Air Pollution Effects on Labour Market: Evidence from Asymmetric Analysis

Abstract

This study aims to evaluate the impact of greenhouse gas emissions on unemployment. Panel data encompassing 180 nations spanning the time frame from 1995 to 2018 has been employed. To capture non-linear and asymmetric effects, a recently developed panel quantile estimator has been used. The findings refute the notion of the environmental Kuznets curve and demonstrate that when emissions are at a low level, an increase in emissions leads to a higher rate of unemployment. Conversely, at greater emission levels, the opposite impact is observed. In addition, the analysis using panel quantile regression reveals that the impact of emissions is significantly greater in countries with higher unemployment rates. Ultimately, the correlation between unemployment and emissions remains consistent when unemployment among those with basic education is examined. However, unemployment among those with higher education has a contrasting association with emissions. In summary, the findings in this study indicate that the relationship between greenhouse gas emissions and labour demand is non-linear and contingent upon certain conditions.

Keywords

Greenhouse gas unemployment panel quantile environmental Kuznets curve labour demand

Introduction

The rationale behind the linkage between the environment in general and pollution in particular and economic growth is that the former deteriorates in the initial stages of economic growth. The situation improves at higher levels of economic growth as development matures, which is demonstrated through the environmental Kuznets curve (EKC; Gill et al., 2019; Grossman & Krueger, 1991; Stagl, 1999). More precisely, the level of pollution grows relatively fast due to the use of traditional engines and technology. In the initial phases, an increase in material output, income, investment and the creation of jobs is the prime concern, leaving less emphasis on a pollution-free environment.¹ A voluminous literature has focused on economic or industrialization impacts on the environment and pollution. What is overlooked in the literature is the environmental or pollution effects on the economy. The fact of the matter is that increasing emissions of greenhouse gases (GHGs), including air pollution and the measures to contain all of that, can potentially affect the economy in several ways. For instance, the development of environment-friendly technology may require a huge amount of resources, which may have to be undertaken at the cost of prime objectives of growth and employment. Hence, instead of being concerned about technological considerations, growth and employment-maximizing strategies may have to be adopted even if the quality of the environment is being compromised. Once enough resources are available from the persistent phases of rapid growth, the economies may innovate better technology, which may sustain the environment.

Economic activities around the globe are causing the emission of GHGs primarily because of the use of fossil fuels in production and transportation (Scholl et al., 1996). Large-scale deforestation, which is taking place in several parts of the world, has raised the supplies of fossil fuels, and it has become cost-effective to use these sources of energy in the process of growth (Lapinskienė et al., 2014). However, on the environmental front, there is a double burden caused by deforestation on the one hand and the use of fossil fuels and CO₂ emissions on the other. The concern has become a central point of discussion in economics and public policy circles. In the face of this, can the waiting period be allowed and the environment be deteriorated exponentially in the name of economic growth? It may get so late that at the higher stages of economic growth, restoration of the environment may become almost impossible. In the backdrop of this issue, a more alarming apprehension that has emerged is about livelihood creation. In the name of growth, the environment-damaging technology that is adopted may not be creating significant levels of employment to justify its continuation. Hence, it becomes pertinent to actually assess if the unemployment rate is falling with the adoption of such environment-damaging technology and energy usage. It may involve losses on both fronts: livelihood creation and sustenance of the environment. The previous studies, such as Fosten et al. (2012), Vicente and Tamarit (2012) and Ronaghi et al. (2020), have mainly focused on the effects of economic activities on the emission of GHGs, while here the effects of GHGs on unemployment have been examined. If GHG emissions are rising because of growth-maximizing technology with little concern for the environment, then the effect of GHG emissions on unemployment will indirectly capture the effect of technology on employment. Or, to put it differently, technology determines both the environment and the livelihood opportunities. Hence, in our unemployment function, with GHG being a determinant, the endogeneity problem is serious, which is addressed through appropriate instrumentation. Taking GHG as a proxy for the volume of economic activities, it may be useful to introduce its square terms in order to assess if its impact on employment/unemployment changes compared to what happens in the initial stages. Even if it reduces the livelihood opportunities in the initial stages beyond a certain threshold level, it may result in enhanced employment levels and a reduced unemployment rate.

Against this backdrop, this study aims to assess the effect of GHGs and CO₂ on unemployment in countries across the world for the period 1995‒2018. The attempt is novel and offers several additions to the related literature. First, as mentioned earlier, unlike the previous studies, the focus is on the labour effect of emissions, as unemployment is a prime indicator of development. Second, this article also contributes in terms of methodology. The distribution of unemployment is widely asymmetrical; consequently, the application of a linear model may not be appropriate. Therefore, the recently developed panel quantile technique has been employed. Finally, a wide range of countries in this analysis has been covered—in fact, all countries across the world if information on related or relevant variables is available.

The rest of the article is structured as follows: the second section reviews some of the studies pursued in this area; the third section focuses on the methodology and data; the empirical results are presented in the fourth section; and finally, the major findings are summarized in the fifth section.

Review of Literature

Environmental concerns arise from the fact that fossil fuel combustion emits GHGs, which are believed by the leading climate scientists to cause global warming and climate instability (Barrett et al., 2002; Ronaghi et al., 2020; Satterthwaite, 2010). In a recent study, Chai et al. (2019) have shown that mitigating the GHG effects may boost economic growth. Policies to reduce carbon emissions or promote new energy sources may impose debilitating costs on the economy, particularly on low-income households, on workers in particular industries and on the economy as a whole. Strict regulations limiting GHG emissions can have serious effects on production, raising the unemployment rate in the economies.

Babiker and Ecquas (2007) argue that with the labour market imperfections, if there were no offsetting policies, the reductions in the gross national product in the US in the first 10 years after emissions restrictions were imposed would be as much as 4%. However, with a counteracting labour market policy and emissions restrictions, the negative direct economic effects on production and employment could be completely eliminated. Sectoral rigidities in labour mobility and sectoral rigidities in wage adjustments are significant factors in determining the character of the economic adjustments to emissions limitations. A labour subsidy policy can reduce the direct and negative economic effects of emissions restrictions. The study by the Organization for Economic Cooperation and Development (OECD) in 2016 pinpoints that energy subsidies are high, particularly in Russia, other non-EU eastern European countries and a number of large developing countries, including India and China. Though emission caps in developed countries are pertinent, removing fossil fuel subsidies in emerging economies and developing countries could lead to a reduction in global GHG emissions by 10% by 2050 compared to the baseline and by as much as 30% in some countries. Aye and Edoja (2017), however, refute the possibility of an EKC: they investigated the effect of economic growth on CO₂ emissions using the dynamic panel threshold framework with data from a panel of 31 developing countries. Economic growth is seen to have a negative effect on CO₂ emissions in the low-growth regime but a positive effect in the high-growth regime, with the marginal effect being higher in the high-growth regime, which rather supports a U-shaped relationship. Climate concerns are naturally not on top of the agenda in European countries at the moment. As Fæhn et al. (2013) argue, having an economically and socially stable future with less unemployment is seen to be more important than the future climate. Particularly, if the climate change policies are not compatible with the pressing need for jobs, countries may postpone them, though such laxity may harm both future growth and the environment.

OECD (2016) projected the total annual market costs, including direct and indirect costs of outdoor air pollution, which is likely to increase from 0.3% of GDP in 2015 to 1.0% of GDP by 2060. The total market costs include both direct and indirect costs. The study also finds that change in labour productivity is one of the predominant costs of pollution. The Congressional Budget Office (2010) estimated that GHGs and processes to control their emissions have serious consequences for employment. It is observed that changes in employment in specific industries would reflect the amounts of GHGs they emit (through production and use of their output) and the difficulty of reducing their emissions of those gases.

Although a majority of studies have examined the effects of economic activities on GHGs or CO₂ emissions, some do attempt to know the causal effects. For instance, Jalil and Mahmud (2009) investigated the causal relationship between CO₂ emissions and economic growth in China. Their estimates show that a unidirectional causality running from economic growth to CO₂ emissions is uncovered in both the short and the long run. Wang et al. (2011), Hossain (2011) and Alam et al. (2012) found unidirectional causality running from CO₂ emissions to economic growth for China, industrialized countries and Bangladesh. On the other hand, Zanin and Marra (2013) investigated the EKC hypothesis for France, Switzerland and Pakistan. Lee (2013) explored the linkage between foreign direct investment (FDI), CO₂ emissions and economic growth using data about the BRICS countries (Brazil, Russia, India, China, South Africa, Iran, Egypt, Ethiopia and the United Arab Emirates [UAE]) from 1971 to 2009 using a panel cointegration approach. The empirical evidence supports the existence of unidirectional causality from FDI to economic growth and from economic growth to CO₂ emissions. Pao and Tsai (2010) also attempted to find the causal links between FDI and CO₂ emissions for a panel of BRICS countries. The results from their causality tests indicate the existence of strong bidirectional causality between these variables. Besides, Ghosh (2010) also documents two-way links between CO₂ emissions and India’s economic growth over the period 1971–2006. Arouri et al. (2012) attempted to find the linkage between CO₂ emissions, energy consumption and real GDP for 12 Middle East and North African (MENA) countries over the period 1981–2005, and they found evidence of bidirectional causality between carbon emissions and energy consumption.

For G7 countries (Canada, France, Germany, Italy, Japan, the United Kingdom and the United States), Ajmi et al. (2015) examine the relationships between energy consumption, CO₂ emissions and GDP. Their findings show the existence of a bidirectional, time-varying causality between energy consumption and CO₂ emissions for the United States. Sbia et al. (2014) tried to understand the connection between FDI, carbon emissions and economic growth in the UAE from 1975 to 2011. It is found that the relationship between carbon emissions and economic growth is bidirectional. Lau et al. (2014) also showed the presence of bidirectional causality between CO₂ and economic growth in Malaysia. Abdouli and Hammami (2017) investigated the causal relationship between environmental quality and economic growth for a panel of 17 MENA countries over the period 1990–2012. Their estimates demonstrate a bidirectional causality between CO₂ emissions and economic growth. Very recently, Sharma (2022) estimated the GHG effect on tourism and found that the effects are negative. The study also finds that the relationship is non-linear and conditional and varies significantly among the countries. Glennerster and Jayachandran (2023) have argued that the effects of GHG vary greatly among low- and high-income countries. Furthermore, the impact of climate change may have greater effects on the countries that are less industrialized and depend more on primary sectors such as agriculture and fisheries.

Methodology and Data

Data

Data have been calculated from more than 180 countries across the world for the period 1995‒2018. These data are extracted from a World Development Indicators database of the World Bank. For the unemployment ratio, the following three measures have been used in this study: total unemployment rate, unemployment ratio with advanced education and unemployment ratio with basic education. For air pollution emissions, the following two emission indicators have been used: CO₂ and GHG emissions. Both of these indicators are used in per capita terms. A range of control variables has also been used, covering aspects related to compensation, education, investment, openness, manufacturing share and price. Details of these variables are discussed in Table 1, and their descriptive statistics are presented in Appendix Table A1.

Table 1.

Variables and Definition.

Abbreviations	Variables	Source
LUE	Unemployment, total (percentage of total labour force) (modelled ILO estimate), converted into logarithmic value	WDI 2020
LUEADV	Unemployment with advanced education (percentage of total labour force with advanced education)	WDI 2020
LUEBAS	Unemployment with basic education (percentage of total labour force with basic education)	WDI 2020
	Greenhouse gas indicators
LCO2CAP	CO₂ emissions (metric tonnes per capita), converted into logarithmic value	WDI 2020
LGHGCAP	(Total greenhouse gas emissions [kt of CO₂ equivalent]/population) × 1,000	WDI 2020 and author calculation
	Control variables
LW	Compensation of employees (percentage of expense), converted into logarithmic value	WDI 2020
LEDU	School enrolment, secondary (percentage gross), converted into logarithmic value	WDI 2020
LOPEN	Trade (percentage of GDP), converted into logarithmic value	WDI 2020
LGCF	Gross capital formation (percentage of GDP), converted into logarithmic value	WDI 2020
LMAN	Manufacturing, value added (percentage of GDP)	WDI 2020
LCPI	Consumer price index, converted into logarithmic value	WDI 2020

Note: WDI – World Development Indicators; ILO – International Labour Organization.

Before shifting to the formulation of empirical models and estimation methods, the distribution of unemployment to understand its pattern is presented, which in turn provides guidance for selecting a specific model and estimation technique. In a quantile plot, each value of unemployment is plotted against the fraction of the data that has values less than that fraction. The diagonal line is a reference line (Figure 1). For example, the unemployment rate (LUE) would be rectangularly distributed if all the data were plotted along the line. Because there are many points above the reference line, it is known that the LUE distribution is skewed to the left while it is rightward skewed after 0.75 fractions. Figure 1 shows that the distribution of the unemployment rate is not symmetric; therefore, a linear model may not be appropriate. Hence, quantile regression in a panel context has been employed.

Figure 1.

Note: In this plot, each value of the unemployment (LUE) is plotted against the fraction of the data that has values less than that fraction. The diagonal line is a reference line.

Figure 2.

Note: Unemployment (LUE) is scatter points, and GHG emission is on the x-axis.

Most of the existing studies have adopted the traditional linear models, that is, fixed, random, two-stage least square and so on, to analyze the panel data. However, this approach yields the conditional expectation (mean value) of the dependent variable. Considering the wide heterogeneity in the labour market across countries and time, the relationships between employment and GHGs are expected to vary across different quantiles. In other words, the relationship between employment and GHGs may fluctuate across countries according to the level of employment or unemployment.

Therefore, to estimate the model, a quantile regression model specially designed for panel data with non-additive fixed effects as developed by Powell (2022) has been used here. Since some of the explanatory variables are endogenous, this model specifically addresses the endogeneity issue. For this, lags of the endogenous variables are used as instruments in a standard two-step process to overcome the problem. The main advantage of the Powell method is that it allows the probability to differ between individual countries in the panel as well as within individual countries. The technique relies on the generalized method of moments (GMM) for final estimation. For robustness, the system GMM estimator proposed by Blundell and Bond (1998) has also been used. A two-step process is adopted in both approaches to deal with the endogeneity problem.

The benchmark model to be estimated here to address the conditional quantile function of the panel data is as follows:

{LN}_{it} (τ x_{it}) = {x^{'}}_{it} β (τ) + ε_{it,}

(1)

where, LN_it (τx_it) is τ quantile of the dependent variable. x_it is a vector of explanatory variables that comprise GHGs and CO₂ indicators (used alternatively in the model) and their square term, wage, education, investment, openness, manufacturing share and price index indicator. The labour demand function follows the standard model developed in the theoretical and empirical literature (e.g., Hassan & Kandil, 2014). It is noteworthy that also included is a square term of pollution indicator to capture the type of effect, that is, linear, concave and convex. ε is the error term that may be a function of several disturbance terms, some fixed and some time-varying; τ refers to the τth quantile; and β(τ) is the regression parameter of the τth quantile.

Empirical Results

Before moving on to the main regression results, Figure 2 presents a scatter diagram and best-fit line between the unemployment rate and per capita GHG emission. The plot shows a positive slope of GHG emissions on the unemployment rate. However, the model does not fit very well (R² is 0.001), and the positive slope is also not very pronounced. Perhaps the plot shows a complicated relationship between these two indicators.

Next, the results of Equation (1) are presented. The effect of CO₂ and GHG emissions on the unemployment rate is assessed separately in Tables 2 and 3, respectively. The results of the models using Sys-GMM and quantiles 0.2, 0.4, 0.6, 0.8 and 0.9 in columns 1 to 6 in Table 2 are presented. While the unemployment ratio is the dependent variable, CO₂ per capita and its square term are the explanatory variables. Sys-GMM-based results suggest that the CO₂ effect is positive at the level and negative for the square term, and both variables are statistically significant. Quantile regression results show a similar pattern, and the estimated coefficients are statistically significant for all the quantiles. Also, the results show that when one moves towards higher quantiles, the size of coefficients starts decreasing. Specifically, for 0.8 and 0.9 quantiles, the effects are relatively smaller. In Table 3, GHG emissions per capita and its square term are used, and a similar type of impact is observed. However, in this instance, the estimated coefficients for higher quantiles (0.8 and 09) are relatively larger, showing that when unemployment is higher, the effect of emissions is more intense. Though in relation to these variables, the quantile regression results based on Powell’s (2022) method relying on the GMM do not differ from the results for system GMM, the significance of other control variables varies between the two methods in both the specifications. Nevertheless, quantile regression estimates provide additional information on how the intensity of effect varies across the unemployment quantiles.

Table 2.

Effects of CO₂ Emission on Unemployment.

	(1)	(2)	(3)	(4)	(5)	(6)
	LUE	LUE	LUE	LUE	LUE	LUE
	(Sys-GMM)	(0.20)	(0.40)	(0.60)	(0.80)	(0.90)
LUE_{(t − 1)}	.934**
	(0.00392)
LGCF	−.230**	−.500**	−.454**	−.0577**	−.251**	−.275**
	(0.0122)	(0.00824)	(0.00706)	(0.000240)	(0.00436)	(0.00180)
LEDU	−.210**	.159**	.0897**	−.118**	−.0626**	.0265**
	(0.0146)	(0.00409)	(0.0100)	(0.00185)	(0.00495)	(0.000847)
LW	−.136**	−.0330**	.0839**	.0446**	−0.000514	.137**
	(0.00540)	(0.0148)	(0.00218)	(0.000640)	(0.00743)	(0.000543)
LOPEN	−.0320**	−.0725**	−.102**	−.0597**	−.0130**	−.0749**
	(0.00805)	(0.0270)	(0.00537)	(0.000271)	(0.00435)	(0.000933)
LCPI	0.00930	−.172**	−.174**	−.149**	−.182**	−.170**
	(0.00645)	(0.0102)	(0.00886)	(0.000646)	(0.00347)	(0.000593)
LMAN	−.0370**	.0667**	.0621**	−.0417**	−.0549**	−.0190**
	(0.00574)	(0.00459)	(0.00306)	(0.000164)	(0.00211)	(0.000287)
LCO2CAP	.141**	.268**	.248**	.259**	.183**	.102**
	(0.00806)	(0.0139)	(0.00297)	(0.000469)	(0.00176)	(0.000306)
LCO2CAP²	−.0325**	−.0755**	−.103**	−.112**	−.135**	−.128**
	(0.00155)	(0.00274)	(0.000645)	(0.000120)	(0.000335)	(0.000177)
_cons	2.315**
	(0.120)
N	942	971	971	971	971	971
Sargan P value:AR(2)	.65:0.84
MAC		0.07	0.11	0.21	0.20	0.42

Notes: (a) MAC is the mean acceptance rate of the fraction of the proposed samples that are accepted. The optimal MAC hinges on the properties of the target distribution and is not known in general. If the target distribution is normal, however, the optimal acceptance rate is known to be 0.234 for multivariate distributions. (b) The dependent variable is the unemployment rate in the form log (LUE). Standard errors are given in parentheses. **P <.05. GMM – generalized method of moments; AR – autoregressive. LCO2CAP² is the square term of LCO2CAP.

Table 3.

Effects of GHG Emission on Unemployment.

	(1)	(2)	(3)	(4)	(5)	(6)
	LUE	LUE	LUE	LUE	LUE	LUE
	(Sys-GMM)	(0.20)	(0.40)	(0.60)	(0.80)	(0.90)
LUE_{(t − 1)}	.930**
	(0.00286)
LGCF	−.289**	−.488**	−.497**	−.307**	−.356**	−.501**
	(0.00630)	(0.00120)	(0.00113)	(0.00194)	(0.000735)	(0.000327)
LEDU	−.0223**	.374**	.413**	.259**	−.0163**	−.119**
	(0.00972)	(0.00155)	(0.000567)	(0.00227)	(0.00104)	(0.000331)
LW	−.157**	−.00944**	.0915**	.109**	−.0214**	.0882**
	(0.00851)	(0.000751)	(0.000502)	(0.00137)	(0.00171)	(0.000147)
LOPEN	0.00439	−.150**	−.125**	−.0846**	.0867**	.301**
	(0.0104)	(0.000716)	(0.000487)	(0.00207)	(0.000597)	(0.000114)
LCPI	−.0309**	−.180**	−.257**	−.323**	−.294**	−.358**
	(0.00471)	(0.00189)	(0.000930)	(0.00162)	(0.000770)	(0.000399)
LMAN	−.0873**	.0207**	−.0429**	−.0271**	−.170**	−.121**
	(0.00541)	(0.000816)	(0.000239)	(0.00108)	(0.000573)	(0.000280)
LGHGCAP	.135**	.761**	.970**	.982**	1.005**	1.029**
	(0.0136)	(0.00128)	(0.000762)	(0.00225)	(0.00224)	(0.000607)
LGHGCAP²	−.0448**	−.208**	−.255**	−.267**	−.283**	−.268**
	(0.00374)	(0.000315)	(0.000211)	(0.000674)	(0.000611)	(0.000215)
_cons	1.871**
	(0.0708)
N	775	804	804	804	804	804
Sargan P value:AR(2)	.62:0.75
MAC		0.08	0.11	0.19	0.18	0.37

Both CO₂ and GHGs show a statistically significant effect on unemployment across countries. However, this effect is non-linear in the sense that in the initial stages, unemployment rises in response to emissions, and at higher levels of emissions, it declines. In other words, evidence shows a concave curve between unemployment and emission. This implies that emissions are leading to higher unemployment initially; however, at a higher level, it has a negative effect. In other words, the marginal effect of the emission initially increases, and with a higher level of emission, it has a dampening effect. Furthermore, it is also seen that the estimated coefficients of emission vary, and they are conditional as per the unemployment level. This perhaps indicates that emissions have a varying effect on the unemployment level; for example, for a high unemployment situation, the relationship weakens in the case of CO₂ and strengthens for GHGs. This may be taken to suggest that at a lower quantum of activities, an increase in growth and the technology and fuel used in the process would raise the rate of unemployment. The capital-intensive technology operated through cheaper energies affects employment adversely. On the other hand, deforestation adopted to extract energy hampers the livelihood of those dependent on forest products. On the whole, environment-damaging economic activities also cause livelihood loss, implying adversity from both the angles of environmental degradation and a rise in unemployment. However, at high levels of emissions, there is an inverse relation between the two, indicating that further increases in emissions reduce the unemployment rate. This reflects largely the scale effect of economic activities at the economy-wide level.

Conducting economic activities widely would generate employment opportunities significantly, even if it may be damaging the environment to a considerable extent. The difference is that the lower scale of activities causes loss in terms of the environment as well as employment, while at higher levels of economic endeavours, the employment gains are at least reaped at the cost of the environment. However, taking these findings on a positive note, it may be suggested that given the urgent need for livelihood creation in the economies with low per capita incomes, the thrust will have to be given to the faster expansion in the volume of economic activities, even if they may be damaging the environment. Once employment and well-being are restored with a huge spreading out of the activities, environment- controlling initiatives may be introduced. However, if massive expansions are not possible, then given the positive relationship between emissions and unemployment, strong regulatory measures are called for so that losses on both counts are minimized. Moderate expansion in economic activities at the initial stages of environmental restoration must be prioritized even if financing growth becomes costlier because, in any case, this growth is not employment-generating—rather, employment-aggravating.

Gross capital formation in both the specifications and as per both the estimation methods is found to reduce unemployment, though the magnitude of impact varies considerably across different quantiles in the equation with CO₂ emissions. Particularly in the lower quantiles, it diminishes unemployment much more than at higher levels of unemployment rates. Countries with lower levels of unemployment witness a complementary relationship between capital and labour. Better human capital formation measured in terms of schooling raises the unemployment rate in countries at lower quantiles, though it is just the opposite in countries with very high rates of unemployment. Possibly countries with lower rates of unemployment are already exhausted in terms of job creation, leaving less scope for further expansion, while in countries with higher rates of unemployment, employability of the labour force rises with higher educational attainments. Compensation to employees has a very mixed impact on unemployment across different quantiles: at lower levels, it is negative, then becomes positive, and again at top levels, labour demand shrinks with an increase in compensation, raising the unemployment rate.

Though the trade openness in the equation with GHG remains insignificant as per the system GMM estimation, for different quantiles, the effect is different: at lower quantiles, it is unemployment-reducing, while at higher levels, international trade is seen to aggravate the employment scenario. However, in the equation with CO₂ emissions, trade openness reduces the unemployment rate in all the quantiles and as per the system estimation as well. Greater trading opportunities seem to have direct and indirect effects on employment gains. Industrialization as a strategy is important for countries for employment creation. Particularly the ones at higher quantiles of the unemployment rate appear to benefit from the expansion in manufacturing production. Hence, in the context of the debate on industry versus services as the engine of growth, the former is still relevant for countries with high rates of unemployment. The consumer price index is seen to have a negative effect on unemployment, implying that price incentives can raise production and employment levels. In fact, the entire literature on Philip’s curve suggests an inverse relationship between inflation and unemployment. Though the excess demand hypothesis can rationalize it easily, the expectation school notes a marked departure. While the adaptive expectation hypothesis envisages a negative relationship between inflation and unemployment at least in the short run, the rational expectation school completely nullifies it.

Table 4.

Effects of GHG Emission on Unemployment with Advanced Education.

	(1)	(2)	(3)	(4)	(5)	(6)
	LUEADV	LUEADV	LUEADV	LUEADV	LUEADV	LUEADV
	(Sys-GMM)	(0.20)	(0.40)	(0.60)	(0.80)	(0.90)
LUEADV_{(t − 1)}	.487**
	(0.0286)
LGCF	−.955**	−.770**	−.614**	−.764**	−.246**	−.426**
	(0.0716)	(0.0270)	(0.0224)	(0.0654)	(0.0327)	(0.00639)
LEDU	0.271	.127**	−.409**	−1.385**	−.577**	−.421**
	(0.214)	(0.0317)	(0.0490)	(0.108)	(0.0582)	(0.00718)
LW	.296**	.195**	.130**	−.126**	0.00881	−0.00406
	(0.0209)	(0.00960)	(0.0276)	(0.0169)	(0.0139)	(0.00566)
LOPEN	−0.0828	−.248**	−.137**	−.181**	−.208**	−.289**
	(0.0582)	(0.00808)	(0.0186)	(0.0116)	(0.0124)	(0.00201)
LCPI	0.0677	.225**	.339**	.417**	−.195**	−.235**
	(0.0777)	(0.0332)	(0.0221)	(0.0515)	(0.0567)	(0.0103)
LMAN	.445**	−.0127	.158**	.0302**	0.00497	−.208**
	(0.133)	(0.0114)	(0.0221)	(0.00958)	(0.00881)	(0.00927)
LGHGCAP	−.527**	−.525**	−.437**	0.0422	−.496**	−1.424**
	(0.258)	(0.0964)	(0.0368)	(0.125)	(0.0166)	(0.0319)
LGHGCAP²	.121*	.0651**	.0959**	−0.0295	.0591**	.217**
	(0.0647)	(0.0227)	(0.00744)	(0.0313)	(0.00438)	(0.00688)
_cons	1.006
	(0.966)
N	360	419	419	419	419	419
Sargan P value:AR(2)	1.00: 0.2460
MAC		0.364	0.376	0.301	0.406	0.312

Notes: (a) MAC is the mean acceptance rate of the fraction of the proposed samples that are accepted. The optimal MAC hinges on the properties of the target distribution and is not known in general. If the target distribution is normal, however, the optimal acceptance rate is known to be 0.234 for multivariate distributions. (b) The dependent variable is the unemployment rate (with advanced education) in the form log (LUEADV). Standard errors are given in parentheses. **P <.05. GMM – generalized method of moments; AR – autoregressive. LCO2CAP² is the square term of LCO2CAP.

While basic education is essential for awareness to improve, higher educational attainments are seen as indicators of enhanced employability. As a distinction is made between the basic and advanced levels of education, the effect of GHGs on unemployment is also assessed with different levels of educational attainments, that is, advanced education and basic education. Table 4 presents regression results on unemployment with advanced education, and Table 5 shows the effect on unemployment with basic education. These findings on the effects of emissions on both types of unemployment are contrary. Though in the system method, neither the advanced nor the basic education impacts unemployment significantly, the quantile regression results confirm the favourable effect of the advanced level of education on employment (except the bottom quantile; Table 4). On the other hand, basic education is seen to raise the rate of unemployment (Table 5). Precisely, the results show a convex effect of emissions on unemployment with advanced education and a concave effect on unemployment with basic education. Also, the results here show that conditional effects vary in both cases when moving towards higher quantile levels.

Table 5.

Effects of GHG Emission on Unemployment with Basic Education.

	(1)	(2)	(3)	(4)	(5)	(6)
	LUEBAS	LUEBAS	LUEBAS	LUEBAS	LUEBAS	LUEBAS
	(Sys-GMM)	(0.20)	(0.40)	(0.60)	(0.80)	(0.90)
LUEBAS_{(t − 1)}	.595**
	(0.0344)
LGCF	−.897**	−.662**	−.439**	−.821**	−.422**	−.261**
	(0.0683)	(0.0892)	(0.0134)	(0.0856)	(0.0379)	(0.0368)
LEDU	−0.0116	.524**	.788**	.596**	.360**	.380**
	(0.104)	(0.0576)	(0.0176)	(0.0343)	(0.0320)	(0.0560)
LW	0.0541	−.110**	.178**	.248**	.353**	.427**
	(0.0943)	(0.0187)	(0.0148)	(0.0252)	(0.0159)	(0.00938)
LOPEN	.154*	0.0718	−.0812**	.0482*	.138**	.124**
	(0.0791)	(0.0463)	(0.00655)	(0.0278)	(0.00606)	(0.0149)
LCPI	0.0274	.102*	.126**	−.160**	.137**	−.235**
	(0.105)	(0.0550)	(0.0354)	(0.0742)	(0.0251)	(0.0287)
LMAN	0.0871	.302**	.191**	.214**	.319**	−.0408**
	(0.158)	(0.0448)	(0.0153)	(0.0278)	(0.0205)	(0.0120)
LGHGCAP	0.260	.894**	.569**	.424**	0.0602	.386**
	(0.204)	(0.0725)	(0.0258)	(0.0379)	(0.0986)	(0.0944)
LGHGCAP²	−0.0515	−.145**	−.0852**	−.0455**	.0938**	−0.00476
	(0.0466)	(0.0205)	(0.00688)	(0.0104)	(0.0246)	(0.0238)
_cons	2.235**
	(0.833)
N	365	424	424	424	424	424
Sargan P value:AR(2)	1.00:0.41
MAC		0.504	0.610	0.417	0.472	0.354

Notes: (a) MAC is the mean acceptance rate of the fraction of the proposed samples that are accepted. The optimal MAC hinges on the properties of the target distribution and is not known in general. If the target distribution is normal, however, the optimal acceptance rate is known to be 0.234 for multivariate distributions. (b) The dependent variable is unemployment (with basic education) in the form log (LUEBAS). Standard errors are given in parentheses. **P <.05. GMM – generalized method of moments; AR – autoregressive. LCO2CAP² is the square term of LCO2CAP.

This research adds to the growing body of evidence showing how emissions and climate change disproportionately impact low-income nations and communities that rely on manual labour, such as those in the agricultural sector (see Glennerster & Jayachandran, 2023). People with lower levels of education and competence experience more hardship, on which these findings provide indirect evidence, and the hypothesis is valid.

Conclusion and Policy Implications

This study, based on the panel data for a large number of countries, focuses on the emission of GHGs effects on unemployment, an issue that has been overlooked in the literature. The results reject the EKC hypothesis and show that at low levels of emissions, an increase in it causes more unemployment, but at higher levels of emissions, the effect is just the opposite. Furthermore, the panel quantile regression results of this study also show that emission effects are much higher in countries where the unemployment rate is relatively higher. Finally, the unemployment and emission relationship does not vary when we consider unemployment data with basic education; however, unemployment with advanced education shares the opposite (convex) relationship with emissions.

Overall, these results do highlight a GHG effect on labour demand. An effort to curtail the GHG emissions leads to a decline in labour demand. Also, better technology in production reduces emissions and causes unemployment in the initial stages to rise. Second, reduced emissions may be an indicator of lesser economic activities, particularly manufacturing and transportation activities, which are relatively more labour-intensive. On the other hand, it could be a reflection of the fact that the economy is moving towards high-technology industries and service industries, which may cause a reduction in employment in the initial stages, though at later stages labour demand may rise.

On the whole, the situation of higher levels of emissions and reduced employment depicts the worst combination of outcomes, though such a possibility is ruled out. Higher emissions may raise unemployment initially, but at later stages, unemployment falls with enhanced economic activities. Sys-GMM-based results suggest that the CO₂ effect is positive at the level and negative for the square term, and both variables are statistically significant. Furthermore, quantile regression results show a similar pattern. Besides, those with basic education may suffer more in comparison with those with advanced levels of education. The major lesson for developing countries is that better technology that uses less cheap fossil oil will have to be followed in a gradual and consecrated manner so that employment losses are minimized and better environmental outcomes are achievable. At the same time, skill formation and educational attainments will have to be enhanced so that better technology does not cause unemployment due to labour market mismatches.

Although the effort in this study is novel, it does have some limitations. There are political and geopolitical risks due to pollution and climatic economic impacts, especially high unemployment in the nation. Geopolitical tensions between major powers are projected to rise in tandem with climate change concerns under the new global order that is taking shape, which is characterized by diminishing natural resource availability. In addition, environmental norms and rules might be poorly implemented during times of political and geopolitical crises. Many countries, like Pakistan, are either strategically involved or unwittingly caught up in domestic and geopolitical games, and these problems can affect all of them. Thus, future studies may consider covering the political, geopolitical and other crisis-joint effects of GHG emissions on the general economy and unemployment. Furthermore, these findings do hint that the less educated suffer more due to GHG emissions. However, the effects on different income groups or countries that depend more on primary activities such as agriculture, fisheries and other allied activities have not been examined. Future studies may look into these aspects.

Footnotes

Declaration of Conflicting Interests

The authors declared no potential conflicts of interest with respect to the research, authorship and/or publication of this article.

Funding

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

Appendix

Table A1.

Descriptive Statistics.

Variable	Observations	Mean	Standard Deviation	Minimum	Maximum
LUE	4,368	1.81	0.8	−1.97	3.64
LUEADV	1,554	1.55	0.69	−3.09	3.83
LUEBAS	1,556	2.07	0.83	−2.12	3.72
LGCF	3,893	3.1	0.39	−2.28	4.44
LEDU	3,032	4.24	0.56	1.66	5.1
LW	2,306	3.11	0.6	0.81	4.25
LOPEN	4,251	4.34	0.62	−3.86	6.76
LCPI	3,987	4.38	0.62	−7.26	8.43
LMAN	4,128	2.3	0.82	−2.14	5.26
LGHGCAP	3,220	1.6	1.1	−1.95	5.11
LCO2CAP	3,866	0.6	1.65	−4.15	4.25

Note

References

Abdouli

, & Hammami

(2017). Investigating the causality links between environmental quality, foreign direct investment and economic growth in MENA countries. International Business Review, 26(2), 264–278.

Ajmi

A. N.

, Hammoudeh

, Nguyen

D. K.

, & Sato

J. R.

(2015). On the relationships between CO₂ emissions, energy consumption and income: The importance of time variation. Energy Economics, 49, 629–638.

Alam

M. J.

, Begum

I. A.

, Buysse

, & Huylenbroeck

G. V.

(2012). Energy consumption, carbon emissions and economic growth nexus in Bangladesh: Cointegration and dynamic causality analysis. Energy Policy, 45, 217–225.

Arouri

M. E.

, Youssef

A. B.

, M’henni

, & Rault

(2012). Energy consumption, economic growth and CO₂ emissions in Middle East and North African countries. Energy Policy, 45, 342–349.

Aye

G. C.

, & Edoja

P. E.

(2017). Effect of economic growth on CO₂ emission in developing countries: Evidence from a dynamic panel threshold model. Cogent Economics & Finance, 5(1), 1–22.

Babiker

M. H.

, & Eckaus

R. S.

(2007). Unemployment effects of climate policy. Environmental Science & Policy, 10(7–8), 600–609.

Barrett

J. P.

, Hoerner

J. A.

, Bernow

, & Dougherty

(2002). Clean energy and jobs: A comprehensive approach to climate change and energy policy. Economic Policy Institute.

Blundell

, & Bond

(1998). Initial conditions and moment restrictions in dynamic panel data models. Journal of Econometrics, 87(1), 115–143.

Chai

H. C.

, Hong

W. H.

, Reilly

J. M.

, Paltsev

, & Chen

Y. H. H.

(2019). Will greenhouse gases mitigation policies abroad affect the domestic economy? The case of Taiwan. Climate Change Economics, 10(4), 1950016.

10.

Congressional Budget Office. (2010). How policies to reduce greenhouse gas emissions could affect employment: A series of issue summaries from the Congressional Budget Office. Retrieved May 5, 2010, from https://www.cbo.gov/sites/default/files/111th-congress-2009-2010/reports/05-05-capandtrade_brief.pdf

11.

Dinda

(2004). Environmental Kuznets curve hypothesis: A survey. Ecological Economics, 49(4), 431–455.

12.

Fæhn

, Gómez-Plana

A. G.

, & Kverndokk

(2013). How can carbon policies impact unemployment? Carbon Management, 4(1), 27–29.

13.

Fosten

, Morley

, & Taylor

(2012). Dynamic misspecification in the environmental Kuznets curve: Evidence from CO₂ and SO₂ emissions in the United Kingdom. Ecological Economics, 76, 25–33.

14.

Ghosh

(2010). Examining carbon emissions economic growth nexus for India: A multivariate cointegration approach. Energy Policy, 38(6), 3008–3014.

15.

Gill

A. R.

, Hassan

, & Viswanathan

K. K.

(2019). Is democracy enough to get early turn of the environmental Kuznets curve in ASEAN countries? Energy & Environment, 30(8), 1491–1505.

16.

Glennerster

, & Jayachandran

(2023). Think globally, act globally: Opportunities to mitigate greenhouse gas emissions in low-and middle-income countries. Journal of Economic Perspectives, 37(3), 111–135.

17.

Grossman

G. M.

, & Krueger

A. B.

(1991). Environmental impacts of a North American free trade agreement (No. w3914). National Bureau of Economic Research.

18.

Hassan

, & Kandil

(2014). Employment fluctuations and sectoral shifts in Egypt: Testing the public/private sectoral shifts hypothesis. International Journal of Development Issues, 13(2), 129–154.

19.

Hossain

(2011). Panel estimation for CO₂ emissions, energy consumption, economic growth, trade openness and urbanization of newly industrialized countries. Energy Policy, 39(11), 6991–6999.

20.

Jalila

, & Mahmudb

S. F.

(2009). Environment Kuznets curve for CO₂ emissions: A cointegration analysis for China. Energy Policy, 37(12), 5167–5172.

21.

Lapinskienė

, Tvaronavičienė

, & Vaitkus

(2014). Greenhouse gases emissions and economic growth—Evidence substantiating the presence of environmental Kuznets curve in the EU. Technological and Economic Development of Economy, 20(1), 65–78.

22.

Lau

L.-S.

, Choong

C.-K.

, & Eng

Y.-K.

(2014). Investigation of the environmental Kuznets curve for carbon emissions in Malaysia: Do foreign direct investment and trade matter? Energy Policy, 68, 490–497.

23.

Lee

J. W.

(2013). The contribution of foreign direct investment to clean energy use, carbon emissions and economic growth. Energy Policy, 55, 483–489.

24.

OECD. (2016). The economic consequences of outdoor air pollution. https://www.oecd.org/environment/indicators-modelling-outlooks/Policy-Highlights-Economic-consequences-of-outdoor-air-pollution-web.pdf

25.

Pao

H.-T.

, & Tsai

C.-M.

(2010). CO₂ emissions, energy consumption and economic growth in BRIC countries. Energy Policy, 38(12), 7850–7860.

26.

Powell

(2022). Quantile regression with nonadditive fixed effects. Empirical Economics, 63(5), 2675–2691.

27.

Ronaghi

, Reed

, & Saghaian

(2020). The impact of economic factors and governance on greenhouse gas emission. Environmental Economics and Policy Studies, 22(2), 153–172.

28.

Satterthwaite

(2010). The contribution of cities to global warming and their potential contributions to solutions. Environment and Urbanization Asia, 1(1), 1–12.

29.

Sbia

, Shahbaz

, & Hamdi

(2014). A contribution of foreign direct investment, clean energy, trade openness, carbon emissions and economic growth to energy demand in UAE. Economic Modelling, 36, 191–197.

30.

Scholl

, Schipper

, & Kiang

(1996). CO₂ emissions from passenger transport: A comparison of international trends from 1973 to 1992. Energy Policy, 24(1), 17–30.

31.

Sharma

(2022). Greenhouse gas effects on international tourism: A panel quantile evidence. Anatolia, 33(2), 266–269.

32.

Stagl

(1999). Delinking economic growth from environmental degradation? A literature survey on the environmental Kuznets curve hypothesis [Working Paper No. 6]. Wirtschafts Universitat Wien.

33.

Vicente

, & Tamarit

(2012). Is there an environmental Kuznets curve for Spain? Fresh evidence from old data. Economic Modelling, 9(6), 2696–2703.

34.

Wang

, Wang

, Maa

, & Gong

(2011). Recent advances in catalytic hydrogenation of carbon dioxide. Chemical Society Reviews, 40(7), 3703–3727. https://doi.org/10.1039/C1CS15008A

35.

Zanin

, & Marra

(2013). Assessing the functional relationship between CO₂ emissions and economic development using an additive mixed model approach. Economic Modelling, 29(4), 1328–1337.