Download as pdf or txt
Download as pdf or txt
You are on page 1of 15

Environmental Science and Pollution Research

https://doi.org/10.1007/s11356-020-09109-0

RESEARCH ARTICLE

Is there a waste Kuznets curve for OECD? Some evidence


from panel analysis
Fırat Yılmaz 1

Received: 11 March 2020 / Accepted: 28 April 2020


# Springer-Verlag GmbH Germany, part of Springer Nature 2020

Abstract
We empirically investigated the existence of a Kuznets curve for OECD area in terms of municipal solid waste generation by
utilizing a panel data covering the 2002–2017 period. Results from GMM and between estimators conclude that an inverted U-
shaped Kuznets relationship holds for OECD countries, but only in relative terms. Results show that consumption expenditure is
one of the main drivers of waste generation. Country-based estimations of random coefficient approach state that 8 out of 16 high-
income countries and only 2 out of low-income countries have achieved decoupling the waste generation from consumption
expenditures. Some policy implications are recommended.

Keywords Kuznets curve . Waste . Decoupling

Introduction population, high-income countries are responsible for 34%


or equally 683 million tonnes of the total waste generated
Solid waste generation is inevitable as long as the human race (Kaza et al. 2018). Figures 1 and 2 show global MSW gener-
exists (Shekdar 2009). Rapid urbanization, increasing popula- ation by country and income groups respectively.
tion, rising prosperity, and consumption levels have positively The MSW generated has increased mostly coherent with
contributed to the amount of solid waste generation and led to private consumption expenditures and GDP in OECD area
a global challenge of sustainable management of waste prob- during the 1990s. This increase seems to have slowed down
lem (Beede and Bloom 1995; Wang and Nie 2001; Saeed et al. after 2000; nevertheless, the total amount of MSW generated
2009; Getahun et al. 2012; Kaartinen et al. 2013; Khan et al. has already surpassed 650 million tonnes (equally 522 kg per
2016) and put the pressure on earth so that the waste generated capita) annually. The amount and components of the MSW
has already gone beyond 30% of Earth’s carrying capacity varies due to income difference, consumption patterns, urban-
(Irwan et al. 2013). ization rates, lifestyles, and waste management practices. The
The amount of waste generated has reached alarming recycling rates have shown significant increase; however,
levels. Each year, 2.01 billion tonnes of municipal solid waste landfilling is still the main way how the MSW is treated in
(MSW) is generated globally. Optimistically, 33% of this some OECD countries (OECD 2016). Figure 3 shows the
waste is not treated with an environmentally safe way. The amount of MSW generated among OECD countries.
average daily per capita MSW generated is around 0.74 kg In this paper, we focus on the MSW generation dynamics
with a variation between 0.11 and 4.54 kg. The amount of of the OECD area. Our contributions to the literature are as
MSW generated is predicted to increase 3.40 billion tonnes follows: firstly, according to our knowledge Cole et al. (1997)
till 2050. Although they form only 16% of the world’s is the only study analyzing the relationship between economic
growth and MSW generation in an environmental Kuznets
curve (EKC) framework for OECD countries. Thus, the liter-
Responsible editor: Nicholas Apergis ature still lacks of empirical evidence. Secondly, even though
some authors stated the benefits of within-country studies
* Fırat Yılmaz (Ercolano et al. 2018), cross-country studies may still be ben-
firatyilmaz@antalya.edu.tr
eficial since they display the “big picture” of the situation.
1
Department of Economics, Antalya Bilim University, Lastly, we utilize a panel dataset which allows to control un-
Antalya, Turkey observed heterogeneity, and provide more variability and
Environ Sci Pollut Res

Fig. 1 Global MSW generation (source: Kaza et al. 2018)

richer observations (Baltagi 2008; Wooldridge 2016). the cross-section units and allows us to run separate regres-
Moreover, econometric techniques allow us to capture the sions for each country. The rest of the paper is organized as
dynamic behavior of the variables and perform separate equa- follows: “Decoupling concept” discusses the decoupling con-
tions for each country within the panel data set. For this pur- cept, “Kuznets curve for municipal solid waste” links EKC
pose, we applied the system generalized method of moments hypothesis to MSW generation and summarizes the previous
(GMM) approach by Arellano and Bover (1995) and Blundell literature, “Variables of interest” derives the variables of inter-
and Bond (1998) and the random coefficient (RC) estimator est, “Methodology and data” clarifies the source of the data
by Swamy (1970) which permits slope heterogeneity among and develops the methodology, “Results” presents the results,
“Discussion” gives a detailed discussion of the results, and
“Policy implications and conclusion” presents a brief conclu-
sion and recommends policy implications.

Decoupling concept

One of the major steps in the transition process towards to


circular economy is the decoupling of material use from eco-
nomic activity (Foundation EM 2015). The risks of unsustain-
able and exponential growth have been spoken out by the
seminal work “Limits of Growth” in 1972 (Meadows et al.
1972). The tangible effects of climate change and environ-
mental degradation have led current growth rates and econom-
ic conditions to be questioned, and decoupling has emerged as
a substantial alternative to the current policies (Ward et al.
2016).
Decoupling can be defined as separating the tie between
Fig. 2 MSW generated by income groups (source: Kaza et al. 2018) economic activity and its environmental impacts. Absolute
Environ Sci Pollut Res

Fig. 3 MSW generated by OECD


countries (source: OECD 2016)

decoupling exists when economic activity increases while its Last decades have also witnessed the significant efforts of
environmental impacts decrease. Decoupling turns into a rel- curbing the MSW generation in OECD area. These efforts
ative form if environmental impacts increase but a slower rate have brought along its fruits so that the amount of MSW
than economic growth (OECD 2009). generated per capita seems to have the signs of relative
The OECD area has experienced a rise in material con- decoupling. The MSW generated decreased by about 3.5%
sumption in the same line with the GDP growth during the in OECD area between 2000 and 2009. Some part of this
past 30 years. Nevertheless, due to increased productivity, the reduction may be due to global financial crises; nonetheless,
material consumption has also been relatively decoupled dur- the role of sustainable waste management policies cannot be
ing the same period. The domestic material productivity has denied (OECD 2009). Figure 5 presents the decoupling trend
jumped from 1 $ per kg to 1.8 $ per kg between 1980 and for MSW generation in OCED area.
2010. The OECD area generates 50% and 30% more value per Even though the relative decoupling of MSW from eco-
unit of material use than in 1990 and 2000 respectively. nomic activity seems to be achieved for OECD area, there is
Decoupling has been achieved across material groups in no clear evidence how far this is decoupling from absolute
OECD area in relative terms, however, not in absolute terms terms. In this study, we aim to fill this gap and provide evi-
(OECD 2015). Figure 4 shows the decoupling trends for ma- dence by analyzing the main drivers of MSW generation in a
terial consumption in OECD area. Kuznets curve framework.

Kuznets curve for municipal solid waste

The debate about how economic growth affects environmen-


tal quality has long been focus of interest among scholars. In
this sense, the EKC concept has emerged after the work of

Fig. 4 Material consumption trend in OECD (source: Agrawala 2017) Fig. 5 Municipal waste trend in OECD (source: Agrawala 2017)
Environ Sci Pollut Res

Grossman and Krueger (1991). The EKC hypothesis, which is However, Gui et al. (2019) found a U-shaped relationship
originally based on Kuznets (1955), formulates an inverted U- rather than an inverted one.
shaped relationship between economic growth and environ- Most of the regional studies also seem to approve the ex-
mental degradation. In early stages of economic development, istence of WKC; nevertheless, some studies show mixed or no
both economic growth and its environmental impacts follow a evidence. Mazzanti et al. (2006) found evidence for only at
same trend; they both increase. After some point of income province level. Ichinose et al. (2011) utilized cross-section
level, namely the “turning point,” the trend turns opposite in data and found that the WKC is valid for Japanese
favor of environmental quality which means environmental municipalities. Using a data set for Lombardy region in
degradation decreases while economic growth increases. Italy, Ercolano et al. (2018) found proof for the WKC curve.
Nonetheless, there is still lack of consensus among scholars Madden et al. (2019) performed analysis for New South Wales
over the existence of the EKC relationship (Stern 2004). state of Australia. They found that municipalities in rural areas
Economic growth can affect the environmental quality with relatively higher MSW generation and lower income
through three ways (Copeland and Taylor 2004; Brock and ratify the WKC curve. On the other hand, for example,
Taylor 2005). Firstly, increasing economic growth needs more Jaligot and Chenal (2018) carried out an analysis for Vaud
inputs which means a rise in environmental degradation. This canton in Switzerland and found evidence against the WKC.
negative impact of economic growth is the scale effect. Additionally, results of Chen (2010) suggested an N-shaped
Secondly, changes in income or preferences may also alter relationship rather than an inverted U-shaped between dispos-
production technologies as well as per capita emissions. This able income and waste generation.
implies a varying relationship between income and environ- Most of the cross-country studies in the literature have been
mental quality since the damage of each pollutant differs. carried out for EU countries. Mazzanti and Zoboli (2008) used
Lastly, the whole structure of economy may change due to a panel data set for EU-25 countries between 1995 and 2005
rise in income. As a consequence, cleaner or dirtier activities and found statistically significant coefficients signalling the
may increase. This may result to a positive or negative impact existence of a WKC relationship. However, their results show
on environment, referred to the composition effect (Tsurumi no traces about absolute delinking. Another study is conduct-
and Managi 2010). ed for EU-15 countries by Nicolli et al. (2012). Their results
The EKC framework has inspired the scholars to investi- are similar to that of Mazzanti and Zoboli (2008). Arbulu and
gate the existence of a waste Kuznets curve (WKC) and an Rey-Maquieira (2015) implemented a panel analysis covering
emerging body of literature have come into existence. Studies 1997–2010 period for EU-32 countries. The main purpose of
have been conducted at city (Trujillo-Lora et al. 2013; Miyata their research, however, is to measure the impact of tourism on
et al. 2013; Gui et al. 2019), regional (Mazzanti et al. 2006; MSW generation. The authors used the EKC framework and
Ercolano et al. 2018; Jaligot and Chenal 2018; Madden et al. confirmed that there exists a WKC curve for MSW
2019), and cross-country levels (Cole et al. 1997; Mazzanti generation.
and Zoboli 2008; Nicolli et al. 2012; Arbulu and Rey- Among the cross-country studies conducted for other coun-
Maquieira 2015; Gnonlonfin et al. 2017). As far as we know, try groups, the study by Cole et al. (1997) is the only paper we
Cole et al. (1997) is the only paper examining the existence of came across for OECD countries. The authors used many
a WKC curve among OECD countries. Another paper by environmental indicators; however, they found no evidence
Johnstone and Labonne (2004) studied the determinants of for MSW generation but rather for local air pollutants. For
MSW generation in OECD countries. However, the authors Mediterranean countries, Gnonlonfin et al. (2017) exploited
did not use the EKC framework. a panel data set covering the 1990–2010 period. The results of
An inverted U-shaped relationship is likely to occur if the their analysis only came up with evidence for high-income
relationship between MSW generation and economic growth countries with very high turning points. A study for OCED
shows a decoupling trend (Ichinose et al. 2011; Montevecchi countries was performed by Johnstone and Labonne (2004) to
2016). According to Ichinose et al. (2011), absolute determine the drivers of MSW generation; nevertheless, the
decoupling and relative decoupling correspond to the de- authors did not use the EKC framework. Karousakis (2009)
scending and ascending sides of the inverted U-shaped curve also performed a study for OECD countries. The author found
respectively. Moreover, they state that absolute decoupling an increasing relationship between waste generation and per
comes into existence if the turning point of the curve falls into capita income.
the range of the economic indicator of the country in question. In the literature, studies also exist for other waste types
Otherwise, relative decoupling comes into question. rather than MSW. Berrens et al. (1997) tested the WKC hy-
The WKC hypothesis is likely to be supported by most of pothesis for hazardous wastes in USA. They used a cross-
the national studies in relative terms. Among city-level stud- sectional data at county level. Their results confirmed the ex-
ies, for instance, Trujillo-Lora et al. (2013) and Miyata et al. istence of WKC hypothesis. Wang and Nie (2001) conducted
(2013) found evidence for the existence of the WKC curve. also an analysis for hazardous wastes in USA. Their results are
Environ Sci Pollut Res

the same to that of Berrens et al. (1997). For medical wastes, Among other covariates, household size, population densi-
Su and Chen (2018) have recently carried out an analysis for ty, urbanization rate, education level, age, employment rate,
Taiwan. According to their results, the WKC relationship and tourism activity have been found to be influential on
holds. waste generation in several micro-level studies (Sujauddin
et al. 2008; Getahun et al. 2012; Beigl et al. 2008; Thanh
et al. 2010; Irwan et al. 2013; Ogwueleka 2013; Oribe-
Garcia et al. 2015; Suthar and Singh 2015; Han et al. 2018;
Variables of interest Pan et al. 2019).
As stated by some authors, tourism may be one of the
A broad range of factors may influence solid waste generation. important drivers of solid waste generation (Teh and
Evidence from micro-level studies analyzing the determinants Cabanban 2007; Llorens et al. 2008; Shamshiry et al. 2011).
of waste generation concluded that income, prosperity, and Moreover, Arbulu and Rey-Maquieira (2015) found that tour-
consumption patterns are crucial factors determining the level ism activities positively contributed to waste generation. So,
of waste generation (Ogwueleka 2013; OECD 2016; Saladie number of tourist arrivals is also added to our model.
2016; Kaza et al. 2018). Researchers found economic indica- As a macroeconomic indicator, unemployment rate is an-
tors like GDP (Namlis and Komilis 2019), disposable income other crucial factor needed to be kept in mind when studying
(Oribe-Garcia et al. 2015); Saladie 2016; Han et al. 2018), and the dynamics of waste generation. The unemployment rate
consumption expenditures (Han et al. 2018) as main drivers of may influence the purchasing power parity (PPP) and thus
solid waste generation. the amount of waste generation (Khajevand and Tehrani
Most of the regional- or city-level studies using the EKC 2019). Several studies considered unemployment rate poten-
framework take income or GDP per capita as the main eco- tial covariate of the waste generation. Of those, Chen (2010);
nomic indicator (Berrens et al. 1997; Wang and Nie 2001; Keser et al. (2012); Arbulu and Rey-Maquieira (2015); and
Mazzanti et al. 2006; Ichinose et al. 2011; Trujillo-Lora Oribe-Garcia et al. (2015) found a negative relationship while
et al. 2013; Su and Chen 2018; Gui et al. 2019; Madden results of Talalaj and Walery (2015) indicated a positive cor-
et al. 2019). On the other hand, for example, Ercolano et al. relation. In this sense, it is included in our analysis.
(2018) used the declared income per capita. Jaligot and The urbanization rate refers to the share of population liv-
Chenal (2018) preferred the tax point value which is the ing in urbans in total population and is often associated with
amount of money collected from local taxes divided by the higher levels of waste generation (Purcell and Magette 2009;
local coefficient. Miyata et al. (2013) used city expenditures as Thanh et al. 2010; Oribe-Garcia et al. 2015; Kaza et al. 2018;
the main economic indicator. Gui et al. 2019). We also think that it may be an important
Among cross-country studies Cole et al. (1997) used in- covariate of MSW generation so it has been added into our
come per capita, Gnonlonfin et al. (2017) took gross national analysis.
product (GNP) per capita, and Arbulu and Rey-Maquieira Mazzanti and Zoboli (2008) stated that the waste produced
(2015) considered GDP per capita as the economic drivers by older people may be less than the younger while on the
of waste generation. On the contrary, Johnstone and contrary they may have less incentive for recycling. So the
Labonne (2004); Mazzanti and Zoboli (2008); Iafolla et al. effects would be highly ambiguous. We thus decided to in-
(2010); and Nicolli et al. (2012) employed final consumption clude the dependency ratio of elderly people (population 60
expenditures in their analyses. and over) to working population as a covariate capturing the
Even though there are no strict borders between GDP, dis- effect of the age structure.
posable income, and consumption expenditures, it is stated by According to Grossman and Krueger (1991) and Antweiler
some authors (Rothman 1998; Andersen et al. 2007; et al. (2001), free trade has positive impacts on environment.
Ramachandra et al. 2018) that consumption is the pushing However, for waste generation the empirical evidence is still
factor of environmental impacts, as well as waste generation. scarce. Gnonlonfin et al. (2017) found ambiguous impacts
Several other works used consumption expenditure as the pri- varying among income groups. We are also interested in the
mary covariate and confirmed its impact by obtaining signif- effects of trade so we included as a share of GDP to our model.
icant results (Weng et al. 2007; Otoniel et al. 2008; Han et al. Finally, we believe that the effect of life expectancy on waste
2018). generation is worth studying. There is a remark stating that as
We also believe that household consumption expenditures people live longer, it would make them treat gently for their and
may provide a more reliable guideline to understand the dy- younger generations’ future. According to this belief, it is expect-
namics behind waste generation and reflect cross-country het- ed that the environmental quality would be better so as the life
erogeneity better than GDP or disposable income. Thus, we expectancy increases (Mariani et al. 2009). Nevertheless, for sol-
included consumption expenditure of households with its id waste generation, the literature still lacks adequate empirical
squared term as economic indicators for our study. evidence. Beigl et al. (2004, 2008) are two studies that found
Environ Sci Pollut Res

positive relationship between life expectancy and waste genera- Table 1 Data source
Dependent variable Source
tion. Thus, it takes place in the analysis.
MSW OECD
Independent variables
Cons World Bank
Methodology and data Unemp World Bank
Oldep World Bank
Data Lifexp World Bank
Tourism World Bank
Our data is an unbalanced panel with cross-section dimension
Urban World Bank
of 32 countries1 and time dimension of 15 years covering the
Trade World Bank
2002–2017 period. Two main sources of our data are OECD
statistics and World Bank open data. Our main variable of
interest in this study is MSW generation per capita and comes
from OECD statistics. All other variables including household MSW ¼ f ðCons; Cons2; Unemp; Oldep; Lifexp; Tourism; Urban; TradeÞ
final expenditure on goods and services, unemployment rate, ð1Þ
old dependency ratio, urbanization rate, number of tourist ar-
rivals, volume of trade, and life expectancy have been drawn where
from World Bank open data. We limited our time period with
the year 2002 since before this date, the data starts to show & MSW is the municipal waste generated per capita (in
gaps in MSW generation. We also excluded Australia, New kilograms),
Zealand, and Chile for this reason. Table 1 shows the sources & Cons is the household final consumption expenditures on
of the data briefly. The summary statistics are also presented in goods and services (in thousand dollars),
Table 2. & Cons2 is the squared term of Cons,
& Unemp is the unemployment rate (percentage share of
employment rate),
Methodology & Oldep is the ratio of people 65 years or older to the people
in age between 25 and 64,
In this study, we utilized different estimation methods. Firstly, & Lifexp is the life expectancy at birth (in years),
we used the system GMM by Arellano and Bover (1995) and & Tourism is the number of tourists arrived to the country,
Blundell and Bond (1998) which has gained popularity & Urban is the urbanization rate (percentage share of urban
among scholars in recent years due to its capability of han- population in total population),
dling the dynamic panel bias (Nickell 1981) as well as unbal- & Trade is the trade volume of the country (the percentage
anced panels and multiple endogenous variables (Roodman share of total imports plus total exports in GDP).
2009). Secondly, although not commonly used, we estimated
the model with the between estimator suggested by Stern Taking the natural logarithm of both sides and using the
(2010). According to Monte Carlo experiments performed panel data notation with EKC framework (Stern 2004; Baltagi
by Hauk and Wacziarg (2009), the between estimator was 2008), Eq. 1 turns into:
found to be the best estimator among many alternative esti-
mators in the presence of measurement error and omitted var-
iables. Lastly, we implemented a random coefficient (RC)
Table 2 Summary statistics
model by Swamy (1970) which takes into account of slope
heterogeneity and allows us the run separate regressions for Variables Mean Std. dev. Min Max
each cross-sectional unit.
MSW 494.34 130.73 254 829
Cons 19,253.89 9440.85 4777 41,636
The model Unemp 7.89 4.28 2.3 527.9
Oldep 23.58 6.05 8.7 45.1
The MSW generation per capita can be expressed as a func- Lifexp 78.74 3.06 70.9 84.1
tion of final consumption expenditures with its squared term Tourism 1.62e+07 1.99e+ 285,000 8.69e+
as well as several other socio-economic variables (Qiao et al. 07 07
2019): Urban 75.52 11.50 50.9 98
Trade 100.13 58.74 20.7 408.4
1
Australia, Chile, and New Zealand are excluded from the analysis.
Environ Sci Pollut Res

lnðMSW it Þ ¼ αi þ β1 lnðConsit Þþ Table 3 VIF values of


the independent Variables VIF
β2 lnðConsit Þ2 þ β3 lnðUnempit Þþ
variables
β 4 lnðOldepit Þ þ β5 lnðLifexpit Þþ ð2Þ Cons 2.89
β 6 lnðTourismit Þ þ β7 lnðUrbanit Þþ Unemp 1.40
β8 lnðTradeit Þ þ uit Oldep 1.38
Lifexp 2.42
with the intercept term αi and the random error term uit which
Tourism 1.64
can be decomposed as uit = μi + δt + vit, where μi refers to un-
Urban 1.48
observable individual heterogeneity, δt is the unobservable
Trade 1.54
time effects, and vit is the remainder error term.
The shape of the WKC curve depends on the estimated Average VIF 1.82
coefficients of the parameters (Lorente and Alvarez-Herranz
2016; Allard et al. 2018):
cross-section dependence. It can be due to spillover effects of
(i) In case β1 = β2 = 0, no relationship between waste gener-
unobserved common or spatial factors. Ignoring cross-section
ation and consumption expenditure,
dependence may cause serious problems (Pesaran 2015).
(ii) In case β1 > 0 and β2 = 0, waste generation increases
In this study, we carried out a test proposed for cross-section
monotonically with the consumption expenditure,
dependence by Pesaran (2004) to check the cross-section depen-
(iii) In case β1 < 0 and β2 = 0, waste generation decreases
dency. The results show that the null hypothesis of cross-section
monotonically with the consumption expenditure,
independence is not rejected which means that it is not an issue
(iv) In case β1 > 0 and β2 < 0, there is an inverted U-shaped
for our data. Table 4 shows the results of the test.
relationship between waste generation and consumption
expenditure; the WKC hypothesis holds,
(v) In case β1 < 0 and β2 > 0, there is a U-shaped relationship
between waste generation and consumption expenditure. Results

The key outcome of our analysis is that the signs of the coef-
ficients of the Cons and Cons2 variables support the WKC
Multicollinearity
hypothesis. Since our econometric model is in double log
form, all coefficients can also be interpreted as elasticities.
The unemployment has been found to be insignificant in all
models. The Oldep variable is significant for only in one step
Before discussing the results, we checked the degree of
system GMM. The Lifexp shows significance with a negative
correlation between the covariates in order to ensure the
sign at 1% level for one-step GMM and 5% level for two-step
stability of our results. If there is a high correlation between
GMM. This variable can be thought as an important determi-
two variables, it may lead to higher standard errors and the
nant of waste generation. The impact of the tourism is insig-
partial effects of the correlated variables are hardly distin-
nificant for both models. Surprisingly, the urbanization level
guishable (Verbeek 2008). As a diagnostic statistic, the
which is captured by Urban variable shows also no signifi-
variance inflation factor (VIF) for each variable in the
cance in all cases. Finally, the volume of trade variable Trade
model is calculated as follows (Greene 2017):
also shows overall insignificance.
For system GMM estimations, we also report the results of
the Sargan and Hansen tests as well as the Arellano-Bond

VIF ¼ 1= 1−Rk 2 ð3Þ AR(1) and AR(2) tests. The results suggest that system
GMM models pass these diagnostic tests.
Values for VIF greater than 20 should be considered further The between estimator can be considered the long-run
(Greene 2017). Since all VIF values presented in Table 3 are estimator since it does not make any assumptions about
under this threshold, multicollinearity is not a major concern
for our analysis.
Table 4 Pesaran cross-
section dependence test H0: Cross-section independence
Cross-section dependence
Test statistic 1.473

Another issue with panel data is the possibility of the correla- Prob. 0.1406
tion of error terms across the cross-section units, namely the
Environ Sci Pollut Res

the nature of the time dimension of the data (Stern, 2009). the average mean GDP based on these mean GDPs which
The results from this estimator support also the WKC is more or less 38.000 $ per capita. Then, we created a
hypothesis. The coefficients of Cons and Cons2 variables dummy variable taking the value of 1 if the country is
are positive and negative respectively. Besides, all vari- above this threshold, otherwise zero. This allows us to
ables except Urban are found to be significant. Lifexp and examine the countries sharing similar development levels
Oldep variables are likely to contribute positively to waste and to mitigate the effects of excess variation.
generation in the long run. Nevertheless, waste generation Nevertheless, the general results of the RC model for all
seems to be negatively affected from an increase in un- variables show no significance, except the Unemp, and
employment, volume of trade, and number of tourist ar- thus are statistically meaningless to interpret in terms of
rivals. Table 5 shows the results. a WKC framework. This result shows that the aggregation
Significance levels are 1%, 5%, and 10% of slopes with heterogenous nature may not bring any
logical sense (List and Gallet 1999; Chakravarty and
RC approach and within-country results Mandal 2019).
When we look at the within-country results, 8 out of 16
Following Halkos (2003); Mazzanti and Musolesi (2013); high-income countries and only 2 out of 16 low-income coun-
Sinha and Bhatt (2017); and Xu (2018), we also estimated tries support the WKC hypothesis and reach decoupling in
the model with random coefficient (RC) approach by absolute terms. Out of 16 low-income group, 5 countries show
Swamy (1970). We examined the countries in two groups a U-shaped relationship. The within-country results are report-
as low income and high income. In this sense, we calcu- ed in Tables 6, 7, and 8. A brief discussion of the country-
lated the mean GDP of each country and then calculated specific results is presented in “Discussion.”

Table 5 The results of general


estimations Independent variable

MSW

İndependent variables One-step system GMM Two-step system GMM Between estimator

Cons 1.128*** 1.264** 9.466**


(0.348) (0.510) (1.328)
Cons2 − 0.049*** − 0.055** − 0.455**
(0.015) (0.023) (0.067)
Unemp − 0.001 − 0.005 0.047**
(0.015) (0.018) (0.023)
Oldep − 0.062** − 0.047 − 0.319**
(0.030) (0.041) (0.028)
Lifexp − 0.960*** − 1.118** − 5.67***
(0.360) (0.541) (0.635)
Tourism 0.009 0.013 0.029***
(0.008) (0.013) (0.005)
Urban − 0.020 − 0.018 0.044
(0.062) (0.076) (0.034)
Trade 0.016 0.020 0.036***
(0.019) (0.031) (0.012)
AR(1) 0.030** 0.050**
AR(2) 0.648 0.641
Hansen test 0.078 0.078
Sargan test 0.000*** 0.000***
Turning point 96.683 $ 86.379 $ 32.555 $

***, **, and * denote statistical significance at the 1%, 5%, and 10% levels, respectively, and the values in
parentheses show robust standard errors for system GMM estimators and bootstrapped std. errors with 100
replications for between estimator. (The turning point can be calculated as follows: exp(− β1/2β2), where β1
and β2 are the coefficients of Cons and Cons2 respectively.)
Environ Sci Pollut Res

Table 6 Country-based estimates—high-income group

High income Cons Cons2 Unemp Oldep Lifexp Tourism Urban Trade

Austria 138.451*** − 6.825*** − 0.094** − 1.135*** 2.651 0.235** − 0.552 − 0.129


(43.138) (2.129) (0.041) (0.331) (2.194) (0.098) (0.824) (0.102)
Belgium − 37.960 1.817 − 0.055 − 0.637** 1.614 − 0.266* 1.803 0.259**
(38.704) (1.915) (0.057) (0.256) (1.739) (0.137) (2.866) (0.114)
Denmark 80.710* − 3.938* − 0.018 − 0.150 − 2.886 − 0.005 9.338*** 0.344***
(46.047) (2.262) (0.042) (0.221) (2.225) (0.040) (3.496) (0.072)
France − 25.056 1.268 − 0.045 0.037 − 0.347 0.095 − 1.357 0.076
(33.301) (1.659) (0.043) (0.150) (2.363) (0.132) (2.274) (0.120)
Finland 17.051 − 0.816 − 0.005 − 0.224 0.798 0.057 − 1.894 0.082
(29.499) (1.470) (0.043) (0.161) (1.028) (0.107) (1.736) (0.079)
Germany 67.782* − 3.409* − 0.198*** − 0.972*** 1.428 0.134 2.955* − 0.135*
(41.043) (2.037) (0.022) (0.245) (1.576) (0.100) (1.520) (0.066)
Iceland 150.105*** − 7.422*** − 0.230*** − 0.631* − 4.049* 0.143*** 12.886*** 0.044
(49.809) (2.468) (0.060) (0.326) (2.215) (0.046) (3.913) (0.086)
Ireland 45.404 − 2.221 − 0.050 − 0.625*** − 0.698 0.240** − 1.775 − 0.139**
(36.121) (1.798) (0.033) (0.198) (1.997) (0.102) (1.092) (0.064)
Japan 59.723* − 2.897* − 0.118*** − 0.478*** − 0.698 − 0.003 − 0.939** − 0.035
(34.666) (1.705) (0.042) (0.123) (1.655) (0.025) (0.367) (0.052)
Luxembourg 12.653 − 0.589 0.042 0.151 − 0.946 − 0.434*** − 1.939 0.295***
(27.655) (1.357) (0.035) (0.255) (1.899) (0.095) (1.951) (0.073)
Netherlands 102.266*** − 5.053*** − 0.011 − 0.492*** 2.916** 0.028 − 0.716 − 0.007
(36.376) (1.807) (0.014) (0.133) (1.171) (0.056) (0.451) (0.048)
Norway 461.79*** − 22.133*** − 0.349*** 3.964*** − 64.956*** 1.044*** 39.773*** − 0.756***
(81.920) (3.933) (0.064) (0.707) (10.638) (0.212) (6.539) (0.184)
Sweden 139.180*** − 6.942*** − 0.124*** − 0.647** − 2.020 0.106 8.239*** 0.151
(46.236) (2.307) (0.046) (0.254) (2.016) (0.091) (3.110) (0.106)
Switzerland − 16.941 0.873 − 0.069* − 0.777*** 1.718 0.186* − 5.672** − 0.132
(13.305) (0.656) (0.036) (0.287) (1.838) (0.105) (2.521) (0.083)
UK 20.489 − 0.990 − 0.044 − 0.054 0.167 0.146 − 6.991*** − 0.143*
(40.635) (2.005) (0.044) (0.227) (1.797) (0.095) (1.912) (0.073)
USA 29.235 − 1.402 − 0.061* − 0.008 − 1.505 − 0.005 − 0.825 0.002
(28.956) (1.397) (0.034) (0.212) (2.030) (0.101) (2.581) (0.064)

***, **, and * denote statistical significance at the 1%, 5%, and 10% levels, respectively, and the values in parentheses show standard errors

Robustness check Another concern is that since our main regressor is per
capita final consumption expenditures, it may be subject to
In order to check the consistency of our results, we follow- endogeneity and simultaneity problems. To examine this, we
ed a couple of steps. Firstly, we wanted to check whether firstly formulated a model for MSW and Cons variables in
the WKC hypothesis still holds if we remove the statisti- which they are jointly determined to take into account of pos-
cally insignificant variables from the models. For this pur- sible simultaneity bias. Clinging to the variables used in the
pose, we removed Unemp, Tourism, Urban, and Trade var- analysis, we developed the following equations:
iables from the one-step system estimator and Oldep vari-
able from the two-step system GMM estimator. The results
are still consistent and support the WKC hypothesis MSW ¼ β1Cons þ β2Urban þ β3Tourism þ β4Trade
(Table 9). þ β5Lifexp þ β6Oldep ð4Þ
We then excluded the three outlier countries, namely the
Luxembourg, Norway, and Switzerland, with relatively high Cons ¼ β1MSW þ β2Oldep þ β3Unemp þ β4Lifexp ð5Þ
income and final consumption expenditure (Table 10). The
We estimated the above equations with three-stage least
results of the system GMM estimators do not change. We also
squares (3SLS) (Greene 2017) (Table 11). Then, we included
repeated the same procedure for the between estimator. The
also the Cons2 to Eq. (4) to check whether the WKC hypoth-
results also do not change.
esis still holds. The coefficients of consumption variables are
Environ Sci Pollut Res

Table 7 Country-based estimates—low-income group

Country Cons Cons2 Unemp Oldep Lifexp Tourism Urban Trade

Czech Rep. − 52.737** 2.943** 0.067* 0.020 1.113 − 0.125 − 1.762 − 0.061
(22.238) (1.217) (0.037) (0.188) (1.919) (0.095) (4.027) (0.079)
Estonia 6.194 − 0.306 0.002 − 0.873** − 1.619 0.135 24.344*** − 0.258**
(12.637) (0.707) (0.045) (0.429) (1.227) (0.190) (5.505) (0.099)
Greece 57.830*** − 2.967*** − 0.111** 1.552*** 0.844 − 0.378*** 1.950 0.011
(19.747) (1.013) (0.048) (0.588) (2.131) (0.089) (1.918) (0.081)
Hungary − 25.585 1.466 − 0.015 − 0.213 − 1.521 − 0.123 − 1.082 0.080
(25.410) (1.431) (0.038) (0.394) (2.046) (0.143) (1.307) (0.084)
Israel − 31.222* 1.618* 0.005 − 0.346 0.822 − 0.004 − 0.623 0.026
(17.231) (0.889) (0.041) (0.276) (1.720) (0.035) (2.112) (0.054)
Italy − 7.712 0.413 − 0.107*** − 0.006 2.337** − 0.101 − 1.194 − 0.083
(11.370) (0.578) (0.041) (0.405) (1.055) (0.073) (1.869) (0.057)
Korea − 43.303 2.353 0.012 − 0.146 2.285 − 0.152 − 1.449 − 0.034
(27.881) (1.522) (0.049) (0.365) (2.187) (0.106) (2.324) (0.072)
Latvia − 27.926*** 1.620*** 0.128** − 0.898** 2.067 0.112 28.972*** − 0.454***
(4.454) (0.258) (0.052) (0.428) (1.261) (0.124) (7.167) (0.103)
Lithuania − 15.513*** 0.867*** − 0.071* 0.642 0.776 − 0.104** − 12.814** 0.296***
(3.390) (0.189) (0.043) (0.472) (1.143) (0.045) (5.423) (0.055)
Mexico 11.441 − 0.649 0.007 0.126 − 0.730 − 0.004 1.926 0.093
(16.269) (0.931) (0.033) (0.370) (1.630) (0.088) (1.220) (0.064)
Poland 14.719 − 0.729 − 0.054 − 0.562 − 5.208** 0.546*** 14.731*** − 0.262**
(17.550) (1.000) (0.044) (0.376) (2.048) (0.186) (4.461) (0.125)
Portugal 23.077 − 1.145 − 0.102** 0.938 − 1.576 − 0.430*** 2.114* 0.006
(22.299) (1.164) (0.043) (0.655) (1.616) (0.108) (1.128) (0.080)
Slovak Rep. − 48.612*** 2.708*** − 0.104** 0.147 − 2.176 − 0.094 − 3.249* 0.105
(12.898) (0.713) (0.046) (0.271) (1.619) (0.072) (1.740) (0.084)
Slovenia 250.090*** − 13.250*** − 0.401*** 6.030*** − 3.530* − 0.931*** − 11.987*** 0.071
(59.375) (3.162) (0.076) (1.389) (2.086) (0.250) (2.870) (0.082)
Spain 2.166 − 0.082 − 0.051 0.939** − 1.501 − 0.062 − 7.763*** − 0.157**
(11.020) (0.565) (0.037) (0.406) (1.695) (0.122) (2.719) (0.079)
Turkey − 5.397 0.311 − 0.016 0.621* 0.407 − 0.069 − 2.284 − 0.080
(7.333) (0.417) (0.039) (0.352) (1.666) (0.072) (1.659) (0.084)

***, **, and * denote statistical significance at the 1%, 5%, and 10% levels, respectively, and the values in parentheses show standard errors

in line with the previous estimates signalling the evidence in Discussion


favor of the WKC hypothesis.
We also consider the endogeneity problem of the Cons The results of one-step system GMM, two-step system
variable in terms of omitted variables bias. Nevertheless, GMM, and the between estimator confirm the validity of
there is no compromised set of instruments for consump- a WKC curve for OECD countries. The general results of
tion expenditure in the literature. We thus instrumented the RC approach do not support the WKC relationship
the Cons variable with its 2-year lagged values (Reed since the coefficients of consumption variables are statis-
2015). The results of the two-stage least squares (2SLS) tically insignificant. The predicted turning points of one-
present evidence in favor of WKC; nevertheless, the co- step and two-step system GMM are 96.683 $ and 86.379
efficients of consumption variables do not show any sta- $ respectively. For the between estimator, the turning
tistical significance. We also instrumented the Cons2 var- point is 32.555 $ which is much lower. This exposes an
iable also with its lagged values and repeated the 2SLS important conclusion. The achievement of decoupling in
procedure. The results are the same. However, we every absolute terms is more costly in short run than in long
time failed to reject the null hypothesis of endogeneity run. One explanation of this might be that the absolute
test which states that the variables in question are exoge- decoupling requires structural reforms like promoting the
nous. Therefore, we think that our results do not suffer circular economy, improving the recycling rates, and in-
from omitted variables problem. creasing the awareness of the whole society which need a
Environ Sci Pollut Res

Table 8 Country-based turning


points High income Turning point ($) Decoupling Low income Turning point ($) Decoupling

Austria 25,410 Absolute Czech Rep. No (U-shaped)


Belgium Estonia
Denmark 28,214 Absolute Greece 17,077 Absolute
France Hungary
Finland Israel No (U-shaped)
Germany 20,777 Absolute Italy
Iceland 24,640 Absolute Korea
Ireland Latvia No (U-shaped)
Japan 29,963 Absolute Lithuania No (U-shaped)
Luxembourg Mexico
Netherlands 24,818 Absolute Poland
Norway 33,933 Absolute Portugal
Sweden 22,572 Absolute Slovak Rep. No (U-shaped)
Switzerland Slovenia 12,548 Absolute
UK Spain
USA Turkey

The turning point can be calculated as follows: exp(− β1/2β2), where β1 and β2 are the coefficients of Cons and
Cons2 respectively

longer time period to put into practice and thus more dynamics of waste generation so we adopted the system
costly in short run. GMM estimators as our main estimators.
When considering that the mean value of household final Since cross-country studies testing the WKC hypothesis
consumption expenditures is 19,253 $ with a min. value of are scarce, we could only compare our results with a couple
4777 $ and max. value of 41,636 $, it can be concluded that of studies. By using gross national income (GNI), Gnonlonfin
the OECD area shows the signs of relative decoupling since its et al. (2017) conclude that turning points for Mediterranean
consumption patterns are in the ascending side of the WKC countries vary from 56,000 $ to 133,000 $ per capita.
curve, but far from absolute decoupling. However, the results Although they did not present this in their paper, the former
obtained from the between estimator state that there is abso- value comes from the fixed-effects estimation. The results
lute decoupling since the value of the turning point falls be- from the Hausman-Taylor estimator range from 109.000 $ to
tween the min. and max. value of the consumption expendi- 133.000 $. These are quite higher than our estimates. The
tures. Nevertheless, we are more interested in short-run authors also classified the countries as high and low income,

Table 9 Robustness check—


exclusion of insignificant Independent variables One-step system GMM Two-step system GMM Between estimator
variables
Cons 1.686*** 1.895** 10.398***
(0.605) (0.805) (1.008)
Cons2 − 0.073*** − 0.083** − 0.503***
(0.027) (0.037) (0.051)
Lifexp − 1.345** − 1.625** − 5.909***
(0.551) (0.778) (0.506)
Oldep − 0.092* − 0.314***
(0.049) (0.024)
AR(1) 0.026** 0.068*
AR(2) 0.598 0.630
Hansen test 0.122 0.150
Sargan test 0.000*** 0.009***

***, **, and * denote statistical significance at the 1%, 5%, and 10% levels, respectively, and the values in
parentheses show robust standard errors for system GMM estimators and bootstrapped std. errors with 100
replications for between estimator
Environ Sci Pollut Res

Table 10 Robustness check—


exclusion of outlier countries İndependent variables One-step system GMM Two-step system GMM Between estimator

Cons 1.056*** 1.234** 10.660***


(0.382) (0.562) (2.145)
Cons2 − 0.046*** − 0.054** − 0.517***
(0.016) (0.025) (0.110)
Unemp − 0.008 − 0.011 0.018
(0.012) (0.014) (0.027)
Oldep − 0.055* − 0.046 − 0.306***
(0.030) (0.037) (0.027)
Lifexp − 0.905** − 1.090** − 6.146***
(0.365) (0.544) (0.903)
Tourism 0.006 0.010 0.028***
(0.007) (0.012) (0.005)
Urban − 0.031 − 0.021 − 0.006
(0.051) (0.056) (0.034)
Trade 0.011 0.016 − 0.009
(0.020) (0.030) (0.021)
AR(1) 0.066* 0.085*
AR(2) 0.261 0.257
Hansen test 0.140 0.140
Sargan test 0.000*** 0.000***

***, **, and * denote statistical significance at the 1%, 5%, and 10% levels, respectively, and the values in
parentheses show robust standard errors for system GMM estimators and bootstrapped std. errors with 100
replications for between estimator

and their results are valid only for high-income countries. high-income and low-income countries. This conflicts with
They also found macroeconomic variables such as trade open- our findings since urbanization rate shows overall no signifi-
ness, urbanization rate, share of working women, and climate cance in our analysis and trade openness has been found to
variable precipitation rate as significant determinants of both contribute positively to waste generation in long run. We did
not include a variable referring to the share of working women
to our model; nevertheless, we added the unemployment rate
Table 11 Results of 3SLS estimation which seems to contribute positively to the waste generation
in the long run.
İndependent variables Without Cons2 With Cons2
The life expectancy at birth represented by the Lifexp var-
Cons 0.285*** 36.244** iable can be thought to be an important determinant of the
(0.053) (17.596) waste generation. However, in contrast to the previous studies
Cons2 − 1.829** like Beigl et al. (2004, 2008), the signs of the coefficients of
(0.900) this variable are negative for all models. This exciting conclu-
Oldep − 0.087** − 0.532*** sion is in the same line with the view stating that as people live
(0.035) (0.202)
longer, they become more conscious and treat environment
Lifexp − 0.430 − 13.604**
(0.477) (5.967)
more gently, resulting a decrease in waste generation.
Similarly, the dependency ratio of elderly people is found to
Tourism 0.026** 0.040
(0.010) (0.024) be highly significant in all models with negative signs.
Urban 0.198*** 0.378 Nonetheless, the results vary in the literature. Mazzanti and
(0.054) (0.239) Zoboli (2008) stated that older people may generate less waste
Trade 0.043* 0.041 than younger. By including the share of elderly people, their
(0.026) (0.059) results pointed out negative coefficients supporting this view.
Ichinose et al. (2011) also came up with the same results.
***, **, and * denote statistical significance at the 1%, 5%, and 10%
levels, respectively, and the values in parentheses show a bootstrapped However, Ercolano et al. (2018) found a positive correlation
std. errors with 100 replications between the share of older people and waste generation. Our
Environ Sci Pollut Res

findings for this variable coincide with Mazzanti and Zoboli relationship. In this sense, the literature still needs to be
(2008) endorsing the notion that elderly people generate less enriched.
amount of waste. When we look at the country-specific results, half of the
The results of within-country estimations based on RC ap- high-income countries seem to have reached a decoupling
proach are presented in Tables 6 and 7. It can be seen from the trend in absolute terms. On the other hand, only two low-
table that Austria, Denmark, Germany, Greece, Iceland, income countries succeeded in achieving an absolute
Japan, Netherlands, Norway, Slovenia, and Sweden support decoupling. Moreover, five countries show a U-shaped rela-
the WKC hypothesis while Czech Rep., Israel, Latvia, tionship which means waste generation increases with the
Lithuania, and Slovak Rep. show a U-shaped relationship be- consumption expenditures. This brings an exciting result.
tween consumption expenditures and waste generation. Our Even though an increase in the per capita consumption expen-
results for Austria, Germany, Greece, and Netherlands overlap diture is associated with more per capita waste generation,
with Iafolla et al. (2010) with slightly higher turning points. high levels of consumption, as well as income, seem to be
However, the authors found evidence of decoupling for necessary to reach the goal of absolute decoupling.
Portugal, Spain, UK, Italy, Finland, and France while we did The inference above also states that countries which have
not. Similarly Nicolli et al. (2012) found evidence also for not reached any decoupling trend yet seem to inevitably bur-
Portugal, Spain, and UK. In contrast, our results do not show den the negative externalities of waste generation after joining
any traces of decoupling for these countries. Moreover, into the trend. The main point that policy makers should take
Nicolli et al. did not find any signs of decoupling for into account that in the short or medium run, policies targeting
Denmark and Sweden. According to our results, Sweden to reduce the waste generation via manipulating the consump-
and Norway achieved an absolute decoupling. It is not surpris- tion expenditures per capita might be ineffective. That is be-
ing that Sweden achieves an absolute decoupling according to cause the estimated turning point is relatively high when com-
our results because the country has already adopted a non- pared with consumption expenditure patterns. In order to mit-
increasing future waste policy as a part of its environmental igate the negative effects and reduce the amount of waste
objectives (Sjostrom and Ostblom 2010). Similarly, Norway generated, policy makers should seek alternative policies such
has announced a Waste Prevention Programme in order to as promoting the circular economy, recycling, and waste
achieve a relative decoupling of economic growth from waste source separation which is crucial to achieve efficient
generation (Papineschi et al. 2019). For Japan and Slovenia recycling levels.
our results are also compatible with OECD (2019). Volume of trade and tourism show also signals of positive
Nevertheless, the waste generation follows a U-shaped rela- contribution of waste generation in the long run. For countries
tionship for Czech Rep., Israel, Latvia, Lithuania, and attracting many visitors such as Italy, Spain, and Turkey, waste
Slovakia. management policies should be extended in the manner that it
embraces sustainable or eco-friendly concepts rather than all
included consumption-based tourism. Although the pure ef-
Policy implications and conclusion fects of trade are out of scope of this paper, we contrasted the
view that trade makes environment better off. It should be
In this study, we aimed to empirically check the validity of a considered for future research.
Kuznets curve for OECD countries in terms of MSW genera-
tion. For this purpose, we utilized the one- and two-step sys-
tem GMM estimators as well as the between estimator sug-
gested by Stern (2010). Then, we relaxed the slope homoge-
References
neity assumption and performed a RC procedure by Swamy
(1970) which allows us to run separate regressions for each Agrawala, S (2017) Waste management and transition to a circular econ-
country. General estimations confirm the validity of a WKC omy: OECD experience and implications for emerging Asia
curve. However, the achievement of absolute decoupling is far Allard A, Takman J, Uddin GS, Ahmed A (2018) The n-shaped environ-
away according to our results. mental Kuznets curve: an empirical evaluation using a panel
quantile regression approach. Environmental Science and Pollution
Decoupling waste generation from economic growth has Research 25(6):5848–5861
gained importance. Many countries have already announced Andersen FM, Larsen H, Skovgaard M, Moll S, Isoard S (2007) A
national waste prevention policies or adopted waste decreas- European model for waste and material flows. Resources,
ing policies as a part of their long-term environmental hori- Conservation and Recycling 49(4):421–435
zons. In the literature, numerous studies testing the EKC hy- Antweiler W, Copeland BR, Taylor MS (2001) Is free trade good for the
environment? American Economic Review 91:877–908
pothesis can be found. However, these studies focused mainly Arbulu LJ, Rey-Maquieira J (2015) Tourism and solid waste generation
on a few indicators. In addition, few studies analyzed the in Europe: a panel data assessment of the environmental Kuznets
dynamics behind waste generation in terms of decoupling curve. Waste Management 46:628–636
Environ Sci Pollut Res

Arellano M, Bover O (1995) Another look at the instrumental variable Hauk WR, Wacziarg R (2009) A Monte Carlo study of growth regres-
estimation of error components models. Journal of Econometrics sions. Journal of Economic Growth 14(2):103–147
68(1):29–51 Iafolla, V., Mazzanti, M., Nicolli, F. (2010) Are you sure you want to
Baltagi B (2008) Econometric analysis of panel data. John Wiley & Sons waste policy chances? Waste generation, landfill diversion and en-
Beede DN, Bloom DE (1995) The economics of municipal solid waste. vironmental policy effectiveness in the eu15. Tech. rep., Nota di
The World Bank Research Observer 10(2):113–150 Lavoro
Beigl, P., Wassermann, G., Schneider, F., Salhofer, S. (2004) Forecasting Ichinose, D., Yamamoto, M., Yoshida, Y., et al. (2011) Reexamining the
municipal solid waste generation in major European cities waste-income relationship. GRIPS National Graduate Institute for
Beigl P, Lebersorger S, Salhofer S (2008) Modelling municipal solid Policy Studies Discussion Paper 10
waste generation: a review. Waste Man-Agement 28(1):200–214 Irwan D, Basri NEA, Watanabe K, Abushammala M (2013) Infuence of
Berrens RP, Bohara AK, Gawande K, Wang P (1997) Testing the income level and age on per capita household solid waste generation
inverted-u hypothesis for us hazardous waste: an application of the in Putrajaya, Malaysia. Jurnal Teknologi:65–62
generalized gamma model. Economics Letters 55(3):435–440 Jaligot R, Chenal J (2018) Decoupling municipal solid waste generation
Blundell R, Bond S (1998) Initial conditions and moment restrictions in and economic growth in the canton of Vaud, Switzerland.
dynamic panel data models. Journal of Econometrics 87(1):115–143 Resources, Conservation and Recycling 130:260–266
Brock, W. A., Taylor, M.S. (2005) Economic growth and the environ- Johnstone N, Labonne J (2004) Generation of household solid waste in
ment: a review of theory and empirics. In: Handbook of economic OECD countries: an empirical analysis using macroeconomic data.
growth, vol 1, Elsevier, pp 1749-1821 Land Economics 80(4):529–538
Chakravarty D, Mandal SK (2019) Environmental Kuznets curve for Kaartinen T, Sormunen K, Rintala J (2013) Case study on sampling,
local and global pollutants: application of GMM and random coef- processing and characterization of landfilled municipal solid waste
ficient panel data models. Journal of Social and Economic in the view of landfill mining. Journal of Cleaner Production 55:56–
Development 21(2):212–233 66
Chen C (2010) Spatial inequality in municipal solid waste disposal across Karousakis, K. (2009) 4 the drivers of msw generation, disposal and
regions in developing countries. International Journal of recycling. Waste Environ Policy p 91
Environmental Science & Technology 7(3):447–456 Kaza, S., Yao, L., Bhada-Tata, P., Van Woerden, F. (2018) What a waste
Cole M, Rayner A, Bates J (1997) The environmental Kuznets curve: an 2.0: a global snapshot of solid waste management to 2050. The
empirical analysis. Environment and Development Economics pp 2: World Bank
401–416 Keser S, Duzgun S, Aksoy A (2012) Application of spatial and non-
Copeland BR, Taylor MS (2004) Trade, growth, and the environment. spatial data analysis in determination of the factors that impact mu-
Journal of Economic literature 42(1):7–71 nicipal solid waste generation rates in turkey. Waste Management
Ercolano S, Gaeta GLL, Ghinoi S, Silvestri F (2018) Kuznets curve in 32(3):359–371
municipal solid waste production: an empirical analysis based on Khajevand N, Tehrani R (2019) Impact of population change and unem-
municipal-level panel data from the Lombardy region (Italy). ployment rate on Philadelphia’s waste disposal. Waste Management
Ecological Indicators 93:397–403 100:278–286
for Economic Co-operation O, Development (2009) Resource productiv- Khan D, Kumar A, Samadder S (2016) Impact of socioeconomic status
ity in the g8 and the OECD: a report in the framework of the Kobe 3r on municipal solid waste generation rate. Waste Management 49:
Action Plan 15–25
for Economic Co-operation O, Sta DO (2016) OECD Factbook 2015- Kuznets S (1955) Economic growth and income inequality. The
2016: economic, environmental and social statistics. OECD American Economic Review 45(1):1–28
Foundation EM (2015) Delivering the circular economy: a toolkit for List JA, Gallet CA (1999) The environmental Kuznets curve: does one
policymakers. Ellen MacArthur Foun-dation size fit all? Ecological Economics 31(3):409–423
Getahun T, Mengistie E, Haddis A, Wasie F, Ale-mayehu E, Dadi D, Van Llorens MCE et al (2008) Characterization of municipal solid waste from
Gerven T, Van der Bruggen B (2012) Municipal solid waste gener- the main landfills of Havana city. Waste Management 28(10):2013–
ation in growing urban areas in Africa: current practices and relation 2021
to socioeconomic factors in Jimma, Ethiopia. Environmental Lorente DB, Alvarez-Herranz A (2016) Economic growth and energy
Monitoring and Assessment 184(10):6337–6345 regulation in the environmental Kuznets curve. Environmental
Gnonlonfin A, Kocoglu Y, Peridy N et al (2017) Municipal solid waste Science and Pollution Research 23(16):16478–16494
and development: the environmental Kuznets curve evidence for Madden B, Florin N, Mohr S, Giurco D (2019) Using the waste Kuznet’s
Mediterranean countries. Region et Developpement 45:113–130 curve to explore regional variation in the decoupling of waste gen-
Greene, W. H. (2017) Econometric analysis (eight edition) Grossman eration and socioeconomic indicators. Resources, Conservation and
GM, Krueger AB (1991) Environmental impacts of a North Recycling 149:674–686
American Free Trade Agreement. Tech. Rep., National Bureau of Mariani, F., Perez-Barahona, A., Raffin, N. (2009) Life expectancy and
Economic Research the environment
Grossman, G. M. and Krueger, A. B. (1991) Environmental impacts of a Mazzanti M, Musolesi A (2013) The heterogeneity of carbon Kuznets
North American free trade agreement (No. w3914). National Bureau curves for advanced countries: comparing homogeneous heteroge-
of economic research. neous and shrinkage/ Bayesian estimators. Applied Economics
Gui S, Zhao L, Zhang Z (2019) Does municipal solid waste generation in 45(27):3827–3842
China support the environmental Kuznets curve? New evidence Mazzanti M, Zoboli R (2008) Waste generation, waste disposal and pol-
from spatial linkage analysis. Waste Management 84:310–319 icy effectiveness: evidence on decoupling from the European union.
Halkos GE (2003) Environmental Kuznets curve for sulfur: evidence Resources, Conservation and Recycling 52(10):1221–1234
using GMM estimation and random coefficient panel data models. Mazzanti M, Montini, A., Zoboli, R. (2006) Municipal waste production,
Environment and Development Economics 8(4):581–601 economic drivers and new waste policies: EKC evidence from
Han Z, Liu Y, Zhong M, Shi G, Li Q, Zeng D, Zhang Y, Fei Y, Xie Y Italian regional and provincial panel data
(2018) Influencing factors of domestic waste characteristics in rural Meadows DH, Meadows DL, Randers J, Behrens WW (1972) The limits
areas of developing countries. Waste Management 72:45–54 to growth. New York 102:27
Environ Sci Pollut Res

Miyata, Y., Shibusawa, H., Hossain, N. (2013) An economic analysis of tourism regions: Langkawi Island, Malaysia. J Environ Public
municipal solid waste management of Toyohashi City, Japan: evi- Health 2011.
dences from environmental Kuznets curve Shekdar AV (2009) Sustainable solid waste management: an integrated
Montevecchi F (2016) Policy mixes to achieve absolute decoupling: a approach for Asian countries. Waste management 29(4):1438–1448
case study of municipal waste management. Sustainability 8(5):442 Sinha A, Bhatt M (2017) Environmental Kuznets curve for CO2 and NOx
Namlis KG, Komilis D (2019) Influence of four socio-economic indices emissions: a case study of India. European Journal of Sustainable
and the impact of economic crisis on solid waste generation in eu- Development 6(1):267–267
rope. Waste Management 89:190–200 Sjostrom M, Ostblom G (2010) Decoupling waste generation from eco-
Nickell S (1981) Biases in dynamic models with fixed effects. Journal of nomic growth - a CGE analysis of the Swedish case. Ecological
the Econometric Society, Econometrica, pp 1417–1426 Economics 69(7):1545–1552
Nicolli F, Mazzanti M, Iafolla V (2012) Waste dynamics, country hetero- Stern DI (2004) The rise and fall of the environmental Kuznets curve.
geneity and European environmental policy e ectiveness. Journal of World development 32(8):1419–1439
Environmental Policy & Planning 14(4):371–393 Stern DI (2010) Between estimates of the environmental Kuznets curve
OECD (2015) Material resources, productivity and the environment Su ECY, Chen YT (2018) Policy or income to a ect the generation of
OECD (2019) Waste management and the circular economy in selected medical wastes: an application of environmental Kuznets curve by
OECD countries using Taiwan as an example. Journal of Cleaner Production 188:
Ogwueleka TC (2013) Survey of household waste composition and quan- 489–496
tities in Abuja, Nigeria. Resources, Conservation and Recycling 77: Sujauddin M, Huda S, Hoque AR (2008) Household solid waste charac-
52–60 teristics and management in Chit-tagong, Bangladesh. Waste
Oribe-Garcia I, Kamara-Esteban O, Martin C, Macarulla-Arenaza AM, Management 28(9):1688–1695
Alonso-Vicario A (2015) Identification of influencing municipal Suthar S, Singh P (2015) Household solid waste generation and compo-
characteristics regarding household waste generation and their fore- sition in different family size and socio-economic groups: a case
casting ability in Biscay. Waste Management 39:26–34 study. Sustainable Cities and Society 14:56–63
Otoniel BD, Liliana MB, Francelia PG (2008) Consumption patterns and Swamy PA (1970) Efficient inference in a random coefficient regression
household hazardous solid waste generation in an urban settlement model. Journal of the Econometric Society, Econometrica, pp 311–
in Mexico. Waste Management 28:S2–S6 323
Pan A, Yu L, Yang Q (2019) Characteristics and forecasting of municipal Talalaj IA, Walery M (2015) The effect of gender and age structure on
solid waste generation in China. Sustainability 11(5):1433 municipal waste generation in Poland. Waste Management 40:3–8
Papineschi, J., Hogg, D., Chowdhury, T., Durrant, C., Thomson, A. Teh L, Cabanban AS (2007) Planning for sustainable tourism in southern
(2019) Analysis of Nordic regulatory framework and its effect on Pulau Banggi: an assessment of biophysical conditions and their
waste prevention and recycling in the region. Nordic Council of implications for future tourism development. Journal of environ-
Ministers mental management 85(4):999–1008
Pesaran, M. H. (2015) Time series and panel data econometrics. Oxford Thanh NP, Matsui Y, Fujiwara T (2010) Household solid waste genera-
University Press tion and characteristic in a Mekong Delta City, Vietnam. Journal of
Purcell M, Magette W (2009) Prediction of household and commercial Environmental Management 91(11):2307–2321
BMW generation according to socio economic and other factors for Trujillo-Lora JC, Carrillo BB, Charris Vizcaino CA, Iglesias Pinedo WJ
the Dublin region. Waste management 29(4):1237–1250 (2013) The environmental Kuznets curve (EKC): an analysis
Qiao H, Chen S, Dong X, Dong K (2019) Has China’s coal consumption landfilled solid waste in Colombia. Revista Facultad de Ciencias
actually reached its peak? National and regional analysis consider- Economicas: Investigacion y Reexion 21(2):7–16
ing cross-sectional dependence and heterogeneity. Energy Tsurumi T, Managi S (2010) Decomposition of the environmental
Economics 84:104509 Kuznets curve: scale, technique, and composition effects.
Ramachandra T, Bharath H, Kulkarni G, Han SS (2018) Municipal solid Environmental Economics and Policy Studies 11(1-4):19–36
waste: generation, composition and GHG emissions in Bangalore, Verbeek M (2008) A guide to modern econometrics. John Wiley & Sons
India. Renew-able and Sustainable Energy Reviews 82:1122–1136 Wang H, Nie Y (2001) Municipal solid waste characteristic and manage-
Reed, W. R. (2015) On the practice of lagging variables to avoid simul- ment in China. Journal of the Air & Waste Management Association
taneity. Oxford Bulletin of Economics and Statistics, 77(6):897-905. 51(2):250–263
Roodman D (2009) A note on the theme of too many instruments. Oxford Ward JD, Sutton PC, Werner AD, Costanza R, Mohr SH, Simmons CT
Bulletin of Economics and statistics 71(1):135–158 (2016) Is decoupling GDP growth from environmental impact pos-
Rothman DS (1998) Environmental Kuznets curves|real progress or pass- sible? PloS one 11(10):e0164733
ing the buck? A case for consumption-based approaches. Ecological Weng YC, Fujiwara T, Matsuoka Y (2007) Analysis of household expen-
Economics 25(2):177–194 diture and household waste emission in Taiwan. In: Proceedings of
Saeed MO, Hassan MN, Mujeebu MA (2009) Assessment of municipal the International Symposium on EcoTopia Science.
solid waste generation and recyclable materials potential in Kuala Wooldridge JM (2016) Introductory econometrics: a modern approach.
Lumpur, Malaysia. Waste management 29(7):2209–2213 Nelson Education.
Saladie O (2016) Determinants of waste generation per capita in Xu T (2018) Investigating environmental Kuznets curve in China- aggre-
Catalonia (north-eastern Spain): the role of seasonal population. gation bias and policy implications. Energy policy 114:315–322
European Journal of Sustainable Development 5(3):489–489
Shamshiry E, Nadi B, Bin Mokhtar M, Komoo I, Saa-diah Hashim H, Publisher’s note Springer Nature remains neutral with regard to jurisdic-
Yahaya N (2011) Integrated models for solid waste management in tional claims in published maps and institutional affiliations.

You might also like