Interrelationships between income, health and the environment: extending the Environmental Kuznets Curve hypothesis

Ecological Economics 36 (2001) 513– 531 www.elsevier.com/locate/ecolecon

ANALYSIS

Interrelationships between income, health and the environment: extending the Environmental Kuznets Curve hypothesis Lata Gangadharan a,*, Ma. Rebecca Valenzuela b a

Department of Economics, Uni6ersity of Melbourne, Melbourne, Vic. 3010, Australia b Department of Economics, Monash Uni6ersity, Vic. 3145, Australia

Received 4 April 2000; received in revised form 19 September 2000; accepted 20 September 2000

Abstract This paper examines the link between the health indicators and the environmental variables for a cross-section of countries widely dispersed on the economic development spectrum. While environment and income are seen to have an inverted-U shaped relationship (Environmental Kuznets Curve (EKC) hypothesis), it is also well established that environment and health are positively related. Our study focuses on the implications of this for the relationship between health and income. In the early phases of income growth, the gains in health and the losses in environmental quality could cancel each other out and this challenges the idea that as incomes increase health would always improve. To empirically analyse these issues, we estimate a two-stage least squares model that focuses on the impact of income and the environment on health status, with environment being an endogenous variable. Our results show that the environmental stress variable has a significant negative effect on health status. At the same time, gross national product (GNP) levels are shown to vary positively with health status variables. We find that the health gains obtained through improved incomes can be negated to a significant extent if the indirect effect of income acting via the environment is ignored. Research findings in this regard would be a useful policy instrument towards maximising both the environmental and health gains that come with economic growth and development. © 2001 Elsevier Science B.V. All rights reserved. Keywords: Environmental stress; Health indicators; Income levels JEL classification: O11; Q25; C30

* Corresponding author. E-mail addresses: [email protected] (L. Gangadharan), [email protected] (M.R. Valenzuela). 0921-8009/01/$ - see front matter © 2001 Elsevier Science B.V. All rights reserved. PII: S 0 9 2 1 - 8 0 0 9 ( 0 0 ) 0 0 2 5 0 - 0

514

L. Gangadharan, M.R. Valenzuela / Ecological Economics 36 (2001) 513–531

1. Introduction Is economic growth part of the solution rather than the cause of environmental problems? This question has been raised very often in recent years as empirical evidence in support of the Environmental Kuznets Curve (EKC) hypothesis mount.1 The EKC (Grossman, 1995; Grossman and Kreuger, 1995) describes the relationship between declining environmental quality and income as an inverted-U, that is, in the course of economic growth and development, environmental quality initially worsens but ultimately improves with improvements in income levels. For instance, Torras and Boyce (1998) show that the level of air pollutants (sulphur dioxide and smoke) peak at a per capita income in the neighbourhood of US $4000, after which they start falling. One of the explanations for the EKC relationship is that the environment can be thought of as a luxury good. In the early stages of economic development, a country would be unwilling to trade consumption for investment in environmental regulation, hence environmental quality declines. Once the country reaches a threshold level of income, its citizens start to demand improvements in environmental quality and this leads to the implementation of policies for environmental protection and, eventually, to reductions in pollution. Increasing levels of pollution are thus strongly associated with both poor and developing economies, while declining levels of pollution are more commonly observed for their developed counterparts.2 Another explanation of the EKC hypothesis is that countries pass through technological life cycles, as they move from agriculturebased economies to service-based systems. As the service sector is associated with lower environmental impact, this transition from high polluting to low polluting technology leads to less environ-

1

The original Kuznets curve refers to an inverted-U shaped relationship between per capita income and inequality observed by Kuznets (1955). 2 For example, Grossman and Kreuger (1995) found evidence in support for the EKC hypothesis for 12 of the 14 air and water quality variables for a cross-section of countries.

mental stress. Hence, in the long run, pollution levels are expected to improve with incomes. This argument has been used to justify the pursuit of growth strategies that do not give due consideration to their effect on the environment. In this paper, we argue that this strategy is not justified and provide some evidence to support our case. We test the hypothesis that larger economic and social gains can be attained by an economy if the growth strategy adopted incorporates, rather than ignores, environmental concerns. To do this, we include health as the intervening variable in the analysis. While environment and income are seen to have an invertedU shaped relationship, it is also well established that health and environment are positively related. What does this imply for the relationship between health and income? It is possible that in the early phases of income growth, the gains in health and the losses in environmental quality cancel each other out and this challenges the idea that as incomes increase, health would always improve. In view of the above, we argue in this paper that the recorded health gains brought about by the improvement in income levels do not represent the total realisable health benefits from having higher per capita income. Without the appropriate environmental protection policies, damages to a country’s physical environment are incurred during the process of income growth and economic development. This negatively affects the health and well-being of individuals in the country and the aggregated impact could negate some of the health gains already derived, and hence dampen achievement levels in the health area. If we find that this argument has some empirical support, it would imply that development policies addressing environmental issues are effectively also addressing the health issues of the economy. In that case, policies that pursue economic development cannot afford to ignore environmental issues, particularly in the early phases of economic growth. We look at recent evidence from a cross-section of countries to determine if this is indeed the case. While there are some studies (for example, Cropper et al., 1997) that look at the incidences of air

L. Gangadharan, M.R. Valenzuela / Ecological Economics 36 (2001) 513–531

or water pollution-related diseases in a particular region or country, this is one of the first papers to analyse the link between the health indicators and the environmental variables for a cross-section of countries widely dispersed on the economic development spectrum. The major contributions of this paper are the following. 1. To establish the link between health and environment. While this link has been suggested before, to the best of our knowledge it has not been explored for different countries. 2. To explore the differential impact on health of several environmental stress variables. The standard practice is to focus on just one variable, normally CO2 emissions. We attempt to make our findings and conclusions more robust by including several other pollutants and environmental damage indicators in the analysis. 3. To analyse the shape of the EKC curve and the health relationship using alternative functional forms to determine which among them best fits the data. The rest of the paper is organised as follows. Section 2 surveys the related literature on the relationship between income levels and environmental stress and the link between income levels and health status. Section 3 describes the analytical framework and the estimation methodology used in the paper. Section 4 summarises the data used in the analysis. Section 5 discusses the results from the estimation, and Section 6 concludes.

2. Related literature survey Panayotou (1993), Selden and Song (1994) and Grossman and Kreuger (1995) presented initial evidence that some pollutants follow an invertedU shaped curve with respect to income. This was widely interpreted (for example, World Bank, 1992) to mean that the surest way to improve a country’s physical environment is to increase income levels. More recent work has focused on factors other than income as explanatory variables in analysing variations in environmental stress in different countries. Kaufmann et al.

515

(1998), Torras and Boyce (1998), Suri and Chapman (1998)and Agras and Chapman (1999) argue that the EKC’s previously estimated could be due to important missing variables. Kaufmann et al. (1998) stress the importance of spatial intensity of economic activity, Torras and Boyce (1998) explore the effects of social factors like civil rights, income inequality and education, while Suri and Chapman (1998)and Agras and Chapman (1999) find that trade-related variables and the price of energy have significant explanatory power. Most of the papers mentioned use linear or a log linear functional relationship between emissions and income. An exception is Galeotti and Lanza (1999), which studied relationships based on the gamma distribution. The current state of the research on the EKC is unable to conclude if the EKC hypothesis is confirmed or rejected. With regards to health, there exists a large literature that has analysed the relationship between income and health using cross-country data (for example, Gerdtham et al., 1992; Chakrabarti and Rao, 1999). A number of previous studies in this literature have found an economically and statistically significant and negative income elasticity of infant mortality rate (see for example, Flegg, 1982; Parpel and Pillai, 1986; Hill and King, 1992; Kakwani, 1993; Subbarao and Raney, 1995; Pritchett and Summers, 1996). Similarly, research on life expectancy and income has shown that there is a positive relationship between increases in income and life expectancy, with income elasticity of life expectancy estimated to be significant and positive (Preston, 1980; Hill and King, 1992). Most of these studies also control for other factors that affect health status such as the accessibility of health services and education levels of the population.

3. Analytical framework In this paper, we are interested in how the interplay between income and the environment affect the health outcomes of a population. Generally, it is assumed that health outcomes for a population improve as the economy grows and develops. Such improvements are facilitated by

516

L. Gangadharan, M.R. Valenzuela / Ecological Economics 36 (2001) 513–531

the rise in general standard of living, including improved access to educational opportunities and health services. One’s health is also seen as dependent on the quality of his or her physical environment — such as the amount of air pollution or the quality of drinking water. At the same time, the quality of a country’s physical environment is a result of certain growth factors in the economy. These include, for instance, the more intensive use of land, forest and water resources to increase overall economic production. Air pollution levels are also bound to increase as production levels rise. Increase in population numbers is another important factor in this context. The relationships discussed above are summarised in the following general model: Hi = f(Xi, Ei (Xi, Zi ), Wi )

(1)

which states that an economy’s health status (Hi ) depends on its level of economic growth (Xi ), the quality of its environment (Ei ) and other social factors (Wi ) including the provision and access to health facilities. Zi is used to denote the factors that determine the quality of the environment. Within this framework, we test the relevance of the EKC hypothesis, captured in the term Ei (Xi, Zi ), and how it impacts on the health outcomes for a country’s population. To empirically analyse these issues, the following econometric model is formulated for country i: Ei = i0 +i11Xi + i12X 2i +i13X 3i +i2Zi +ei

(2)

Hi = h0 + h1Xi +h2Wi +h3Ei +ui

(3)

where Ei refers to the overall level of environmental stress in the economy; Hi refers to health status of the population; Xi pertains to the country’s level of economic growth; Zi, are non-income variables that impact on the environment; Wi, are variables that directly influence health such as provision and access to medical facilities, etc.; ui, ei, are the error terms. Eq. (2) is the EKC, where the dependence of environmental quality on economic growth is represented in a cubic relationship. The inverted-U shaped EKC requires i11 to be positive and i12 to be negative. A cubic income term is added to test

the proposition from recent research that environmental quality tends to increase once again with extremely high incomes (De Bruyn et al., 1998). The upward bend of the Kuznets curve at the very high-income levels will be captured by the i13 term, which is expected to have a positive sign. The Zi term captures the effects of non-income variables such as population levels, literacy rates and income inequality, which are thought to significantly influence environmental outcomes. For example, countries with higher population densities are expected to suffer from greater environmental stress as there would be more people sharing the existing environmental resources. A country’s level of urbanisation is also thought to negatively impact on the environment, although it is possible that urbanisation can bring improved waste disposal and sanitation provisions for the urban areas, and hence mitigate the detrimental environmental effects of an increase in population. Further, there are strong grounds for believing that the education levels of the population are positively associated with better environmental quality, while is not clear in which direction the gap between the rich and the poor influences the environment. In previous research, increases in the inequality of income distribution have been associated with higher levels of pollution, as those who benefit from pollution-generating activities are better able to prevail against those who bear the costs (Torras and Boyce, 1998). However, this hypothesis has been challenged by Scruggs (1998), who suggests that it is impossible to make generalisations about the effect of income distribution on environmental degradation without having more information about the preferences of different income groups. Eq. (3) postulates that the population’s overall health and general well-being is dependent on three factors, the country’s level of economic growth; the availability and accessibility of medical facilities; and the quality of the country’s physical environment. Xi and Ei are the same variables defined in Eq. (2) while Wi captures all other variables thought relevant to health such as number of doctors and other medical workers, immunisation levels, as well as literacy rates and other population-related factors.

L. Gangadharan, M.R. Valenzuela / Ecological Economics 36 (2001) 513–531

The quality of the physical environment is included here as an endogenous variable. We expect that health indicators would show an improvement with better availability of health care services and with higher rates of child immunisation. Urbanisation being an indicator of modernisation could lead to improved health, as there might be easier access to sophisticated health facilities. However, it is also possible that the same could impact negatively on health quality as urbanisation very often leads to overcrowding, poor sanitation levels and hence more health ailments. We expect higher educational levels to improve health quality as education lowers the cost of information, and people with higher levels of education might have a better understanding of the value of public health infrastructure and are better able to locate and utilise these services. In particular, education has often been cited for its strong effect on reducing child mortality (for example, Mellington and Cameron, 1999). In the above model, the structural equations are clearly identified given that Ei and Hi are the only endogenous variables in the system while the rest (Xi, Zi and Wi ) are taken to be exogenous. The equations are then estimated using general twostage least squares estimation methods. The results obtained are subjected to a robustness test with regards to functional forms and to different assumptions made regarding the type and nature of the variables used. The coefficients from the estimated equations will indicate if environmental variables play an important role in improving health outcomes in a country.

4. Data The data used in this paper are obtained from the World Development Indicators (1998), compiled by the World Bank. Summary statistics of these data and the units of measurement used are presented in Table 1. There are a total of 51 countries included in the analysis, covering more than 70% of the world’s population (Appendix A).3 3 The countries in the sample are listed in the Table of Appendix A.

517

Twenty-two of the 51 countries (43%) come from the high-income OECD set, but the population composition is dominated by those in the low-income countries, with the inclusion of China and India in this group. There are a number of environmental stress variables that are available for analysis. The air pollutants for which data are available are carbon dioxide (CO2), sulphur dioxide (SO2), nitrogen oxides (NOx) and total suspended particulates (TSP).4 Commercial energy use (ENPC) is taken to be another environmental indicator. For water pollution, we use data on emission levels of organic pollutants (EMW) while data on deforestation rates (DEFRTE) have also been obtained from the World Bank. CO2 emissions and ENPC are very high for richer countries compared with low-income countries. Other air pollutants like SO2, NOx and TSP are significantly higher for low-income countries than for the high-income countries in the sample. This could be due to the fact that richer countries have already in place environmental regulations targeting these pollutants, while this has yet to be implemented for poorer countries. EMW and DEFRTE are much higher for low-income countries for similar reasons. The link of the environment with the country’s level of economic growth is analysed using the country’s gross national product (GNP) per capita, purchasing power parity, as the proxy variable for the latter.5 Population factors are controlled for the estimation through the coun-

4 The data for SO2, NOx and TSP were provided at the city level not the country level. We obtain the country level data using the city’s population as a proportion of the country’s population. 5 It is acknowledged here that gross national product is just one aspect of economic growth and development, and that other more broad based measures (for example, the Human Development Index or HDI) are available. We nonetheless use GNP in this analysis to facilitate comparison with previous EKC studies. It is noted that GNP is highly correlated with HDI values for the countries covered in this study (z= 0.82), therefore, results should be robust between these two variables. For more details on the importance of alternative measures of income and growth on the EKC relationship, see Munasinghe (1999).

518

Table 1 Variable means and summary statisticsa Variable

Code

All countries

Income groupb

High

Total number Share (%) Size of Total population number (in millions) Share (%)

Means Population density (number per km2) Urban population (%) GNP purchasing power parity p.c. Gini coefficient Life expectancy (years) Healthy life expectancy (years, 1997–1999) Physicians/ 100 000 (1993)

Medium

Low

Africa

Asia

Eastern Europe

Latin America

Middle East

OECD

51

23

16

12

3

9

8

7

2

22

100 4326

45 875

31 706

24 2745

6 83

18 2685

16 260

14 374

4 122

43 802

100

20

16

63

2

62

6

9

3

19

206

354

65

112

53.33

751.11

89.88

30.00

50.00

117

41

76

65

48

11.03

100.74

23.28

41.81

31.90

74

GNP

12 204

20151

7514

3228

3457

10 252

5579

7830

4110

18 733

GINI

36

32

43

37

52

39

27

53

38

32

LEXP

72

77

71

66

61

70

70

71

68

76

HLE

65

71

62

58

42

63

65

64

60

71

DOC

202

257

157

19

26

72

309

147

126

263

POPDEN

UPOP

L. Gangadharan, M.R. Valenzuela / Ecological Economics 36 (2001) 513–531

Totals Number of countries

Region

Table 1 (Continued) Variable

Code

All countries

Income groupb

High

Medium

Low

Africa

Asia

Eastern Europe

Latin America

Middle East

OECD

87

88

87

84

57

91

93

84

95

87

86

109

71

61

48

69

84

56

72

109

19

6

22

40

59

27

13

26

44

8

24

7

26

53

89

34

17

31

52

9

6.99

10.07

5.61

2.92

2.93

5.88

7.48

3.49

2.95

9.30

1.997

0.394

0.992

6.411

0.834

7.931

1.061

1.118

1.143

0.426

0.477

0.160

0.411

1.174

0.159

1.377

0.371

0.279

1.173

0.190

L. Gangadharan, M.R. Valenzuela / Ecological Economics 36 (2001) 513–531

Child IMM immunisatio n rate (% of all under 12 months) Gross EDU enrollment ratio (% of school-age childrenc) Infant IMR mortality rate (per 1000 live births) Child CMR mortality rate (per 1000 children under age 5) Carbon CO2 dioxide emission p.c.c Total TSP suspended particles emissionc Sulfur SO2 dioxide emissionc

Region

519

520

Variable

Code

All countries

Income groupb

High

Nitrogen NOx dioxide emissionc Emission of EMW organic water pollutantsc Commercial ENPC energy use p.c.c Deforestation DEFRTE rate (average% change, 1990–1995)

0.693

0.534

Region

Medium

0.558

Low

1.176

Africa

0.289

Asia

Eastern Europe 1.417

0.456

Latin America

Middle East

0.687

0.00

OECD

0.603

430

382

217

800

148

1092

217

216

150

336

2850

4437

1940

1020

869

2067

2852

1169

985

4144

0.19

−0.36

0.53

0.80

0.60

1.13

−0.09

0.76

0.90

−0.39

a Data were derived from World Development Indicators, WDI (1998), except for HLE and DOC which were taken from the World Health Organisation. All variables are for year 1996 unless otherwise indicated. Gini coefficients for a few countries (missing in the WDI) were obtained from Deininger and Squire (1996). b Denote x as the per capita income. This grouping thus classifies countries in the high income category if (x]$12 000), medium if ($45005xB$12 000) and low if (xB$4500). c EDU refers to secondary level (1995), CO2 expressed in metric tons (1995), TSP, SO2 and NOx in kg/m3 (1995), EMW in kg/day (1993), and ENPC in kg of oil equivalent (1995).

L. Gangadharan, M.R. Valenzuela / Ecological Economics 36 (2001) 513–531

Table 1 (Continued)

L. Gangadharan, M.R. Valenzuela / Ecological Economics 36 (2001) 513–531

521

disability adjusted life expectancy (DALE) and it summarises the expected number of years to be lived in what might be termed the equivalent of ‘full health’.7 For all the countries in the sample, the average rate of infant mortality is 19 deaths per 1000 live births. However, this rate shoots up to 59 deaths per 1000 live births in Africa and 44 deaths per 1000 births in the Middle East. The African rate is extremely high, and far exceeds the average levels computed for all the other regions — more than double the rate for Asia and Latin America and seven times higher than that of the advanced countries of Europe and North America. Income inequality, measured using Gini coefficients, was supplemented from Deininger and Squire (1996). The data show that income inequality is highest in Latin America and Africa, and lowest in Europe and North America.8 Table 2 presents the correlation between the environmental indicators relating to air pollution. CO2 emissions and ENPC are noted to have a very high positive correlation, while ENPC and CO2 emissions have a weak negative relationship with the emission levels of SO2, TSP and NOx, which are local pollutants. This may be a bit

try’s population density levels (POPDEN) and the overall degree of urbanisation (UPOP). Secondary school enrollment ratios (EDU) proxy for the literacy rate in the country, and the gap between the rich and the poor as measured by the Gini coefficient (GINI) is used as the indicator of income inequality in the country. The main health status indicators used are life expectancy (LEXP) and infant mortality rates (IMR). Life expectancy is a popular indicator of health although it is not without problems. Feachem et al. (1992) show that the causes of death in adults are much less likely to decrease with increases in per capita income; it may, in fact, increase. For example, many adult deaths could be due to motor vehicle accidents, use of tobacco and alcohol, excessive consumption of food products related to heart disease, and all these tend to rise with income. Infant mortality is a good alternative indicator as it avoids the potentially more severe reverse causation problems associated with the relationship between adult health and income growth. The mortality rate of children under 5 years of age is usually used as an indicator of child well-being (UNICEF, 1991 and 1992). This welfare measure, which we refer to as child mortality rate (CMR), is used as another main health indicator here.6 To account for ill health in life expectancy, we use a new variable developed by the World Health Organisation called healthy life expectancy (HLE). To calculate HLE, the years of ill health are weighted according to severity and subtracted from the expected overall life expectancy to give the equivalent years of healthy life. This indicator is also known as the

CO2 ENPC TSP SO2 NOx

6 There exist alternative data sources for the variables used in this paper. For example, the World Resources Institute (WRI) provides data on the environment, health and the economy (http://www.wri.org/wri/facts/data-tables.html). WRI data on air pollution is very similar to the kind we use as it is available for a few years, but not over time. The data on water pollution is of better quality in our current source (i.e. World Development Indicators, 1998) compared with that of the WRI. Further, WRI data for water pollution is for different years for different countries, with a time span difference of up to 13 years in some cases. Such disparate data would not lend itself to cross-country comparisons, as we require in this study.

7 We would like to thank an anonymous referee for bringing this to our attention. 8 The Gini coefficient does have some limitations as a measure of income inequality. Torras and Boyce (1998), Magnani (2000) discuss some of these limitations, and it is sometimes suggested that the ratio of the income shares of the first and fourth quintile of the income distribution might be a better measure of income inequality. In this paper, however, the use of a general measure such as the Gini coefficient is sufficient, as the focus of this paper is not on inequality but on impact of environmental stress on health outcomes.

Table 2 Correlation between environmental indicators (air pollution) CO2

ENPC

TSP

SO2

NOx

1.0 0.8909 −0.2871 −0.2176 −0.1060

1.0 −0.3193 −0.2648 −0.1451

1.0 0.8905 0.8216

1.0 0.8677

1.0

522

L. Gangadharan, M.R. Valenzuela / Ecological Economics 36 (2001) 513–531

Fig. 1. Estimated relationship between GNP and CO2, ENPC level.

surprising, as we would expect that a rise in energy consumption would be accompanied by a rise in pollutant emissions. Suri and Chapman (1998) explain this seemingly inconsistent result by suggesting that it is possible for energy consumption to keep rising but for emission levels of local pollutants to fall, as would be the case when end-of-pipe technology like scrubbers are used to reduce local pollutants. As the existing policies to abate local pollution often concentrate on end-ofpipe methods and not on reducing energy consumption or emission levels, it should not surprising that energy use and CO2 emissions are not being reduced along with reductions in the levels of local pollutants. 5. Results Table 3 presents results from the estimation of the environmental equation (Eq. (2)). It is seen that per capita income, population density, country’s level of urbanisation, inequality in the distribution of income as well as level of education exert significant influences on a country’s level of environmental stress. The results are particularly strong when CO2 emissions and ENPC are used as the dependent variables. In the case of CO2, we find that when all other influences are taken to be constant, a $1000 increase in per capita GNP increases the per capita CO2 emission level by 1 metric ton. For ENPC, a 255 kg in oil equivalent increase in energy use results from a $1000 increase in per capita GNP.9 The F-test shows that the 9 The coefficients do not come out to be statistically significant for ENPC, however, the sign of the income coefficients are very similar to the signs of the income coefficients for CO2 emissions.

coefficients are jointly significant. Fig. 1 shows how CO2 emissions and ENPC levels change with increasing per capita income. At income levels below $5000, environmental stress is seen to increase with energy use increasing much more than carbon dioxide emissions. These emission levels peak at around the $6000 mark, after which energy use plateaus while CO2 emissions decrease until income levels reach $18 000. After this critical income point, rapid rates of increase in energy use and CO2 emission levels are observed. These results contrast with the standard EKC curve in that we do not find an inverted-U curve.10 Rather, we find that the curve is a flattened inverse-S shaped curve where the slope is mostly positive everywhere, except for the inflection point where the slope is zero. The inverse-S shape is observed for both CO2 and ENPC variables, with ENPC levels showing larger rates of change over the income scale, that is, the CO2 curve is flatter and less variable. The results imply that we can partition the environmental stress experience of countries into distinct phases. During the first phase when per capita incomes are low, environmental stress is shown to increase but at a diminishing rate. During the second phase when per capita incomes are higher, environmental stress levels appear controlled and no increases are observed. The third phase occurs at extremely high incomes when emissions increase again and escalate rapidly. This implies that the impact of income on the environment is more significant at the extreme ends of the income scale. In particular, the results show that very low and very high-income countries tend to experience increasing stress levels in their environmental conditions, while there is relatively little change in the environmental stress levels for the middle-income countries. The results further show that population density level and levels of urbanisation are both positively related to environmental stress while the level of income inequality is inversely related to environmental quality. Hence, as a country gets more 10 We do not find an inverted-U shaped relationship between income and any of the pollutants. In the literature, the EKC relationship is supported for pollutants like SO2 and NOx, but not always for CO2, which is usually seen to increase over income.

Variable

CO2

ENPC

TSP

SO2

NOx

EMW

Constant

−0.895 (3.401) 0.001 (0.001) c −1.12E-07 (5.62E-08)c 3.02E-12 (1.47E-12)c 0.001 (0.001)b 5.034 (2.869)b −0.090 (0.059) 0.034 (0.034) 4.44

−233.047 (897.500) 0.255 (0.208) −2.11E-05 (1.88E-05) 6.53E-10 (4.81E-10) 0.199 (0.213) 1989.433 (795.405)b −36.024 (14.988)b 18.411 (7.757)b 6.96

13.760 (7.941)b −0.002 (0.001) 1.21E-07 (1.11E-07) −2.41E-12 (2.38E-12) 0.001 (0.001) −9.916 (7.185) 0.020 (0.096) 0.012 (0.046) 0.71

1.034 (1.011) −1.93E-04 (2.40E-04) 6.91E-09 (1.70E-08) −5.38E-14 (3.60E-13) 9.84E-05 (1.43E-04) −1.421 (1.736) 0.017 (0.021) 0.010 (0.011) 0.83

1.513 (1.519) −1.85E-04 (2.61E-04) 1.07E-08 (2.01E-08) −1.65E-13 (4.65E-13) −7.21E-05 (1.82E-04) −1.016 (1.956) 0.011 (0.024) 0.002 (0.016) 0.53

844416.1 (697726.6) −129.00 (201.261) 0.002 (0.016) 1.13E-07 (3.68E-07) −152.317 (143.116) −1010368 (1152394) 11058.34 (14480.17) 6086.675 (7135.055) 1.08

GNP GNP2 GNP3 POPDEN UPOP GINI EDU F-test a

Figures in parenthesis indicate robust S.E. Significant at 10% level. c Significant at 5% level. b

DEFRTE 0.485, 0.651 −4.05E-05 (2.33E-04) −9.46E-10, 1.52E-08 6.71E-14, 3.12E-13 −5.45E-05, 1.20E-04 0.201, 1.288 0.018, 0.015 −0.008, 0.008 2.45

L. Gangadharan, M.R. Valenzuela / Ecological Economics 36 (2001) 513–531

Table 3 Impact of GNP and other explanatory variables on the environmenta

523

524

L. Gangadharan, M.R. Valenzuela / Ecological Economics 36 (2001) 513–531

crowded (more people on a fixed area of land), the higher will be their CO2 emissions and per capita energy use. This can be due to the fact that as population density increases, there is increasing pressure to use the existing land more intensively. The creation of multi-storey residential and commercial buildings in high population density countries is a good example of this problem. Lifestyle adjustments for residents in these countries imply more energy consumption and this leads to abnormally high levels of CO2 emissions. Singapore is a case in point; its population density in 1996 was 4990 persons per km2 and the commercial energy use was 7162 kg of oil equivalent per capita. In contrast, the corresponding average levels for our sample of 51 countries are 206 and 2850, respectively. Clearly, the high population density in Singapore exerts a major influence on its extremely high level of energy use. Further, the percentage of population living in urban areas (UPOP) impacts positively on the levels of CO2 and ENPC, with emissions rising as urban population increases. We observe a positive coefficient for the education variable (EDU), which runs counter to expectations. The results show that higher levels of education aggravate, rather than improve, environmental conditions. On the other hand, any improvement in the inequalities between the rich and the poor is found to be detrimental to the environment. While counter intuitive in the first instance, this makes empirical sense because a move towards more equal standards of living implies more people are able to afford the use of electricity, cars and other luxuries — which leads to increased energy use.11 For such a cross-section of countries, the 11 This issue has been analysed in greater detail in Torras and Boyce (1998), Scruggs (1998) with mixed results. While Torras and Boyce (1998) find that more equitable distributions of income and power tend to result in better environmental quality, Scruggs (1998) shows that equality does not necessarily lead to lower environmental degradation. Magnani (2000) finds that higher levels of income would increase environmental quality provided the negative effect of production of goods and services on pollution levels is more than counterbalanced by the positive effect of growth on the demand for pollution abatement policy. The demand for environmental quality will be affected by inequality levels in the country. As the level of per capita income increases above a critical level, income equality becomes positively correlated with environmental protection; however, beyond a certain threshold level of income, the correlation between income and environmental protection turns negative.

explanatory power of these two models is fairly high (adjusted R 2 = 67% for CO2 and 79% for ENPC). Eq. (2) was also estimated using data on other specific pollutants such as TSP, SO2, NOx, EMW and DEFRTE. The magnitude and signs of the estimated coefficients are very sensitive to the pollutant used, and are very unstable. Further, the explanatory power of the models are greatly reduced with F-test results simultaneously indicating inappropriate models. We note that many environmental studies used CO2 and ENPC precisely because the data on these variables are well developed. Also, we note here that trend results are similar for CO2 and ENPC because CO2 is a major component of ENPC. As seen in Table 2, these two environmental stress variables have a high and positive correlation between them.

5.1. Impact on health Results of the two-stage least squares (2SLS) estimation of Eq. (3) are presented in Tables 4 and 5. In these estimations, we use alternative indicators of a population’s health status — namely, life expectancy (LEXP), healthy life expectancy (HLE), infant mortality rate (IMR) and child mortality rate (CMR) — and treat the environmental stress variable as endogenous. We find that if we ignore the potential endogeniety of the environmental variables, the results obtained are inconsistent. The Davidson and MacKinnon (1993) augmented regression test shows that the null hypothesis of an exogenous environmental stress variable is strongly rejected for all the alternative types of pollutants.12 Table 6 presents the coefficients of the environmental stress variables for the different health indicators and compares the OLS and the 2SLS estimates. The coefficients obtained from the 2SLS estimation have signs in the expected direction and the magnitudes are larger compared with the coefficients from the OLS estimation. This implies that the impact of the environmental stress variable on

12 The augmented regression is formed by including the predicted value of the endogenous right-hand side variable as a function of the all exogenous variables, in a regression of the original model.

Table 4 Impact of GNP and environment on health: with LEXP and HLE as dependent variablesa CO2

(A) Dependent 6ariable: life expectancy Constant 57.228 (3.597)b Environmental −0.380 (0.156)b stress variable GNP 0.001 (9.3E-05)b IMM 0.064 (0.038)c DOC 0.009 (0.004)b EDU 0.012 (0.022) UPOP 3.846 (2.813) F-test 23.60

ENPC

TSP

SO2

NOx

EMW

DEFRTE

53.078 (3.250)b −0.002 (0.001)b

62.143 (5.048)b −0.623 (0.440)

62.728 (6.948)b −5.144 (5.156)

65.456 (12.590)b −4.152 (5.439)

63.381 (5.336)b −4.220E-06 (0.470E-06)c

58.002 (3.962)b −1.803 (3.430)c

4.1E-04 (7.4E-05)b 0.095 (0.062) 0.006 (0.005) 0.008 (0.022) −5.383 (4.526) 16.16

3.3E-04 (1.2E-04)b 0.110 (0.082) 0.005 (0.007) 0.016 (0.021) −5.541 (5.614) 9.40

4.6E-04 (1.4E-04)b 0.056 (0.104) 0.012 (0.009) −0.004 (0.040) −4.054 (6.700) 9.43

5.01E-04 (1.1E-04)b 0.063 (0.069) 0.007 (0.004)c 0.006 (0.025) −3.737 (4.363) 12.59

4.0E-04 (8.7E-05)b 0.087 (0.081) −0.001 (0.017) 0.002 (0.041) 3.664 (4.506) 21.61

44.883 (8.480)b −0.603 (0.627)

47.426 (10.820)b −7.256 (7.622)

52.610 (20.266)b −6.626 (8.987)

47.126 (9.874)b −5.890E-06c (3.690E-06)

41.164 (6.873)b 0.320 (4.680)

4.352E-04 (1.115E-04)b 0.157 (0.097)c 0.019 (0.009)b 0.004 (0.038) −2.234 (7.330) 14.11

3.053E-04 (1.807E-04)c 0.196 (0.138) 0.017 (0.011) 0.012 (0.042) −5.738 (8.015) 7.72

5.029E-04 (2.108E-04)b 0.120 (0.172) 0.028 (0.016)c −0.020 (0.070) −4.771 (10.689) 7.14

5.604E-04 (1.665E-04)b 0.146 (0.123) 0.020 (0.009)c −0.004 (0.048) −3.463 (6.259) 9.43

4.652E-04 (1.564E-04)b 0.112 (0.099) 0.021 (0.026) 0.014 (0.039) 4.662 (4.797) 17.97

0.001 0.091 0.013 0.030 4.932 17.27

(1.6E-04)c (0.033)b (0.004)b (0.023) (3.301)

(B) Dependent 6ariable: healthy life expectancy Constant 40.236 (6.759)b 37.139 (5.767) −0.326 (0.237) Environmental − 0.002b (00.001) stress variable GNP 0.001 (1.357E-04)b 7.393E-04 (1.678E-04)b IMM 0.126 (0.070)c 0.146 (0.061)b DOC 0.022 (0.008)b 0.025 (0.006)b EDU 0.008 (0.039) 0.023 (0.037) 6.504 (3.639)c UPOP 7.176 (3.348)b F-test 18.16 19.50 a

Figures in parenthesis indicate robust S.E. Significant at 5% level. c Significant at 10% level. b

L. Gangadharan, M.R. Valenzuela / Ecological Economics 36 (2001) 513–531

Variable

525

526

Table 5 Impact of GNP and environment on health: with IMR and CMR as dependent variablesa Variable

TSP

SO2

NOx

EMW

DEFRTE

(A) Dependent 6ariable: infant mortality rate Constant 74.065 (11.594)b Environmental stress 0.173 variable (0.619) GNP −0.001 (3.285E-04)b IMM −0.214 (0.129)c DOC −0.056 (0.011)b EDU −0.064 (0.066) UPOP −13.266 (9.619) F-test 22.62

82.829 (13.419) 0.004 (0.003) −0.002 (0.001)b −0.275 (0.135)b −0.066 (0.013)b −0.091 (0.071) −17.604 (10.654)c 17.98

60.419 (15.342)b 2.113 (1.178)c −0.001 (2.181E-04)b −0.348 (0.159)b −0.051 (0.014)b −0.038 (0.064) 12.711 (12.768) 17.64

59.308 (20.929)b 16.439 (14.902) −0.001 (3.586E-04)c −0.388 (0.209)c −0.047 (0.020)b −0.066 (0.066) 11.772 (16.183) 10.85

45.716 (46.084)b 16.079 (18.806) −0.001 (4.523E-04)b −0.216 (0.349) −0.073 (0.031)b 0.011 (0.132) 11.144 (22.007) 7.64

56.816 (12.095)b 1.540E-05 (8.420E-06)c −1.270E-03 (3.580E-04)b −0.283 (0.141)b −0.048 (0.016)b −0.023 (0.074) 10.838 (15.122) 16.53

75.416 (12.086)b 12.889 (10.253) −0.001 (3.391E-04)b −0.442 (0.256)c 0.003 (0.053) 0.034 (0.133) −24.073 (16.440) 13.39

(B) Dependent 6ariable: child mortality rate Constant 112.428 (21.052)b Environmental stress 0.416 variable (0.851) GNP −0.001 (4.639E-04)b DPT −0.468 (0.223)b DOC −0.071 (0.017)b SEC −0.063 (0.095) UPOP −22.663 (13.364)c F-test 17.97

127.268 (23.089)b 0.007 (0.005)c −0.002 (0.001)b −0.571 (0.227)b −0.088 (0.021)b −0.111 (0.103) −29.605 (15.172)c 13.91

89.890 (27.228)b 3.433 (1.948)c −0.001 (3.176E-04)b 0.683 (0.299)b −0.062 (0.020)b −0.022 (0.094) 20.191 (19.265) 11.99

90.029 (36.188)b 24.473 (22.826) −0.001 (0.001) −0.723 (0.363)b −0.057 (0.030)c 0.068 (0.095) 15.370 (24.349) 6.44

71.226 (71.380) 23.110 (28.130) −0.001 (0.001)b −0.467 (0.547) −0.095 (0.046)b 0.043 (0.195) 13.224 (33.370) 5.36

82.166 (22.437)c 2.190E-05 (1.250E-05)c −0.002 (0.001)b −0.512 (0.266)c −0.059 (0.021)b −0.010 (0.107) 12.990 (21.196) 10.61

113.820 (21.322)b 17.854 (15.862) −0.001 (4.825E-04)c −0.779 (0.419)b 0.011 (0.080) 0.071 (0.208) −36.789 (24.261) 10.82

Figures in parenthesis indicate robust S.E. Significant at 5% level. c Significant at 10% level. b

L. Gangadharan, M.R. Valenzuela / Ecological Economics 36 (2001) 513–531

ENPC

a

CO2

Environmental

Health indicator

stress variable LEXP

HLE

IMR

OLS

2SLS

t-Valueb

OLS

CO2

−0.16c

−0.38b

6.63c

−0.35c

ENPC

−6.02E-04c (3.67E-04) −0.002b (0.001)

6.45c

−1.33E-03b (5.75E-04) −0.002b (0.001)

(0.09)

(0.16)

(0.20)

2SLS

t-Valueb

−0.33 (0.24)

3.51c

OLS −0.22 (0.20)

3.43c

3.00E-04 (8.12E-04)

CMR t-Valueb

OLS

0.17 (0.62)

3.78c

−0.08 (0.39)

0.004 (0.003)

3.76c

2SLS

1.50E-03 (1.09E-03)

2SLS

t-Valueb

0.42 (0.85)

4.19c

0.007c (0.005)

4.03c

SO2

0.21 (0.24)

−5.14 (5.16)

8.35c

0.42 (0.33)

−7.26 (7.62)

3.81c

0.53 (1.11)

16.44 (14.90)

3.56c

NOx

0.28 (0.19)

−4.15 (5.44)

8.48c

0.46 (0.31)

−6.63 (8.99)

4.01c

0.31 (0.98)

16.08 (18.81)

3.62c

−0.19 (1.35)

23.11 (28.13)

4.13c

TSP

0.02 (0.07)

−0.62 (0.44)

8.45c

0.10 (0.10)

−0.60 (0.63)

3.74c

0.33 (0.37)

2.11c (1.18)

3.45c

0.34 (0.53)

3.43c (1.95)

3.96c

6.86E-08 (4.58E-07)

−4.22E-06c (2.47E-06)

6.98c

3.45E-07 (5.99E-07)

−5.89E-06c (3.69E-06)

3.53c

1.17E-06 (1.53E-06)

2.19E-05c (1.25E-05)

4.21c

−1.80c (3.43)

7.46c

3.43c

0.03 (2.10)

EMW DEFRTE a b c

−0.25 (0.44)

−0.16 (0.77)

0.32 (4.68)

Figures in parenthesis indicate robust S.E. *, Significant at 10%; **, significant at 5%. t-Value from Davidson–Mackinnon augmented regression test for exogeneity. Indicates the null hypothesis of exogeneity is rejected at 5% significance level.

1.54E-05c (8.42E-06) 12.89 (10.25)

3.56c 3.70c

2.34E-03 (1.40)

1.20E-06 (2.11E-06) −0.27 (3.08)

24.47 (22.83)

17.85 (15.86)

4.06c

4.29c

L. Gangadharan, M.R. Valenzuela / Ecological Economics 36 (2001) 513–531

Table 6 Comparative test results from alternative estimation methods useda

527

528

L. Gangadharan, M.R. Valenzuela / Ecological Economics 36 (2001) 513–531

health is bigger when we take the endogeniety into account. The t-statistic from the Davidson – Mackinnon augmented regression test for exogeneity shows that the null hypothesis of exogeneity is rejected at the 5% significance level. Panel A in Table 4 presents the results when life expectancy is the health indicator used. In general, the environmental stress variables have the correct sign (negative) and are significant for certain pollutants such as CO2, EMW, ENPC and DEFRTE.13 GNP is always significant and increases in income levels lead to an increase in life expectancy. The level of immunisation (IMM) is seen to increase life expectancy when the pollutant used is CO2 or ENPC. Education level (EDU) does not seem to be significant for improving health. The availability of doctors (DOC) as a proportion of the population has a significant impact on improving life expectancy, particularly for the CO2, ENPC and EMW pollutants. When the health indicator used is healthy life expectancy (panel B of Table 4), the results are similar. The availability of doctors increases healthy life expectancy significantly; the absolute impact (measured by size of the coefficients) of this variable is also greater with HLE, as the dependent variable. The impact of urbanisation (UPOP) on health is large and significant particularly for HLE as the dependent variable. This positive impact reflects the society’s benefits from improvements in the provision of better waste disposal and sanitation facilities, which would come with urbanisation. When infant mortality is taken as the health indicator (panel A of Table 5), we find that increases in TSP emissions and water pollutant emissions levels lead to significantly high infant mortality. Coefficients derived from the estimation of the model using the child mortality rates are very similar to the infant mortality results (panel B of Table 5). In general, the results

show that income, immunisation rates, access to doctors and urbanisation levels all make large positive and significant impacts on both infant and child mortality rates. Only the education variable fails to make a significant impact on mortality rates. We also use log linear models to check for robustness of results. Per capita GNP, purchasing power parity, is found to be significant in all cases in improving health. Using log TSP, log SO2, log NOx and log EMW as the environmental stress variable, it is found that the coefficients are negative and significant (for log SO2 and log NOx) in explaining health outcomes. Hence, the environmental variable is significant in explaining changes in health levels in a population. The estimated coefficients for log CO2 and log ENPC, however, have positive coefficients and are significant. This is contrary to what we would expect. We find, therefore, that in some cases the results could be sensitive to the functional form used.14

6. Conclusion and further research In this paper, we examine the links between health status, income and environmental indicators of a country. We first look at the relationship between environment and income — the EKC hypothesis. We find that low-income countries cannot simply postpone attending to environmental concerns in the hope that the environment will eventually improve with increased incomes. Health is a significant intervening variable and isolating the impact of environment on health is very important, particularly in the context of developing countries. Our results show that the gains in health obtained through improved incomes can be negated to a significant extent if the indirect effect of income, acting via the environ-

13

The economic relationship between deforestation and health indicators is a bit ambiguous. As deforestation is an indicator of ecological balance in the country, this could have long-term effects on health, however, the impact in the short term is not very clear.

14

Results for log linear models are not presented in the paper, however these are available from the authors on request.

L. Gangadharan, M.R. Valenzuela / Ecological Economics 36 (2001) 513–531

ment, is ignored. This study thus shows that policy makers who have chosen to pursue rapid growth strategies at the expense of the environment are not delivering the full realisable health gains that can be derived from higher incomes. Also environmental damage is bound to result in health problems for the domestic population. A less healthy labour force will not be able to increase productivity levels, and hence result in lesser income for the economy. Addressing chronic health problems for the population is also costly and will divert valuable resources from income generating investment projects. Clearly, policies for growth must incorporate appropriate programs for protection of the country’s natural environment and this does not have to be at odds with growth and development targets. One of the ways this research can be extended is to obtain time series data on environmental indicators and health status for varied countries along the development spectrum. As we are interested in different kinds of environmental and health indicators, obtaining data for all these indicators for many years is quite challenging. Most developing countries do not keep records of environmental variables, and this hampers our objective here. However, with the continued improvements in data availability, over time this problem will be reduced and research on this can be encouraged. A second extension is to create a single indicator or index that could capture the overall quality of a country’s physical environment. Such an index would be useful for analysing environmental issues within a country and can also provide important insights for cross-country trends. A third extension would be to study the impact of environmental policies on income or growth in a country and whether it has an impact on the health outcomes. This could help us in understanding whether the pursuit of pro-environmental policies would have beneficial or adverse effects on growth itself. This could be examined using a three-equation system with income, environmental stress and health being endogenous variables.

529

Acknowledgements We would like to thank Pete Summers, three anonymous referees, seminar participants at the Research School of Pacific and Asian Studies, Australian National University, the participants at the Conference of Economists, 1999, Latrobe University and at the Sixth Biennial Meeting of International Society of Ecological Economics, Canberra, 2000, for their comments. We are, however, responsible for all remaining errors. Funding for this research was provided by the Faculty Research Grant Scheme, Faculty of Economics and Commerce, University of Melbourne.

Appendix A. Countries included in the study. 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29

Argentina Australia Austria Belgium Brazil Bulgaria Canada Chile China Colombia Croatia Czech Republic Denmark Ecuador Egypt, Arab Republic Finland France Germany Greece Hungary India Indonesia Iran, Islamic Republic Ireland Italy Japan Kenya Korea, Republic Malaysia

530

30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51

L. Gangadharan, M.R. Valenzuela / Ecological Economics 36 (2001) 513–531

Mexico Netherlands New Zealand Norway Philippines Poland Portugal Romania Russian Federation Singapore South Africa Spain Sweden Switzerland Thailand Turkey Ukraine United Kingdom United States Venezuela Ghana Slovak Republic

References Agras, J., Chapman, D., 1999. A dynamic approach to the Environmental Kuznets Curve hypothesis. Ecol. Econ. 28, 267– 277. Chakrabarti, A., Rao, D.N., 1999. Measuring the impact of policy variables on the burden of disease. Mimeo, Jawaharlal Nehru University. Cropper, M.L, Simon, N.B., Alberini, A., Sharma, P.K., 1997. The health effects of air pollution in Delhi, India. Policy Research Working Paper 1860, The World Bank. Davidson, R., MacKinnon, J.G., 1993. Estimation and Inference in Econometrics. Oxford University Press, New York. De Bruyn, S.M., van den Bergh, J.C.J.M., Opschoor, J.B., 1998. Economic growth and emissions: reconsidering the empirical basis of Environmental Kuznets Curve. Ecol. Econ. 25, 161– 175. Deininger, K., Squire, L., 1996. Measuring Income Inequality: A New Data-base. World Bank, Washington, DC., mimeographed. Feachem, R., Kjellstrom, F., Murray, C., Over, M., Phillips, M., 1992. The Health of Adults in the Developing World. Oxford University, Oxford. Flegg, A.T., 1982. Inequality of income, illiteracy and medical care as determinants of infant mortality in underdeveloped countries. Popul. Stud. 36 (3), 441–458. Galeotti, M., Lanza, A., 1999. Desperately seeking (Environmental) Kuznets. Mimeo, International Energy Agency.

Gerdtham, U., Sogaard, J., Andersson, F., Jonsson, B., 1992. An econometric analysis of health care expenditure: a cross-section study of the OECD countries. J. Health Econ. 11, 63 – 84. Grossman, G., 1995. Pollution and growth. In: Goldin, I., Winters, L.A. (Eds.), The Economics of Sustainable Development. OECD, Paris, pp. 19 – 46. Grossman, G., Kreuger, A.B., 1995. Economic growth and the environment. Q. J. Econ. 112, 353– 377. Hill, K., King, E., 1992. Women’s education in the third world: an overview. In: King, E., Hill, A.M. (Eds.), Women’s Education in Developing Countries: Barriers, Benefits and Policy. John Hopkins University Press for the World Bank, Baltimore, pp. 1 – 50. Kakwani, N., 1993. Performance in living standards: an international comparison. J. Dev. Econ. 41 (2), 307– 336. Kaufmann, R.K., Davidsdottir, B., Garnham, S., Pauly, P., 1998. The determinants of atmospheric SO2 concentrations: reconsidering the Environmental Kuznets Curve. Ecol. Econ. 25, 209– 220. Kuznets, S., 1955. Economic growth and income inequality. Am. Econ. Rev. 45 (1), 1 – 28. Magnani, E., 2000. The Environmental Kuznets Curve, environmental protection policy and income distribution. Ecol. Econ. 32 (3), 431– 443. Mellington, N., Cameron, L., 1999. Female education and child mortality in Indonesia. Bull. Indonesian Stud. 35 (3), 115– 144. Munasinghe, M., 1999. Is environmental degradation an inevitable consequence of economic growth: tunneling through the Environmental Kuznets Curve. Ecol. Econ. 29, 89 – 109. Panayotou, T., 1993. Empirical tests and policy analysis of environmental degradation at different stages of economic development. Working Paper, Technology and Environment Programme, International Labour Office, Geneva. Parpel, F., Pillai, V., 1986. Patterns and determinants of infant mortality in developing nations, 1950– 1975. Demography 23 (4), 525– 542. Preston, S., 1980. Mortality declines in less developed countries. In: Easterlin, R. (Ed.), Population and Economic Change in Developing Countries. The University of Chicago, Chicago, pp. 289– 360. Pritchett, L., Summers, L.H., 1996. Wealthier is healthier. J. Hum. Resour. XXXI (4), 841– 868. Scruggs, L., 1998. Political and economic inequality and the environment. Ecol. Econ. 26 (3), 259– 275. Selden, T.M., Song, D., 1994. Environmental quality and development: is there a Kuznets Curve for air pollution emissions. J. Environ. Econ. Manage. 27 (2), 147– 162. Subbarao, K., Raney, L., 1995. Social gains from female education: a cross-national study. Econ. Dev. Cult. Change 44 (1), 105– 128. Suri, V., Chapman, D., 1998. Economic growth, trade and energy: implications for the Environmental Kuznets Curve. Ecol. Econ. 25, 195– 208.

L. Gangadharan, M.R. Valenzuela / Ecological Economics 36 (2001) 513–531 Torras, M., Boyce, J.K., 1998. Income, inequality and pollution: a reassessment of the Environmental Kuznets Curve. Ecol. Econ. 25, 147–160. UNICEF, 1991 and 1992. The State of the World’s Children.

531

Oxford University Press, Oxford. World Bank, 1992. World Development Report 1992: Development and the Environment. Oxford University Press, New York.

.