See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/340938630

The Macroeconomic and Social Impact of COVID-19 in Ethiopia and Suggested

Directions for Policy Response

Preprint · April 2020

CITATIONS READS

0 11,670

1 author:

Alemayehu Geda
Addis Ababa University
115 PUBLICATIONS   771 CITATIONS   

SEE PROFILE

Some of the authors of this publication are also working on these related projects:

Growth and Poverty in Fragile State of Africa (With African Economic Research Consortium). I am currently coordinating this project which has about 14 country case studies
View project

Modelling the Trade Effect of the Continental Free Trade Area (AfCFTA) Agreement Among African Countries View project

All content following this page was uploaded by Alemayehu Geda on 26 May 2020.

The user has requested enhancement of the downloaded file.

 1 
 

 
 
The COVID-19 Damage on the Ethiopian Department of Economics
Addis Abeba University
Service Sector: A Supplement Using  
 
Google Search Trend “Big data”  
 
 
 

kk
Alemayehu Geda
([email protected])

Professor of Economics, Department of Economics, Addis Ababa University

Research Associate, School of Oriental and African Studies, University of London &
African Economic Research Consortium, AERC, Nairobi

A Service Sector Supplement

 
Department of Economics
Addis Abeba University
May 2020
 

 
 2 
 

The COVID-19 Damage on the Ethiopian Service Sector: A Supplement Using

Google Search Trend “Big data”

Alemayehu Geda
Department of Economics
Addis Abeba University
May, 2020

Abstract
In the last week alone (May 19 to 25, 2020) the Ministry of Health data shows that the number of confirmed
cases has climbed from 365 last week to 655 on May, 25, 2020. This shows an average daily growth rate of
9.4 percent and a weekly growth rate of 79.5 percent. With the elasticity figures estimated in this paper
using google search data, this rate basically means the demand for the service sectors examined in this
study has already collapsed. The model results are realistic as that is also corroborated by the recent actual
data reported by the Ethiopian Hotels association that reported a decline in occupancy rate by 98 percent.
It is imperative for government to realize the extremely challenging situation the sector has found itself
and with more challenging time in the horizon. Thus, given the significant contribution of the sector to
GDP growth (46 percent) and urban employment (70 percent), it is imperative to take appropriate actions
now. In this effort, time is of the essence to save firms from collapsing and bankruptcy with the
determinantal impact for its work force.

I. Introduction
In my recent study about the social and economic impact of COVID-19 in Ethiopia
(Alemayehu. 2020), I have used local and global media-based information, information
obtained from interaction with some of the industry actors as well as my owen research
about the Ethiopian economy to come up with the possible impact of the virus on the
country. That study noted that GDP may contract by 11.2 percent in 2020/21 fiscal year.
In the best-case scenario of the effect of the virus being limited to the first quarter of the
next Ethiopian fiscal year that runs from July,2020- June 2021, the GDP may contract by
5.6 percent, instead. In the worst-case scenario of the effect hanging around for the
coming three quarters, the decline could be as high as 16.7 percent. In the same study the
service sector is, on average, is forecasted to contract by about 15.6 percent (Figure 1)
 3 
 

Figure 1: Scenarios for GDP & Sectoral Growth Effect COVID in Ethiopia

 GDP Growth      Agricultre Growth     Industry Growth      Services Growth

0.0%

‐1.6% ‐0.8% ‐2.4%

‐5.0%
‐5.6%
‐10.0% ‐8.5% ‐7.8%

‐11.1%
‐15.0%
‐15.6%
‐16.7% ‐17.0%
‐20.0%

‐25.0% ‐23.4%
‐25.5%
‐30.0%
GDP Growth wt COVID‐19 Best Case Worst

Source: Alemayehu, 2020

This paper is a further elaboration (and hence, a supplement) to the previous paper by
focusing in detail on the service sector. The reasons for this choice are three. First, the
service sector is the top contributor to the GDP growth as well as to urban employment.
Second, it is a sector that will be hit hard by the economic effect of the virus. Third, the
“big data” used for this paper is not capable of showing similar trends in the agriculture
and industrial sectors.

The contribution of the service and industrial sector (which are invariably located in the
urban areas and hence the urban economy) to GDP has increased from its level of 46
percent in 1995/96 to 67 percent in 2018/19, overtaking the historically dominant
contribution of the agricultural sector. Within the urban economy, manufacturing is
usually taken as the most dynamic sub-sector within the industrial sector. Ethiopia’s
urban economy is dominated by the service sector, however. In 1995/96 the service sector
contributed about 34 percent to GDP. This has increased to 39 percent in 2018/19. The
comparable figure for the agriculture and industrial value added had been 54 and 12
percent, respectively, in 1995/96; and became 33 and 29, respectively, today – (it has to
be noted, however, that about 70 percent of the industrial sector’s share of this 29 percent
in 2018/19 came from the construction sub-sector ) (Figure 2).
 4 
 

Figure 2: The Place of the Service Sector in Ethiopian Economy

Contribution to GDP (2018/19) Contribution to GDP Growth (2018/19)

Agriculture
15%
Agriculture
33% Services
Services
39% 46%
Indistry
39%

Indistry
28%

Sub‐Sector Share in Services (2028/19 Share in Urban Employment, 2018

Others* Whole Sale & 
22% Retail Trade Industry
36% 21%
Agricuture
Public  2%
Adminstratio
n & Defense
11% Services Households 
67% as employers
10%
Hotels and 
Transports &  Resturants
Real Estate, 
Communcation 6%
Renting & 
14%
Bussines Activites
11%

*Includes financial intermediaries, education, health, social work, private households with employed person and other community, social and personal services

Within the service sector, distributive services (trade, hotel and restaurant, followed by
transport and communication) at 36 percent share are the most important. This is
followed by a collection of sub-sectors under the name “others “at 22 percent in 2018/19.
The latter includes financial intermediaries, education, health, social work, private
households with employed person and other community, social and personal services.
This is followed by public administration and defense (Figure 2).

Such shift in sectoral share of GDP over time is usually taken as an indicator of structural
change an economy. Structural changed is generally defined as a shift in economic
 5 
 

activity from low productivity areas (historically the agricultural sector in the history of
today’s developed countries) towards high productivity sectors (the industrial sector
historically). The pattern of change in Ethiopia that is given above defies this historical
pattern as the economy is generally transforming towards the service sector.
Unfortunately, this service sector is not as sophisticated and as productive as the service
sector in today’s developed countries. It is rather dominated by small traders, hotels,
restaurants and administrative organ of the state – thus dashing the hope for true
structural transformation/industrialization that has been observed historically in today’s
developed countries.

Notwithstanding this, there are two interesting features of this service sector
development in Ethiopia. First, its contribution to GDP growth is significant - being a
staggering 46 percent today (Figure 2). Second, its contribution to urban employment, at
the staggering rate of about 67, is also the most important one (Figure 2). Thus, the effect
of COVID-19 on the service sector will have far reaching implications to total GDP growth
as well as to urban employment. This points at the need to focus on the sector urgently
now. To indicate the likely damage of the virus’ economic effect on the sector, the next
section has used a unique data and a model to predict this effect.

II. The Data and Results

2.1 The Data
The data used for this analysis is obtained from a recent paper by Kibrom et al (2020)
which is a comprehensive study using google search “big data” for 182 countries. The
analysis below is based on Ethiopia’s data from this database. Kibrom et al (2020) has
made an excellent description of the data as well as the number of adjustments measures,
they took to make the data comparable across countries. They also adjusted the data so
as to make it reflect the independent effect of COVID’s economic impact on the service
sector of the world. Thus, readers are advised to refer to this study for detailed
information about the data as well as the authors’ result about the impact of the virus on
the global service sector.

The Google search data measures the popularity of a search term in its search engine from
billions of google search engine users worldwide. A number of studies are now beginning
to appear using such data to forecast economic outcomes (see Kibrom et al, 2020 for such
studies). Such data is available since 2004. Google provides search index that measures
the popularity of a search term in a specific geographic area and at specific time relative
to the most popular search item in the same location and time. The index is then
constructed by giving this popular search term 100, and setting the rest relative to this
level. Thus, a search item that has half the popularity of the most popular term in Ethiopia
will have a value of 50 – a process that normalizes the data across time and space for
comparison (see Kibrom et al, 2020 for detail).
 6 
 

Such data is extracted by Kibrom et al (2020) from the Google server. They compiled the
data on a weekly basis, from January 1, 2004 to May 1, 2020 for 182 countries. It is the
Ethiopian part of this dataset that is used for this study.

One of the major weaknesses of this data is its limitation in reflecting the true picture of
the service sector in Ethiopia. That is, at the country’s internet penetration rate of about 4
percent of the population (of 108 million), google search-based information cannot reflect
the bulk of activities in the service sector of Ethiopia. Nonetheless, it crudely shows the
trend which is based on real and near real time “big data” which is important for making
a reasonable forecasting amidst problem of getting real data related to COVID today.
2.2 The Pattern of the Data, the Model and Estimated Results
I have generated elasticity values for the Ethiopian service sector that are given in Table
1. These results are based on auto-regressive distribution lag model (ADL) formulation
of a vector auto-regressive (VAR) model specified below. In general, in ADL formulation,
a long run (equilibrium) relationship between two variables, Y and X, could be given by
equation [1]

: , & [1]

As this equilibrium relationship cannot be observed, the observable disequilibrium

formulation of this long run (equilibrium) relationship between Y and X, in a simplified
from, can be given by equation [2]. Equation [2] is a simple ADL (m,n,p) [where m is the
number of lags, n &p the number of endogenous and exogenous variables, respectively],
ADL(1,1,1), formulation of equation [1]:
0 1 [2]
With some reparameterization1, the ECM representation of equation [2], which is the
estimable version that generated the results in Table 1, could be given by equation [3]
(Benerjee et al, 1993; Thomas, 1993; Hendry, 1995; cited in Alemayehu, 2002; Alemayehu
et al, 2012; Morales and Raei, 2013)
∆ ∆ 1 [3]

Where: ; ; 1 is the ECM term and is expected to be

negative and less than one.
This is the dynamic error correction model estimated using the Pesaran approach to co-
integration (Pesaran and Shin, 1999; Pesaran et al, 2001). The result from fitting this model
for the eight sub-sectors indicators of the service sector are given in Table 1.

 
1
  Subtracting Yt-1 from either side of equation [3] and adding and subtracting Xt-1 in the right-hand side of the resulting equation gives
equation [4].
 7 
 

Table 1: Estimated Elasticities of the service sub-sectors with respect to confirmed cases of the COVID Virus
Elasticities Hates and Restaurants Air Travel/Transport Tourism ITC
n=222(Observation) Hotels Restaurants Airport Flight Tourists Museums Zoom Skype
Short-term -0.11 -0.85 -0.67 -0.37 -1.66 -1.5 0.56 0.11
Long Terms -0.68 00* 000* -1.42 -0.21* 00* 0.57 0.25
Adjustment Coefficient -0.17 -0.70 -0.86 -0.25 -0.81 -0.58 -0.98 -0.45
* Not statistically significant at the conventional rates

The results in Table 1 show that demand for hotels and restaurants, transport and tourism
have contracted significantly due to COVID-19 while the demand for ITC services has
expanded significantly. A ten percent increase in confirmed weekly cases in Ethiopia is
found to lead to a 1.1 and 6.8 percent reduction in demand for hotels in the long and short
run, respectively. This became -8.5 percent for restaurants in the short run.
Similarly, the demand shock effect of a 10 percent weekly increase in the confirmed cases
on reducing the demand for air travel is found to range from 6.7 (using the search term
“airport”) to 3.7 (using the search term “flight”) percent in the short run and to about 14.2
percent in the long run.
The effect of a 10 percent increase in a weekly confirmed case is found to reduce the
demand in the tourism sector by about 15 to 17 percent in the short run.
This demand contraction in the service sector can also be read from Figures 3a and 3b. As
can be read form these figures, the declining trend of demand for these services began in
February with the highest decline being observed since March, 2020.

Figure 3a: Ethiopia: The Effect of COVID-19 on the Service Sector

120 120

100 100
airport_raw hotel_raw
flight_raw restaurant_raw
80 80

60 60

40 40

20 20

0 0
I II III IV I II I II III IV I II
2019 2020 2019 2020
Source: Author’s computation using Google Search data from Kibrom et al, 2020

On the positive side, the social distancing measures taken and the related condition of
working at home has led to a surge in demand for ITC services in Ethiopia. This can be
read from the increased demand for Zoom and Skype services following the
intensification of the virus’ effect as depicted in Figure 3b. The estimated model above
 8 
 

also shows that a ten percent increase in confirmed weekly cases leads to about 5.6
percent increase in demand for Zoom services both in the short and in the long run. This
effect became 1.1 and 2.5 percent surge in demand for Skype services in the short and
long run, respectively. Zoom seems twice popular compared to skype in Ethiopia. The
right panel in Figure 3b shows this positive demand effect in the ICT sector.

Figure 3b: Ethiopia: The Effect of COVID-19 on the Service Sector

90 120

80
tourist_raw 100
70 museum_raw zoom_raw
80 skype_raw
60

50
60
40
40
30

20 20

10
0
0 I II III IV I II
I II III IV I II 2019 2020
2019 2020
Source: Author’s computation using Google Search data from Kibrom et al, 2020

III. Conclusion
In this study, by focusing on the service sector, an attempt to complement my previous
study on the economic and social impact of COVID-19 on Ethiopian economy is made.
The choice of the sector is informed (a) by the data availability – which is google search
trend “big data” and (b) the significant national growth and urban employment effect of
the service sector. By estimating a forecasting model, the study has managed to come up
with specific magnitude of the effect of COVID-19 on the Ethiopian service sector that
can be used by policy makers. The study also shows the potential use of “big data” for
forecasting economic outcomes.

To give a perspective about the result, it is imperative to look the latest infection rate
(confirmed cases) in the country. This past week (May 19 to 25, 2020) the Ministry of
Health data shows that the number of confirmed cases has climbed from 365 last week to
655 on May, 25, 2020. This shows an average daily growth rate of 9.4 percent and a weekly
growth rate of 79.5 percent. With the elasticity figures estimated in this paper, this rate
basically means the demand for the various sub-sectors of the service sector examined in
this study has nearly collapsed (the exception being the demand for ICT services which
has expanded). This model’s results make sense as it is also corroborated by the recent
actual data reported by the Ethiopian Hotels association, for instance. The association
noted that Hotel occupancy rate among its member, that employed about 15, 000 work
force, has collapsed by 98 percent.
 9 
 

I hope the government realized the extremely challenging situation the sector has found
itself (and is going to be hit further each passing day) and takes appropriate actions now.
In this effort, time is of the essence so as to save firms from collapsing and bankruptcy
with detrimental implications of this for their work force. Given the significant effect of
COVID-19 on the service sector, and the enormous contribution of the sector for overall
GDP/economic growth of the country (46 percent) and urban employment (70 percent),
it is imperative to make sure that the sector is operating at a highest possible level of
capacity. When the latter is not feasible (as in the case of hotels & restaurants), finding
alternative engagement for firms (such as using them to service the health sector) is
important. It is also important to provide the necessary credit and financial help to the
sector so that firms may not go bankrupt or fail. This should be done while at the same
time providing COVID related health services at the premises of firms that could work
at some capacity. Apart from its damaging implication for employment, since most of the
new big firms in the sector, in particular big hotels, are indebted to the banking sector,
their situations may have a knock-on determinantal effect for the banking sector as well.

Acknowledgment
I am grateful to Kibrom Abay, Kibrom Tafere, Andinet Woldemichael for allowing me to access
the Ethiopian data from their global database. Any errors are mine.

Reference
Alemayehu Geda (2020). The Economic and Social Impact of COVID-19 and a Direction for Policy
Response (unpublished, Department of Economics, Addis Abeba University/
www.researchgate.net/profile/Alemayehu_Geda )
Alemayehu Geda, 2002. Finance and Trade in Africa: Modelling Macroeconomic Response in the World
Economy Context. London: Palgrave McMillan
Alemayehu Geda., N. Nudng’u & D. Zerfu. 2012. Applied Time Series Econometrics: A Practical
Guide for Macro-economic Researchers in Africa. Nairobi: University of Nairobi Press.
Kibrom Abay, Kibrom Tafere, Andinet Woldemichael (2020, forthcoming). Winners and losers
from COVID-19: Global Evidence from Google search
Morales, R and F. Raei2013 The Evolving Role of Interest Rate and Exchange Rate Channels in
Monetary Policy Transmission in EAC, IMF Research Paper NO 7, IMF, Washington, D.C.
Pesaran, M.H., Shin, Y., 1999. An autoregressive distributed lag modeling approach to
cointegration analysis. In: Strøm, S. (Ed.), Econometrics and Economic Theory in the 20th
Century: The Ragnar Frisch Centennial Symposium, Econometrics and Economic Theory in the
20th Century. Cambridge: Cambridge University Press.
Pesaran, M.H., Shin, Y., Smith, R.J. 2001. Bounds Testing Approaches to the Analysis of Level
Relationships, Journal of Applied Econometrics, 16: 289–326.

View publication stats