.

Employing micro-level and exogenous variation data for the USA, the UK, and France, Simion (2015) investigates the impact of cohort size on labor market outcomes and finds that ballooning generations spent less time in school and faced persistently high unemployment rates. Applying data from 49 European regions, for the period 2005–2012, Moffat and Roth (2017) estimate the effect of cohort size on youth unemployment and employment outcomes. The article finds a positive (negative) effect of cohort size on unemployment (employment) among individuals aged 18–22 years but finds opposite effects among other individuals. Roth (2017) evaluates the relationship between the duration of search for employment and apprenticeship graduates (who entered German labor market) over the period 1999–2012. The study finds that it takes less time for apprentices from a larger cohort to find employment.

In summary, from theoretical and empirical viewpoints, the debate on the impact of the size of one’s generation on unemployment and employment outcomes has been a subject of controversy, which is extensively studied, and generated a considerable body of economic literature from different countries and time periods. However, while the debate is still inconclusive, a cautious reading of the literature reveals that, aside from the contradictory evidences, studies assessing the aforesaid relationship within the context of African countries have received relatively little attention as the bulk of the extant studies largely focused on American, European, and Asian economies. Likewise, the focus of these studies has been heavily biased toward cross-country/cross-section/panel econometric analysis and often failed to use long period data. To give intuitive insights into the nature of the said relationship and related policy issues, country-specific studies are needed. Albeit, in Nigeria, there is a substantial body of literature (Aiwone, 2016; Ashford, 2007; Bloom et al., 2010; Canning et al., 2015; Cleland, 2012; Dramani & Mbacké, 2017; Drummond et al., 2014; Jimenez & Pate, 2017; Olaniyan et al., 2012; Omoju & Abraham, 2014; Soyibo et al., 2008) assessing the implications of this specifically defined age group in terms of economic growth opportunities it presents; however, the question of whether the size of one’s generation affects the labor market opportunities of its members has so far been left largely unaddressed. The current article thus fills this gap.

Data and Methodology

Data Sources

Annual secondary data, spanning the period 1970–2019, sourced primarily from World Development Indicators database, African Development Indicators (both maintained by the World Bank), the publications of Central Bank of Nigeria (CBN) Statistical Bulletin, and Annual Abstract by National Bureau of Statistics were utilized. The specific source for each series in presented in Table 1.

Table 1.

Data Sources and Description Variables.

Variable	Proxy	Sources
Unemployed labor force	Unemployment rate (%)	National Bureau of Statistics (2010, p. 2; 2003–2018)
Cohort size	Age groups 15–24 and 25–34 (% of population) and growth rates of age groups 15–24 and 25–34	Africa Development Indicators (2019)
Economic growth	Real GDP growth rate (Annual %)	World Development Indicators (2019)
Inflation rate	Consumer Price Index (2010 = 100)	CBN Statistical Bulletin (2019)
Foreign direct investment {FDI) net inflows	FDI Inflows {BoP, current US$}	World Development Indicators (2019)
Total external debt stocks	Nigeria’s external debt outstanding (₦' Billion)	CBN Statistical Bulletin (2019)
Government expenditure	Capital and recurrent expenditure (₦' Billion)	CBN Statistical Bulletin (2019)

Source: The author.

Note: CBN, Central of Nigeria.

Model Specification

This study takes after the works of Rosen (1972), Berger (1985), Zimmermann (1992), Schmidt (1993), Eguía and Echevarria (2004) and adopts and builds on career phase model, advanced by Welch (1979), in which workers follow a given career path independent of cohort size (Schmidt, 1993). Basically, according to this model, the easiest view of the way cohort affects unemployment rate follows from the notion that work careers comprise of a series of more or less distinct phases and at any moment in a career (Welch, 1979), a member of the profession is in transit between two of these phases (often viewed as a convex combination). For simplicity, we assume that each of these distinct phases is marked and characterized by an age segment to which the individual belongs (e.g., , aged 15–24 years designated by, $j = 1$ , aged 25–34 years by $j = 2$ , aged 35–44 years by $j = 3$ , and aged 45–64 years by $j = 4$ , for both sexes). And while there is imperfect substitutability across age segments, individuals sharing certain similar characteristics, primarily age, (though, in some studies, the level of education, gender, and/or experience) are perfectly substitutable. Besides, building on Easterlin’s (1978) and Welch’s (1979) views, we assume that each of these four age segments/phases is, as well, associated with distinct degrees of labor market experience. In essence, the older the worker, the greater is his/her experience.

Thus, as advanced by Welch (1979), as the worker commences the $j t h$ phase at age $x_{j}$ , he/she leaves the preceding phase $j - 1$ (unless when $j th$ phase is the first one) behind, and begins transition to the following phase $j + 1$ (unless when $j t h$ phase is the last one). In more specific terms, as soon as phase $j th$ is entered at age $x_{j}$ , the worker begins a transition into the phase $j + 1$ . When in $j th$ phase, the worker is assumed to assign a (declining with age $x$ ) share $p_{j}^{i} \{x\}$ of his or her time to this phase (Eguía & Echevarría, 2004); the rest of the time, in an increasing share $1 - p_{j}^{i} \{x\}$ is assigned to human capital accumulation requisite to access the next phase, $j + 1$ (Eguía & Echevarría, 2004). Mathematically, $p_{j}^{i} (x)$ is such that $\frac{\partial p_{j}^{i}}{\partial x} < 0; for x \in \{x_{j}, x_{j + 1}\}, p_{j}^{i} (x_{j}) = 1 , and p_{j}^{i} (x_{j + 1}) = 0$ . Hence, in line with these assumptions, if $m^{i} (x)$ depicts the number of active populations of sex $i$ and age $x$ , then the number of employees supplying their labor services in group $j i$ is expressed as follows:

M_{j}^{i} = \int_{x_{j - 1}}^{x_{j}} \{1 - p_{j - 1}^{i} (x)\} • m^{i} (x) d x + \int_{x_{j}}^{x_{j + 1}} \{p_{j}^{i} (x)\} • m^{i} (x) d x

(1)

where $j = 1,2,3,4$ ; $i = w, m$ depict, respectively, women and men. Note that both workers in $j age segment$ and workers in the preceding $j - 1 age segment$ offer their services to $j th$ segment. While the latter do it in $1 - p_{j}^{i} \{x\}$ , the former do it in $p_{j}^{i} \{x\}$ share.

For simplicity, we assume the aggregate production function of the economy can be expressed as follows:

Q = f (L, R)

(2)

where $L$ and $R$ depict, respectively, the total amount of labor supply, assumed to be a function of all groups labor in the economy, that is, $\{L = f (L_{1}^{M}, L_{1}^{W}, L_{2}^{M}, L_{2}^{W}, L_{3}^{M}, L_{3}^{W})\}$ and all other factors of production posits to be fixed in short-run, representing the state of the economy (Eguía & Echevarría, 2004). Following Schmidt (1993), Eguia and Echevarria (2004), we assume that for given wages ( $w_{j}$ ) for the four age segments $j = 1,2,3,4$ and for a given $R$ , firms face the problem of choosing levels of labor inputs $L_{j}^{i}$ corresponding to the eight worker groups as follows:

{Max}_{(L_{j}^{i})} f (L, R) - \sum_{j = 1}^{j = 3} w_{j} (L_{j}^{m} + L_{j}^{w})

(3)

subject to

L = f (L_{1}^{M}, L_{1}^{W}, L_{2}^{M}, L_{2}^{W}, L_{3}^{M}, L_{3}^{W})

(4)

Maximizing Equation (3) subject to Equation (4), a system of first-order necessary condition (FOC) is obtained. Upon solving the resulting system of FOCs, labor demand functions for each group (which depend on wages for the four age segments and on economic activity), as in Equation (5), are derived.

L_{j}^{i *} = f (w_{1}, w_{2}, w_{3}, w_{4}, R) j = 1,2,3,4; i = w, m

(5)

Having ascertained the supply of and demand for labor functions (i.e., Equations 1 and 5, respectively), overall unemployment rate is then expressed as follows:

U_{j}^{i} = \frac{M_{j}^{i} - L_{j}^{i}}{M_{j}^{i}}, i = m, w j = 1, 2, 3, 4

(6)

Equation (6) then becomes the basis of our empirical estimation. It says that overall rate of unemployment is determined by the wages associated with age segments, the level of economic activity, and the number of active populations of sex $i$ and age $x$ expressed as follows:

U_{j}^{i} = f_{j}^{i} (w_{1}, w_{2}, w_{3}, w_{4}, R, m_{}^{i} (x))

(7)

Following the previous literature, we assume there is a union that represents the whole labor force of the economy and sets the wages for all workers (Eguía & Echevarría, 2004). The union maximizes preferences, $Z$ , defined over both wages and employment (Eguía & Echevarría, 2004), subject to the labor demand of firms, $L_{j}^{i *}$ , for every group which is expressed as follows:

\begin{matrix} M a x_{(w_{j})} Z (w_{1}, L_{1}^{M}, L_{1}^{W}, w_{2}, L_{2}^{M}, L,_{2}^{W}, w_{3}, L_{3}^{M}, L_{3}^{W}, w_{4}, L_{4}^{M}, L_{4}^{W}) \\ s . t . L_{j}^{i *} = f (w_{1}, w_{2}, w_{3}, w_{4}, R) \end{matrix}

(8)

Accordingly, the resulting system of FOCs obtained from Equation (8) is written as follows:

\frac{\partial Z}{\partial w_{j}} + \sum_{k = 1}^{3} \sum_{i = m, w} \frac{\partial Z}{\partial L_{k}^{i}} \frac{\partial L_{k}^{i}}{\partial w_{j}} = 0, j = 1, 2, 3, 4

(9)

Based on these arguments, thus, optimal wages are functions of the level of economic activity alone. That is:

w_{j}^{*} = w_{j} (R)

(10)

Incorporating (10) into (7) yields

U_{j}^{i} = f_{j}^{i} \{m^{i} (x), R\}

(11)

In summary, the overall unemployment rate (i.e., the unemployment rate for every age segment and sex) is determined by the age structure of active population and level of economic activity.

Before leaving the subsection, the variables in Equation (11) need to be well defined. In the empirical literature on the analysis of determinants of unemployment rate, when specifying the age structure of the active population, several studies have used different proxies to measure this variable. The main explanation for this, other than the difference in the conceptualization of the age structure of the active population, has been lack of reliable time series data, particularly for most developing countries. Nevertheless, four distinct variables have been the most commonly used proxies to measure the age structure of active population, namely the relative size of adult population (often depicted as the ratio of the size of the population between ages 40 and 65 to the population between ages 16 and 39 years; see Gil and Bailén, 1997), the relative mean age of adult population (approximate as the ratio of the mean age among people between ages 40 and 60 to the mean age among people between ages 16 and 39; see Eguía & Echevarría, 2004), the relative cohort size (defined as a ratio of the size of population between ages 16 and 24 to the population between ages 16 and 64; see Simion, 2015), and cohort size (which, by far the most widely used measure in the literature, is the one adopted and incorporated in the current study, and defined more precisely later on).

Further, as regards the specification of the level of economic activity $\{R\}$ , from theoretical and empirical perspectives, typical variables commonly included in the vector $\{R\}$ are measures of institutional quality, human capital accumulation, foreign direct investment (FDI hereafter) inflow, labor productivity growth, economic growth, government expenditure, external debt stocks, trade openness, gross fixed capital formation, inflation, and population growth rates. For comprehensive analysis, this article (following Adachi, 2007; Folawewo & Adeboje, 2017; Raifu, 2017) incorporates economic growth, inflation rate, FDI inflow, external debts stocks, and government expenditure.

Theoretically, in his path-breaking work, “Potential GNP: Its measurement and significance” published in 1962, Arthur Okun postulates that when an economy is increasing at a certain percentage (Raifu, 2017), unemployment rate is expected to decrease by a certain percentage. A plethora of empirical studies have examined the prediction and implication of this exposition. However, different studies have come up with varied conclusions (Folawewo & Adeboje, 2017). While one strand of the literature finds evidence consistent with the negative relationship, as posited by Okun’s law, others have found contrary evidence (Folawewo & Adeboje, 2017). Thus, the debate on whether inflation restricts, promotes, or is independent of unemployment rate remains, at best, inconclusive.

Also, starting with Phillips’ (1958) pioneering work, empirical studies on the nature of the relationship between inflation and unemployment rates occupy a considerable portion of economic literature. However, despite a large and burgeoning literature, the debate is far from being settled (Akinlo, 2009), with one strand of the literature finding support for an inverse relationship, as advanced by (Folawewo & Adeboje, 2017) Phillips’(1958) curve, while the other reports the opposite. Hence, following Adachi (2007) and Folawewo and Adeboje (2017), an attempt is made to explain the variation (Jamal, 2006) in overall unemployment rate in accordance with Phillips’ (1958) and Okun’s (1962) innovative works.

With respect to the unemployment effects of FDI inflow, available empirical evidence, as noted in the literature, is quite divergent. Central questions in the unemployment–FDI relationship debate are whether FDI replaces local investment in the host country; what are the particular sectors that are targeted by the investors; whether FDI has been oriented toward the acquisition of existing facilities (Ernst, 2005) or construction of new plants; are the investments from abroad export oriented (Zdravković et al., 2017). While most of the studies found strong evidence of significant and positive impacts of FDI, in particular to real GDP growth, its impacts on overall rate of unemployment are doubtful. Thus, it is included in the equation as a determinant of unemployment.

Further, there are two contrasting notions on unemployment–public debts relationship. The first holds the view that the two are inversely related. Advocates of this strand of literature emphasize that, during periods of unemployment, borrowing can be considered as a substitute to money creation and by implication as a tool of fiscal policy (Ogonna et al., 2016). If funds are borrowed and appropriately channeled into productive ventures, not only will this create employment in the economy, but the profit generated from such investments can also be used to service such debt (Ogonna et al., 2016). Proponents of the second view maintain that, unlike tax finance, public debt passes the burden of collective action on future generations (Ogonna et al., 2016). Through debt financing, existing taxpayers had their taxes reduced. The tax reduction was offset by higher taxes levied on taxpayers in the future to payback the debt. The obtained empirical results are quite divergent (Zdravković et al., 2017).

Lastly, two strands of theoretical literature on impacts of government expenditure on unemployment rate have evolved separately in development economics discourse, namely the classical economic growth models and Keynesian theories. The former holds the view that overall unemployment rate can be alleviated by cutting down wages, which would increase labor demand, and by extension, stimulate economic activity as well as employment. In contrast, the latter rejected the classical exposition of wage flexibility and inbuilt power of the invisible hand to restore employment (Onodugo et al., 2017) and as an alternative suggest fiscal policy measures in the form of increased government expenditure (Onodugo et al., 2017). In accordance with the theoretical ambiguity, empirical findings are mixed and do not provide conclusive evidence. Keeping all the above arguments in view, the study specifies an econometrically estimate equation as follows:

\partial \ln {ump}_{t} = α_{0} + α_{1} \ln {csi}_{t} + α_{2} (\partial \ln {gdp}_{t}) + α_{3} (\partial \ln \inf_{t}) + α_{4} \ln f {di}_{t} + α_{5} \ln {eds}_{t} + α_{6} \ln {gve}_{t} + ε_{1 t}

(12)

where $\partial \ln ump, \ln csi, \partial \ln gdp, \partial \ln \inf, \ln f di, \ln eds, \ln gve,$ and $ε_{1 t}$ denote, respectively, unemployment rate, cohort size, economic growth (proxied by real GDP growth rate), inflation rate, FDI inflow, external debts stock, government expenditure, and error term; $α_{i}$ for $(i = 1, ..., 6)$ represent the numerical estimates; while $t$ denotes time and $α_{0}$ denotes the drift component. All the variables (except unemployment, real GDP growth, and inflation rates) were expressed in natural logarithm $(\ln)$ to minimize the scale effect. It is anticipated that GDP growth and government expenditure will lessen aggregate unemployment rate. Cohort size, inflation rate, and FDI inflow may or may not benefit aggregate unemployment rate; therefore, the expected a priori is, respectively, negative or positive. External debt stock is detrimental to aggregate unemployment rate. Hence, the expected a priori is positive.

To retain simplicity, four model versions of Equation (12) were estimated. The four model versions are hereafter denoted as A, B, C, and D. In version A, the logarithm of youth-aged group 15–24 was proxied as cohort size. In model B, the exercise was repeated using log of middle-aged (25–34) population as measure for cohort size. The key reason for defining cohort this way is based on the assumption of potential endogeneity of schooling/education decision. By considering individuals between 25 and 34 years (age by which most people with at least/less than an undergraduate degree would have finished their education), this issue of endogeneity is addressed. In model versions C and D, respectively, growth rates of the “youth” (aged 15–24) and “middle” (aged 25–34) population are incorporated as proxies for flow of cohort size. The main motivation for this, from economic intuition standpoint, has been that growth rates of “youth” (aged 15–24) and “middle” (aged 25–34) labor force seem more plausible than the absolute size. This is because it is the rapid entry of youth labor force that is likely to put pressure on labor market.

Estimation Techniques and Empirical Analysis

Stationarity Tests, Vector Autoregression (VAR) Lag Length Selection, and Cointegration Results

Prior to a detailed estimation of model (12), the study performed stationarity and cointegration tests on all the series. To test for stationarity, the study employed two main tests, specifically the augmented Dickey–Fuller (ADF), advanced by Dickey and Fuller (1979), and the Phillips–Perron (PP), developed by Phillips and Perron (1988). Both were conducted at 1%, 5%, and 10% levels of significance. First, the tests were undertaken with intercept only, and thereafter with intercept and trend. The results obtained and presented in Tables 2 and 3 clearly indicate that, other than real GDP growth rate, inflation rate, growth rate of youth-aged group, and growth rate of middle-aged group which were stationary at levels, all other variables were found to be integrated of order one.

Next, having ascertained the order of integration of the variables, the existence of long-run cointegrating relationship among the variables was also established using ARDL procedure of Pesaran and Shin (1998) and Pesaran et al. (2001). The article considered the approach apt because of the following reasons. First, given the nature of interrelation between aggregate unemployment rate, real GDP growth, and inflation rates, included in the model, this technique appears suitable to address any possible endogeneity issue. Second, since the set of variables used in the study is a mix of I(1) and I(0) variables, the method is more reliable as compared to Johansen and Juselius (1990) and Johansen (1992) maximum likelihood procedure as well as the residual-based method put forward by Engle and Granger (1987), as it does not pose a strict classification of regressors (Adediran et al., 2019) to be of the same order. Third, it is apposite for finite and small sample size (Pesaran et al., 2001). Fourth, unlike other techniques, each series can have different number of lags. Finally, it enables short- and long-run parameters of the model to be estimated simultaneously. In line with these, the dynamic ARDL version of model (12) is expressed as follows:

\begin{array}{l} Δ (\partial \ln {ump}_{t}) = α_{0} + \sum_{i = 1}^{p} α_{1 i} Δ (\partial \ln {ump}_{t - i}) + \sum_{i = 0}^{q} α_{2 i} Δ \ln cs i_{t - i} + \sum_{i - 0}^{r} α_{3 i} Δ (\partial \ln {gpd}_{t - i}) + \sum_{i = 0}^{s} α_{4 i} Δ (\partial \ln \inf_{t - i}) + \\ \sum_{i = 0}^{v} α_{5 i} Δ \ln f {di}_{t - i} + \sum_{i = 0}^{w} α_{6 i} Δ \ln {eds}_{t - i} + \sum_{i = 0}^{x} α_{7 i} Δ \ln {gve}_{t - i} + γ_{1} {(\partial \ln ump)}_{t - 1} + γ_{2} {(\ln csi)}_{t - 1} + γ_{3} {(\partial \ln gdp)}_{t - 1} + \\ γ_{4} {(\partial \ln \inf)}_{t - 1} + γ_{5} {(\ln f di)}_{t - 1} + γ_{6} {(\ln eds)}_{t - 1} + γ_{7} {(\ln gve)}_{t - 1} + ε_{2 t} \end{array}

(13)

where $p, q, r, s, v, w, x$ denote the maximum number of lags, $Δ$ denotes the first difference operator, $ε_{2 t}$ denotes white noise error term, $γ' s$ correspond to long-run estimates, $α' s$ capture the short-run parameters. Thus, to identify cointegration among the variables, a two-step procedure was followed. First, we estimated Equation (13) using ordinary least squares procedure. Second, the null hypothesis of no cointegration $H_{0} : γ_{1} = γ_{2} = γ_{3} = γ_{4} = γ_{5} = γ_{6} = γ_{7} = 0$ was tested against the alternative hypothesis of the existence of a long-run relationship $H_{1} : γ_{1} \neq γ_{2} \neq γ_{3} \neq γ_{4} \neq γ_{5} \neq γ_{6} \neq γ_{7} \neq 0$ . Nonetheless, before the test was applied, the optimal lag lengths of order 1 incorporated for each variable in the model was first ascertained using Schwarz information criterion. The results are presented in Table 4.

With this lag length: 64 distinct ARDL specifications for models A, B, C and D, respectively, were considered and the most suitable models (1,0,0,1,1,0,0), (1,0,1,1,0,0,0), (1,0,0,1,0,0,0) and (1,0,0,1,0,0,1) were, respectively, selected for A, B, C, and D. Figures 1 –4 depict the graphs of the top 20 models, showing the relative superiority of the selected models against other possible models. Afterward, we applied the ARDL procedure to examine the cointegration relationship. The results are depicted in Table 5. As can be seen, in all cases, the computed F-statistics are greater than the upper bound values at 5% level of significance. This implies that there is a long-run cointegrating relationship between aggregate unemployment rate, cohort size, real GDP growth rate, inflation rate (proxied by consumer price index), FDI inflow, external debts stocks, and government expenditure.

Table 2.

Stationarity Tests of Variables.

ADF Test (intercept only)

Variable

LVL

First Diff.

ADF Statistic

CRV (%)

PVL

Rmks

ADF Statistic

CRV (%)

PVL

Rmks

ɖlnump

1.2682

−3.5713

−2.9224

−2.5992

0.6370

−8.7825

−3.5744

−2.9238

−2.5999

0.0000

I(1)

lncsi1524

−1.6028

−3.5812

−2.9266

−2.6014

0.4731

−8.8972

−3.5812

−2.9266

−2.6014

0.0000

I(1)

lncsi2534

−1.6862

−3.5713

−2.9224

−2.5992

0.4318

−7.0304

−3.5744

−2.9238

−2.5999

0.0000

I(1)

ɖlncsi1524

−8.8871

−3.5812

−2.9266

−2.6014

0.0000

I(0)

***

I(0)

ɖlncsi2534

−7.0447

−3.5744

−2.9238

−2.5999

0.0000

I(0)

***

I(0)

ɖlngdp

−5.6131

−3.5713

−2.9224

−2.5992

0.0000

I(0)

***

I(0)

ɖlninf

−3.6893

−3.5713

−2.9224

−2.5992

0.0073

I(0)

***

I(0)

Lnfdi

−1.3189

−3.5744

−2.9238

−2.5999

0.6134

−12.8631

−3.5744

−2.9238

−2.5999

0.0000

I(1)

Lneds

−1.7503

−3.5713

−2.9224

−2.5992

0.4003

−5.2027

−3.5744

−2.9238

−2.5999

0.0001

I(1)

Lngve

−1.4600

−3.5713

−2.9224

−2.5992

0.5452

−7.7152

−3.5744

−2.9238

−2.5999

0.0000

I(1)

ADF Test (intercept and trend)

Variable

LVL

First Diff.

ADF Statistic

CRV (%)

PVL

Rmks

ADF Statistic

CRV (%)

PVL

Rmks

ɖlnump

−2.6338

−4.1567

−3.5043

−3.1818

0.2679

−8.7134

−4.1611

−3.5064

−3.1830

0.0000

I(1)

lncsi1524

−1.6282

−4.1706

−3.5107

−3.1855

0.7662

−8.7945

−4.1706

−3.5107

−3.1855

0.0000

I(1)

lncsi2534

−0.7436

−4.1611

−3.5064

−3.1830

0.9996

−8.6282

−4.1611

−3.5064

−3.1830

0.0000

I(1)

ɖlncsi1524

−8.7847

−4.1706

−3.5107

−3.1855

0.0000

I(0)

***

I(0)

ɖlncsi2534

−8.6573

−4.1611

−3.5064

−3.1830

0.0000

I(0)

***

I(0)

ɖlngdp

−5.6288

−4.1567

−3.5043

−3.1818

0.0001

I(0)

***

I(0)

ɖlninf

−3.6782

−4.1567

−3.5043

−3.1818

0.0334

I(0)

***

I(0)

Lnfdi

−0.9572

−4.1611

−3.5064

−3.1830

0.9403

−12.9112

−4.1611

−3.5064

−3.1830

0.0000

I(1)

Lneds

−0.9824

−4.1567

−3.5043

−3.1818

0.9371

−5.4026

−4.1611

−3.5064

−3.1830

0.0003

I(1)

Lngve

−1.5447

−4.1567

−3.5043

−3.1818

0.8001

−7.9677

−4.1611

−3.5064

−3.1830

0.0000

I(1)

Source: The author.

Note: NS denotes non-stationary.

Table 3.

Stationarity Tests of Variables.

PP Test (intercept only)

Variable

LVL

First Diff.

PP Statistic

CRV (%)

PVL

Rmks

PP Statistic

CRV (%)

PVL

Rmks

ɖlnump

−1.1002

−3.5713

−2.9224

−2.5992

0.7086

−8.8535

−3.5744

−2.9238

−2.5999

0.0000

I(1)

lncsi1524

−1.3203

−3.5713

−2.9224

−2.5992

0.6129

−5.4938

−3.5744

−2.9238

−2.5999

0.0000

I(1)

lncsi2534

−1.6959

−3.5713

−2.9224

−2.5992

0.4270

−7.2336

−3.5744

−2.9238

−2.5999

0.0000

I(1)

ɖlncsi1524

−5.4664

−3.5744

−2.9238

−2.5999

0.0000

I(0)

***

I(0)

ɖlncsi2534

−7.2466

−3.5744

−2.9238

−2.5999

0.0000

I(0)

***

I(0)

ɖlngdp

−5.6257

−3.5713

−2.9224

−2.5992

0.0000

I(0)

***

I(0)

ɖlninf

−3.5693

−3.5713

−2.9224

−2.5992

0.0101

I(0)

***

I(0)

Lnfdi

−1.8227

−3.5713

−2.9224

−2.5992

0.3655

−12.5860

−3.5744

−2.9238

−2.5999

0.0000

I(1)

Lneds

−1.6809

−3.5713

−2.9224

−2.5992

0.4345

−5.1945

−3.5744

−2.9238

−2.5999

0.0001

I(1)

Lngve

−1.5429

−3.5713

−2.9224

−2.5992

0.5037

−7.6689

−3.5744

−2.9238

−2.5999

0.0001

I(1)

PP Test (trend and intercept)

Variable

LVL

First Diff.

PP Statistic

CRV (%)

PVL

Rmks

PP Statistic

CRV (%)

PVL

Rmks

ɖlnump

−2.5057

−4.1567

−3.5043

−3.1818

0.3242

−8.7915

−4.1611

−3.5064

−3.1830

0.0000

I(1)

lncsi1524

−1.1555

−4.1567

−3.5043

−3.1818

0.9084

−5.5796

−4.1611

−3.5064

−3.1830

0.0002

I(1)

lncsi2534

−0.5592

−4.1567

−3.5043

−3.1818

0.9992

−8.4786

−4.1611

−3.5064

−3.1830

0.0000

I(1)

ɖlncsi1524

−5.5505

−4.1611

−3.5064

−3.1830

0.0002

I(0)

***

I(0)

ɖlncsi2534

−8.5045

−4.1611

−3.5064

−3.1830

0.0000

I(0)

***

I(0)

ɖlngdp

−5.6288

−4.1567

−3.5043

−3.1818

0.0001

I(0)

***

I(0)

ɖlninf

−3.5115

−4.1567

−3.5043

−3.1818

0.0492

I(0)

***

I(0)

lnfdi

−3.1338

−4.1567

−3.5043

−3.1818

0.1101

−12.6321

−4.1611

−3.5064

−3.1830

0.0000

I(1)

lneds

−1.1018

−4.1567

−3.5043

−3.1818

0.9183

−5.3500

−4.1611

−3.5064

−3.1830

0.0003

I(1)

lngve

−1.6025

−4.1567

−3.5043

−3.1818

0.7776

−7.9206

−4.1611

−3.5064

−3.1830

0.0000

I(1)

Source: The author.

Note: NS denotes non-stationary at LVL.

Table 4.

VAR Lag Length Selection Results.

First specification (model A)
LAG	LR.	FPE.	AIC.	SC.	HQ.
0	NA	1.34E+02	24.76205	25.03493	24.86517
1	489.2257*	0.005143*	14.57307*	16.75614*	15.39806*
2	61.56094	6.98E-03	14.74925	18.84251	16.29610
Second specification (model B)
LAG	LR.	FPE.	AIC.	SC.	HQ.
0	NA	4.50E+01	23.67202	23.94491	23.77515
1	455.8556	3.98E-03	14.31730	16.50037*	15.14229*
2	82.28018*	0.002885*	13.86563*	17.95888	15.41247
Third specification (model C)
LAG	LR.	FPE.	AIC.	SC.	HQ.
0	NA	1.21E+01	22.3584	22.63395	22.46209
1	371.2067*	0.007326*	14.92538*	17.12981*	15.75492*
2	63.33507	0.009352	15.03127	19.16458	16.58666
Fourth specification (model D)
LAG	LR.	FPE.	AIC.	SC.	HQ.
0	NA	1.12E+01	22.28105	22.5566	22.38474
1	383.9565	0.004890*	14.52112*	16.72555*	15.35066*
2	68.16957*	0.005367	14.47592	18.60923	16.03131

Source: The author.

Table 5.

Cointegration Results.

First specification (model A)
Level of Significance (%)	Bounds Value		F Statistic
	Lower I(0)	Upper I(1)	3.63
10	1.99	2.94
5	2.27	3.28
2.5	2.55	3.61
1	2.88	3.99
Second specification (model B)
Level of Significance (%)	Bounds Value		F Statistic
	Lower I(0)	Upper I(1)	3.53
10	1.99	2.94
5	2.27	3.28
2.5	2.55	3.61
1	2.88	3.99
Third specification (model C)
Level of Significance (%)	Bounds Value		F Statistic
	Lower I(0)	Upper I(1)	3.51
10	1.99	2.94
5	2.27	3.28
2.5	2.55	3.61
1	2.88	3.99
Fourth specification (model D)
Level of Significance (%)	Bounds Value		F-statistic
	Lower I(0)	Upper I(1)	3.41
10	1.99	2.94
5	2.27	3.28
2.5	2.55	3.61
1	2.88	3.99

Source: The author.

Figure 1.

Source: Author’s computation using E-view 10 (2020).

Figure 2.

Source: Author’s computation using E-view 10 (2020).

Figure 3.

Source: Author’s computation using E-view 10 (2020).

Figure 4.

Source: Author’s computation using E-view 10 (2020).

Long-run and Short-run Estimates

Having established the existence of cointegrating relationship, the numerical estimates associated with the selected models were ascertained. The estimated long- and short-run coefficients obtained (presented in Tables 6 and 7, respectively) are broadly similar, albeit with different magnitudes. Beginning from error correction term (ECM) in Table 7, in all the specifications, the unrestricted ECM (−1) coefficients follow a priori expectation in that they are all negative and statistically significant at 5% level. This suggests that short-run disequilibrium is corrected in the long-run equilibrium.

With respect to the relationship between cohort size and overall rate of unemployment in Nigeria, as anticipated, in all the specifications, for the period 1970–2019, the results revealed that cohort size has a positive and significant impact on overall rate of unemployment in both the long-run and short-run periods, albeit with distinct magnitudes. More specifically, regarding model C, a 1% increase in growth rate of youth-aged group (15–24) will bring about 15.8795% and 18.8803% increase in overall rate of unemployment, respectively, in the short- and long run. By implication, in line with the Easterlin’s (1968) proposition, an increase in the size of one’s generation, notably the share of young persons in the working-age population, does have repercussions not only on the (un-) employment outcomes of that particular group, but also on the other age group in Nigeria. Similar findings were also observed by Moffat and Roth (2017).

With regard to the control variables, an examination of the results presented in Tables 6 and 7 revealed that economic growth (proxied by real GDP growth rate) has negative but statistically insignificant impacts on overall rate of unemployment, as evidenced by t-statistics and p-values. In more specific terms, with respect to model A, ceteris paribus, a 1% increase in economic growth will bring about 0.0247% and 0.1894% decrease in overall rate of unemployment in the short run and in the long run, respectively. This negative but statistically insignificant unemployment impact of real GDP growth rate suggests low employment elasticities of growth in Nigeria. A plausible explanation of this reducing but statistically insignificant effect growth is that output growth in Nigeria is insensitive to the unemployment situation in the economy. Likewise, the high growth rate and a high unemployment level indicate the economy’s over reliance on oil sector as its major source of revenue. Few proportions of the country’s labor force are captured in this sector. This result is consistent with Okun’s (1962) law. Moreover, the result is in congruence with the findings of Lee (2000), Knotek (2007), Villaverde and Maza (2009), Adawo et al. (2012), Doğru (2013), Ogbeide et al. (2016).

Further, with respect to unemployment effect of inflation rate, a careful look at Tables 6 and 7 below revealed that, in all the specifications, the elasticity coefficients of inflation rate are positive and statistically significant as anticipated. From the estimated model A, over the period 1970–2019, the result suggests that for a 1% decrease in inflation rate, 2.0228% and 0.1140% in overall rate of unemployment are induced, respectively, in the short and in the long run. This result corroborates the empirical findings of Sir (2014) and Raifu (2017). However, it contradicts the findings of Elliot (2015) and Folawewo and Adeboje (2017). Moreover, the result is consistent with the Phillips (1958) curve hypothesis.

Making reference to Tables 6 and 7 again, an insight from the short-run and long-run estimates suggests that the effect of FDI inflow on overall rate of unemployment apparently depends on its greenfield (whether the investments are oriented toward the construction of new plants) and brownfield (transnational mergers and acquisitions of the existing facilities) structure. As anticipated, according to the four specifications, the elasticity coefficients of inflation rate have the expected sign but are statistically insignificant, as shown by the t-statistics and p-values. With respect to model A, we found that an increase in FDI inflow by 1% will bring about decrease in rate of unemployment by 1.4701% and 0.2724%, respectively, in the short run and in the long run. This result is consistent with economic theory and the findings of Ogbeide et al. (2016) and Zdravković et al. (2017).

With respect to unemployment effects of external debt stocks, the empirical results not only validate Keynesian’s proposition but also invalidate the Monetarist and New Classical economists’ expositions. As shown in Tables 6 and 7, in all specifications, the coefficients of external debt stocks remain negative as anticipated and statistically significant as evidenced by the t-statistics and p-values. In specification A, the results revealed that an increase in external debt stocks by 1% will bring about a decrease in unemployment by 1.9534% in the long run. By implication, an upsurge in unemployment due to lack of demand or recession may be prevented by means of fiscal policies of the state aimed at employment boost. The result supports the findings of Okonjo-Iweala et al. (2003) and Folawewo and Adeboje (2017). Unlike its long-run negative significant impact, in all four specifications, external debt stocks had a negative but insignificant short-run effect on rate of unemployment, suggesting that borrowing (public) should exclusively be for capital projects that have the potential to create jobs (Ogonna et al., 2016).

Finally, an insight from the results depicted in Tables 6 and 7 suggests that government expenditure in the long run seems to worsen and aggravate unemployment problem in the country, reflecting prolonged deficit financing and rent-seeking behavior of Nigeria’s economic and political elites. From the estimated model A, as evidenced in the Tables 6 and 7, in contrast to the study’s expectation, keeping other variables constant, the results revealed that an increase in government expenditure by 1% results in an increase in unemployment by 4.9962% in the long run. This evidence of positive and significant unemployment impact of government spending undoubtedly reflects the state of Nigerian economy where government workers and elected officials believe they have the right to share government revenue and use it to help their co-religionists, supporters, and members of their ethnic groups.

In conclusion, a variety of diagnostics tests were conducted to check the adequacy of selected ARDL models and reliability of the results. The results of these various diagnostics tests are depicted in the lower segment of Table 6 and graphs shown in Figures 5 –12. All the results disclosed that the models possess the appropriate best linear unbiased estimator (BLUE) properties.

Table 6.

Long-run Estimates.

Explanatory Variables	Model A First (baseline) Specification	Model B Second Specification	Model C Third Specification	Model D Fourth Specification
lncsi1524	23.9904*
	{4.4630}
	[5.3754]
	{{0.0441}}
lncsi2534		16.4093*
		{3.0241}
		[5.4262]
		{{0.0113}}
ɖlncsi1524			18.8803*
			{4.5021}
			[4.1937]
			{{0.0069}}
ɖlncsi2534				27.8233*
				{9.1203}
				[3.0507]
				{{0.0307}}
ɖlngdp	−0.1894*	−0.1785*	−0.1799*	−0.0818*
	{0.1647}	{0.1624}	{0.1886}	{0.1823}
	[−1.1496]	[−1.0994]	[−0.9544]	[−0.4485]
	{{0.2573}}	{{0.2784}}	{{0.3456}}	{{0.6563}}
ɖlninf	−0.1140*	−0.1119*	−1.0599*	−1.0671*
	{0.0659}	{0.0654}	{0.0682}	{0.6629}
	[−1.7291]	[−1.7112]	[−15.5544]	[−1.6097]
	{{0.0317}}	{{0.0359}}	{{0.0342}}	{{0.0232}}
lnfdi	−0.2724*	−0.0186*	−0.5951*	−1.3366*
	{1.7175}	{1.8567}	{1.9882}	{1.8701}
	[−0.1586]	[−0.0100]	[−0.2993]	[−0.7147]
	{{0.8748}}	{{0.9920}}	{{0.7662}}	{{0.4792}}
lneds	−1.9534*	−1.7339*	−2.1582*	−1.9969*
	{0.7025}	{1.1469}	{0.8839}	{0.7556}
	[−2.7808]	[−1.5118]	[−2.4418]	[−2.6429]
	{{0.0083}}	{{0.0088}}	{{0.0191}}	{{0.0119}}
lngve	4.9662*	4.3636*	5.1236*	5.0375*
	{1.2415}	{1.5250}	{1.5204}	{1.3493}
	[4.0000]	[2.8615]	[3.3698]	[3.7336]
	{{0.0003}}	{{0.0067}}	{{0.0017}}	{{0.0006}}
C	8.2323*	−17.3838*	6.4432*	22.2130*
	{6.5613}	{5.4592}	{3.3378}	{12.4166}
	[1.2547]	[3.1843]	[1.9304]	[0.6272]
	{{0.0482}}	{{0.0207}}	{{0.0439}}	{{0.0343}}
Goodness of fit measures
R²	0.9128	0.9127	0.9083	0.9134
Adjusted R²	0.8927	0.8926	0.8900	0.8929
F-statistic	45.3566	45.3073	49.5331	44.5598
F-statistic (Prob)	0.0000	0.0000	0.0000	0.0000
Durbin-Watson Stat	2.0687	2.0258	2.0729	2.0911
Diagnostic Statistical Checking
Breusch–Godfrey serial correlation LM test	10.9923*** {{0.4100}}	12.0350*** {{0.2400}}	9.96644*** {{0.6900}}	3.8914*** {{0.1429}}
Breusch-Pagan-Godfrey test for heteroskedasticity	11.9690*** {{0.2151}}	10.2633*** {{0.3296}}	11.2799*** {{0.1863}}	10.5629*** {{0.3069}}
ARCH test for heteroskedasticity	0.9550*** {{0.3285}}	0.6405*** {{0.4235}}	0.4378*** {{0.8061}}	0.0666*** {{0.7963}}
Jarque-Bera normality test	0.2633** {{0.8766}}	0.6256** {{0.7314}}	0.5788*** {{0.7487}}	0.0922** {{0.9549}}
Ramsey RESET specification test	[0.9872] {{0.3372}}	[0.9720] {{0.3298}}	[0.7059] {{0.4845}}	[0.6678] {{0.5084}}

Source: The author.

Note: {{ }}, [ ], { } and * Depict, probability, t-statistic, Std. error, and coefficient, respectively.

** and *** Denote obs. R-squared and Jarque–Bera statistic, respectively.

Table 7.

Short-run Estimates.

Explanatory Variables	Model A First (baseline) Specification	Model B Second Specification	Model C Third Specification	Model D Fourth Specification
ECM(−1)	−0.4606*	−0.5269*	−0.5204*	−0.5128*
	{0.0818}	{0.1485}	{0.1171}	{0.1138}
	[−5.6323]	[−3.5481]	[−4.4431]	[−4.5060]
	{{0.0000}}	{{0.0012}}	{{0.0001}}	{{0.0001}}
lncsi1524	3.7108*
	{0.7370}
	[5.0354]
	{{0.0001}}
lncsi2534		24.4380*
		{7.6286}
		[3.2035]
		{{0.0007}}
ɖlncsi1524			15.8795*
			{5.4949}
			[2.8899]
			{{0.0018}}
ɖlncsi2534				7.9231*
				{5.0962}
				[1.5547]
				{{0.0003}}
ɖlngdp	−0.0247*	−0.0709*	−0.0474*	−0.0423*
	{0.0960}	{0.0791}	{0.0770}	{0.0761}
	[−0.2571]	[−0.8962]	[−0.6155]	[−0.5561]
	{{0.7987}}	{{0.3764}}	{{0.5424}}	{{0.5819}}
ɖlninf	−2.0228*	−1.3031*	−1.0198*	−1.0258*
	{0.2029}	{0.3343}	{0.2061}	{0.2059}
	[−9.9653]	[−3.8979]	[−4.9484]	[−4.9824]
	{{0.0017}}	{{0.0098}}	{{0.0043}}	{{0.0062}}
lnfdi	−1.4701*	−0.8064*	−1.5168*	−1.2020*
	{1.2868}	{1.1631}	{0.9222}	{0.9244}
	[−1.1425]	[−0.6933]	[−1.6447]	[−1.3003]
	{{0.2615}}	{{0.4928}}	{{0.1095}}	{{0.2025}}
lneds	−1.8618*	−2.3853*	−1.3060*	−1.3006*
	{1.1391}	{0.9920}	{0.8865}	{0.8734}
	[−1.6344]	[2.4046]	[−1.4732]	[−1.4892]
	{{0.1117}}	{{0.2018}}	{{0.1502}}	{{0.1459}}
lngve	−1.2235*	−0.8767*	−1.3014*	−2.0379*
	{2.3927}	{2.0246}	{1.7388}	{1.7552}
	[−0.5113]	[−0.4330]	[−0.7485]	[−1.1611]
	{{0.6125}}	{{0.6678}}	{{0.4595}}	{{0.2539}}

Source: The author.

Notes: {{ }}, [ ], { } and * Depict, probability, t-statistic, Std. error, and coefficient, respectively.

Figure 5.

Source: Author’s computation using E-view 10 (2020).

Figure 6.

Source: Author’s computation using E-view 10 (2020).

Figure 7.

Source: Author’s computation using E-view 10 (2020).

Figure 8.

Source: Author’s computation using E-view 10 (2020).

Figure 9.

Source: Author’s computation using E-view 10 (2020).

Figure 10.

Source: Author’s computation using E-view 10 (2020).

Figure 11.

Source: Author’s computation using E-view 10 (2020).

Figure 12.

Source: The author.

Conclusions

The article empirically examines the nature of the relationship between cohort size and aggregate rate of unemployment in Nigeria in a dynamic and multivariate framework. ARDL technique was applied to assess the cointegrating, long- and short-run relationship among the series. Using time series data, for the period 1970–2019, the article finds that cohort size has positive and significant impact on rate of unemployment in Nigeria in both the long- and short-run periods. These results, therefore, lend credence to cohort-crowding hypothesis of Easterlin. In view of these findings, several policy implications are deduced. First, there is need for properly designed and well implemented functional microcredit schemes for the youth. Such schemes should be flexibly structured to avoid institutional bottlenecks and enhance accountability and transparency in management. Second, apart from the inclusion of entrepreneurial skills and training modules in the curriculum of all levels of education, prioritizing vocational and technical education training institutions to prepare youth for labor market can help alleviate youth unemployment rate in the country. The success of Canada, Australia, Germany, Japan, South Korea, Singapore, Taiwan, and Hong Kong provide a very clear evidence.

Contribution to Knowledge

This article contributes to the literature in several respects. First, the Nigerian labor market is examined. To date, extant evidence on Nigeria lacks intuitive insights on the subject. While there is a sizable body of literature assessing the macroeconomic implications of this specifically defined age group in terms of economic growth opportunities it presents, the question of whether the size of this specifically defined age group affects labor market prospects of its members has been left largely unaddressed. Second, in the extant literature, cohort size is often defined as a proportion of the population of youngsters aged 16–24 relative to the population size of individuals aged 16–24. One aspect not taken into account in these estimations is the fact that an individual’s investment in education is endogenous and can be one of the channels in which individuals manage their timing of entry into the labor market, due to the larger sizes of the cohort. By considering individuals between 25 and 34 years of age (age by which most people with less than an undergraduate degree and those with at least an undergraduate degree would have finished their education), this issue of endogeneity is addressed. Lastly, the study demonstrates the empirical validity of the cohort-crowding hypothesis.

Limitation and Future Research Direction

One of the major limitations of time series analysis is the fact it is difficult to capture all variables influencing a particular variable of interest. Given that the study employed time series analysis, it bears the same limitation. Indeed, apart from cohort size, inflation rate, economic growth, total external debts stocks, FDI inflow, and government expenditure, there are other factors affecting unemployment rate. These include among others human capital, trade openness, and current account balance variables. Therefore, it would be of interest to find out the contribution of education, health, and international trade as essential determinants of un-(employment) which is beyond the scope of this study. These are areas open to other researchers to contribute.

Footnotes

Declaration of Conflicting Interests

Funding

The author received no financial support for the research, authorship and/or publication of this article.

References

Abraham

K. G.

, & Shimer

(2001). Changes in unemployment duration and labor force attachment (No. w8513). National Bureau of Economic Research. https://www.nber.org/papers/w8513

Adachi

(2007). Economic growth and unemployment-theoretical foundations of Okun’s Law . The University of Marketing and Distribution Science, pp. 1–14. https://repository.kulib.kyoto-u.ac.jp/dspace/bitstream/2433/81030/1/1557-01.pdf

Adawo

M. A.

, Essien

E. B.

, & Ekpo

N. U.

(2012). Is Nigeria’s unemployment problem unsolvable. Current Research Journal of Social Sciences , 4(6), 389–395. https://citeseerx.ist.psu.edu/viewdoc/download?doi=10.1.1.1041.4065&rep=rep1&type=pdf

Adediran

O. S.

, George

E. O.

, Alege

P. O.

, & Obasaju

B. O.

(2019). Is there any relationship between monetary policy tools and external credit-growth nexus in Nigeria. Cogent Economics & Finance , 7(1), 1–14. https://doi.org/10.1080/23322039.2019.1625100

Aiwone

(2016). Demographic dividends and economic growth in Nigeria (Doctoral Dissertation). Department of Economics, Faculty of Social Sciences, University of Ibadan.

Akinlo

A. E.

(2009). Electricity consumption and economic growth in Nigeria: Evidence from cointegration and co-feature analysis. Journal of Policy Modeling , 31(5), 681–693. https://doi.org/10.1016/j.jpolmod.2009.03.004

Anderson

J. M.

(1982). An economic-demographic model of the United States labor market. Research in Population Economics , 4, 117–153. https://pubmed.ncbi.nlm.nih.gov/12264899/

Ashford

L. S.

(2007). Africa’s youthful population: Risk or opportunity . Population Reference Bureau. http://www.brac.net/sites/default/files/glm/AfricaYouth.pdf

Barwell

(2000). Age structure and the UK unemployment rate’. Bank of England Quarterly Bulletin , 40(3), 57–65. https://doi.org/10.2139/ssrn.258022

10.

Ben-Porath

(1985). Market, government, and Israel's muted baby boom (Working Paper No. 1569). National Bureau of Economic Research. https://www.nber.org/papers/w1569

11.

Berger

M. C.

(1985). The effect of cohort size on earnings growth: A reexamination of the evidence. Journal of Political Economy , 93(3), 561–573. https://doi.org/10.1086/261315

12.

Biagi

, & Lucifora

(2008). Demographic and education effects on unemployment in Europe. Labour Economics , 15(5), 1076–1101. https://doi.org/10.1016/j.labeco.2007.09.006

13.

Bloom

D. E.

(2010). Nigeria: The next generation report (Working Paper 6210, Program on the Global demography of aging). Harvard University. https://cdn1.sph.harvard.edu/wp-content/uploads/sites/1288/2013/10/PGDA_WP_62.pdf

14.

Bloom

D. E.

, Freeman

R. B.

, & Korenman

S. D.

(1988). The labour-market consequences of generational crowding. European Journal of Population/Revue Européenne de Démographie , 3(2), 131–176. https://link-springer-com-443.web.bisu.edu.cn/article/10.1007%2FBF01796774

15.

Bloom

, Finlay

, Humair

, Mason

, Olaniyan

, & Soyibo

(2010). Prospects for economic growth in Nigeria: A demographic perspective . In IUSSP Seminar on Demographics and Macroeconomic Performance held at Novotel, Gare de Lyon, Paris, France (pp. 4–5). https://ntaccounts.org/doc/repository/BFHMOS2010.pdf

16.

Canning

, Raja

, & Yazbeck

A. S.

(2015). Africa's demographic transition: Dividend or disaster?

The World Bank.

https://doi.org/10.1596/978-1-4648-0489-2

17.

CBN. (2019). Central Bank of Nigeria Statistical Bulletin . https://www.cbn.gov.ng

18.

Cleland

(2012). Will Africa benefit from a demographic dividend. The Health & Education Advice & Resource Team (HEART) . https://www.heart-resources.org/wp-content/uploads/2012/12/Demographic-Dividend-November-2012.pdf

19.

Dickey

D. A.

, & Fuller

W. A.

(1979). Distribution of the estimators for autoregressive time series with a unit root. Journal of the American Statistical Association , 74(366a), 427–431. https://doi.org/10.1080/01621459.1979.10482531

20.

Doğru

(2013). The link between unemployment rate and real output in Eurozone: A panel error correction approach. Procedia-Social and Behavioral Sciences , 99, 94–103. https://doi.org/10.1016/j.sbspro.2013.10.475

21.

Dramani

, & Mbacké

(2017, October). Africa’s demographic dividend-An elusive window of opportunity? In International Population Conference.

22.

Drummond

M. P.

, Thakoor

, & Yu

(2014). Africa rising: Harnessing the demographic dividend (No. 14–143). International Monetary Fund. https://www.imf.org/en/Publications/WP/Issues/2016/12/31/Africa-Rising-Harnessing-the-Demographic-Dividend-41819

23.

Easterlin

R. A.

(1968). The American baby boom in historical perspective. In Easterlin

R. A.

, (Ed.), Population, labor force, and long swings in economic growth: The American experience (pp. 77–110). National Bureau of Economic Research, Inc. https://www.nber.org/chapters/c1119.pdf

24.

Easterlin

R. A.

(1978). What will 1984 be like? Socioeconomic implications of recent twists in age structure. Demography , 15(4), 397–432. https://doi.org/10.2307/2061197

25.

Eguía

, & Echevarría

C. Á

. (2004). Unemployment rates and population changes in Spain. Journal of Applied Economics , 7(1), 47–76. https://doi.org/10.1080/15140326.2004.12040603

26.

Elliot

(2015). The relationship between inflation and unemployment in Ghana: Analysis of the Philips Curve. African Journal of Economic Review , 3(2), 117–124.

27.

Engle

R. F.

, & Granger

C. W.

(1987). Co-integration and error correction: Representation, estimation, and testing. Econometrica: Journal of the Econometric Society , 55(2), 251–276. https://doi.org/10.2307/1913236

28.

Ernst

(2005). The FDI–employment link in a globalizing world: The case of Argentina, Brazil and Mexico. Employment Strategy Papers , 17, 1–45. https://www.ilo.org/empelm/pubs/WCMS_114029/lang–en/index.htm

29.

Flaim

P. O.

(1990). Population changes, the baby boom, and the unemployment rate. Monthly Labor Review , 113, 3.

30.

Folawewo

A. O.

, & Adeboje

O. M.

(2017). Macroeconomic determinants of unemployment: Empirical evidence from economic community of West African states. African Development Review , 29(2), 197–210. https://doi.org/10.1111/1467-8268.12250

31.

Foote

C. L.

(2007). Space and time in macroeconomic panel data: Young workers and state-level unemployment Revisited (Working Paper 07-10). Federal Reserve Bank of Boston. https://dx-doi-org.web.bisu.edu.cn/10.2139/ssrn.1021697

32.

Freeman

R. B.

(1979). The effect of generational crowding on the labor market for young male workers . Proceedings of the American Statistical Association, pp. 45–53. https://scholar.harvard.edu/freeman/publications/effect-generational-crowding-labor-market-young-male-workers

33.

Fuchs

, & Weyh

(2014). Demography and unemployment in East Germany: How close are the ties? (IAB-Discussion Paper No. 26/2014). http://hdl.handle.net/10419/108833

34.

Garloff

, Pohl

, & Schanne

(2013). Do small labor market entry cohorts reduce unemployment. Demographic Research , 29, 379–406. https://doi.org/10.4054/DemRes.2013.29.15

35.

Gil

J. T.

, & Bailén

J. M.

(1997). Transitional effects of a pension system change in Spain . Documents de treball (Facultat d'Economia i Empresa. Espai de Recerca en Economia), 1997, E97/018, pp. 1–49. http://hdl.handle.net/2445/12050

36.

Jamal

(2006). Does inequality matter for poverty reduction? Evidence from Pakistan's poverty trends. The Pakistan Development Review , 45(3), 439–459. https://doi.org/10.30541/v45i3pp.439-459

37.

Jimenez

, & Pate

M. A.

(2017). Reaping a demographic dividend in Africa’s largest country: Nigeria. In Groth

& May

J. F.

(Eds.), Africa's population: In search of a demographic dividend (pp. 33–51). Springer. https://link-springer-com-443.web.bisu.edu.cn/chapter/10.1007/978-3-319-46889-1_3

38.

Johansen

(1992). Cointegration in partial systems and the efficiency of single-equation analysis. Journal of Econometrics , 52(3), 389–402. https://doi.org/10.1016/0304-4076(92)90019-N

39.

Johansen

, & Juselius

(1990). Maximum likelihood estimation and inference on cointegration—with applications to the demand for money. Oxford Bulletin of Economics and Statistics , 52(2), 169–210. https://doi.org/10.1111/j.1468-0084.1990.mp52002003.x

40.

Knotek

E. S.

(2007). How useful is Okun's law. Economic Review-Federal Reserve Bank of Kansas City , 92(4), 73. https://ideas.repec.org/a/fip/fedker/y2007iqivp73-103nv.92no.4.html

41.

Kochar

(2007). The “youth bulge” in developing economies: Can this be an advantage in the labor market? In Jimenez

(Ed.), Development and the next generation (pp. 87–99). Berlin Workshop Series 2007. https://doi.org/10.1596/978-0-8213-6833-6

42.

Korenman

, & Neumark

(2000). Cohort crowding and youth labor markets: A cross-national analysis. In Blanchflower

D. G.

& Freeman

R. B.

(Eds.), Youth employment and joblessness in advanced countries . NBER Comparative Labor Markets Series (pp. 57–106). University of Chicago Press. https://doi.org/10.3386/w6031

43.

Lam

(2014). Youth bulges and youth unemployment. IZA World of Labor , 26, 1–10. https://wol.iza.org/articles/youth-bulges-and-youth-unemployment/long

44.

Lam

, & Leibbrandt

(2013). Global demographic trends and their implications for employment . Background Research Paper for United Nations High Level Panel on the Post-2015 Development Agenda, United Nations. https://www.post2020hlp.org/wp-content/uploads/docs/Lam-Leibbrandt_Global-Demographic-Trends-and-their-Implications-for-Employment.pdf

45.

Lee

(2000). The robustness of Okun's law: Evidence from OECD countries. Journal of Macroeconomics , 22(2), 331–356. https://doi.org/10.1016/S0164-0704(00)00135-X

46.

Leveson

(1980). Generational crowding: Economic, social and demographic effects of changes in relative cohort size (pp. 115–158). Hudson Institute.

47.

Moffat

, & Roth

(2017). Cohort size and youth labour-market outcomes in Europe. Economics Bulletin , 37(4), 2735–2740. http://www.accessecon.com/pubs/eb/default.aspx?topic=AbstractPaperID=EB-17-00465

48.

Mortensen

D. T.

, & Pissarides

C. A.

(1994). Job creation and job destruction in the theory of unemployment. The Review of Economic Studies , 61(3), 397–415. https://doi.org/10.2307/2297896

49.

Nardone

(1987). Decline in youth population does not lead to lower jobless rates. Monthly Labor Review , 110(6), 37–41.

50.

Newhouse

, & Wolff

(2014). Cohort size and youth employment outcomes . The World Bank. https://doi.org/10.1596/1813-9450-6848

51.

Nordstrom Skans

(2002). Age effects in Swedish local labour markets (Working Paper, 2002, 11). Institute for Labour Market Policy Evaluation (IFAU). https://dx-doi-org.web.bisu.edu.cn/10.2139/ssrn.319080

52.

Ochsen

(2009). Regional labor markets and aging in Germany (Thünen-Series of Applied Economic Theory Working Paper No. 102). Department of Economics, Faculty of Economics and Social Sciences, University of Rostock. http://hdl.handle.net/10419/39739

53.

Ochsen

(2011). Relative cohort size and unemployment in the United States . 4th World Conference of the Spatial Econometrics Association. https://paa2012.princeton.edu/papers/121008

54.

Odedokun

M. O.

, & Round

J. I.

(2004). Determinants of income inequality and its effects on economic growth: Evidence from African countries. African Development Review , 16(2), 287–327. https://doi.org/10.1111/j.1017-6772.2004.00093.x

55.

Ogbeide

F. I.

, Kanwanye

, & Kadiri

(2016). Revisiting the determinants of unemployment in Nigeria: Do resource dependence and financial development matter. African Development Review , 28(4), 430–443. https://doi.org/10.1111/1467-8268.12222

56.

Ogonna

I. C.

, Idenyi

O. S.

, Ifeyinwa

A. C.

, & Gabriel

N. U.

(2016). The implications of rising public debt on unemployment in Nigeria: An auto regressive distributed lag approach. Asian Research Journal of Arts & Social Sciences , 1(1), 1–15. https://doi.org/10.9734/ARJASS/2016/26394

57.

Okonjo-Iweala

, Soludo

C. C.

, & Muhtar

(Eds.). (2003). The debt trap in Nigeria: Towards a sustainable debt strategy . Africa World Press.

58.

Okun

A. M.

(1962). Potential GNP: Its measurement and significance . In Proceedings of the Business and Economic Statistics Section, pp. 98–104, American Statistical Association.

59.

Olaniyan

, Soyibo

, & Lawanson

A. O.

(2012). Demographic transition, demographic dividend and economic growth in Nigeria. African Population Studies , 26(2), 159–176. https://doi.org/10.11564/26-2-211

60.

Omoju

O. E.

, & Abraham

T. W.

(2014). Youth bulge and demographic dividend in Nigeria. African Population Studies , 27(2), 352–360. https://doi.org/10.11564/27-2-480

61.

Onodugo

V. A.

, Obi

K. O.

, Anowor

O. F.

, Nwonye

N. G.

, & Ofoegbu

G. N.

(2017). Does public spending affect unemployment in an emerging market. Risk Governance & Control: Financial Markets & Institutions , 7(1), 32–40. https://doi.org/10.22495/rgcv7i1art4

62.

Organization for Economic Co-operation and Development (OECD). (1980). Youth unemployment: Causes and consequences . OECD.

63.

Perry

G. L.

(1970). Changing labor markets and inflation. Brookings Papers on Economic Activity, Economic Studies Program, The Brookings Institution , 1(3), 411–448.

64.

Perry

G. L.

, Schultze

, Solow

, & Gordon

R. A.

(1970). Changing labor markets and inflation. Brookings Papers on Economic Activity , 1970(3), 411–448.

65.

Pesaran

M. H.

, & Shin

(1998). An autoregressive distributed-lag modelling approach to cointegration analysis. Econometric Society Monographs , 31, 371–413. https://doi.org/10.1017/CCOL521633230.011

66.

Pesaran

M. H.

, Shin

, & Smith

R. J.

(2001). Bounds testing approaches to the analysis of level relationships. Journal of Applied Econometrics , 16(3), 289–326. https://doi.org/10.1002/jae.616

67.

Phillips

A. W.

(1958). The relationship between unemployment and the rate of change of money wages in the United Kingdom 1861–1957. Economica , 25(100), 283–299. https://doi.org/10.2307/2550759

68.

Phillips

P. C.

, & Perron

(1988). Testing for a unit root in time series regression. Biometrika , 75(2), 335–346. https://doi.org/10.1093/biomet/75.2.335

69.

Raifu

I. A.

(2017). On the determinants of unemployment in Nigeria: What are the roles of trade openness and current account balance? Review of innovation and competitiveness. A Journal of Economic and Social Research , 3(4), 5–30. https://doi.org/10.32728/ric.2018.34/1

70.

Reid

, & Smith

D. A.

(1981). The impact of demographic changes on unemployment. Canadian Public Policy/Analyse de Politiques , 7(2), 348–351. https://doi.org/10.2307/3549234

71.

Rodriguez-Palenzuela

, & Jimeno-Serrano

J. F.

(2002). Youth unemployment in the OECD: demographic shifts; Labour market institutions, and macroeconomic shock (Working Paper Series 155). European Central Bank.

72.

Rosen

(1972). Learning and experience in the labor market. Journal of Human Resources , VII(3), 326–342. https://doi.org/10.2307/145087

73.

Roth

(2017). Cohort size and transitions into the labour market (MAGKS Papers on Economics 201708). Philipps-Universität Marburg, Faculty of Business Administration and Economics, Department of Economics (Volkswirtschaftliche Abteilung), pp. 1–40. https://ideas.repec.org/p/mar/magkse/201708.html

74.

Russell

L. B.

(1982). The baby boom generation and the economy. Brookings Studies in Social Economics. Washington, D.C.: Brookings, 1982. Pp. xiii, 183. The Journal of Economic History , 44(1), 224–226. https://doi.org/10.1017/S0022050700031739

75.

Schmidt

C. M.

(1993). Ageing and unemployment. In Johnson

(Ed.), Labor markets in an ageing Europe (pp. 216–252). Cambridge University Press.

76.

Shimer

(2001). The impact of young workers on the aggregate labor market. The Quarterly Journal of Economics , 116(3), 969–1007. https://doi.org/10.1162/00335530152466287

77.

Simion

S. R.

(2015). Demographic bulges and labor market outcomes (Doctoral Dissertation). Queen Mary University of London. https://econ.ubbcluj.ro/ermas2015/lucrari/A2_Stefania_Simion.pdf

78.

Sippel

, Kiziak

, Woellert

, & Klingholz

(2011). Africa’s demographic challenges: How a young population can make development possible . Berlin Institute for Population and Development. https://www.berlin-institut.org/en/publications/studies-in-english/africas-demographic-challenges.html

79.

Sir

A. T.

(2014). Impact of inflation and economic growth on unemployment in Sri Lanka: A study of time series analysis. Global Journal of Management and Business Research , 13(5), 1–11 https://www.journalofbusiness.org/index.php/GJMBR/article/view/1346

80.

Skans

O. N.

(2005). Age effects in Swedish local labor markets. Economics Letters , 86(3), 419–426. https://doi.org/10.1016/j.econlet.2004.09.004

81.

Soyibo

, Olaniyan

, & Lawanson

A. O.

(2008). National transfer accounts of Nigeria, 2004 . Draft research report submitted to the East-West Center.

82.

United Nations Department of Economic and Social Affairs Population Division. (2015). Youth population trends and sustainable development. Population Facts , 2015(1), 1. https://www.un.org/esa/socdev/documents/youth/fact-sheets/YouthPOP.pdf

83.

Villaverde

, & Maza

(2009). The robustness of Okun's law in Spain, 1980–2004: Regional evidence. Journal of Policy Modeling , 31(2), 289–297. https://doi.org/10.1016/j.jpolmod.2008.09.003

84.

Wachter

M. L.

, Hall

R. E.

, & Holt

C. C.

(1976). The changing cyclical responsiveness of wage inflation. Brookings Papers on Economic Activity , 1976(1), 115–167.

85.

Welch

(1979). Effects of cohort size on earnings: The baby boom babies' financial bust. Journal of Political Economy , 87(5, Part 2), S65–S97. https://doi.org/10.1086/260823

86.

World Bank International Economics Dept Development Data Group. (2019). World development indicators . World Bank. https://databank.worldbank.org/reports.aspx?source=world-development-indicators

87.

Zdravković

, Đukić

, & Bradić-Martinović

(2017). Impact of FDI on unemployment in transition countries: Panel cointegration approach. Industrija , 45(1), 161–174. https://doi.org/10.5937/industrija45-13548

88.

Zimmermann

K. F.

(1991). Ageing and the labor market. Journal of Population Economics , 4(3), 177–200. https://doi.org/10.1007/BF00602428

89.

Zimmermann

K. F.

(1992). Ageing and the labor market. In Johnson

(Ed.), Fiscal implications of an aging population (pp. 113–136). Springer. https://link-springer-com-443.web.bisu.edu.cn/chapter/10.1007/978-3-642-77250-4_6