1

ECONOMIC GROWTH AND ENVIRONMENTAL QUALITY: SEARCHING FOR ENVIRONMENTAL

KUZNETS CURVES IN AFRICA

Christopher O. Orubu, Douglason G. Omotor Department of Economics,

Delta State University, Abraka, Nigeria.

E-mail: corubu@yahoo.com* yomotor@yahoo.com

&

Patience O. Awopegba

UNESCO Office, Addis Ababa,

Ethiopia.

Text of Paper Presented at CSAE Conference, University of Oxford, UK. March 22 –24 2009.

2

ECONOMIC GROWTH AND ENVIRONMENTAL QUALITY: SEARCHING FOR ENVIRONMENTAL KUZNETS CURVES IN AFRICA

By

Christopher O. Orubu, Douglason G. Omotor Department of Economics,

Delta State University, Abraka, Nigeria.

&

Patience O. Awopegba

UNESCO Office, Addis Ababa,

Ethiopia.

Key Words: Growth, Environmental Degradation, Kuznets Curves Abstract

This study investigated the relationship between per capita income and environmental degradation in selected African counties, using longitudinal data spread generally between 1975 and 2002. The specific objective was to estimate environmental Kuznets curves for five indicators of environmental quality, namely: suspended particulate matter, carbon dioxide emissions, access to sanitation, access to safe water, and emissions of organic water pollutants, and to establish whether the estimated relationships conform to the inverted U-shape hypothesis. The results of the empirical investigation generally suggest the existence of environmental Kuznets curves for suspended particulate matter, carbon dioxide, organic water pollutants and access to sanitation. Other factors such as population density, education as well as policy regimes were also found to affect environmental quality. Specifically, population density has a positive effect on environmental degradation, particularly for suspended particulate matter, while education tends to reduce environmental pollution from the same source. An N-shaped pollution – income curve was also indicated for organic water pollutants – an indication that more stringent policy measures may be required to stem pollution from this source, as incomes rise to higher bounds. The turning levels of income for the various indicators of environmental quality were however generally low, when compared to evidence from existing studies on the environmental Kuznets curve, thus suggesting that African countries may be turning the corner of the environmental Kuznets curve, much faster, and at lower levels of income than expected. Indirectly, this is also evidence that policy measures already put in place may be working well, and there is the need to continue with such policy measures. The influence of other factors such as population and education on environmental quality provides justification for mainstreaming the environment into the entire process of planning for development in order to ensure environmental sustainability in Africa.

We acknowledge with profound thanks and gratitude the financial support of the African Economic Research Consortium (AERC) for this study. We are also grateful to all the resource persons and other members of Group B of the AERC biannual Research Workshops for their useful comments and suggestions. We however remain solely responsible for any errors and short-comings of this work.

3

1. Introduction

The environment encapsulates the natural capital stock of an economy. It comprises all land and

natural resources, water and related resources as well as the atmosphere which envelopes the earth. It

provides the natural material inputs for the production of goods and services and nature’s sink into

which the wastes arising from economic activities – including production and consumption – are

dumped. Environmental degradation or decline occurs with the deterioration of the quality of

environmental resources or services from their original states. On the other hand, economic growth

involves sustained increase in the aggregate scale of production, and by implication greater use of

material resources and inputs. While a rapidly growing economy is desirable due to its beneficial

social and economic effects, it also presupposes a well functioning environment and it is now

recognized that the quality of the environment and economic growth are intricately related to each

other over time. However, there does not seem to be a hard-and-fast rule about the effects of

economic growth on the quality of the environment. Thus while a number of researchers (e.g.

Georgescu-Roegen, 1971, Hall, Cleveland and Kaufmann, 1986) suggest that higher incomes reduce

environmental quality, others (e.g. Beckerman, 1992; World Bank, 1992) are of the view that higher

levels of income are associated with improvements in environmental quality.

During the 1970s, and for a significant part of the 1980s, the debate on the relationship between the

environment and economic growth had been largely influenced by the materials balance paradigm. An

essentially physico- economic correlate of the law of mass conservation, the materials balance

approach to the economic analysis of environmental problems posits that as the volume of the

economy’s output increases due to expanding industrial activities and use of more material and

natural resources as inputs, the quantity of effluents and other associated wastes also increases,

thereby leading to intense pollution of the physical environment (Seneca and Tausig, 1984; Kneese

and Bower, 1979). In effect, increased economic activities create growing demand for the use of the

environment, not only in terms of explicit use of its resources, but also either consciously or

unconsciously as a medium of waste disposal. Pollution arising from human-induced activities

therefore involves an extractive use of the environment, to the extent that each unit of pollution

generated either through production or consumption processes is bound to reduce the quality of the

environment (Smulders, 2000). An economic system can therefore only be environmentally

sustainable if it is physically in a steady state in which the amount of resources it utilizes to generate

its output and welfare for its citizens is constrained to the size and quality that does not overexploit its

resources and overburden nature’s sinks (Stagl, 1999). Within the framework of the materials

paradigm, economic growth, ceteris paribus necessarily leads to a deterioration of environmental

quality. It could also be shown that pollution arising from economic activities could exert undesirable

4

price and output effects, with definite consequences on sustainable development (Orubu, 2004; Orubu

et al. 2002). Unfortunately, environmental quality shares in the essential characteristics of public

goods, and so, unlike other scarce goods, it is not easily traded in the market! The pervasiveness of the

negative externality effects associated with anthropogenic economic activities therefore justifies the

need for environmental policy intervention1.

An important milestone in understanding the relationship between economic growth and the

environment was laid during the second quinquennium of the 1980s, which recognizes the

complementarities that exist between them, with an emphasis on the need to mainstream

environmental concerns into the planning process in order to ensure sustainable development (WCED,

1987; Pearce and Warford, 1993). Grossman and Krueger (1991), in their path-breaking work on the

potential environmental impacts of the North American Free Trade Agreement (NAFTA) had

extended this milestone, by providing seminal evidence in support of an inverted U-shaped

relationship between economic growth (measured by increases in per capita income) and some

indicators of environmental quality. This relationship is the so-called ‘environmental Kuznets curve

(EKC)’2. The EKC phenomenon received further popularization by the World Bank (1992) in its

World Development Report. This study represents an empirical contribution to the existing knowledge

and debate on the environmental Kuznets curve (EKC) hypothesis, and searches specifically for EKCs

for five indicators of environmental quality for selected groups of African countries. Following this

introductory section, the remaining of the paper is structured into five sections, beginning with

section2 which examines the nature of the research problem. The literature is reviewed in section3,

while section4 presents the framework of analysis and methodology adopted. The empirical results

and discussions are contained in section5, while section6 concludes the paper by drawing out the

policy implications of the findings.

2. Nature of the Research Problem

Environmental degradation has definite adverse effects on human welfare, mainly in terms of reduced

wellbeing and livelihood opportunities of individuals who are directly affected. The World Bank

(1992) has documented these in terms of health and productivity effects for different categories of

environmental degradation, including water pollution and water scarcity, air pollution, solid and

hazardous wastes, soil degradation, deforestation, loss of biodiversity and atmospheric changes. Air

pollution resulting mainly from suspended particulate matter and volatile organic compounds is

known to cause chronic health problems among people who live close to industrial areas in urban

centres. A large number of rural dwellers, particularly women and children suffer from the adverse

effects of indoor air, while vehicular and industrial activity also affect productivity as does the effect

of acid rain on forests, crops, aquatic life and metal roofs of buildings. Water pollution is responsible

for a number of water-borne diseases. Ill-health resulting from either air or water pollution, through its

5

effect on human strength and concentration reduces productivity in diverse ways. Water pollution can

increase the cost of providing safe water for the human population, while its shortage can constrain

economic activity through its adverse effect on other environmental resources as in declining fisheries

and aquifer depletion (Orubu, 2006). Solid and hazardous wastes also affect human health and

productivity through the pollution of ground water resources, while soil degradation may increase the

risk of reduced living standards among the vulnerable segments of the population who depend on

farming for their living. One other critical consequence of environmental degradation that is generally

less visible arises from the rapid rate of depletion of the economy’s natural capital base.

More recently, it has also been observed that as a result of increasing economic activities, world

consumption of fossil fuels such as oil, natural gas and coal, the amount of carbon dioxide in the

atmosphere has increased substantially over its level over the past century (Sengupta, 1996; Ciegis,

Stremikiene and Mativsaityte, 2007), stretching the natural greenhouse effect beyond its beneficial

bound. It has been estimated that significant changes in global climate could result from the

intensified greenhouse effect as a result of continued increase in carbon dioxide concentration in the

atmosphere and that this accounts for more than half of global warming that is already taking place

(Shi, 2004). Global warming has potential grave consequences for climate change that could lead to

increased incidents of natural disasters such as sea level rise, floods, displacement of human and

animal populations, possible food insecurity, and ultimate disorganization of life on earth as we

presently know it!

To what extent then, does the proposition of the environmental Kuznets curve provide an antidote for

reversing the undesirable consequences of man’s economic activities? Grossman and Krueger (1991)

provide an optimistic view. They argue that economic growth, taking place at the intermediate stage

does increase pollution, hence deterioration in environmental quality. However, the capacity to offset

this relationship tends to increase in later stages of the growth process. Thus during the initial stage of

the developmental process, when the typical economy is dominated by agriculture and allied

activities, pollution intensity will be generally low. But as the economy moves into heavy industry,

pollution will tend to increase. Furthermore, as the economy shifts into high technology and services,

pollution intensity will tend to fall. What is implied in this observation is that pollution intensity is

likely to be increasing in countries at the lowest rung of the development ladder, up to the

intermediate stage, before possibly declining after reaching a threshold level of income. If the EKC

proposition holds true, then economic growth becomes an eventual means of achieving environmental

improvement over time!

11

An Overview of Current Environmental Issues in Africa

According to the African Development Bank (2006), the continent is characterized by a number of

environmental problems, which include: soil erosion, desertification, deforestation, water and air

pollution, relatively high carbon intensity, habitat loss and threatened wild life population, and poor

sanitation facilities and practices. While soil erosion generally results from overgrazing and other

poor farming practices induced mainly by population pressure, desertification has been traced to rapid

deforestation arising from intense use of forest wood as fuel and timber. As noted by the ADB (2006),

increasing scarcity of forests in many African countries is illustrated by a generally declining forest-

to-people ratio, which is currently less than half of what it was four decades ago.

Air pollution in African countries usually arises from two major sources – primary and secondary

sources respectively. The major primary source of air pollution in African countries is suspended

particulate matter (SPM), which consists of chemically stable substances such as dust, soot, ash,

smoke, and liquid droplets from fuel consumption, industrial production and construction activities

(Schwela, 2002), and is capable of being transported to other localities by fugitive winds (Dimai et al.

2008). Generally, the state of a country’s development and pollution control technology are important

determinants of particulate matter concentration (World Bank 2003:169). Comparatively, measures of

SPM for African countries are relatively high, compared to what obtains in a number of industrial

countries. For example, SPM for Sweden, United Kingdom, United States, stood at 15, 17, and 27

( µg/m3 ) respectively, in 2003 (World Bank, 2003: 169). Compare these figures to those for Angola,

Nigeria, and Sudan, measured at 112.8, 94.8, and 219 respectively for the year 2002. Most African

countries are still at the intermediate stage of development, where industrial production and

transportation is energy-intensive. As in other developing countries, indoor pollution from cooking

fuels is also relatively high in African countries (Schwela, 2002); hence SPM levels tend to be high. It

should also be noted that air pollution from particulate matter is higher when the primary fuel for

power plants, industrial boilers, steel mills and domestic heating and cooking is based on coal. In the

developed countries, the use of coal as a source of fuel is rapidly giving way to cleaner forms of

energy such as natural gas, solar energy and wind-driven systems; hence pollution from particulate

matter in these countries is relatively lower. As incomes rise due to economic growth in African

countries, air pollution from this source should decline, as households switch from cooking with

firewood to gas, drive newer cars, and as production technologies become cleaner.

The major source of secondary air pollution in African countries is through a number of gaseous

emissions that are characterized by high diffusion rates and instability, resulting largely from

transportation and production technologies. In this category are such gases like carbon monoxide,

sulphur oxides, oxides of nitrogen, and hydrocarbons such as methane. The combustion of the latter

produces a large quantum of carbon dioxide as a by-product. The flaring of natural gas is a major

12

source of secondary air pollution, particularly in oil-producing countries in Africa such as Nigeria (see

for example, Orubu, 2005).Generally, as economic growth progresses and incomes increase we expect

the composition and technology effects to reduce air pollution from these sources. However, since this

study examines more specifically the problem of carbon emissions, it is necessary for us to focus on

the issue of carbon intensity. Emissions of carbon dioxide result mainly from the burning of fossil

fuels such as natural gas and industrial production (particularly cement production).It could be argued

that, given the current level of economic development in the African continent, recorded carbon

intensities (in per capita terms) are relatively high - although less than world levels. For example,

Sub-Sahara’s carbon dioxide emissions in per capita metric tons recorded at 0.9 was higher than the

average of 0.5 for all low income countries in 1980 - although the Sub- Saharan figure dropped to 0.8

in 1999, while the average for low income countries at the world level went up to 1.0 in the same year

(World Bank, 2003). Compare this to the average of 11.6 metric tons per capita recorded for high

income countries in 1999. It is of interest to note that there is an on-going important debate on the

role, which developing countries should play in curbing CO2 emissions. For example, the Kyoto

Protocol contains specific commitments taken by industrialized and transitional economies to reduce

their emissions of CO2 over the period 2008 – 2012 – a crucial factor in global warming and climatic

change - but no serious commitment exists for developing countries up to the end of the 20th century

(Anderson and Cavendish, 2001). This suggests that economic development may continue to drive up

carbon intensities in spite of changing output composition in many developing countries (including

African countries) even in the 21st century, in the absence of appropriate abatement policies

(Martinez-Zarzoso and Bengochea-Morancho, 2003:4). The difficulty currently being experienced

with the United States of America in the process of implementing the Kyoto Protocol is an indication

that reducing carbon emissions even in a developed country is not an easy task!

Water pollution in African countries is mainly due to unhygienic and poor sanitation practices and

emissions of organic water pollutants from industrial processes. Like in the developed countries, food

and beverages account for the largest share of emissions of organic water pollutants in Africa.

However, a comparative examination of the existing data (see World Bank, 2003; ADB, 2006) shows

that per capita emissions of organic water pollutants in African countries rank relatively higher than

what obtain in some industrial countries. For example, per capita emissions of organic water

pollutants (measured in kg/day/worker) in 2000 for Algeria, Angola, Burundi, Cote d’Ivoire, Kenya,

Malawi, Namibia and Nigeria stood at 0.24, 0.20, 0.24, 0.24, 0.25, 0.29, 0.35 and 0.17 respectively.

Compare these figures to those of Canada, France, Germany, United Kingdom and the USA, which

stood at 0.15, 0.10, 0.13, 0.15, and 0.12 (kg/day/worker) respectively in the same year (World Bank,

2003). Generally, water pollution from industrial production and consumption intensifies as the

economy moves into the intermediate stage of development; but as incomes rise further with

13

economic growth, there is the tendency for investment in effluent-reducing production technologies

and increased pressure for stricter environmental regulation and enforcement.

Access to safe water and sanitation services are two other important indicators of environmental

quality in any country. Access to safe water generally refers to the proportion of the population with

(or without) reasonable access to an adequate amount of water from improved sources such as

household connections, public taps, boreholes, protected wells, or spring or rain water connections.

On the other hand, access to improved sanitation services refers to the proportion of the population

with at least adequate access to excreta disposal that can prevent human, animal and insect contact

with excreta. These two measures of environmental quality indirectly indicate a country’s disease

prevention capability. For, over the period 1985 – 1989, a number of African countries were

characterized by a situation in which less than 70% of the population had no access to safe water,

whereas, in many developed countries, more than 95% of the population has access to safe water (see

World Bank, 2003; Orubu, 2006). In this category are such countries as Benin, Burundi, Central

African Republic, Cote d’Ivoire, Ethiopia, Guinea-Bissau, Mali, Mozambique, Nigeria, Sierra-Leone,

Sudan and Uganda. However, over the years, the proportion of the population in African countries

without access to safe water has been on the decline from about 43% in 1990 to 36% in 2000. The

trend for access to sanitation follows generally the same pattern; with the proportion of the population

without access falling from 50% in 1990 to about 43% in 2000 (see ADB, 2006). Generally as

incomes rise, individuals, on the aggregate are able to afford water services, or even sink their

personal water boreholes; in the same way, such individuals are able to provide better sanitary

facilities for themselves at the household level, or use their lobbying power to influence general

environmental policy at the local or state level.

It should be noted that there are other intervening influences on the environment in the typical

developing country. The United Nations Research Institute for Social Development (UNRISD) has

particularly noted the role of population pressure on environmental resources as in Asian and African

countries. Generally, as population density increases, the pressure on environmental resources

increases, leading to environmental degradation, particularly in the absence of appropriate

complementary policies (UNRISD, 1994). Implicitly therefore, rising incomes may not necessarily be

associated with improved environmental quality in the absence of policy mainstreaming to address the

adverse effects of population pressure. Since the 1990s, environmental policy advocacy, particularly

at the international level, has been in favour of integrating the principles of environmental

sustainability into the overall process of planning for sustainable human development at all levels.

The Millennium Development Goal Number 7 is the ultimate expression of this vision (see World

Bank, 2003:12). Many African countries, based on the acceptance of the concept of sustainable

development, have therefore come to recognize the need to mainstream environmental sustainability

14

into the process of planning for development. Thus by the beginning of this 21st century, most African

countries had established national environmental policy institutions/ and completed the preparation of

their environmental plans, apart from participation in a number of international treaties on the

environment such as those for Climatic Change, Ozone Layer, Chlorofluorocarbon, Law of the Sea,

and Biological Diversity. Appendix 1 summarizes the commitment to environmental concerns by way

of creation of environmental agencies, and international conventions and treaties signed up to the

early 2000s. There is therefore sufficient reason to suppose that African countries quite appreciate the

need for well-directed policies of environmental protection and management. However, as the African

Development Bank (ADB, 2004) has observed, in spite of the significant strides made at the national

and regional levels in establishing policy frameworks for environmental management and control,

environmental degradation and low quality continue to pose significant constraint on sustainable

development in Africa.

4. Framework of Analysis and Methodology

As already demonstrated in the review of the theoretical literature, the scale, composition,

technological effects, and input mix are critical proximate causes of the EKC relationship.

Specifically, higher scale implies higher output levels (and hence income) capable of yielding

pollution abatement economies. In later stages of the development process, with yet higher incomes,

changing output mix as well less residual-producing input mix are associated with decreasing

pollution-intensive methods of production that are significantly propelled by improvements in

technology. For these reasons, it is expected that during the later stages of development,

environmental degradation will begin to decline. A number of concrete attempts have been made to

provide a theoretical framework, on the basis of which, the existence of the EKC phenomenon can be

formally rationalized.

Theoretical Framework

The critical issue is to explain how environmental degradation relates to income, producing an

inverted U-shape. Lopez (1994) sees the environment (and implicitly its quality) as a factor of

production, whose efficiency could improve over time. Stokey (1998) shows that the inverted U-shape

of the EKC could emerge from a situation in which pollution control effort is not expended until a

certain pollution threshold is reached as income increases with economic growth. Beyond this

threshold, environmental degradation presumably begins to decline, as abatement effort begins to

increase with rising incomes. This view is not significantly different from that of Lieb (2001), who

argues that an EKC can only be generated when society reaches a point of satiation in consumption.

Magnani (2001) posits that the EKC results when the collective preferences of individuals for better

environmental quality are converted into public policy, while Kemp-Benedict (2003) formulates the

15

EKC hypothesis using a model in which environmental impact is seen as the product of such factors

as population pressure, the degree of societal affluence, as well as the state of technology.

Most of the existing studies of the EKC – both theoretical and empirical – make different simplifying

assumptions about the economy, in terms of how preferences, technology and other factors interact to

produce an inverted U-shaped curve. Some of these assumptions include those of infinitely lived

agents, exogenous/endogenous technological change, and whether or not pollution or environmental

degradation is as a result of production activities or consumption (Selden and Song, 1994).

McConnell (1997), for instance considers a model based on overlapping generations in which

pollution is assumed to be generated by consumption, rather than by production (see Stern,

2004a:1422; John and Peachenino, 1994). Andreoni and Levinson (2001) have however argued that

none of these special assumptions is needed to explain the existence of the EKC, and that increasing

returns to scale in abatement are sufficient to generate the inverted U-shaped relationship between

environmental degradation and income. In an earlier paper, Levinson (2000) had derived a polynomial

pollution-income curve from a model based on the utility maximizing behavior of economic agents, in

which pollution rises at lower levels of income, but falls at higher levels. Although Levinson’s model

appears plausible and conforms to the standard specification of the EKC model, it is doubtful if

economic agents, in the aggregate in a typical African economy abate pollution through voluntary self

efforts the way the model conceives it5.

A much more appealing explanation of the EKC, particularly from the point of view of the typical

developing country is that based on willingness to pay for environmental quality and services ((Boyce

and Torras, 2002; Ekins 1997; Munasinghe, 1996, 1998, 1999). As pointed out by Stagl (1999), the

common assumption is that the poor have little demand for environmental quality. Consequently, they

are constrained by their current income level and consumption needs to do nothing about improving

the environment. But as society gets richer, its members have the capacity to intensify demand for a

healthier and sustainable environment, and by calling upon government to impose more stringent

environmental control measures6. Thus at higher levels of income, the income elasticity of demand for

environmental quality is higher, and economic agents on the aggregate are not only willing and able to

pay for a ‘green’ environment, they are also expected to exert pressure on the authorities to enforce

environmental regulations. In the strict sense therefore, the EKC may well be evidence that in some

cases, institutional reforms as income increases have made private users of environmental resources to

internalize the social costs of their activities (Arrow et al., 1995). The relationship between

environmental degradation and income as depicted in the EKC (in the face of a multiplicity of

intervening factors), is therefore intended only to represent a long term relationship between the two

variables7.

16

Data Sources and Description

The data used for the empirical analysis in this study were obtained from World Bank sources, the

main source being several issues of World Development Indicators. Annual longitudinal series

(panels) of the data are used for the empirical analysis. Where necessary, complementation was from

African Development Bank’s publication, Gender, Poverty and Environmental Indicators on African

Countries, and other World Bank (2000) data files. We discuss briefly the nature of the environmental

indicators used in this study, namely: suspended particulate matter, carbon emissions, organic water

pollutants, lack of access to sanitation, and lack of access to safe water respectively, in terms of unit

of measurement, span, descriptive statistics, as well as their simple correlation with per capita GDP.

Suspended Particulate Matter (SPM): This refers to fine suspended particles less than 10 microns in

diameter, capable of penetrating deeply into the respiratory tract and causing significant health

damage in human beings and animals. It is measured in micrograms per cubic meter (µg/m3).

A consistent annual data series is available for 47 African countries for the period 1990 – 2002, as

indicated in Appendix2, making a total of 13 cross-sectional observations for each country, and total

balanced panel observations of 611for all the countries included in the sample. Appendix3 provides

the necessary summary statistics in terms of means, standard deviations (SD), and minimum and

maximum values respectively. Each annual statistic as shown in Appendix3 is for the 47 countries,

taken as a group. The smallest minimum value recorded for SPM occurred in 2002, while the highest

maximum value occurred in 1990. This observation is consistent with the trend of the mean values

recorded for SPM. The highest mean value occurred in 1990 (106.5965 µg/m3), and progressively

declined until the smallest observed mean value of 68.82277 µg/m3 in 2002. If these observations are

anything to go by, the indication is that air pollution, as measured by SPM has, on the average, been

on the decline in the countries included in the study sample over time (see Fig. 1). A glimpse into the

associative relationship between SPM and per capita income is captured by the Pearson correlation

coefficient, calculated at -0.4809 (see Table 1), thus indicating a possible inverse relationship between

primary air pollution and per capita income in African countries over the period 1990 – 2002.

Emissions of Carbon Dioxide (CO2): Emissions of CO2 are measured in thousand metric tons per

annum over the period 1975 – 2002 for 34 countries as shown in Appendix2. This makes a total of 28

cross-sectional observations for each country, and total balanced panel observations of 952. The

descriptive statistics for CO2 emissions are summarized in Appendix4. An examination of Appendix4

shows that carbon emissions, have on the average been on a progressive increase in the 34 countries

included in the study sample (see Fig. 2). For example, in 1975, average annual carbon emissions for

these countries as a group stood at 10720.86 thousand metric tons. By 1990, this has risen to 17737.20

thousand metric tons, and up to 23436.28 thousand metric tons on the average by 2002. This

17

observation seems to lend simple credence to the view that carbon emissions tend to rise over time. A

positive correlation coefficient between CO2 and per capita income is calculated at 0.6965, as shown

in Table1.

Organic Water Pollutants (OWP): Degradation of water by organic pollutants is measured in terms of

biochemical oxygen demand (BOD) – measured in kilograms/day. BOD is defined as the amount of

oxygen required by aquatic bacteria in breaking down waste. The data available for this

environmental indicator are relatively sparse, a consistent series being available for only six countries

(Botswana, Ethiopia, Kenya, Senegal, South Africa, and Morocco over the period 1980 – 2002),

making a total of 23 cross-sectional observations, and total balanced panel observations of 138. The

summary statistics are as shown in Appendix5. An examination of the raw data, at first sight, does not

indicate a definite pattern as to whether OWP is increasing or decreasing over time. For example,

while mean OWP stood at 53159.57 kg/day) in 1980, it rose up to 56847.37 kg/day in 1982; and then

fell to 55987.67 kg/day in 1983, before rising steadily up to 64865.69 kg/day in 1989. It fell slightly

in 1990, before picking up again to 70976.23 kg/day in 1993. In 2002, OWP stood at 65265.59

kg/day, higher than its value of 63383.54 kg/day in 2001 (see Fig. 3). However, the correlation

coefficient between OWP and per capita income measured negative (at -0.3299), which may well be

an indication that OWP might decline over time as per capita GDP increases.

Access to Sanitation (ASN): In this work, access to sanitation (ASN) is measured as the proportion of

the population without adequate access to faecal disposal services or systems that can prevent human,

animal and insect contact with faeces. The data series used for the analysis covers 20 African

countries. Because of the absence of relatively consistent time series for the individual countries, the

cross-sections are spaced over 1990, 1995, and 2000 respectively, making a total balanced panel of 60

observations. The summary statistics for ASN are as shown in Appendix6. Among the countries

covered in the sample, on the average, Egypt has the smallest proportion of population without access

to sanitation (6%), while Ethiopia recorded the largest proportion of population without access to

sanitation (92%). The mean values recorded for each country also reflect these trends (see Fig. 4). As

for the correlation between access to sanitation and per capita GDP is negative. This is shown by the

Pearson correlation coefficient of -0.3556, in Table 1. The negative sign of the coefficient is a

possible indication that as per capita income increases; the number of persons without access to

sanitation facilities will tend to decline.

Access to Safe Water (ASW): Access to safe water (ASW) is measured by the proportion of the

population without reasonable access to an adequate amount of water from improved sources such as

household connections, public taps, boreholes, protected wells, or spring or rain water connections.

As in the case of access to sanitation, the data series for access to safe water covers the cross-sectional

18

points in 1990, 1995 and 2000, but for 24 African countries (see Appendix2), making a total balanced

panel of 72 observations. The relevant summary descriptive statistics are shown in Appendix7 (see

Fig. 5). Among the countries listed in this data set, Angola has the highest proportion of the

population without access to safe water (65%), while Egypt records the lowest mean of 15%. The

smallest minimum value for the proportion of the population without access to safe water is recorded

by Lesotho (9%), while the largest maximum value is recorded by Central Africa Republic. The

Pearson correlation coefficient between per capita GDP and ASW is -0.6278, thus also indicating a

possible inverse relationship between the two variables, in other words, as income increases, the

proportion of the population without access to safe water will tend to decline.

Table 1: Pearson Correlation between PCGDP and Indicators of Environmental Quality

S/N Environmental Indicator Correlation Coefficient P – Value Significance

1 SPM - 0. 4809 0.0006* Sig. at 1% level 2 CO2 0.6965 0.0000* Sig. at 1% 3 OWP -0.3299 0.0612 Sig. at 6% 4 ASN -0.3556 0.1239 Sig at 12% 5 ASW -0.6278 0.0010 Sig. at 1%

Source: Authors’ computations from the raw data

19

20

Specification of the Empirical Model

Following standard practice, the framework of which we have already considered above, the basic

EKC equation, can be specified in quadratic form as,

ttt ePCGDPPCGDPED +++= 2321 )ln()ln()ln( βββ

(3)

Where:

ED = indicator of environmental degradation/

PCGDP = per capita GDP (PPP in constant US dollars)

t = time

ln = natural logarithm of the relevant variable

e = disturbance term with zero mean and finite variance

For the EKC hypothesis to hold, 0;0 32 <> ββ , and both must be statistically significant. In this

study, longitudinal data are used for the empirical analysis. One advantage of such data over cross-

sectional and time series data is that they have both cross-sectional and time series characteristics, and

implicitly rely on the premise that differences across the units under investigation can be captured by

differences in the intercept term. Such models can be generally specified to take care of fixed and

random effects. Taking into account the quadratic form of the EKC equation specified above, the

basic estimable equation could be re-expressed as;

ititittiit PCGDPPCGDPED εββγα ++++= 221 ))(ln()ln()ln( (4)

Where, i = 1, 2, 3, …, n; t = 1, 2, 3, …, T; N = nT (for a balanced panel).

In equation (4), the first two terms on the right hand side are intercept parameters that vary across

countries (i), and years (t). Here, the implicit assumption is that, although environmental

degradation/quality may be different between one country and the other at any given level of income,

the income elasticity is the same for all countries at a given level of income. On the other hand, the

time specific intercepts take care of time-varying variables that are omitted from the model, including

stochastic shocks. It is theoretically plausible to have a situation whereby income increases beyond a

threshold, and environmental quality begins to deteriorate thereafter. In such a case the EKC equation

will assume a cubic form such as,

ititittiit PCGDPPCGDPPCGDPED εβββγα +++++= 33

221 ))(ln())(ln()ln()ln( (5)

21

In equation (5), if β3 > 0, this would be symptomatic of an N-shaped curve. A few of the models tested

in the literature have included the cubic term (see for example, Shafik and Bandyopadhyay, 1992;

Bengochea-Morancho, Higon-Tamarit and Martinez-Zarzoso, 2001; Dijkgraaf and Vollebergh, 2001;

Martinez-Zarzoso and Bengochea-Morancho, 2003). In this work we estimate both quadratic and

cubic versions of the EKC equation, solving them to derive the relevant turning points. Generally,

equations of this type (4 and 5) can be estimated, taking into account, both fixed and random effects.

Fixed effects models treat αi and γt as regression parameters, while random effects models treat them

as components of the random disturbance. In this study, we use the Hausman test to choose between

the relevant fixed and random effects models. Khanna (2002) has pointed out that income is only one

of the several factors which help to determine exposure to pollution, or declining environmental

quality generally, identifying such factors as race, education, population density, housing tenure and

the structural composition of the workforce as also critical (see also Panayotou, 1997; Torras and

Boyce, 1998). Technically, finding an EKC in the presence of other modifying factors provides a

more persuasive basis for validating the hypothesis. We therefore experiment by expanding the basic

model to include such factors as population density (POPDEN), the literacy rate (EDUC), and a

dummy variable (DUM), to reflect the composite influence of internationally coordinated

environmental policy pressure. As already indicated most African countries had by 1995 signed a

number of international environmental treaties and prepared their own environmental strategy

frameworks. We therefore assign a value of 0 to this dummy for years before 1995, and 1 thereafter.

The higher the population density, the greater will be the intensity of pollution, as well as the pressure

brought to bear on environmental services and resources. On the other hand, high literacy rates exert a

definite effect on the willingness to maintain or create improved environmental quality, hence it is

expected that pollution should fall as educational attainment increases. One other purpose for

hybridizing the basic model is to establish if the EKC phenomenon is stable in the presence of other

variables. The estimable equation (dropping the cubic term, for simplicity) is,

iti

n

jitjittiti uXPCGDPPCGDPED +++++= ∑

=

γϕβββ1

2210 )ln())(ln()ln()ln( (6)

Where,

X = vector of other explanatory variables that include POPDEN and EDUC, DUM,

Where,

POPDEN = population density; EDUC = literacy rate, DUM = policy dummy variable.

22

5. Empirical Results and Discussion

The regression estimates of the environmental Kuznets curves for different environmental indicators

are summarized in Tables 2 and 3, using the traditional quadratic and augmented forms, as well as

their corresponding fixed and random effects presentations.

In the traditional model, the estimated coefficients of the two income variables in the equations for

SPM, CO2, OWP and ASN are significant and their signs imply inverted U-shaped relationships

between income and the measures of environmental quality. This is in line with the a priori

expectations that follow the EKC hypothesis. In other words, we can tentatively accept that there is an

inverted U-shaped relationship between environmental degradation and income per capita in the

countries covered in the study, for the given environmental indicators.

Table 2a: Summary Estimates for Traditional Quadratic Environmental Kuznets Equations for SPM and CO2

Independent Variables

Traditional Model (SPM)

Random Effects (SPM)

Fixed Effects (SPM)

Traditional Model (CO2)

Random Effects (CO2)

Fixed Effects (CO2)

Constant 0.222866 (1.107)

0.9774 (5.708)

1.03916 (5.954)

-30.015 (5.8535)

24.54829 (8.74949)

25.0397 (8.9429)

PCGDP 0.448202 (6.989)

0.234139 (4.3289)

0.22415 (4.04764)

8.719863 (6.5714)

-5.188169 (6.9646)

-5.308957 (7.0863)

PCGDP 2 -0.0396 (7.9216)

-0.02524 (-5.86260

-0.02527 (-5.69287)

-0.485758 (5.71339)

0.38786 (7.78796)

0.395134 (7.8736)

R -2 0.23 0.93 0.92 0.22 0.96 0.96 F-stat 93.589 52.177 7660.156 134.47 55.41337 22736.05 Turning-Point $286.887 $103.325 $84.32 $7,923.196 $802.86 $827.79 Hausman Chi2 (prob.) 4.5138

(0.1047) 5.2838 (0.0712)

Source: Source: Authors’ computations from the raw data

The estimated turning points for environmental degradation for the different indicators with respect to

per capita income are $286.89 for SPM, $7,923.2 for CO2, $739.93 for OWP and $1,160.95 for ASN

respectively in the traditional model. It is not surprising that the CO2 per capita turning point is on the

average higher than those for SPM, OWP and ASN, as CO2 is much more difficult to control.

23

Table 2b: Summary Estimates for Traditional Quadratic Environmental Kuznets Equations for OWP and ASN

Independent Variables

Traditional Model (OWP)

Random Effects (OWP)

Fixed Effects (OWP)

Traditional Model (ASN)

Random Effects (ASN)

Fixed Effects (ASN)

Constant 5.0913 (3.76767)

15.06223 (13.67166)

15.07383 (17.16452)

-57.99161 (4.577357)

-41.50595 (2.63121)

-22.2636 (-0.9811)

PCGDP 1.6123 (3.8589)

-1.483101 (5.65167)

-1.486652 (-5.65005)

16.28183 (4.591156)

11.65457 (2.642146)

6.246095 (0.988214)

PCGDP 2 -0.122014) (3.94937)

0.107677 (5.679485)

0.107937 (5.67766))

-1.153507 (4.672275)

-0.830891 (2.706959)

-0.453281 (1.032338)

R -2 0.097 0.18 0.97 0.29 0.905 0.86 F-stat 8.3733 16.13547 557.3805 13.29244 4.685 397.4202 Turning-Point $739.93 $979.26 $981.21 $1,160.95 $1,344.628 $982.27 Hausman Chi2 (prob.)

1.3405 (0.5116)

1.4622 (0.4814)

Source: Authors’ computations from the raw data

Table 2c: Summary Estimates for Traditional Quadratic Environmental Kuznets Equations for ASW

Independent Variables

Traditional Model (ASW)

Random Effects(ASW)

Fixed Effects (ASW)

Constant -2.697944 (0.944284)

-3.787135 (0.981013)

-7.1335 (-0.7101)

PCGDP 0.9422968 (0.995115)

1.331461 (1.040841)

3.331671 (1.030967)

PCGDP 2 -0.10739 (1.386198)

-0.141248 (1.35119)

-0.377295 (1.452059)

R -2 0.32 0.21 0.63 F-stat 17.82454 10.548 147.6552 Turning-Point $80.64 $111.408 $82.67 Hausman Chi2 (prob.)

5.0938 (0.0783)

Source: Authors’ computations from the raw data

These results generally agree with the findings of some earlier studies (for example, Stern and

Common, 2001; Cole, Rayner and Bates, 1997; Panayotou, 1993; Shafik and Bandhopadhyay, 1992

among others). Although the traditional model for ASW reported the expected signs; the estimated

coefficients are not statistically significant.

The results for the random and fixed effects models are reported in the appropriate columns of Table

2a – Table 2c. The Hausman test statistic is reported in the last row of each table. As can be seen

from the results, the inverted U-shaped hypothesis only holds true for suspended particulate matter

(SPM) and access to sanitation (ASN) for the random effects variant; while the expected signs of the

coefficients of the income variable are not realized, for the carbon dioxide (CO2) and organic water

pollutants (OWP) equations. The results for the access to water (ASW) equations follow the same

pattern as in the traditional equations; the coefficients, even though properly signed, are not

statistically significant. In order to establish if the EKC phenomenon is stable in the presence of other

variables, three additional factors viz; population density (POPDEN), literacy rate (EDUC) and a

dummy variable (DUM) were included in the analyses. The dummy variable assumes a value of 1 for

all years including 1995 and after and 0 for all years before 1995, to capture the composite influence

of internationally-coordinated environmental policies and related conventions and initiatives since the

second quinquennium of the 1990s. Tables 3a-c summarize the estimates.

24

For the SPM model (Table 3a), the EKC has an inverted – U shape for the traditional estimation and

for both the random and fixed effects formulations. The composite SPM model coefficients for

POPDEN, EDUC, and DUM are all rightly signed as expected a priori and statistically significant.

However the coefficient attached to DUM is not statistically significant in the fixed effects estimation.

If these hybridized results particularly for the traditional model are anything to go by, they establish a

relatively stable EKC phenomenon for SPM in the presence of other augmenting factors in addition to

income. The Hausman test indicates that the effects are correlated with the explanatory variables so

that the random effects model cannot be estimated consistently implying that the fixed effects model

SPM is consistently estimated. The implications of the SPM results are interesting. Firstly, population

density in African countries tends to intensify pollution from SPM sources. Secondly, an increase in

the literacy rate tends to reduce environmental pollution. In the same way, the negative sign attached

to the dummy variable is an indication that pollution intensity seems to be on the downward trend in

African countries since the mid- 1990s. This may be as a result of general environmental awareness

and environmental policies pursued over the years by individual African countries. Conversely, the

results of the hybridized model for the other indicators of environmental quality, (viz: CO2, OWP,

ASN and ASW) as shown in Tables 3a-c did not find any significant support for the EKC hypothesis

in the face of other augmenting variables.

Table 3a: Summary estimates for Traditional Quadratic Environmental Kuznets Equation Involving Additional Explanatory Variables for SPM and CO2

Independent Variables

Traditional Model (SPM)

Random Effect (SPM)

Fixed Effects (SPM)

Traditional Model (CO2)

Random Effects (CO2)

Fixed Effects (CO2)

Constant

0.3275 (1.511)

1.1784 (8.653)

1.7504 (8.324)

22.5515 (3.769)

11.1086 (2.569)

8.1349 (1.846)

PCGDP 0.3495 (5.123)

0.1442 (3.552)

0.16705 (2.776)

-5.2824 (3.355)

-2.0801 (1.783)

-1.4006 (1.176)

PCGDP2

-0.0296 (5.507)

-0.016 (4.921)

-0.0182 (5.522)

0.456314 (4.430)

0.1835 (2.30)

0.1351 (1.656)

POPDEN

0.0247 (6.716)

-0.0179 (1.510)

-0.1921 (4.726)

0.0044 (0.099)

0.6037 (5.619)

0.7668 (6.291)

EDUC

-0.0441 (5.561)

-0.2608 (6.163)

-0.0247 (0.491)

2.0876 (6.658)

0.1688 (0.887)

-0.0142 (0.072)

DUM

-0.0441 (4.182)

-0.0318 (8.661)

-0.002 (0.561)

0.0384 (0.312)

0.0906 (2.428)

0.0668 (1.715)

R-2 0.25 0.52 0.98 0.47 0.25 0.94 F-stat

37.734 (0.0000)

119.489 (0.0000)

511.3581 (0.000)

125.56 (0.000)

47.658 (0.000)

905.672 (0.0000)

Turning point $366.39 $90.58 $98.38 Hausman

108.01 (0.0000)

21.167 (0.0003)

Source: Authors’ computations from the raw data

25

Table 3b: Summary Estimates for Traditional Quadratic Environmental Kuznets Equation Involving Additional variables for OWP and ASN

Independent Variables

Traditional Model (OWP)

Random Effect (OWP)

Fixed Effects (OWP)

Traditional Model (ASN)

Random Effects (ASN)

Fixed Effects (ASN)

Constant

15.9932 (5.436)

11.8954 (12.593)

11.685 (9.840)

-1.1504 (0.951)

-1.3645 (0.789)

5.439 (1.199)

PCGDP -2.2343 (2.610)

-1.1859 (5.066)

-1.1878 (4.948)

0.1295 (0.296)

0.2635 (0.438)

0.3035 (0.327)

PCGDP2

0.1425 (2.346)

0.0811 (4.812)

0.0824 (4.760)

-0.0342 (0.880)

-0.0448 (0.840)

-0.0575 (0.711)

POPDEN

1.14268 (14.552)

0.7796 (6.663)

0.8102 (3.941)

0.1453 (2.010)

0.1057 (0.832)

-1.653 (2.044)

EDUC

3.1797 (9.288)

0.3266 (2.252)

0.15586 (0.959)

-16924 (3.758)

-1.14851 (2.194)

1.4236 (0.843)

DUM -4.4368 (2.632)

-0.0301 (0.572)

-0.0247 (0.374)

-0.0468 (0.277)

-0.0536 (0.629)

0.0454 (0.343)

R-2 0.63 0.39 0.97 0.31 0.16 0.89 F-stat

48.637 (0.000)

18.839 (0.000)

542.97 (0.000)

5.7045 (0.0003)

2.975 (0.0203)

20.445 (0.0000)

Hausman Chi2 (Prob)

0.000 (1.0000)

0.000 (1.000)

Source: Authors’ computations from the raw data

Table 3c: Summary Estimates for Traditional Quadratic Environmental Kuznets Equation Involving Additional Variables for ASW

Independent Variable

Traditional Model (ASW)

Random Effect (ASW)

Fixed Effects (ASW)

Constant 0.91813 (0.322)

-0.0277 (0.008)

-6.8674 (0.619)

CGDP 0.0778 (0.084)

0.36905 (0.341)

3.9859 (1.153)

PCGDP2 -0.0472 (0.627)

-0.0732 (0.786)

-0.4441 (1.586)

POPDEN -0.10955 (2.844)

-0.1121 (2.278)

-0.6684 (0.640)

EDUC 0.7300 (2.086)

0.61130 (1.470)

-1.1902 (.551)

DUM -0.09139 (0.8242)

-0.0817 (0.869)

0.2094 (1.211)

R-2 0.43 0.30 0.62 F-stat 10.751

(0.000) 6.5882

(0.0006) 5.094

(0.000) Hausman Chi2 (Prob)

9.12964 (0.104)

Source: Authors’ computations from the raw data

Tables 4a – 4b provide the parameter estimates of the third order (cubic) polynomial regressions for

the five environmental indicators under alternative specifications. The first column under each

indicator is the traditional (cubic) regression estimates, while columns two and three under each

indicator provide the parameter estimates of the fixed and random effects models respectively, with

the Hausman statistic reported in the last row of each table. An examination of the results in Table 4a

for SPM shows that that the signs do not turn out as expected. Rather, they suggest that environmental

degradation due to suspended particulate matter is low at the lower levels of income; high at higher

levels of income; and falls at highest levels of income.

26

Table 4a: Summary Estimates for Cubic Environmental Kuznets Equations for SPM, CO2 and OWP

Source: Authors’ computations from the raw data

Table 4b: Summary Estimates for Cubic Environmental Kuznets Equations for ASN and ASW

Source: Authors’ computations from raw data

No specific turning points are determined for the relationship between SPM and per capita GDP. The

sign expectations are also not met in the case of carbon. However, the results for OWP appear to fit

the expectations quite fairly, reflecting an N-shape for the environmental Kuznets curve. In other

words, there is first observed an increase in environmental degradation from organic water pollutants,

up to a per capita income level of about $253.75, at which point, pollution from this source begins to

decline. But at a higher turning point level of per capita income ($2,603.75), environmental

degradation from this source begins to increase once again! An examination of the results summarized

in Table 4b also shows that no significant EKC relationships are established for ASW in the

polynomial model in the three alternative specifications (traditional cubic, random, and fixed effects).

As for the ASN model, only the traditional specification yielded the expected result of an N- shaped

Independent Variable

SPM CO2 OWP

Traditional Random Fixed Tradition

al Random Fixed Traditional Random Fixed

Constant 5.81 (5.577)

2.485 (3.063)

1.6949 (1.8512)

287.48 (6.82)

20.319 (1.0959)

17.1424 (0.9145)

-42.8734 (-7.3029)

-14.554 (-2.824)

-14.513 (-2.848)

PCGDP -2.312 (-4.540)

-0.562 (-1.332)

-0.1255 (-0.260)

-114.63 (-7.02)

-3.5361 (-0.490)

-2.2172 (-0.304)

24.9295 (8.9559)

11.972 (5.208)

11.955 (5.198)

PCGDP2 0.407 (4.970)

0.110 (1.544)

0.034 (0.419)

15.36 (7.34)

0.17569 (0.190)

-0.0027 (-0.003)

-3.7964 (-8.8004)

-1.8865 (-5.560)

-1.885 (-5.550)

PCGDP3 -0.04 (-5.461)

-0.007 (-1.903)

-0.003 (-0.730)

-0.673 (-7.58)

0.0090 (0.2294)

0.0168 (0.4261)

0.188 (8.624)

0.0964 (5.885)

0.0963 (5.875)

R-2 0.27 0.15 0.93 0.27 0.10 0.96 0.38 0.35 0.98 Total pooled observations

611 611 611 924 924 924 138 138 138

Turning-Point $134.98 $724.98

- - - $693.53 $788.41 $288.52 $2,603.75

$253.75 $2,603.75

$231.18 $2008.8

1 Hausman Chi2 (prob.)

3.8114 (0.283)

8.3742

(0.0389)

1.113 (0.775)

Independent Variable ASN ASW

Traditional Random Fixed Traditional Random Fixed Constant -334.306

(-2.198) 3.561

(0.022) 299.04 (1.176)

24.813 (1.287)

34.053 (1.392)

27.11 (0.516)

PCGDP 132.59 (2.074)

-7.313 (-0.110)

-99.32 (-1.22)

-13.11 (-1.34)

-17.97 (-1.45)

-14.19 (-0.53)

PCGDP2 -17.410 (-1.951)

1.82 (0.20)

14.265 (1.258)

2.256 (1.38)

3.098 (1.50)

2.547 (0.578)

PCGDP3 0.754 (1.822)

-0.123 (-0.287)

-0.681 (-1.30)

-0.131 (-1.44)

-0.179 (-1.566)

-0.159 (-0.67)

R-2 0.32 0.87 0.33 0.63 Total observations 60 60 60 72 72 72 Turning-Point $1018.98

$14,346.3 - - - - -

Hausman Chi2

(prob.) 4.677

(0.20) 3.584

(0.31)

27

environmental Kuznets curve, with the lower and upper turning points measured at $1018.98 and

$14,436.34 respectively.

A comparison of the results obtained from alternative specifications of the EKC model, suggests that

the basic quadratic variant has performed more to expectation than the augmented basic model and the

cubic polynomial case. Specifically, the possibility that the EKC hypothesis holds is suggested in the

quadratic cases for SPM, CO2, OWP, and ASN. On the other hand, N-shaped environmental Kuznets

curves are only indicated in the cubic polynomial cases for OWP and ASN. In the case of the random

and fixed effects estimations, the results in the quadratic variant obtained are fairly consistent with

those obtained in the traditional model for SPM, ASN, and ASW, while perverse results in terms of

unexpected signs are obtained for CO2 and OWP (though the coefficients obtained in the OWP

equation are generally statistically significant). For the SPM and ASN random effects equations,

where the results are characterized by properly signed and significant coefficients, the estimated

Hausman statistics are 4.5138 and 1.4622, with probability values of 0.1047 and 0.4814 respectively.

This implies that the results of the fixed effects model are to be preferred to those of the random

effects model.

As for the random and fixed effects estimates of the cubic polynomial equations, only the equations

for OWP produced properly signed and statistically significant coefficients, in consonance with the N-

shaped hypothesis, with a Hausman statistic of 1.113 for the random effects model – with a

probability value of 0.775, implying also that the fixed effects estimates are preferred. Figures 6 - 12

depict the shapes of the EKCs for some of the environmental indicators, based on selected regression

results (in their logarithm transformation).

28

Fig. 9a Basic quadratic ASN model

-50-40-30-20-10

01020

1 2 3

PCGDP

ASN

Fig. 11a Basic quadratic ASW model

-3

-2

-1

0

1

2

1 2 3

PCGDP

ASW

Fig. 11b ASW quadratic fixed effects model

-8-6-4-2024

1 2 3

PCGDP

ASW

Fig.6 Basic quadratic SPM model

-0.10

0.10.20.30.40.5

1 2 3

PCGDP

SPM

Fig. 7 Basic quadratic CO2 model

-40-30

-20-10

010

20

1 2 3

PCGDP

CO2

Fig. 9b ASN quadratic fixed effects model

-30

-20

-10

0

10

1 2 3

PCGDP

ASN

Fig. 8 OWP standard cubic model

-60

-40

-20

0

20

40

1 2 3 4

PCGDP

OW

P

Fig. 9

Fig. 10 Fig. 11

Fig. 12

29

6. Policy Implications of the Findings

The results of our empirical investigation suggest the existence of EKCs for four, out of the five

indicators of environmental quality chosen for this study for African countries, based on the

traditional quadratic specification. The environmental indicators, for which EKCs exist, are

suspended particulate matter (SPM), carbon dioxide (CO2), organic water pollutants (OWP), and

access to sanitation (ASN). The EKC relation was not established for access to safe water

(ASW). Augmenting the basic quadratic formulation of the EKC model lends some credence to

the view that population density, education and policy regimes do influence environmental

quality. However, the augmentation process did not produce consistent results for the EKC

hypothesis, except in the case of suspended particulate matter. On the other hand, using the cubic

polynomial specification of the EKC model, N- shaped EKCs are indicated for organic water

pollutants and access to sanitation.

The results imply generally that economic growth, as proxied by increases in per capita income

over time, may well reduce environmental degradation due to air and water pollution in African

countries. In the same vein, increased income, will also improve access to sanitation facilities

over time. N-shaped EKCs indicated for OWP and weakly for ASN imply further that as income

grows substantially, stringent environmental policy measures might be required to stem pollution

arising from organic water pollutants, and to improve access to sanitation. More specifically, the

N-shapes could be interpreted to mean that the decline in water pollution as income increases

with economic growth may well be a temporary observation, as pollution intensifies at yet higher

income levels. The same applies to the case of access to sanitation. One observation made about

the income turning points estimated for the various indicators of environmental quality is that

they are generally low, compared to what obtains for similar indicators in studies for developed

industrial countries. This is not surprising, for as the records so far indicate, apart from carbon

dioxide, the other indicators of environmental quality considered in this study are already

showing signs of secular improvement. It must be pointed out that due to increasing knowledge

of the impacts of environmental degradation, stricter policy measures as well as the availability

of pollution control technologies, developing countries may be turning the corner along the

environmental Kuznets curve faster than expected, and at far lower levels of income than those

suggested by existing empirical evidence, for the different environmental indicators . This

implies technically that the EKC shifts over time – an observation that has yet received any

significant attention in the existing literature!8

It was observed that no EKCs were found for access to safe water in this empirical investigation.

The non-existence of EKCs may well be an indication, that for African countries, rising average

incomes may not necessarily improve access to safe water, thereby raising fundamental questions

30

about the distributional consequences of economic growth. Thus it is possible for a relatively

large proportion of the population to suffer deprivation, even when on the average, incomes are

rising. Furthermore, the observation that other factors such as population and education have

impacts on environmental quality indicates the need to mainstream income distribution policies

into environmental policy measures so as to improve the quality of life among the vulnerable

segments of society.

Notes

1. Broadly speaking, traditional policy response to environmental degradation has taken three

major forms, namely conservationism, primary environmental care (which emphasizes the

human cost of environmental degradation) and a wide range of tax, pricing and accounting-

based instruments). See Utting (1993), and UNRISD (1994).

2. According to Grossman and Krueger, this inverted U-shape is analogous to the relationship

propounded to exist between income inequality and per capita national income by Kuznets

(1955), hence the environmental quality- income curve is dubbed as the “environmental

Kuznets curve”.

3. For helpful surveys of the EKC phenomenon, see Dinda (2004), and Stern (2004b). See also

Hilton (2006) and Ciegis, Stremikiene, and Mativsaityte (2007).

4. See Stern (2004b: 1421 – 1422).

5. We are grateful to an anonymous AERC referee for pointing this out to us. The modified

Levinson’s explanation can be collapsed into five basic equations (a social utility function, a

pollution function, a modified pollution function, an abatement function, and a constraint,

respectively);

),,( PCUU = ),( FCPP = , βα FCCP −= , βα FCA = , YFC =+

Where, U = total utility, C = consumption, F = effort expended in abating pollution, A = total abatement, Y = income, whileα and β are parameters. From these equations, the consumption-income, and pollution-income equations can be derived as,

YYYC

YC

YCC

+

=+

=+

=

=

+

=+

βαα

αβα

αβ

αβ

αβ

1

1

31

( )βαβα

βαβ

βαα

βαα +

+

+

−

+

= YYP

Note that if ( βα + )>1, abatement will reflect increasing returns to scale, and the pollution curve will correspond to the EKC.

6. This is the view originally expressed by Grossman and Krueger (1991).

7. For a similar view, see Ciegis, Stremikiene and Mativsaityte (2007: 46).

8. Hilton (2006) has hinted at this possibility.

REFERENCES

Adreoni, J. and A. Levinson. 2001. ‘’The simple analysis of the environmental Kuznets curve’’. Journal of Public Economics 8, 269-286

African Development Bank. 2002 and 2006: Gender, Poverty and Environmental Indicators on African Countries. Tunis.

African Development Bank. 2004: Africa Development Bank Group’s Policy on the Environment. Tunis.

Agras, J. and D. Chapman. 1999: “A dynamic approach to the environmental Kuznets curve hypothesis’’. Ecological Economics. 28(2): 267-277, February

Anderson, D. and W. Cavendish. 2001. ‘’Dynamic simulation and environmental policy analysis: Beyond comparative statics and the environmental Kuznets curve’’. Oxford Economic Papers, 53(4): 721-746

Arrow,K., Bolin, R. Costanza, P., Dasgupta, C. Folke, C.S. Holling, B.O. Jansson, S.Levin, K.G. Mater, C.A. Perrins and D. Pimentel. 1995. ‘’Economic growth, carrying capacity to the environment’’. Science , 268, 520-521

Barbier, I.B; 1997. ‘’Introduction to the environmental Kuznets curve’’. Environment and Development Economics, 2: 369-381

Beckerman, W. 1992. ‘’Economic growth and the environment: Whose growth? Whose environment?’’. World Development. 20:481-496

Bengochea-Morancho A.; F. Higon-Tamarit and I. Martinez-Zarzoso. 2001. ‘’Economic growth and CO2 emissions in the European Union’’. Environmental and Resource Economics.19(2): 165-172

Bhattarai, M. and M. Hamming. 2001. ‘’Institutions and the environmental Kuznets curve for deforestation: A cross-country analysis for Latin America, Africa, and Asia’’. World Development. 29(6): 995-1010

Boyce, J. K. and M. Torras. 2002. ‘’Rethinking the environmental Kuznets curve’’. In J.K. Boyce (ed.), The Political Economy of the Environment, Cheltenham, Edward Elgar, 47-66.

32

Carson, R.T; Y. Jeon, and D.R. McCubbin. 1997: ‘’The relationship between air pollution and income: US data’’. Environment and Development Economics, 2: 433-450.

Ciegis, R., D. Stremikiene & R. Mativsaityte 2007: ‘’The Environmental Kuznets Curve on Environmental Policy’’. Environmental Research, Engineering and Management, No.2 (40): 44-51.

Cole, M.A; and A.J. Rayner and J.M. Bates. 1997. ‘’The environmental Kuznets curve: An empirical analysis’’. Environment and Development Economics, 2: 401-416

Copeland, B.R, and M.S. Taylor. 2004. ‘Trade, growth and the environment. Journal of Economic Literature”. 42; 7-71

Daly, H.E., 1977. Steady State Economics: The Economics of biophysical equilibrium and moral growth. San Francisco: W.H. Freeman

Economics. 25(2):209-220.

de Bruyn, S.M; vanden Bergh, and J.B. Opschoor. 1998. “Economic growth and emissions: Reconsidering the empirical basis of the environmental Kuznets curve”. Ecological Economics. 25(2): 161-176,May.

Dijkgraaf, E. and H.R.J. Vollebergh. 2001. “A note on testing for environmental Kuznets curves with panel data’’. Nota di Lavoro, 63.01, Fonazione Eni Ennco Mattei

Dimai, G.A., S.S. Hati, M. Waziri and O.N. Maitera, 2008. ‘’Pollution synergy: the harmattan, fugitive dust and combustion emissions in Maiduguri Metropolis, Nigeria. European’’ Journal of Scientific Research, Vol. 23 No. 3 465-473

Dinda, S. 2004. ‘’Environmental Kuznets Curve Hypothesis’’: A Survey. Ecological Economics, N. 49, 431-455

e Barros, F. G., A. F. Mendonca & J. M. Nogueira. 2002. “Poverty and environmental degradation: The Kuznets environmental curve for the Brazilian case’’. Department of Economics Working Paper 267; University of Brasilia. December.

Ekins, P.1997. ‘’The Kuznets curve for the environment and economic growth: Examining the evidence’’. Environment and Planning, Vol.29: 805-830.

Field, B. C., 1997. Environmental Economics: An Introduction, Boston, Irwin/McGraw Hill.

Georgescu-Rogen, N., 1971. The Entropy Law and the Economic Process. Cambridge, Mass, Harvard University Press

Grossman, G.M. and A.B. Krueger. 1991: ‘’Environmental impacts of a North American Free Trade Agreement’’, Discussion Paper No. 158, Woodrow Wilson School, Princeton University, Princeton, N.J.

Grossman, G.M. and A.B. Krueger.1995: “Economic growth and the environment’’. Quarterly Journal of Economics, May, 353-377

Hall, C. A. S., C. I. Cleveland & R. Kaufmann. 1986. Energy and Resource Quality – The Ecology of the Economic Process. New York, John Wiley and Sons

33

Halkos, G.E. and E.G. Tsiones. 2001. ‘’Environmental Kuznets curves: Bayesian evidence from switching regime models’’. Energy Economics. 23(2) 191-210

Harbaugh, W., A. Levinson and D. Wilson. 2002. ‘’Re-examining the empirical evidence for an environmental Kuznets curve’’. Review of Economics and Statistics, 84(3): 541-551

Heerink, N.; A. Mulatu and E. Bulte. 2001. ‘’Income inequality and environment: Aggregation bias in environmental Kuznets curves: Evidence from automobile lead emissions’’. Journal of Environmental Economics and Management, 35(2):126-141

Hilton, F.G. 2006. ‘’Poverty and pollution abatement: Evidence for lead phase-out”. Ecological Economics, N. 56, 125- 131

Holtz-Eakin, D. and T.M. Selden. 1995. ‘’Stoking the fires? CO2 emissions and economic growth’’. Journal of Public Economics. 57(1): 185-101

John, A. and, R. Peachenino. 1994. ‘’An overlapping generations model of growth and the environment’’. Economic Journal 104: 1393-1410

Kemp-Benedict, E. 2003. ‘’Traveling along the environmental Kuznets curve’’. http://www.kb- creative.nets/IPats, Case Study No. 2.

Khanna, N. (2002). ‘’The income elasticity of non-point source air pollutants: Revisiting the environmental Kuznets curve’’. Economics Letters, 77 387-392

Kneese, A.V. and B.T. Bower 1979. Environmental Quality and Residuals Management, Baltimore, John Hopkins University Press.

Kuznets, S., 1955. ‘’Economic growth and income inequality’’. American Economic Review, 45(1): 1-28

Levinson, A. 2000. ‘’The ups and downs of the environmental Kuznets Curve’’. UCF/Center Conference on Environment, Orlando, FL.

Lieb, C.M (2001). “The environmental Kuznets curve and satiation. A simple static model” Environment and Development Economic, 7, 429-448

List, J.A. and C.A. Gallet.1999. ‘’The environmental Kuznets curve: Does one size fit all?’’ Ecological Economics. 2 :383-399.

Lopez, R. 1994. ‘’The environment as a factor of production: The effects of economic growth and trade liberalization’’. Journal of Environmental Economics and Management. 27(2): 185- 204.

Magnani, T., 2001, ‘’The Environmental Kuznets curve: Development path or policy result’’. Environmental Modelling and Software, 16, 157-166

Martinez-Zarzoso, I. and A. Bengochea-Morancho. 2003: ‘’Testing for an environmental Kuznets curve in Latin American Countries’’. Revista de An’alisis Economico, 18(1): 3-26, June

McConnell, K.E., 1997. ‘’Income and the demand for environmental quality’’. Environment and Development Economic, 2, 383-399

34

Munasinghe, M., 1996. ‘’An overview of environmental impacts of macroeconomic and sectoral policies’’. In M. Munasinghe (ed). Environmental Impacts of Macroeconomic and Sectoral Policies, Washington DC. World Bank, 1-14

Munassinghe, M. 1998. ‘’Countrywide policies and sustainable development: Are the linkages perverse?’’. The International Yearbook of Environmental and Resource Economics. 1998/1999: A Survey of Current Issues, 33-88.

Munassinghe, M.; 1999. ‘’Is environmental degradation an inevitable consequence of economic growth: Tunnelling through the environmental Kuznets curve’’. Ecological Economics. 29: 89-109.

Orubu, C. O., 2006. ‘’Water resources, environment and sustainable development in Nigeria’’. Journal of Human Ecology. 19(3), 169-181

Orubu, C.O. 2005. ‘’A quantitative assessment of gas utilization and flare reduction policies in the Nigerian petroleum industry’’. Petroleum Training Journal. 2(1): 63-74.

Orubu, C. O., 2004. ‘’Using transportation measures and economic instruments to reduce pollution due to automobile emissions’’. Journal of Social Sciences, vol. 2: 227-236.

Orubu, C. O., A. Fajingbesi, A. Odusola and N.O. Magbagbeola. 2002. ‘’Environmental regulations in the Nigerian petroleum industry: Compliance and implications for sustainable development’’. Research Report, NCEMA/ACBF, Ibadan.

Panayotou, T. 1997. ‘’Demystifying the environmental Kuznets curve: Turning a black box into a policy tool’’. Environment and Development Economics, 2: 465-484

Panayotou, T; 1993. ‘’Empirical tests and policy analysis of environmental degradation at different stages of economic development’’. Working Paper WEP2-22/WP. 238, International Labour Office, Geneva.

Pearce, D.W. and J.J. Warford. 1993. World Without End: Economics, Environment and Sustainable Development. Oxford, Oxford University Press/World Bank

Schmalensee R; T.M. Stoker and R.A. Judson. 1998 ‘’World carbon dioxide emissions: 1980- 2050’’. Review of Economics and Statistics, 80(1) 15-27

Schwela, J. 2002. ‘’Pollution and human health in developing countries’’. Journal of Human Environmental System, 6, 1-7

Selden, T.M. and D. Song. 1994. ‘’Environmental quality and development: Is there an environmental Kuznets curve for air pollution?’’. Journal of Environmental Economics and Management, vol. 27: 147-162

Seneca, J. J. and M. K. Tausig. 1984. Environmental Economics. Engle Wood Cliffs, Prentice Hall.

Sengupta, R. 1996. ‘’CO2 emission-income relationship: Policy approach for climate control’’. Pacific and Asian Journal of Energy, 7(2): 207-229

35

Shafik, N. and S. Bandhopadyay. 1992. ‘’Economic growth and environmental quality: Time series and cross country evidence’’. Working Paper WPS 904: Background Paper for World Development Report, World Bank, Washington, D.C.

Shi, J. 2004. ‘’Tests of the EKC Hypothesis using CO2 Panel Data’’. Working Paper 2004-03 Resources and Environmental Economics and Policy Analysis (REPA) Research Group: Department of Economics University of Victoria, Canada

Smulders, S. 2000. ‘’Economic growth and environmental quality’’’. in Henk Folmer and Landis Gabel (eds.), Principles of Environmental and Resource Economics. Chaltenham, Edward Elger, 602-664.

Stagl, Sigrid. 1999. ‘’Declining economic growth for environmental degradation? A literature survey on the environmental Kuznets curve hypothesis’’. Wirtschaftsuniversitat Wien Working paper No 6. http://epub.wu-wien.ac.at epub. Downloaded 25/26 June 2008.

Stern, D.I. (2004a). ‘’Comment on: Cole M.A. (2003). “Development, trade and the environment: how robust is the environmental Kuznets curve?’’ Environment and Development Economics, 8, 557-580

Stern, David I. 2004b. ‘’The rise and fall of the environmental Kuznets curve’’. World Development, Vol 32, No 8, 1419-1439.

Stern, D.I. and M.S. Common. 2001. ‘’Is there an environmental Kuznets curve for sulphur?’’. Journal of Environmental Economics and Management: 41(2): 162-178.

Stokey, N., 1998. “Are there limits to growth?”. International Economic Review, 39(1), 1-31

Torras, M. and J.K. Boyce. 1998. ‘’Income, inequality and pollution: A reassessment of the environmental Kuznets curve’’. Ecological Economics: 25(2): 195-208. May.

UNRISD, 1994. Environmental Degradation and Social Integration, Briefing Paper No 3, Geneva

Utting, Peter (1993). Trees, People and Power: Social Dimensions of Deforestation in Central America, Earthscan, London

Vincent, J.R. 1997. ‘’Testing for environmental Kuznets curve within a developing country’’. Environment and Development Economics: 2: 471-431.

Vincent, J. R., R. M. Ali & Associates. 1997, Environment and Development in a Resource-Rich Economy: Malaysia under the New Economic Policy. Harvard, Institute for International Development.

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

World Bank. 2000. Data Files. New York.

World Bank. 2003: World Development Indicators. New York.

World Commission on Environment and Development (WCED). 1987: Our Common Future, Oxford University, Press, New York

36

Appendix 1: Government Commitment to the Environment in African Countries

S/N Country Main Government Environmental Authority Any Envir. Strategy/Action Plan ?*

Treaties: Year of Participation Climatic Biodiversity Change

1 Algeria Na. Agency for Protection of the Environment

Yes 1994 1995

2 Angola Ministry of Environment - 2000 1998 3 Benin Ministry of Environment Yes 1994 1994 4 Botswana Ministry of Environment Yes 1994 1996 5 Burkina Faso Ministry of Environment and Water Affairs Yes 1994 1993 6 Burundi Environmental Agency Yes 1997 1997 7 Cameroon Ministry of Environment and Forestry Yes 1995 1995 8 Cape Verde - Yes - - 9 C. Afr. Rep. - - 1995 1995 10 Chad Environmental Protection Agency - 1994 1994 11 Comoros Ministry of Production and Environment Yes - - 12 Congo Environmental Agency Yes 1997 1996 13 Congo DRC Ministry of Environment Yes 1995 1995 14 Cote d’Ivoire National Environmental Agency Yes 1995 1995 15 Djibouti Environmental Protection Agency Yes - - 16 Egypt Egyptian Environmental Affairs Ministry Yes 1995 1994 17 Eq. Guinea - Yes - - 18 Eritrea Eritrean Agency for the Environment Yes 1995 1996 19 Ethiopia Environmental Protection Agency Yes 1994 1994 20 Gabon - Yes 1998 2000 21 Gambia National Environmental Agency - 1994 1994 22 Ghana Environmental Protection Agency Yes 1995 1994 23 Guinea National Directorate of the Environment Yes 1994 1993 24 Guinea-Bis. National Council for the Environment - 1996 1996 25 Kenya Ministry of Environment & Nat. Resources Yes 1994 1994 26 Lesotho Ministry of Tourism, Environment & Culture Yes 1995 1995 27 Liberia Environmental Protection Agency - 2002 2000 28 Libya Dept. of Environment & Nat. Resources - 1999 2001 29 Madagascar National Office for the Environment Yes 1996 1996 30 Malawi Min. of Nat. Resources & Env. Affairs Yes 1994 1994 31 Mali Environmental Agency Yes 1995 1995 32 Mauritania Min. of Rural Dev. & Environment Yes 1994 1996 33 Mauritius Ministry of Environment Yes 1994 1993 34 Morocco Department of the Environment Yes 1996 1995 35 Mozambique Environmental Agency Yes 1995 1995 36 Namibia Ministry of Environment Yes 1995 1997 37 Niger National Directorate of Meteorology Yes 1995 1995 38 Nigeria Ministry of Environment Yes 1994 1994 39 Rwanda Min. of Lands, Env. Forestry & Water Res. Yes 1998 1996 40 Sao T. & P. Min. of Social Infrastructure & the Envir. Yes - - 41 Senegal Min. of Environment / Protection of Nature Yes 1995 2001 42 Seychelles Ministry of Environment Yes - - 43 Sierra Leone Min. of Lands, …, and the Environment Yes 1995 1995 44 Somalia - - - - 45 South Africa Dept. of Environmental Affairs and Tourism Yes 1997 2000 46 Sudan Env. Research & Wildlife Dev. Agency - 1994 1996 47 Swaziland Min. of Tourism, Env., and Communications - 1997 1995 48 Tanzania Min. of Natural Resources and Tourism Yes 1996 1996 49 Togo National Meteorological Services of Togo Yes 1995 1996 50 Tunisia Min. of Environment & Regional Planning Yes 1994 1993 51 Uganda National Env. Management Authority Yes 1994 1993 52 Zambia Environmental Council of Zambia Yes 1994 1993 53 Zimbabwe Env. Management Agency - 1994 1995 Source: World Economic Indicators, 2003. * Environmental Action Plan Status as at 2003.

37

Appendix 2. African Countries Covered in the Study for Different Environmental Indicators (marked)

Country SPM CO2 OWP ASN ASW Algeria * * Angola * * * Benin * * * * Botswana * * * Burkina Faso * * Burundi * * * Cameroon * * * * Cape Verde * C.Afr. Republic * * * Chad * * Comoros * Congo * * * Congo (DRC) * * Cote d’Ivoire * * * Egypt * * * * Eq. Guinea * Ethiopia * * * Gabon * * Gambia * * Ghana * * * * Guinea * * * Guinea-Bissau * * * * Kenya * * * Lesotho * * Liberia * Madagascar * * * * Malawi * * * * Mali * * * Mauritania * * Mauritius * Morocco * * * * * Mozambique * Namibia * * Niger * * * * Nigeria * * * * Rwanda * * Sao T. & Principe * Senegal * * * * Seychelles * * Sierra Leone * * South Africa * * * Sudan * * * Swaziland * * Tanzania * * Togo * * * * Tunisia * * * Uganda * * Zambia * * * Zimbabwe * * *

Sources: World Development Indicators (World Bank) and, Gender and Environmental Indicators on African Countries (African Development Bank).

38

Appendix 3: SPM Summary Statistics for Cross-Sections of Selected African Countries (1990 – 2002)

Year Mean Std Dev. Min. Value Max. Value 1990 106.5965 58.33957 20..65835. 290..9602 1991 100.0955 55.99181 20.45422 274.2179 1992 98.43288 54.97791 21.06273 289.8728 1993 96.57594 52.84471 21.54999 278.0982 1994 91.69567 51.53528 18.74770 275.3971 1995 86.74005 49.32354 19.07402 277.7722 1996 81.93073 44.71985 20.15645 252.2695 1997 78.29003 42.88747 18.62464 239.7411 1998 75.51525 39.59751 19.71713 220.8533 1999 75.97452 39.38208 16.03642 206.1345 2000 74.20941 38.85613 13.34442 201.9959 2001 71.51760 37.95424 12.94789 200.4462 2002 68..82277 38.65634 11.91941 219.0774

Source: Authors’ computations from raw data

Appendix 4: CO2 Summary Statistics for Cross-Sections of Selected African Countries (1975 – 2002)

Year Mean SD Min. Value Max. Value 1975 10720.86 32508.78 54.96000. 181888.3 1976 11623.76 34362.20 73.28000 189905.1 1977 11791.75 34576.94 73.28000 191513.6 1978 12518.56 35367.27 76.94400 192231.8 1979 13387.70 38009.29 87.93600 206678.9 1980 14299.81 39351.18 95.26400 211266.2 1981 14496.23 42398.98 98.92800 234056.3 1982 14688.64 43620.17 84.27200 241736.1 1983 15428.44 46045.44 98.92800 256047.6 1984 16988.19 49916.63 98.92800 273546.9 1985 17283.31 50555.13 146.5600 277401.4 1986 17865.44 52238.30 150.2240 284487.6 1987 17886.98 51938.37 157.5520 283663.2 1988 18965.97 54525.69 62.28800 297813.6 1989 18183.65 53463.00 102.5920 296754.7 1990 17737.20 51836.52 113.5840 285480.6 1991 18250.16 54272.17 62.28800 299334.2 1992 18501.53 51709.24 76.94400 279735.4 1993 19428.30 54786.83 91.60000 296787.7 1994 19540.78 56878.11 95.26400 312506.2 1995 20111.81 59465.16 95.26400 425213.0 1996 20684.72 59763.11 102.5920 324696.3 1997 21252.87 61491.89 113.5840 333530.2 1998 21848.77 62455.91 113.5840 335387.9 1999 21800.36 61664.50 120.9120 332269.8 2000 21970.12 61255.06 54.96000 326770.2 2001 22403.69 62552.40 131.9040 332284.5 2002 23436.28 65037.72 131.9040 344819.0

Source: Authors’ computations from raw data

39

Appendix 5: OWP Summary Statistics for Cross-Sections of Selected African Countries (1980 – 2002)

Year Mean SD Min. Value Max. Value 1980 53159.57 90890.01 1306.570 237599.1 1981 55668.14 93674.97 1530.190 245623.4 1982 56847.37 94960.33 1806.950 249441.4 1983 55987.67 93190.67 1880.450 244901.4 1984 58081.83 94481.99 2121.610 249220.2 1985 57681.20 93608.27 2187.210 246725.6 1986 58798.79 92543.32 2700.910 245184.2 1987 60653.09 96656.45 2622.500 255515.1 1988 62814.86 98045.60 2967.900 259909.3 1989 64865.69 97945.72 4236.320 261827.9 1990 63221.14 98476.48 4509.080 261617.6 1991 66330.58 95865.78 4664.800 256686.4 1992 67903.09 95158.22 4003.060 253462.7 1993 70976.23 101406.30 3966.970 269343.3 1994 70546.37 98265.04 4150.160 262177.6 1995 72053.26 99269.87 4543.980 265534.2 1996 71858.80 97996.50 4386.050 262696.3 1997 71141.52 95969.90 4388.870 257030.5 1998 70124.07 94216.66 4634.620 251085.1 1999 70255.29 91710.22 5182.530 245891.2 2000 70170.10 91679.48 5398.670 245637.8 2001 63383.54 83170.63 5589.780 225079.9 2002 65265.59 82042.09 5203.990 221256.2

Source: Computed by Authors from raw data

Appendix 6: ASN Summary Statistics for Cross-Sections of Selected African Countries (1990, 1995, and 2000)

Country Mean SD Min. Value Max. Value Angola 0.743667 0.159124 0.560 0.84 Benin 0.778000 0.019287 0.764 0.80 Cameroon 0.173333 0.081445 0.080 0.230 Chad 0.773333 0.056862 0.710 0.820 Egypt 0.093333 0.035119 0.060 0.130 Ethiopia 0.896667 0.040415 0.850 0.920 Ghana 0.380000 0.010000 0.370 0.390 Guinea 0.433333 0.015275 0.420 0.450 Guinea-Bissau 0.620000 0.147309 0.530 0.790 Madagascar 0.610000 0.030000 0.580 0.640 Malawi 0.250000 0.020000 0.230 0.270 Mali 0.303333 0.005774 0.300 0.310 Morocco 0.330000 0.085440 0.250 0.420 Niger 0.826667 0.025166 0.800 0.850 Nigeria 0.422333 0.050163 0.370 0.870 Sudan 0.526667 0.220303 0.380 0.780 Tanzania 0.133333 0.030551 0.100 0.160 Togo 0.626667 0.035119 0.590 0.660 Zambia 0.326667 0.143643 0.220 0.490 Zimbabwe 0.396667 0.124231 0.320 0.540

Source: Derived by authors from the raw data. Each country value is obtained as average for the three sample points of 1990, 1995, and 2000, respectively.

40

Appendix 7: ASW Descriptive Statistics for Cross-Sections of Selected African Countries (1990, 1995, 2000)

Country Mean SD Min. Value Max. Value Angola 0.6500 0.0300 0.6200 0.6800 Benin 0.4220 0.0530 0.3700 0.4760 Burundi 0.3867 0.0666 0.3100 0.4300 Cameroon 0.4367 0.0551 0.3800 0.4900 Cape Verde 0.3600 0.1179 0.2600 0.4900 C. African Rep. 0.5800 0.2163 0.4000 0.8200 Congo 0.4767 0.0611 0.4100 0.5300 Cote d’Ivoire 0.2033 0.0252 0.1800 0.2300 Egypt 0.1567 0.1762 0.0500 0.3600 Ethiopia 0.7500 0.0100 0.7400 0.7600 Ghana 0.4200 0.0557 0.3600 0.4700 Guinea 0.5067 0.0513 0.4500 0.4700 Guinea-Bissau 0.5367 0.0833 0.4700 0.5500 Lesotho 0.3400 0.2326 0.0900 0.6300 Madagascar 0.5467 0.0153 0.5300 0.5500 Malawi 0.4000 0.0500 0.3500 0.5600 Morocco 0.2867 0.1289 0.1800 0.4500 Namibia 0.2567 0.0252 0.2300 0.2800 Niger 0.4533 0.0379 0.4100 0.4800 Nigeria 0.4670 0.0356 0.4300 0.5010 Senegal 0.2500 0.0300 0.2200 0.2800 Togo 0.4667 0.0208 0.4500 0.4900 Tunisia 0.1867 0.0777 0.1000 0.2500 Uganda 0.5267 0.0252 0.5000 0.5500

Source: Derived by authors from raw data. Each country value is obtained as average for the three sample points of 1990, 1995, and 2000, respectively.