Climate and Development
ISSN: 1756-5529 (Print) 1756-5537 (Online) Journal homepage: https://www.tandfonline.com/loi/tcld20
The nexus of economic growth and environmental
degradation in Ethiopia: time series analysis
Mohammed Adem, Negasi Solomon, Saghi Movahhed Moghaddam,
Alexandru Ozunu & Hossein Azadi
To cite this article: Mohammed Adem, Negasi Solomon, Saghi Movahhed Moghaddam, Alexandru
Ozunu & Hossein Azadi (2020): The nexus of economic growth and environmental degradation in
Ethiopia: time series analysis, Climate and Development, DOI: 10.1080/17565529.2020.1711699
To link to this article:  https://doi.org/10.1080/17565529.2020.1711699
Published online: 31 Jan 2020.
Submit your article to this journal 
Article views: 26
View related articles 
View Crossmark data
Full Terms & Conditions of access and use can be found at
https://www.tandfonline.com/action/journalInformation?journalCode=tcld20
CLIMATE AND DEVELOPMENT
https://doi.org/10.1080/17565529.2020.1711699
The nexus of economic growth and environmental degradation in Ethiopia: time
series analysis
Mohammed Adema,b, Negasi Solomonc,d, Saghi Movahhed Moghaddame, Alexandru Ozunuf and Hossein Azadig,h,i
aDepartment of Economics, Mekelle University, Mekelle, Ethiopia; bDepartment of Agricultural Economics and Agribusiness, University of Ghana, Accra,
Ghana; cDepartment of Land Resources Management and Environmental Protection, Mekelle University, Mekelle, Ethiopia; dInstitute for Environment
and Sanitation Studies, University of Ghana, Accra, Ghana; eDepartment of Agroecology, Environmental Sciences Research Institute, Shahid Beheshti
University, Tehran, Iran; fFaculty of Environmental Science and Engineering, Babeş-Bolyai University, Cluj-Napoca, Romania; gDepartment of
Geography, Ghent University, Ghent, Belgium; hResearch Group Climate Change and Security, Institute of Geography, University of Hamburg,
Hamburg, Germany; iCzech University of Life Sciences Prague, Faculty of Environmental Sciences, Prague, Czech Republic
ABSTRACT ARTICLE HISTORY
Economic growth and environmental degradation are like two sides of a coin implying that these two Received 6 March 2019
variables are interdependent. Therefore, the main objective of this study is to investigate the Accepted 2 January 2020
relationship between economic growth and environmental quality. Moreover, this study attempted to
identify the relationships between the environmental quality, population growth, and economic KEYWORDSEconomic growth;
growth using a time series data from 1981 to 2012. The result showed that the Vector Error correction environmental quality; vector
model was appropriate to show both short-run and long-run relationships between variables. The error correction model;
results also indicated that the long-run economic growth helps to improve the environmental quality population growth
and/or reduce environmental degradation, while population growth worsens the environmental
degradation. This study concluded that a major cause of bio-diversity loss is the degradation of
vegetation in order to expand agriculture production for a rapidly growing population. Therefore, the
country should formulate an appropriate policy for monitoring population growth.
1. Introduction
Exploitation of these natural resources may generate large
Ethiopia issued a Climate Resilient Green Economy Strategy in economic benefits in the short term. However, in the long
2011, which aims at transforming Ethiopia into a carbon-neu- term, unsustainable use of these natural resources increases
tral middle-income country by 2025. Even though there has not not only environmental degradation, but it also decreases econ-
been a major legislative development with regard to the omic growth and livelihood opportunities.
environment in the last two years, the country issued a new The existing economic growth is based on agricultural devel-
investment law in 2012 which could have important ramifica- opment lead industrialization which has put mounting pressure
tions in relation to the environment and sustainable develop- on natural resources and environmental quality. If environ-
ment (Birhanu, 2014). In terms of institutional infrastructure, mental degradation is proportional to economic growth, the
one notable development has been the transformation of the only way to avoid environmental disasters is to limit economic
Ethiopian Environmental Protection Authority to a new Min- growth. If this theory is true, the future is gloomy for Ethiopia’s
istry, the Ministry of the Environment and Forests in 2013. environment. On the other hand, many economists optimisti-
The creation of this new institution, by upgrading the former cally believe in technological progress and reliance on the mar-
agency to a Ministry, appears to be an indication of an ket. Technology has a positive impact on resource conservation
increased willingness on the part of the government to give and pollution abatement, which might offset the adverse conse-
more attention to environmental protection. The expectation quences of population and income growth. The market mech-
is that the new Ministry will be able to make a meaningful anism dictates that the explicit or implicit price for
difference in environmental protection in the country by environmental goods and services will increase as the environ-
addressing the institutional and capacity challenges of the for- ment continues to deteriorate (Gradstein, 2004; Huber, Hunzi-
mer Environmental Protection Agency (EPA). ker, & Lehmann, 2011; Zhang, 2012).
During the last couple of years, Ethiopia has had a remark- According to the World Bank (2013), between 2004 and
able economic growth, and the country has recorded notable 2011, Ethiopia had achieved high economic growth averaging
progress towards many of the millennium development goals, 10.7 percent per year. As compared to the sub-Saharan African
including poverty reduction, access to primary education, and average of 5.4 percent during the same decade, the figure
access to health service. Improvements have also been made (10.7%) shows the existence of a great potential for further pro-
in basic infrastructure and in strengthening both regional and gress in the country. In 2012, Ethiopia was the 12th fastest
national policies and governance capacity. Ethiopia’s economy growing economy in the World, and the Bank forecasts that
is highly dependent on the exploitation of natural resources. if the country is able to continue this impressive growth
CONTACT Hossein Azadi hossein.azadi@ugent.be Department of Geography, Ghent University, Krijgslaan 281, S8, B-9000 Ghent, Belgium
© 2020 Informa UK Limited, trading as Taylor & Francis Group
2 M. ADEM ET AL.
performance, it could potentially reach middle income status Ethiopia is still classed as a least-developed country,
by 2025. The composition of economic growth source was although poverty and, to some extent, inequality are diminish-
through a mix of factors including agricultural modernization, ing notably in urban areas. With per-person GDP of only $380,
development of new export sectors, strong global commodity the proportion of the population below the poverty line in
demand, and government-led development investments 2010/2011 was 30.4% in rural areas and 25.7% in urban
(Allanson, 1996; World Bank, 2013). areas. Poverty has declined substantially according to the
Extensive contributions have been made to the literature on National Household Income Consumption Expenditure Sur-
the determinants of environmental degradation, which has vey. Trends in total, rural and urban Gini coefcients of the dis-
been centred around the nonlinearities in the relationship tribution of household consumption per capita were computed
between environmental indicators and average income. Much on nationally representative Household Income and Consump-
of the literature suggests that environmental degradation tion Expenditure Surveys (HICESs) conducted by the Central
initially increases in the early stages of economic development Statistical Agency of Ethiopia for 1995/1996, 1999/2000,
and then eventually declines with further growth beyond a 2004/2005, and 2010/2011 (MoFED, 2012).
national income threshold. This widely cited inverted U-shaped Over the period 1995–2011, the economy experienced only a
relationship between growth and environment is called the moderate structural transformation; over 1995–2005, the urban
Environmental Kuznets Curve (EKC) after Kuznets (1955) sector recorded a 10-point rise in urban Gini and a subsequent
who introduced an analogous relationship between economic 6-point fall. Moreover, the transformation of the economy over
growth and income inequality. 1995–2011 entailed a 14.0 percentage point decline in agricul-
The investigation of environmental degradation determi- tural employment, accompanied by a 4.3 employment rise in
nants has received considerable attention in the literature, but manufacturing (mostly urban), 2.0 in construction, and 6.7
the main focus has been on countries or regions with beneficial points in commerce and restaurants. No considerable changes
agroecological conditions and high technological adoption. were recorded in other sectors (UNDESA, 2017).
Thus, the main objective of this paper is to investigate the Ethiopia is one of the most well endowed countries in Sub-
relationship between economic growth and environmental Saharan Africa in terms of natural resources (Gete, Menale, &
degradation in the context of Ethiopia, which has received Mahmud, 2006). However, natural resource degradation in
less attention. Ethiopia has been going on for centuries. Similarly, Berry
Specifically, the paper attempts to show: (2003) stated that loss of land resource productivity is an
important problem in Ethiopia, and with continued population
(1) the link between economic growth and environmental growth, the problem is likely to be even more important in the
quality in Ethiopia and future. The major causes of land degradation in Ethiopia are
(2) the links among population growth and environmental rapid population increase, severe soil loss, deforestation, low
quality in Ethiopia. vegetative cover, and unbalanced crop and livestock production
(Girma, 2001).
Mulugeta (2004) argued that land degradation is a biophysi-
2. Literature review cal process driven by socioeconomic and political causes in
which subsistence agriculture, poverty, and illiteracy are impor-
Ethiopia is the second most populous country in Africa. The tant causes of land and environmental degradation in Ethiopia.
country’s dramatic and diverse landscape encompasses low- On the other hand, Gebreyesus and Kirubel (2009) reported
lands, deserts, canyons, and high plateaus. Its current climate that the heavy reliance of around 85% of Ethiopia’s growing
varies from very dry to very wet, and five types of agro-ecology population on an exploitative kind of subsistence agriculture
zones are identified: moisture-reliable humid lowlands, moist- is a major reason behind the current state of land degradation.
ure-sufficient highlands, drought-prone highlands, and arid Environmental degradation has already had visible impacts
lowland plains. on Ethiopia, with devastating droughts in some areas, associ-
Ethiopia is an ecologically diverse country of 1.1 million ated crises in food security, and health impacts. The impact
square kilometres, with a population of 99 million people grow- of environmental degradation on Ethiopia is already apparent
ing at 2.5% a year (UNDESA, 2017). The country occupies in the increasing temperature and declining rainfall, particu-
much of the Horn of Africa, a drought-prone area frequently larly in northern parts, which are exceptionally vulnerable to
affected by food crisis. Despite these structural handicaps, drought, with annual rainfall being only 100mm in the
between 2000 and 2011, Ethiopia recorded GDP growth of north-east. Such changes can be catastrophic for agricultural
8%, 6% growth in agriculture, a rapid increase in cereal pro- production; they can deteriorate infrastructure and worsen
duction, stable and low inequality, and declining poverty. the livelihoods of the rural poor (Bewket, 2009; Demeke,
The economy is highly dependent on agriculture and cur- Guta, & Ferede, 2004). For instance, evidence indicates that a
rent climatic variability as well as future climate change, and 10% decrease in seasonal rainfall compared with its long-
these issues pose challenges in achieving Ethiopia’s green econ- term average leads to a 4.4% decline in food production (Von
omy objectives. Ethiopia has a largely rural population with Braun, 1991). Studies made by Gebregziabher, Stageand, and
84% living in rural areas. 13.6 million people lived in urban Mekonnen (2011), Bewket (2009), World Bank (2008), and
areas in 2011/12, the majority in Addis Ababa, and urbaniz- Von Braun (1991) concluded that the frequency and intensity
ation was proceeding at a rapid 4.4% rate (Steve, Tadele, Sha- of drought are likely to increase over the coming decades,
mon, & Fikreyesus, 2013). which will present a serious threat to biodiversity, ecosystems,
CLIMATE AND DEVELOPMENT 3
water, and agricultural and human health. Because of the stra- (3) If B1 < 0 and B2 = 0, there is a monotonically decreasing
tegic importance of agriculture to the national economy, as well relationship between growth and environmental
as its sensitivity to water availability, this sector has been given degradation.
priority by the government. (4) If B1 > 0 and B2 < .0, there is an inverted-U-shaped
The most important environmental problems in Ethiopia relationship, i.e. Environmental Kuznet Curve.
include (i) land degradation, mainly due to physical soil (5) If B1 < 0 and B2 > 0, there is a U-shaped relationship.
loss and depletion of soil nutrients, overgrazing, and defor-
estation; (ii) environmental vulnerability due to climate 3.3. Estimation of models
variability causing droughts and floods; (iii) indoor air pol-
The stationary of the data was tested using the Augmented
lution; and (iv) water pollution. Other, somewhat less
Dickey Fuller (ADF) for unit root. This is because if a time
important, environmental problems include the potentially
series data is non-stationary, the study indicates only the
negative effects on livelihoods and food security of agro-
behavioural relationship at the time period under consider-
fuel investments, loss of biodiversity, spread of invasive
ation. A stationary (time) series is the one whose statistical
alien species, urban outdoor air pollution (mainly in
characteristics (such as mean, variance, and autocorrelation)
Addis Ababa), and toxic and household wastes (European
are all constant over time. Thus, a non-stationary series is the
Commission, 2007).
one whose statistical properties change over time (Chatfield,
2003). In this context, the ADF test is a unit root test for statio-
3. Materials and methods narity. Unit roots can cause unpredictable results in the time
series analysis. Therefore, each set of time series data can be
3.1. Data source used for a particular episode. As a consequence, it is not poss-
The data used for this study is secondary data. The data ible to generalize it to other time periods. Therefore, for the
obtained from the World Bank website contains a lot of purpose of forecasting, the non-stationary series may be of little
detailed country-specific information from 1981 to 2012. practical value.
In order to discover the long-run relationships between the
appropriate variables, a test for cointegration was conducted
3.2. Model specification using the Engle-Granger test, and the number of cointegration
Earlier analyses use a variety of speci cations to estimate the vectors was tested using the Johansen test. The Engle-Grangerfi
relationship between economic growth and environmental method first constructs residuals (errors) based on a static
degradation. Here the study develops the speci cations that regression. Residuals are tested for the presence of unit rootsfi
are consistent with both previous studies and theoretical frame- using ADF or similar tests. If the time series is merged, the
work. In the theoretical part of this research, the study speci es residuals will be practically stationary. A major issue with thefi
that environmental degradation uses a proxy variable of total Engle-Granger method is that the choice of the dependent vari-
CO emission per capita, which has a non-linear functional able may lead to different conclusions (Armstrong, 2001), as2
relationship with growth. Non-linear means that the graph is corrected by more recent tests such as Johansen’s. Johansen’s
not a straight line. The graph of a non-linear function is a test is another improvement over the Engle-Granger test. It
curved line that its direction constantly changes. Functional avoids the issue of choosing a dependent variable as well as
relationship refers to a class of statistical models in which a issues created when errors are carried from one step to the
functional relationship is assumed to exist between two arith- next. As such, the test can detect multiple cointegrating vectors.
metic variables, but the two arithmetic variables can only be In addition, different related diagnoses have been tested like
viewed with measurement error and/or natural variability hetroskedasticity, autocorrelation, and Ramsey model specifi-
(Kimura, 2000). In previous researches, the Environmental cation test. To be clear, Ramsey model is designed to detect
Kuznet Curve of conventional inverted-U shape is modelled whether there are any neglected nonlinearities in the model.
as a second-degree polynomial in logarithmic terms (Hilton The results of the data analyses and model diagnostic tests
& Levinson, 1998). Therefore, using this concept the model is were achieved using STATA 13.
specified as follows:
E = 4. Result and discussiont b0 + b 21Yt + b2Yt + . . . . . . . . . . . .+ biXt + ei (1)
4.1. Descriptive result analysis
where E is the total CO2 emission per capita, Y is the income
per capita, X refers to other factors (population growth rate), Before the investigation of econometric results, it would be bet-
the subscript t is a time index, and e is a normally distributed ter to look at the simple descriptive statistics of the basic expla-
error term. According to a prior expectation, the signs of the natory variables-GDP per capita and population growth rate
coefficients must be as follows: (Table 2, Appendix 1).
(1) If B1 = B2= 0, it means that there is no relationship 4.2. CO
between growth and environmental degradation. 2
emission level
(2) If B1 > 0 and B2 = 0, it means that there is a monotonical This paper considers a time series of data from 1981 to 2012.
increase or linear relationship between growth and The average CO2 emission level of the country for the study
environmental degradation. period is 0.0623 metric tons per capita. Additionally, Table 1
4 M. ADEM ET AL.
Table 1. Variables definition and possible effects on the dependent variable.
Expected
Variable name Definition sign
CO2 percapita Carbon dioxide emission is a dependent variable
which is used as proxy variable to measure
environmental degradation. It includes CO2
emission from the burning of fossil fuels and the
manufacture of cement. These include carbon
dioxide produced during consumption of solid,
liquid, and gas fuels and gas flaring. It is
measured in Metric tones Figure 2. GDP per capita growth rate.
GDPpercapita GDP per capita is gross domestic product divided +/−
by midyear population. GDP is the sum of gross
value added by all resident producers in the
economy plus any product taxes and minus any ruling party took the power from the dictatorial ‘Durge’ regime
subsidies not included in the value of the through military force.
products. Data are in constant 2005 U.S. dollars.
POPULgrowth Annual population growth rate for year t is the +
exponential rate of growth of midyear
population from year t−1 to t, expressed as a 4.4. Population growth
percentage. Population is based on the de facto
definition of population, which counts all Ethiopia is Africa’s second most populous nation. Current esti-
residents regardless of legal status or
citizenship. mation indicated that around 90 million people live in Ethiopia,
and over 80% of the population is living in rural areas (USAID,
2012). The average population growth rate for the study period
Table 2. Descriptive summary of the variable. indicated in Table 1 is 2.99%, and it varies from a minimum of
Variable Obs (years) Mean Min Max 2.4% to a maximum of 3.5% which implies the existence of a
CO2 32 0.0622894 .0377481 .088309
high population growth rate in the country. The main reason
GDPprercpita 32 159.0877 113.8756 273.687 for high population growth is the existance of a high fertility
POPULgrowth 32 2.98944 2.392229 3.51924 rate, and low contraceptive prevalence contributes to a high
Source: Own computation, 2015. annual population growth rate. This high population growth
rate in rural and urban areas contributes to environmental
indicated that the CO emission level varies from a range of degradation in different ways. However, as the graph below2
0.037 metric tons per capita to 0.0883 metric tons per capita. indicated, there is a declining trend for the population growth
Recently, the CO emission level has declined in comparison rate for the last two decades starting from the mid 1990s since2
with the last 15 years since the country gives more emphasis the current government designed and implemented a different
to climate change resilient green economy policy that focused population policy (Figure 3).
on the reduction of pollutant emission. This declining trend The population growth rate is another variable included in
was indicated in Figure 1 below. the model to show its impact on environmental degradation.
As a result, the high population growth rate has shown a signifi-
cant contribution to environmental degradation. This is con-
4.3. Gross domestic product per capita sistent with the outcome of another study by Drechsel,
According to César and Ekbom (2013), during the last decade, Gyiele, Kunze, and Cofie (2001), which confirmed a significant
Ethiopia has had remarkable economic growth rates with an relationship between population pressure, reduced fallow
increase in GDP by an average of 10% every year. This high periods, and soil nutrient depletion (including erosion), indi-
growth rate is expected to continue, and the GDP per capita cating a generally unsustainable dynamic between population,
will also rise in the same direction as the past decades. As agriculture, and the environment in Sub-Saharan Africa.
shown in Table 1 above, the average GDP per capita for the Shaw (2009) also found that rapid population growth as a prox-
period of 1981–2012 is USD 157.087, and it ranges from the imate cause of environmental degradation has shown to be
lowest of USD 113.87 to the maximum of USD 273.687 GDP more situation-specific for the Ethiopian highlands using quan-
per capita (Figure 2). During the mid 1990s, the GDP per capita titative methods. Based on a study by Cooper (2018), forests
reached its minimum point because at that time the country and wetlands are partly driven by population growth as land
faced severe political turmoil. During that period, the current
Figure 1. CO2 emission metric ton per capita. Figure 3. Population growth rate.
CLIMATE AND DEVELOPMENT 5
is cleared for agriculture or settlement. This land conversion The Johansen co-integration test was also carried out, and
has implications for ecosystem services, agricultural pro- the result indicated the existence of two co-integrating vectors
ductivity, and biodiversity in Uganda. As Kalipeni (1992) pre- between the variables of interest. Such results show that the
sented, rapid population growth and immigration resulted in variables have a long-term interaction relationship (i.e. there
massive environmental degradation in Malawi. Ray (2011) is a long-term relationship between economic growth, popu-
investigated the contribution of population pressure to land lation growth, and environmental quality) so that the vector
degradation and soil erosion and found that it affects the pro- error correction model suits the data set. However, the exist-
ductive resource base of the economy in India. However, Harte ence of the two cointegrating vectors can be seen as an issue
(2007) found that the widely assumed notion that environ- of identification that can be resolved by selecting a particular
mental degradation grows in proportion to population size – cointegrating vector, where the long-term estimate corresponds
assuming fixed per capita consumption and fixed modes of to that prediction by economic theory.
production – is shown to be overly optimistic. In particular, Additionally, the Lagrange multiplier test was carried out to
feedbacks, thresholds, and synergies generally amplify risk check whether or not the autocorrelation problem exists. The
and cause degradation to grow disproportionately faster than result of the Lagrange multiplier test showed that we failed to
the growth in population size (Dang, Li, Nuberg, & Bruwer, reject the null hypothesis of no autocorrelation with the
2014; Mancosu et al., 2015). variables. Similarly, the normality test of the disturbance term
and the stability test of the model were carried out, and
4.5. Econometrics analysis the results showed that the stability of the model was fine
(Appendix 5).
Stationary Test: According to Table 3 and Appendix 2, the null
hypothesis of no unit roots for all the time series is rejected at
the first differences. The reason is that the ADF test statistic 4.6. Vector error correction model estimation
values are less than the critical values at 1% level of significance
The study showed that all variables are cointegrated in the first
except for GDP per capita variable, which is significant at 5%.
difference or I (1). This indicates that there is a long run equi-
Thus, the variables are stationary and integrated in the same
librium or long run relationship among the variables. Of
order, i.e. I (1). In short, all variables have become stationary
course, in the short run, there may be disequilibrium. There-
and do not contain a unit root in the first difference.
fore, the error term can be treated as the ‘equilibrium error’.
Determination of Lags: As proposed by Hussain (2009), there
The study used this error term to tie the short-run behaviour
are different criteria to determine the number of lags. These are
of environmental degradation with long-run value. The error
Akaike Information Criterions (AIC), Hanna-Quinn Infor-
correction model states that the long run equilibrium depends
mation criterions (HQIC), and Schwarz Information Criterions
on the equilibrium error term. If the error term is nonzero and
(SBIS) that strongly advise the inclusion of the appropriate lag
positive, then the model is out of equilibrium. This means that
in the analysis. All the three lag selection criteria were used to
there is a very high possibility to be in equilibrium. Thus, the
choose the appropriate lag lengths, and the result rec-
study preferred to use the error correction model (ECM) rather
ommended to include three lags for all variables in the model
than the Autoregressive Distributed Lag Model (ARDLM).
(see Appendix 3).
The presence of cointegration among the variables indicates
Johansen Co-Integration Test: The co-integration rank is esti-
a long term relationship among the CO emission level, GDP
mated using the Johansen methodology. This methodology 2
per capita, and population growth rate. Therefore, the VEC
drives two likelihood estimators for the co-integration test,
model was applied to forecast the long run relationship
i.e. the trace and max-Eigen statistics. These statistics indicated
among these variables from 1981 to 2012.
the existence of two cointegrating equations among variables of
In Table 4 and Appendix 6, all the coefficients are significant
interests which means the vector error correction model fits
at a 1% level of significance and positively influenced the
with the data set and shows the long run relationship among
environmental quality except the GDP percapita variable. In
economic growth, population growth, and environmental qual-
addition, GDP per capita square is significant at a 1% level of
ity (see Appendix 4). Engle and Granger (1987) suggested that
significance but negatively related to the CO emission level
if co-integration exists among variables in the long run, then, 2
per capita in the long run. This result confirms the Environ-
there must be Granger-causality among these variables. Engel
mental Kuznets Curve hypothesis that in states with a lower
and Granger illustrated that the cointegrating variables must
level of economic growth, the economy positively contributes
have an error correction model representation, and the evi-
dence of co-integration excludes the possibility of the estimated
relationship being spurious. Table 4. Vector error correction model (Identification: beta is exactly identified,
Johansen normalization restrictions imposed).
Table 3. Augmented Dickey Fuller test. Beta Coef. Std. Err. Std. Err. P>|z|
ADF test _ce1
CO2 1 . . .
Variable I(0) Level I(1) First difference GDPpercapita .0038541 .0033450 1.152 0.272
CO2 −2.223 −6.383*** GDPprercpita2 −.0024588 .0004629 −5.31 0.000***
GDPpercapita 2.674 −3.193** POPULgrowth .1002098 .018132 5.53 0.000***
POPULgrowth −1.565 −5.221*** _cons −.724223 . . .
***Denote significant at 1% and **Significance at 5% using t-stat approach. ***Denote significant at 1%.
Source: Own computation. Source: Own computation, 2015.
6 M. ADEM ET AL.
to environmental pollution, whereas in the long run economic between CO2 emissions and GDP in both the short and long-
growth, it contributes to the reduction of environmental degra- run, thus supporting the EKC hypothesis.
dation. This means that at the early stage, economic growth However, our result was completely different from the result
inevitably contributes to environmental degradation. Later of Özokcu and Özdemir (2017) that was in contrast with the
on, environmental degradation starts to decrease with the EKC hypothesis, implying that environmental degradation can-
increase in economic growth. not be solved automatically by economic growth. Similarly,
Population growth rate is another variable included in the Almeida, Cruz, Barata, and García-Sánchez (2017) used a
model to show its influence on environmental degradation. panel data composed of 152 countries in a period of 6 years
Thus, the result shows that the high population growth rate and revealed that the EKC hypothesis is not proved, suggesting
has a positive and significant contribution to environmental that economic growth alone is not enough to improve environ-
degradation. This is because the high population growth rate mental quality. Jaunky (2011) attempted to test the EKC
creates pressure in both urban and rural areas, and this worsens hypothesis for 36 high-income countries for the period 1980–
the problem in rural areas since the pressure creates farming 2005. He found the evidence of an EKC in Greece, Malta,
land scarcity. In such a case, farmers in the rural area tried to Oman, Portugal, and the United Kingdom. However, it can
expand their farming land through clearing the existing forests be observed that for the whole panel, a 1% increase in GDP
which serve to absorb the CO2 emission. This resulted in an generates a 0.68% increase in CO2 emissions in the short-run
increase in the stock of CO2 level in the environment. and a 0.22% increase in the long-run. The lower long-run
Prior to the econometrics, model estimation of all necessary income elasticity does not provide evidence of an EKC but
tests of time series data structure has been done in order to get a does indicate that, over time, CO2 emissions are stabilizing in
valid, unbiased, and consistent estimate of the covariates. The rich countries. Al-mulali, Weng-Wai, Sheau-Ting, and
estimation results in Tables 4 and 5 (short-run and long-run Mohammed (2015) investigated the EKC hypothesis using a
result) revealed that GDP per capita has a negative significant country’s ecological footprint as an indicator of environmental
effect on the environmental quality, while in the short run, degradation for ninety-three countries, categorized by income.
GDP per capita has a positive significant effect which is consist- Their results do not support the EKC hypothesis in low- and
ent with our prior expectation (B1>0 and B2<0). The result con- lower-middle-income countries.
forms with the renowned Environmental Kuznets Curve (EKC)
hypothesis that at a lower level of economic growth, the econ-
4.7. Granger causality tests
omy has an adverse effect on environmental pollution, whereas
in the long run, economic growth supports the rehabilitation of The study used a Granger causality test to show the short run
environmental quality. This means that at the early stage, econ- effect of GDP per capita, and population growth rate on the
omic growth inevitably contributes to environmental degra- environmental degradation.
dation due to the dependency of the economy on agriculture Chi Square test statistics were constructed to show the short-
in Ethiopia. Later on, environmental degradation starts to run causal relationship between the CO2 emission level and
decrease with an increase in economic growth. This is in line other variables. Table 5 and Appendix 7 provide the results
with the results of Wolde (2015) who indicated that early of pairwise analyses. Significant probability values denote rejec-
stage economic growth positively contributes to environmental tion of the null hypothesis. This study rejects the null hypoth-
degradation, whereas later on, economic growth reduces esis if the probability value is more than 1%, otherwise the
environmental degradation in Ethiopia. Likewise, Kwabena study fails to reject the null hypothesis if the probability value
Twerefou, Danso-Mensah, and Bokpin (2017) found a positive is less than 1%. The study found that POPUL growth and
relationship between economic growth and environmental GDP per capita have ‘Granger cause’ on CO2 emission level
quality in the short term. Ahmad et al. (2017) also portrayed in the short-run. It means that the CO2 emission level follows
the link between economic growth and environmental quality, its mature counterparts in the short-run, and there exists a
and they proved the existence of an inverted U-shape relation- lead-lag relationship among population growth, GDP per
ship between CO2 emissions and economic growth. Narayan capita, and environmental quality. The remaining variables,
and Narayan (2010) also tested the applicability of EKC using i.e. GDP per capita has no causality effect in the short run on
longitudinal data in developing countries. They revealed that the level of CO2 emission.
carbon dioxide emissions have fallen over the long run as econ- The study found that POPUL growth and GDP per capita
omies have grown. Saboori, Sulaiman, and Mohd (2012) found have ‘Granger cause’ on CO2 emission level in the short-run.
the existence of a long-run relationship between per capita CO2 It means that the CO2 emission level follows its mature
emissions and real per capita Gross Domestic Product (GDP) counterparts in the short-run, and there exists a lead-lag
in Malaysia. They also found an inverted-U shape relationship relationship among population growth, GDP per capita, and
environmental quality. The remaining variables, i.e. GDP per
capita has no causality effect on the level of CO2 emission in
Table 5. Granger causality test. the short run. Similarly, Ahmad et al. (2017) revealed a bi-
Null hypothesis Chi2 Probability Decision directional causality between CO2 emissions and economic
GDPpercapita does not Granger-cause on CO2 11.964 0.000*** Reject growth in Croatia in the short run. However, Saboori et al.
GDPpercapita2 does not Granger cause CO2 0.98 0.6118 Accept
POPULgrowth does not Granger-cause on CO 485.49 0.0000*** Reject (2012) found an absence of causality between CO emissions2 2
***Denote significant at 1%. and economic growth in the short-run while demonstrating
Source: Own computation, 2015. uni-directional causality between economic growth and CO2
CLIMATE AND DEVELOPMENT 7
emissions in the long-run. Finally, the graph of the eigenvalues observations, experiments and household survey methods to study the
shows that none of the remaining eigenvalues appears close to role of forest management to climate change mitigation and livelihoods.
the unit circle. The stability check does not indicate that the Saghi Movahhed Moghaddam is a graduated scholar in agroecology from
model is mis specified. Shahid Beheshti University in Iran. Her research interests include ecologi-
cal modeling, sustainable natural resource management, climate change,
agroecology, sustainable development, and socio-economic indicators.
5. Conclusion Alexandru Ozunu a full professor and Dean of the Faculty of Environ-
mental Science and Engineering of Babeş-Bolyai University in Romania.
This study has attempted to explore the drivers of environ- His research interests include disaster and risk mitigation, sustainable
mental quality in Ethiopia, and it explores the link between development and environmental management. He is editor-in-chief of
economic growth and environmental quality (considering the Journal of Environmental Research and Protection and co-editor of
the CO emission level as a proxy for environmental quality). the Procedia Environmental Science, Engineering and Management Jour-2
It found a positive and signi cant relationship between nal. He also has a great deal of experience at the international level coor-fi
dinating numerous projects and multiple projects at national level in
environmental quality and GDP per capita in the short run Europe.
while they have an inverse relationship in the long term.
Hossein Azadi is a senior researcher at the Department of Geography,
This result is conformable with several studies in environ- Ghent University in Belgium. He was also a Fulbright fellow at the State
mental economics, including the EKC hypothesis. The University of New York at Binghamton. During his academic career, he
country should therefore pursue an existing environmental- has applied a variety of mixed methods on socio-economic impact assess-
friendly economic policy in order to strengthen its environ- ments of sustainable agri-rural development projects.
mental standards and participate in globalization. These
actions should be carried out through the use of cleaner
advances and products, as well as by reducing the deterio- References
ration of natural and fixed capital through investments in Ahmad, N., Du, L., Lu, J., Wang, J., Li, H.-Z., & Hashmi, M. Z. (2017).
green infrastructure and nature-based projects, while at the Modelling the CO2 emissions and economic growth in Croatia: Is
same time increasing expenditure on education and combat- there any environmental Kuznets curve? Energy, 123, 164–172. doi:10.
ing ecological crises resulting from environmental degra- 1016/j.energy.2016.12.106
dation. Therefore, creating a consistent, coherent, and Allanson, P. (1996). Agricultural land use change in England and Wales,
1892–1992. Journal of Environmental Planning and Management, 39
effective environmental policy framework is essential in (2), 243–254.
order to improve environmental quality that supports well- Almeida, T. A. D. N., Cruz, L., Barata, E., & García-Sánchez, I.-M. (2017).
being and enables long-term economic development. Economic growth and environmental impacts: An analysis based on a
Population growth rates is one of the main threats of composite index of environmental damage. Ecological Indicators, 76,
environmental quality in the long run. Furthermore, the rising 119–130. doi:10.1016/j.ecolind.2016.12.028
Al-mulali, U., Weng-Wai, C., Sheau-Ting, L., & Mohammed, A. H. (2015).
population is a major source of environmental degradation. Investigating the environmental Kuznets curve (EKC) hypothesis by
Population affects the environment through the use of natural utilizing the ecological footprint as an indicator of environmental
resources and production of wastes. These lead to loss of bio- degradation. Ecological Indicators, 48, 315–323. doi:10.1016/j.ecolind.
diversity and air and water pollution and increased pressure 2014.08.029
on land. Excessive deforestation and overgrazing by the grow- Armstrong, J. (2001). Principles of forecasting: A handbook for researchers
and practitioners. Springer Science & Business Media. Springer Nature
ing population have led to land degradation. A major cause of Switzerland AG. doi:10.1007/978-0-306-47630-3
the loss of biodiversity has been the depletion of vegetation in Berry, L. (2003). Land degradation in Ethiopia: Its impact and extent. In L.
order to expand agriculture by the rapidly rising population. Berry, J. Olson, & D. Campbell (Eds.), Assessing the extent, cost and
Thus, the country should design an appropriate population impact of land degradation at the national level: Findings and lessons
growth policy. learned from seven pilot case studies (p. 204). Ethiopia: Commissioned
by global mechanism with support from the World Bank.
Bewket, W. (2009). Rainfall variability and crop production in Ethiopia
Disclosure statement case study in the Amhara region. In S. Ege, H. Aspen, B. Teferra, &
S. Bekele (Eds.), Proceedings of the 16th international conference of
No potential conflict of interest was reported by the authors. Ethiopian studies. Trondheim: Ethiopian Journal of Development
Research.
Birhanu, F. M. (2014). Green growth, investment, environment and sustai-
Notes on contributors nabledevelopment in Ethiopia: Country Report of Ethiopia. Addiss
Ababa: 5 IUCNAEL EJournal.
Mohammed Adem is a senior researcher and lecturer at the Department of César, E., & Ekbom, A. (2013). Ethiopia environmental and climate change
Economic, College of Business and Economics Mekelle University in policy brief.
Ethiopia. His research interests include resources and environmental econ- Chatfield, C. (2003). The analysis of time series: An introduction. United
omics, production economics technology adoption, food security, impact States: Chapman & Hall/CRC.
and poverty analysis. In his studies, he also focuses on field observations, Cooper, R. (2018). Current and projected impacts of renewable natural
experiments and household survey methods to explore various socio-econ- resources degradation on economic development in Uganda (K4D emer-
omic issues. ging issues report). Brighton: Institute of Development Studies.
Negasi Solomon is a senior researcher and lecturer at the Department of Dang, H. L., Li, E., Nuberg, I., & Bruwer, J. (2014). Understanding farmers’
Land Resources Management and Environmental Protection, Mekelle adaptation intention to climate change: A structural equation modelling
University in Ethiopia. His researches focus on the interface between study in the Mekong Delta, Vietnam. Environmental Science & Policy,
land resource management and climate change mitigation and adaptation. 41, 1–10.
He is particularly interested in how dry forest management contributes to Demeke, M., Guta, F., & Ferede, T. (2004). Agricultural development and
climate change mitigation and adaptation. He also performs field food security in sub-Saharan Africa: Building a case for more public
8 M. ADEM ET AL.
support-the case of Ethiopia working paper no. 02, policy assistance Research in International Business and Finance, 42, 939–949. doi:10.
unit of the FAO sub-regional office for East and Southern Africa, 2006. 1016/j.ribaf.2017.07.028
Drechsel, P., Gyiele, L., Kunze, D., & Cofie, O. (2001). Population density, Mancosu, E., Gago-Silva, A., Barbosa, A., de Bono, A., Ivanov, E.,
soil nutrient depletion, and economic growth in sub-Saharan Africa. Lehmann, A., & Fons, J. (2015). Future land-use change
Ecological Economics, 38(2), 251–258. doi:10.1016/S0921-8009 scenarios for the Black Sea catchment. Environmental Science &
(01)00167-7 Policy, 46, 26–36.
Engle, R. F., & Granger, C. J. (1987). Co-integration and error correction: MoFED. (2012). Ethiopia’s progress towards eradicating poverty: An
Representation, estimation, and testing. Econometrica, 55(2), 251–276. interim report on poverty analysis study (2010/11). Addis Ababa:
European Commission. (2007). Ethiopia – country environmental profile. IFPRI E-BARAY.
Addis Ababa: European Commission. Mulugeta, L. (2004). Effects of land use change on soil quality and native
Gebregziabher, Z., Stageand, J., & Mekonnen, A. (2011). Climate change flora degradation and restoration in the highlands of Ethiopia.
and the Ethiopian economy: A computable general equilibrium analysis Implication for sustainable land management (Ph.D Thesis). Uppsala:
(Discussion paper series EfD DP 11-09). Environment for Swedish university of Agricultural Science.
Development. Narayan, P. K., & Narayan, S. (2010). Carbon dioxide emissions and econ-
Gebreyesus, B., & Kirubel, M. (2009). Estimating soil loss using universal omic growth: Panel data evidence from developing countries. Energy
soil loss equation (USLE) for soil conservation planning at Medego Policy, 38(1), 661–666. doi:10.1016/j.enpol.2009.09.005
Watershed, Northern Ethiopia. Journal of American Science, 5(1), Özokcu, S., & Özdemir, Ö. (2017). Economic growth, energy, and environ-
58–69. mental Kuznets curve. Renewable and Sustainable Energy Reviews, 72,
Gete, Z., Menale, K. P., & Mahmud, Y. (2006). Stakeholder analysis for 639–647. doi:10.1016/j.rser.2017.01.059
sustainable land management (SLM) in Ethiopia: Assessment of Ray, S. (2011). Impact of population growth on environmental degra-
opportunities, strategic constraints, information needs, and knowledge dation: Case of India. Journal of Economics and Sustainable
gaps. Environmental Economics Policy Forum for Ethiopia (EEPFE), 1, Development, 2(8).
4–91. Saboori, B., Sulaiman, J., & Mohd, S. (2012). Economic growth and CO2
Girma, T. (2001). Land degradation: A challenge to Ethiopia. Addis Ababa: emissions in Malaysia: A cointegration analysis of the environmental
International Livestock Research Institute. pp. 815–823. Kuznets curve. Energy Policy, 51, 184–191. doi:10.1016/j.enpol.2012.
Gradstein, M. (2004). Governance and growth. Journal of Development 08.065
Economics, 73(2), 505–518. Shaw, R. P. (2009). Rapid population growth and environmental degra-
Harte, J. (2007). Human population as a dynamic factor in environmental dation: Ultimate versus proximate factors. Environmental
degradation. Population and Environment, 28(4), 223–236. doi:10.1007/ Conservation, 16(3), 199–208. doi:10.1017/S0376892900009279
s11111-007-0048-3 Steve, B., Tadele, F., Shamon, W., & Fikreyesus, D. (2013). Making growth
Hilton, F., & Levinson, A. (1998). Factoring the environmental Kuznets green and inclusive: The case of Ethiopia’ OECD green growth papers.
curve: Evidence from automobile lead emissions. Journal of Paris: OECD Publishing.
Environmental Economics and Management, 35(2), 126–141. United Nations Department of Economic and Social Affairs
Huber, R., Hunziker, M., & Lehmann, B. (2011). Valuation of agricultural (UNDESA) Population Division. (2017). World population prospects,
land-use scenarios with choice experiments: A political market share the 2017 revision. Retrieved from https://www.un.org/development/
approach. Journal of Environmental Planning and Management, 54 desa/publications/world-population-prospects-the-2017-revision.
(1), 93–113. html
Hussain, K. (2009). Monetary policy channels of Pakistan and their impact United State Agency Internationa Development. (2012). Country develop-
on real GDP and inflation (CID graduate student working paper series ment cooperation strategy, 2011–2015.
no. 40). Center for International Development at Harvard University. Von Braun, J. (1991). A policy agenda for famine prevention in Africa
Jaunky, V. C. (2011). The CO2 emissions-income nexus: Evidence from (Food policy statement no. 13). Washington, DC: IFPRI.
rich countries. Energy Policy, 39(3), 1228–1240. doi:10.1016/j.enpol. Wolde, E. T. (2015). Economic growth and environmental degradation in
2010.11.050 Ethiopia: An environmental Kuznets curve analysis approach. Journal
Kalipeni, E. (1992). Population growth and environmental degradation in of Economics and International Finance, 7(4), 72–79. doi:10.5897/
Malawi. Africa Insight, 22(4), 273–282. JEIF2015.0660
Kimura, D. K. (2000). Using nonlinear functional relationship regression World Bank. (2008). Ethiopia: A country study on the economic impacts of
to fit fisheries models. Canadian Journal of Fisheries and Aquatic climate change (Environment and natural resource management report,
Sciences, 57, 160–170. no. 46946-ET). Washington, DC: World.
Kuznets, S. (1955). Economic growth and income inequality. The World Bank. (2013). Ethiopia economic update II: Laying the foundation
American Economic Review, 45(1), 1–28. for achieving middle income status. Addis Ababa: Public Disclosure
Kwabena Twerefou, D., Danso-Mensah, K., & Bokpin, G. A. (2017). The Authorize.
environmental effects of economic growth and globalization in sub- Zhang, J. (2012). Delivering environmentally sustainable economic growth:
Saharan Africa: A panel general method of moments approach. The case of China.
CLIMATE AND DEVELOPMENT 9
Appendices
Appendix 1: Summary of descriptive statistics
. sum CO2 GDPprercpita GDPpercapit2 POPULgrowth
Variable Obs Mean Std. Dev. Min Max
CO2 32 .0622894 .0377481 .0123237 .088309
GDPprercpita 32 159.0877 39.37845 113.8756 273.687
GDPpercapit2 32 26811.09 15051.96 12967.65 74904.59
POPULgrowth 32 2.98944 .3053983 2.392229 3.51924
Appendix 2: ADF test
. dfuller CO2,lags(0)
Dickey-Fuller test for unitroot Number of obs = 31
——— Interpolated Dickey-Fuller ———
Test statistic 1% Critical value 5% Critical value 10% Critical value
Z(t) −2.223 −3.709 −2.983 −2.623
MacKinnon approximate p-value for Z(t) = 0.1981
. dfuller GDPpercpita,lags(0)
Dickey-Fuller test for unit root Number of obs = 31
——— Interpolated Dickey-Fuller ———
Test statistic 1% Critical value 5% Critical value 10% Critical value
Z(t) 2.674 −3.709 −2.983 −2.623
MacKinnon approximate p-value for Z(t) = 0.9991
. dfuller POPULgrowth,lags(0)
Dickey-Fuller test for unit root Number of obs = 31
——— Interpolated Dickey-Fuller ———
Test statistic 1% Critical value 5% Critical value 10% Critical value
Z(t) −1.565 −3.709 −2.983 −2.623
MacKinnon approximate p-value for Z(t) = 0.5010
. dfuller DCO2,lags(0)
Dickey-Fuller test for unit root Number of obs = 30
——— Interpolated Dickey-Fuller ———
Test statistic 1% Critical value 5% Critical value 10% Critical value
Z(t) −6.383 −3.716 −2.986 −2.624
MacKinnon approximate p-value for Z(t) = 0.0000
. dfuller DGDPprercpita,lags(0)
Dickey-Fuller test for unit root Number of obs = 30
——— Interpolated Dickey-Fuller ———
Test statistic 1% Critical value 5% Critical value 10% Critical value
Z(t) −3.893 −3.716 −2.986 −2.624
MacKinnon approximate p-value for Z(t) = 0.0204
. dfuller DPOPULgrowth,lags(0)
Dickey-Fuller test for unit root Number of obs = 30
——— Interpolated Dickey-Fuller ———
Test statistic 1% Critical value 5% Critical value 10% Critical value
Z(t) −5.221 −3.716 −2.986 −2.624
MacKinnon approximate p-value for Z(t) = 0.0000
10 M. ADEM ET AL.
Appendix 3: Lag selection
. varsoc CO2 GDPprercpita GDPpercapit2 POPULgrowth, maxlag(3)
Selection-order criteria
Sample: 1984–2012 Number of obs = 29
Lag LL LR df p FPE AIC HQIC SBIC
0 −398.236 822533 27.8094 27.8832 28.0451
1 −251 294.47 25 0.000 184.978 19.3793 19.8223 20.7938
2 −199.478 103.04 25 0.000 35.1729 17.5502 18.3624 20.1434
3 −128.275 142.41* 25 0.000 2.37679* 14.3638* 15.5451* 18.1357*
Endogenous: CO2 GDPprercpita GDPpercapit2 INDUSTgrowth POPULgrowth
Exogenous: _cons
Appendix 4: Co-integration test
. vecrank CO2 GDPprercpita POPULgrowth, trend(trend) max
Johansen tests for cointegration
Trend: trend Number of obs = 30
Sample: 1983–2012 Lags = 2
Maximum rank Parms LL Eigenvalue Trace statistic 5% Critical value
0 24 −45.594762 . 94.6427 54.64
1 31 −21.818506 0.79507 47.0902 34.55
2 36 −2.2097884 0.72944 7.8727* 18.17
3 39 1.6668001 0.22774
Maximum rank Parms LL Eigenvalue Max statistic 5% Critical value
0 24 −45.594762 . 47.5525 30.33
1 31 −21.818506 0.79507 39.2174 23.78
2 36 −2.2097884 0.72944 7.7532 16.87
3 39 1.6668001 0.22774
Appendix 5: Stablity test
.vecstable, graph
Eigenvalue stability condition
Eigenvalue Modulus
1.220589 1.22059
1 1
1 1
.7399887+.5129264i .900376
.7399887−.5129264i .900376
−.2872844–.7749413i .826478
−.2872844–.7749413i .826478
−.06337681+.7539965i .756655
−.06337681–.7539965i .756655
.6174076 .617408
−.3979601 .39796
.2207236 .220724
The VECM specification imposes 2 unit moduli.
CLIMATE AND DEVELOPMENT 11
Appendix 6: Vector error correction model result
. vec CO2 GDPpercapita GDPprercpita2 POPULgrowth,trend(constant)rank(2)lags(3)
Vector error-correction model
Sample: 1984–2012 No. of obs = 29
AIC = 15.35134
Log likelihood =−151.5944 HQIC = 16.39974
Det(Sigma_ml) = .0238828 SBIC = 18.69886
Equation Parms RMSE R-sq chi2 P > chi2
D_CO2 11 .003748 0.9031 139.7563 0.0000
D_GDPpercapita 11 2698.82 0.7816 53.68113 0.0000
D_GDPprercpita2 11 9.20797 0.6590 28.99243 0.0066
D_POPULgrowth 11 .013442 0.9775 650.7887 0.0000
Coef. Std. Err. z P>|z| [95% Conf. Interval]
D_CO2
ce1– −1.307262 .1696029 −7.71 0.000 −1.639677 −.9748464
L1.
Ce2– −7.87e–06 9.80e–07 −8.03 0.000 −9.79e–06 −5.95e–06
L1.
CO2
LD. .1741836 .1029986 1.69 0.091 −.0276899 .3760571
L2D. .09578 .0944214 1.01 0.310 −.0892824 .2808425
GDPpercapita
LD. 2.68 7.32e–06 2.73e–06 0.007 1.96e–06 .0000127
L2D. .0000159 3.23e–06 4.94 0.000 9.62e–06 .0000223
GDPprercpita2
LD. −.0023645 .0008024 −2.95 0.003 −.0039372 −.0007918
L2D. −.0046965 .0009253 −5.08 0.000 −.0065102 −.0028829
POPULgrowth
LD. .086624 .0323866 2.67 0.007 .0231474 .1501005
L2D. −.1407006 .031825 −4.42 0.000 −.2030764 −.0783249
cons .0106587 .0015593 6.84 0.000 .0076025 .0137149
D_GDPpercapita
ce1– 49160.64 122111.8 0.40 0.687 −190174.2 288495.4
L1.
Ce2– .0870647 .7053679 0.12 0.902 −1.295431 1.46956
L1.
CO2
LD. −35194.38 74157.61 −0.47 0.635 −180540.6 110151.9
L2D. −42415.88 67982.13 −0.62 0.533 −175658.4 90826.65
GDPpercapita
LD. 1.972897 1.96862 1.00 0.316 −1.885527 5.831321
L2D. −1.789102 2.322911 −0.77 0.441 −6.341924 2.76372
(Continued )
12 M. ADEM ET AL.
Continued.
Coef. Std. Err. z P>|z| [95% Conf. Interval]
GDPprercpita2
LD. −561.0311 577.7294 −0.97 0.332 −1693.36 571.2977
L2D. 324.5863 666.2294 0.49 0.626 −981.1993 1630.372
POPULgrowth
LD. −17425.89 23317.92 −0.75 0.455 −63128.18 28276.4
L2D. −2800.475 22913.56 −0.12 0.903 −47710.22 42109.27
_cons .0027853 1122.688 0.00 1.000 −2200.425 2200.43
Appendix 7: Granger causality test
. test ([D_GDPpercapita]: LD.GDPpercapita L2D.GDPpercapita)
(1) [D_GDPpercapita]LD.GDPpercapita = 0
(2) [D_GDPpercapita]L2D.GDPpercapita = 0
chi2( 2) = 11.964
Prob > chi2 = 0.000
test ([D_GDPprercpita]: LD.GDPprercpita2 L2D.GDPprercpita)
(1) [D_GDPprercpita2]LD.GDPprercpita2 = 0
(2) [D_GDPprercpita2]L2D.GDPprercpita2 = 0
chi2( 2) = 0.98
Prob > chi2 = 0.6118
test ([D_POPULgrowth]: LD.POPULgrowth L2D.POPULgrowth)
(1) [D_POPULgrowth]LD.POPULgrowth = 0
(2) [D_POPULgrowth]L2D.POPULgrowth = 0
chi2( 2) = 485.49
Prob > chi2 = 0.0000