Oil Prices, Labour Market Institutions, and Unemployment: Evidence from African Oil-Exporting Economies

Abstract

The volatility of oil prices has a considerable impact on the economies of oil-exporting countries, making it critical to understand how price variations affect labour markets and unemployment. This study investigates the distinct role of labour market institutions in moderating the effects of oil price volatility on unemployment. Using the Cross-Sectionally Augmented Autoregressive Distributed Lag Model (CS-ARDL) on a panel dataset of nine African oil-exporting countries from 1994 to 2024, the study establishes a strong negative link between oil price changes and unemployment. Furthermore, the results show that real GDP growth leads to a reduction in unemployment in the long run, while the labour market institutional index has a negative impact on unemployment. Interacting the oil price with the labour market institutional index causes a further reduction in unemployment. These results suggest that good labour market institutions and macroeconomic stability are essential for reducing unemployment. While increases in oil prices directly stimulate a reduction in unemployment in African oil-exporting countries, this impact is reinforced by the presence of good labour market institutions in an economy. Therefore, the results suggest that countries with strong labour market institutions are more resilient in reducing the negative impact of oil price volatility on employment. As such, policymakers must prioritise labour market institutional reforms to enhance countries’ capacity to absorb oil price shocks and reduce unemployment during periods of oil prosperity and shield against employment declines when oil prices drop. Furthermore, the creation of oil stabilisation funds in these countries may serve a similar purpose. Contribution/originality: Against a background of inconclusive empirical evidence in the literature and a dearth of research on African countries, this study investigates the role of labour market institutions (LMIs) in the oil price–unemployment nexus in African oil-exporting countries. While highly dependent on oil revenue, these countries record persistent structural unemployment. Therefore, the study provides critical evidence to guide the formulation of policies necessary to deal with external shocks and facilitate structural shifts required for employment growth. Existing studies consider general institutional variables such as democratic accountability and the rule of law and do not assess the effect of labour market institutions. The current study fills in this gap by assessing the distinct role of labour market institutions that are specifically designed to regulate only work-related activities, such as quality of labour regulations, adequacy of social protection and unemployment benefits. Furthermore, this study employed the cross-sectionally augmented autoregressive distributed lag (CS-ARDL) for econometric estimations. Compared to previous studies, this is a more appropriate method that accounts for unobserved common factors such as oil price shocks affecting all oil-exporting countries simultaneously.

1. Introduction and Background

The role of energy as a fundamental resource and an important driver in the growth process in the global economy is undisputed. As such, any changes in the global price of oil, a major source of energy, have significant implications for macroeconomic fundamentals in both oil-importing and oil-exporting nations. The implications are more crucial for oil-exporting economies given that oil revenue influences fiscal policy and overall macroeconomic stability. Additionally, the impact of oil price fluctuations on macroeconomic fundamentals has become more important and compelling to policymakers in oil-dependent countries in Sub-Saharan Africa. Despite its resource endowments, the region faces a high unemployment rate, one of the most critical global economic problems, and its concomitant challenges1.

Theoretically, an increase in oil prices translates to increases in oil revenue, improved balance of payment, higher consumption, and investment in the oil sector, which potentially creates jobs in extractive and related industries. Similarly, higher oil revenues lead to increased economic activity and, consequently, higher economic growth and employment across different sectors in oil-exporting countries. On the contrary, sharp declines in oil prices can lead to reduced investments, fiscal constraints, and job cuts in the oil industry, consequently affecting other sectors and employment opportunities.

There is existing empirical evidence linking positive oil price shocks to job creation and consequently lower unemployment (Alfalih, 2024; Adeosun et al., 2023; Tien, 2022). For instance, Karlsson et al. (2018) found a negative relationship between oil price increases and unemployment in Norway, while Tien (2022), using data from Vietnam, also found that oil price increases have a significant negative impact on unemployment. Another strand of the literature found evidence linking positive oil price shocks to higher unemployment (Abdelsalam, 2023; Zivkov & Duraskovic, 2023; Tien & Hung, 2022). These studies argue that rising oil prices can significantly increase unemployment when, for instance, the price increase results in currency appreciation that reduces the competitiveness of other export industries, leading to job losses, called the Dutch disease effect (Adeosun et al., 2023; International Monetary Fund, 2016). Similarly, oil price increases can lead to inflation, which reduces competitiveness in other sectors and potentially leads to job losses in those sectors, thereby increasing unemployment.

Therefore, the literature presents two opposing perspectives on the impact of oil price changes on unemployment: one suggesting that a positive oil price shock reduces unemployment, while the second suggests that rising oil prices lead to more job losses.

The third group of studies focused on identifying the channels through which oil price fluctuations are transmitted to affect unemployment (African Union, 2015; O. K. Kocaarslan, 2019). Several variables, such as economic diversification, trade openness, institutional quality, FDI, and labour market regulations, have been identified as factors that mitigate the positive impact of oil price increases on unemployment (African Development Bank, 2024).

Motivated by these inconclusive results in the literature, this study investigates the distinct role of labour market institutions in moderating the effects of oil price volatility on unemployment in nine African oil-exporting countries, including Angola, Algeria, Cameroon, Egypt, Gabon, Niger, Nigeria, the Republic of Congo, and Sudan. While studies such as O. K. Kocaarslan (2019) suggest that different countries experience the effects of oil price changes on unemployment differently due to the quality of the institutions, previous research has not adequately investigated how labour market institutions can moderate the effects of oil price volatility on unemployment in African oil-exporting countries. This study addresses this gap by focusing specifically on the role of these institutions in a region that has not been well-studied. It argues that the different results found in studies on the link between oil price shocks and unemployment are likely due to the different labour market institutions in each country. Therefore, the study hypothesises that the impact of oil price changes on employment is moderated by the presence of labour market institutions. The study seeks to answer the following research questions:

What is the effect of oil price shocks on unemployment in Africa’s oil-exporting countries? Does the presence of good labour market institutions moderate the effect of oil price shocks on unemployment? Systematically analysing how different institutional settings influence the transmission of oil price shocks on the labour market provides valuable insights into the effectiveness and potential improvements of labour market institutions in mitigating the adverse effects of oil price fluctuations on employment.

This study is important for net oil-exporters in Africa and other developing countries for three compelling reasons. Firstly, these countries suffer from persistently high unemployment, a structural issue that is frequently aggravated by weak LMIs that fail to properly manage the labour force and absorb shocks (International Labour Organization, 2023); therefore, addressing this institutional weakness is critical for focused reform. Secondly, because these economies rely heavily on oil, their labour markets are extremely exposed to volatile oil price swings (OPEC, 2023), and therefore, it is essential to determine how LMIs may be effectively exploited to insulate workers and stabilise employment. Finally, the results from this study are critical for oil-exporting African governments’ initiatives to diversify economies from natural resource dependency (African Union, 2015). The findings will provide policymakers with critical, evidence-based guidance on reforming specific LMIs to both cushion the labour market against external shocks and facilitate the structural shifts required for long-term growth and resilience across all oil and non-oil industries.

While the influence of oil price variations on unemployment has been researched in both developed and rising nations, African economies have been mostly disregarded. This study has become even more relevant and timely, considering new findings showing high levels and incidences of poverty strongly linked to unemployment in Africa (Zaman et al., 2023). The rest of this paper is organised as follows. The next session presents a brief description of oil prices and unemployment trends in the studied countries. Section 2 provides a review of the related literature, Section 3 presents the empirical methodology adopted in this study, while Section 4 presents the results, summary, conclusion and policy recommendations.

Stylised Facts

Africa’s crude oil exports rose from 4.8 million barrels per day in 1995 to a high of 7.53 million barrels per day (mbd) in 2010, with the top exporters of crude oil being Nigeria, Angola, Libya, Algeria and the Republic of Congo. However, crude oil exports dropped significantly to 4.91 mbd in 2024 from their 2010 level. This was mainly attributed to the reduced demand from the US, volatile and lower oil prices and the consequent effects of the COVID-19 pandemic. On the contrary, oil exports from the Middle East, a region with the highest crude oil exports in the world, rose from 13.82 mbd in 1995 to 16.44 in 2024.

Regarding unemployment patterns, African oil-exporting countries have historically recorded higher unemployment rates than the world average (International Labour Organization, 2023). Between 1990 and the early 2000s, many of these countries experienced relatively high unemployment rates, with Nigeria having an average jobless rate of 10.5% (World Bank, 2023). Similarly, the average unemployment rate for the 2013 to 2023 period was around 12%, with Nigeria and Libya having the lowest of 4.1% and the highest of 19.2%, respectively (International Monetary Fund, 2024). On the other hand, lower unemployment rates for some oil exporters were attributed to the oil boom in the mid-2000s that spurred tremendous economic growth and job creation. For instance, Angola’s unemployment rate dropped from 26% in 2000 to 7.3% in 2014 (World Bank, 2023). However, the oil price crash from 2014 to 2016 had severe consequences for many African oil exporters, leading to economic contractions, fiscal deficits, and rising unemployment. These statistics sharply contradict global unemployment rates that averaged 6% over the 1990 to 2023 period.

On the other hand, the quality of labour market institutions in African oil-exporting economies varies significantly across countries, with notable differences in labour market flexibility, wage structures, education and skills development programmes, and social safety nets. For example, a 2018 World Bank report highlighted Nigeria’s relatively flexible labour market, characterised by a large informal sector and limited minimum wage regulations. In contrast, Algeria’s labour market is known for its rigidities, including extensive regulations on hiring and firing, which could hinder job creation and economic dynamism (OECD, 2020). Additionally, the strength of social safety nets varies across countries, with some providing comprehensive unemployment benefits and social assistance programmes, while others have limited coverage and inadequate funding (International Labour Organization, 2019). For instance, Nigeria and Angola faced significant challenges in establishing comprehensive social protection due to high levels of informality and weak administrative capacity, leaving large segments of their populations highly exposed to the full brunt of oil price downturns (International Monetary Fund, 2024; UNDP, 2025).

These variations in labour market institutions have significant implications for employment outcomes, wage levels, and how economies respond to oil price shocks, thereby determining the resilience and adaptability of their labour markets. Countries with flexible labour markets, strong education systems, and robust social safety nets are generally better equipped to adjust to oil price fluctuations and mitigate their negative employment effects. Conversely, countries with rigid labour markets, weak education systems, and inadequate social protection are more vulnerable to oil price shocks and may experience persistent unemployment and social unrest. Therefore, labour market institutions influence labour market flexibility, wage determination, skill development, and social protection mechanisms, all of which can moderate or exacerbate the effects of oil price shocks on unemployment levels.

2. Literature Review

2.1. Theoretical Literature

An effective theoretical framework that combines macroeconomic linkages with the micro-foundations of labour market dynamics is necessary to comprehend the intricate interactions between changes in oil prices and unemployment. Several important theoretical frameworks serve as the foundation for this study, each of which provides a unique perspective for examining this crucial link, especially in economies that export oil.

Firstly, a basic macroeconomic connection is offered by Okun’s Law (Okun, 1963), which asserts an empirically inverse relationship between changes in the unemployment rate and the growth rate of real GDP. Shocks to oil prices have a direct effect on GDP growth in countries that export oil; a drop in oil prices can cause an economic contraction, which raises unemployment in accordance with Okun’s Law. This study builds on this fundamental knowledge by analysing how the Okun coefficient, which measures how responsive unemployment is to changes in output, is not constant but rather mediated by the institutions that currently govern the labour market. In contrast to more flexible labour markets where adjustments may be quicker and more noticeable, strict employment protection laws (EPL) may cause labour hoarding, which would initially reduce the immediate unemployment response to negative shocks but could ultimately result in greater unemployment persistence or reduced job creation.

Secondly, the classical and neoclassical labour market theories offer micro-foundations to explain the persistence of unemployment. According to the classical approach, markets clear automatically at the real wage, suggesting that any protracted unemployment is either frictional or voluntary, induced by rigidities such as trade union involvement or minimum wage regulations, which keep the real wage above the market-clearing level. The neoclassical method refines this by claiming that firms will hire workers until the real pay matches the marginal output of labour. From this standpoint, an oil price crash that affects overall economic productivity necessitates a lower real wage to preserve employment. When institutional constraints such as inflexible EPL prohibit actual wages from declining, businesses respond with involuntary layoffs rather than wage reduction. This supply-side perspective is critical for understanding why oil-exporting countries’ persistent structural unemployment in their labour market institutions has prevented the pay flexibility needed to adapt to productivity shocks caused by oil price volatility.

Thirdly, theoretical assertions attribute the negative impact of oil price increase on employment to the Dutch Disease experienced by resource-rich countries. The influx of oil revenues following a positive oil price shock in oil-rich nations strengthens the local currency. Such an increase in exchange rates results in exports from other non-oil tradeable sectors becoming more expensive and uncompetitive internationally, leading to the de-industrialisation of these sectors and consequent job losses.

Lastly, the vast body of research on policies and institutions demonstrates how institutions such as good governance, transparency and upholding the rule of law enhance proper resource allocation and lower transaction costs in resource-rich countries. On the contrary, poor institutions would foster rent-seeking behaviour, leading to inefficiencies, economic declines and job losses in these countries. Similarly, certain labour market institutions2 influence unemployment outcomes in response to external shocks such as changes in the price of oil. For instance, rigid employment protection legislation (EPL) (Nickell, 1997; Blanchard & Wolfers, 2000) could impede job creation during upturns but postpone layoffs during downturns. On the other hand, strong social safety nets can reduce the social costs of unemployment, promote easier labour market transitions, and avoid long-term unemployment by protecting human capital (International Labour Organization, 2019).

2.2. Conceptual Framework

The review presented above is the basis for the conceptual framework on how changes in oil prices affect unemployment rates in African oil-exporting countries. The main theme is that the oil price–unemployment relationship is greatly influenced by high-quality labour market institutions. These institutions enhance the positive relationship between oil price increases and job creation through their flexibility and effectiveness. On the other hand, these high-quality institutions are also expected to reduce the adverse effects of a negative shock by ensuring smoother labour market operations, thereby reducing sudden increases in unemployment. In addition, real gross domestic product is incorporated as a control variable to isolate the structural effects of the oil price and institutional quality from the overall macroeconomic cycle, thereby supporting the theoretical underpinning of Okun’s Law’s principles.

Through the integration of these theoretical insights and the derived conceptual framework, this paper develops a thorough empirical framework for analysing how various labour market institution configurations in oil-exporting economies interact with changes in oil prices to influence unemployment outcomes. This, in turn, illuminates effective policy levers for promoting sustainable development and strengthening labour market resilience in resource-dependent countries.

2.3. Empirical Literature

Several studies have examined the general impact of oil price changes on unemployment rates, but the findings are conflicting. Using wavelet methodologies on data from the US, Canada, France, Italy, the UK, Germany and Japan, Adeosun et al. (2023) reported that oil prices and uncertainty significantly drive unemployment. Similarly, Ahmed et al. (2023) employed the logistic smooth transition autoregressive (LSTAR) process and found that high oil price volatility states increase levels of unemployment in the United States. A study by Koirala and Ma (2020) applied a bivariate GARCH-in-mean VAR model and reported an increase in unemployment following a positive price shock in the United States for the period 1974 and 2018. Similarly, O. K. Kocaarslan (2019) used a GARCH-in-mean VAR model for the period 1974 to 2017 in the US and results confirmed that oil price uncertainty is positively related to unemployment. In a separate study, B. Kocaarslan et al. (2020) used a nonlinear autoregressive distributed lag (NARDL) on US data and concluded that oil price increases further increase unemployment in oil sectors while having no significant effect on unemployment following an oil price decline.

Several empirical studies used data from other regions outside the US to analyse the relationship between oil price shocks and unemployment. For instance, Cuestas and Gil-Alana (2018) employed the NARDL model on data from Central and Eastern European countries for the period from 2000 to 2015. The study found that oil prices and the natural rate of unemployment flow in the same direction: a positive oil price shock increases unemployment, while decreases in the oil price reduce unemployment. Similarly, Raifu et al. (2020) found a long-run positive relationship between oil price and unemployment in Nigeria after applying a non-linear ARDL. Cheratian et al. (2019) also employed the NARDL model on data from oil-importing and oil-exporting countries in the MENA region. The study reveals that an increase has a long-run positive effect on unemployment in both oil-importing and oil-exporting countries.

On the other hand, some studies found a negative relationship between oil price and unemployment. For instance, a study by Tien (2022) indicates that an increase in oil prices led to a decline in the unemployment rate in Vietnam between 1999 and 2020. This was supported by Nusair (2020) in a study that was carried out in both Canada and the US. Karlsson et al. (2018) also found a negative relationship between oil price increases and unemployment in Norway between 1997 and 2015. Similarly, Alfalih (2024) adopted the ARDL model using time series data for the period 1991 to 2019 from Saudi Arabia and found a negative relationship between oil price increases and unemployment. In the case of the Gulf Cooperation Council (GCC) countries, Cheikh et al. (2018) suggest that higher oil prices can lead to lower unemployment due to increased economic activity and government spending.

The findings from the reviewed empirical studies highlight the complex and context-dependent nature of the relationship between oil prices and unemployment, showing that under certain conditions, positive oil price shocks can indeed reduce or increase unemployment in oil-exporting economies.

While some studies have explored the role of institutional quality on the oil price–unemployment nexus, the specific role of labour market institutions remains unclear, especially within the context of African oil-exporting economies. Raifu (2021) employed Pooled OLS and panel ARDL on data from 24 oil-exporting African and Asian countries for the 1991 to 2019 period. The study investigated the role of institutional quality on the oil price–unemployment nexus. The results revealed that the effect of an increase in oil prices on unemployment is enhanced by the presence of good institutions such as democratic accountability and the rule of law. However, the study did not assess the effect of labour market institutions. Similarly, Agboola et al. (2024) found that increasing institutional quality minimises the negative impact of oil price shocks on macroeconomic variables in a panel of eight non-OECD emerging economies. This suggests that the effectiveness of oil price changes in reducing unemployment may depend on the quality of labour market institutions.

Table 1. Summary of empirical studies reviewed.

Authors	Scope	Methods	Findings: Relationship Between Oil Price and Unemployment
Alfalih (2024)	Saudi Arabia 1991 to 2019	ARDL	Negative
Ahmed et al. (2023)	USA 1974 to 2019	Logistic smooth transition autoregressive (LSTAR)	Positive
Tien (2022)	Vietnam 1999 to 2020	Quantile-on-quantile regression	Negative
Raifu (2021)	28 African and Asian oil-importing countries 1991 to 2019	POLS and PARDL	POLS: Negative for Asian economies PARDL: Negative in the short run, positive in the long run
Cuestas and Gil-Alana (2018)	Central and Eastern European countries 2000 to 2015	NARDL	Unidirectional
B. Kocaarslan et al. (2020)	USA Monthly data from 2007:M5 to 2019:M4	NARDL	Positive
Koirala and Ma (2020)	USA Monthly data from 1974 M2 to 2018 M11.	Bi-variate GARCH-in mean VAR	Positive
Nusair (2020)	Canada, USA Monthly data: 1960:M1 to 2018:M4.	NARDL	Negative
Raifu et al. (2020)	Nigeria 1979 Q1 to 2018 Q4	Linear and non-linear ARDL	Positive in the long run Long-run asymmetric relationship
Cheratian et al. (2019)	MENA 1991 to 2017	NARDL	Positive short-run impact on oil-exporting nations. Positive long-run effect on oil-importing and oil-exporting nations
O. K. Kocaarslan (2019)	US 1974 to 2017	GARCH-in-mean VAR model	Positive
Karlsson et al. (2018)	Norway 1997 to 2015	Wavelet multi-resolution analysis	Negative

below summarises the key studies reviewed above.

The existing literature provides a foundation for understanding the complex relationship between oil prices and unemployment. However, there is a need for further research to specifically examine the role of labour market institutions in oil-exporting economies, particularly within the African context. This research can help policymakers design and implement effective labour market policies to mitigate the negative impacts of oil price fluctuations on employment and promote sustainable economic development.

3. Research Methodology

3.1. Theoretical Framework

A simple theoretical model of unemployment as a function of oil price fluctuations and other determinants, as reviewed in the literature, is specified as follows.

$$UN=f(OP, LMII, RGDP,)$$

(1)

where UMP is unemployment, OP represents oil price, LMII is labour market institutional index, and RGDP is the real GDP.

3.2. Empirical Model

The study used the Cross-sectionally Augmented Autoregressive Distributed Lag (CS-ARDL) model to investigate how Labour Market Institutions (LMIs) influence the link between oil prices and unemployment in oil-exporting economies. This study estimates CS-ARDL, developed by Chudik and Pesaran (2015). This study favours the CS-ARDL over conventional linear panel ARDL because it effectively accounts for unobserved common factors by incorporating cross-sectional averages in the ARDL framework. Furthermore, the CS-ARDL model estimates long-run and short-run relationships on data that exhibits cross-sectional dependence, non-stationarity and slope heterogeneity. CS-ARDL also works well with series that are integrated of different orders, I (0) and I (1). Since African oil-exporting countries depend heavily on crude oil for foreign exchange earnings, the countries are exposed to common oil price shocks. In addition, the price of crude oil is determined globally, and therefore any price shock affects all the oil-exporting countries simultaneously. This legitimises the selection of CS-ARDL. In other words, the CS-ARDL is the most appropriate technique given that African oil-exporting countries are exposed to common shocks. Adopting the CS-ARDL approach solves the problems of cross-sectional dependence and heterogeneity in the data.

The specific CS-ARDL used in this study is as follows.

$${UN}_{i,t}={\beta Y}_{i,t-1}+{\beta }^{\prime }{X}_{i,t}+{\pi }_i+{\epsilon }_{i,t}$$

(2)

where is the ${UN}_{it}$ unemployment, which is the dependent variable, $X_{it}$ represents a set of explanatory variables, ${\pi }_i$ denotes unobserved country-specific effects, and denotes other factors not included in the model. Subscripts t and i represent time in years and country, respectively. Specifically, $X_{it}$ incorporates oil price, LMII, OP, OP-LMII as well as other determinants of unemployment such as RGDP. This model is generalised to include the following equation.

$$\begin{gathered} {UN}_{it}=\sum _{j=1}^p{\Pi }_{1, i, j}{UN}_{1, t-j}+\sum _{J=0}^{q_1}{\beta }_{1, i, j}{OP}_{i, t-j}+ \sum _{j=0}^{q_2}{\gamma }_{2, i, j}{LMII}_{i, t-j}+\sum _{j=0}^{q_3}{\Phi }_{3, i, j}{RGDP}_{i, t-j}\sum _{j=0}^{q_4}{\vartheta }_{4, i, j}{OP\ast LMII}_{i, t-j} \\ \hspace{1em}\hspace{1em}\hspace{1em}\hspace{1em}\hspace{1em}+\sum _{J=0}^p{\overline{UN}}_{t-j}+\sum _{j=0}^{q_1}{\overline{OP}}_{t-j}{\underset{\_}{OP}}_{t-j}+\sum _{j=0}^{q_2}{\overline{LMII}}_{t-j}+\sum _{j=0}^{q_3}{\overline{RGDP}}_{t-j}\sum _{j=0}^{q_4}{\overline{OP\ast LMII}}_{t-j}+{\mu }_{i, t} \end{gathered}$$

(3)

where $i$ is the number of oil-exporting economies, $t$ represents a 30-year period from 1995 to 2025, ${\epsilon }_{i,t}$ is the error term and ${\gamma }_{ij}$, ${\pi }_{ij}$, ${\Phi }_{i, j}$ ${ß}_{ij}$, ${\vartheta }_{ij}$, ${\phi }_{ij}$, are long-run coefficients. Furthermore, UN: unemployment rate, OP: oil price, LMII: labour market institutions, Index OP*LMII is the interaction term, and ${\alpha }_i$: Country-specific fixed effects

$$\begin{gathered} {∆UMP}_{it}={\alpha }_i+{\lambda }_1({UMP}_{it-1}-{\theta }_0-{\theta }_1{OP}_{it-1}-{\theta }_2{LMII}_{it-1}-{\theta }_3({OP\ast LMII)}_{it-1} \\ \hspace{1em}\hspace{1em}\hspace{1em}\hspace{1em}\hspace{1em}-{\theta }_{4}R{GDP}_{it-1})+{\sum }_{j=1}^p{\gamma }_{i,j}{∆UMP}_{i,t-j}+{\sum }_{k=0}^{q-1}{\delta }_{1,i,j}{∆OP}_{i,t-j} \\ \hspace{1em}\hspace{1em}\hspace{1em}\hspace{1em} +{{\sum }_{k=0}^{q-1}{\pi 2,}_{i,j}∆LMII}_{i,t-j}+{\sum }_{k=0}^{q-1}{\vartheta 3,}_{i,j}∆({OP\ast LMII)}_{i,t-j} \\ +{\sum }_{k=0}^{q-1}{\phi 4,}_{i,j}{∆RGDP}_{i,t-j}+{\epsilon }_{i,t} \end{gathered}$$

(4)

where ${\lambda }_1$ is the error correction term coefficient, which represents the speed of adjustment of UMP back to the long-run equilibrium following a shock, and ${\theta }_0$:, ${\theta }_1$:, ${\theta }_2$:, ${\theta }_3$: ${\theta }_4$ are the long-run coefficients.

These equations can be estimated using the CS-ARDL methodology, which allows for the inclusion of both I (0) and I (1) variables. The optimal lag lengths (p and q) can be determined using information criteria such as AIC or BIC.

The analysis makes use of a balanced panel dataset that spans nine African oil-exporting countries from 1994 to 2024. The use of a panel dataset is critical because it accounts for both cross-sectional heterogeneity and time-series dynamics, resulting in more robust estimates than static models. The period 1994 to 2024 was chosen specifically because it encompasses the whole cycle of key macroeconomic shocks, including the massive oil boom, the 2014 to 2016 oil price fall, and the COVID-19 pandemic, as well as the long-term effects of structural and labour market institutional reforms.

3.3. Data

The data for the selected variables were sourced from the World Bank’s World Development Indicators (WDI), except for oil price, which has been sourced from Global Primary Energy Data (GPED). The study focused on African oil-exporting economies over the period 1994 to 2024, a range determined by data availability. To analyse the short- and long-term relations among variables across the sampled countries, the CS-ARDL technique was employed.

Table 2. Data and sources of variables.

Variables	Abbreviation	Data Metrics	Source
Unemployment	UN	% of unemployed to labour force	WDI
Real Gross Domestic Product	RGDP	Natural logs of absolute figures	WDI
Oil Price	OP	Natural logarithms of absolute prices	GPED
Labour Market Institutional Index	LMII	Index number	WDI

provides a summary of specific data metrics and their sources.

Labour Market Institutional Index

The labour market institutional index was constructed using principal component analysis (PCA) based on Collective Bargaining Coverage rate (CBCR), Adequacy of Social Protection and labour programmes (ASPLP), Active Labour market Policies (ALMP), and the Coverage of unemployment benefits (CUB).

Table 3. LMII constructs.

Pillar	Proxy	Source
ASPLP	transfer received by population participating in social safety as % to total welfare	WDI
ALMP	% of total welfare of beneficiary households	WDI
CBCR	% of employees whose labour conditions are determined by collective agreements	WDI
CUB	% of population participating in unemployment compensation	WDI

Source: World Bank (2025).

shows the proxy indicators and the source of the constructs of the LMII.

To ensure that variables measured on different scales do not disproportionately influence the index, a critical pre-processing step is the standardisation of each indicator. This is achieved by calculating Z-scores, which transform each variable to have a mean of zero and a constant standard deviation of one. The mathematical expression for this transformation is

$$Z_i={\displaystyle \frac{X_i-{\cup }_i}{{\sigma }_i}}$$

The dimensionality of the data is reduced by employing the PCA. This method is effective for institutional data as it accounts for inherent correlations between different labour market policies. The LMII is derived from the first component, which accounts for Y% of the total variance. The factor loadings have been used to determine the weight of each component. The standardisation of each variable was done before the calculation of the PCA through the following formula:

$${PC}_1={\sum }_{i=1}^n{w}_1{Z}_i$$

Finally, the min-max normalisation was then used to rescale the final index to enhance the interpretability of results and allow meaningful cross-country comparisons. The min-max normalisation was used to rescale the PC1 score. This allowed the transformation of the index into a standardised range between 0 and 1, where values closer to 1 represent good quality labour market institutions. The rescaling is as follows

$$LMII={\displaystyle \frac{{PC}_1-\min \left({PC}_1\right)}{\max \left({PC}_1\right)-\mathrm{m}\mathrm{i}\mathrm{n}\left({PC}_1\right)}}$$

This LMII provides a composite measure that reflects the overall structural environment of the labour market. The reliability of the LMII was carried out using sensitivity analysis, where each of the variables was left out before recalculating.

Table 4. Correlation matrix of sub-indices.

Excluded Construct	Correlation with Baseline Equation	Impact on Variance Explained
Baseline equation	1	65.78%
ASPLP	0.92	66.64%
ALMP	0.93	67.23%
CBCR	0.85	68.36%
CUB	0.88	64. 66%

Source: Author Compilation, 2025.

below indicates the robustness of the LMII. This is indicated by the high value of the correlation between the baseline index and the reduced versions of the baseline equations. This implies that LMII captures the general institutional landscape of the labour market even if one of the constructs is excluded. Moreover, the stability of the variance explained by the first principal component remained relatively constant across all iterations, thereby confirming the robustness of the LMII.

4. Empirical Results

This section summarises the findings of econometric research conducted to investigate the short-term and long-term links between unemployment, oil price, labour market institutions, and economic output (RGDP) in the sampled African oil-exporting economies. The discussion follows a methodical format, beginning with descriptive statistics and correlation analysis to describe the data. This is followed by important panel unit root and cointegration tests to choose the suitable econometric framework for investigation. Finally, the short-run dynamics and long-run coefficients from the cross-sectionally augmented Autoregressive Distributed Lag (CS-ARDL) model are provided and interpreted to address the study’s fundamental objectives in terms of equilibrium and adjustment mechanisms.

4.1. Descriptive Statistics

The data in

Table 5. Descriptive statistics.

Variable	Observations	Mean	Standard Deviation	Minimum	Maximum
UN	270	0.84	0.32	0.51	1.67
Oil price	270	1.56	0.31	1.09	2.04
LMII	270	0.02	0.94	0.01	0.89
RGDP	270	8.23	0.64	−1.44	10.38

reveal that the average log unemployment rate is 0.84, with a standard deviation of 0.32, indicating substantial volatility around the mean. The difference between the minimum log unemployment rate of 0.51 and the maximum of 1.67 shows that unemployment rates vary greatly across the sample but tend to cluster around the mean. The average price for log oil is $1.56. With a low standard deviation of 0.31, the data shows that oil prices are largely steady across the sample period and countries, with a minimum of 1.09 and a maximum of 2.04. The average labour market institution index (LMII) score is 0.02, which is very close to zero, indicating that labour market institutions in these economies are generally poor. However, the high standard deviation of 0.94 and the wide range of 0.01 to 0.89 show that institutional strength varies substantially, with some countries having far stronger institutions than others. The average log real GDP (RGDP) is 8.23, suggesting a positive economic outlook. The standard deviation of 0.64 shows moderate fluctuations in economic output. The wide range, from −1.44 to 10.38, indicates that, while most countries’ economies are thriving, some have experienced significant economic hardship.

4.2. Correlation Analysis

Table 6. Correlation matrix results.

	UN	Oil Price	LMII	RGDP
UMP	1	−0.216	−0.046	−0.394
Oil price	−0.216	1	−0.301	0.291
LMII	−0.467	−0.301	1	−0.072
RGDP	−0.394	0.291	−0.072	1

Source: Own figure drawn from figures from the World Bank (2025).

indicates that the variables are most likely weakly correlated because the panel data contains cross-country heterogeneity, which obscures country-specific dynamics. Furthermore, the weak simple correlation reflects the critical role of labour market institutions as mediators, reducing the direct link between oil prices and unemployment. Finally, the observed relationship is most likely non-linear or subject to time lags that simple correlation analysis cannot detect. It can be confidently stated that there is no multicollinearity.

4.3. Cross-Sectional Dependence Results

The aim of a cross-sectional dependence test is to ascertain whether the cross-sectional units within a panel dataset exhibit any association. Detecting cross-sectional dependence is critical for the validity and efficiency of econometric models. The findings in

Table 7. Cross-sectional dependence results.

Pesaran’s CSD Tests	UN	LMII	RGDP	OP
Pesaran scaled LM test	27.3984 (0.0000) ***	15.4674 (0.0000) ***	53.3715 (0.0000) ***	123.0366 (0.0000) ***
Pesaran CD test	8.6132 (0.0000) ***	0.4586 (0.6465)	13.2138 (0.0000) ***	32.8633 (0.0000) ***

Notes: The value inside the parentheses is the corresponding probability value for the t statistic parenthesis. *** indicates significance at 1%.

illustrate that rejecting the null hypothesis of no cross-sectional dependence at a 1 percent significance level indicates that there is presence of cross-sectional dependence in the panel. Therefore, this study considers second-generation techniques or methods, such as a second-generation unit root test, to account for cross-sectional dependence.

4.4. Unit Root Results

Table 8. Cross-sectionally augmented Im-Pesaran-Shin (CIPS) unit root test results.

Variables	At Level		At 1st Difference		Order of Integration
	Stat	Critical Value at 5%	Stat	Critical Value at 5%
UN	−1.7346	−2.33	−3.0906	−2.33 **	I (1)
LMII	−2.3486	−2.33 **	N/A	N/A	I (0)
RGDP	−2.1952	−2.33	−2.6782	−2.33 **	I (1)
OP	−0.1795	−2.33	−5.6465	−2.33 **	I (0)

Source: Own figure drawn from figures from the World Bank (2025). Note: ** indicates the significance of statistics at 5% significance level.

presents formal unit root results to determine the order of integration for all the variables. Table 8 shows different integration orders. The panel unit root test findings indicate that the unemployment series (UN) becomes stationary after first differencing, which means that the series is integrated of order 1. In addition, RGDP shows that the series becomes stationary after first differencing. LMII and OP are stationary in levels. Second-generation unit root testing legitimises the use of Cross-sectional ARDL estimators.

4.5. Cointegration Results

Once the order of integration has been established, it is essential to examine cointegration among the variables. The Westerlund cointegration technique is applied in this study. The Westerlund cointegration test results are presented in

Table 9. Westerlund cointegration test.

Stat	Value	z-Value	p-Value	Robust p-Value
Gt	−3.75 ***	3.99	0.099	0.00
Ga	−0.666	3.432	1.000	0.380
Pt	−4.354 ***	−0.31	0.06	0.00
Pa	0.861	2.437	0.993	0.620

Source: Own figure drawn from figures from the World Bank (2025). Note: *** indicates the significance of statistics at 10% significance level.

. The results show a strong long-run equilibrium link between unemployment (the dependent variable) and the independent variables: oil price, labour market institutional index, and GDP. This conclusion is reached after rejecting the null hypothesis of no cointegration at Gt and Pt. The t-values and p-values of the Gt and Pt tests are statistically significant at the 1 percent level.

Ga and Pa show that we fail to reject the null hypothesis. Ga and Pa are insignificant, showing that there is no long-run relationship between the variables. The Gt and Pt are generally considered to take precedence. Gt and Pt tests are robust, powerful and more reliable than the Ga and Pa tests. The Gt and Pt provide more reliable and robust cointegration results than the Ga and Pa because the tests are based on t-statistics, which provide better size and properties in samples as compared to a-statistics. In addition, Gt and Pt perform better in panels with small time spans. Therefore, we conclude that there is strong evidence of cointegration among variables.

4.6. Econometric Estimation Results

The results presented in

Table 10. Cross-sectional ARDL model.

VARIABLES	Oil-Exporting
Long-run equation Dependent variable: UEMP
Oil price	−0.428 ***
	(0.107)
LMII	−0.272 **
OP*LMII	(0.034) −0.386 ** (0.094)
RGDP	−0.106 **
	(0.051)
Short-run equation
ECT	−0.319 **
	(0.172)
O P	−0.141 *
	(0.048)
LMII	−0.015 *
OP*LMII	(0.011) −0.095 ** (0.012)
RGDP	−0.0531
Constant	(0.085) 0.871
	(0.087)
Observations	270

Source: Author’s computation from World Bank’s Development Indicators. Notes *, **, ***, indicate the significance of statistics at 1%, 5%, and 10% significance levels, respectively. The standard errors are in parentheses. The dependent variable is unemployment.

show that the long-run coefficients are consistent with established economic theory and signal a clear basis for policy recommendations. The coefficient for oil price is much higher and has a more significant negative value (−0.428 ***), showing that a 1% prolonged increase in oil prices results in a significant 0.428% decrease in unemployment, confirming the positive labour market impact of oil sector expansion. The LMII coefficient is also negative (−0.272 **), implying that better labour market institutions result in lower long-term unemployment. The significance of the interaction term indicates that the moderating influence of LMII is immediate and not just a gradual phenomenon. This implies that the institutions have a capacity to prevent unnecessary job losses during adverse oil price shocks. The real GDP coefficient (−0.106 **) is also negative and significant, showing the strong association that economic expansion reduces unemployment.

The short-term dynamics are also consistent. The ECT is substantial (−0.319 **), showing the rate of recovery to equilibrium. The negative and significant coefficients for oil prices (−0.141 *) and LMII (−0.015 *) indicate that both variables have an immediate impact on unemployment reduction. This implies that the institutions have a capacity to prevent unnecessary job losses during adverse oil price shocks. The small GDP impact in the short run reinforces the view that its full effects on the labour market are realised over time.

4.7. Granger Causality Test

In African economies that export oil, the results of the Dumitrescu and Hurlin Granger causality tests clearly show a unidirectional causal relationship between unemployment and oil prices. With a low p-value of 0.0016 (see

Table 11. Dumitrescu and Hurlin Granger’s non-causality results.

	Oil-Exporting
Wald test	2.484
$\underset{\_}{Z}$	3.147
$\tilde{Z}$	2.67
p-values	0.0016
H1: Oil price does cause UMP
Wald test	0.523
$\underset{\_}{Z}$	−1.011
$\tilde{Z}$	−1.029
p-values	0.312
H0: UMP does not cause oil price
H1: UMP does cause oil price

Source: Author’s computation from the World Bank’s Development Indicators.

) for the hypothesis that oil prices cause unemployment, the conclusion that changes in oil prices are appropriate for forecasting changes in unemployment is statistically sound, and the null hypothesis of no causality is strongly rejected. On the other hand, Table 11 indicates that there is no significant causality running in that direction, as indicated by the high p-value of 0.312 for the reverse test (unemployment causing oil price). This creates a one-way predictive relationship, indicating that during the study period, labour market outcomes were externally driven by shocks to the price of oil. However, it should be noted that Granger causality provides a temporal lead lag relationship rather than a structural mechanism. Since oil prices are exogenous, labour market institutions, subsidy structure and probably fiscal structures transmit changes in oil prices to the labour market. Therefore, oil prices should be viewed as the main driver within a broader policy-dependent framework.

4.8. Diagnostic Tests

Table 12. Diagnostic test.

Test	Test Statistic	p-Value
Breusch–Godfrey LM test	2.419	0.121
Shapiro–Wilk test for normality	0.827	0.000
Breusch–Godfrey Heteroscedasticity test	1.025	0.306

Source: Own figure drawn from figures from the World Bank (2025).

shows that the calculated PARDL model meets the two criteria for efficient coefficient estimation: no serial correlation and no heteroskedasticity. The Breusch–Godfrey LM test for serial correlation gave a p-value of 0.121, while the Breusch–Godfrey heteroskedasticity test gave a p-value of 0.306. Because both values are greater than 5%, we cannot reject the null hypotheses, implying that the model residuals are not serially correlated or affected by non-constant variance. However, the model does have non-normal residuals, as evidenced by the Shapiro–Wilk test, which yielded a highly significant p-value of 0.000, contradicting the normality hypothesis. This distributional failure indicates the presence of outliers; however, the coefficient estimates remain consistent in large samples due to the central limit theorem.

4.9. CUSUM Square Stability Results

Figure 1 represents the stability and precision of the model. The results confirm that the regression coefficients remain stable over the entire study period, despite the same external shocks. The implication is that the relationships between unemployment and its independent variables within the model are consistent regardless of changes in the independent variables’ conditions having been met. Figure 1 below represents the corresponding CUSUM of the UN variable when accounting for oil price, RGDP and LMII.

5. Discussion of Results

The empirical findings presented make an important contribution to the research, notably by emphasising the critical and different roles of institutional characteristics and commodity prices in influencing macroeconomic outcomes for African oil-exporting states. The long-run finding of a strong negative and significant coefficient for oil price provides strong support for the resource-driven economic expansion hypothesis, directly aligning with empirical work by Alfalih (2024) on Saudi Arabia and Cheikh et al. (2018) on MENA countries, which found that higher oil revenues stimulate aggregate demand and government spending, resulting in lower unemployment rates. This finding, however, contrasts significantly with research on developed oil-importing countries (O. K. Kocaarslan, 2019; Koirala & Ma, 2020), in which oil prices function as a cost-push shock.

The study’s most important conclusion is a negative and substantial association between the labour market institutional index (LMII) and long-term unemployment. Together with the result on unemployment, the negative and significant coefficient on the interaction term confirms that unemployment is further reduced in economies with good labour market institutions. The results are plausible given that countries with good labour market institutions are generally better equipped to adjust to oil price fluctuations and mitigate their negative employment effects. Intuitively, high-quality labour market institutions not only stimulate job creation during periods of oil prosperity but also shield against employment declines when commodity prices drop. The result also echoes Agboola et al.’s (2024) more general claim that institutional quality reduces the adverse effects of external shocks and is supported by the beneficial role of labour market institutions, as reiterated by Nusair (2020). In addition, the results of this study suggest that quality institutions may improve labour market resilience by lowering the non-accelerating inflation rate of unemployment and allowing workers to adjust to productivity shocks brought on by fluctuations in the price of oil, thereby averting the structural unemployment that the rigidity models predicted.

Additionally, Okun’s Law (Okun, 1963) is easily validated by the long-run coefficient for RGDP, which affirms the basic inverse relationship that unemployment decreases with economic expansion. The idea that the full effects of output on employment are realised over time, however, is supported by the short-run RGDP impact, which is in line with micro-foundations such as labour hoarding that prevent immediate layoffs or new hiring in response to transient output fluctuations. Lastly, a stable and statistically significant adjustment process is confirmed by the significant and negative error correction term, which gives strong confidence in the reliability of the estimated long-run relationships. The system corrects more than 10% of any unemployment disequilibrium in the subsequent period. In contrast to the oil price effect, the short-run LMII coefficient has immediate significance, indicating that policymakers can address unemployment more quickly and easily through institutional reform.

6. Conclusions and Recommendations

This study examined the distinct role of labour market institutions in moderating the effects of oil price volatility on unemployment in nine African oil-exporting countries for a 30-year period from 1994 to 2024. This study was motivated by inconclusive results in the literature on the oil price–unemployment nexus, and suggestions that differences in labour market institutions are the main cause of the mixed results. To appropriately assess how labour market institutions exacerbate or alleviate the effects of oil price shocks on unemployment over time, the study employed a Cross-Sectionally Augmented Autoregressive Distributed Lag model (CS-ARDL) on a panel dataset of nine African oil-exporting countries from 1994 to 2024.

The main hypothesis was that countries with quality labour market institutions would experience less unemployment during periods of oil price increases. The study establishes the existence of a reduction in unemployment following oil price increases. Furthermore, a negative impact of labour market institutions on unemployment was established in this study. Of importance is the fact that good-quality institutions reduce unemployment. In addition, RGDP was found to have a long-run negative and significant effect on unemployment, confirming that unemployment falls as output grows. Our results also suggest that economies with stronger labour market institutions exhibit greater resilience, effectively cushioning the negative consequences of oil price shocks. These institutions appear to facilitate a more efficient reallocation of oil revenues and labour resources towards job creation in non-extractive sectors. In other words, countries with high-quality labour market institutions perform better in diverting oil revenues into sectors that create employment.

The econometric results from this study have clear policy implications. Firstly, overall quality labour market institutions play a pivotal role in the extent to which the oil price affects unemployment. While oil price increases stimulate a reduction in unemployment in African oil-exporting countries, the presence of quality labour market institutions further enhances this unemployment-reducing effect. Therefore, policymakers must prioritise labour market institutional reforms to absorb oil price shocks and reduce unemployment. Furthermore, policy reforms should be targeted at the specific aspects of the labour market institutional index, i.e., wage bargaining systems, the generosity and design of social safety nets and unemployment benefits, active labour market policies, employment protection legislation, and trade union density and power.

Secondly, African oil-exporting countries are challenged to create and promote institutions that ensure oil revenues are channelled towards productive sectors to further enhance reduction in unemployment during periods of oil prosperity. The creation of oil stabilisation funds will enable oil-exporting economies to absorb oil price shocks as well as shield against employment declines when commodity prices drop.

One of the limitations of this study is employing an aggregate index for labour market institutions instead of disaggregated measures. Individual aspects of labour market institutions would have accounted for more specific causal effects of institutions on the oil price–unemployment nexus. The study therefore recommends that future studies consider analysing individual aspects separately to assist policymakers in focusing on targeted institutional aspects.

Another limitation of this study is that it focuses only on oil-exporting countries in Africa. Future studies may consider countries outside Africa for comparative analysis. Another option may be a comparative analysis between net oil importers and net oil importers from across the globe.