African Journal of

Agricultural and Resource Economics

Volume 17, Number 4 (2022), pp 355–366

Determinants of bilateral trade flow between Ethiopia and its major

trading partners: A gravity model approach

Firomsa Mersha Tekalign*

School of Agricultural Economics and Agribusiness, Haramaya University, Haramaya, Ethiopia. E-mail:
firomer2008@gmail.com

Abule Mehare
Ethiopian Economic Association, Addis Ababa, Ethiopia. E-mail: abule.mehare@gmail.com

* Corresponding author

Received: December 2022

Accepted: March 2023

DOI: https://doi.org/10.53936/afjare.2022.17(4).24

Abstract

This study seeks to identify the internal and external factors determining Ethiopia’s bilateral exports
and total trade flows. It uses panel data covering 21 major trading partners of Ethiopia from 2000 to
2017 and estimates an augmented fixed effects gravity model. The results reveal that domestic and
foreign revenues increase Ethiopia’s bilateral exports, whereas the country’s foreign direct
investment and the population size of the trading partners decrease bilateral exports. The results
further show that both the domestic and foreign income and similarity endowment of Ethiopia
increase the country’s total trade. The study provides recommendations for the effective
implementation of supply side policies to enhance trade.

Key words: determinants, bilateral, trade flow, Ethiopia, gravity model

1. Introduction

Foreign trade is understood as a country’s trade with other countries. It is the legal exchange of
capital, goods and services across international borders or territories, consisting of imports and
exports flowing in and out of the country respectively. International trade arises because no country
can be completely self-sufficient. Over the years, international agricultural trade has allowed
countries to obtain the benefits of specialisation, such as increases in the output of goods and services
and obtaining those commodities and services that they do not produce, or do not produce in sufficient
quantities (Arene 2008).

As nations are generally not self-sufficient in the global economy, they are involved in trade on
various levels to sell what they produce in order to obtain what they require. In terms of conventional
economics theory, trade eventually promotes economic efficiency, and therefore enhances global
trade (International Monetary Fund [IMF] 2016). However, a combination of tariffs, quotas and
 AfJARE Vol 17 No 4 (2022) pp 355–366 Tekalign & Mehare

subsidies can act as economic and, in some cases, political barriers, imposing significant trade
restrictions (Caliendo et al. 2017).

By importing the required raw materials, intermediate and capital goods, consumer goods and
services, , if these goods and services are not available domestically, a country is able to enlarge its
productive capacity, foster export growth, meet the growing domestic demand and raise the living
standards and economic well-being of its populace. Exports, on the other hand, generate the foreign
exchange necessary to increase the import capacity of the country, and boost its industrialisation and
overall economic activity, which, in turn, augment its economic growth. Exports also enable the
expansion of markets and hence allow for economies of scale. In 2010, world trade recorded its largest
ever annual increase, as merchandised exports surged 14.5%. This was buoyed by a 3.6% recovery
in global output, after it took a major tumble in 2009, when it declined by 12%, with world gross
domestic product (GDP) also waning, but at a much lower rate of 2.4% (World Trade Organization
2011).

Growth in world trade, in turn, is the result of both technological developments and concerted efforts
to reduce trade barriers (IMF and World Bank 2001). In poor developing countries, agricultural trade
is important because most of the world’s poor live in rural areas, where agriculture is a key source of
income and consumption (United States Agency for International Development [USAID] 2010).

In Sub-Saharan Africa, the various nations’ share of goods exported to Europe fell from 31% in 2005
to 25% in 2010, and East Asia is rapidly replacing North America and Europe as Sub-Saharan Africa's
key trading partner in both intermediate and capital goods trade (World Bank 2022). Deep-rooted
structural problems, weak policy frameworks and institutions, protection at home and abroad and the
structure of African exports, which are characterised by dependence on primary commodities
(Alemayehu 2006; Biggs 2007; United Nations Conference for Trade and Development [UNCTAD]
2008), are considered the reasons for Africa’s poor export performance.

The performance of foreign trade in Ethiopia has increased significantly in recent times. The available
evidence shows that the value of both exports and imports improved tremendously, and government
has implemented many export-incentive packages besides the reduction in the tariff rate for the import
of raw materials and capital goods by the manufacturing sector. According to the annual report of the
National Bank of Ethiopia (NBE), Ethiopia’s total export earnings by value increased by 21% from
2019 to 2020 and, despite the decrease in imports from 2019 to 2020, the country’s imports have
increased steadily over the past decade, with a fivefold increase from 2007 to 2020. As a result,
Ethiopia faces a growing trade deficit, with total imports increasing by 12.5% per year on average
over the previous ten years (NBE 2021).

The trade deficit and its economic and social implications are matters of concern for both the public
and private sectors and therefore requires concerted efforts in terms of trade strategy. There is an
urgent need to address the trade deficit not only from the export side, but also from the import side,
by identifying products that can be produced locally to reduce the deficit. The export basket of the
country is concentrated on a few agricultural products such as coffee, oilseeds, pulses and semi-
processed leather. The export destinations of the country’s products are also very limited. On the
other hand, as a consequence of the growing domestic economy, the demand for consumer and capital
goods, as well as various services, is growing (Yeshineh 2016).

Ethiopia has a trade relationship with many of the nations of the world. The country’s exports
included coffee, leather and leather products to Europe (41.3%), Asia (31%), Africa (17.4%), and the
Americas (9.4%). The vast majority of Ethiopia’s imports (61.3%) come from Asia, followed by
Europe (22%), the Americas (7.6%) – of which the United States account for a sizable share (3.4%),

356
 AfJARE Vol 17 No 4 (2022) pp 355–366 Tekalign & Mehare

and other African countries (8.9%). Imports from China accounted for 22.8% of Ethiopia’s total
foreign supplies (NBE 2021).

Despite these opportunities, the country’s export performance is sub-optimal. Dealing with the
underperformance and constraints of the external trade sector, particularly the export sector, is critical
for maximising the country’s trade potential and leveraging trade to benefit the entire economy. To
optimise benefits from trade requires easing restrictions in order to take advantage of opportunities.
Given the potential benefits of trade, countries are eager to liberalise their economies in order to reap
these benefits of trade and globalisation via bilateral and multilateral means. It is critical that each
country understands its full trade potential in relation to other countries or regions in order to begin
the process of engagement.

The increasing volume and value of trade performance requires good trade policies based on reliable
information. In this regard, although there have been some studies on trade issues, they are not current
and not all trade factors have been explained sufficiently.

Therefore, the objective of this paper is to investigate the major factors influencing Ethiopian bilateral
export and total trade performance with its major trading partners and to provide a suitable
recommendation for the development of the export sector of the country.

2. Methodology

2.1 The gravity model

The gravity model explains export flows between two countries by their economic size (GDP or
GNP), population, and direct geographical distances between them. The gravity model predicts that
the flow of people, ideas or commodities between two locations is positively related to their size and
negatively related to their distance, based on Newton’s law of gravitation. They specified the
following gravity model equation in its original form:
𝐺𝐷𝑃𝑖𝑡 𝐺𝐷𝑃𝑗𝑡
𝐹𝑖𝑗𝑡 = 𝐶 , (1)
𝐷𝑖𝑗 2

where 𝐹𝑖𝑗𝑡 are the bilateral trade flows between country i and country j; C is the constant of the
equation; 𝐺𝐷𝑃𝑖𝑡 is the gross domestic product of country i; 𝐺𝐷𝑃𝑗𝑡 is the gross domestic product of
country j; and 𝐷𝑖𝑗 is the distance between the capitals of the two partner countries.

The gravity model in log linear form is typically used for empirical estimation, with the coefficients
representing the elasticity of bilateral trade to estimated parameters (Butt 2008). Taking this fact into
consideration, and applying it to the fundamental concept of the gravity model, yields the following
linear form of the equation:

𝑋𝑖𝑗𝑡 = 𝛼0 (𝐺𝐷𝑃𝑖𝑡 )𝛼1 (𝐺𝐷𝑃𝑗𝑡 )𝛼2 (𝐷𝐼𝑆𝑖𝑗 )𝛼3 , (2)

where 𝑋𝑖𝑗𝑡 is the trade flow between countries, which may be represented by total or average total
trade, or only the export or import flow of a country, 𝐺𝐷𝑃𝑖𝑡 and 𝐺𝐷𝑃𝑗𝑡 are the economic size of the
two countries, which can be represented by economic variables such as GDP, GNP, per capita GDP,
per capita GNP and population, and the variable 𝐷𝑖𝑗 stands for distance between trading countries as
a proxy for transportation cost.

357
 AfJARE Vol 17 No 4 (2022) pp 355–366 Tekalign & Mehare

To analyse the determinants of Ethiopia’s bilateral trade flows within the framework of the gravity
model, this study employed a panel dataset of annual observations on a cross-section of 21 major
trading partners of Ethiopia collected from different secondary sources over a period of 18 years,
from 2000 to 2017. The choice of the sample period and countries in the cross-section was influenced
by the availability of data on all the variables used in the study, and the relative importance of each
country (measured in terms of its percentage share) in Ethiopia’s total merchandise trade over the
sample period. Pooled ordinary least squares, fixed effect and random effect models were applied to
this panel dataset to investigate factors influencing trade.

2.2 Pooled ordinary least squares (OLS)

The simplest, and possibly naïve, estimation approach is the pooled OLS estimator, which proceeds
by essentially ignoring the panel structure of the data (the space and time dimensions of the pooled
data) and just estimates the usual OLS regression. The pooled specification can be written as:

𝑦𝑖𝑡 = 𝑋𝑖𝑡 𝛽 + 𝛼 + 𝑢𝑖𝑡 , (3)

where 𝑦𝑖𝑡 is the observation on the dependent variable for the cross-sectional unit (country) i in period
t, 𝑋𝑖𝑡 is the 1 x k vector of the explanatory variables observed for country i in period t, 𝛽 is a 1 x k
vector of parameters, and 𝑢𝑖𝑡 is an error or disturbance term specific to country i in period t. This
approach assumes that the intercept (𝛼) and all the coefficients (𝛽) are constant or identical for all
individuals across time, and that 𝑢𝑖𝑡 ~ 𝑖𝑖𝑑(0, 𝜎𝑢2 ) for all i and t, implying that the observations are
serially uncorrelated. Furthermore, the errors are homogenous across individuals and time. As
Gujarati (2004) indicates, these assumptions are highly restrictive, as the pooled regression ignores
the ‘individuality’ of each country and distorts the true picture of the relationship between the
dependent and independent variables.

2.3 The fixed effects model (FEM)

In the formulation of the fixed effects model, the intercept in the regression is allowed to differ among
individual units in recognition of the fact that each cross-sectional unit might have some special
characteristics of its own. That is, the model assumes that differences across units can be captured in
differences in the constant term. The 𝛼𝑖 are random variables that capture unobserved heterogeneity.
The model allows each cross-sectional unit to have a different intercept term even though all slopes
are the same, so that

𝑦𝑖𝑡 = 𝑥 ′ 𝑖𝑡 𝛽 + 𝛼𝑖 + 𝜇𝑖𝑡 , (4)

where 𝜀𝑖𝑡 is iid for all i and t.

The subscript i on the intercept term suggests that the intercepts across the individuals are different,
but that each individual intercept does not vary over time. The FEM is appropriate in situations where
the individual specific effect might be correlated with one or more regressors (Greene 2003; Gujarati
2003).

2.4 The random effects model (REM)

In contrast to the FEM, the random effects model (REM) assumes that the unobserved individual
effect is drawn randomly from a much larger population with a constant mean (Gujarat 2003). The
individual intercept is then expressed as a deviation from this constant mean value. The REM has an
advantage over the FEM in that it is economical in terms of degrees of freedom, since we do not have
358
 AfJARE Vol 17 No 4 (2022) pp 355–366 Tekalign & Mehare

to estimate N cross-sectional intercepts. The REM is appropriate in situations where the random
intercept of each cross-sectional unit is uncorrelated with the regressors. The basic idea is to start with
Equation (3). However, instead of treating β1i as fixed, it is assumed to be a random variable with a
mean value of β1. Then the value of the intercept for an individual entity can be expressed as:

𝛼𝑖 = 𝛼 + 𝜀𝑖 , where i = 1, 2, 3 … n. (5)

The random error term is assumed to be distributed with a zero mean and constant variance.
Substituting Equation (2) into Equation (1), the model can be written as:

𝑦𝑖𝑡 = 𝑥 ′ 𝑖𝑡 𝛽 + 𝛼 + 𝜀𝑖 + 𝜇𝑖𝑡
𝑦𝑖𝑡 = 𝑥 ′ 𝑖𝑡 𝛽 + 𝜔𝑖 . (6)

The composite error term, 𝜔𝑖𝑡 , consists of two components: 𝜀𝑖𝑡 is the cross-sectional or individual-
specific error component, and 𝑢𝑖𝑡 is the combined time-series and cross-sectional error component,
given that 𝜀𝑖 ~ (0, 𝜎𝜀2 ) and 𝜇𝑖 ~ (0, 𝜎𝜇2 ), where 𝜀𝑖 is independent of the 𝑋𝑖𝑡 (Gujarati 2003).

Generally, the FEM is held to be a robust method of estimating gravity equations, but it has the
disadvantage of not being able to evaluate time-invariant effects, which are sometimes as important
as time-varying effects. Therefore, for the panel projection of potential bilateral trade, researchers
have often concentrated on the REM, which requires that the explanatory variables be independent
of the 𝜀𝑖𝑡 and the 𝑢𝑖𝑡 for all cross-sections (i, j), and all time periods (Egger 2000). If the intention is
to estimate the impact of both time-variant and invariant variables in trade potential across different
countries, then the REM is preferable to the FEM (Ozdeser & Ertac 2010).

2.5 Model specifications

The gravity model in its most basic form explains bilateral trade (TTij) as being proportional to the
product of GDPi and GDPj, and inversely related to the distance between them. The static general
basic gravity model that we applied in this paper has the following log linear form:

𝑙𝑛𝑇𝑖𝑗 = 𝛽0 + 𝛽1 𝑙𝑛𝐺𝐷𝑃𝑖 + 𝛽2 𝑙𝑛𝐺𝐷𝑃𝑗 + 𝛽3 𝑙𝑛𝑊𝐷𝐼𝑆𝑇𝑖𝑗 + 𝜀𝑖 (7)

To account for other factors that may influence trade activities, other variables have been added to
the basic model to form the augmented gravity equation.

2.6 Definition of variables

The following definitions are provided for the variables used in this study.

2.6.1 Bilateral exports (X) and total trade (TT)

The above are the annual values (in US dollars) of Ethiopian exports to each of the 21 trading partners.
Bilateral exports are measured as the total value of all goods and services in US dollars flowing out
of Ethiopia to the given 21 trading partners. The total trade is the sum of bilateral exports and imports.
The data was collected mainly from the IMF DOTS 2019 CD-ROM.1

1
Available at http://www.imf.org/data

359
 AfJARE Vol 17 No 4 (2022) pp 355–366 Tekalign & Mehare

2.6.2 Gross domestic product (GDP)

The gross domestic product is the market value of the total production of goods and services in a
country. Data on the GDP of Ethiopia and its trading partners (in US dollars) was collected from the
FAOSTAT2 Outlook database.

2.6.3 Distance (WDIST)

Distance is the geographical distance between Addis Ababa (the capital city of Ethiopia) and the
capital cities of its trading partners, measured in kilometres (km). Data on distance was sourced from
an online distance calculator website, which can be found at www.distancefromto.net. Based on the
distance data and the GDP, as measured above, the weighted distance between Ethiopia and its trading
partners for each year in the observation period were calculated and used. A long distance between
Ethiopia and its trading partner would result directly in a high cost of transportation, which means
there will also be a reduction in demand for Ethiopian products and services, which implies this
variable is expected to have negative effect on trade.

2.6.4 Foreign direct investment (FDI)

Foreign direct investment is the total annual inward flow of international investment. FDI flows are
defined as investments that acquire a lasting management interest (10% or more of voting stock) in a
local enterprise by an investor operating in another country. Such investment is the sum of equity
capital, reinvestment of earnings, other long-term capital and short-term capital, as shown in the
balance of payments and both short-term and long-term international loans. Data on FDI inflows to
Ethiopia was sourced from the FAOSTAT database.

2.6.5 Real bilateral exchange rate (RBER)

The real bilateral exchange rate is the bilateral exchange rate between country i and country j at time
t. The depreciation of the real exchange rate enhances the competitiveness of domestic goods vis-à-
vis foreign goods. On the other hand, however, an appreciation in real exchange rate will decrease
the competitiveness of home goods in international markets. Data on the nominal real exchange rate
and price indices was collected from the FAOSTAT database.

2.6.6 Population (POPjt)

The effect of the population of the importing country is indeterminate where the absorption effects
and effects of the economies of scale are expected to affect their imports positively and negatively,
respectively.

2.6.7 Similarity endowment (SIMijt)

With reference to the similarity endowment, Linder (1961) says that the more similar the demand
structure of the two countries, the more intensive the trade between these two countries potentially
is. The traditional way of testing the similarity of demand structure or preferences, as suggested by
Linder, is by comparing the average (per capita) income of each country.

On the other hand, the Heckscher-Olin theorem postulates that trade patterns are determined by the
comparative advantage arising from differences in the relative factor endowments of different nations.

2
Available at http://www.fao.org/faostat

360
 AfJARE Vol 17 No 4 (2022) pp 355–366 Tekalign & Mehare

This difference in factor endowments of nations, in turn, produces differences in average income
across countries. Thus, by predicting that nations with dissimilar factor endowments will trade more
intensively with each other than countries with identical resource endowments, the Heckscher-Ohlin
hypothesis deductively also predicts that countries with dissimilar levels of per capita income will
trade more than countries with similar levels of per capita income. To summarise, a negative effect
of per capita GDP differential between Ethiopia and its partners on Ethiopia’s bilateral trade in this
study suggests that Ethiopia’s trade pattern follows the Linder hypothesis, whilst a positive effect
implies that the country’s trade pattern follows the H-O hypothesis.

2.7 Augmented gravity model for exports and total trade

The augmented gravity model that this paper used to estimate the determinants of trade is as follows:

𝑙𝑛𝑇𝑖𝑗𝑡 = 𝛽0 + 𝛽1 𝑙𝑛𝐺𝐷𝑃𝑖𝑡 + 𝛽2 𝑙𝑛𝐺𝐷𝑃𝑗𝑡 + 𝛽3 𝑙𝑛𝑊𝐷𝐼𝑆𝑇𝑖𝑗𝑡 + 𝛽4 𝑙𝑛𝐹𝐷𝐼𝑖𝑡 + 𝛽5 𝑙𝑛𝑅𝐵𝐸𝑅𝑖𝑗𝑡 +

𝛽6 𝑙𝑛𝑃𝑂𝑃𝑗𝑡 + 𝛽7 𝑙𝑛𝑆𝐼𝑀𝑖𝑡 + 𝜀𝑖𝑡 , (8)

where 𝑇𝑖𝑗𝑡 is total trade between country i and j at time t; GDPi and GDPj represent the GDP of
Ethiopia and the trading partners at current market prices (in USD) at time t; WDISTijt represents the
weighted distance between Ethiopia and her trading partner j at time t, which is defined as
(𝐷𝐼𝑆𝑇𝐼𝐽 𝑋𝐺𝐷𝑃𝑖𝑡 )
𝑊𝐷𝐼𝑆𝑇𝑖𝑗𝑡 = ∑ 𝐺𝐷𝑃
; FDI𝒊𝒕 represents FDI stock in Ethiopia (in USD) at time t; RBER ijt
𝑖𝑡
represents the real bilateral exchange rate between country i and j at time t, measured by the formula
𝑇𝐶𝑁
𝑖/$ 𝑗𝐶𝑃𝐼
𝑅𝐵𝐸𝑅𝑖𝑗𝑡 = (𝑇𝐶𝑁 ) 𝑥( 𝐶𝑃𝐼 ), where TCN is the nominal exchange rate vis-à-vis the dollar and CPI is
𝑖/$ 𝑖
the price index, notably the GDP deflator; POPjt is the total population of the trading partners at time
𝐺𝐷𝑃𝑖𝑡 𝐺𝐷𝑃𝑗𝑡
t, and SIM is defined as 1 − (𝐺𝐷𝑃 )2 − (𝐺𝐷𝑃 )2, which is the similarity in absolute factor
𝑖𝑡 +𝐺𝐷𝑃𝑗𝑡 𝑖𝑡 +𝐺𝐷𝑃𝑗𝑡
endowments between economies to test the Debaere transformation of the Helpman theorem.

In this paper, an attempt is made to devise a model for export and total trade to identify the major
determinants of bilateral trade. Thus, an estimation was done of the two trade models, as set out
below.

The bilateral export flow can be modelled as:

𝑙𝑛𝑋𝑖𝑗𝑡 = 𝛽0 + 𝛽1 𝑙𝑛𝐺𝐷𝑃𝑖𝑡 + 𝛽2 𝑙𝑛𝐺𝐷𝑃𝑗𝑡 + 𝛽3 𝑙𝑛𝑊𝐷𝐼𝑆𝑇𝑖𝑗𝑡 + 𝛽4 𝑙𝑛𝐹𝐷𝐼𝑖𝑡 + 𝛽5 𝑙𝑛𝑅𝐵𝐸𝑅𝑖𝑗𝑡 +

𝛽6 𝑙𝑛𝑃𝑂𝑃𝑗𝑡 + 𝛽7 𝑙𝑛𝑆𝐼𝑀𝑖𝑡 + 𝜀𝑖𝑡 , (9)

where all the variables are as defined above.

For the purpose of estimation, we modelled the bilateral total trade (exports plus imports) as follows:

𝑙𝑛𝑇𝑇𝑖𝑗𝑡 = 𝛽0 + 𝛽1 𝑙𝑛𝐺𝐷𝑃𝑖𝑡 + 𝛽2 𝑙𝑛𝐺𝐷𝑃𝑗𝑡 + 𝛽3 𝑙𝑛𝑊𝐷𝐼𝑆𝑇𝑖𝑗𝑡 + 𝛽4 𝑙𝑛𝐹𝐷𝐼𝑖𝑡 + 𝛽5 𝑙𝑛𝑅𝐵𝐸𝑅𝑖𝑗𝑡 +

𝛽6 𝑙𝑛𝑃𝑂𝑃𝑗𝑡 + 𝛽7 𝑙𝑛𝑆𝐼𝑀𝑖𝑡 + 𝜀𝑖𝑡 , (10)

where 𝑇𝑖𝑗𝑡 is total trade between country i and j at time t, and the other variables are as defined above.

361
 AfJARE Vol 17 No 4 (2022) pp 355–366 Tekalign & Mehare

3. Model estimation and interpretation of empirical results

The main concern of this section is to present and analyse the estimated results of the gravity models
of bilateral trade flow. The empirical analyses and discussions in this section are presented in two
main parts. The first of these presents the results of the pooled OLS, fixed effects (FE) and random
effects (RE) estimators for the determinants of Ethiopian exports. In the second part, the results of
the pooled OLS, fixed effects (FE) and random effects (RE) estimators for the determinants of
Ethiopian total trade (exports plus imports) are presented. The two parts are also devoted to the choice
of the appropriate estimator based on the Hausman test and a discussion of the results.

3.1 Analysis of the estimated pooled OLS, fixed effects and random effects models for Ethiopian
exports

In view of the nature of the dataset employed in this study, it was imperative that we select an
appropriate estimation method that accounts for the heterogeneity in the gravity models resulting
from the presence of individual and time effects in the panel data. In so doing, we first estimated the
pooled OLS model, along with the fixed effects (FE) and random effects (RE) models, with bilateral
exports as the regressand. The preliminary results of these models are presented in Table 1 below.

Table 1: Pooled OLS, fixed effects (FE) and random effects (RE) estimates of the augmented
gravity models of Ethiopia’s exports
Dependent variable: Ethiopia’s bilateral exports
Estimation method
Variables Pooled OLS Fixed effects model Random effects (GLS) model
lnGDPi 1.753*** 1.113*** 1.737***
(0.211) (0.366) (0.277)
lnGDPj -0.048 1.441*** 0.796***
(0.118) (0.230) (0.171)
lnWDIST -0.261 -0.273 -0.634**
(0.214) (0.366) (0.317)
lnFDI -0.216** -0.123* -0.160**
(0.085) (0.064) (0.066)
lnRBER 0.0147 -0.152 -0.284***
(0.042) (0.137) (0.082)
lnPOPj -0.192** -1.829*** -0.692***
(0.081) (0.619) (0.169)
lnSIM -0.549*** 0.015 -0.064
(0.072) (0.153) (0.126)
_cons -12.297** -1.906 -17.261***
(3.343) (10.492) (4.741)
Notes: ***,**, and * indicate statistical significance at the 1%, 5% and 10% error levels respectively. The values in
parenthesis are the standard errors associated with the parameters. The results were obtained with the aid of Stata13.
Source: Results from model, 2019

In estimating the FE model, we treated the country-specific effects as fixed. From the results, the
Hausman chi2 statistic for the export model is 34.38 (with a p-value of 0.0000). Since the associated
p-values are less than the 1% error level, the Hausman test strongly rejects the null hypothesis that
both the FE and RE estimators are consistent and that there is no significant difference between their
respective coefficients. In other words, this leads to strong rejection of the null hypothesis that RE
estimator provides consistent estimates. Thus, based on the Hausman test, we conclude that the FE
estimator is appropriate for the estimation of the export’s models. Consequently, the remainder of
this section is devoted to analysing the results of the gravity models of bilateral trade and exports as
yielded by FE estimator.

362
 AfJARE Vol 17 No 4 (2022) pp 355–366 Tekalign & Mehare

Table 2: Hausman specification test results for Ethiopia’s exports

Variables _Coefficients_
(b) (B) (b - B) sqrt (diag(V_b-v_B))
Difference
lnGDPi 1.113 1.737 -0.623 0.239
lnGDPj 1.441 0.796 0.644 0.154
lnWDIST -0.273 -0.634 0.361 0.183
lnFDI -0.123 -0.160 0.037
lnRBER -0.152 -0.284 0.132 0.110
lnPOPj -1.829 -0.692 -1.137 0.595
lnSIM 0.015 -0.064 0.079 0.087
Notes:
b = consistent under Ho and Ha; obtained from xtreg
B = inconsistent under Ho; efficient under Ho; obtained from xtreg
Test: Ho: difference in coefficients not systematic
chi2(6) = (b-B)'[(V_b-B_)^(-1)](b-B) = 34.38
Prob > chi2 = 0.0000
(V_b-V_B is not positive definite)

According to the FE estimates, Ethiopia’s bilateral trade export flows increase significantly with the
economic mass of its GDPi and the trading partners’ GDPj. The results fittingly concur with the
theoretical postulation of the gravity model of trade. Specifically, the results show that a 1% increase
in domestic income (GDPi) and foreign income (GDPj) significantly increases Ethiopia’s total
bilateral exports, by 1.11% and 1.44% respectively. This suggests that Ethiopia’s elasticities of
exports with respect to domestic and foreign incomes are highly elastic.

Foreign direct investment is found to exert a negative and significant impact on Ethiopia’s bilateral
exports. By increasing capital stock and enhancing the transfer of technology, new processes,
managerial skills and know-how in the domestic market, FDI is expected to result in a more efficient
utilisation of domestic resources and higher absorption of unemployed resources. This, in turn, will
lead to increased productivity, especially of the country’s comparative advantage export products.

The population size of the trading partners of Ethiopia was found to have a negative and statistically
significant effect on the bilateral exports of Ethiopia. The coefficients of population of the trading
partners imply that, all other things being equal, a 1% growth in population size of the trading partners
results in a decrease in bilateral exports by 1.83% (which is significant at the 1% error level).

3.2 Analysis of the estimated pooled OLS, fixed effects and random effects models for Ethiopia’s
total trade

To addresses the biased estimates of the pooled ordinary least square (POLS) estimator due to the
omission of country-specific effects, the results using the fixed effects (or within-group) and random
effects (generalised least squares (GLS)) estimators are presented in Table 3. From the results of the
Hausman specification, the Hausman chi 2 statistic for the export model is 37.88 (with a p-value of
0.0000) (Table 4). Since the associated p-values are less than the 1% error level, the Hausman test
strongly rejects the null hypothesis that both estimators are consistent and that there is no significant
difference between their respective coefficients. In other words, this leads to the strong rejection of
the null hypothesis that the RE estimator provides consistent estimates. Thus, based on the Hausman
test, we can again conclude that the FE estimator is appropriate for the estimation of the total trade
models. Consequently, the remainder of this section is devoted to analysing the results of the gravity
models of bilateral trade as yielded by the FE estimator.

363
 AfJARE Vol 17 No 4 (2022) pp 355–366 Tekalign & Mehare

Table 3: Pooled OLS, fixed effects (FE) and random effects (RE) estimates of the augmented
gravity models of Ethiopia’s total trade
Dependent variable: Ethiopia’s bilateral total trade
Estimation method
Variables Pooled OLS Fixed effects model Random effects (GLS) model
lnGDPi 0.815*** 0.517*** 0.909***
(0.134) (0.177) (0.147)
lnGDPj -0.062 0.917*** 0.578***
(0.075) (0.111) (0.091)
lnWDIST 0.143 -0.207 -0.333**
(0.136) (0.177) (0.167)
lnFDI -0.048 0.001 -0.012
(0.054) (0.031) (0.032)
lnRBER 0.009 0.089 -0.088*
(0.027) (0.066) (0.048)
lnPOPj 0.216*** -0.026 -0.097
(0.051) (0.299) (0.116)
lnSIM -0.270*** 0.223*** 0.145**
(0.046) (0.074) (0.067)
_cons -5.206** -12.716** -10.704***
(2.12) (5.067) (2.702)
***, ** and * indicate statistical significance at the 1%, 5% and 10% error level respectively. The values in parenthesis
are the standard errors associated with the parameters. The results were obtained with the aid of Stata13.
Source: Model results, 2019

Table 4: Hausman specification test result for Ethiopia’s total trade

Variables Coefficients_
(b) (B) (b - B) sqrt (diag(V_b-v_B))
Difference
lnGDPi 0.517 0.909 -0.392 0.098
lnGDPj 0.917 0.578 0.339 0.064
lnWDIST -0.207 -0.333 0.126 0.059
lnFDI 0.001 -0.012 0.013
lnRBER 0.089 -0.088 0.178 0.046
lnPOPj -0.026 -0.097 0.071 0.275
lnSIM 0.223 -0.145 0.079 0.030
Notes:
b = consistent under Ho and Ha; obtained from xtreg
B = inconsistent under Ho; efficient under Ho; obtained from xtreg
Test: Ho: difference in coefficients not systematic
chi2(7) = (b-B)'[(V_b-B_)^(-1)](b-B) = 37.88
Prob > chi2 = 0.0000
(V_b-V_B is not positive definite)

According to the FE estimates, Ethiopia’s bilateral total trade flows increased significantly with
Ethiopia’s GDP and its trading partners’ incomes (as measured by GDPj). The results follow the
theoretical postulation of the gravity model of trade. Specifically, the results show that a 1% increase
in domestic and foreign GDP significantly increases Ethiopia’s total bilateral trade, by 0.52% and
0.92% respectively.

Ethiopia’s similarity in endowment is also found to have a positive and statistically significant effect
on total trade in Ethiopia. The coefficients of Ethiopia’s similarity endowment imply that, all other
things being equal, a 1% growth in similarity endowment increases its total trade by 0.22% (which is
significant at the 1% error level).

364
 AfJARE Vol 17 No 4 (2022) pp 355–366 Tekalign & Mehare

4. Conclusions and policy implications

This study set out to analyse the determinants of Ethiopia’s bilateral trade flows within the gravity
model of trade, using panel data covering a cross-section of 21 major trading partners of Ethiopia for
the period 2000 to 2017. The estimation of the model was made using the fixed effects gravity model,
since it is an appropriate model according to the Hausman specification test.

The empirical results show that the gravity model is very successful in explaining the pattern of
Ethiopia’s bilateral trade flows. This is because the coefficients of the standard gravity variables
(domestic and foreign incomes and distance) were found to be consistent with the predictions of the
gravity model.

The results in relation to Ethiopia’s bilateral exports confirm that the elasticities of the conventional
gravity variables, domestic income and foreign income, where statistically significant and had their
theoretically stipulated sign. However, even though the geographical distance had its expected sign,
it was insignificant. This suggests that Ethiopia’s elasticities of export with respect to domestic and
foreign incomes are highly elastic. In addition to the basic gravity model variables, foreign direct
investment in Ethiopia and the population size of its trading partners were also statistically significant
in determining the country’s bilateral exports.

On the other hand, the results on the bilateral total trade of Ethiopia show that, as in the cast of
bilateral exports, they correspond with the theoretical postulation of the gravity model of trade. In
other words, Ethiopia’s bilateral total trade is positively and significantly determined by the country’s
productive capacity and foreign income, but not by the geographical distance between Ethiopia and
its trading partners. Furthermore, Ethiopia’s similarity endowments are found to exert negative effects
on the country’s total trade, while being statistically significant.

The main limitation of this study was that it examined the determinants of Ethiopia’s bilateral trade
flows using aggregated data on bilateral exports and total trade. However, the effective
implementation of the supply-side policies (such as increasing the productive capacity of the
agricultural sector) recommended in this study necessitates the identification and detailed
understanding of factors that significantly affect specific export sectors of Ethiopia. Thus, analysing
the bilateral flows of Ethiopian trade from within the gravity model using disaggregated data specific
to the sector can be considered in future studies.

Another limitation of the study is that it failed to examine Ethiopia’s trade potential with its partners.
That is, this study is unable to indicate with which countries Ethiopia has unexploited trade potential
and with which it has exhausted its trade potential. A consideration of this in future studies will help
the nation to identify the countries in which there are high prospects for expanding exports in order
to maximise its gains from bilateral trade.

References

Alemayehu G, 2006. Openness, inequality and poverty in Africa. DESA Working Paper No. 25,
United Nations Department of Economic and Social Affairs, New York NY.
Arene CJ, 2008. Agricultural economics: A functional approach. Abuja: Price Publishers.
Biggs T, 2007. Assessing export supply constraints: Methodology, data, and measurement. AERC
Research Project on Export Supply Response Capacity Constraints in Africa, Paper No.
ESWP_02, African Economic Research Consortium. Available at
http://publication.aercafricalibrary.org/bitstream/handle/123456789/375/2BiggsT_Assessing.pdf
?sequence=1&isAllowed=y

365
 AfJARE Vol 17 No 4 (2022) pp 355–366 Tekalign & Mehare

Butt WA, 2008. Pakistan’s export potential: A gravity model analysis. SBP Working Paper Series 23,
State Bank of Pakistan Research Department, Karachi, Pakistan.
Caliendo M, Schröder C, Fedorets A, Wittbrodt L & Preuss M, 2017. The short-run employment
effects of the German minimum wage reform. IZA Discussion Paper No. 11190, IZA Institute of
Labor Economics, Bonn, Germany.
Egger P, 2000. A note on the proper econometric specification of the gravity equation. Economic
Letters 66(1): 25–31.
Greene WH, 2003. Econometric analysis. 5th edition. Upper Saddle River NJ: Prentice Hall.
Gujarati DN, 2003. Basic econometrics. Fourth edition. New York NY: McGraw-Hill.
International Monetary Fund (IMF) and World Bank, 2001. Market access for developing countries’
exports. Available at http://imf.org/external/np/madc/eng/042701.pdf
International Monetary Fund (IMF), 2016. IMF annual report 2016. Available at
https://www.imf.org/external/pubs/ft/ar/2016/eng/pdf/ar16_eng.pdf
Linder SB, 1961. An essay on trade and transformation. New York: Wiley.
National Bank of Ethiopia (NBE), 2021. 2020/21 annual report. https://nbebank.com/wp-
content/uploads/pdf/annualbulletin/Annual%20Report%202020-2021/2020-
21%20Annual%20Report.pdf
Ozdeser H & Ertac D, 2010. Turkey’s trade potential with Euro zone countries: A gravity study.
European Journal of Scientific Research 43: 15–23.
United Nations Conference for Trade and Development (UNCTAD), 2008. Export competitiveness
and development in LDCs: Policies, issues and priorities for least developed countries for action
during and beyond UNCTAD XII. Geneva: United Nations. Available at
http://www.unctad.org/en/docs/aldc20081_en.pdf
United States Agency for International Development (USAID), 2010. Strategic review: Feed the
Future. USAID Report Paper. Available at https://pdf.usaid.gov/pdf_docs/Pdacr819.pdf
World Bank, 2022. Annual report 2022: Helping countries adapt to a changing world. Available at
https://www.worldbank.org/en/about/annual-report
World Trade Organization, 2011. World Trade Report 2011. Geneva: World Trade Organization.
Yeshineh AK, 2016. Determinants and potentials of foreign trade in Ethiopia: A gravity model
analysis. MPRA Paper No. 745090, Munich Personal RePEc Archive. Available at
https://mpra.ub.uni-muenchen.de/74509/

366