Environmental Science and Pollution Research

https://doi.org/10.1007/s11356-019-06586-w

RESEARCH ARTICLE

The relationship between cement production, mortality rate, air

quality, and economic growth for China, India, Brazil, Turkey,
and the USA: MScBVAR and MScBGC analysis
Melike E. Bildirici 1

Received: 21 March 2019 / Accepted: 24 September 2019

# Springer-Verlag GmbH Germany, part of Springer Nature 2019

Abstract
The related literature reveal that the papers on environmental pollution do not sufficiently analyse the cement production which is
an important determinant of air pollution and health problems by using econometric methods. To fill this gap, this paper aims to
examine the relationship between cement production, air pollution, mortality rate, and economic growth by employing MS
Bayesian Vector Autoregressive (MScBVAR) and Markov Switching Bayesian Granger causality (MScBGC) approaches from
1960 to 2017 for China, Brazil, India, Turkey and the USA. MSIA(2)-BVAR(1) model for China, MSIAH(2)-BVAR(3) models
for India, MSIAH(3)-BVAR(2) for Brazil, and MSIAH(3)-BVAR(1) for Turkey, and MSIAH(2)-BVAR(2) for the USA were
selected. The MScBGC results revealed that the cement production is granger cause of mortality rate, air pollution, and economic
growth in all regimes for China, India, Brazil, Turkey, and the USA.

Keywords MScBVAR . MScBGC . Cement production . CO2 emissions

Introduction 2016; Lei et al. 2011; Chen et al. 2015). Accordingly,

Habert (2013), the effect of cement sector on CO2 emissions,
The 20th century witnessed the increase of big constructions, is more vital than the other energy-intensive sectors.
high bridges, hydroelectric dams, and highways around the Moreover, among the non-metallic mineral manufacturing
world. In the last of the 20th century especially in Asia, the processes, cement production is the most energy-consuming
more skyscrapers rose. This all conversions enhanced cement process. According to the ECA (European
production, and this increment has been continued since that CementAssociation), each ton of cement manufactured needs
time. USGS (2017) determined that through the World, during range from 60 to 130 kg for fuel oil, that base on the cement
1990–2014, the production increased from 209.7 million tons diversity and the process, and ~105 KWh of electricity (Xi
to 2476 million tons. Especially, in the effects of the increased et al. 2016; Lei et al. 2011; Chen et al. 2015).
cement demand since 1990, many cement factories were Especially, the workers and people living near the cement
established. plants are exposed to dust and gases resulting from chemical
Cement plants are major contributors of sulfur oxides reactions, polychlorinated biphenyls (PCBs), dibenzofurans
(SOx), carbon monoxide (CO), carbon dioxide (CO2), nitro- (PCDD/Fs), polychlorinated dibenzo-p-dioxins, NOx, Sox,
gen oxides (NOx), and PM2.5. For example 1 kg of cement CO, CO2, and hydrogen chloride, chlorine, ammonia (Leone
releases ~ 0.8 kg of CO2, In China, the CO2 releases from the et al. 2016; Mosca et al. 2014; Zou et al. 2018; Chinyama
cement comprise 5% of universal CO2 emissions (Xi et al. 2011; Raffetti et al. 2019), and additionally thallium.
Dolgner et al. (1983) and Brockhaus et al. (1981) determined
that in Germany, the cement factories were the source of thal-
Responsible editor: Philippe Garrigues lium. Moreover, during the production and transfer process of
cement, particulate matter and gaseous substances are emitted
* Melike E. Bildirici
into the environment.
[email protected]
Cement production is one of the factors causing un-
1
Economics Department, FEAS, Yıldız Technical University, healthy environment. According to the World Health
Istanbul, Turkey Organization (2018), 7 million persons die every year in
 Environ Sci Pollut Res

consequence of working or living in an unhealthy envi- capture Markov switching Bayesian Granger causalities
ronment. This rate equals to 23% of adult deaths and 26% (MScBGC) for different regimes. Sims and Zha (1998) used
of child deaths under 5. to estimate the MScBVAR models and also, MScBGC models.
Thus, in the effect of these processes, cement production has The literature review part of the study is given in the second
impacts of various diseases such as respiratory and section. While in the third part the data and econometric meth-
cardiovascular diseases, cancer, etc., and mortality. Some odology were presented, in the fourth part, model specifica-
papers showed that respiratory diseases such as, lung cancer, tions are defined. In the fifth section, the empirical results are
obstructive pulmonary disease, bronchitis, asthma, chest pain, supplied. In the last section, the policy discussion and the
pneumonia, emphysema, and cardiovascular diseases were conclusion are covered.
impacted by cement plants. Others found that cement
production causes mortality. Aydin et al. (2010) and Peters
et al. (2009) showed increased plasma lipid peroxidation and Literature review
reduced antioxidant capacity caused cerebrovascular mortality.
Although cement plants cause adverse effect on air quality, The literature review is given in two subtitles: the first subtitle
environmental pollution, and some health distortions on both gives the papers that analyze the relation between cement
the general inhabitants and workforces, the related literature production and environmental pollution and the second ones
revealed that the papers on environmental pollution do not suf- present the papers that analyze the relation between cement
ficiently analyze the relation between cement production, eco- production and health risk.
nomic growth, air and environmental pollution, and mortality
rate by using the econometric methods. To fill this gap, this The relation between cement production
study aims to determine the nonlinear causality approach based and environmental pollution
on Bayesian regime-switching method for analyzing the rela-
tionship among the economic growth, mortality rate, cement The literature generally addresses the relation between envi-
production, and air pollution. With this purpose, the MS ronmental damages and economic growth and focus on the
Bayesian Vector Autoregressive (MSc-BVAR) and MS species of contaminants such as suspended particles, carbon
Bayesian Granger causality (MSc-BGC) methods are dioxide (CO 2 ), and sulfur dioxide emissions. Firstly,
employed for China, Brazil, India, Turkey, and the USA from Grossman and Krueger (1991) explored the relation between
1960 to 2017. The selected countries for analysis are the most CO2 emission and GDP growth by using regression analyses
cement producing countries over the world. According to and they found the pollution rises at low levels of GDP and
Global cement report (2017) by Edwards, China, India, decreases in high level of GDP. And Shafik and
Brazil, Turkey, and the USA are among the world’s top cement Bandyopadhyay (1992) corrected the presence of inverted
producers in 2017. China is the first-largest cement producer by U-shaped EKCurve. Selden and Song (1994), Cole et al.
installed cement capacity in 2017. While India is the second- (1997), and Stern et al. (1996) obtained the significant results
largest cement producer, the USA is the third-largest producer. in the context of the EKCurve relationship.
Brazil is the sixth-largest producer. Turkey is in seventh rank. While in pursuit of these papers, many papers analyzed the
Russia as the fourth-largest producer and Vietnam as the fifth relation between economic growth and environmental quality;
were removed from the analysis due to data scarcity problem. some papers focused on the relation between environmental
The real GDP that was employed as a measure of economic pollution and sectoral production. Hanle (2004) in the USA;
growth has a nonlinear structure, asymmetric behavior in re- Mandal and Madheswaran (2010) in India; Hasanbeigi et al.
gimes to be regarded as business cycle. The fluctuations in (2010), Xu et al. (2014), and Lin and Zhang (2016) in China;
economic growth caused business cycles. Besides, the business and Dirik et al. (2018) in Turkey explored environmental ef-
cycles are the significant structures that can affect the relation- fects caused by the cement sector. These papers determined
ship between the cement production and CO2 releases over a the environmental effects of CO2 emission released by their
given period. In addition to economic growth, cement produc- cement sectors. Bildirici (2019) for China and the USA stud-
tion is effected by some other events as impact of the people’s ied Granger causality relation among cement production, eco-
psychological expectations, construction sector, etc. To deter- nomic growth, and environmental pollution by using Granger
mine non-linear structure, BDS and Tsay NL tests were applied. causality methods. And she determined unidirectional causal-
In non-linear structure, regime-dependent asymmetry is very ity from cement production to environmental pollution.
important. If regime-dependent asymmetry and regime changes On the other hand, some papers tested GHG releases in
are ignored, the policy recommendations will be wrong. In this cement sector. Wang et al. (2013) explored GHG releases for
condition, it was based on the Bayesian regime-switching meth- the cement sector and explained the causes of change in GHG
od and Markov switching Bayesian vector autoregressive releases. Ali et al. (2015) focused on the synergy between
(MScBVAR) methodology that also allows the researcher to LAPs (local air pollutants) and found the evidence of the
Environ Sci Pollut Res

strong link; moreover, they showed that air pollutant control between cement production, cardiovascular symptoms, and
can help as an endogenous instrument to mitigate GHG. mortality. Dab et al. (2011) showed cement workers have
Some papers focused on the other polluters such as PM2.5 higher risk of cancer mortality than the general population.
that leads to lung disease, lung cancer, heart disease and Rachiotis et al. (2012), Giordano et al. (2012a), and Koh
stroke, and mortality. Van Donkelaar (2010) found that in et al. (2013) explored the relation between the cancer risk
the world, the uppermost concentrations of PM2.5 were in and mortality for employment in cement production. They
Eastern China. Between 2001 and 2006, only 24 of 350 showed the emergence of different cancer risks caused by
Chinese regions have annual average PM2.5 concentrations emissions of the cement plant. Aydin et al. (2010) and Peters
below 10 μg/m3 value that is the advancement determined et al. (2009) showed increased plasma lipid peroxidation and
by the World Health Organization (Han et al., 2014). Keene reduced antioxidant capacity caused cerebrovascular
and Deller (2015) tested an EKC analysis of PM2.5 for USA. mortality. Giordano et al. (2012a) determined mortality and
They determined that the topmost EKC happens between the neoplasia for people living near cement plants. Smailyte et al.
range US$24,000 and US$25,500, basing upon the estimator (2004) and Rachiotis et al. (2012) found a rise in mortality rate
employed. Hao and Liu (2016) analyzed an EKC model for in regard to lung cancer mortality since first and cumulative
PM2.5 concentrations for 73 Chinese cities in 2013 and deter- exposure to dust for people living near cement plants.
mined an inverted U shape curve.

The relation between cement production and health The cement production, mortality rate,
risk and environmental damage

McDowall (1984), Amandus (1986), and Jakobsson et al. Cement production has increased through the twentieth cen-
(1990) determined health risk estimations for employed peo- tury and continues to increase. In the selected countries, espe-
ple in cement production. Mehraj et al. (2013) found that cially in China, India, and Turkey, the cement industry has
cement dust threatens health and that the cement industries developed rapidly in the recent 30 years in effects of rapid
have important adverse impacts on health. Meo et al. (2013) economic growth and urbanization.
explored the health problems caused by cement dust on lung After the 1950s, cement production has become one of the
functions. Bertoldi et al. (2012) showed that emissions in ce- main reasons environmental pollution rose from 209.7 million
ment plants are reasons of cardiovascular or respiratory tons (Mt) to 2476 Mt (USGS 2017). And huge quantities of air
diseases. pollutants, including NOX, SO2, CO2, and PM2.5, were emit-
Some papers accented the relation between cement produc- ted from cement production in the selected countries.
tion and cancer risk. Magnani et al. (2001), Amendola et al. However, when the cement was produced, the high-
(2003), Bilancia et al. (2003), and Musti et al. (2009) deter- temperature calcination of carbonate minerals caused clinker,
mined tumor risk for people living near cement plants. Maier and CO2 is emitted into the atmosphere (Xi et al. 2016). The
et al. (1992, 1999) found that the cancer risk caused by cement calcination is the reason for 50–60% of CO2 emission and the
production exhibits a positive correlation with the duration of rest of the emission is the result of burning fossil fuels used for
exposure. Jakobsson et al. (1993) for colon cancer and heating the raw materials in the kiln (NRMCA 2012) .
Magnani and Leporati (1998), Eom et al. (2017) and Lopez- Figure 1 gives the knowledge about the growth of CO2
Cima et al. (2013) for lung cancer found that the individuals emissions from cement production.
exposed to cement dust have a risk factor. Yhdego (1992), Especially in China and India, CO2 emissions from cement
and Yang et al. (1996), found that the risks of laryngeal cancer production have a high rate. Teller et al. (2000) showed that
and supra glottis cancer were increased by cement sector. Koh the CO2 emission of China’s cement sector is higher than the
et al. (2011), in six Portland cement plants in Korea, found a CO2 releases of many countries. And China is one of the
potential relation between stomach cancer and cement expo- countries that has the worst air quality in the world (Kan
sure. Garcia-Perez et al. (2017) found that the people exposed et al. 2012). In January 2013, a dangerous dense enclosed
to cement production have fourfold higher frequency of pedi- 1.4 million km2 of China and impacted more than 800 million
atric bone tumors. persons (Xu et al. 2013).
Some papers accented the nexus between cement Figure 2 a and b exhibit the growth of PM2.5 air pollution.
production and mortality. Legator et al. (1998) determined Figure 2 a exhibited the population rate exposed to greater
the risk of cardiovascular symptoms and mortality. Knox levels than WHO standard value for PM2.5. For Brazil, China,
and Gilman (1997) found a higher mortality for leukemia or India, and Turkey, the exceeding level from the WHO guideline
solid cancers in children living within 3 km from the cement value is very high, while for the USA, this value is increasing.
factory. Legator et al. (1998), Bertoldi et al. (2012), and In China, about 1.6 million individuals exposed to PM2.5 die
Giordano et al. (2012a) determined the positive relation per year from heart, stroke, and lung problems (Rohde and
 Environ Sci Pollut Res

Fig. 1 CO2 emissions from 1000000

cement production, Sources:
Andrew (2018) 800000

600000

400000

200000

0
Brazil China India Turkey USA

1960 1970 1980 1990 2000 2010 2011 2012 2013 2014 2015 2016

Muller 2015). In China, in the context of short-run effects, it taken from USGS and CEIC for China. The CO2 releases from
was found that a 10 μg/m3 increase in PM2.5 causes a 0.40% cement production were taken from Andrew (2018). Per capita
rise in non-accidental death, a 0.63% rise in death rate from GDP and mortality rate data were taken from World Bank.
cardiovascular disease, and a 0.75% rise in death caused by The cement production is measured as lcpt = ln(cement
respiratory disease (Lu et al. 2015). On the other hand, for the productiont) and ln(.) demonstrates the natural logarithms.
USA, it was found that in the individuals exposed to PM2.5, Other variables were transformed as lyt = ln(real GDPt),
both the mortality of lung cancers and cardiopulmonary prob- lmort = ln(mortality ratet), and lcot = ln(CO2emissions from
lems (Schwartz 2000) (Franklin et al. 2008) increase. cement production).
Cement production causes serious health problems and
mortality in the selected countries in effects of air pollutions Empirical methodology The MScBVAR model allows
such as CO2, PM2.5, etc. Some papers showed the significant obtaining the characteristics of the economy under different
relationships between air pollution and some diseases includ- regimes in pursuit of Hamilton (1990) and Krolzig (1996) and
ing shortness of breath, lung cancer, bronchitis, asthma, chest Krolzig (1997). The direction of causality is determined for
pain, and sore throat. Some other papers showed the levels of different regimes with MScBVAR causality approach.
SOx, NOx, and SPM cause higher frequencies of respiratory The MSVAR and MSGC models were discussed by Fallahi
diseases (Hart 1970; Dockery et al. 1993a, b; Schwartz 2002; (2011) and Bildirici (2012) to test the causal nexus between
Dockery and Pope 1994; Vestbo and Rasmussen 1990). And energy usage and economic growth based on MSIA(.)-VAR(.)
Pope et al. (2002), Bertoldi et al. (2012), and Giordano et al. and MSIAH(.)-VAR(.) models. The MSIAH(.)-VAR(.) model
(2012b) found that air pollution can cause high blood pressure is given as
and cardiovascular problems.
i
yt ¼ μðst Þ þ ∑ Aiðst Þ xt þ utðst Þ ð1Þ
i¼0

Dataset and empirical methodology where ut/st~N(0, δ2(st)) and Ai (.) show the coefficients of the

0 0
The data of real GDP as a measure of economic growth (y), CO2 lagged values of the variable in different regimes. xt ¼ xt
emissions (co) from cement production as a measure of air  0
¼ yt−1 ; :::; yt−p ; xt−1 ; :::; xt−p in matrix form for t = 1,2,...,n
pollution, mortality rate (mor), and cement production (cp) data
were employed for China, Brazil, India, Turkey, and the USA define the input variables. Pij = Pr[st = j| st − 1 = i] is transition
between the years 1960 and 2017. Cement production data were probabilities matrix:

Fig. 2 a PM2.5 air pollution, a b

population exposed to greater 120 120
levels than WHO standard value 100 100
(% of total). b PM2.5 air 80 80
pollution, mean annual exposure 60 60
(micrograms per cubic meter) 40 40
Sources: Worldbank (2019) 20 20
World Bank (2019), World 0 0
Development Indicators https:// Brazil China India Turkey United Brazil China India Turkey United
databank.worldbank.org/reports. States States
aspx?source=world- 1990 2000 2010 2017 1990 2000 2010 2017
development-indicators accessed
date:06/06/2019
Environ Sci Pollut Res
 
ut ∼i:i:d:N 0; δ2 ðst Þ if jϕj < 1 ð2Þ M
pðθi jy; S Þ∞ ∏ pðyt jθi ; yt−1 ; S Þpðθi Þ
t:S t ¼i
st is governed by a Markov chain.:
is the posterior density of the state−dependent parameters
ð10Þ
And it is defined as P(yt|Yt − 1, st − 1) = Pr(yt|Yt − 1) 3
pij is characterized as In this paper, it employed Sims and Zha (1998) methods. The
Prðst ¼ jjst−1 ¼ i; st−2 ¼ k; :::; s0 ¼ hÞ ¼ Prðst ¼ jjst−1 ¼ iÞ ¼ pij and posterior distribution of a is
1−p22 1−p11
Prðst ¼ 1Þ ¼ ; Prðst ¼ 2Þ ¼ pða=yt Þ ¼ πða0 Þπðaþ =a0 Þ ð11Þ
2−p11 −p22 2−p11 −p22
ð4Þ and

 0 
Dependent upon the coefficients of the lagged values of xt and 1
yt, the estimation of the MSIA(H)-VAR(q) models is given as πða0 Þ∞jA0 jT exp − trace A0 SA0 ð12Þ
2
follows:
πðaþ ja0 Þ ¼ φððI⊗U Þa0 ; I⊗V Þ
, and φ(μ; ∑) means the nor-
  " ðk Þ ðk Þ
#  mal density function in Eq. (12) (Akgul et al. 2017).
xt μ11;st q ϕ11;st ϕ12;st xt−k e
¼
μ22;st
þ ∑ ðk Þ ðk Þ þ 1;t ð5Þ
yt k¼1 ϕ ϕ22;st yt−k e2;t Specification of the model It tested the relation between ce-
21;st
ment production, economic growth, environmental pollution,
where s1,t and s2, t are variables which reflect the regime of the and mortality rate by using the MS Bayesian Vector
system at time t which take values in the zero to one interval; Autoregressive (MSc-BVAR) and MS Bayesian Granger cau-
where sality (MSc-BGC) methods for China, Brazil, India, Turkey,
and the USA from 1960 to 2017.
In pursuit of Fallahi (2011), the Granger causalities are MSIA(H)(.)-VAR(.) models are built as follows:
ðk Þ
noticed by examining the following hypotheses: H 0 : ϕ12 ¼ 2 3 2 3 2 ð jÞ ð jÞ ð jÞ ð jÞ 32 3
ðk Þ lyt μ1ðstÞ ϕ11 st ϕ12 st ϕ13 st ϕ14 st ly
0 and H 0 : ϕ21 ¼ 0 6 lcot 7 6 μ2ðstÞ 7 p 6 ð jÞ ð jÞ ð jÞ ð jÞ 76 t− j 7
6 7 6 7 ∑ 6 ϕ21 st
6 ϕ22 st ϕ23 st ϕ24 st 76 lcot− j 7
ð13Þ
likelihood function is given. If θ ∈ Θ ⊂i and θst :μst, Aist , ∑st 4 lcpt 5 ¼ 4 μ3ðstÞ 5 þ k¼1 4 ϕð31jÞ st ð jÞ ð jÞ ð jÞ
ϕ32 st ϕ33 st ϕ34 st
74
5 lcpt− j 5
lmort μ4ðstÞ ð jÞ ð jÞ ð jÞ ð jÞ lmort− j
for st=1, K, M where i=1, K, p. The likelihood function is ϕ st41 ϕ42 st ϕ43 st ϕ44 st
2 3
equal to the joint sampling distribution. It is decomposed into ε1;st
6 ε2;st 7
a product of a conditional distribution of y given S and θ, and þ6 7
4 ε3;st 5
a conditional distribution of S given θ (Akgul et al. 2017). ε4;st

pðS; yjθÞ ¼ pðyjS; θÞpðS jθÞ and Markov process is

For causality analysis, Fallahi (2011) and Bildirici (2012)
M M N ðsÞ ð6Þ
pðS jθÞ ¼ pðso jPÞ ∐ ∏ Pij ij utilized the MScGC for MSIA(H)(.)-VAR(.) models. In the
i¼1 j¼1
dlyt vector, dlco2t and/or dlcpt is/are Granger cause of dlyt in
ð jÞ ð jÞ
The Bayesian inference dependent upon the posterior dis- each jth regime if the ϕ12 and ϕ21 are statistically different
tribution of θ is employed. The likelihood function is contin- than zero. Accordingly, Granger causality is noticed by exam-
ðk Þ ðk Þ ðk Þ
ued by the selection of the prior distribution in Eq. 7: ining H 0 : ϕ12 ¼ 0 and H 0 : ϕ21 ¼ 0, H 0 : ϕ13 ¼ 0 and
ðk Þ ðk Þ ðk Þ
M H0 : ϕ31¼ 0, and H 0 : ϕ14
¼ 0 and H 0 : ¼ 0 for the ϕ41
pðθÞ ¼ ∏ pðθi ÞpðPÞ ð7Þ vector of dlyt (Bildirici and Gokmenoglu 2017).
i¼1
The Bayesian inference dependent upon the posterior dis-
This state permits the possibility of prior knowledge about θst tribution of θ is employed by depending upon equations be-
(Akgul et al. 2015) tween 7 and 12.

M
pðθjy; S Þ∞ ∏ pðθi jy; SÞpðPjy; SÞ define the posterior distribution ð8Þ Empirical results
i¼1

M M M N ðS Þ
pðPjS Þ∞ ∏ pðs0 jPÞ ∏ ∏ pij ij pð P Þ ð9Þ The following stages were applied.
i¼1 i¼1 j¼1
show posterior density of the transition probabilities 1. The results determined by descriptive statistics and
unit root tests are exhibited in Tables 1 and 2. ADF
and test and Kapetanios, Shin and Shell (KSS) non-linear
 Environ Sci Pollut Res

Table 1 ADF and KSS results

ADF KSS ADF KSS ADF KSS ADF KSS ADF KSS
China Brazil India Turkey USA

lyt 0.68 − 1.27 − 1.51 − 1.04 − 1.31 − 1.05 − 0.91 − 0.12 − 1.28 − 1.21
Δlyt − 4.25 − 5.89 − 7.76 − 5.91 − 6.42 − 5.86 − 4.772 − 5.61 − 5.79 − 4.76
lcot 0.181 − 1.19 − 1.17 − 1.01 − 1.41 − 1.19 − 1.41 − 1.75 − 1.35 − 1.65
Δlcot − 4.23 − 5.77 − 6.25 − 5.47 − 7.12 − 4.77 − 7.46 − 3.31 − 3.89 − 5.79
lcp 0.31 − 1.02 − 1.01 − 1.02 − 1.51 − 1.31 − 1.59 − 1.84 − 0.91 − 1.32
Δlcp − 3.11 − 5.12 − 4.92 − 4.67 − 5.98 − 3.54 − 5.07 − 4.66 − 3.99 − 4.98
lmort − 1.66 − 0.52 − 1.01 − 1.01 − .042 − 0.89 0.89 − 1.59 − 2.11 − 1.12
Δlmort − 6.34 − 4.18 − 5.16 − 6.17 − 3.13 − 4.331 − 4.86 − 3.99 − 5.78 − 5.98
Test critical values: 1% level − 3.552666
5% level − 2.914517
10% level − 2.595033

unit root test were employed to evaluate the station- In Table 1, the results of ADF and KSS tests are given.
arity of the variables. The sample period has non- The results of unit root test determined that all variables are
linear structure because ADF test can cause spurious at first-order stationary.
results. In this state, KSS test must be used as The results in Table 2 show descriptive statistics.
complimentary. The JB statistics showed the robust evidence that the null
2. Secondly, the BDS and TsayNL tests were applied to de- hypothesis of the normal distribution can be rejected.
termine nonlinear structure. And then, whether the series was exhibited the structures of
3. Johansen test was used to find the cointegration. When linear or non-linear were determined by the BDS and Tsay NL
the Johansen test does not determine a cointegrated tests. In Table 3. The results of BDS and Tsay NL tests exhib-
vector, the innovations of the series are employed to ited that the series are non-linear dependent on all dimensions.
determine the results from MSc Bayesian Granger The results in the tables 1, 2, and 3 determined that all
causality. variables have nonlinear structure at all dimensions and all
4. After examining the MScBVAR model with LR and LL series are stationary at the first difference.
tests, and posterior estimates, the MScBVAR models are In the second step, the presence of cointegration among lyt,
selected. lcpt, lmort, and lcot variables was explored by Johansen’s pro-
5. The direction of nexus via MSc Bayesian causality was cedure. Dependent upon Table 4, the null hypothesis of no
determined. cointegration was not rejected. It was not determined the

Table 2 Descriptive statistics

Brazil China
Series Skewness JB ARCH Skewness JB ARCH Skewness JB ARCH
0.307210 1.84680 1.47115
Δly − 0.8756 1.5510 1.569 0.609990 2.33071 1.0856 0.84782 2.69299 1.1766
Δlco − 0.722498 2.5731 1.756 0.84569 3.2569 1.15369 0.151308 1.96888 1.33277
Δlcp 0.266555 2.4751 1.121 − 0.2966 1.11304 1.1044 0.215770 2. 6451 1.0848
Δlmor 0.87100 3.3265 1.569 0.5377 1.52043 1.0350 0.455316 2.52043 1.0350
Turkey USA
Series Skewness JB ARCH Skewness JB ARCH
Δly − 0.61533 1.46415 1.3269 − 0.460946 2.12512 1.1236
Δlco 0.462489 3.280685 1.1145 − 1.221782 1.3852 1.09715
Δlcp − 0.524373 2.369797 1.5463 0.439235 3.38946 1.07236
Δlmor − 0.118837 1.600751 1.2369 0.210125 2.03890 1.0269
Table 3 Nonlinearity test results
Environ Sci Pollut Res

Dimension Δlco Δly Δlmor Δlcp Lag Δly ΔLco Δlcp Δlmor

BDS test Brazil

2 0.2074 (0.00) 0.2227 (0.00) 0.2810 (0.00) 0.1883 (0.000) Tsay NL test 1 2.43 (0.042) 6.881 (0.00) 3.84 (0.00) 4.011 (0.006)
3 0.3549 (0.00) 0.3699 (0.00) 0.3430 (0.00) 0.3126 (0.000) 2 6.7 (0.002) 7.451 (0.00) 4.424 (0.00) 4.465 (0.006)
4 0.4578 (0.00) 0.4574 (0.00) 0.4426 (0.00) 0.3916 (0.000) 3 3.544 (0.001) 6.089 (0.00) 2.668 (0.00) 3.879 (0.006)
5 0.5280 (0.00) 0.5279 (0.00) 0.5147 (0.00) 0.4402 (0.000) 4 2.995 (0.00) 5.442 (0.00) 2.645 (0.00) 3.201 (0.006)
6 0.5748 (0.00) 0.5739 (0.00) 0.5694 (0.00) 0.4707 (0.000) 5 2.648 (0.00) 4.334 (0.00) 1.758 (0.08) 2.052 (0.091)
BDS test 6 2.113 (0.00) 3.04 (0.00) 1.923 (0.011) 1.134 (0.423)
China
2 0.1939 (0.00) 0.1871 (0.00) 0.2010 (0.00) 0.2012 (0.00) Tsay NL test 1 4.5502 (0.00) 5.69794 (0.00) 1.410672 (0.496) 2.1854
3 0.3277 (0.00) 0.3113 (0.00) 0.3417 (0.00) 0.3407 (0.00) 2 6.1497 (0.00) 6.14374 (0.00) 3.699824 (0.00) 4.0869 (0.00)
4 0.4202 (0.00) 0.3922 (0.00) 0.4436 (0.00) 0.4369 (0.00) 3 3.3491 (0.00) 5.716 (0.00) 3.432 (0.00) 2.202 (0.00)
5 0.4855 (0.00) 0.4459 (0.00) 0.5166 (0.00) 0.5039 (0.00) 4 3.09229 (0.00) 5.006 (0.00) 2.2923 (0.00) 2.1915 (0.00)
6 0.5315 (0.00) 0.4797 (0.00) 0.5698 (0.00) 0.5514 (0.00) 5 2.9523 (0.042) 4.57702 (0.00) 2.216001 (0.00) 1.9826 (0.13)
6 1.9985 (0.013) 3.95589 (0.00) 1.9856 (0.13) 1.1772 (0.00)
India
BDS test 2 0.1987 (0.00) 0.1806 (0.00) 0.2029 (0.00) 0.1971 (0.00) Tsay NL test 1 5.244117 (0.00) 9.12778 (0.00) 6.56180 (0.00) 4.9544 (0.00)
3 0.3331 (0.00) 0.2977 (0.00) 0.3424 (0.00) 0.3301 (0.00) 2 3.914959 (0.00) 4. 6080 (0.00) 5.54450 (0.00) 4.0014 (0.00)
4 0.4283 (0.00) 0.3759 (0.00) 0.4399 (0.00) 0.4209 (0.00) 3 3.268569 (0.00) 2. 91445 (0.00) 2.1103 (0.00) 3.1270 (0.00)
5 0.4980 (0.00) 0.4267 (0.00) 0.5100 (0.00) 0.4834 (0.00) 4 3.058638 (0.00) 1.99457 (0.108) 1.9122 (0.18) 2.2789 (0.00)
6 0.5492 (0.00) 0.4602 (0.00) 0.5625 (0.00) 0.5264 (0.00) 5 2.65912 (0.00) 1.3631 (0.254) 1.19947 (0.33) 2.1001 (0.00)
6 2.08569 (0.00) 1.0075 (0.43) 1.00366 (0.43) 1.8856 (0.0289)
Turkey
BDS test 2 0.1997 (0.00) 0.1794 (0.00) 0.2017 (0.00) 0.1978 (0.00) Tsay NL test 1 3.777641 (0.00) 4.81554 (0.00) 1.8287 (0.09) 5.563898 (0.00)
3 0.3396 (0.00) 0.3051 (0.00) 0.3391 (0.00) 0.3345 (0.00) 2 5.500860 (0.00) 7.78277 (0.00) 4.2957 (0.00) 4.7272 (0.00)
4 0.4377 (0.00) 0.3897 (0.00) 0.4350 (0.00) 0.4304 (0.00) 3 2.56912 (0.00) 3.87186 (0.00) 3.5046 (0.00) 3.2889 (0.00)
5 0.5074 (0.00) 0.4468 (0.00) 0.5037 (0.00) 0.4988 (0.00) 4 1.89711 (0.021) 2.0737 (0.00) 3.5236 (0.00) 2.1995 (0.00)
6 0.5577 (0.00) 0.4866 (0.00) 0.5540 (0.00) 0.5485 (0.00) 5 1.19021 (0.22) 1.1456 (0.430) 1.8965 (0.029) 1.8965 (0.0291)
6 0.823790.541 0.96894 (0.641) 0.9986 (0.130) 1.0785 (0.436)
USA
BDS test 2 0.1897 (0.00) 0.2080 (0.00) 0.1990 (0.00) 0.1460 (0.00) Tsay NL test 1 3.99899 (0.00) 7.0136 (0.00) 6.5370 (0.00) 5.1339 (0.00)
3 0.3211 (0.00) 0.3535 (0.00) 0.3380 (0.00) 0.2387 (0.00) 2 3.87193 (0.00) 5.8701 (0.00) 3.6674 (0.00) 3.3926 (0.00)
4 0.4103 (0.00) 0.4560 (0.00) 0.4348 (0.00) 0.2939 (0.00) 3 3.65858 (0.00) 4.2849 (0.00) 2.3035 (0.08) 2.4804 (0.00)
5 0.4677 (0.00) 0.5287 (0.00) 0.5019 (0.00) 0.3247 (0.00) 4 3.11238 (0.00) 2.8763 (0.08) 2.0056 (0.19) 1.5604 (0.023)
6 0.5008 (0.00) 0.5808 (0.00) 0.5486 (0.00) 0.3399 (0.00) 5 1.81022 (0.009) 1.4624 (0.156) 1.4196 (0.156) 1.0803 (0.436)
6 0.56981 (0.930) 1.0536 (0.311) 1.1429 (0.223) 0.5698 (0.741)

Probability values are in parenthesis

 Environ Sci Pollut Res

Table 4 Johansen cointegration

test Brazil China India Turkey USA Trace statistic

r=0 45.99561 40.1493 34.07525 45.99556 41.55924 47.85613

r≤1 20.23477 24.296 15.09678 22.17412 26.86809 29.79707
r≤2 7.817571 13.2090 6.854731 8.485087 4.992105 15.49471
r≤3 0.274858 1.87 0.089440 2.035954 0.287009 3.841466

evidence of cointegration between the variables. The selected In Table 6, for India, Sims and Zha (1998) test determined
series are I(1), but they are not cointegrated. In these conditions, log posteriors as 70.12 for regime 1, 182.81 for regime 2, and
according to Table 4, the innovation or first differences of series, posterior estimates as 798.345 and 2208.98 for regimes 1 and
Δlyt, Δlcpt, Δlmort, and Δlcot, can be explored for MScBGC. 2, respectively. The results of the MSIAH(2)-BVAR(3) model
are given for India (Table 6). Prob(st = 2|st−1 = 2) = 0.79, and
MSc BVAR results1 AIC, Log-likelihood, and LR test statistics Prob(st = 1|st−1 = 1) = 0.68 gave the computed regime proba-
were used to select the models. Sims and Zha (1998) test was bilities. According to our results, regime 2 is persistent.
used to estimate the models. The models are tested for linear- In Table 7, for Brazil, Sims and Zha (1998) test determined
ity using the LR linearity statistics (and Davies) assuming the log posteriors as 63.99 for regime 1, 196.99 for regime 2, and
H0 and H1 hypotheses to be a linear model and a MS model, 74.552 for regime 3, and posterior estimates as 901.05,
respectively. And the optimal models were determined. In this 2211.07, and 1091.88, respectively. The results of the
process, the models were selected among the ones which com- MSIAH(3)BVAR(2) model determined the regime probabili-
plete their best, dependent on the results of the diagnostic ties as Prob(st = 3|st−1 = 3) = 0.83, Prob(st = 2|st−1 = 2) = 0.75,
tests. Moreover, it was tested the modified critical values as and Prob(st = 1|st−1 = 1) = 0.68.
the Davies upper bound. It was employed the modified LR In Table 8, for Turkey, log posteriors were determined as
criterion together with the properties of the given results to 125.66 for regime 1, 172.88 for regime 2, and 152.06 for
discriminate between 2 and 3 regime models (see between regime 3, and posterior estimates as 1614.05, 2098.36, and
Table 5, 6, 7, 8, and 9). Ang and Bekaert’s (2002) suggestion 1882.23, respectively.
as the simple modification was applied. According to their And lastly, for the USA (Table 9), log posteriors were de-
opinion, the modified freedom degrees for the LR test is equal termined as 82.88 for regime 1, 135.76 for regime 2, and
to the number of restrictions acquired from constraining the posterior estimates as 988.16 and 1875.99, respectively.
model to two regimes, p21 + p22 = 1 and p11 + p12 = 1, plus In all models, for regime 2, the growth regime is persistent.
the nuisance parameters in each model that are undefined un- In the situations described above, the presence of asymmetry
der the null hypothesis. cannot be rejected. The second regime is dominant regime
After this process, it was estimated MSIA(2) BVAR(1) according to the ergodic probabilities. And these states indi-
model for China, MSIAH(2) BVAR(3) models for India, cate the asymmetries.
MSIAH(3)–BVAR(2) for Brazil, MSIAH(3) BVAR(1) for The first dependent variable is dlcot, the innovations of
Turkey, and MSIAH(2)–BVAR(2) for the USA. Regime 1 CO2 emissions; the third one is dlmort, the innovations of
describes the recession stage and/or high-volatility regime; mortality rate; and the last one is dlcpt, the innovations of
regime 2 depicts low volatility and/or the growth regime. cement production. The selected countries are a cement pro-
However, in the presence of three regimes, regime 1 defines ducer and additionally, by considering the size of the coun-
the recession stage and/or high-volatility regime; the moderate tries, covering important effects of the economic growth on
growth and/or moderate volatility regime are known as regime cement production is expected. The second dependent vari-
2, and regime 3 symbolizes low volatility and/or high-growth able in all regimes is dlyt, innovations of per capita GDP. In the
stage regime. Total period spent in regimes 3 and 2 is longer second vector, the majority of the coefficients are statistically
than regime 1. significant at the conventional levels, the innovations of ce-
The results of the MSIA(2)BVAR(1) model for China are ment production and economic growth on CO2 releases and
shown in the Table 5. Sims and Zha (1998) test determined log mortality rate cannot be irrefutable.
posteriors as 92.66 for regime 1 and 123.88 for regime 2, and
posterior estimates as 1009.88 and 1423.77 for regime 1 and Traditional and MSc causality results It will be compared the
2, respectively. Prob(st = 2|st−1 = 2) = 0.811 and Prob(st = 1|st possible differences and similarities of causality results, be-
−1 = 1) = 0.621 show the computed regime probabilities. cause the direction of causality is very important for recom-
mendations about the policy suggestions. In Table 10, the
traditional results are displayed. For China, it was determined
1
The MScVAR and MScGC analyses are realised in Oxmetrics package 3. that there is no causality relation between mortality rate and
Environ Sci Pollut Res

Table 5 MSIA(2)-BVAR(1) model estimation results for China

Regime 1 Regime 2

Variables Δlcot Δlyt Δlmort Δlcpt Δlcot Δlyt Δlmort Δlcpt

c 0.32 (1.6) 0.21 (1.28) − 0.52 (− 1.82) 0.44 (1.89) 0.22 (1.81) 0.51 (1.74) 0.144 (2.22) − 0.44 (2.23)
Δlco t-1 0.571 (2.32) 0.33 (1.30) 0.21 (2.11) − 0.33 (0.19) − 0.56 (0.22) − 0.128 (0.52) 1.017 (2.66) 1.56 (1.21)
Δlyt-1 0.08 (0.19) 0.57 (1.95) 0.12 (2.21) 0.33 (1.99) − 0.87 (2.66) − 0.72 (2.12) − 0.451 (2.75) 0.58 (2.32)
Δlmort -1 − 0.25 (0.61) 0.74 (1.14) 0.77 (1.98) 0.27 (0.55) 0.85 (1.25) 0.375 − 0.102 (1.21) − 2.84 (0.13)
(0.59)
Δlcpt -1 0.0961 (2.5) 0.23 (2.58) 0.33 (3.71) 0.085 (0.77) 0.828 (2.23) 0.72 (1.88) 0.725 (2.76) 1.22 (1.32)
se 0.064 0.073 0.015 0.075 0.064 0.073 0.015 0.075
Trans.Prob. To determine the number of regimes
P11 0.621 AIC − 11.10 LR test
P22 0.811 LL 425.22 H0: Linear VAR(1) 222.33
H1: MSIA(2)-BVAR(1)
DAVIES = [0.0000]
Sims-Zha (normal-Wishart)
Log posterior Posterior estimate
92.66 1009.88
123.88 1423.77

cement production and bi-directional causality for Turkey, and pollution to mortality for the USA, China, India, Brazil,
Brazil, and unidirectional causality from mortality rate to ce- and Turkey. The evidence of feedback relation was found
ment production for India and USA. There is no causality between cement production and real GDP in China, India,
relation between pollution and economic growth for Brazil, Turkey, Brazil, and the USA. According to the results of this
India, and Turkey while there is a bidirectional relation for paper, cement production is granger cause of air pollution
China and the USA. And there is no causality relation between and mortality. The results of this paper are similar to Lin and
economic growth, and cement production for the USA, Zhang (2016) that determine a significant role on CO2 emis-
Turkey, India, and China. In Brazil, the cement production is sion of the cement sector in China. And Bildirici (2019)
Granger cause of CO2 emission, but there is unidirectional found cement production is granger cause of environmental
causality from air pollution to cement production in India, pollution; this result is similar to the results of this paper.
Turkey, and the USA and bi-directional causality in China. The results found by MScBGC and traditional causality
There is a feedback relationship between mortality rate and tests are severely differentiated from each other. The
economic growth only for Turkey. For China and India, there MScBGC takes into account the fluctuations of the economy,
are unidirectional causality relations from mortality to GDP while the traditional causality test does not. Traditional cau-
and no causality relation for Brazil and the USA. There is bi- sality test does not consider the different regime and regime-
directional causal relation between mortality rate and CO2 dependent asymmetry and analyze the economy in the same
emission in Brazil, Turkey, and the USA. regime through whole period. When the regimes are taken into
In Table 11, the results of MSc Bayesian causality consideration, the results and policy recommendations are
(MScBGC) are given. differentiated.
For India, USA, and Brazil, in all regimes, the MScBGC
results found the evidence of feedback relation between air
pollution and real GDP, and only in regime 1 for Turkey and Policy discussion and conclusions
no causality for regime 1 in China. Different result in regime
1 for China emerged from China’s economic performance In this paper, it was analyzed the MScGC relation between
that does not experience the crisis for long period and from cement production, mortality rate, economic growth, and air
China’s rapid urbanization period. It was determined the quality via MScBVAR and MScBGC methods for the period
one-way causal relation running from cement production from 1960 to 2017 for China, India, Brazil, Turkey, and the
to air pollution for China, India, Brazil, Turkey, and the USA.
USA in all regimes except regime 2 for Turkey and Brazil. According to the results of this paper, the cement sector is
In all regimes, the MScBGC results determined the the important reason of mortality rate and air pollution. Health
evidence of unidirectional causal relation from air risk is increased by the emissions of the cement production for
Table 6 MSIAH(2)-BVAR(3) model estimation results

Regime 1 Regime 2

Variables Δlcot Δlyt Δlmort Δlcpt Δlcot Δlyt Δlmort Δlcpt

c 0.22 (0.27) 0. 87 (0.44) 0.113 (0.64) 0.43 (0.011) 0.33 (1.41) 0.51 (2.64) − 0.81 (2.11) 0.106 (1.83)
Δlco t-1 − 0.37 (1.21) 0.62 (0.84) 0.55 (2.45) − 0.22 (0.147) 0.74 (1.36) 0.25 (2.29) 0.75 (2.33) − 0.35 (1.04)
Δlco t-2 − 0.42 (1.31) 1.38 (2.43) 0.41 (1.83) − 0.33 (1.28) − 0.88 (0.61) 1.36 (2.14) 0.73 (2.84) 0.58 (1.34)
Δlco t-3 0.44 (0.14) 1.6 (0.17) 0.146 (0.71) 0.31 (1.23) 0.51 (0.21) − 1.01 (0.41) 0.11 (1.22) 0.863 (0.61)
Δlyt-1 0.77 (2.86) 0.54 (1.28) 0.86 (1.73) − 0.136 (1.01) − 0.76 (2.84) 0.21 (1.05) − 0.59 (2.38) − 0.22 (1.18)
Δlyt-2 − 0.84 (0.32) − 0.32 (0.87) − 0.42 (0.86) 0.49 (2.44) − 0.54 (0.83) − 0.76 (1.38) − 0.41 (1.31) 0.28 (2.46)
Δlyt-3 0.52 (2.17) − 0.38 (1.09) − 0.23 (0.01) 0.55 (1.81) 0.63 (1.87) − 0.21 (0. 16) − 0.18 (1.08) − 0.03 (0.88)
Δlmort -1 1.09 (0.21) − 1.21 (0.87) 0.35 (1.07) 0.17 (1.09) 0.41 (0.73) − 1.02 (0.17) 0.68 (1.26) − 0.26 (1.01)
Δlmort -2 1.42 (0.95) 1.07 (1.41) − 0.25 (1.09) 0.54 (0.97) 1.03 (1.27) − 0.98 (1.26) − 0.38 (0.93) 1.73 (1.28)
Δlmort -3 − 3.08 (1.05) 0.44 (1.36) 0.121 (1.07) − 0.33 (1.11) − 0.71 (1.04) 0.92 (1.06) 0.56 (0.81) − 0.38 (0.19)
Δlcpt-1 0.62 (2.71) 1.08 (2.93) 0.82 (2.89) − 0.34 (1.23) 0.512 (1.12) 1.28 (1.97) − 0.36 (1.27) − 0.72 (0.08)
Δlcpt -2 0.64 (1.14) 0.71 (1.99) 0.161 (1.77) − 0.32 (1.06) 0.23 (0.88) − 0.61 (1.01) 0.386 (2.77) − 0.19 (1.34)
Δlcpt -3 − 0.81 (1.22) − 0.31 (0.45) 0.21 (0.76) − 0.51 (1.88) 0.53 (1.99) 0.78 (1.86) 0.43 (0.01) − 0.56 (0.28)
se 0.018 0.06 0.003 0.03 0.02 0.058 0.0038 0.026
Trans.Prob. AIC − 12.01 The number of regimes and LR test
P11 0.68 LL 595.321 Test LR
P22 0.79 DAVIES = [0.0000] H0: Linear VAR(3) 128.74
H1: MSIAH(2)-BVAR(3)
Sims-Zha (normal-Wishart)
Log posterior Posterior estimate
70.12 798.345
182.81 2208.98
Environ Sci Pollut Res
 Environ Sci Pollut Res

Table 7 MSIAH(3)-BVAR(2) model estimation results for Brazil, 1960–2017

Regime 1 Regime 2 Regime 3

Variables Δlcot Δlyt Δlmort Δlcpt Δlcot Δlyt Δlmort Δlcpt Δlcot Δlyt Δlmort Δlcpt

c 0.27 (1.86) 0.23 (1.15) − 0.22 (0.89) − 0.37 (2.12) 0.47 (1.15) 0.33 (0.82) − 0.23 (0.19) − 0.18 (1.91) − 0.17 (0.33) − 0.79 (0.03) − 0.77 (2.13) 0. 21 (1.93)
Δlco t-1 0.512 (1.97) 0.541 (0.21) 0.75 (1.86) − 0.51 (0.62) − 0.33 (1.33) 1.28 (1.04) − 0.23 (0.77) 0.27 (0.89) 0.56 (1.07) − 1.33 (3.47) 0.23 (1.93) − 1.07 (0.994)
Δlco t-2 − 0.44 (3.42) 1.25 (2.55) 0.53 (2.64) − 0.17 (0.08) 0.34 (1.98) 1.56 (2.11) − 0.61 (2.22) 0.56 (2.33) − 0.13 (1.42) − 1.12 (2.54) − 0.23 (1.88) − 1.27 (1.38)
Δlyt-1 0.51 (2.16) 0.39 (1.92) − 1.03 (0.78) 0.72 (1.95) − 0.112 (0.51) 0.29 (1.18) − 0.73 (2.61) 0.29 (0.15) 0.68 (1.87) 1.67 (2.44) 0.23 (0.56) 0.64 (1.22)
Δlyt-2 0.06 (0.45) 0.33 (1.07) − 0.09 (0.88) 0.12 (0.77) 0.289 (2.07) 0.07 (0.13) 0.19 (0.31) 0.55 (2.19) 1.17 (1.11) 2.16 (1.99) 0.27 (1.33) 0.93 (1.99)
Δlmort -1 0.59 (0.51) 0.15 (0.44) − 0.52 (2.11) 0.15 (0.23) 0.11 (0.19) 1.17 (1.02) − 0.12 (0.33) 0.17 (0.95) 0.86 (0.27) 1.13 (1.44) − 0.36 (0.41) 0.86 (1.188)
Δlmort -2 0.55 (1.02) − 0.65 (1.22) − 0.38 (1.01) 0.71 (0.92) − 0.46 (0.29) 0.12 (1.07) 0.41 (1.21) − 0.45 (0.21) − 0.42 (0.52) 1.01 (1.18) − 0.27 (1.43) 0.93 (1.33)
Δlcp t-1 0.48 (3.14) − 0.13 (0.08) − 0.56 (2.42) 0.76 (2.44) 0.52 (2.28) 0.39 (1.98) 0.61 (1.93) 0.44 (1.22) 0.71 (2.33) 0.76 (1.88) 0.61 (2.33) 0.17 (1.51)
Δlcpt-2 0.45 (1.34) 0.15 (1.901) − 0.19 (0.75) − 0.25 (0.89) − 0.51 (1.77) 0.38 (2.79) − 0.17 (1.04) − 0.55 (0.99) 0.06 (0.11) 0.51 (0.08) − 0.24 (0.37) 0.82 (2.51)
se 0.038 0.051 0.188 0.044 0.037 0.055 0.0144 0.033 0.036 0.031 0.019 0.032
Trans.Prob. The number of regimes and LR test
P11 0.68 AIC − 10.17 LR test
P22 0.75 LL 605.328 H0: Linear VAR(2) 111.56
H1: MSIAH(2)-BVAR(2)
P33 0.83 DAVIES = [0.0000] H0: MSIAH(2)-BVAR(2) 301.881
H1: MSIAH(3)-BVAR(2)
Sims-Zha (normal-Wishart)
Log posterior Posterior estimate
63.99 901.05
196.99 2211.07
74.552 1091.88
Table 8 MSIAH(3)-BVAR(1) model estimation results for Turkey, 1960–2017

Regime 1 Regime 2 Regime 3

Variables Δlcoo Δlyt Δlmort Δlcpt Δlc0o Δlyt Δlmort Δlcpt Δlcoo Δlyt Δlmort Δlcpt

c 0.32 (0.3) 0.58 (1.71) − 0.041 (0.71) 0.35 (1.58) 0.15 (1.01) − 0.33 (0.41) − 0.53 (0.23) 0.23 (1.44) 0.11 (4.45) 0.42 (0.02) − 0.221 0.56 (2.01)
(0.51)
Δlcot-1 − 0.98 (1.11) 0.98 (3.15) − 0.40 (1.37) 0.54 (0.65) 0.13 (0.81) 0.67 (0.34) 0.77 (2.74) − 0.55 (2.14) − 0.31 (0.11) 0.14 (1.32) 0.32 (1.97) − 0.28 (0.66)
Δlyt-1 0.58 (2.23) − 0.72 (1.01) 0.452 (1.07) 0.54 (2.88) 0.42 (2.57) 0.112 (1.28) − 0.23 (2.01) 0.56 (4.13) 0.47 (2.56) 0.412 (0.91) 0.42 (0.88) 0.42 (3.07)
Δlmort -1 0.99 (0.15) 1.23 (1.42) 0.41 (0.61) 0.34 (0.31) 0.23 (0.72) − 1.17 (0.21) 0.27 (0.44) 1.01 (0.14) − 1.25 (0.11) 1.05 (0.87) 0.51 (0.63) − 2.01 (0.34)
Δlcpt-1 0.56 (1.97) 1.48 (2.81) 0.48 (1.98) − 0.45 (0.12) 0.31 (2.83) 0.57 (2.34) 0.52 (1.23) − 0.45 (0.11) 0.83 (2.23) − 1.45 (2.82) 0.56 (2.44) 0.11 (0.43)
se 0.031 0.123 0.03 0.06 0.024 0.111 0.005 0.064 0.021 0.212 0.0027 0.063
Trans.Prob. The number of regimes and LR test
P11 0.68 AIC − 18.77 LR test
P22 0.86 LL 550.01 H0 : Linear VAR(1) 94.75
H1 : MSIAH(2)-BVAR(1)
P33 0.61 DAVIES = [0.0000] H0 : MSIAH(2)-BVAR(1) 120.56
H1 : MSIAH(3)-BVAR(1)
Sims-Zha (normal-Wishart)
Log posterior Posterior estimate
125.66 1614.05
172.88 2098.36
152.06 1882.23
Environ Sci Pollut Res
Environ Sci Pollut Res

Table 9 MSIAH(2)BVAR(2) model estimation results for USA, 1960–2017

Regime 1 Regime 2

Variables Δlcot Δlyt Δlmort Δlcpt Δlcot Δlyt Δlmort Δlcpt

c 0.1 (1.07) 0.14 (0.18) 0.22 (1.88) − 0.22 (0.41) 0.223 (1.85) 0.75 (1.12) 0.72 (1.88) 0.32 (2.55)
Δlcot-1 − 0.12 (1.42) 0.24 (1.29) 0.54 (2.64) 0.173 (0.11) 0.44 (0.75) 0.42 (0.141) 0.38 (2.24) 0.47 (0.97)
Δlcot-2 − 0.34 (0.77) 0.31 (2.61) 0.27 (2.33) − 0.46 (1.27) 0.23 (0.62) 0.75 (1.904) 0.53 (2.12) 0.68 (0.37)
Δlyt-2 − 0.63 (2.42) − 0.44 (1.17) 0.25 (2.94) − 0.55 (2.88) 0.43 (0.18) 0.63 (0.43) − 0.23 (2.81) − 0.56 (2.71)
Δlyt-2 1.05 (2.46) 1.17 (1.45) 0.63 (1.97) 1.48 (1.93) 0.33 (1.98) 0.82 (2.47) − 0.11 (1.17) 0.82 (2.66)
Δlmort -1 0.42 (1.03) 1.23 (1.08) 0.17 (1.22) 0.12 (0.76) 0.142 (0.86) − 0.38 (1.37) − 0.22 (1.02) − 0.17 (0.19)
Δlmort -2 0.35 (0.11) − 0.87 (0.11) − 0.46 (0.92) 0.44 (1.11) 0.275 (0.61) − 0.77 (0.11) 0.22 (0.14) − 1.28 (0.57)
Δlcp t-1 0.63 (1.89) 0.94 (2.41) 0.62 (1.89) 0.24 (0.77) − 0.36 (2.44) 1.06 (3.09) 0.84 (0.87) − 0.55 (1.07)
Δlcp t-2 0.48 (0.87) 1.02 (2.87) 0.22 (0.13) 0.55 (0.77) − 0.28 (1.43) − 0.41 (2.72) 0.92 (2.76) 0.114 (0.72)
se 0.035 0.022 0.009 0.029 0.018 0.019 0.005 0.023
TransProb. The number of regimes and LR test
P11 0.66 AIC − 14.12 Test LR test
P22 0.84 LL 595.88 H0: Linear VAR(2) 294.75
H1: MSIAH(2)-BVAR(2)
DAVIES = [0.0000]
Sims-Zha (normal-Wishart)
Log posterior Posterior estimate
82.88 988.16
135.76 1875.99

Table 10 Linear Granger causality results

Δ lcp → Δ lco Δlcp → Δly Δly → Δlco Δlcp → Δlmor Δlmor → Δly Δlmor → Δlco
Δlco → Δlcp Δly → Δlcp Δlco → Δly Δlmor → Δlcp Δly → Δlmor Δlco → Δlmor

China
7.67 0.143 8.69 0.52 7.88 0.12
12.61 0.778 5.75 0.82 0.332 7.66
Direction of causality Δlcp↔Δlco Δlcp ≠ Δly Δ ly ↔Δlco Δlcp≠Δlmor Δlmor→ Δly Δlco→Δlmor
India
0.456 1.23 0.14 0.53 3.028 1.01
5.115 0.89 1.327 6.88 0.112 7.569
Direction of causality Δlco → Δlcp Δlcp ≠ Δly Δly ≠ Δlco Δlmor → Δlcp Δlmor → Δly Δlco → Δlmor
Brazil
7.125 7.89 0.442 9.89 0.61 9.18
1.112 0.23 0.971 5.56 0.081 5.71
Direction of causality Δlcp → Δlco Δlcp → Δly Δly ≠ Δlco Δlmor ↔ Δlcp Δlmor ≠ Δly Δlco ↔ Δlmor
Turkey
1.045 1.22 0.04 7.71 5.71 8.52
10.112 1.56 1.17 5.05 4.12 5.66
Direction of causality Δlco → Δlcp Δlcp ≠ Δly Δly ≠ Δlco Δlmor ↔ Δlcp Δlmor ↔ Δly Δlco ↔ Δlmor
US
0.562 0.445 6.45 0.02 1.11 3.98
11.67 0.62 6.24 3.78 1.33 4.13
Direction of causality Δlco → Δlcp Δlcp ≠ Δly Δly ↔ Δlco Δlmor → Δlcp Δlmor ≠ Δly Δlco ↔ Δlmor

Δ was used to show first differences

 Environ Sci Pollut Res

Table 11 MSc Bayesian Granger causality results

China India USA

First regime Second regime First regime Second regime First regime Second regime
Δly≠Δlco Δly→Δlco Δly↔Δlco Δly↔Δlco Δly↔Δlco Δly↔Δlco
Δlcp→Δlco Δlcp→Δlco Δlcp→Δlco Δlcp→Δlco Δlcp→Δlco Δlcp→Δlco
Δlco →Δlmor Δlco →Δlmor Δlco →Δlmor Δlco →Δlmor Δlco →Δlmor Δlco →Δlmor
Δly→Δlmor Δly→Δlmor Δly→Δlmor Δly→Δlmor Δly→Δlmor Δly→Δlmor
Δlcp↔Δly Δlcp↔Δly Δlcp↔Δly Δlcp↔Δly Δlcp↔Δly Δlcp↔Δly
Δlcp→Δlmor Δlcp→Δlmor Δlcp→Δlmor Δlcp→Δlmor Δlcp→Δlmor Δlcp→Δlmor
Brazil Turkey
First Regime Second Regime Third Regime First Regime Second Regime Third Regime
Δly↔Δlco Δly↔Δlco Δly↔Δlco Δly↔Δlco Δly→Δlco Δly→Δlco
Δlcp→Δlco Δlcp↔Δlco Δlcp→Δlco Δlcp→Δlco Δlcp↔Δlco Δlcp→Δlco
Δlco →Δlmor Δlco→Δlmor Δlco →Δlmor Δlco →Δlmor Δlco→Δlmor Δlco →Δlmor
Δly≠Δlmor Δly→Δlmor Δly≠Δlmor Δly≠Δlmor Δly→Δlmor Δly≠Δlmor
Δly↔Δlcp Δlcp↔Δly Δlcp↔Δly Δlcp↔Δly Δlcp↔Δly Δlcp↔Δly
Δlcp→Δlmor Δlcp→Δlmor Δlcp→Δlmor Δlcp→Δlmor Δlcp≠Δlmor Δlcp→Δlmor

Δ was used to show first differences

all individuals. The estimated results demonstrated that it is usage of renewable energy to support by eliminating the reli-
not possible to reduce CO2 releases and mortality rate, if the ance on fossil fuels, and the policy makers should submit a
cement production is not decreased. But if cement production subsidy program that encourages the firms to grip renewable
was decreased, the economic growth will be reduced because energy.
of the evidence of bi-directional causality between economic And lastly, if the governments focus on solve environmen-
growth and cement production for China, India, Brazil, tal problems, some diseases can be prevented. Moreover, they
Turkey, and the USA. can contribute to the design of preventive health strategies in
The policy makers must consider the environment and context of the environment-health relations.
health impacts of cement production. Short, medium, and
long-term government policies must focus on improvement
and control of cement production. The governments must en- References
courage the firms to provide technological improvement by
ordering the development of environmental technologies. In Akgul I, Bildirici ME, Ozdemir SO (2017) Evaluating the nonlinear link-
the perspective of environmental pollution, these countries age between gold prices and stock market index using Markov-
must investigate different methods in cement production tech- Switching Bayesian VAR Models. In: Istanbul as s Global
Financial Center, Ed. by, Edited by M. Bildirci, C. Zehir, F.
nique to decrease air pollutions as well as CO2 emissions. To Kayıkcı, M.Karagoz, and T. Bakirtas. Cambridge Scholars
reduce high pollution caused by the cement sector, it must be Publishing, Cambridge
stimulated to invest in new technologies and to enhance ener- Akgul I, Bildirici M, Özdemir S (2015) Evaluating the nonlinear linkage
gy efficiency. It was eradicated the old-fashioned production between gold prices and stock market index using markov-
switching Bayesian VAR models. Procedia - Social and
methods, and it was supported the cement production method
Behavioral Sciences 210:408–415
by dry rotary kiln as an alternative to cement production by Ali N, Jaffar A, Anwer M, Alwi SKK, Anjum MN, Ali N, Raja MR,
shaft kiln as an efficient way to decrease the national and Hussain A, Ming X (2015) The greenhouse gas emissions produced
general environmental damages. Moreover, air pollution con- by cement production and its impact on environment: a review of
trol systems, and optimizing transport distance can be global cement processing. International Journal of Research (IJR)
2(2):488–500
suggested.
Amandus HE (1986) Mortality from stomach cancer in United States
Because decreased cement production can cause slowing cement plant and quarry workers, 1950–80. Br J Ind Med 43:526–
economic growth, to decrease releases from cement produc- 552
tion is probable with new production systems and energy ef- Amendola P, Belli S, Binazzi A, Cavalleri A, Comba P, Mastrantonio M
ficiency. The energy efficiency must be enhanced by new et al (2003) Mortality from malignant pleural neoplasms in Broni
(Pavia), 1980–1997. Epidemiol Prev 27(2):86–90
technology. And non-renewable energy should be substituted
Andrew R (2018) Global CO2 emissions from cement production. Earth
by renewable energy. And policymakers must stimulate the Syst Sci Data 10:195–217
Environ Sci Pollut Res

Ang A, Bekaert G (2002) Regimes switches in interest rates. J Bus Econ areas: new findings from a case-control study. Sci Total Environ
Stat 20:163–182 579:1333–1342
Aydin S, Croteau G, Sahin I, Citil C (2010) Ghrelin, nitrite and paraox- Giordano F, Dell’Orco V, Fantini F et al (2012a) Mortality in a cohort of
onase/arylesterase. Concentrations in cement plant workers. J Med cement workers in a plant of Central Italy. Int Arch Occup Environ
Biochem 29(2):78–83 Health 85:373–379
Bertoldi M, Borgini M, Tittarelli A, Fattore E, Cau A, Fanelli R, Giordano F, Grippo F, Perretta V, Figa-Talamanca I (2012b) Impact of
Crosignani P (2012) Health effects for the population living near a cement production emissions on health: effects on the mortality
cement plant: an epidemiological assessment. Environ Int 41:1–7 patterns of the population living in the vicinity of a cement plant.
Bilancia M, Cavone D, Pollice A, Musti M (2003) Valutazione del rischio Fresenius Environ Bull 21(7A):1905–1909
di mesotelioma: il caso di una fabbrica per la produzione di Grossman G, Krueger A (1991) Environmental impacts of a North
cemento-amianto nella citta` di Bari. Epidemiol Prev 27(5):277–282 American Free Trade Agreement., no. 3914: 1-57: National
Bildirici M (2012) The relationship between economic growth and elec- Bureau of Economic Research NBER WP
tricity consumption in Africa: MS-VAR and MS-Granger causality Habert G (2013) Environmental impact of Portland cement production,
analysis. Journal of Energy and Development 37(2):179–207 eco-efficient concrete. Woodhead Publishing Series in Civil and
Bildirici (2019) Cement production, environmental pollution, and eco- Structural Engineering. Pacheco-Torgal F, Jalali S, Labrincha J,
nomic growth: evidence from China and USA. Clean Techn Environ John VM (Ed.), pp 3–25
Policy 21:783–793 Hamilton JD (1990) Analysis of time series subject to changes in regime.
Bildirici M, Gokmenoglu S (2017) Environmental pollution, hydropower J Econ 45(1–2):39–70
energy consumption and economic growth: Evidence from G7 Hanle LJ (2004) CO2 Emissions Profile of the U.S. Cement Industry.
countries. Renew Sust Energ Rev 75:68–85 https://p2infohouse.org/ref/43/42654.pdf
Brockhaus A, Dolgner R, Ewers U, Kramer U, Soddemann H, Wiegand Hao Y, Liu Y-M (2016) The influential factors of urban PM 2.5 concen-
H (1981) Intake and health effects of thallium among a population trations in China: a spatial econometric analysis. J Clean Prod 112:
living in the vicinity of a cement plant emitting thallium containing 1443–1453
dust. Int Arch Occup Environ Health 48:375–389 Hart RH (1970) The concept of APS: air pollution syndromes. Journal of
Chen W, Hong J, Xu C (2015) Pollutants generated by cement production South Carolina Medical Association 66:71–73
in China, their impacts, and the potential for environmental im- Hasanbeigi A, Price L, Lu H, Lan W (2010) Analysis of energy-
provement. J Clean Prod 103:61–69 efficiency opportunities for the cement industry in Shandong
Chinyama M (2011) Alternative fuels in cement manufacturing. In: Province, China: a case study of 16 cement plants. Energy 35:
Manzanera M (ed) Alternative fuel, pp 263–284 3461–3473
Cole MA, Rayner AJ, Bates JM (1997) The environmental kuznets curve:
Jakobsson K, Attewell R, Hultgren B, Sjöland K (1990) Gastrointestinal
an empirical analysis. Environ Dev Econ 2:401–416
cancer among cement workers, Int. Arch Occup Environ Heath
Dab W, Rossignol M, Luce D et al (2011) Cancer mortality study among
62(4):337–340
French cement production workers. Int Arch Occup Environ Health
Jakobsson K, Horstmann V, Welinder H (1993) Mortality and cancer
84:167–173
morbidity among cement workers. Br J Ind Med 50:264–272
Dirik C, Şahin S, Engin P (2018) Environmental efficiency evaluation of
Kan H, Chen R, Tong S (2012) Ambient air pollution, climate change,
Turkish cement industry: an application of data envelopment analy-
and population health in China. Environ Int 42:10–19
sis. Energy Efficiency. https://doi.org/10.1007/s12053-018-9764-z
Dockery DW, Pope CA (1994) Acute respiratory effects of particulate air Keene A, Deller SC (2015) Evidence of the environmental Kuznets’
pollution. Animal Review of Public Health 1(5):107–132 curve among US counties and the impact of social capital. Int Reg
Dockery DW, Schwartz J, Spengler D (1993a) Air pollution and daily Sci Rev 38(4):358–387
mortality: associations with particulates acid aerosols. Environ Res Knox EG, Gilman EA (1997) Hazard proximities of childhood cancers in
59:362–373 Great Britain from 1953-80. J Epidemiol Commun Health 51:151–
Dockery DW, Pope CA, Xu X, Spengler JD, Ware JH, Fay ME, Ferris 159
BG, Speizer FE (1993b) An association between air pollution and Koh D-H, Kim T-W, Jang S-H, Ryu H-W (2011) Cancer mortality and
mortality in six U.S. cities. New England Journal of Medicine 329: incidence in cement industry workers in Korea. Saf Health Work 2:
1753–1759 243–249
Dolgner R, Brockhaus A, Ewers U, Wiegand H, Majewski F, Soddeman Koh D-H, Kim T-W, Jang S-H et al (2013) Dust exposure and the risk of
H (1983) Repeated surveillance of exposure to thallium in a popu- cancer in cement industry workers in Korea. Am J Ind Med 56:276–
lation living in the vicinity of a cement plant emitting dust contain- 281
ing thallium. Int Arch Occup Environ Health 52:79–94 Krolzig HM (1996) Statistical analysis of Cointegrated VAR processes
Edwards P (2017) Global Cement Top 100 Report 2017 – 2018. https:// with Markovian regime shifts. SFB 373 discussion paper. No:25/
www.globalcement.com/magazine/articles/1054-global-cement- 1996
top-100-report-2017-2018. Accessed 2/2/2019 Krolzig HM (1997) Markov switching vector autoregressions: modeling,
Eom SY, Cho EB, Oh MK, Kweon SS, Nam HS, Kim YD, Kim H (2017) statistical inference, and application to business cycle analysis.
Increased incidence of respiratory tract cancers in people living near Springer Verlag
Portland cement plants in Korea. Int Arch Occup Environ Health 90: Legator MS, Singleton CR, Morris DL, Philips DL (1998) The health
859–864 effects of living near cement kilns; a symptom survey of Midlothian,
Fallahi F (2011) Causal relationship between energy consumption (EC) Texas. Toxicol Ind Health 14:829–842
and GDP: a Markov-switching (MS) causality. Energy 36(7):4165– Lei Y, Zhang Q, Nielsen C, He K (2011) An inventory of primary air
4170 pollutants and CO2 emissions from cement production in China,
Franklin M, Koutrakis P, Schwartz P (2008) The role of particle compo- 1990–2020. Atmos Environ 45:147–154
sition on the association between PM2.5 and mortality. Leone V, Cervone G, Iovino P (2016) Impact assessment of PM10 ce-
Epidemiology 19:680–689 ment plants emissions on urban air quality using the SCIPUFF dis-
Garcia-Perez J, Morales-Piga A, Gomez-Barroso D, Tamayo-Uria I, persion model. Environ Monit Assess 188:499
Pardo Romaguera E, Lopez-Abente G, Ramis R (2017) Risk of bone Lin B, Zhang Z (2016) Carbon emissions in China's cementindustry: a
tumors in children and residential proximity to industrial and urban sector and policy analysis. Renew Sust Energ Rev 58:1387–1394
 Environ Sci Pollut Res

Lopez-Cima MF, Garcıa-Perez J, Perez-Gomez B, Aragones N, Lopez Cardiovasculardisease(CVD) Cardiopulmonary Disease Lung
Abente G, Pascual T, Tardon A, Pollan M (2013) Lung cancer risk canCVD). Chapel Hill, NC,, s.n.
associated with residential proximity to industrial installations: a Schwartz J (2000) Harvesting and long term exposure effects in the rela-
spatial analysis. Int J Environ Sci Technol 10:891–902 tion between air pollution and mortality. Am J Epidemiol 151:440–
Lu F, Dongqun X, Cheng Y et al (2015) Systematic review and meta- 448
analysis of the adverse health effects of ambient PM2.5 and PM10 Schwartz J (2002) Particulate air pollution and chronic respiratory dis-
pollution in the Chinese population. Environ Res 136:196–204 eases. Environ Res 62:7–13
Magnani C, Leporati M (1998) Mortality from lung cancer and popula- Selden TM, Song D (1994) Environmental quality and development: is
tion risk attributable to asbestos in an asbestos cement manufactur- there a Kuznets curve for air pollution? J Environ Econ Environ
ing town in Italy. Occup Environ Med 55:111–114 Manag 27:147–162
Magnani C, Dalmasso P, Biggeri A, Ivaldi C, Mirabelli D, Terracini B Shafik N, Bandyopadhyay S (1992) Economic growth and environmental
(2001) Increased risk of malignant mesothelioma of the pleura after quality: time series and cross-country evidence, background paper
residential or domestic exposure to asbestos: a case–control study in for World Development Report. The World Bank, Washington
Casale Monferrato, Italy. Environ Health Perspect 109(9):915–919
Sims C, Zha T (1998) Bayesian methods for dynamic multivariate
Maier H, Gewelke U, Dietz A, Thamm H, Heller WD, Weidauer H
models. Int Econ Rev 39:949–968
(1992) Laryngeal cancer and occupation: results of the Heidelberg
Smailyte G, Kurtinaitis J, Andersen A (2004) Mortality and cancer inci-
laryngeal cancer study. HNO 40(2):44–51
dence among Lithuanian cement producing workers. Occup
Maier H, Tisch M, Dietz A, Conradt C (1999) Construction workers as an
Environ Med 61:529–534
extreme risk group for head and neck cancer. HNO 47(8):730–736
Mandal SK, Madheswaran S (2010) Environmental efficiency of the Stern D, Common M, Barbier E (1996) Economic growth and environ-
Indian cement industry: an interstate analysis. Energy Policy mental degradation: the Environmental Kuznets Curve and sustain-
38(2):1108–1118 able development. World Dev 24(27):1151–1160
McDowall ME (1984) A mortality study of cement workers. Br J Ind Teller P et al (2000) Use of LCI for the decision-making of a Belgian
Med 41:179–182 cement producer: a common methodology for accounting CO2
Mehraj SS, Bhat GA, Balkhi HM, Gu T (2013) Health risks for popula- emissions related to the cement life cycle.. s.l., s.n
tion living in the neighborhood of a cement factory. Afr J Environ USGS (2017) Mineral commodity summaries 2017, s.l.: U.S. Geological
Sci Technol 7(12):1044–1052 Survey
Meo SA, Al-Drees AM, Al Masri AA, Al Rouq FA, Azeem MA (2013) van Donkelaar A, Martin RV, Brauer M, Kahn R, Levy R, Verduzco C,
Effect of duration of exposure to cement dust on respiratory function Villeneuve PJ (2010) Global estimates of ambient fine particulate
of non-smoking cement mill workers. Int J Environ Res Public matter concentrations from satellite-based aerosol optical depth: de-
Health 10(1):390–398 velopment and application. Environ Health Perspect 118:847–855
Mosca S, Benedetti P, Guerriero E, Rotatori M (2014) Assessment of Vestbo J, Rasmussen FV (1990) Long-term exposure to cement dust and
nitrous oxide emission from cement plants: real data measured with later hospitalization due to respiratory disease. International
both Fourier transform infrared and nondispersive infrared tech- Archives of Occupational and Environmental Health 62(3):217–220
niques. J Air Waste Manag Assoc 64:1270–1278 Wang Y, Zhu Q, Geng Y (2013) Trajectory and driving factors for GHG
Musti M, Pollice A, Cavone D, Dragonieri S, Bilancia M (2009) The emissions in the Chinese cement industry. J Clean Prod 53:252–260
relationship between malignant mesothelioma and an asbestos ce- WHO (2018) WHO Global Ambient Air Quality Database. https://www.
ment plant environmental risk: a spatial case–control study in the who.int/airpollution/data/en/. Accessed 03/01/2019
city of Bari (Italy). Int Arch Occup Environ Health 82:489–497 Xi F et al (2016) Substantial global carbon uptake by cement. Nat Geosci
NRMCA (2012) Concrete CO2 fact sheet,, National Ready Mixed 9:1–8
Concrete Association, No. PCO2 Xu P, Chen Y, Ye X (2013) Haze, air pollution, and health in China.
Peters S, Thomassen Y, Fechter-Rink Y, Kromhout H (2009) Personal Lancet 382(9910):2067
exposure to inhalable cement dust among construction workers. J Xu J-H, Fleiter T, Fan Y, Eichhammer W (2014) CO2 emissions reduc-
Environ Monit 11(1):174–180 tion potential in China’s cement industry compared to IEA’s Cement
Pope, C.A, Burnett, R.T., Thun, M.J., Calle, E.E., Krewski, D., Ito, K. and Technology Roadmap up to 2050. Appl Energy 130:592–602
Thurston, G.D. (2002). Lung Cancer, cardiopulmonary mortality,
Yang CY, Huang CC, Chiu HF, Chiu JF, Lan SJ, Ko YC (1996) Effects of
and long-term exposure to fine particulate air pollution. J Am Med
occupational dust exposure on the respiratory health of Portland
Assoc 287:1132–1141
cement workers. J Toxicol Environ Health 49:581–588
Rachiotis G, Drivas S, Kostikas K et al (2012) Respiratory tract mortality
in cement workers: a proportionate mortality study. BMC Pulm Med Yhdego M (1992) Epidemiology of industrial environmental health in
12:30 Tanzania. Environ Int 18(4):381–387
Raffetti E, Treccani M, Donato F (2019) Cement plant emissions and Zou L, Ni Y, Gao Y, Tang F, Jin J, Chen J (2018) Spatial variation of
health effects in the general population: a systematic review. PCDD/F and PCB emissions and their composition profiles in stack
Chemosphere 218(2019):211–222 flue gas from the typical cement plants in China. Chemosphere 195:
Rohde R, Muller R (2015) Air pollution in China: mapping of concen- 491–497
trations and sources. PLoS One 10(8):1–15 -60000-50000-40000-
30000-20000-10000 0 10000 20000 2020BL 2020EEPCP Publisher’s note Springer Nature remains neutral with regard to jurisdic-
2020EEPTP 2030BL 2030EEPCP 2030EEPTP tional claims in published maps and institutional affiliations.