Australian Journal of Basic and Applied Sciences, 9(1) January 2015, Pages: 272-282
AENSI Journals
Australian Journal of Basic and Applied Sciences
ISSN:1991-8178
Journal home page: www.ajbasweb.com
Determinants of Poverty and Pollution: An Application of a Model of Simultaneous
Equations Units in Brazil
1
André Luiz Marques Serrano, 2Marcelo Driemeyer Wilbert, 1Carlos Rosano Peña, 1Patrícia Guarnieri dos Santos, 1Luiz Medeiros de Araujo
Neto
1
University of Brasília. Department of Business Administration. Faculty of Economics, Business Administration and Accounting.70910-900.
Brasília. Brazil
2
University of Brasília. Department of Accounting. Faculty of Economics, Business Administration and Accounting.70910-900. Brasília.
Brazil
ARTICLE INFO
Article history:
Received 19 September 2014
Received in revised form
19 November 2014
Accepted 22 December 2014
Available online 2 January 2015
Keywords:
Income. Poverty. Industrial Pollution
ABSTRACT
Background: This article offers empirical evidence regarding one of the oldest
controversies in the economics literature: interfaces between income level and the
environment. In spite of the significant debate about possible causal relations among
income, pollution and environmental degradation, it has been dominated by
oversimplifications of the complex realities involved. They allege, at one extreme, that
accelerated population growth is the cause of poverty which is seen as cause of
environmental degradation. At the other extreme, the concentration of wealth in the
hands of a small percentage of the population is alleged to be the cause of an immense
demand of natural resources. Such simplifications have survived for many years
because of the scarcity of empirical work testing those hypotheses. Objective: The
present empirical research was designed to test the relationship among income levels,
poverty, inequality and environmental degradation. Multiple linear regression analysis
models were used to handle the data gathered from secondary Brazilian sources.
Results: In spite of most of the robust results obtained, there is persistent evidence of a
considerable conceptual hiatus to be overcome if any robust understanding is to be
achieved regarding the correlations that exist among income level, income distribution
and the use of natural capital by society. Conclusion: Poverty may be both a cause and
a consequence of high level of environmental degradation just as income distribution
inequality may lead to the coexistence of both degradation and conservation of natural
capital.
© 2015 AENSI Publisher All rights reserved.
To Cite This Article: André Luiz Marques Serrano, Marcelo Driemeyer Wilbert, Carlos Rosano Peña, Patrícia Guarnieri dos Santos, Luiz
Medeiros de Araujo Neto, Determinants of Poverty and Pollution: An Application of a Model of Simultaneous Equations Units in Brazil.
Aust. J. Basic & Appl. Sci., 9(1): 272-282, 2015
INTRODUCTION
Generally speaking, developing countries are characterized by high levels of poverty and social inequality.
Such countries are faced with the historical challenge of confronting social injustice that has excluded, and still
excludes a considerable part of the population from access to a minimum standard of life and well-being. In
recent decades, there has been a perception that those problems faced by developing countries are correlated to
the degradation of their natural resources and in some cases, to increases in pollution levels (STERN, 1998).
That proposition has motivated a quest for a greater understanding of the interfaces among poverty,
pollution and environmental degradation. It should be pointed out that economic analysis occupies itself
primarily with addressing the problem of how best to allocate increasingly scarce resources and with their most
efficient alternative use, in an effort to maximize social well-being. In developing countries, the economic wellbeing of many families depends directly on the quantity and the quality of natural resources and the environment
(CHIMELI, 2007). For 25 years the relation between income levels and concentration and the use of natural
resources has been subordinated to the concept of pollution control. The outreach of that concept has led some
analysts, as Cole (2004) in particular, to the simplistic belief that achieving pollution control will also lead to
reductions in poverty and social inequality levels.
Against that background, this article aims to investigate, in all 26 Brazilian states and the Federal District,
the effects of the relation between income level and industrial pollution on the poor population. To do so, a
system of simultaneous equations was estimated relating poverty determinant factors to an industrial pollution
Corresponding Author: Luiz Medeiros de Araujo Neto, University of Brasília. Department of Business Administration.
Faculty of Economics, Business Administration and Accounting.70910-900. Brasília. Brazil
E-mail: medeiros2002@hotmail.com
273
Luiz Medeiros de Araujo Neto et al, 2015
Australian Journal of Basic and Applied Sciences, 9(1) January 2015, Pages: 272-282
factor. The methodology was based on the one proposed by Griliches (1977) who first perceived the need to
elaborate a system of simultaneous equations that would make it possible to estimate an indeterminate
endogenous variable.
In that context, over the last few years there has been a significant interest in the relation between poverty
and environmental pollution. In the academic literature various studies have focused on the possible relation of
income levels to people’s increasing exposure to pollutants being discharged into the atmosphere (PERSSON
and TABELLINI, 1994; CARD, 2001; DALY, 2002; COLE, 2004; PERRY, 2006; MURAD, HASHIM and
MUSTAPHA, 2010).
Just as an example, it is worth mentioning a study conducted by Salama (2006) that highlighted the
importance of the question of environmental sustainability for a country’s development and warned about the
lack of information on polluting emissions and their consequences for human health
This article is organized as follows: section 2 sets out an analysis of how income and industrial pollution
affect poverty levels in the 26 Brazilian states and the Federal District. Section 3 presents the research methods
and procedures used; the strategy was to elaborate an analysis based on the use of simultaneous equations to
estimate a poverty and pollution system (model) and applying the Generalized Method of Moments to panel
data. In section 4 there is the discussion of the results which reveal that all the variables considered in the
analysis are in compliance with economic theory and also the existence of a positive significant effect on
poverty associated to the states’ participation in the Brazilian GDP whereby growth in State GDP is associated
to increased industrial pollution levels but also to a lower unemployment rate. Section 5 sets out the final
remarks.
A Reminder of the Possible Relationship between Income and Pollution:
The quest for economic development presupposes that there will be economic growth and improvement in
individuals’ quality of life (SWAN, 1956). Apart from changes in the composition of the gross product and in
the allocation of resources along different sectors of the economy, a society’s economic development ought to
achieve reductions in poverty, unemployment and social inequality, as well as improvement in health, nutrition
and housing standards. In the last five decades yet another objective has been associated to a society’s economic
development, namely, a reduction in the deterioration of its natural capital (TAROZZI and DEATON, 2009).
Thus natural resources represent the stock of potential wealth on which humanity depends to generate its
production and income and to raise its level of well-being. Accordingly, economists have stressed the
importance of an efficient allocation of natural resources to maximize the population’s social well-being as the
basis for economic, social and environmental relations. However, that natural capital is becoming increasingly
scarce due to population growth, and the accelerated rhythm of wealth generation and accumulation. In their
efforts to achieve economic development, human societies are expanding primary production processes and
diffusing the process of industrialization (SHAFIK and BANDYOPADHYAY, 1992; KUBRUSLY, 2011).
The use of natural resources to increase economic growth may have negative effects, or externalities, in the
form of increase in pollution (DALY, 2002). Those negative externalities are not captured or reflected by the
price system, given the technical difficulties that exist to determine exclusive or rival rights to resource use.
There is therefore no possibility of establishing exchange relations among those rights that would ensure the
optimum use of natural resources. In such cases, the economic value being practiced in the market usually
underestimates the cost of opportunity and thus natural resources are overused causing inefficiency and
environmental degradation.
While internalizing environmental externalities does increase the efficiency, the gains are perceived
differently by the economic agents and they are also dispersed over time so that they affect the intra and intertemporal income distribution. Hence, economic theory proposes that correcting that market failure should
involve defining property rights in such a way that exchanging them via the market establishes a balanced price
that effectively represents their social costs (BRUYN, 2000; TEIXEIRA and PEREIRA, 2011).
In his study of pollution, environmental degradation and urban poverty in Brazil, Mueller (1993) argued
that most of the problems related to urban poverty and environmental degradation can be can be explained as
being the result of unequal economic development.
Degradation may be highly concentrated in the consumption patterns of the high income groups, who have
high consumption levels of durable goods, electricity and fresh water, as well as high levels of waste and
sewage generation and accumulation. On the other hand, Datt and Ravallion (1992) found inadequate sanitation,
overloading and congestion of the sanitation system, degradation of the land surrounding sewage treatment
installations and the presence of diseases associated to the inadequate basic sanitation services, especially
among people in low income brackets.
Lower income levels may also lead to more pollutant consumer patterns inducing more environmental
degradation (MURAD, HASHIM and MUSTAPHA, 2010). When the rate of increase of environmental
degradation surpasses the income growth rate, total degradation may increase, accompanying increased
consumption levels. Even though, in the case of most of the environmental problems, the technological process
274
Luiz Medeiros de Araujo Neto et al, 2015
Australian Journal of Basic and Applied Sciences, 9(1) January 2015, Pages: 272-282
affects the intensity of degradation of all social classes indiscriminately, thereby reducing the possibility of
imposing a threshold income level, the standard and quantity of consumption in each income bracket group are
most certainly varying and directly affecting their environmental impact (ALESINA and RODRIK, 2004).
Nevertheless, income constraints drastically reduce poor people’s capacity to do any defensive spending
against the negative effects of degradation such as paying for medical care or improving their housing
conditions. In other words, the poor probably have to meet a proportion of the environmental costs that is
greater than the proportion of the overall population they represent, that is to say they may be consuming the
degradation of the rich. Thus economic growth and increasing industrialization in the developing economies
have led to an intensification of polluting activities (MOLINA and RAO,2010).
Against that background it must be stressed that such a complex situation cannot be oversimplified. There
are other factors in it which must be considered in any attempt to gain an understanding of the interfaces
between poverty and environmental degradation. For example, the absence of people’s empowerment to
participate in political and decision making processes that affect their well-being is increasingly identified as
one of the cause of the poorer segments of society living on increasingly degraded environments (BESLEY,
BURGESS and VOLART, 2005).
When attention is focused on the quality of the environment, however, the question of its relation to growth
is not a consensus. Some researchers like Georgescu-Roegen (1971), Bérgolo et al. (2010) and Cechin and
Veiga (2010) initially proposed the hypothesis that the higher a country’s income level was, the greater would
be the extent of environmental degradation through either more pollution or more intensive use of natural
resourses. On the other hand, authors like Beckerman (1993), Lopez and Serven (2004), as well as the study
from the World Commission on Environment and Development Report (WORLD BANK, 1992), argue that
countries with higher income levels tend to reduce environmental degradation as a result of the use of less
pollutant technology and the society`s effort to preserve the environment and, therefore, firmly assert that the
best way for a country to improve the quality of its environment, in the long run, is for it become wealthier.
Not all researchers agree that there is a direct link among poverty, pollution and environmental degradation.
For example Murad, Hashim and Mustapha (2010) state that poverty plays an uncertain or minor role in regard
to pollution and environmental degradation. That is because the relation between poverty and the environment is
measured by institutional, socio-economic and cultural factors, whereas environmental degradation, in areas of
endemic poverty, is brought about by the effects of poor management of macro-economic and institutional
policies. To those authors, an improved management of those factors would lead such poor communities to
value the environment. In the same way, Wei (2011) asserts that institutional measures must be taken to permit
poor communities to improve their resistance in the face of economic shocks and growing environmental risks.
In the light of all the discussions it is not clear whether there is a negative or positive relation between
national income growth and the quality of the environment. If it is plausible that environmental quality
deteriorates with each product unit produced, it is equally plausible that there should be a decline in the
environmental deterioration rate when a certain higher income threshold is attained (BHATTACHARYA and
LUECK, 2009).
Fonseca and Ribeiro (2005) and Texeira and Bertella (2010) state that around the year 1970 there was a
generalized belief, that a nation’s economic growth was largely responsible for environmental problems.
According to Deacon and Norman (2004), beginning in the 1990s, economists started to consider the opposite
view, that technological innovation, improved education, economic progress and the improvement of institutions
resulting from economic growth would in fact induce less environmental degradation. An example of this
changing view is Jorgenson and Wilcoxen (2007), who believe that a policy ensuring economic growth lead to
environmental protection. Nonetheless, that idea is contested by the works of researchers like Margulis (1992)
and Clark (1996), who feel that economic growth performs indiscriminately in relation to protecting the
environment.
In that context, the production process normality generates some negative externalities regardless of how
much clean production technology is introduced. Reaction against those negative externalities associated to
production (and consumption) tends to be positively related to income levels (AFONSO et al., 2011). In the
same vein, Aghion, Caroli and Peñalosa (1999) stress that higher income levels give rise to environmentally
cleaner production and consumption patterns, thereby introducing a technological trajectory of less intense
degradation and that whenever the rate of reduction of that intensity becomes greater than the annual income
growth rate, then total environmental degradation goes down too, in spite of increased production and
consumption levels.
Nevertheless, the effects of pollution are not neutral in terms of the social groups it affects. Gürlük (2009)
shows how economic costs associated to urban pollution tend to be greater among the poorer strata of the
population. Given that the poor have fewer means of meeting the expense of defensive spending against
pollution, they tend to suffer more from its effects. That inequality in the distribution of environmental pollution
among the different income level groups also occurs in greater spatial spheres.
275
Luiz Medeiros de Araujo Neto et al, 2015
Australian Journal of Basic and Applied Sciences, 9(1) January 2015, Pages: 272-282
It is evident that poorer countries find it more difficult to adopt clean technology due to the high costs
involved. That same difficulty applies to the management of post-consumption consequences (COLEA and
FREDRIKSSON, 2009). The same argument can be applied to different geographic spaces in Brazil. The fact is,
however, that there is a generalized absence of statistics on pollutant emissions and their impacts on social
segments in Brazil. That absence makes it difficult to conduct a systematic analysis of the Brazilian industrial
sector’s environmental performance and its effects on the population’s health in the various regions of Brazil. It
is known, however, that according to the Brazilian Geography and Statistics Institute (InstitutoBrasileiro de
Geografia e Estatística – IBGE) (IBGE, 2010), large urban areas put great pressure on natural resources due to
the absence of basic sanitation, the low quality of housing, the difficulty of accessing public health services and
other problems stemming from the low income levels of our population (MOLINA and RAO, 2010).
One of the facts that has aggravated the consequences of environmental impacts for the populations of
Brazilian cities has been the extremely rapid transition, in less than a century, from being a predominantly rural
society to being an essentially urban one (Mueller, 1993). In 70 years the total population increased almost
fivefold while at the same time the rural population went down both in absolute numbers and as a percentage of
the total in comparison with the baseline reality of 1940. According to the preliminary data of the 2010
demographic census, 84% of Brazilians, that is 160.1 million people live in the urban milieu. In the year 2000,
81% of Brazilians lived in urban areas.
Again according to the 2010 demographic census, income inequality is still very high in the Brazilian cities
in spite of the trend to reduction observed over the last 10 years. Even though the average per capita income for
households in 2012 was 668 Brazilian reals, in that same year 25% of the population received 188 reals or less
and half of all Brazilians received 483; an amount lower than the official minimum salary at the time (622
reals).
It is important to reiterate that there is a generalized absence of statistics on pollutant emissions, making it
difficult to conduct a systematic analysis of the Brazilian industrial sector’s environmental performance and its
effects on the population’s health in the various regions of Brazil. Nevertheless it is possible to measure the
expansion of those industrial sectors with the greatest emissions potential and that has been done by the
indicators constructed by the IBGE’s Department of Industry. The building of those indicators is based
aggregating the physical industrial production in classifications reflecting the industrial activities’ power to
pollute (high, medium, low or negligible).
Methods and Procedures:
The simultaneous equations method is used in this article to estimate a model that relates poverty
determinant factors to an industrial pollution factor as shown in Equations 1 and 2. The estimators for the
equations’ coefficients were calculated using the methods of Ordinary Least Squares (OLS) and Fixed Effects
Least Squares (OLS-FE), and the Generalized Method of Moments (GMM).
The OLS method is used to estimate a classic linear regression model. In general terms and with due
observation of the Gauss-Markov theorem, the OLS parameters are the best unbiased linear estimators, that is to
say, the expected value for each estimator is equal to the parameter for the model that is to be estimated, so that
the OLS may be the most efficient. The OLS-Fixed Effects method is used as a way to minimize the effects of
any variability present in the OLS model. Thus the OLS-FE method is the one that offers the most plausible
estimators (HAYASHI, 2000).
GMM offers an efficient alternative for handling and solving the problems of endogeneity because it
performs well in the presence of homoscedasticity and leads to the orthogonality of the results with out of phase
models. Furthermore, the estimated coefficients are similar to those of the fixed effects model. In the case of
GMM estimations it is possible to include variables that are strictly endogenous, that is to say, not correlated to
error in any period.
On the basis of those reflections, the following two multiple linear regression models were estimated.
��� =
+
� �� � +
���� + �� + + ��
(1)
��� =
+
� �� � + ��� +
���� + � + + ��
(2)
Equation 1 represents the expansion of those industries with the greatest emissions potential used as a proxy
for pollution levels pl . The variable(xplit ) represents the exogenous factors that determine pollution:
urbanization rate (urban) and the unemployment rate (un). Equation 2 represent the level of poverty (pv) of the
population as a function of pl. The variable (xpvit )represents the factors determining poverty: Gini coefficient
(Gini), infant mortality rate (mr) and education (edu). The variable stategndit represents the participation of
each state (i) in the national gross domestic product (GND)
Parameters β and represent the consistent estimators. The terms ε and represent the matrixes of the
dummy variable for the states and Federal District (the individual effects). Parameters andγ stand for the
dummy temporal variables. The stochastic terms υand ξ mirror the variables that have been omitted or
incorrectly calculated (HAYASHI, 2000).
276
Luiz Medeiros de Araujo Neto et al, 2015
Australian Journal of Basic and Applied Sciences, 9(1) January 2015, Pages: 272-282
The aim of the estimations made using OLS and OLS-FE was to identify whether the coefficients are
predominantly significant. According to economic theory, it can be expected that any increase in the state
industrial GDP’s participation in the national GDP (stategnd) will lead to an increase in pollution levels. In
Equation 1, it is presumed that any increase in the urbanization rate (urban) will give rise to an increase in
pollution levels and that an increase in the unemployment rate (un), can be expected to be associated to a
reduction of the state GDP participation and to a reduction in pollution levels. An investigation will also made to
verify whether the GMM regression revealed any significant degree of relation between the unemployment rate
(un), the state GDP’s participation in the national GDP (stategdp),and the urbanization rate (urban) to the
pollution of the Brazilian states, that is, whether there is any positive or negative causality among the
explanatory variables defined by the model. An analysis will be made as to whether the variable urbanization
rate (urban) is, in turn, statistically significant when the other control variables are statistically relevant.
In Equation 2, it is expected that any increase in the state industrial GDP’s participation in the national GDP
(stategdp) will lead to a reduction in the poverty level (pv), that a rise in the urbanization rate (urban) lowers the
GDP participation and the pollution levels (pl), that years of schooling (edu) have a negative relation to the level
of poverty (pv) and that fetal mortality (mr) have a positive relation with population poverty (pv). A verification
of the GMM regression was made to check for any evidence of the importance of the unemployment rate (un),
the state industrial GDP’s participation in the National GDP (stategdp),the urbanization rate(urban), years of
schooling (edu),and number of fetal deaths (mr) for the poverty levels in the respective Brazilian states and
Federal District, that is, whether there is any positive or negative causality among the explanatory variables
defined by the model. An analysis will be made as to whether the variable urbanization rate (urban) is, in turn,
statistically significant when the other control variables are statistically relevant.
Table 1 displays the variables of Equations 1 and 2, the expected signs associated to the coefficients
according to economic theory, estimated using OLS, OLS-FE, and GMM in addition to the descriptions of the
variables and the data sources.
Equação 2
Equation 1
Table 1: Variables in the estimation of Equations 1 and 2 by OLS, OLS-FE and GMM and the expected signs of the coefficients.
Estimated Equation Parameters
Var
OLS
OLS
GMM
Descripton
Source
(EF)
pl
Dependent variable - Expansion of industrial sectors with
Ipeadata greatestemissions potential.
Pnad.
un
Unemployment rate.
Ipeadata.
urban
+
+
+
Urbanization rate.
Pnad.
stategdp
+
+
+
Participation of state GDP in National GDP.
Ipeadata.
yr1990
yr1991
yr1992
yr1993
yr1994
yr1995
yr1996
yr1997
yr1998
yr1999
yr2000
Cons
pv
pl
un
urban
stategdp
edu
mr
yr1998
yr1999
yr2000
yr2001
yr2002
yr2003
yr2004
yr2005
yr2006
cons
+
+
-
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
Pnad.
Pnad.
Pnad.
Pnad.
Pnad.
Pnad.
Pnad.
Pnad.
Pnad.
Pnad.
Pnad.
Dependent variable - Poverty
Expansion of industrial sectors with greatest emissions potential.
Unemployment rate.
Urbanization rate.
Participation of state GDP in National GDP.
Years of Schooling.
Number of fetal deaths.
IpeadataPnad.
IpeadataPnad.
Ipeadata.
Ipeadata.
Ipeadata.
Pnad.
Pnad.
Pnad.
Pnad.
Pnad.
Pnad.
Pnad.
Pnad.
Pnad.
Pnad.
+
For each dependent variable appearing in Table 1 there is an expectation according to economic theory set
out in Table 2.
277
Luiz Medeiros de Araujo Neto et al, 2015
Australian Journal of Basic and Applied Sciences, 9(1) January 2015, Pages: 272-282
Table 2: Expectations or expected patterns for the dependent variables.
Dependent
Economic Expectation
Variables
It is intended to confirm whether there is a positive relation between pollution and income in regard to the states and
pl
whether environmental conservation is directly related to the participation of the state’s industrial GDP in the national
GDP (stategdp).
It is intended to confirm whether there is a negative relation between poverty and pollution in the states and whether
pv
environmental conservation is directly related to the participation of the state’s industrial GDP in the national GDP
(stategdp).
The validity of the results obtained using GMM is verified subjecting them to the Hansen test and Arellano
and Bond’s (1991) serial correlation test. The serial correlation test was used to test the null hypothesis that no
serial correlation exists while the Hansen technique tests the hypothesis that the instrumental variables are
correlated with the error term and correctly excluded from the estimated equation. The idea is to verify whether
the statistics F shows itself to be globally significant and to what level of significance for all the regressions
carried out and also whether the value of R2 to be adjusted by the fixed effect model is capable of detecting
whether the explanatory variables can indeed evaluate the total poverty. The models were estimated using the
10.0 version of the statistics software Stata.
RESULTS AND DISCUSSIONS
The estimates of the coefficients of the proposed models (Equations 1 and 2) proved to be consistent with
economic theory expectations. The result of the F test applied to the estimates associated to Equation 1
generated by OLS (143.46) and the OLS-FE (65.42) showed that the model is globally significant.
In the estimations obtained using OLS-FE it can be seen that the coefficients are predominantly significant
(t statistics) to significance levels of 5%, 10% and 15%. (with the exception of dummy temporal variable for
the year 2000, statistically significant to the level of 10%). Furthermore, it can also be seen that the explanatory
variable participation of the state industrial GDP in the National GDP (stategdp) had a high statistical
significance (t=15.38). The adjusted R2 value of 0.70 shows that the equation’s set of determinant factors
(unemployment rate – des; urbanization rate – urban; state industrial GDP’s participation in the national GDP
(stategdp) explains the Brazilian state’s pollution to the extent of 70% (Table 3). It should be borne in mind that
the temporal effects have significant negative impacts throughout the temporal series (except in the year 1993
when the tendency changes but the trajectory is taken up once again in the following years) and there is an
overall reduction of poverty somewhere between 1.64% and 0.43%.
The validity of the results obtained using GMM is verified subjecting them to Hansen’s test and Arellano
and Bond’s (1991) serial correlation test. Hansen’s test indicates the probability of the instruments being 16.5%
(21.36) orthogonal and the Arellano-Bond correlation tests indicates a 79.5% (z = 0.26) probability of there
being no second order correlation (Table 1).
Table 3: Estimated parameters for Equation 1 (pollution) using OLS, OLS-FE and GMM method and the t Test (in brackets).
(1-OLS)
(2-OLS-FE)
(3-SYSTEM GMM)
un
0.0734***
-0.0437**
0.0856*
(3.70)
(-1.93)
(-1.89)
urban
0.0124*
0.0174***
0.123**
(1.84)
( 2.86)
(2.47)
stategdp
0.137***
0.148***
0.0953*
(12.86)
(15.38)
(1.82)
yr1990
-1.637***
(-5.68)
yr1991
-1.582***
(-5.86)
yr1992
-1.389***
(-5.58)
yr1993
-1.504***
(-6.17)
yr1994
-1.502***
(-6.18)
yr1995
-1.440***
(-5.92)
yr1996
-1.330***
(-5.52)
yr1997
-1.322***
(-5.52
yr1998
-0.844***
(-3.58)
yr1999
-0.609**
278
Luiz Medeiros de Araujo Neto et al, 2015
Australian Journal of Basic and Applied Sciences, 9(1) January 2015, Pages: 272-282
(-2.59)
-0.433*
(-1.84)
_cons
0.461
2.194***
(0.95)
(4.62)
R20.63 R2aj 0.70 F= 143.36 65.42 Arellano-Bond test for AR(2) z = 0.26 (0.795)
t statistics in parentheses. * p<0.10 ** p<0.05,*** p<0.01
yr2000
-6.602*
(-2.22)
A Equation 1 shows the existence of a positive relation between pollution and the relative incomes of the
states. Environmental conservation is directly related to the states’ relative participations in the national GDP
(stategdp). The GMM regression reveals the importance of the unemployment rate (un), the states’ participation
in the national GDP(stategdp) and the urbanization rate(urban) for the level of pollution generated by the
Brazilian states. Causality between the level of pollution and the unemployment rate is negative. Between
pollution level and all the others it is positive.
The variable urbanization rate, in turn, is statistically significant to a level of 5% while the other variables
are statistically relevant to a level of significance of 10%. The research made it possible to verify that the
urbanization rate had an impact to the level of 12.3% in relation to pollution. An annual increase of 1% in the
urbanization rate leads to an annual increase in pollution of 0.12%. Urbanization and industrialization can alter
the characteristics of the global climate which in turn can have direct effects on air pollution. As an example, the
formation of the so-called “Urban Heat Islands” is a phenomenon that occurs in urban areas and brought about
by pollution which causes temperatures to rise and air humidity to fall (SANTAMOURIS, 2001, COLEA and
FREDRIKSSON, 2009).
In the case of Equation 2, the statistical value F generated by the OLS method (344.83) and by the OLS-EF
method (215) showed that the model is globally significant to a significance level of 1%. The value obtained for
the adjusted R2 of the model estimated by the fixed effect method showed that the set of explanatory variables
explains 0.81% of the total variation in the level of poverty.
The GMM tools were shown to be consistent by the Arellan and Bond (1991) serial correlation test with a
value for p = 0.271 (z=1.10) and by Hansen’s test with p-val = 0.115 (24.16) at a significance level of 5%
(Table 4). Those values guarantee that the term of error is not serially correlated and that the moment conditions
are correctly specified. The Hansen’s test shows there is a 24.16% probability of orthogonality for the GMM
tools and the Arellan-Bond test shows an 11.5% probability of second order correlation’s not occurring.
The results show that there is an obvious correlation between poverty levels and income levels because the
coefficients are statistically significant to the level of 5% for all three estimation methods used (OLS, OLS-FE
and GMM). Exceptions appear among the dummies for temporal effects for the years 1998 and 1999 for the
Fixed Effect models. The explanatory variables pollution (pl), states’ participation in the national GDP
(stategdp) and schooling years (edu) are negative, while the coefficients for unemployment rate (un) and infant
mortality rate (mr) are positive. The signs obtained for the explanatory variables estimated by the three methods
are in accordance with the hypotheses formulated and with the majority of the analyses undertaken and
described in the specialized literature.
Table 4: Equation 2 parameters estimated by the OLS, OLS-FE and GMM methods and t Test (in brackets).
(1-OLS)
(2-OLS-FE)
(3-SYSTEM GMM)
pl
-0.0149***
-0.0210***
-0.0513***
(-5.24)
(-6.71)
(-2.43)
stategdp
-0.00662***
-0.00589***
-0.0484***
(-4.92)
(-4.57)
(-2.44)
edu
-0.640***
-0.669***
-0.466***
(-30.19)
(-32.19)
(-4.51)
un
0.0153***
0.0143***
0.0173***
(11.03)
(9.40)
(3.13)
mr
0.0000235***
0.0000274***
0226***
(4.24)
(5.06)
(2.75)
yr1998
-0.0106
(-0.66)
yr1999
0.00933
(0.56)
yr2000
-0.0433**
(-2.58)
yr2001
0.0336**
(2.01)
yr2002
0.0311**
(1.88)
yr2003
0.0695***
(4.11)
yr2004
0.0710***
279
Luiz Medeiros de Araujo Neto et al, 2015
Australian Journal of Basic and Applied Sciences, 9(1) January 2015, Pages: 272-282
(4.22)
0.0479**
(2.75)
yr2006
0.0459**
(2.68)
_cons
1.401***
1.460***
1.122***
(46.45)
(48.36)
(7.13)
R2 0.79 R2ajust 0.81 F= 344.83 215.00 Arellano-Bond test for AR(2) z = 1.10 (0.271)
Hansen test of over-identifying restrictions chi2(17)_ (24.16) 0.115 t statistics in parentheses. * p<0.10 ** p<0.05,*** p<0.01
yr2005
Years of Schooling is the explanatory variable with the greatest impact on poverty reduction indicated by
the fact that its coefficient values are higher and more significant than all the others for the three estimation
methods. The negative relation of this variable to poverty [(β = – 0.640 (OLS); β = – 0.669 (OLS-FE); β = –
0.466 (GMM)] means that the higher the level of education, the lower the poverty index. The poverty index goes
down consistently as the number of schooling years increases. As an example, based on GMM values it can be
seen each additional year of schooling, on average, reduces the poverty index by 0.47%.
Econometrics theory suggests that the economic interactions involved in the relations between pollution
levels and poverty levels are complex and so the most suitable means of studying that interdependence was to
make use of a set of simultaneous equations, as represented by equation (1) and equation (2), to identify any
causality among them. Such models are known as “simultaneous equations systems” because the exogenous
variables are determined simultaneously and the equations that are part of the system are estimated
simultaneously (GRILICHES, 1977).
According to economic theory, the explanatory variable pollution (polluting industries) has a negative
relation with the poverty index. The greater the amount of pollution generated by industrial production the more
the poverty index will go down. The negative sign of the estimated coefficient (– 0.0513 by the GMM) tells us
that an increase of 1% in the production of those polluting industries will tend to reduce the poverty index by
5.1% a year. That is indicative of poor people’s tolerance in regard to living and working in places with the
highest levels of pollution and environmental degradation. Here we have two sides of the same coin: the
company that pollutes also generates employment and income.
In none of the three regressions did the state GDP’s participation in the national GDP show itself to be
significant. In the equation estimated by the OLS method it can be seen that an increase of 1% in the states’
participation in the national GDP leads to a 4.8% reduction in the poverty index mirroring the poor population’s
improved living conditions.
Parallel to to that, a 1% increase in the unemployment rate implies a possible drop in the GDP value which
in turn would be accompanied by a reduction of pollution by 8.56%. The higher the unemployment rate, the
lower the pollution rate. More specifically, the unemployment leads to a reduction of pollution to the order of
0.09% a year. The negative relation between the unemployment figures and pollution is therefore directly
related to industrial production. The results reveal a strong negative association between the unemployment rate
(5.56%) and the participation of the state’s industrial GDP in the National GDP (9.53%) thus corroborating the
postulate of economic theory that positive variations in economic activity provoke negative variations in the
unemployment rate.
Generally speaking, the results indicate that there is a significant positive effect of the states’ industrial
GDP’s participation in the national GDP on the pollution index. In other words, whenever the state industrial
GDP participation increases, the there is an increase in the pollution indicators and consequently that may lead
to a reduction in the unemployment rate (improved income conditions).
Conclusions:
The links between income (level and distribution) and natural capital (use, degradation, pollution and
conservation) has challenged economists for decades. Even so the thinking on those connections has presented
chronic differences and has generated extremist stances that cannot be reconciled. This paper has verified that
one of the reasons for those differences and incompatibilities stems from the relative scarcity of empirical
verifications of the theoretical variations that have been put forward by environmental economists. The
interfaces of income (level and distribution) and natural capital (use, degradation and conservation) have been
relatively neglected topics in the empirical studies of economists interested in environmental problems. This
article has sought to close that gap by means of an empirical test of the arguments usually put forward in the
specialized literature.
The study reported here investigates the relations between poverty (low income level) and industrial
pollution in the Brazilian states and Federal District. This empirical verification sought to identify possible
causality relations by using a set of simultaneous equations whereby exogenous factors are determined
simultaneously and the equations that are part of the system are estimated simultaneously. The results suggest
that there is a negative relation between the levels of industrial pollution and poverty rates among the Brazilian
states and Federal District: the greater the amount of pollution stemming from industrial production, the more
280
Luiz Medeiros de Araujo Neto et al, 2015
Australian Journal of Basic and Applied Sciences, 9(1) January 2015, Pages: 272-282
the poverty level will go down. That is indicative of poor people’s tolerance in regard to living and working in
places with the high levels of pollution and environmental degradation. Here we have two sides of the same
coin: the company that pollutes also generates employment and income.
The models proposed by Equations 1 and 2 obtained using the simultaneous equations method confirm the
correlation that existed between income level, pollution levels and poverty levels in the Brazilian states and
Federal District for the period 1990 to 2008. Against that background, a review of the theoretical elements
present in the discussions on income level, pollution and poverty clearly reveal the complexity involved in
trying to fully apply them to the conflicts that are inherent to the inter-relations between the social sphere and
the environmental sphere. The analysis of the data on the Brazilian states shows that the variables associated to
income may be positively related to some of the environmental variables.
In light of the proposed objective of examining possible correlations among variables associated to the
concept of pollution and poverty - social indicators such as schooling years, urbanization rate and others
mentioned above – it was found that they were, in general, correlated to economic and environmental aspects.
That underscores the need to create mechanisms that will foster public policies that do not set priority on the
states’ economic growth alone, but above all, on improving the social indicators, especially in regard to
improved income distribution patterns, pollution control and access to education, all of which are of
fundamental importance to ensure that the environment is respected and that future generations will not have to
pay the price.
There are various comments that can be made on the basis of the test results set out in this article. One is
that economists in general and environmental economists in particular have neglected crucial aspects of the link
between income (level and distribution), the environment and pollution processes associated to economic
growth. There are various reasons for that negligence. Two of them can be derived from our research: one is
conceptual and the other stems from the limitations of official statistics.
The conceptual reason suggests that, in the course of the last few decades, there has been little evolution in
terms of analyzing the links between income level, inequalities in income distribution and the use of natural
capital. It was found that many analysts view the environment as a kind of “luxury goods”; a view that was
popularized by the World Bank’s 1992 World Development Report which suggested that there was an empirical
relationship between per capita GDP and concentrations of industrial pollutants. The presumption was that it
will always be possible to recuperate the stock in the future because natural capital is capable of recuperating
itself. According to Fare et al. (2004), biodiversity studies have shown that to be a false assumption: the
existence of ecological thresholds means that damage to ecosystems may be irreversible.
This study sets out numerous observations regarding empirical research involving income, pollution and the
use of natural resources (capital). Its outstanding feature is that it shows that the creation of wealth may collide
with environmental conservation and may co-exist with locations whose poverty is a current realty. There can be
no doubt that the enrichment of our conceptual framework in regard to the theme of this article is an essential
component of any agenda for future research. Thus, the practical reason explaining why there is poverty,
namely, inequality, environmental degradation and the influence both of them have on natural capital is still not
on the economists’ research agenda and that is related to the quantitative and qualitative limitations of the
empirical databases for those variables, especially in a country’s less developed regions. While there is
admittedly a large number of sources of information on poverty and pollution, the same cannot be said of the
information available on income distribution or on the impacts that different social groups’ activities have on
natural capital. As long as that scarcity of data sources continues to be the rule in the reality of regions and/or
countries, certain hypotheses that have never actually been tested will continue to be held to be true.
Finally, it must be stressed that the main limitation to this work has been not making a comparison of the
results obtained with the classic approach of economics. It is worth stating that the empirical verification sought
to illuminate possible causal relations by means of a set of simultaneous equations whereby the exogenous
variables are determined simultaneously and the equations that are part of the system are estimated
simultaneously. The results suggest that there is a negative relation between the levels of industrial pollution and
poverty rates among the Brazilian states and Federal District: the greater the amount of pollution stemming from
industrial production, the more the poverty level will go down. That is indicative of poor people’s tolerance in
regard to living and working in places with the high levels of pollution and environmental degradation. Here we
have two sides of the same coin: the company that pollutes also generates employment and income. In that light
it is suggested that future studies should make a comparison between the two approaches.
REFERENCES
Afonso, L.E., P.C. Pereda, F. Giambiagi, S. Franco, 2011. O salário mínimo como instrumento de combate
à pobreza extrema: estariam esgotados seus efeitos? EconomiaAplicada, 15:559-593.
Aghion, P., E. Caroli, CG. Peñalosa, 1999. Inequality and Economic Growth: The Perspective of the New
Growth Theories. Journal of Economic Literature, 37(4): 1615-1660.
281
Luiz Medeiros de Araujo Neto et al, 2015
Australian Journal of Basic and Applied Sciences, 9(1) January 2015, Pages: 272-282
Alesina, A., D. Rodrik, 1994.Distributive Politcs and Economic Growth.QuaterlyJounal of Economics,
109(2): 465-490.
Arellano, M., S. Bond, 1991. Some Tests of Specification for Panel Data: Monte Carlo Evidence and an
Application to Employment Equations. The Review of Economic Studies, 58: 277-297.
Beckerman, W., 1993. The Environmental Limits to Growth: a fresh look. In: H. Giersch (Ed.). Economic
Progress and Environmental Concerns. Berlin: Springer.
Bérgolo, M., 2010. Vulnerability to Poverty in Latin America: Empirical Evidence from Cross-Sectional
Data and Robustness Analysis with Panel Data. Argentina: CEDLAS and CONICET.
Besley, T., R. Burgess, B.E. Volart, 2005. Operationalising Pro-Poor Growth: India Case Study.Department
of Economics of the London School of Economics, London.
Bhattacharya, H., D. Lueck, 2009. The role of property rights in determining the environmental qualityincome relationship. Ecological Economics, 68: 2511-2524.
Bruyn, S.M., 2000. Economic Growth and the Environoment. Dordrecht: Kluwer Academic Publishers.
Card, D., 2001. Estimating the return to schooling: Progress on some persistent econometric
problems.Econometrica, 69(5): 1127-1160.
Cechin, A.D., J.E. Veiga, 2010. A economia ecológica e evolucionaria de GeorgescuRoegan. Revista de
EconomiaPolítica, São Paulo, 30: 438-454.
Chimeli, A.B., 2007. Growth and the environment: Are we looking at the right data? Economics Letters, 96:
89–96.
Clark, C.W., 1996. Operational Environmental Policies.Environment and Development Economics, 1: 11013.
Cole, M.A., 2004. Trade, the pollution haven hypothesis and the environmental Kuznets curve: examining
the linkages, Ecological Economics, 48(1): 71-80.
Colea, M.A., P.G. Fredriksson, 2009. Institutionalized pollution havens.Ecological Economics, 68: 12391250.
Daly, H.E., 2002.Elements of Environmental Macroeconomics. In: Munasinghe, M. Macroeconomics and
the Environment. UK/USA: Elgar.
Datt, G., M. Ravallion, 1992.Growth and redistribution components of changes in poverty: a decomposition
with application to Brazil and India. Journal of Development Economics, 38(2): 275-295.
Deacon, R., C.S. Norman, Is the Environmental Kuznets Curve an empirical regularity? In: Halvorsen, R.;
Layton, F.D. (Eds.) 2006. Explorations in Environmental and Natural Resource Economics: essays in honor of
Gardner M. Brown, Jr. Massachusetts: Edward Elgar Publishing Limited. Cap., 5: 97-121.
Deacon, R., Norman, C.S., 2006. Is the Environmental Kuznets Curve an empirical regularity? In:
Färe R, Grosskopf S, Weber WL. Shadow prices and pollution costs in U.S. agriculture.Ecol Econ, 56: 89–
103.doi:10.1016/j.ecolecon.2004.12.022
Fonseca, L.N., E.P. Ribeiro, 2005. Preservação ambiental e crescimento econômico no Brasil. In: Encontro
de Economia da Região Sul, 7. Rio Grande do Sul.
Georgescu-Roegen, N., 1971. The Entropy Law and the Economic Process. Massachusetts: Harvard
University Press.
Griliches, Z., 1977. Estimating the returns to schooling: Some econometrics problems.Econometrica, 45(1):
1-22.
Gürlük, S., 2009. Economic growth, industrial pollution and human development in the Mediterranean
Region.EcologicalEconomics, 68: 2327–2335.
Gylfason, T., 2001. Natural Resources, Education, and Economic Development. European Economic
Review, 45(6): 847-859.
HAYASHI, F., 2000. Econometrics is published by Princeton University Press and copyrighted, by
Princeton. University Press
Instituto Brasileiro de Geografia E Estatística (IBGE). Diretoria de Pesquisas. Departamento de População
e Indicadores Sociais. Pesquisa Nacional de Saneamento Básico – 2000. Rio de Janeiro: IBGE, 2002. Available
in: <http://www.ibge.gov.br/home/estatistica/populacao/condicaodevida/pnsb/pnsb.pdf>.
____. Departamento de Estatística. Relatório dos Resultados do Censo Populacional, 2010. Rio de Janeiro:
IBGE,2011a.Available in: <http://www.censo2010.ibge.gov.br/resultados>.
____. Departamento de Estatística. Pesquisa Nacional por Amostra de Domicílios (Pnad). 1970 – 2010.
Brasília: IBGE,2011b. Available in:
<http://www.ibge.gov.br/home/estatistica/pesquisas/pesquisa_resultados.php?id_pesquisa=40>.
Instituto de Pesquisa Econômica Aplicada. Base de Dados Social (IPEADATA). Diretoria de Políticas
Regionais, Urbanas e Ambientais. Pesquisa sobre pagamento por serviços ambientais urbanos para gestão de
resíduos sólidos. Relatório de Pesquisa, maio 2010. Brasília: Ipea, 2010. Available in:
<http://agencia.ipea.gov.br/images/stories/PDFs/100514_relatpsau.pdf>.
282
Luiz Medeiros de Araujo Neto et al, 2015
Australian Journal of Basic and Applied Sciences, 9(1) January 2015, Pages: 272-282
Jorgenson, D., P.J. Wilcoxen, 2007.Environmental Regulation and U.S. Economic Growth.The
RandJournalofEconomics, 21(2): 314-340.
Kubrusly, L.S., 2011. A população ocupada e a renda no Brasil: encontros e desencontros. Economia e
Sociedade, 20: 567-600.
Lopez, H., L. Serven, 2004.The Mechanics of Growth-Poverty-Inequality Relationship.The World Bank.
Mimeo.
Margulis, S., 1992. Causas do Desmatamento da Amazônia Brasileira. Brasília: Banco Mundial.
Molina, I., J.N.K. Rao., 2010. Small Areas Estimation of Poverty Indicators.The CanadianJournal of
Statistics, 38: 369–385.
Mueller, C.C., 1993. Problemas ambientais de um estilo de desenvolvimento: a degradação da pobreza no
Brasil. Ambiente e Sociedade, 1(1): 81-101.
Murad, W., N. Hashim, N. Mustapha, 2010. Does poverty cause environmental degradation? Evidence from
waste management practices of the squatter and low-cost flat households in Kuala Lumpur.World Journal of
Science,Technology and Sustainable Development, 7(3): 275-289.
Persson, T., G. Tabellini, 1994. Is inequality Harmful for Growth? The American Economic Review, 84(3):
600-621.
Perry, G.E., 2006. Poverty Reduction And Growth: Virtuous And Vicious Circles. Washington: The World
Bank.
Salama, A., 2006. A Life Style TheoriesApproachforAffordableHousing in Saudi Arabia.
EmiratesJournalforEngineeringResearch, College of Engineering, UnitedArabEmiratesUniversity, UAE, 11(1):
67-76.
Shafik, N., S. Bandyopadhyay, 1992.Economic Growth and Environmental Quality: a time series and crosscountry evidence. Journal of Environmental Economics and Management, 4: 1-24.
Stern, D.I., 1998. Progress on the environmental Kuznets curve? Environmental and Development
Economics, 3(2): 173-196.
Swan, T.W., 1956. Economic growth and capital accumulation.Economic Record, 32(13): 334-361.
Tarozzi, A., A. Deaton, 2009.Using Census and Survey Data to Estimate Poverty and Inequality for Small
Areas.Review of Economics and Statistics, 91: 773–792.
Texeira, R.F.A.P., MA. Bertella, 2010. Curva de Kuznets Ambiental para o Estado de Mato Grosso:
Modelagem Espacial. In: Encontro Regional de Economia – Anpec Sul, 13, Porto Alegre. Available in:
<http://www.ppge.ufrgs.br/anpecsul2010/artigos/40.pdf>.
Teixeira, M.D.J., B.D. Pereira, 2011. Os resíduos sólidos e a renda per capita no Brasil: uma análise sobre a
Curva de Kuznets Ambiental. In: Congresso da Sociedade Brasileira de Economia, Administração e Sociologia
Rural, 49., Belo Horizonte. Anais do Congresso Sociedade Brasileira de Economia, Administração e Sociologia
Rural. Belo Horizonte: SOBER, 2011.
Wei, T., 2011. What Stirpat tells about effects of population and affluence on environment? Ecological
Economics, 72: 4-70.
World Bank, 1992. World Bank Development Report: Development and the Environment. New York:
Oxford University Press.