Convergence analysis of ecological footprint at different time scales: Evidence from Southern Common Market countries

Abstract

This study investigates the convergence in ecological footprint per capita across Southern Common Market countries over the period 1961–2016 within the framework of the environmental convergence hypothesis. However, unlike the existing literature, which mainly tests the convergence for the overall period, this study follows a different path. First, the time series is decomposed into different frequencies using the discrete wavelet transform method. Then, using the Fourier Augmented Dickey-Fuller and Augmented Dickey-Fuller unit root tests, convergence in ecological footprint per capita is tested for different time scales; short-run, medium-run, and finally long-run. The results indicate that countries show different convergence tendencies at different time scales. While the results support the convergence hypothesis for all countries in the short-run, the convergence hypothesis holds for only four and three of the five countries in the medium and long-run, respectively. Besides, the results show that the convergence hypothesis holds for only Uruguay for the whole period.

Keywords

Ecological footprint convergence discrete wavelet transform MERCOSUR countries environmental degradation

Introduction

The neoclassical growth theory, introduced by Solow¹ and Swan,² and later refined by Cass³ and Koopmans,⁴ assumes that developing countries tend to grow faster than rich countries, and eventually, the income levels of these countries will converge to a steady state due to decreasing return to scale. Particularly after the seminal works of Baumol⁵ and Barro and Sala-i-Martin,^6,7 the convergence hypothesis has been empirically tested by numerous studies.^8,9 The convergence hypothesis has been tested not only in terms of per capita income but also in terms of happiness,¹⁰ labour productivity,¹¹ interest rates,¹² house prices,¹³ etc. In recent years, countries have faced serious environmental issues such as global warming, climate change, shortages of natural resources, water pollution, etc., which have led researchers to focus on environmental studies. In this context, numerous studies have tested the convergence hypothesis for several environmental indicators within the framework of the environmental convergence hypothesis, which assumes that initially developing countries experience higher environmental degradation than richer countries; however, differences in the quality of their environments diminish over time.¹⁴

Testing the existence of convergence in environmental indicators is important for policymakers in both developed and developing countries. For example, the convergence in per capita emissions may affect the political economy of negotiating multilateral climate agreements. Countries with lower per capita emissions may expect countries with higher per capita emissions to make more effort toward mitigating climate change.¹⁵ If there is no convergence in terms of emissions, the principle of forcing equal per capita emissions would result in significant international transfers of rents through carbon allowance trading or the relocation of pollution-intensive industries.¹⁶ Besides, testing the environmental convergence hypothesis can help researchers and policymakers to develop more efficient policies. One could say that policies considering countries converging in terms of environmental quality could be more successful than policies considering a more heterogeneous group of countries in terms of environmental quality.¹⁷

Southern Common Market (MERCOSUR) is a South American regional economic organization founded in 1991 to encourage the free movement of trade, goods and services, and people among member countries and promote regional growth. Argentina, Brazil, Paraguay, and Uruguay are founding countries and full members of MERCOSUR. Venezuela joined as a full member in 2012, but it was suspended indefinitely in 2016. The motivation to select the MERCOSUR countries as a sample is based on two reasons. First, the MERCOSUR countries have recorded rapid economic growth for over two decades, and it is one of the largest trade organizations in the world. Several factors may affect environmental degradation, such as renewable energy consumption,¹⁸ environmental regulations,¹⁹ and increasing economic activities.^20,21 In the last three decades, the total carbon dioxide (CO₂) emissions of MERCOSUR countries have nearly doubled.²² Thus, one can say that the MERCOSUR countries, as developing countries have an important role in global environmental quality. Second, the member countries of MERCOSUR have a similar economic and social structure, geography, and colonial history, and they have abundant natural resources and a large consumer market.^22,23 MERCOSUR is a trade organization; however, since these countries may not be homogeneous regarding environmental degradation, implying the same environmental policies in the organization may not be effective. Therefore, the convergence process of their environmental degradation is important.

Motivated by this context, the main objective of this study is to test the convergence process in ecological footprint (EF) per capita across five MERCOSUR countries for different time scales. More specifically, to test the existence of convergence, the Fourier Augmented Dickey-Fuller (FADF) and Augmented Dickey-Fuller (ADF) unit root tests are used. Besides, by using the wavelet transform (DWT) technique, first, the time series is decomposed into different scales. Next, EF per capita across MERCOSUR countries is analysed for different time scales, short-run, medium-run, and finally, long-run. The contributions of this paper to the literature are three-fold. First, unlike the existing literature, which mainly tests the convergence for the entire period, this study follows a different path. As mentioned above, first, the time series is decomposed into different scales, short-run, medium-run, and long-run, using the wavelet transform technique. Then, the convergence process of EF per capita at different time scales among the MERCOSUR countries is tested by using the FADF and Augmented Dickey-Fuller unit root tests. Second, the EF per capita data is used to proxy for environmental degradation. While early studies mostly used CO₂, recent studies have focused on the EF introduced by Mathis Wackernagel and William Rees.^24,25 Since the EF comprises six components: the footprints of fishing ground, carbon, forest land, cropland, grazing land, and built-up, it is a more comprehensive indicator than CO₂ as a proxy for environmental degradation. The EF allows for tracking aggregate human pressure on the biosphere's capacity.^26,27 Third, to the authors' best knowledge, this is the first to analyse the environmental convergence process among MERCOSUR countries. The next part of the study summarizes the related empirical literature. Section three presents the data set and econometric methods. Section four provides the empirical results. Section five presents discussions, and finally, section six concludes the paper.

Literature review

Numerous studies in the literature test the environmental convergence hypothesis, using various environmental degradation indicators and econometric methods. The literature on the environmental convergence hypothesis follows three paths regarding the environmental degradation indicator: (i) studies using CO₂ emissions, (ii) studies using EF, and (iii) studies using other environmental indicators, such as sulfur dioxide, nitrous oxide, methane emissions, biomass material footprint, etc.

The first group of studies has used CO₂ emissions as a proxy for environmental degradation. While most of these studies^28–35 have tested convergence across countries, the others focus on convergence in CO₂ emissions across states or regions^36–38 or convergence across sectors.^39–43 Strazicich and List²⁸ test the convergence in CO₂ emissions across industrial countries over the period 1960–1997. Tiwari and Mishra³² study the convergence in CO₂ emissions across Asian countries. Li et al.³⁶ test the convergence in CO₂ for the United States (US). Apergis and Payne³⁹ analyse the convergence in CO₂ across sectors for US states. Yu et al.⁴² test the convergence of carbon emissions for 24 industrial sectors in China.

However, in recent years, the studies testing the environmental convergence hypothesis have focused on EF which is more comprehensive than CO₂. These studies follow two main lines regarding the econometric methodology adopted. The first group of studies applies several unit root tests to test the convergence process. Solarin⁴⁴ tests the convergence in CO₂, carbon footprint, and EF across 27 Organisation for Economic Co-operation and Development (OECD) countries using several unit root tests. Yilanci and Pata⁴⁵ investigate the convergence in EF across the ASEAN-5 countries from 1961 to 2016 using a two-regime threshold autoregressive (TAR) panel unit root test Işık et al.⁴⁶ test the convergence in EF per capita across the North American Free Trade Agreement countries from 1961 to 2016 by employing the TAR panel unit root test Yilanci et al.⁴⁷ investigate the convergence in EF and carbon footprint across Group of Seven (G7) countries over the period 1961–2016 using the Panel Fourier TAR methodology. The second group of studies on EF convergence applies the nonlinear time-varying methodology developed by Phillips and Sul.⁴⁸ Ulucak and Apergis⁴⁹ analyse convergence in EF across European Union (EU) countries from 1961-to 2013. Haider and Akram⁵⁰ analyse the club convergence in EF and carbon footprint per capita among 77 countries over the period 1961–2014. Solarin et al.⁵¹ analyse the convergence in per capita EF and its six components for 92 countries over the period 1961–2014. Ulucak et al.⁵² test the convergence clubs for EF across Sub-Saharan African countries. Erdogan and Okumus⁵³ test the convergence process in EF for different income groups over the period 1961–2016. Apaydin et al.⁵⁴ test the convergence in EF across 130 countries over the period 1980–2016. Tillaguango et al.⁵⁵ test the convergence of per capita EF in 16 Latin American countries from 1990 to 2016. Yildirim et al.⁵⁶ test the club convergence for EU and candidate countries. Besides, Abbassi and Haq⁵⁷ show the existence of β-convergence for 88 developed and developing countries from 1978 to 2017 using spatial econometric models.

As mentioned above, there are also other ecological indicators used as a proxy for environmental degradation indicators. El-Montasser et al.⁵⁸ analyse the convergence in greenhouse gas emissions across G7 countries over the period 1990–2011. Apergis and Garzon⁵⁹ test the convergence in greenhouse gas emissions across 19 Spanish regions from 1990 to 2017 using the club convergence methodology. Solarin and Tiwari⁶⁰ investigate the convergence in sulfur dioxide across 32 OECD countries for the 1850–2005 period using the panel stationary test Ivanovski and Churchill⁶¹ analyse the convergence in nitrous oxide, methane, and CO₂ emissions at the regional level in Australia over the period 1990–2017. Zhang et al.⁶² investigate the convergence in sulfur dioxide emissions in 74 cities in China from December 2014 to June 2019. Haider et al.⁶³ test the convergence in biomass material footprint convergence across 172 countries from 1990 to 2017. Cui et al.⁶⁴ show the existence of absolute and conditional β-convergence for NO₂ in Chinese regions. The studies testing the convergence process of environmental indicators report mixed results. For instance, the studies by Lee et al.⁶⁵ and Lee and Chang⁶⁶ investigate the convergence in CO₂ emissions for the same countries (21 OECD) and the same period (1960–2000). However, while Lee et al.⁶⁵ show the existence of convergence, Lee and Chang⁶⁶ show a weak convergence.

Data and methodology

To test the environmental convergence hypothesis across MERCOSUR countries, the EF per capita measured in global hectares is used from 1961 to 2016. The data is gathered from the Open Data Platform of the Global Footprint Network¹. Table 1 presents descriptive statistics of the data.

Table 1.

Descriptive statistics.

	Mean	Median	Maximum	Minimum	Skewness	Kurtosis	Jarque-Bera	Probability
Argentina	3.5088	3.4409	4.6472	2.8844	1.0500	4.1130	13.1810*	0.0014
Brazil	2.8126	2.8919	3.1350	2.3405	−0.6810	2.2403	5.6757***	0.0586
Paraguay	3.6900	3.6799	4.8235	2.8998	0.1393	2.1012	2.0663	0.3559
Uruguay	4.4802	4.5856	6.6010	1.9202	−0.4948	3.1423	2.3320	0.3116
Venezuela	2.9015	2.9666	3.8512	1.6518	−0.3963	2.2522	2.7705	0.2503

Note: * and *** show the significance at the 1% and 10% levels.

The statistics in Table 1 show that only the EF series of Argentina and Brazil are distributed non-normally; the remaining series are normally distributed. By using the EF series, the following variable is created:

e_{n, t} = e f_{n, t} - \begin{matrix} \bar{e f_{t}} \end{matrix}

(1)

where

e f_{n, t}

is defined as

e f_{n, t} = \log (E F_{n, t})

E F_{n, t}

indicates the per capita EF series of country n, and

\begin{matrix} \bar{e f_{t}} \end{matrix}

shows the cross-country average log of per capita EF at time t. If

e_{n, t}

is found to be stationary at the level, it is concluded that the EF series of the country n, is converging to the mean; otherwise, it is concluded the series of the country n, is diverging.

FADF unit root test

Ignoring structural changes in unit root testing may lead to a false conclusion about the stationarity characteristics of the considered series, as emphasized by Perron⁶⁷ who suggests to allow structural changes by employing dummy variables and determine the breakpoint dates exogenously. Most of the subsequent studies allow endogenously determined structural breaks with dummy variables. There are mainly two limitations of these unit root tests; first, most of them allow a predefined number of breaks which causes biased results in the case of the different number of breaks in the data generation process. Second, dummy variables are capable of capturing only sharp breaks, and most economical series change slowly over time, as suggested by Hyndman.⁶⁸ So, in this study, instead of allowing breaks via dummy variables, the FADF unit root test is applied to consider multiple smooth changes. To implement the FADF unit root test, the following test equation is estimated:

Δ e_{t} = α_{0} + α_{1} e_{t - 1} + α_{2} \sin (\frac{2 π k t}{T}) + α_{3} \cos (\frac{2 π k t}{T}) + \sum_{i = 1}^{p} θ_{i} Δ e_{t - i} + u_{t}

(2)

where

e_{t}

is defined as in Eq. 1, trigonometric terms, sin and cos, constitute the Fourier function which is used for capturing multiple smooth changes. k, t,

T

, and p show a particular frequency, a trend term, sample size, and optimal lag length, respectively. By using the FADF unit root test, there is no need to determine the location or number of breaks. Only the optimal Fourier frequency,

k *

, is defined by estimating Eq. 2 for all values in the interval [0.1, 0.2, 0.3, …, 5] with 0.1 increments, and the value that minimizes the sum of squared residuals is selected. The value of p is determined by using information criteria such as Akaike or Schwarz.

After finding the $k *$ and p, Eq.2 is estimated, and the null hypothesis of the non-significance of the Fourier function, $H_{0} : α_{2} = α_{3} = 0$ , is tested using the usual F test. In the case of rejection of the null, the null of a unit root, $H_{0} : α_{1} = 0$ , can be tested employing the FADF test; otherwise, the traditional ADF unit root test should be used. The critical values for the F test are obtained from Enders and Lee,⁶⁹ while critical values for the FADF test for the fractional frequencies are obtained from Bozoklu et al.⁷⁰

Wavelet transform

By using the ADF or FADF unit root test, one can only test the convergence hypothesis for the overall period. However, the convergence on different time scales can be tested by using wavelet transformation. This study uses the DWT method to decompose the time series into different frequencies. The orthogonal approximation of the DWT is presented as:

e (t) = \sum_{k} s_{j, k} ϕ_{j, k} (t) + \sum_{k} d_{j, k} ψ_{j, k} (t) + \sum_{k} d_{j - 1, k} ψ_{j - 1, k} (t) + \dots + \sum_{k} d_{1, k} ψ_{1, k} (t)

(3)

where

ϕ_{j, k}

and

ψ_{j, k}

indicate the scaling and wavelet functions, respectively. The former and the latter capture long-run and short-run behaviour, respectively.

s_{j, k}

and

d_{j, k}

are the smoothing and detail coefficients that contain information about the trend and short-term components, respectively.

This study uses the maximal overlap DWT (MODWT) proposed by Walden,⁷¹ since DWT limits the ability to perform statistical analysis as the number of wavelet and scaling coefficients decreases. The wavelet ( $w_{j, t}$ ) and scaling ( $v_{j, t}$ ) coefficients of the MODWT can be presented as follows:

w_{j, t} = \sum_{l = 0}^{L - 1} \frac{ψ_{j, l}}{2^{j / 2}} Y_{t - l mod N}

(4)

v_{j, t} = \sum_{l = 0}^{L - 1} \frac{ϕ_{j, l}}{2^{j / 2}} Y_{t - l mod N}

(5)

where

Y_{t}

(

t = 0, 1, 2, 3, \dots, L - 1

) denotes the time series. By following the suggestion of Gencay et al.,⁷² Daubechies Least Asymmetric as wavelet filter and eight as wavelet length (LA8) are employed. To find the largest scale, the following formula is employed:

J = \log_{2} (T)

(6)

where T is the sample size. So, the ADF or FADF unit root tests are applied to test the unit root in the original series and also in the decomposed series.

By considering only the original series, the convergence hypothesis is tested only for the considered time period. However, by using the wavelet transform technique, the series can be decomposed into different scales. The maximum scale is found to be six by using the formula in (5); thus, six different series from the original series, denoted as d1, d2, d3, d4, d5, and s5, are obtained. d1 shows the lowest time scale, while d5 presents the largest time scale and s5 represents the trend of the original series. By following Andersson⁷³ and Ha et al.,⁷⁴ instead of using all the decomposed components individually, the components are combined to get three time horizons; while the short-run (less than eight years) is obtained by combining d1 and d2, the medium-run (8–32 years) is obtained by considering d3, long-run (over 32 years) is obtained by combining d4 and d5.

Empirical results

To test the convergence hypothesis, the stationarity of each $e_{t}$ series computed as in Eq. 1 must be tested. First, the unit root tests are applied for comparison purposes without wavelet decomposition. Table 2 provides the test results:

Table 2.

Unit root test results for original series.

Series	Frequency	Min. SSR	F Test Stat.	p	FADF Test Stat.	ADF Test Stat.
Argentina	1.2	2.646	4.131	1	−3.394	−2.592 (0) [0.101]
Brazil	3.2	0.389	5.310	2	−2.632	−1.998 (0) [0.287]
Paraguay	1	2.129	7.187***	4	−4.039	−1.675 (1) [0.438]
Uruguay	0.1	13.673	8.573**	1	−4.094**
Venezuela	2.1	1.992	3.542	4	−2.365	−2.365 (4) [0.157]

Note: ** and *** show the significance at the 5% and 10% levels, respectively. The critical values of the F test at the 5% and 10% levels are 7.58 and 6.35, respectively. The critical value at the 5% level for the FADF test is −3.88 for the 0.1 frequency. Numbers in the parentheses and brackets show the optimal lag length and p-values, respectively.

The findings in Table 2 show that trigonometric terms are significant only for Uruguay's series. So, the convergence hypothesis for Uruguay can be tested using the FADF unit root test; for the remaining countries, the ADF unit root test should be used. Overall, test results present evidence of convergence only for Uruguay since the test statistic of the FADF test is higher than the critical value. The findings support the evidence of divergence for the remaining countries. Table 3 presents the convergence test results for the short-run.

Table 3.

Convergence results for short-run.

Countries	Frequency	Min. SSR	F Test Stat.	p	FADF Test Statistics	ADF Test Statistics
Argentina	0.1	0.0103	0.670	8	−6.440	−6.439* (8)[0]
Brazil	3.8	0.009	0.167	8	−5.762	−5.887* (8)[0]
Paraguay	5	0.010	0.289	10	−5.159	−5.347* (10)[0.0001]
Uruguay	4.4	0.035	0.305	10	−5.856	−6.005* (10)[0]
Venezuela	5	0.023	4.083	10	−5.326	−4.993* (10)[0.0002]

Note: * shows the significance at the 1% level. The critical value of the F test at the 10% level is 6.35. The numbers in the parentheses and brackets show the optimal lag length and p-values, respectively.

Findings in Table 3 show that the trigonometric terms are not statistically significant for all the considered countries. So, the ADF test is applied for these series. The ADF unit root test results show that all the considered series are stationary; that is, there is strong evidence of convergence for all countries in the short-run. Next, the convergence hypothesis is tested for the medium-run. Table 4 shows the results for the medium-run.

Table 4.

Convergence results for Medium-run.

Countries	Frequency	Min. SSR	F Test Stat.	p	FADF Test Statistics	ADF Test Statistics
Argentina	5	0.002	1.042	8	−2.463	−3.392** (8)[0.0162]
Brazil	5	0.001	19.148*	10	−1.869
Paraguay	3.7	0.001	7.384	8	−4.208	−2.718*** (8)[0.0786]
Uruguay	3.5	0.001	2.163	7	−6.133	−8.452* (7)[0]
Venezuela	5	0.002	1.183	7	−5.699	−6.837 *(7)[0]

Note: *, ** and *** show the significance at the 1%, 5%, and 10% levels, respectively. The critical value of the F test at the 1% level is 10.35. The critical value at the 10% level for the FADF test is −2.588 for Frequency five. Numbers in the parenthesis and brackets show the optimal lag length and p-values, respectively.

Trigonometric terms are significant for only the series of Brazil in the medium-run. So, the FADF unit root test is applied to Brazil, which shows that the series of Brazil has a unit root in the medium-run. Besides, the ADF unit root test is applied to the remaining countries. The results support the evidence of stationarity for the $e_{t}$ series of these countries, so one can conclude that the convergence hypothesis is valid for Argentina, Paraguay, Uruguay, and Venezuela in the medium-run. Finally, Table 5 shows the convergence process for the long-run.

Table 5.

Convergence results for long-run.

Countries	Frequency	Min. SSR	F Test Stat.	p	FADF Test Statistics	ADF Test Statistics
Argentina	0.9	0.001	12.628*	10	−7.273*
Brazil	3.6	0.001	2.0185	10	−1.222	−0.593 (10)[0.862]
Paraguay	3.1	0.001	1.7850	10	−0.167	−3.718* (10)[0.007]
Uruguay	0.5	0.001	21.119*	10	−2.331
Venezuela	2	0.001	1.321	10	−4.945	−5.095* (10)[0.000]

Note: *, ** and *** show the significance at the 1%, 5%, and 10% levels, respectively. The critical value of the F test at the 1% level is 10.35. The critical value at the 1% level for the FADF test at the 0.9 frequency is −4.461, and at the 10% level for the 0.5 frequency is −3.636. The numbers in the parenthesis and brackets show the optimal lag length and p-values, respectively.

According to the test results in Table 5, the null of non-significance of the Fourier function can be rejected for only Argentina and Uruguay. So, the FADF unit root test is applied to these countries. The results of the FADF test show that the series of Argentina is stationary. For the remaining series, the ADF unit root test is applied. The results of the ADF test support the evidence of stationarity for only Paraguay and Venezuela's series; so it is concluded that the convergence hypothesis holds for Argentina, Paraguay, and Venezuela in the long-run.

The descriptive statistics in Table 1 indicate large disparities between mean values of EF across countries. It is observed that Uruguay has the highest mean value of EF per capita. Uruguay has been growing rapidly in recent years. In other words, increasing economic activity may lead to greater economic degradation. Besides, Brazil has the lowest mean value. It is observed that these two countries, Uruguay and Brazil, are not converging to average in the long-run. All these results provide important insights into the convergence process of environmental degradation. When the different frequencies are not considered, the convergence process is valid only for Uruguay. However, by analysing the convergence process in different frequencies using wavelet decomposed series, more evidence of convergence is observed in the MERCOSUR countries. For instance, while the findings support the evidence of convergence for all countries in the short-run, four countries converge in the medium-run, and only three countries converge in the long-run, which indicates that implementing common environmental policies to reduce EF may be effective only in the short-run, and they are not effective in a longer term. Besides, the divergence among countries in the long term may be explained by differences in structural characteristics and growth dynamics across countries.

Discussion

The results show that the convergence process in EF across MERCOSUR countries differs across different time scales. This difference among different time scales deserves further explanation. First, the results show the validity of convergence hypothesis in the short-run, which is reasonable. The impact of growth fundamentals, structural characteristics of countries, or environmental policies on environmental quality may not emerge in the short-run. However, one can expect that the effect of growth fundamentals, structural differences of economies, or environmental policies on environmental degradation become important in the medium or long-run, which is supported by results in the study stating that EF across MERCOSUR countries shows the different convergence tendencies in the medium or long-run.

It is observed that while the convergence hypothesis holds in the short-run, some countries show divergence in the medium and long-run. One of these countries is Brazil which stands out from other countries in a positive way since it has the lowest EF in the sample. Brazil's government put into force the National Environmental Policy in 1981 for sustainable environmental development. The act aims to preserve, improve and recover the environmental quality conducive to a healthy life, to ensure socio-economic development, the interests of national security, and the protection of human life (Act No. 6.938).⁷⁵ According to the findings, Uruguay, which has the maximum EF in the sample, also shows a divergence in the long-run. Uruguay has struggled with serious environmental threats, such as air and water pollution, the effect of livestock on the environment, deforestation, etc. Besides, Uruguay has had higher growth rates in recent years. Besides, economic activities in Uruguay have been increasing recently, which can directly increase EF levels. Therefore, the result of the divergence of Uruguay seems reasonable. Furthermore, the EF series show fluctuations over time, and disparities in the EF series across countries are not stable over time, which supports that it is important not only to test the convergence process for the whole period but different periods.

To the authors' best knowledge, there is no study on the convergence process of EF per capita across MERCOSUR countries. Therefore, the results of this study can be compared with studies investigating convergence in EF across Latin American countries. The results are partially consistent with the study of Tillaguango et al.,⁵⁵ which identifies multiple convergence clubs for EF across 16 Latin American countries, including four MERCOSUR countries. Although the short-run results of this study are not consistent with the results of Tillaguango et al.,⁵⁵ the long-run results are consistent with the authors' study. There are a few studies investigating convergence in CO₂ emissions among MERCOSUR (or Latin America) countries. Panopoulou and Pantelidis⁷⁶ show evidence of convergence across 24 Latin America and the Caribbean over the period 1960–2003, which implies that the long- run results in this study are not consistent with the results of Panopoulou and Pantelidis.⁷⁶ Christidou et al.⁷⁷ test the convergence for 36 countries, including four MERCOSUR countries, Argentina, Brazil, Uruguay, and Venezuela. The results of Christidou et al.⁷⁷ which show that the null of a unit root cannot be rejected for Uruguay and Venezuela are not consistent with the long-run results in this study. Acaravci and Erdogan⁷⁸ test the convergence in CO₂ emissions for seven regions, including Latin America and the Caribbean region. The results of the study show the regional convergence for Latin America and Caribbean, which is not consistent with the results of this study in the long-run. Robalino-López et al.⁷⁹ test the convergence process in CO₂ emissions per capita across 10 South American countries, including five MERCOSUR countries, over 1980–2010 using Phillips and Sul⁴⁸ methodology. The results for long-run of the study are consistent with the study of Robalino-López et al.,⁷⁹ which identifies multiple convergence clubs. Similar to Robalino-López et al.,⁷⁹ the results of this study are partially consistent with the study of Belloc and Molina,⁸⁰ which identify multiple convergence clubs in greenhouse gas emissions across Latin American countries. Even if the results of this study are compared with the existing literature, there is no study testing convergence in EF across MERCOSUR countries for different time scales. Therefore, comparing the results of this study with the current literature may cause misleading interpretations. Since the studies on convergence in ecological indicators have used various environmental indicators, different country samples, including some MERCOSUR countries, and other econometric methods, inconsistent findings across studies seem reasonable.

Conclusion

This study analyses the convergence in the EF per capita across MERCOSUR countries over the period 1961–2016, using the FADF and ADF unit root tests. However, unlike the existing literature, which mainly tests the convergence for the overall period, this study follows a different path regarding the time scales. First, the DWT method is employed to decompose the time series into different frequencies. Then, the convergence in EF per capita is tested among MERCOSUR countries for different time scales, short-run, medium-run, and finally, long-run.

The results of unit root tests for the overall period show that the convergence hypothesis holds only for Uruguay; that is, Uruguay converges to the group average. The findings support the evidence of divergence for the remaining countries. After testing the convergence for the overall period, using the DWT method, three time horizons are identified; the short-run (less than eight years), medium-run (8–32 years), and long-run (more than 32 years). In the short-run, the unit root test results support the evidence of convergence for all countries. The unit root test results for the medium-run reveal that the convergence hypothesis is valid for all countries except for Brazil, which shows divergence tendencies. Finally, the results show that the convergence hypothesis holds for Argentina, Paraguay, and Venezuela in the long-run.

In line with the major findings of this study, several relevant policies can be recommended. Based on the results in this study, since different convergence patterns are identified for different time periods, common and similar environmental policies may not be effective for MERCOSUR countries. Furthermore, MERCOSUR countries should consider the time periods when implementing environmental policies. For instance, since the convergence hypothesis is valid for the short-run, the common environmental policies may be effective in the short-run. However, in the medium and long-run, countries may consider implementing different policies for different time periods. However, several comprehensive policies can be recommended from a broader perspective, not only considering the results in this study. Since environmental degradation is affected by several factors, policymakers should implement comprehensive policies to mitigate environmental degradation. For instance, governments should encourage the private sector to invest in discovering technologies to reduce environmental degradation. Governments should enforce rigid environmental protection laws to prevent dirty foreign direct investments. Intra-regional trade across the member states of MERCOSUR is also an important factor in environmental degradation. Promoting intra-regional trade, especially through cross-border, can facilitate renewable energy trade to decrease the environmental pollution. Within this context, governments of members of MERCOSUR should increase intra-regional trade and implements policies and regulations that increase intra-regional trade. One could say that institutional structure is one of the significant obstacles to maintaining environmental quality and reducing environmental pollution. At this point, governments and policymakers should focus on policies that increase institutional quality and build a strong legal system.

Footnotes

Author contributions

Uğur Ursavaş: Conceptualization, Methodology, Writing- Original draft preparation, Editing. Veli Yilanci: Conceptualization, Methodology, Software, Editing. All authors read and approved the final manuscript.

Declaration of Conflicting Interests

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

Funding

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

ORCID iD

Uğur Ursavaş

Data availability

The data used in this study can be found from Global Footprint Network

Notes

Uğur Ursavaş is a research assistant at Zonguldak Bulent Ecevit University and a researcher at the Quantitative Regional and Computational Science (QuaRCS) Lab. His research topics are economic growth, regional development, environmental economics, and spatial analysis.

Veli Yilanci is Professor at Çanakkale Onsekiz Mart University, Turkey. His research topics include time series analysis, panel data analysis, applied economics, energy economics, and environmental economics.

References

Solow

. A contribution to the theory of economic growth. Q J Econ 1956; 70: 65–94.

Swan

. Economic growth and capital accumulation. Econ Rec 1956; 32: 334–361.

Cass

. Optimum growth in an aggregative model of capital accumulation. Rev Econ Stud 1965; 32: 233–240.

Koopmans

. On the concept of optimal growth. In: Johansen

(ed.) Econometric approach to development planning. 1st ed. North Holland Amsterdam, 1965, pp.225–287.

Baumol

. Productivity growth, convergence, and welfare: what the long-run data show. Am Econ Rev 1986; 76: 1072–1085.

Barro

Sala-i-Martin

. Convergence across states and regions. Brookings Pap Econ Act 1991; 22: 107–182.

Barro

Sala-i-Martin

. Convergence. J Polit Econ 1992; 100: 223–251.

Islam

. What have we learnt from the convergence debate? J Econ Surv 2003; 17: 309–362.

Johnson

Papageorgiou

. What remains of cross-country convergence? J Econ Lit 2020; 58: 129–175.

10.

Apergis

Georgellis

. Does happiness converge? J Happiness Stud 2015; 16: 67–76.

11.

Mendez

Kataoka

. Disparities in regional productivity, capital accumulation, and efficiency across Indonesia: a club convergence approach. Rev Dev Econ 2021; 25: 790–809.

12.

Kisswani

Nusair

. Nonlinear convergence in Asian interest and inflation rates: evidence from Asian countries. Econ Chang Restruct 2014; 47: 155–186.

13.

Maynou

Monfort

Morley

, et al. Club convergence in European housing prices: The role of macroeconomic and housing market fundamentals. Econ Model 2021; 103: 1-11

14.

Brock

Taylor

. The Kindergarten rule of sustainable growth. NBER Working Paper 9597, 2003.

15.

Aldy

. Per capita carbon dioxide emissions: convergence or divergence? Environ Resour Econ 2006; 33: 533–555.

16.

Acar

Söderholm

Brännlund

. Convergence of per capita carbon dioxide emissions: implications and meta-analysis. Clim Policy 2018; 18: 512–525.

17.

Bilgili

Ulucak

. Is there deterministic, stochastic, and/or club convergence in ecological footprint indicator among G20 countries? Environ Sci Pollut Res 2018; 25: 35404–35419.

18.

Xue

Haseeb

Mahmood

, et al. Renewable energy use and ecological footprints mitigation: evidence from selected South Asian economies. Sustain 2021; 13: 1613.

19.

Murshed

Rahman

Alam

, et al. The nexus between environmental regulations, economic growth, and environmental sustainability: linking environmental patents to ecological footprint reduction in South Asia. Environ Sci Pollut Res 2021; 28: 49967–49988.

20.

Murshed

Ahmed

Kumpamool

, et al. The effects of regional trade integration and renewable energy transition on environmental quality: evidence from South Asian neighbours. Bus Strategy Environ 2021; 30: 4154–4170.

21.

Murshed

Nurmakhanova

Al-Tal

, et al.

Can intra-regional trade, renewable energy use, foreign direct investments, and economic growth mitigate ecological footprints in South Asia?

Energ Source Part B 2022: 1–26.

22.

Koengkan

Fuinhas

Santiago

. The relationship between CO₂ emissions, renewable and non-renewable energy consumption, economic growth, and urbanization in the southern common market. J Environ Econ Policy 2020; 9: 383–401.

23.

Koengkan

Fuinhas

Marques

. Does financial openness increase environmental degradation? Fresh Evidence from MERCOSUR Countries. Environ Sci Pollut Res 2018; 25: 30508–30516.

24.

Wackernagel

. Ecological footprint and appropriated carrying capacity: a tool for planning toward sustainability . PhD. Thesis, The University of British Columbia, Canada, 1994.

25.

Wachernagel

Rees

. Our ecological footprint: reducing human impact on the earth. Canada: New Society Publishers, 1998.

26.

Kitzes

Wackernagel

Loh

, et al. Shrink and share: humanity’s present and future ecological footprint. Philos Trans R Soc Lond, B, Biol Sci 2008; 363: 467–475.

27.

Galli

Wackernagel

Iha

, et al. Ecological footprint: implications for biodiversity. Biol Conserv 2014; 173: 121–132.

28.

Strazicich

List

. Are CO₂ emission levels converging among industrial countries? Environ Resour Econ 2003; 24: 263–271.

29.

Jobert

Karanfil

Tykhonenko

. Convergence of per capita carbon dioxide emissions in the EU: legend or reality? Energy Econ 2010; 32: 1364–1373.

30.

Camarero

Picazo-Tadeo

Tamarit

. Are the determinants of CO₂ emissions converging among OECD countries? Econ Lett 2013; 118: 159–162.

31.

Solarin

. Convergence of CO₂ emission levels: evidence from African countries. Int J Econ Res 2014; 19: 65–92.

32.

Tiwari

Mishra

. Testing the CO₂ emissions convergence: evidence from Asian countries. IIM Kozhikode Soc Manag Rev 2017; 6: 67–72.

33.

Churchill

Inekwe

Ivanovski

. Conditional convergence in per capita carbon emissions since 1900. Appl Energy 2018; 228: 916–927.

34.

Payne

Apergis

. Convergence of per capita carbon dioxide emissions among developing countries: evidence from stochastic and club convergence tests. Environ Sci Pollut Res 2020; 28: 33751–33763.

35.

Erdogan

Solarin

. Stochastic convergence in carbon emissions based on a new Fourier-based wavelet unit root test Environ Sci Pollut Res 2020; 28: 21887–21899.

36.

Tang

Chang

. CO₂ Emissions converge in the 50 US states—sequential panel selection method. Econ Model 2014; 40: 320–333.

37.

Burnett

. Club convergence and clustering of US energy-related CO₂ emissions. Resource Energy Econ 2016; 46: 62–84.

38.

Guo

, et al. Convergence of carbon dioxide emissions in Chinese cities: a continuous dynamic distribution approach. Energy Policy 2016; 91: 207–219.

39.

Apergis

Payne

. Per capita carbon dioxide emissions across US states by sector and fossil fuel source: evidence from club convergence tests. Energy Econ 2017; 63: 365–372.

40.

Wang

Zhang

. Convergence of carbon dioxide emissions in different sectors in China. Energy 2014; 65: 605–611.

41.

Brännlund

Lundgren

Söderholm

. Convergence of carbon dioxide performance across Swedish industrial sectors: an environmental index approach. Energy Econ 2015; 51: 227–235.

42.

Fan

, et al. Convergence of carbon emissions intensity across Chinese industrial sectors. J Clean Prod 2018; 194: 179–192.

43.

Liu

. Convergence of energy carbon emission efficiency: evidence from manufacturing sub-sectors in China. Environ Sci Pollut Res 2022: 1–15.

44.

Solarin

. Convergence in CO₂ emissions, carbon footprint and ecological footprint: evidence from OECD countries. Environ Sci Pollut Res 2019; 26: 6167–6181.

45.

Yilanci

Pata

. Convergence of per capita ecological footprint among the ASEAN-5 countries: evidence from a nonlinear panel unit root test Ecol Indic 2020; 2020: 106178.

46.

Işık

Ahmad

Ongan

, et al. Convergence analysis of the ecological footprint: theory and empirical evidence from the USMCA countries. Environ Sci Pollut Res 2021; 28: 32648–32659.

47.

Yilanci

Gorus

Solarin

. Convergence in per capita carbon footprint and ecological footprint for G7 countries: evidence from panel Fourier threshold unit root test Energy Environ 2021; 2021:1-19. https://doi.org/10.1177/0958305X211011461

48.

Phillips

Sul

. Transition modeling and econometric convergence tests. Econometrica 2007; 75: 1771–1855.

49.

Ulucak

Apergis

. Does convergence really matter for the environment? An application based on club convergence and on the ecological footprint concept for the EU countries. Environ Sci Policy 2018; 80: 21–27.

50.

Haider

Akram

. Club convergence analysis of ecological and carbon footprint: evidence from a cross-country analysis. Carbon Manag 2019; 10: 451–463.

51.

Solarin

Tiwari

Bello

. A multi-country convergence analysis of ecological footprint and its components. Sustain Cities Soc 2019; 46: 101422.

52.

Ulucak

Kassouri

İlkay

SÇ

, et al. Does convergence contribute to reshaping sustainable development policies? Insights from Sub-Saharan Africa. Ecol Indic 2020; 112: 106140.

53.

Erdogan

Okumus

. Stochastic and club convergence of ecological footprint: an empirical analysis for different income group of countries. Ecol Indic 2021; 121: 107123.

54.

Apaydin

Ursavaş

Koç

. The impact of globalization on the ecological footprint: do convergence clubs matter? Environ Sci Pollut Res 2021; 28: 53379–53393.

55.

Tillaguango

Alvarado

Dagar

, et al. Convergence of the ecological footprint in Latin America: the role of the productive structure. Environ Sci Pollut Res 2021; 28: 59771–59783.

56.

Yildirim

DÇ

Yildirim

Erdoğan

, et al. Time-Varying convergences of environmental footprint levels between European countries. Energies 2021; 14: 1813.

57.

Abbassi

Haq

. A spatial econometric analysis of convergence in global environmental performance. Environ Sci Pollut Res 2022: 1–14.

58.

El-Montasser

Inglesi-Lotz

Gupta

. Convergence of greenhouse gas emissions among G7 countries. Appl Econ 2015; 7: 6543–6552.

59.

Apergis

Garzón

. Greenhouse gas emissions convergence in Spain: evidence from the club clustering approach. Environ Sci Pollut Res 2020; 27: 38602–38606.

60.

Solarin

Tiwari

. Convergence in sulphur dioxide (SO₂) emissions since 1850 in OECD countries: evidence from a new panel unit root test Environ Model Assess 2020; 25: 665–675.

61.

Ivanovski

Churchill

. Convergence and determinants of greenhouse gas emissions in Australia: a regional analysis. Energy Econ 2020; 92: 104971.

62.

Zhang

Zhao

. The convergence of sulphur dioxide (SO₂) emissions per capita in China. Sustain 2020; 12: 1781.

63.

Haider

Akram

Ali

. Does biomass material footprint converge? Evidence from club convergence analysis. Environ Sci Pollut Res 2021; 28: 27362–27375.

64.

Cui

Wang

Jiang

, et al. Dynamic spatial analysis of NO₂ pollution over China: satellite observations and spatial convergence models. Atmos Pollut Res 2021; 12: 89–99.

65.

Lee

Chang

Chen

. Do CO₂ emission levels converge among 21 OECD countries? New evidence from unit root structural break tests. Appl Econ Lett 2008; 15: 551–556.

66.

Lee

Chang

. New evidence on the convergence of per capita carbon dioxide emissions from panel seemingly unrelated regressions augmented Dickey-Fuller tests. Energy 2008; 33: 1468–1475.

67.

Perron

. The great crash, the oil price shock, and the unit root hypothesis. Econometrica 1989; 57: 1361–1401.

68.

Hyndman

. Structural breaks. https://robjhyndman.com/hyndsight/structural-breaks/ (2014, accessed 30 December 2021).

69.

Enders

Lee

. The flexible Fourier form and Dickey–Fuller type unit root tests. Econ Lett 2012; 117: 196–199.

70.

Bozoklu

Yilanci

Gorus

. Persistence in per capita energy consumption: a fractional integration approach with a Fourier function. Energy Econ 2020; 91: 1-12.

71.

Walden

. Wavelet analysis of discrete time series. In European Congress of mathematics, Birkhäuser, Basel, 2001, pp. 627–641.

72.

Gencay

Gradojevic

Selcuk

, et al. Asymmetry of information flow between volatilities across time scales. Quant Finance 2010; 10: 895–915.

73.

Andersson

. Identifying and modelling cycles and long waves in economic time series. In: Ljungberg

(ed) Structural analysis and the process of economic development. London: Routledge, 2016, pp.54–75.

74.

Tan

Goh

. Linear and nonlinear causal relationship between energy consumption and economic growth in China: new evidence based on wavelet analysis. PloS one 2018; 13: e0197785.

75.

Environmental Policy Act No. 6.938., https://www.fao.org/faolex/results/details/en/c/LEX-FAOC012932/ (1981, accessed 8 December 2021).

76.

Panopoulou

Pantelidis

. Club convergence in carbon dioxide emissions. Environ Resour Econ 2009; 44: 47–70.

77.

Christidou

Panagiotidis

Sharma

. On the stationarity of per capita carbon dioxide emissions over a century. Econ Model 2013; 33: 918–925.

78.

Acaravci

Erdogan

. The convergence behavior of CO₂ emissions in seven regions under multiple structural breaks. Int J Energy Econ Policy 2016; 6: 575–580.

79.

Robalino-López

García-Ramos

Golpe

, et al. CO₂ Emissions convergence among 10 South American countries. A study of Kaya components (1980–2010). Carbon Management 2016; 7: 1–12.

80.

Belloc

Molina

. Are Greenhouse Gas Emissions Converging in Latin America?. GLO Discussion Paper Series 1037, 2022.