AMERICAN ASSOCIATION

OF WINE ECONOMISTS

AAWE WORKING PAPER

No. 169
Economics

STRUCTURE AND PERFORMANCE OF

ETHIOPIA’S COFFEE EXPORT SECTOR

Bart Minten, Seneshaw Tamru,

Tadesse Kuma and Yaw Nyarko

Sep 2014
ISSN 2166-9112 www.wine-economics.org
 WORKING  PAPER

Structure and Performance of Ethiopia’s

Coffee Export Sector

Bart Minten – International Food Policy Research Institute (IFPRI)

( [email protected])

Seneshaw Tamru – LICOS, University of Leuven

Tadesse Kuma – Ethiopian Development Research Institute

(EDRI) ( [email protected])

Yaw Nyarko –Department of Economics, NYU

([email protected])

 
  
TABLE OF CONTENTS
Abstract ................................................................................................................................................................................ 1  
1.   Introduction ...................................................................................................................................................................... 1  
SUMMARY  |  APRIL  22010  
2.   Coffee in Ethiopia ............................................................................................................................................................  
3.   Data ................................................................................................................................................................................. 5  
4.   Structure of the coffee export sector in Ethiopia .............................................................................................................. 6  
4.1.  Characteristics of coffee transactions and coffee exporters ..................................................................................... 6  
4.2.  Firm concentration and dynamics ............................................................................................................................. 6  
4.3.  Seasonality ................................................................................................................................................................ 8  
4.4.  Private versus public coffee exporting firms .............................................................................................................. 9  
5.   Coffee Quality ................................................................................................................................................................ 10  
6.   Coffee Export destination markets................................................................................................................................. 16  
7.   Performance of Ethiopia’s coffee export sector ............................................................................................................. 20  
7.1.  Values and quantities .............................................................................................................................................. 20  
7.2.  Changes in international coffee prices .................................................................................................................... 21  
7.3.  Quality premiums offered for Ethiopian coffee ........................................................................................................ 23  
8.   Discussion and conclusions........................................................................................................................................... 29  
References ......................................................................................................................................................................... 32  

LIST OF TABLES
Table 1—Ethiopia’s coffee exports – descriptive statistics ................................................................................................... 6  
Table 2—Factors associated with the export of washed and certified coffee – probit model ............................................. 15  
Table 3—Multinomial logistic regression model of destination markets for Ethiopian coffee ............................................. 19  
Table 4—Determinants of coffee prices (log(US cents per lb)); hedonic price model ........................................................ 25  
Table 5—Testing the benefits of vertical integration and traceability in the Ethiopian coffee export sector ....................... 27  
Table 6—Determinants of coffee prices (log(US cents per lb)) by destination market; hedonic price model..................... 28  
Table 7—Determinants of coffee prices (log(US cents per lb)); quantile hedonic price models ........................................ 29  

LIST OF FIGURES
Figure 1—Location of coffee production in Ethiopia ............................................................................................................. 3  
Figure 2—Value of coffee exports from Ethiopia and year-on-year growth in coffee exports, compared to total
exports, 2002/03 to 2012/13 .......................................................................................................................... 4  
Figure 3—Number of exporters and annual average/median value of coffee exports per exporter, 2007 to 2012 .............. 7  
Figure 4—Concentration ratios in Ethiopia’s coffee export sector, 2006 to 2013 at six-month intervals.............................. 7  
Figure 5—Evolution of incumbent (present year before and after), exiter (not present year after), and entrant (not
present the year before) firms in the coffee export sector, by number and by share in coffee exports ......... 8  
Figure 6—Monthly seasonality index for quantity of coffee exports for period 2003 to 2012 ............................................... 9  
Figure 7—Share of coffee exports by cooperatives and parastatal firms, 2006 to 2013 at six-month intervals ................. 10  
Figure 8—Quality premiums for coffee certification and washing, 2006 to 2013 ............................................................... 11  
Figure 9—Shares of washed coffee in total exported coffee quantities, monthly, 2006 to 2013 ........................................ 12  
Figure 10—Evolution of grades within washed and unwashed market segment, 2006 to 2013 at six-month intervals ..... 12  
Figure 11—Export share of the different producing regions by quantity and value, 2006 to 2013 at six-month intervals .. 14  
Figure 12—Share of different destination markets for Ethiopia’s coffee exports in 2012 ................................................... 16  
Figure 13—Share of different destination markets in Ethiopia’s coffee exports ................................................................. 17  
Figure 14—Cumulative price distribution by destination market, 2005 to 2013 ................................................................. 17  
Figure 15—Share of exports of washed and unwashed coffee by export destination, 2005 to 2013 ................................. 20  
Figure 16—Trends in real value and quantities of coffee exports from Ethiopia, 2003 to 2012 (Jan. 2003=100; 12-
month moving average) ............................................................................................................................... 21  

 
  
Figure 17—Trends in Ethiopian export coffee price versus reference international coffee price, 2002 to 2013 ................ 22  
Figure 18—Difference between Addis Ababa retail coffee price and the Ethiopia coffee export price, 2002 to 2013 by
month ........................................................................................................................................................... 23  
Figure 19—Regional coffee premiums over time, compared to Sidama coffee ................................................................. 26  
SUMMARY  |  APRIL  2010  

 
  
ABSTRACT
We study the structure and performance of the coffee export sector in Ethiopia, Africa’s most
important coffee producer, over the period 2003 to 2013. We find an evolving policy environ-
ment leading to structural changes in the export sector, including an elimination of vertical SUMMARY  |  APRIL  201
integration for most exporters. Ethiopia’s coffee export earnings improved dramatically over this
period, i.e. a four-fold real increase. This has mostly been due to increases in international
market prices. Quality improved only slightly over time, but the quantity exported increased by
50 percent, seemingly explained by increased domestic supplies as well as reduced local
consumption. To further improve export performance, investments to increase the quantities
produced and to improve quality are needed, including an increase in washing, certification, and
traceability, as these characteristics are shown to be associated with significant quality premi-
ums in international markets.

1. INTRODUCTION
Coffee is one of the most important traded commodities in the world. The sector’s trade struc-
ture and performance have large development and poverty implications, given the high concen-
tration of production by smallholders in poor developing countries. Coffee’s global value chains
are quickly transforming because of shifts in demands and an increasing emphasis on product
differentiation in importing countries (Ponte 2002; Daviron and Ponte 2005). There is a growing
willingness-to-pay for premium, high quality coffee by rich consumers and the demand for
specialty and certified coffee is on the rise.1 Moreover, international coffee markets have experi-
enced significant price variation over the last decade – prices were five times higher in 2011
than in 2002.
These changes have important implications for a number of the poorest developing
countries, as most coffee production takes place in these countries, even though most coffee
consumption is in developed countries (Pendergrast, 2010; Ponte, 2002). While there are a
number of studies that have looked at price formation for different types of coffee at the retail
consumption level in importing countries (e.g. Teuber and Herrmann, 2012), important ques-
tions remain on who benefits from this increasing willingness-to-pay for coffee and on how
changes in global coffee markets are transmitted to producing countries. Moreover, few re-
searchers have looked at how domestic policy change is affecting the performance of the coffee
sector in these exporting countries.
In this paper, we look at the structure and performance of the coffee export sector in
Ethiopia. In 2012, Ethiopia exported 3.2 million bags, making it the most important African
coffee exporter and the tenth largest exporter in the world (ICO, 2013). Its share of the interna-
tional coffee trade that year was about 3 percent. Coffee is the most important export product of
the country, accounting for about a quarter of the value of all exports in 2012. Coffee is cultivat-
ed by over 4 million primarily smallholder farming households (CSA, 2013) and, with those
employed in ancillary activities to coffee production, even more households are dependent on
coffee for part of their livelihoods (LMC, 2003). Furthermore, coffee plays an important role in
social gatherings and is important in local consumption, as more than half of Ethiopia’s coffee
production is consumed locally (CSA, 2013).
We study the structure and performance of Ethiopia’s coffee export sector over the last
decade. First, we document the characteristics of the sector, the policy shifts affecting the
                                                                                                                       
1
 Specialty  coffee  and  high-­‐quality  coffee  are  typically  synonymous.  However,  certified  coffee  is  not  necessarily  specialty  coffee.  

1  
  
sector, and the important structural changes that have occurred in Ethiopia’s coffee export
market during the study period. We then study how price, quantity, and quality variation is
associated with export performance. Local policy changes in Ethiopia, primarily those related to
the start of the Ethiopian Commodity Exchange (ECX) at the end of 2008, as well as export
SUMMARY  |  APRIL  201
license suspensions, have contributed to structural shifts in the sector. We find a lower concen-
tration ratio in the export sector, an inflow of smaller firms, a slowly increasing importance of
cooperatives and parastatals, and a slight diversification in the countries to which coffee is
exported subsequent to these policy changes. Over the last decade, we also note a large
increase in the value of coffee exports over time. This change has mostly been driven by
increases in international prices and to a lesser degree by increased quantity and quality of
exported coffee from Ethiopia. While the exported quantity has increased, this has seemingly
been driven by both increased production and reduced local consumption. We further note
important premiums being paid in international markets for washing, certification, vertical
integration, and for geographical indications of origin. The latter two are especially rewarded in
emerging high-end markets.
The structure of the paper is as follows. First, we give some background information on
the coffee sector in Ethiopia. Thereafter, in Section 3, we describe the data used for this study.
Structural characteristics of the coffee sector are discussed in Section 4. We discuss quality
issues in the Ethiopian coffee market and export destination markets in Sections 5 and 6,
respectively. The performance of the sector over the last decade is considered in Section 7,
particularly examining export prices and quality and their associations with different influencing
factors. We finish with a discussion of the findings and conclusions.

2. COFFEE IN ETHIOPIA
Ethiopia is endowed with a good production environment for growing coffee with a combination
of appropriate altitude, temperature, rainfall, soil type, and pH. Ethiopia is the center of origin for
Coffea arabica. The country possesses a diverse genetic base for this Arabica coffee with
considerable heterogeneity. Ethiopia produces a range of distinctive Arabica coffees and has
considerable potential to sell a large number of specialty coffees (Nure, 2008).2 Little of the
lower-value Robusta coffee is produced in Ethiopia, being better suited for production in lower
altitude equatorial climates. Coffee production in Ethiopia is almost exclusively situated in the
two regions of Oromia and the Southern Nations, Nationalities, and People Regions (SNNPR) in
the south and west of the country (Figure 1).

                                                                                                                       
2
 The  Specialty  Coffee  Association  of  America  (SCAA)  defines  specialty  coffee  in  its  green  stage  as  coffee  that  is  free  of  primary  
defects,  contains  no  unripened  beans,  is  properly  sized  and  dried,  is  free  of  faults  and  taints,  and  has  distinctive  attributes  
(Rhinehart,  2012).  

2  
  
Figure 1—Location of coffee production in Ethiopia

SUMMARY  |  APRIL  201

Source: Authors’ calculations based on data from the Central Statistical Agency (CSA)

Smallholder farmers produce 95 percent of Ethiopia’s coffee (Tefera and Tefera, 2013).
It is produced under several types of production systems, including forest, semi-forest, garden,
and plantation coffee (Tulu, 2008). Forest coffee is grown in the wild under natural forest cover
and is gathered by farmers from trees with minor tree maintenance. Semi-forest coffee is also
grown in forest conditions, but there is some limited maintenance by farmers, mostly annual
weeding. This type of coffee has clearly delineated boundaries of ownership, although the trees
usually are located away from agricultural plots. Garden coffee is defined as coffee from trees
planted by farmers in the vicinity of their residences. It is often intercropped with other crops or
trees. Plantation coffee is grown on large commercial farms, private as well as state farms.
Modern production practices – such as irrigation, modern input use, mulching, stumping, and
pruning - are often applied in this case. While reliable recent statistics are lacking, it is estimated
that these different production systems make up about 10, 35, 50, and 5 percent, respectively,
of total coffee production in the country (Kufa, 2012).
Figure 2 shows the variation in the value (in nominal USD) of both overall exports and
coffee exports from Ethiopia over the last decade. The figure illustrates two main patterns. First,
coffee makes up an important part of overall exports. Second, we see strong growth rates in
both over the last decade. The overall value of exports grew at an average compounded annual
rate of 21 percent, while coffee exports grew at 16 percent. This slightly slower growth rate of
coffee exports compared to overall exports implies that the share of coffee exports in total
exports has decreased over time. While coffee made up almost 35 percent of the value of total
export in 2002/03, this came down to 24 percent for the period 2012/13, which suggests that
export commodities have diversified in recent years.

3  
  
Figure 2—Value of coffee exports from Ethiopia and year-on-year growth in coffee exports, com-
pared to total exports, 2002/03 to 2012/13

3,500   70.0  
coffee  
total   60.0   SUMMARY  |  APRIL  201
3,000  
growth  coffee   50.0  
growth  total   40.0  
2,500  
30.0  
1000  USD  

percent  
2,000   20.0  
1,500   10.0  
0.0  
1,000   -­‐10.0  
-­‐20.0  
500  
-­‐30.0  
0   -­‐40.0  

 
Source: Authors’ calculations based on data from the National Bank of Ethiopia

There have been significant domestic policy reforms in the last decade that affected the
structure and performance of the coffee export sector. First, from December 2008 onwards it
became mandatory for private traders to sell their coffee through the Ethiopian Commodity
Exchange (ECX), a new modern commodity exchange.3 ECX trades standard coffee contracts,
based on a warehouse receipt system, with standard parameters for coffee grades, transaction
size, payment, and delivery. The first level quality control is decentralized and undertaken in
nine liquoring and inspection units in major production areas.4 The establishment of the ECX
has led to important changes in the structure of the coffee value chain (Gabre-Madhin, 2012).
Second, the government intervened in the coffee market on several occasions in an ef-
fort to reduce hoarding by exporters. In April 2009, six large traders were banned from exporting
coffee because of their presumed excessive hoarding. The government revoked their licenses,
closed down their warehouses, seized their coffee stocks, and sold them on their behalf (Alemu,
2009). A policy was further implemented in May 2011 that limited the amount of coffee an
exporter can store. An exporter, for example, selling and buying coffee on the ECX will have his
or her right to trade on the commodity exchange revoked if found to be storing more than 500
metric tons of coffee without a shipment contract with an importer (Tefera and Tefera, 2013).5
Failing to adhere to these regulations has led to the banning of coffee exporters, as seen in
2011 and 2013 (Araya, 2011; Yewondwossen, 2014).
Third, there have been a number of changes regarding export taxes on coffee over time.
Core changes include the removal of entry barriers (Proclamation No. 70/1993); the consolida-
tion of all taxes and duties levied on coffee export into a single tax family (Proclamation No.
99/1998), which consolidated all taxes on coffee export to 6.5 percent; and, following the 2002
international coffee crisis, the waiving of all export taxes on coffee exports.

                                                                                                                       
3
 Producers  who  are  exporters  can  bypass  the  ECX,  as  can  farmer  cooperatives.  
4
 Before  the  establishment  of  ECX,  all  first  and  second  level  quality  inspection  was  carried  out  in  Addis  Ababa.  
5
 A  directive  requiring  the  shipment  of  coffee  in  bulk  containers,  instead  of  60-­‐kilogram  jute-­‐bags,  was  put  in  place  in  mid-­‐
November  2011.  However,  it  was  revoked  soon  after  because  of  pressures  from  exporters.  Such  bags  are  widely  used  in  
international  trade  and  help  to  better  maintain  the  identity  of  the  coffee  (Tefera  and  Tefera,  2013).  

4  
  
Finally, an Ethiopian Fine Coffee Trademark Licensing Institute was set up in February
2005 with the purpose of setting up a system to secure legal ownership in international markets
of specialty coffee names (especially Sidamo, Harar, and Yirgacheffe) (Agrer, 2004). There was
initial resistance against this initiative, but they were ultimately settled. The goal of this effort
SUMMARY  |  APRIL  201
was to add brand value to Ethiopian coffee. Signatories entered into a brand management
strategy with the government with the purpose of achieving better farm-gate and export prices
for coffee (Arslan and Reicher, 2010).

3. DATA
To understand Ethiopia’s coffee export sector, we relied on different sources of information and
methods. First, a large number of key informants in the sector were contacted and interviewed.
These included employees of private exporters, the Coffee Liquoring Unit (CLU), the Coffee
Plantation and Development Enterprise, the Coffee Processing and Warehouse Enterprise,
government entities, ECX, coffee producers and retailers, as well as coffee researchers.
In addition, we used a number of secondary data sources. First, data were obtained from
the National Bank of Ethiopia (NBE) on monthly coffee exports for the period 2002 to 2013.
These data were used to calculate trends, 12-month moving averages, as well as seasonal
movements. Second, the International Coffee Organization (ICO) calculates an international
composite price for coffee based on future contract prices recorded on the London and New
York stock exchange. Historical monthly data were downloaded from their webpage.6 Third, a
database of coffee export transactions is maintained by the Ministry of Trade. This export
transactions dataset for the period July 2006 to June 2013 was used.
An important aspect in coffee exports is quality. Coffee quality assessments for exports
are conducted by the Coffee Liquoring Unit (CLU) to ensure that it meets export standards. After
buying coffee on the ECX trading floor (or, before 2009, at the coffee auction), exporters pick up
the coffee from regional ECX warehouses and it undergoes further processing to meet export
standards.7 On reaching this export standard, a quality certificate is issued by the CLU. A quality
inspection sheet is prepared and is attached to the lot to be exported.8 These quality indicators,
as well as others (such as washing, certification, and origin), are part of the coffee export
transactions dataset, which was used for this study. We also obtained a list of private commer-
cial coffee farms (with cultivated areas of 40 hectares and above) from its association. This
information was integrated into the analysis as well.

                                                                                                                       
6
 http://www.ico.org/  
7
 It  is  estimated  that  there  are  about  80  of  these  processing  units,  mostly  based  in  Addis.  The  most  sophisticated  processing  
machine  is  held  by  the  Coffee  Processing  and  Warehouse  Enterprise.  It  is  estimated  that  they  process  in  their  unit  about  one-­‐
quarter  of  all  coffee  exported  from  Ethiopia.  
8
 The  analysis  of  coffee  quality  by  the  CLU  is  based  on  two  measures,  the  raw  and  physical  inspection  and  the  cup  inspection.  
The  raw  and  physical  inspection  contributes  for  40  percent  to  the  final  quality  grade,  while  the  cup  inspection  contributes  for  
60  percent.  However,  moisture  and  screen  analysis  are  the  two  requisites  before  grading  any  coffee.  The  moisture  content  
should  be  less  than  11.5  percent,  while  the  size  of  the  bean  should  be  above  screen  size  14  for  85  percent  of  the  bean  sample.  
In  the  case  of  unwashed  coffee,  raw  quality  is  determined  based  on  defect  count  of  the  beans  and  on  odor.  In  the  case  of  
washed  coffee,  the  raw  quality  is  based  on  an  assessment  of  shape  and  make,  color,  and  odor.  Cup  quality  is  assessed  along  
four  criteria,  including  cup  cleanliness,  acidity,  body,  and  flavor.  Each  characteristic  counts  for  15  percent  of  the  60  percent  of  
the  cup  quality  value.  The  washed  coffee  export  standards  vary  between  grade  1  and  2,  as  well  as  under-­‐grade  (UG)  while  
unwashed  coffee  is  graded  3,  4,  5,  or  under-­‐grade.  Within  the  under-­‐grade  category,  a  further  distinction  is  made  for  under-­‐
grade  type  ‘inferior,  but  exportable’  coffee,  while  the  worst  under-­‐grade  coffee  is  not  exportable  and  is  destined  for  domestic  
consumption.  

5  
  
4. STRUCTURE OF THE COFFEE EXPORT SECTOR
IN ETHIOPIA
4.1. Characteristics of coffee transactions and coffee exporters
SUMMARY  |  APRIL  201
Table 1 gives an overview of some of the characteristics of coffee export transactions and of
exporters over the period 2006 to 2013. The average coffee export transaction over this period
involved 37 metric tons with a value of 133,000 USD. The average price was 173 US cents per
pound (lb). The large standard deviations indicate significant variations in quantity, value, and
price between transactions. 51 percent of the export transactions were destined for Europe, 14
percent for North America, 12 percent for Saudi Arabia, and 10 percent for Japan. Sudan
accounted for 4 percent of all the export transactions over that period.
Transactions were aggregated by coffee exporter for every year to give an idea of the
scale of operations of the exporters. An average exporter over that period exported 1,266 metric
tons of coffee per year for a value of 4.5 million USD. Again, we see large variability in scale of
operations across exporters.
Table 1—Ethiopia’s coffee exports – descriptive statistics
Mean Median Standard
Unit deviation
Transaction  data  (07/2006  -­‐  06/2013)
Quantity Metric  tons 37.8 19.2 36.1
Price US  c ents/lb 173.2 159.0 66.8
Value 1000  USD   132.7 91.5 131.4
Destination  markets:  
Europe %  of  transactions 51.2
Japan %  of  transactions 9.6
North-­‐America %  of  transactions 13.9
Saudi-­‐Arabia %  of  transactions 12.0
Sudan %  of  transactions 4.1
Other %  of  transactions 9.1
Exporters  annually  (01/2007  -­‐12/2012)
Quantity  exported  by  a ctive  exporters  per  year Metric  tons 1,265.7 340.5 2,610.1
Value  of  exports  by  a ctive  exporters  per  year 1000  USD   4,516.3 1,062.5 8,966.2  
Source: Authors’ calculations based on data from the Ministry of Trade

4.2. Firm concentration and dynamics

We note large changes over the study period in the number of exporters that are active in coffee
export markets. In the beginning of the period (2007-2008), there were about 100 active coffee
exporters (Figure 3). This increased to 175 exporters by 2012, an increase of 75 percent. The
evolution of the median value of exports per exporter shows a slight decline over time, seeming-
ly indicating an inflow of smaller exporters. While the average value of exports per exporter
came down in 2009 (the first year of the ECX), it has rebounded since, although the median
value has not.

6  
  
Figure 3—Number of exporters and annual average/median value of coffee exports per exporter,
2007 to 2012

200   6  
180   SUMMARY  |  APRIL  201
5  
160  

Number  of  exporters  

140  
4  

million  USD  
120  
100   Number  of  exporters   3  
80   Average  value  
2  
60   Median  value  
40  
1  
20  
0   0  
2007   2008   2009   2010   2011   2012  
 
Source: Authors’ calculations based on data from the Ministry of Trade

We look at the concentration ratio of export firms in Ethiopia. Common measures used
to measure the concentration of firms in economic sectors are the CR4 and CR8, the share of
business in the sector handled by the four and by the eight largest firms in the sector, respec-
tively. These concentration ratios have decreased significantly over time, especially after 2008
(Figure 4). Before 2008, the CR4 on average was about 40 percent, while the CR8 approached
60 percent. Since the ban of export licenses of six major exporters in early 2009 (Alemu, 2009),
this share came down by 16 percent and 14 percent for the CR8 and CR4, respectively, com-
pared to the end of 2008.9 Over the period considered, the lowest concentration was at the
beginning of 2011, but has been slowly increasing since. At the beginning of 2013, the CR8 was
almost 40 percent, while the CR4 was 25 percent.
Figure 4—Concentration ratios in Ethiopia’s coffee export sector, 2006 to 2013 at six-month
intervals

70  

60  

50  
 percent  

40  

30  

20   CR4  
CR8  
10  

0  

 
Source: Authors’ calculations based on data from the Ministry of Trade

                                                                                                                       
9
 The  license  ban  was  triggered  by  the  perception  that  these  exports  were  involved  in  excessive  hoarding,  contributing  to  a  
general  lack  of  foreign  exchange  in  Ethiopia  (Alemu,  2009).  

7  
  
The World Bank (2014) shows that the largest exporters, i.e. those selling more than 5
million USD per year, account for nearly 80 percent of coffee exports. They further argue that
the coffee market is hard to enter – new firms have limited opportunities to enter into the coffee
export business. This is illustrated in Figure 5, which plots the share of entrants (coffee export
SUMMARY  |  APRIL  201
firms that were not in business the year before), exiters (firms that were not there the year after),
and incumbents (firms that are there both the year before and after) for the period 2007 to 2012.
On the left (Figure 5a), we see that a relatively large number of new entrants came into the
market in 2010/11, possibly attracted by the high prices in that year. In 2007/08, we find that a
large number of firms exited the market the year after, possibly related to the start of the ECX.
However, in the graph on the right (Figure 5b), the share in total coffee exports of firms that
moved in and out of the market is shown to be small, with entrants and exit firms typically
making up less than 10 percent of the market. The graph also illustrates the disruptive effect of
the ban on large exporters in the year 2008/09. The six banned firms made up a large propor-
tion of the 35 percent of firms in that year who were exiters and would not export coffee any-
more.

Figure 5—Evolution of incumbent (present year before and after), exiter (not present year after),
and entrant (not present the year before) firms in the coffee export sector, by number
and by share in coffee exports
a. Number  of  coffee  export  firms     b. Share  in  coffee  exports  
200   100%  
Share  in  total  coffee  exports  
Number  of  firms  

150   80%  
60%  
100  
40%  
50   20%  
0   0%  
2007/08   2008/09   2009/10   2010/11   2011/12   2007/08   2008/09   2009/10   2010/11   2011/12  

incumbents   entrants   exiters   incumbents   entrants   exiters  

   
Source: Authors’ calculations based on data from the Ministry of Trade

4.3. Seasonality
As for most agricultural crops, the production, processing, and marketing of coffee is character-
ized by important seasonal patterns. Coffee is mostly harvested during the period October to
December. It can be sold immediately in the form of red cherries, or farmers might dry the
cherries and sell them later in the year as whole dried cherries. Figure 6 shows how coffee
exports vary over the year. The peak is achieved in the period from March until June when
exports are on average more than twice as high as during the lean period, i.e. October to
January.10 The timing of the peak of exports indicate a significant lag between production and
exports, linked to time consuming post-harvest and processing activities, but also due to stor-
age. If the share of washed coffee in total exports had increased over time, one might have
expected a shift forward in the timing of exports, since washed coffee is sold slightly earlier in
the season. However, such a shift is not noted. In fact, the reverse is observed. Comparing
exports for the period 2003 to 2007 to those of the period 2008 to 2012, a shift in the main

                                                                                                                       
10
 During  that  period,  left-­‐over  stocks  of  the  previous  year  are  mostly  sold.  It  is  not  yet  the  new  harvest.  

8  
  
export season to about 2 months later in the year is seen (Figure 6).11 While the peak month of
exports was April for the period 2003 to 2007, this had shifted to June for the period 2008 to
2012.

Figure 6—Monthly seasonality index for quantity of coffee exports for period 2003 to 2012 SUMMARY  |  APRIL  201

1.8  
1.6  
1.4  
1.2  
1.0  
0.8  
0.6   average  2003-­‐2007  
0.4   average  2008-­‐2012  
overall  average  2003-­‐2012  
0.2  
0.0  
J   F   M   A   M   J   J   A   S   O   N   D  
 
Yearly average = 1.0
Source: Authors’ calculations based on data from the National Bank of Ethiopia for the period January 2003 to
December 2012

4.4. Private versus public coffee exporting firms

Figure 7 shows the extent to which cooperatives and parastatals, including the Ethiopian Grain
Trade Enterprise (EGTE) and state farms, play a role in coffee exports from Ethiopia. Three
main findings appear from the graph. First, the share of cooperatives and parastatals in export
trade is relatively small. The large majority of coffee exports is in the hands of the private sector.
The lowest share in total exports for the private sector was achieved at the end of 2009, when
its share was 83 percent. However, in most years the private sector largely dominated, with a
share close to 90 percent.
Second, we see significant variation in the share of exports for cooperatives and para-
statals over time. In the case of cooperatives, we see a steadily increase from between 3 and 4
percent in 2006/07 to between 5 and 6 percent in 2012/13. The most important cooperative
involved in coffee exports is the Oromia Coffee Cooperative Union, which over this period
accounted for 57 percent of the export transactions made by cooperatives. Other important
cooperatives include the Yirgacheffe, Sidama, and Kafa Forest Coffee Cooperatives.
Third, there also is large variation in the share of exports made by parastatal firms. They
were the source of 3 to 4 percent of all coffee exports until the end of 2008. However, their
share increased to more than 10 percent at the end of 2009. This was mostly driven by the
seizure of coffee stocks of a number of large traders by the government.12 These stocks were
then consequently exported by EGTE on behalf of the government (Alemu, 2009). Since the
                                                                                                                       
11
 Part  of  the  explanation  is  seemingly  linked  to  late  exports  in  2009.  As  there  were  wide  expectations  of  currency  depreciation  
and  as  there  was  a  decline  in  international  coffee  prices,  exporters  held  on  longer  than  usual  to  their  coffee  stocks.  This  led  the  
government  to  take  the  drastic  measure  of  banning  licenses  of  major  coffee  exporters  in  April  2009.  
12
 Alemu  (2009)  states  that  the  government  received  21  million  dollars  from  exporting  the  seized  stock  owned  by  the  six  coffee  
exporters.  

9  
  
end of 2009, the share of coffee exports coming from parastatals has been diminishing as many
state farms were privatized over this period. These include the Bebeka and Teppi Coffee
Plantations of the Coffee Plantation and Development Enterprise, which were privatized in
2011/12 and 2012/13, respectively. The Limu Coffee Plantation was the last to be privatized in
SUMMARY  |  APRIL  201
2014. Currently, there are no more state coffee plantations in Ethiopia.

Figure 7—Share of coffee exports by cooperatives and parastatal firms, 2006 to 2013 at six-month
intervals

12  

10   cooperaives  
parastatal  
8  
percent  

6  

4  

2  

0  

 
Source: Authors’ calculations based on data from the Ministry of Trade

In short, we see a significant number of structural changes in the coffee sector over the
last decade. First, we see a decline in concentration ratios in the export sector over time.
Second, seasonality in coffee exports is pronounced with most exports being shipped between
March and June. There also has been a recent shift in the major peak of exports to a later
period in the year. Third, we find variation in the share of cooperatives (increasing over time)
and parastatals (first an increase and then a decrease) in total exports over time. However, their
share in total exports is relatively small. Overall, we note an increase in the number of exporters
and increasing diversity of players (e.g. private sector, cooperatives, parastatals) in the export
market. However, the shares of incumbent firms are large, possibly because of the expertise
and reputation required to gain market share in the coffee export business, as well as problem-
atic access to trade credit for new entrant firms, especially (World Bank, 2014).

5. COFFEE QUALITY
There are several measures of coffee quality in the Ethiopian market place. They include, most
importantly, certification, which affects marketability and prices, but not necessarily the intrinsic
quality of the coffee; geographical indications of origin; grades; and washing. Each is discussed
in turn.
Certification and traceability have become major new requirements in the global food
trade (Swinnen, 2007), with such certification schemes often implemented to add value to a
product (Jena et al., 2012). By guaranteeing the product origin, fair prices to producers, ethical
standards of production and processing, environmental sustainability in production, and safety
and quality safeguards for a product, international buyers and consumers are often willing to
pay extra for a product. Conversely, adhering to those new requirements can be costly. In the
global coffee sector, it is estimated that around 16 percent of current coffee production is

10  
  
certified. This share should reach over 25 percent by 2015 (Panhuysen and Van Reenen,
2012). There are currently a number of certification schemes in place, most importantly Fair
Trade, Organic Coffee, Bird-Friendly, UTZ, and Global Forest Alliance. In the case of Ethiopia,
the share of certified coffee is increasing, but is significantly lower than in other countries. For
SUMMARY  |  APRIL  201
example, data from the Ministry of Trade show that only 2 percent of coffee transactions over
the period 2006-2013 were done under the Fair Trade scheme. This suggests that Ethiopia
likely foregoes the commercial rewards of the price premiums associated with these certification
schemes. Figure 8a illustrates the size of the certification premium over the period 2006 to
2013. It is observed that the density function of prices of certified coffee is distinctively to the
right of non-certified coffee.

Figure 8—Quality premiums for coffee certification and washing, 2006 to 2013
a. Premium  certified  coffee   b. Premium  washed  coffee  
.008

.01
.006
Density

Density
.004

.005
.002

0
0

0 100 200 300 400 500 0 100 200 300 400 500
US cents/lb) US cents/lb)

non-certified certified unwashed washed

   
Source: Authors’ calculations based on data from the Ministry of Trade

Second, the quality of coffee can be increased by washing, i.e. processing red cherries
immediately after harvest in wet mills, instead of sun-drying the cherries (Nure, 2009). Washed
coffee preserves the intrinsic quality of the bean better than unwashed beans, and the process
leads to homogenous coffee with fewer defective beans. The washing process is carried out in
washing stations where cherries are pulped immediately after harvesting, fermented in tanks,
and washed in clean water to remove the mucilage. The wet parchment coffee is then dried in
the sun. For unwashed coffee, cherries are dried on mats or concrete floors. After drying, the
outer layer of the cherries is removed by hulling in coffee processing plants. Figure 8b (the right
panel) is a graph of the price density functions for washed and unwashed coffee over the period
2006 to 2013. The graph reflects the significant rightward shift for the distribution of the export
prices of washed coffee, illustrating the premium paid for washed coffee over unwashed coffee.
However, this price difference might be linked to other explanatory variables as well. We explore
this below through the use of multivariate regression analysis.
Given the sizable premiums for washed coffee in international markets, investments in a
number of coffee producing countries have focused on setting up more wet mills. Figure 9
shows the share of coffee exports from Ethiopia that were washed. As shown in the linear trend
line, the share of washed coffee in total exports has not changed significantly over time, remain-
ing close to 30 percent. We see strong seasonality in the exports of washed coffee. The im-
portance of washed coffee is higher at the beginning of the year, reflecting the fact that washed
coffee is coming earlier to market than the unwashed coffee.

11  
  
Figure 9—Shares of washed coffee in total exported coffee quantities, monthly, 2006 to 2013

35,000   unwashed   80  
washed  
30,000   share  washed   70  
Linear  (share  washed)   SUMMARY  |  APRIL  201
60  
25,000  
50  

 percent  
20,000  
tons  

40  
15,000  
30  
10,000  
20  
5,000   10  
0   0  
2006   2007   2008   2009   2010   2011   2012   2013  
 
Source: Authors’ calculations based on data from the Ministry of Trade

The Coffee Liquoring Unit (CLU) grades the quality of each exported lot of coffee beans,
for both washed and unwashed coffee, based on physical and cup inspection. Grades range
from 1 to 5, where 1 is the best and highest grade and 5 is the lowest. Ungraded (U.G.) is the
worst quality and this coffee is destined to those export markets where there is less demand for
quality or alternatively, this type of coffee quality is not allowed to be exported and is sold on the
local market. Grades 1 and 2 and grades 3, 4, and 5 are assigned to washed and unwashed
coffee, respectively. Overall, we find a slight increase in quality over the study period for both
washed and unwashed coffee. For the unwashed bean market segment, the share of the worse
qualities (grade 5) has been decreasing slightly over time, while the share of grade 4 has been
increasing slightly (Figure 10b). In 2006/07, 24 percent of unwashed coffee was grade 4, but
this increased to 31 percent in 2012/13. We see little changes over time in the quality grades for
washed coffee (Figure 10a). However, there is a small decline in the low quality washed U.G.
Figure 10—Evolution of grades within washed and unwashed market segment, 2006 to 2013 at six-
month intervals
a. Washed   b. Unwashed  
100%   100%  
90%   90%  
Washed  U.G.   80%   Unwashed  U.G.  
80%  
70%   Grade  2   70%   Grade  5  
60%   60%  
Grade  1     Grade  4  
50%   50%  
40%   40%   Grade  3  
30%   30%  
20%   20%  
10%   10%  
0%   0%  
2006-­‐2  
2007-­‐1  
2007-­‐2  
2008-­‐1  
2008-­‐2  
2009-­‐1  
2009-­‐2  
2010-­‐1  
2010-­‐2  
2011-­‐1  
2011-­‐2  
2012-­‐1  
2012-­‐2  
2013-­‐1  
2006-­‐2  
2007-­‐1  
2007-­‐2  
2008-­‐1  
2008-­‐2  
2009-­‐1  
2009-­‐2  
2010-­‐1  
2010-­‐2  
2011-­‐1  
2011-­‐2  
2012-­‐1  
2012-­‐2  
2013-­‐1  

   
Source: Authors’ calculations based on data from the Ministry of Trade

Third, the geographic origin of coffee is an important quality consideration, as it is

strongly related to taste. In our analysis, we distinguish between Sidama, Jimma, Wollega

12  
  
(Nekempt), Yirgacheffe, Limu, Harar, and other coffees.13 Kufa (2012) associates tastes and
regions as follows: spicy for Sidama, fruity for Wollega (Nekempt), floral for Yirgacheffe, winy for
Limu and Jimma, and mocha for Harar. Figure 11a shows the export share in quantity over time
of the different origins of coffee. The three major coffee types that are exported are Sidama,
SUMMARY  |  APRIL  201
Jimma, and Nekempt. While our data shows significant variation of the shares of different
origins over the time period, there are no systematic trends. The combined share of these three
major regions has stayed stable over the years at around 80 percent. Figure 11b further illus-
trates that, while the shares of the premium quality coffees of Harar and Yirgacheffe in total
exports is relatively small in quantity, they are higher in value terms given their relatively high
prices.

                                                                                                                       
 It  is  to  be  noted  that  some  coffees  are  put  in  these  categories  even  though  they  might  be  produced  in  areas  that  are  located  
13

 
outside  these  geographical  delineations.

13  
  
Figure 11—Export share of the different producing regions by quantity and value, 2006 to 2013 at
six-month intervals
a. Quantity   b. Value  
100   100%   SUMMARY  |  APRIL  201
90   Other   90%   Other  
80   80%  
70  
Harar   70%  
Harar  
60   Limu   60%   Limu  
50   50%  
40   Yirgacheffe   40%   Yirgacheffe  
30   30%  
20   Nekempt   20%   Nekempt  
10   10%  
Jimma   Jimma  
0   0%  
2006-­‐2  
2007-­‐1  
2007-­‐2  
2008-­‐1  
2008-­‐2  
2009-­‐1  
2009-­‐2  
2010-­‐1  
2010-­‐2  
2011-­‐1  
2011-­‐2  
2012-­‐1  
2012-­‐2  
2013-­‐1  

Sidama   Sidama  
   
Source: Authors’ calculations based on data from the Ministry of Trade

To understand the links of different associated variables with measures of coffee quality,
we ran a regression on two main quality indicators, i.e. certification and washed coffee (Table
2). In the left columns, we present the results of a probit regression where we regress washed
(1) versus unwashed (0) on a number of associates. As expected, we find that washed coffee is
exported earlier in the year. The coefficients for the dummy variables for February, March, and
April are positive and significant, while those for the rest of the year are negative. This indicates
that washed coffee is significantly less likely to be exported in the latter period of the year
compared to the default month January. We also find that there are strong linkages of washed
coffee with regional dummies. Compared to the default region Sidama, Yirgacheffe has a
significantly higher likelihood to export washed coffee. On the other hand, Jimma and Wollega
(Nekempt) coffee exports are much less likely to be washed. No washed coffee is exported from
the region of Harar and no coefficient estimates are therefore available. We also note that
certification schemes as well as the type of exporters are strongly linked with washed coffee
exports. Parastatals, cooperatives, state farms, and commercial private farms are significantly
more likely to export washed coffee than are private sector traders. Fair Trade coffee is also
much more likely to be washed.

14  
  
Table 2—Factors associated with the export of washed and certified coffee – probit model
Washed  c offee Certification
Unit Coef. z-­‐value* Coef. z-­‐value*
Type  of  exporter  (default=private)
Cooperative yes=1 0.530 8.06 3.452 61.39 SUMMARY  |  APRIL  201
Parastatal yes=1 1.042 7.53 0.000 -­‐
State  farm yes=1 1.358 9.54 0.000 -­‐
Private  c ommercial  farm yes=1 0.389 4.85 1.924 25.34
Region  of  origin  (default=Sidama)
Jimma yes=1 -­‐5.098 -­‐22.23 -­‐0.832 -­‐12.41
Wollega/Lekempte yes=1 -­‐3.126 -­‐26.51 -­‐0.958 -­‐8.35
Yirgacheffe yes=1 1.495 33.15 0.137 2.51
Limu yes=1 0.936 13.36 -­‐0.852 -­‐8.14
Harar yes=1 -­‐ -­‐ -­‐1.016 -­‐8.44
Other yes=1 -­‐0.572 -­‐8.48 -­‐0.339 -­‐2.43
Certificate
Fair  Trade yes=1 0.799 8.06 -­‐ -­‐
Organic yes=1 -­‐0.131 -­‐1.91 -­‐ -­‐
Year  (default=2006)
2007 yes=1 0.210 2.77 0.057 0.38
2008 yes=1 0.188 2.32 0.015 0.09
2009 yes=1 0.174 2.15 -­‐0.372 -­‐2.48
2010 yes=1 0.153 2.11 0.126 0.91
2011 yes=1 -­‐0.123 -­‐1.62 0.060 0.42
2012 yes=1 0.039 0.49 0.243 1.74
2013 yes=1 -­‐0.253 -­‐2.86 -­‐0.067 -­‐0.42
Month  (default=January)
February yes=1 0.472 4.46 0.067 0.47
March yes=1 0.436 4.28 0.014 0.09
April yes=1 0.084 0.89 -­‐0.024 -­‐0.18
May yes=1 -­‐0.290 -­‐3.11 0.016 0.12
June yes=1 -­‐0.332 -­‐3.59 0.012 0.09
July yes=1 -­‐0.351 -­‐3.72 -­‐0.190 -­‐1.38
August yes=1 -­‐0.463 -­‐4.80 -­‐0.567 -­‐4.15
September yes=1 -­‐0.435 -­‐4.38 -­‐0.284 -­‐2.01
October yes=1 -­‐0.542 -­‐5.34 -­‐0.170 -­‐1.11
November yes=1 -­‐0.549 -­‐5.07 -­‐0.007 -­‐0.05
December yes=1 -­‐0.413 -­‐2.82 -­‐0.352 -­‐1.75
Intercept 0.395 3.71 -­‐2.513 -­‐14.07
Number  of  observations 28220 28893
Wald  c hi2(29) 3455 5716
Prob>Chi2 0 0
Pseudo  R2 0.6147 0.7277
*  r obust  s tandard  errors;  z-­‐values  i n  bold  a re  s ignificant  a t  the  5  percent  l evel  
Source: Authors’ calculations based on data from the Ministry of Trade

For the factors associated with certified coffee, Yirgacheffe coffee is more likely to be
certified than is Sidama coffee, the default category of origin. Coffee from other zones of origin
is less likely to be certified. We note higher shares of certified coffee are exported at the begin-
ning of the year, which also suggests that most certified coffee is washed. Although there is high
volatility, the prevalence of certification in general is increasing over time, as seen by the
generally increasing positive coefficients linked with the year dummy variables. Cooperatives
and private commercial farms are significantly more likely to export certified coffee. This is
presumably because they are not required to transact on the ECX platform, and trading certified
coffee on the ECX platform was not an option over the period considered.

15  
  
6. COFFEE EXPORT DESTINATION MARKETS
In this section, we look more closely at the export destination markets. Ethiopian coffee was
exported on average to about 50 countries annually over the period 2005 to 2012. Figure 12
shows the share of different destination markets for coffee exported from Ethiopia in 2012. The
SUMMARY  |  APRIL  201
largest share of coffee exports went to Germany. It accounts for one-third of Ethiopia’s coffee
exports. The second most important importing country of Ethiopian coffee is Saudi Arabia,
accounting for 12 percent of coffee exports.

Figure 12—Share of different destination markets for Ethiopia’s coffee exports in 2012

AUSTRALIA   U.K.  
JORDAN   2%   1%  
2%  
KOREA   Other  
2%   9%  
SWEDEN  
3%   SUDAN   GERMANY  
4%   33%  
ITALY  
5%  
FRANCE  
6%  
U.S.A.   S/ARABIA  
6%   12%  
JAPAN   BELGIUM  
7%   8%  
 
Source: Authors’ calculations based on data from the Ministry of Trade

Figure 13 shows how the share of the different destination markets has evolved over the
period 2005 to 2012. We note a number of changes. While the share of Japan was almost 20
percent in the year 2005, this declined to considerably lower levels in 2009, as Japan halted
deliveries of coffee from Ethiopia in May 2008 after finding abnormally high pesticide residues in
a shipment of coffee beans (Oakes and Thompson, 2009). This problem has since been ad-
dressed and the share of coffee exports to Japan is slowly increasing. We also see some
increase in diversification of destinations. While exports to countries outside Europe, North
America, Japan, and Saudi Arabia made up 5 percent of the export market in 2005, this share
was more than 10 percent in 2012. Moreover, while no coffee was exported to Sudan in 2006,
its share rose to 4 percent in 2012.

16  
  
Figure 13—Share of different destination markets in Ethiopia’s coffee exports
100%  
90%  
80%   Europe   SUMMARY  |  APRIL  201
70%   Japan  
60%   North  America  
50%   Saudi  Arabia  
40%  
Sudan  
30%  
Other  
20%  
10%  
0%  
2005   2006   2007   2008   2009   2010   2011   2012  
 
Source: Authors’ calculations based on data from the Ministry of Trade

Figure 14 shows the cumulative price distributions over the period that data were col-
lected for five major destination markets: North America, Europe, Saudi Arabia, Japan, and
Sudan. It shows different price patterns for these markets. The lowest price is noted for exports
to Sudan. These exports also show relatively little price variation, possibly because Sudan came
later into the market and was therefore less exposed to international price changes. The varia-
tion in prices for exports to North America is much wider than for all other countries. The graph
also indicates that North American prices are superior to the other countries as the cumulative
price distribution curve for North America lies beneath all other distribution curves. European
prices are second to North American ones, followed by Saudi Arabia and then Japan. This
variation in prices between countries and continents reflects mainly different quality demands.

Figure 14—Cumulative price distribution by destination market, 2005 to 2013

1
.8
.6
.4
.2
0

0 100 200 300 400 500

USC/LB

Europe Japan
NorthAmerica SaudiArabia
Sudan
 
Source: Authors’ calculations based on data from the Ministry of Trade

To analyze to what extent different characteristics of coffee exports are associated with
the different destination markets, we run a multinomial logistic regression with the destination
markets as dependent variables (using Europe as the default market) and quality characteristics
as explanatory variables, using a similar model to Boger (2001). The results are presented in

17  
  
Table 3. Overall, we find strong effects of different quality indicators on specific destination
markets, indicating possible market segmentation.

SUMMARY  |  APRIL  201

18  
  
Table 3—Multinomial logistic regression model of destination markets for Ethiopian coffee
(default  =  Europe)  
Mean Japan North-­‐America Middle  East Africa Other
Unit share Coef. z-­‐value Coef. z-­‐value Coef. z-­‐value Coef. z-­‐value Coef. z-­‐value
Washed  coffee  (grade  2  =  default) 0.37
SUMMARY  |  APRIL  201
Grade  1 yes=1 0.01 2.43 10.39 1.57 7.57 -­‐14.91 -­‐0.01 -­‐13.47 0.00 1.74 7.71
Un-­‐graded yes=1 0.02 -­‐1.20 -­‐4.04 -­‐0.84 -­‐6.49 -­‐1.03 -­‐2.02 2.16 2.84 -­‐1.40 -­‐5.89
Unwashed  coffee
Grade  3 yes=1 0.02 1.49 7.96 1.60 12.54 -­‐1.07 -­‐1.05 -­‐14.50 -­‐0.01 1.84 12.84
Grade  4 yes=1 0.18 2.66 37.69 0.81 12.67 1.36 9.61 -­‐1.48 -­‐1.07 1.73 25.02
Grade  5 yes=1 0.36 -­‐0.23 -­‐2.13 -­‐1.46 -­‐12.20 2.75 16.26 -­‐1.75 -­‐1.63 -­‐0.58 -­‐4.66
Un-­‐-­‐graded yes=1 0.05 1.18 6.77 -­‐0.47 -­‐2.35 2.99 12.89 5.18 4.87 -­‐0.39 -­‐1.50
Region  of  origin  (default=Sidama)
Jimma yes=1 0.24 0.55 5.57 0.93 7.92 -­‐1.94 -­‐12.55 4.60 4.90 0.82 6.96
Wollega/Lekempte yes=1 0.15 0.76 9.22 -­‐0.32 -­‐2.58 0.97 6.92 1.06 1.00 -­‐0.31 -­‐2.66
Yirgacheffe yes=1 0.12 0.69 7.21 1.41 25.70 -­‐1.04 -­‐3.12 0.74 0.86 1.39 19.75
Limu yes=1 0.05 -­‐0.34 -­‐1.98 0.76 9.92 -­‐0.76 -­‐2.04 0.66 0.58 0.89 9.22
Harar yes=1 0.07 0.02 0.11 1.90 15.96 3.53 24.76 -­‐15.10 -­‐0.01 1.61 13.46
Other yes=1 0.04 0.54 4.75 0.43 4.19 -­‐0.62 -­‐3.16 -­‐15.87 -­‐0.01 -­‐0.32 -­‐2.10
Certificate
Fair  Trade yes=1 0.02 -­‐2.19 -­‐3.69 -­‐0.14 -­‐1.27 -­‐0.65 -­‐0.81 -­‐14.04 -­‐0.01 0.08 0.53
Organic yes=1 0.09 0.15 0.97 0.72 8.41 -­‐4.70 -­‐4.59 -­‐15.28 -­‐0.01 0.57 4.73
Type  of  exporter  (default=private) 0.81
Cooperative yes=1 0.10 -­‐0.60 -­‐3.99 0.58 6.93 -­‐0.09 -­‐0.46 1.12 1.48 -­‐0.16 -­‐1.34
Parastatal yes=1 0.03 -­‐1.50 -­‐6.27 -­‐0.76 -­‐5.33 -­‐1.41 -­‐6.88 -­‐19.58 -­‐0.01 -­‐0.36 -­‐2.68
State  farm yes=1 0.03 0.06 0.38 -­‐0.63 -­‐4.49 -­‐0.90 -­‐2.11 -­‐18.37 -­‐0.01 -­‐0.10 -­‐0.69
Private  c ommercial  farm yes=1 0.03 -­‐0.09 -­‐0.55 1.44 16.80 -­‐0.20 -­‐0.73 -­‐18.36 -­‐0.01 0.09 0.65
Intercept -­‐2.70 -­‐45.09 -­‐1.93 -­‐48.57 -­‐3.67 -­‐35.87 -­‐7.34 -­‐12.74 -­‐2.46 -­‐47.58
Number  of  observations 30354
LR  c hi2(106) 27945
Prob  >  c hi2 0
Pseudo  R2 0.32
*  z-­‐values  i n  bold  a re  s ignificant  a t  the  5  percent  l evel  
Source: Authors’ calculations based on data from the Ministry of Trade

First, Japan and the Middle East are more likely to import unwashed coffee than are
other countries. The impression of washed coffee being cleaner is strong in the US and Europe.
Japan and the Middle East generally prefer unwashed coffee for a presumed better and richer
natural taste (Tefera and Tefera, 2013). These different preferences for washed and unwashed
coffee are illustrated in Figure 15. Second, we see significant differences in demand for different
grades. For example, Sudan is more likely to import ungraded coffee, both washed and un-
washed. Third, different destination markets import coffees from different specific regions. The
Middle East is much more likely to import coffee that originates from Harar and from Wollega
(Nekempt). Japan and North America import relatively more coffee from Yirgacheffe than do
European importers. African importers focus mostly on the lower priced coffees from Wollega
and Jimma. Fourth, coffee with Fair Trade certificates are more likely to be exported to Europe
than to other destinations. Organic coffee on the other hand is in greatest demand by North
American markets. These markets also import relatively more coffee from cooperatives than the
other destination markets. Overall, these results suggest that quality differences explain an
important part of the significant differences in prices between destination markets. This will be
further discussed below.

19  
  
Figure 15—Share of exports of washed and unwashed coffee by export destination, 2005 to 2013
0%   20%   40%   60%   80%   100%  

Other  
Africa   SUMMARY  |  APRIL  201
Middle  East  
North-­‐America  
Japan  
Europe  
Washed   Unwashed    
Source: Authors’ calculations based on data from the Ministry of Trade

7. PERFORMANCE OF ETHIOPIA’S COFFEE EXPORT SECTOR

7.1. Values and quantities
We start with a discussion of the performance of the coffee sector through an analysis of the
changes in quantities and values of coffee exports. Figure 16 shows how the values, in real
USD14, and quantities of coffee exports have evolved over the last 10 years. To illustrate trends
over time, we plot 12-month moving averages for value and quantity that we equate to 100
percent for January 2003 and then add linear trend lines to these plots. There are two main
findings. First, we see large growth rates in the values and quantities of coffee exported. Figure
16 illustrates that the real value of Ethiopia’s coffee exports at the end of 2012 was four times
higher than it was at the beginning of 2003. The value of exports in nominal terms was five
times as high. Quantities exported also increased by 50 percent over the same period. Even
though the growth rate in quantity terms is much smaller than for real values, this is still an
impressive performance. Second, we see a significant drop in the real value and the quantity of
exports from the trend line in 2009, indicating a bad production year as well as disruption to
coffee exports with the revocation of export licenses. However, exports have increased and
remained on an upward trend since then.

                                                                                                                       
14
 Deflated  by  the  US  CPI  downloaded  from  http://www.usinflationcalculator.com/inflation/consumer-­‐price-­‐index-­‐and-­‐annual-­‐
percent-­‐changes-­‐from-­‐1913-­‐to-­‐2008/  

20  
  
Figure 16—Trends in real value and quantities of coffee exports from Ethiopia, 2003 to 2012 (Jan.
2003=100; 12-month moving average)
450  
400   value   SUMMARY  |  APRIL  201
volume  
350  
Linear  (value)  
300   Linear  (volume)  
250  
Index  

200  
150  
100  
50  
0  

Source: Authors’ calculations based on data from the Ministry of Trade

The change in the real value of exports is largely driven by the increasing international
prices offered for Ethiopian coffee. In the next two sections, we explore to what extent interna-
tional price changes and quality changes (rewarded in quality premiums) in exported coffee
from Ethiopia, have contributed to this change.

7.2. Changes in international coffee prices

Figure 17 compares the export prices, as reported by the Ministry of Trade, and the composite
price of Brazilian natural Arabica coffee, as constructed by the International Coffee Organization
(ICO), based on coffee futures contracts in commodity exchanges in London and New York.
The graph illustrates the large variations found in coffee prices over time. Coffee prices in 2011
were five times higher than prices in 2002, with the price in 2011 at around 2.5 USD per lb,
compared to 0.5 USD per lb in 2002. While the price has come down significantly since its peak
in 2011, prices in mid-2013 had not yet declined to the low level of 2002. This large price
variability in recent years has led both to increasingly unpredictable windfalls and losses in
Ethiopia’s export sector. Several key informants reported that a lack of hedging instruments in
local coffee markets, such as futures contracts, was a source of weakness in the coffee sector
in the face of this price variability in Ethiopia. The graph further shows a strong correlation and
synchrony between export and international prices. The reported export prices are mostly higher
than the ICO composite price, reflecting the relatively good quality Arabica coffee produced in
Ethiopia. However, a drop below the ICO price is seen in 2010 and 2011. The reasons for this
are unclear, although this pattern in the relationship of the two prices has since been reversed.

21  
  
Figure 17—Trends in Ethiopian export coffee price versus reference international coffee price, 2002
to 2013

300  

NY/London  price   SUMMARY  |  APRIL  201

250  
export  price  
US  cents  per  lb  

200  

150  

100  

50  

0  

 
Source: Authors’ calculations based on data from the Ministry of Trade and International Coffee Organization (ICO)

In principle, all marketed coffee in Ethiopia has to go through the ECX to determine the
quality standard of the coffee. If the quality of coffee is judged to be of export quality, then it
must be exported and cannot be sold locally. This makes the local market a residual market,
where prices are in theory detached from international prices. Given the government’s need of
foreign currency to finance its ambitious development agenda, it may be the case that this
export requirement has become increasingly stringent over time and there has been less coffee
supplied to local markets. This pattern is possibly confirmed in Figure 18 where we compare the
retail prices for coffee in Addis Ababa markets, as collected by the Central Statistical Agency,
with export prices for coffee. We plot monthly differences as well as 12-month moving averages
of that difference to more accurately discern the trend.15 While the retail prices for coffee beans
in Addis Ababa was 20 cents lower per pound than export prices in the years 2002 and 2003,
this difference was reversed in 2011 and 2012 when local prices in Addis Ababa were 20 cents
or higher per pound than export prices. More recently, the price difference has readjusted so
that the prices are now considerably closer, possibly because of the development of a black
market (Fikade, 2014). Such informal markets are encouraged to flourish, given the possible
rewards obtained from channeling export quality coffee to domestic markets.

                                                                                                                       
 Unfortunately,  we  cannot  control  for  changes  in  coffee  quality  over  time,  but  it  seems  safe  to  assume  that  these  do  not  
15

explain  these  trends.  

22  
  
Figure 18—Difference between Addis Ababa retail coffee price and the Ethiopia coffee export price,
2002 to 2013 by month
60  
50   Dashed  line  is  actual  price  difference;  solid  line  is  
12-­‐month  moving  average  of  the  price  difference   SUMMARY  |  APRIL  201
40  
30  
US  cents  per  lb.  

20  
10  
0  
-­‐10  
-­‐20  
-­‐30  
-­‐40  

 
Source: Authors’ calculations based on data from the Ministry of Trade and CSA

In contrast with most coffee exporting countries, Ethiopia is itself a major consumer of
coffee. The improved export performance over the last decade raises the important question
about the extent to which this change is linked to high international prices and a subsequent
shift from domestic consumption to export markets. A negative price elasticity, as is commonly
seen for most agricultural products, would suggest that part of the total increase of coffee
exports is explained by reduced consumption. This is especially the case in a situation where
prices have changed dramatically, as in the last decade.
The analysis of the data from the National Household and Income Surveys (HICES) of
2000 and of 2011, indicate that coffee expenditures were 40 Birr per adult equivalent in 2000
and 206 Birr in 2011. When these expenditures are divided by the average annual retail prices
for whole dried coffee beans, this implies a consumption per adult equivalent of 8.7 kg in 2000
and 5.9 kg in 2011, or a decline by one-third in coffee consumption by Ethiopian households.
However, given that different methodologies were followed for the implementation of these
surveys over time, the interpretation should be treated with caution (Stifel and Woldehanna,
2013). However, the Ethiopian Rural Household Survey (ERHS), a comprehensive panel survey
in fifteen Ethiopian villages, shows that weekly household coffee consumption declined from 1.0
kg in 1997 to 0.7 kg in 2009, again a decline by 30 percent. Evidence from these consumption
surveys therefore seem to suggest that the higher exported quantities in the last decade are at
least partly driven by reduced local consumption of coffee.

7.3. Quality premiums offered for Ethiopian coffee

To arrive at the implicit price of product attributes, the hedonic pricing method is widely used. If
the marginal yield of most characteristics and the implicit price for each attribute is assumed
constant, a hedonic price regression can be estimated where the price of a product is a function
of the characteristics of the product, for example, through variety choices or post-harvest
technologies. A simple model of the following form can then be run:
N
p h = ∑ β kh X hk + ν
k =0

23  
  
where ph is the price of the product h, Xkh is the quantity of the attribute k of the product h, βkh
the implicit price, and ν a stochastic error term. Similar approaches have been used in coffee
pricing studies by, for example, Teuber (2010), Teuber and Herrmann (2012), and Donnet et al.
(2007, 2008). We discuss the results of similar exercises below. In all the regressions that were
SUMMARY  |  APRIL  201
run, robust standard errors were estimated.
In a first specification of the hedonic price model presented in Table 4, we regress the
logarithm of prices in US cents per lb on quantity exported, the origin of the coffee, certification,
type of exporters, and yearly dummy variables. We find a strong effect of origin on export prices.
Compared to the Sidama default value, Jimma coffee is exported at a 32 percent lower price. It
is the least valuable export coffee in Ethiopia. Wollega (Nekempt) also has a negative coeffi-
cient – its price is 23 percent lower than the Sidama price. In terms of place of origin, the most
valued coffees are from Yirgacheffe and Harar, with premiums of 19 percent and 7 percent,
respectively, over Sidama. We also find that certification raises prices significantly. Fair Trade
and Organic certification generate premiums of about 9 percent, other things being equal.
Finally, coffee exported by cooperatives and private commercial farms obtains higher export
prices compared to that exported by private exporters. The premium for cooperatives and
private commercial farms is 16 percent and 5 percent, respectively. As we control in the regres-
sion for the quality of the coffee and certification, this premium might partly reflect the willing-
ness of buyers to pay for vertical integration in sourcing their coffee from Ethiopia. We further
note significant changes in the year dummies, reflecting large international price variability over
the period considered.
In a second specification, we add a dummy variable to reflect the washing of coffee. We
find that washed coffee raises the export price by a premium of 20 percent on average over the
period considered. We also note that the coefficients on the region of origins change significant-
ly with the inclusion of the washed coffee dummy in the hedonic price model. There is now
significantly less variability between the regions. While the price range between the highest
(Yirgacheffe) and the lowest (Jimma) priced regions was almost 51 percent in the first model
specification, this was reduced to 38 percent in the second specification (Harar versus Jimma),
indicating that a significant part in regional price variation is explained by the processing method
followed. By controlling for washing, coffee originating from Harar is shown to be the most
expensive in the country, with an average price premium of 20 percent over Sidama.
In a third specification, we add within the washed and unwashed categories the different
quality grades as measured by the CLU. We find that the grades within each segment lead to
significantly different premiums, and that the premiums attached to the grades are consistent
with what would have been expected, i.e. better prices offered for better quality grades. In the
washed segment, grade 1 coffee is sold at a premium of 20 percent while un-graded washed
coffee is sold at a price that is 23 percent lower than grade 2 washed coffee. In the unwashed
segment, grade 5 is sold at a price 8 percent lower than grade 4, while grade 3 has a price that
is 32 percent higher than grade 4. Note that ungraded washed coffee is more valuable than
ungraded unwashed coffee.
Fourth, we interact the washing dummy with the year dummy, reflecting the returns to
washing for each year separately. We find that washing in all years adds a premium relative to
the price of unwashed coffee. In all cases, this interacted coefficient is significant and positive.
However, the premium is highly volatile between years, seemingly linked with the high volatility
of international prices. The premium was lowest during 2006 and 2007 at between 15 and 18
percent. The premium then increased to 34 percent and 24 percent in 2009 and 2010, respec-

24  
  
tively. In 2012, when prices decreased during the period of sales of the unwashed coffee, the
premium of washed coffee came down but still resulted in a premium of 21 percent over un-
washed coffee. This volatility in the price premium associated with washing coffee leads to
uncertain pay-offs to investments in coffee washing mills.
SUMMARY  |  APRIL  201
Table 4—Determinants of coffee prices (log(US cents per lb)); hedonic price model
Specification  1 Specification  2 Specification  3 Specification  4
Unit Coef. t-­‐value* Coef. t-­‐value* Coef. t-­‐value* Coef. t-­‐value*
Quantity  exported log() -­‐0.038 -­‐26.82 -­‐0.037 -­‐26.53 -­‐0.023 -­‐18.30 -­‐0.036 -­‐26.55
Washed  c offee yes=1 0.202 55.41
Washed  coffee  (grade  2  =  default)
Grade  1 yes=1 0.196 13.29
Un-­‐graded yes=1 -­‐0.230 -­‐39.77
Unwashed  coffee
Grade  3 yes=1 0.057 4.86
Grade  4 yes=1 -­‐0.259 -­‐88.13
Grade  5 yes=1 -­‐0.344 -­‐67.38
Un-­‐-­‐graded yes=1 -­‐0.497 -­‐64.17
Region  of  origin  (default=Sidama)
Jimma yes=1 -­‐0.320 -­‐116.37 -­‐0.188 -­‐52.02 -­‐0.051 -­‐10.00 -­‐0.186 -­‐52.13
Wollega/Lekempte yes=1 -­‐0.234 -­‐76.94 -­‐0.106 -­‐28.39 -­‐0.015 -­‐3.19 -­‐0.103 -­‐27.98
Yirgacheffe yes=1 0.192 55.25 0.136 37.34 0.104 35.63 0.135 37.73
Limu yes=1 0.006 1.34 -­‐0.041 -­‐9.26 -­‐0.055 -­‐13.65 -­‐0.043 -­‐9.83
Harar yes=1 0.068 14.04 0.200 37.10 0.290 51.52 0.202 36.65
Other yes=1 -­‐0.071 -­‐9.90 -­‐0.030 -­‐4.50 -­‐0.011 -­‐1.69 -­‐0.029 -­‐4.41
Certificate
Fair  Trade yes=1 0.089 13.43 0.053 8.21 0.057 8.97 0.043 6.53
Organic yes=1 0.095 13.46 0.096 13.94 0.076 12.07 0.098 14.22
Type  of  exporter
Cooperative yes=1 0.157 23.21 0.146 22.01 0.154 25.38 0.146 21.79
Parastatal yes=1 -­‐0.008 -­‐1.44 -­‐0.029 -­‐6.04 -­‐0.042 -­‐9.24 -­‐0.024 -­‐5.17
State  farm yes=1 -­‐0.018 -­‐3.04 -­‐0.063 -­‐11.04 -­‐0.082 -­‐14.02 -­‐0.065 -­‐11.55
Private  c ommercial  farm yes=1 0.048 6.33 0.041 5.38 0.030 5.08 0.041 5.51
Year  (default=2006)
2007 yes=1 0.112 26.57 0.101 27.47 0.093 27.43 0.115 27.13
2008 yes=1 0.343 83.49 0.332 92.90 0.319 98.31 0.335 83.60
2009 yes=1 0.222 43.77 0.208 47.40 0.204 51.22 0.140 26.64
2010 yes=1 0.335 79.37 0.324 89.45 0.319 97.00 0.305 71.68
2011 yes=1 0.753 157.44 0.745 171.10 0.744 181.94 0.756 134.65
2012 yes=1 0.671 150.9 0.664 167.81 0.651 177.18 0.655 138.13
2013 yes=1 0.445 83.35 0.436 89.10 0.415 93.68 0.444 64.34
Year  interacted  with  washing
2006 yes=1 0.183 23.21
2007 yes=1 0.155 30.43
2008 yes=1 0.180 34.29
2009 yes=1 0.345 51.42
2010 yes=1 0.237 49.16
2011 yes=1 0.160 26.09
2012 yes=1 0.212 34.68
2013 yes=1 0.174 21.33
Intercept 5.131 343.94 5.001 331.51 5.096 393.46 5.002 333.27
Number  of  observations 30333 30333 30207 30333
F() 6830.0 7870.6 8035.9 6239.5
Prob>F 0.00 0.00 0.00 0.00
R-­‐squared 0.78 0.81 0.85 0.81
*  r obust  s tandard  errors;  t-­‐values  i n  bold  a re  s ignificant  a t  the  5  percent  l evel

25  
  
Source: Authors’ calculations based on data from the Ministry of Trade

In a fifth model specification, we use the second model specification, but interacted year-
ly dummies with geographical indicators and split up the regression between washed and
unwashed coffee. In Figure 19, we plot the quality premiums compared to coffee of Sidama SUMMARY  |  APRIL  201
origin. We see increasing divergence of quality premiums for unwashed coffee over time. While
the range in quality premiums between 2006 and 2008 was about 20 percent, this had in-
creased to 90 percent in 2012 and 2013. This increase in variation has been driven largely by
the substantial upswing in the Harar coffee premium. However, even if we discard Harar coffee
from our analysis, we still see slightly increasing divergence over time, with the premiums
offered for unwashed coffee from Wollega (Nekempt) and Jimma not keeping up with premiums
offered for unwashed Sidama coffee. This divergence might possibly be explained by the rise of
specialty coffee which fetches very high prices in international markets (Teuber and Herrmann,
2012). In the case of washed coffee, changes in premiums over time are less pronounced.

Figure 19—Regional coffee premiums over time, compared to Sidama coffee

a. Washed   b. Unwashed  
70%   70%  
Yirgacheffe   Jimma  
50%   Limu   50%   Wollega  
Harar  
30%   30%  

10%   10%  

-­‐10%   -­‐10%  

-­‐30%   -­‐30%  
2006  2007  2008  2009  2010  2011  2012  2013     2006  2007  2008  2009  2010  2011  2012  2013    
Source: Authors’ calculations based on data from the Ministry of Trade

Sixth, we run a test on the rewards for vertical integration and traceability. Commercial
private farms, state farms, and cooperatives produce coffee themselves or buy it directly from
producers, and are permitted to sell directly to international buyers. This gives them the benefits
of assuring traceability, which is increasingly demanded in international markets (Meijerink et
al., 2010; Swinnen, 2007). Since the introduction of the ECX, other buyers and exporters can no
longer sell directly to international buyers, but must sell their coffee through the ECX. To test to
what extent management structures that allow vertical integration have increased quality
premiums after the introduction of the ECX compared to the situation before, we interact in the
regression the management structures with a dummy variable for before and after the ECX and
compare differences through an F-test.16 The results are presented in Table 5. While coopera-
tives obtained a price premium compared to other traders before the introduction the ECX, this
premium gap has widened significantly subsequently. The price premium of cooperatives
compared to private traders widened by 9 percent, from 9 percent before ECX to 18 percent in
the period afterwards. Significant but smaller changes are also found for other vertically inte-
grated structures – 3 percent for state farms and 5 percent for commercial private farms.
However, they both started from significantly lower levels of sales.

                                                                                                                       
 However,  caution  in  interpretation  is  required  as  it  is  not  directly  possible  to  separate  ECX,  time,  and  cohort  effects.  
16

Untangling  this  is  left  for  future  analysis.  

26  
  
Table 5—Testing the benefits of vertical integration and traceability in the Ethiopian coffee export
sector

Before  ECX After  ECX

Unit Coef. t-­‐value* Coef. t-­‐value* SUMMARY  |  APRIL  201
Regression  coefficients
Type  of  exporter
Cooperative yes=1 0.091 13.72 0.181 23.38
Private  c ommercial  farm yes=1 -­‐0.007 -­‐0.49 0.055 6.30
State  farms yes=1 -­‐0.082 -­‐11.93 -­‐0.047 -­‐6.22
Control  variables  s imilar  a s  i n  s pecification  2
F-­‐test  of  differences  between  period  before  ECX  and  after  ECX
F() Prob>F
Cooperative 200.62 0.00
State  farms 15.28 0.00
Private  c ommercial  farm 15.26 0.00
*  r obust  s tandard  errors;  t-­‐values  i n  bold  a re  s ignificant  a t  the  5  percent  l evel  
Source: Source: Authors’ calculations based on data from the Ministry of Trade

Seventh, to understand international market segmentation, different hedonic regression

models for different destination markets are required (Teuber and Herrman, 2012). The results
of these regression models are presented in Table 6. A number of interesting findings are
obtained. First, similar patterns to those detected in Table 4 emerge over different countries –
better grades are associated with higher prices; there are significant differences in prices for
coffee from different regions; and certification raises the price of coffee. However, the size of
these coefficients differs over destination markets. Second, ungraded coffee has the lowest
negative premium in the case of exports to other African countries. The ungraded coffee that is
exported to Japan gets the best price of all ungraded exported coffee. Third, organic certificates
are rewarded the highest prices in the Middle Eastern and North American markets, while Fair
Trade coffee gets the best prices in Europe. Fourth, compared to Europe, price rises over time
have been higher in Japan. Prices for coffee exported to North America, on the other hand
follows, the prices for coffee exported to European markets more closely, while Middle Eastern
and African markets follow them less closely, other things being equal. Fifth, while vertical
integration is rewarded in the European, Japanese, and North American markets, this is seem-
ingly not the case in the Middle East.

27  
  
Table 6—Determinants of coffee prices (log(US cents per lb)) by destination market; hedonic price
model
Europe Japan North-­‐America Middle  East Africa
Unit Coef. t-­‐value* Coef. t-­‐value* Coef. t-­‐value* Coef. t-­‐value* Coef. t-­‐value*
Quantity  exported log() -­‐0.02 -­‐13.43 -­‐0.02 -­‐6.24 -­‐0.02 -­‐7.08 0.00 SUMMARY  
-­‐0.20 -­‐0.01|  APRIL  
-­‐1.45 201
Washed  coffee  (grade  2  =  default)
Grade  1 yes=1 0.23 6.55 0.12 3.79 0.19 9.63
Un-­‐graded yes=1 -­‐0.24 -­‐41.60 -­‐0.16 -­‐2.73 -­‐0.23 -­‐9.25 -­‐0.27 -­‐2.58 -­‐0.61 -­‐6.66
Unwashed  coffee
Grade  3 yes=1 0.06 2.23 -­‐0.03 -­‐0.62 0.06 3.93 -­‐0.32 -­‐20.05
Grade  4 yes=1 -­‐0.29 -­‐65.52 -­‐0.25 -­‐33.87 -­‐0.19 -­‐22.61 -­‐0.18 -­‐8.57 -­‐0.58 -­‐46.67
Grade  5 yes=1 -­‐0.32 -­‐37.59 -­‐0.32 -­‐19.69 -­‐0.28 -­‐11.42 -­‐0.34 -­‐13.59 -­‐0.63 -­‐12.92
Un-­‐-­‐graded yes=1 -­‐0.40 -­‐28.30 -­‐0.21 -­‐8.93 -­‐0.44 -­‐7.11 -­‐0.43 -­‐13.76 -­‐0.68 -­‐14.21
Region  of  origin  (default=Sidama)
Jimma yes=1 -­‐0.08 -­‐9.46 -­‐0.01 -­‐0.87 -­‐0.13 -­‐5.42 -­‐0.08 -­‐3.81 -­‐0.09 -­‐2.56
Wollega/Lekempte yes=1 -­‐0.05 -­‐5.50 0.02 3.18 -­‐0.12 -­‐6.18 -­‐0.05 -­‐2.61
Yirgacheffe yes=1 0.15 34.60 0.09 6.19 0.06 10.10 0.10 2.37 0.06 3.54
Limu yes=1 -­‐0.06 -­‐12.71 -­‐0.05 -­‐1.94 -­‐0.07 -­‐8.28 0.01 0.29 -­‐0.69 -­‐17.24
Harar yes=1 0.33 18.38 0.30 11.68 0.22 19.49 0.25 12.72
Other yes=1 -­‐0.06 -­‐7.01 0.10 5.80 -­‐0.06 -­‐3.62 0.01 0.43
Certificate
Fair  Trade yes=1 0.08 9.21 -­‐0.12 -­‐2.15 0.03 2.20 0.00 -­‐0.02
Organic yes=1 0.05 5.49 0.06 2.78 0.08 7.17 0.13 2.62
Type  of  exporter
Cooperative yes=1 0.16 18.28 0.12 6.02 0.15 14.01 0.02 0.53 0.25 15.37
Parastatal yes=1 -­‐0.04 -­‐7.36 -­‐0.06 -­‐3.50 -­‐0.07 -­‐3.84 -­‐0.10 -­‐4.18
State  farm yes=1 -­‐0.07 -­‐11.42 -­‐0.20 -­‐9.49 -­‐0.06 -­‐2.90 -­‐0.06 -­‐0.77
Private  c ommercial  farm yes=1 0.00 0.03 0.07 2.34 0.05 4.22 -­‐0.04 -­‐1.23
Year  (default=2006)
2007 yes=1 0.08 19.49 0.13 27.54 0.03 1.84 0.13 12.19 0.09 4.54
2008 yes=1 0.31 79.35 0.35 63.99 0.30 16.34 0.34 39.33 0.20 9.54
2009 yes=1 0.19 40.08 0.38 28.49 0.23 11.72 0.22 21.61 -­‐0.07 -­‐2.08
2010 yes=1 0.31 80.24 0.43 47.79 0.32 17.07 0.36 40.91 0.06 2.65
2011 yes=1 0.75 157.49 0.83 61.03 0.69 35.64 0.74 56.82 0.56 20.48
2012 yes=1 0.62 144.74 0.60 61.85 0.68 35.06 0.72 64.85 0.58 25.06
2013 yes=1 0.37 72.32 0.43 41.52 0.44 21.58 0.56 37.82 0.24 10.51
Intercept 5.10 306.68 5.03 156.20 5.14 127.53 4.85 91.39 5.34 63.84
Number  of  observations 15540 2885 4167 3643 1229
F() 4834 872 966 . .
Prob>F 0 0 0 . .
R-­‐squared 0.87 0.87 0.84 0.76 0.60
*  r obust  s tandard  errors;  t-­‐values  i n  bold  a re  s ignificant  a t  the  5  percent  l evel  
Source: Source: Authors’ calculations based on data from the Ministry of Trade

Eighth, we run a simultaneous quantile regression in order to understand the relative im-
portance of different price determinants at lower and higher ends of the markets (e.g. Costani-
gro et al., 2010). The results for 5 quantile - at 0.10, 0.25, 0.50, 0.75 and 0.90 quantiles -
regressions are presented in Table 7. We find that quality premiums for indications of origins are
increasing from lower to higher quantiles, indicating the greater importance of origin effects in
high-end markets. While the difference between the lowest (Limu) and the highest (Harar)
coefficients for the 0.10 quantile regression shows a difference of 27 percent, this gap increases
to 44 percent for the 0.90 quantile regression. We also note increasing willingness to pay for
vertical integration at the higher end of the market as shown by higher positive coefficients for

28  
  
cooperatives (17 percent and 11 percent for the 0.90 and 0.10 quantile regression respectively)
and for commercial private farms (7 percent and 2 percent for the 0.90 and 0.10 quantile
regressions, respectively). Parastatals are less able to compete at the high end of the market as
shown by its larger negative coefficient in that market segment when compared to other export-
SUMMARY  |  APRIL  201
ers as well as to lower market segment. Rewards to grades and certificates decline from low to
high quantiles, possibly indicating absolute (a fixed amount) instead of relative rewards.

Table 7—Determinants of coffee prices (log(US cents per lb)); quantile hedonic price models
quantile  0.10 quantile  0.25 quantile  0.50 quantile  0.75 quantile  0.90
Unit Coef. t-­‐value* Coef. t-­‐value* Coef. t-­‐value* Coef. t-­‐value* Coef. t-­‐value*
Quantity  exported log() -­‐0.013 -­‐12.05 -­‐0.018 -­‐15.53 -­‐0.018 -­‐17.39 -­‐0.021 -­‐18.49 -­‐0.031 -­‐14.88
Washed  coffee  (grade  2  =  default)
Grade  1 yes=1 0.125 5.87 0.181 10.51 0.205 13.19 0.227 8.16 0.245 6.42
Un-­‐graded yes=1 -­‐0.232 -­‐39.49 -­‐0.244 -­‐48.99 -­‐0.243 -­‐34.07 -­‐0.211 -­‐29.54 -­‐0.187 -­‐22.17
Unwashed  coffee
Grade  3 yes=1 -­‐0.144 -­‐12.95 -­‐0.012 -­‐0.54 0.087 4.82 0.151 19.24 0.179 12.41
Grade  4 yes=1 -­‐0.296 -­‐103.60 -­‐0.289 -­‐90.16 -­‐0.267 -­‐67.19 -­‐0.229 -­‐59.69 -­‐0.226 -­‐34.63
Grade  5 yes=1 -­‐0.363 -­‐48.59 -­‐0.379 -­‐66.64 -­‐0.350 -­‐56.46 -­‐0.314 -­‐48.32 -­‐0.319 -­‐28.49
Un-­‐-­‐graded yes=1 -­‐0.624 -­‐37.02 -­‐0.570 -­‐60.64 -­‐0.495 -­‐51.87 -­‐0.415 -­‐43.74 -­‐0.406 -­‐29.55
Region  of  origin  (default=Sidama)
Jimma yes=1 -­‐0.015 -­‐2.04 -­‐0.024 -­‐4.41 -­‐0.048 -­‐8.07 -­‐0.053 -­‐8.21 -­‐0.040 -­‐3.99
Wollega/Lekempte yes=1 0.012 1.63 0.008 1.81 -­‐0.012 -­‐1.89 -­‐0.011 -­‐1.77 0.002 0.17
Yirgacheffe yes=1 0.101 20.25 0.114 28.65 0.102 32.47 0.116 28.36 0.177 16.30
Limu yes=1 -­‐0.067 -­‐9.17 -­‐0.046 -­‐10.74 -­‐0.048 -­‐13.40 -­‐0.061 -­‐13.75 -­‐0.065 -­‐11.11
Harar yes=1 0.202 21.95 0.266 46.92 0.271 46.20 0.290 38.80 0.380 29.15
Other yes=1 -­‐0.060 -­‐7.35 -­‐0.035 -­‐3.83 -­‐0.041 -­‐5.51 -­‐0.003 -­‐0.28 0.048 3.44
Certificate
Fair  Trade yes=1 0.106 15.42 0.051 4.79 0.052 5.44 0.037 7.17 0.034 3.45
Organic yes=1 0.067 6.52 0.059 7.74 0.073 10.49 0.065 11.35 0.038 4.58
Type  of  exporter
Cooperative yes=1 0.115 10.42 0.161 21.79 0.170 27.44 0.177 33.22 0.175 23.99
Parastatal yes=1 -­‐0.010 -­‐1.34 -­‐0.020 -­‐4.71 -­‐0.043 -­‐6.76 -­‐0.041 -­‐6.82 -­‐0.071 -­‐14.01
State  farm yes=1 -­‐0.069 -­‐8.32 -­‐0.089 -­‐10.94 -­‐0.077 -­‐13.94 -­‐0.052 -­‐8.57 -­‐0.077 -­‐8.21
Private  c ommercial  farm yes=1 0.019 2.62 0.008 1.52 0.016 2.75 0.066 7.55 0.069 6.72
Year  (default=2006)
2007 yes=1 0.070 14.33 0.070 27.72 0.089 29.31 0.112 34.14 0.136 16.94
2008 yes=1 0.286 53.22 0.315 119.94 0.318 106.88 0.327 109.20 0.341 42.27
2009 yes=1 0.085 14.64 0.138 21.83 0.227 46.38 0.264 64.55 0.296 28.34
2010 yes=1 0.268 54.24 0.292 88.15 0.313 125.15 0.349 97.27 0.370 45.14
2011 yes=1 0.621 102.36 0.690 160.71 0.775 182.57 0.833 198.02 0.852 114.84
2012 yes=1 0.552 111.04 0.584 177.93 0.631 106.36 0.732 175.36 0.747 97.58
2013 yes=1 0.339 67.26 0.352 117.12 0.383 84.56 0.449 89.70 0.513 52.33
Intercept 4.918 416.91 5.004 412.17 5.044 465.55 5.098 385.94 5.253 231.49
Number  of  observations 30207
Pseudo  R2 0.60 0.63 0.64 0.64 0.62
*  s multaneous  quantile  r egressions;  s tandard  errors  with  100  bootstrap  r eplications;  
t-­‐values  i n  bold  a re  s ignificant  a t  the  5  percent  l evel  
Source: Source: Authors’ calculations based on data from the Ministry of Trade

8. DISCUSSION AND CONCLUSIONS

We study the structure and performance of the coffee export sector over the period from 2003 to
2013 in Ethiopia. Important structural changes are happening in coffee markets internationally
and locally within Ethiopia with significant implications on performance and, ultimately, on the

29  
  
livelihoods of local smallholder coffee producers, given their high dependence on coffee as a
major source of income. However, few studies have looked at the impact of these changes on
the organization of export markets for coffee exporting countries.
Major changes have been seen in international coffee markets in recent years. First, SUMMARY  
the |  APRIL  201
specialty coffee market has quickly grown in global markets. Coffee differentiation, often based
on product origin, is becoming increasingly important (Reuber, 2010). Second, there has been
high price variability in global coffee markets with prices on international markets increasing
five-fold between 2000 and 2012. Third, the demand for certified coffee is on the rise (Jena et
al., 2012; Meijerink et al., 2010). Domestically, there have also been important policy changes
related to the start of the Ethiopian Commodity Exchange (ECX), a modern marketing system
based on warehouse receipts, interventions by the government in the issuing of export licenses,
and changes in the export tax structure.
We note important structural changes in the export sector in Ethiopia. We see a larger
number of exporters and an increasing diversification in the types of exporters – such as coop-
eratives and commercial farms - in the export sector, with lower concentration ratios in the
export sector overall. On the other hand, the share of the incumbents in the local coffee market
is large. This is possibly driven by reputation effects in these international markets (Macchi-
avelo, 2010), the complications of understanding coffee markets, as well as lack of access to
credit for emerging exporters (World Bank, 2014). There is over time relatively little change in
the countries that Ethiopia exports to, except for a small increase of exports to Sudan. We also
note the small but increasing share of parastatals and cooperatives in the coffee export sector
as well as more emphasis on certification of coffee, although the share of coffee exports from
Ethiopia that is certified is still small compared to other producing countries.
We have seen an impressive performance of Ethiopia’s coffee export sector over the last
decade with the real value of coffee exports rising four-fold between 2003 and 2012. The
increases in export values have largely been driven by increases in international coffee prices
between 2003 and 2012. However, it should be noted that increases in quantity also have
played a role, as the quantity exported has gone up as well – exported quantities from Ethiopia
were 50 percent higher in 2012 than 10 years earlier. The increasing quantity exported may
have partly come at the expense of local consumption, with coffee in local markets becoming
increasingly rationed over time, as shown in relatively important domestic price increases. We
also note a small increase in the quality of the coffee exported, reflected in the share of certified
coffee as well as in the better quality of unwashed coffee. The share of the premium washed
coffee in total exports, however, has not changed over time.
We show that the coffee export market is highly differentiated in Ethiopia, with quality
premiums being offered for washing, grades, certification, and specific geographical indications.
About 30 percent of the coffee is washed, leading to high but variable (due to international price
variation) premiums compared to unwashed coffee. There are strong effects of geographic
indications of origin with Yirgacheffe and Harar coffee commanding large premiums over
coffees originating from elsewhere in Ethiopia. Certified coffee and coffee marketed by coopera-
tives are being sold at higher prices than coffee sold by the private sector. We also find that all
exporters that are vertically integrated are able to obtain significantly higher prices over time
than those that do not. This is especially valued in the differentiated high-end market.
The findings from this research point to a number of important implications for efforts to
stimulate Ethiopia’s coffee export sector. First, emphasis on stimulating washing of coffee by
creating the right investment incentives is important, as it leads to significantly higher export

30  
  
prices and, thus, higher foreign exchange earnings for the country. As only part of the exported
coffee is sold as washed coffee, there is still room for growth in this area. However, profitability
for the setting up of wet mills needs to be carefully assessed, as does the environmental impli-
cations of more widespread washing (Beyene et al., 2012; Endris et al., 2008). Improved
SUMMARY  |  APRIL  201
processing methods, relying on ecological pulpers which reduce water demand significantly,
should be more widely adopted. The growth of washing coffee in Ethiopia may have been
constrained because of regulations and the difficulty of access to foreign exchange to import
wet milling machines. However, due to climatic constraints, coffee cherries produced in some
areas of Ethiopia are not suitable for washing, thus limiting spatial growth in wet milling services
to coffee producers.
Second, it has been estimated that between 20 and 30 percent of Ethiopian coffee could
qualify as specialty coffee17, opening up important export opportunities (Chemonics, 2010).
Greater efforts to gain increased access for Ethiopian coffee to global specialty coffee markets
could be made. Ethiopia could also benefit from increasing the profile of the Ethiopian coffee
brand, increased participation in certification schemes, and emphasizing trade in traceable
coffee products. This is especially important given the growing emphasis of international mar-
kets on certification of sustainable coffee production practices.18 However, as certification is
often a costly process that might also limit benefits for producers (Jena et al., 2012; Cramer et
al., 2014), setting up cheaper local certification schemes that would be credible in international
markets and that would benefit local producers directly should be considered. Furthermore, as
many of these international certification schemes require collaboration with cooperative institu-
tions, further capacity building of these institutions will be required.
Finally, coffee yields are very low in Ethiopia compared to other countries. There seem-
ingly are significant opportunities for productivity growth (Adugna et al., 2008; Kufa, 2012).
Increasing support at the farm level and training towards higher adoption of improved technolo-
gies, such as mulching, pruning, rejuvenation of trees, planting of improved varieties, and
modern input use, have been shown to be associated with higher productivity (Adugna et al.,
2008) and could lead to higher local supply and, therefore, quantities of coffee exported. Moreo-
ver, more efforts should be done with the research and agricultural extension services to
address widespread concerns related to coffee diseases and adaptation and mitigation of
climate change that are likely to have major impacts on the coffee sector in Ethiopia (Davis et
al., 2012).
 

                                                                                                                       
 Carl  Cervone,  Technoserve,  personal  communication  
17

 For  example,  the  Sustainable  Trade  Initiative  has  an  agreement  with  major  roasters  towards  increasing  global  sustainable  
18

coffee  sales  to  25  percent  in  2016.  See  http://www.idhsustainabletrade.com/koffie-­‐news  

31  
  
REFERENCES
Adugna, G., B. Bellachew, T. Shimber, E. Taye, and, T. Kufa. 2008. Coffee diversity & knowledge.
Ethiopian Institute of Agricultural Research. Addis Ababa, Ethiopia
SUMMARY  |  APRIL  201
Agrer, 2014. Formulation of Coffee Development Strategy for Ethiopia, mimeo.

Alemu, H. 2009. Through the grinds, Fortune, Edition of 06/21/2009, downloaded from:
http://www.addisfortune.com/vol%2010%20no%20476%20archive/Through%20the%20Grinds.htm

Araya, E. 2011. Ethiopia: Trade Ministry bans 16 coffee exporters from exchange floor, Addis Fortune,
st
November 21 , 2011, accessed at http://allafrica.com/stories/201111240122.html

Arslan, A., and, C.P. Reicher. 2010. The effects of the coffee trademarking initiative and Starbucks
publicity on export prices of Ethiopian coffee. Kiel Institute for the World Economy, Kiel Working Paper
1606

Beyene, A., Kassahun, Y., Addis, T., Assefa, F., Amsalu, A., Legesse, W., Kloos, H., and, Triest, L. 2012.
The impact of traditional coffee processing on river water quality in Ethiopia and the urgency of adopt-
ing sound environmental practices. Environmental Monitoring and Assessment, 184(11): 7053–7063.
doi:10.1007/s10661-011-2479-7

Boger, S. 2001. “Quality and contractual choice: A transaction cost approach to the Polish hog market.”
European Review of Agricultural Economics 28(3): 241-261

Chemonics. 2010. Ethiopia coffee industry value chain analysis: Profiling the actors, their interactions,
costs, constraints and opportunities, COMPETE/USAID, mimeo

CSA (Central Statistical Agency). 2013. Agricultural Sample Survey 2012/2013 (2005 E.C.), Report on
area and production of major crops. Central Statistical Agency, Addis Ababa

Costanigro, M., J. McCluskey, and, Goemans, C. 2010. “The economics of nested names: Name specific-
ity, reputations, and price premia”. American Journal of Agricultural Economics 92(5): 1339–1350

Cramer, C., Johnston, D., Oya, C., Sender, J. 2014. Fair trade, employment and poverty reduction in
Ethiopia and Uganda, Final report submitted to DfID, SOAS, University of London, April 2014

Daviron, B., and, S. Ponte. 2005. The Coffee Paradox, Global Markets, Commodity Trade and the Elusive
Promise of Development. Zed Books, London, New York.

Davis, A.P., Tadesse, W., Baena, S., Moat, J. 2012. The impact of climate change on indigenous Arabica
coffee (Coffee arabica): Predicting future trends and identifying priorities. PLoS ONE 7(11):e4798.
Accessed at doi:10.1371/journal.pone.0047981

Donnet, M.L., Weatherspoon, D.D., Hoehn, J.P. (2008), Price determinants in top-quality e-auctioned
specialty coffee, Agricultural Economics, 38:267-276

Donnet, M.L., D.D. Weatherspoon, and, Hoehn, J.P. 2007. “What adds value in specialty coffee? Mana-
gerial implications from hedonic price analysis of Central and South American E-auctions”. Interna-
tional Food and Agribusiness Management Review 10(3):1-18

Endris, S., T. Kebede, B. W/Sember, W. Legesse, Y. Kassahun, T. Yakob, and A. Sualeh. “Environmental
contamination from by-products of wet coffee processing.” in edited by G. Adugna, B. Bellachew, T.
Shimber, E. Taye, T. Kufa. 2008. Coffee diversity & knowledge. Ethiopian Institute of Agricultural Re-
search. Addis Ababa, Ethiopia: 335-344
st st
Fikade, B. 2014. Trouble brewing for coffee exporters, Fortune, February 1 , 2014, accessed on June 1 ,
2014 at http://www.thereporterethiopia.com/index.php/news-headlines/item/1572-trouble-brewing-for-
coffee-exports

32  
  
International Coffee Organization (ICO) (2013), http://www.ico.org/new_historical.asp

Gabre-Madhin, E.Z. 2012. A market for Abdu: Creating a Commodity Exchange in Ethiopia. Occasional
Essays Series. Washington, DC: International Food Policy Research Institute (IFPRI).

Jena, P.R., B.B. Chichaibelu, T. Stellmacher, and U. Grote. 2012. “The Impact of Coffee Certification SUMMARY  
on |  APRIL  201
Small-Scale Producers’ Livelihoods: A Case Study from the Jimma Zone”. Ethiopia, Agricultural Eco-
nomics 43(4): 429-440.

Kufa, T. 2012. Recent coffee research development in Ethiopia, Presentation at the “Ethiopian Coffee
Export Conference: Strengthening the Legacy of Our Coffee”, November 8-9, 2012, Hilton, Addis Ab-
aba

LMC. 2003. Review of the Ethiopian Coffee Market. Report prepared for DFID, Programme of Advisory
Support Services for Rural Livelihoods. LMC International Ltd, Oxford. December 2003.

Macchiavelo, R. 2010. Development uncorked: Reputation acquisition in the new market for Chilean
wines in the UK, mimeo

Meijerink, G.W., Alemu, D., Van der Mheen-Sluier, J., Wijnands, J. (2010). Institutional innovation in
African markets: Can commodity exchanges address sustainability issues? Paper presented at the
th th nd
119 EAAE seminar, Capri, Italy June 30 -July 2 , 2010

Nure, D. 2008. Mapping quality profiles of Ethiopian coffee by origin, in Eds. Adugna, G., Bellachew, B.,
Shimber, T., Taye, E., Kufa, T. 2008. Coffee diversity & knowledge, Ethiopian Institute of Agricultural
Research, Addis Ababa, Ethiopia: 317-227

Oakes, N., and, M. Thompson. 2009. Ethiopian coffee: Brewing a recovery, The Africa Group, accessed
at:
http://www.theafricagroup.com/static/django_sites/consulting/cms_page_media/8/Ethiopian%20Coffee
.pdf

Panhuysen, S., and, M. Van Reenen, 2012. Coffee Barometer 2012, Tropical Commodity Coalition

Pendergrast, M. 2010. Uncommon grounds: The history of coffee and how it transformed our world, Basic
Books, New York, 424p.

Petit, N. 2007. “Ethiopia’s coffee sector: A better or bitter future?” Journal of Agrarian Change. 7(2): 225-
263

Ponte, S. 2002. “The ‘Latte Revolution’? Regulation, Markets and Consumption in the Global Coffee
Chain”. World Development 30(7): 1099-1122

Rhinehart, R. 2012. What is special about specialty coffee? Paper presented at the SCAA symposium,
th st
April 18/19 2012, accessed on May 1 , 2014 at
http://www.scaaevent.org/PDF/Press%20Kit/2012/What%20is%20special%20about%20specialty.pdf

Stifel, D., and, T. Woldehanna. 2013. Utility-consistent poverty in Ethiopia, 2000-2011: Welfare improve-
ments in a changing economic landscape, paper prepared for the UNU-Wider Growth and Poverty
Project, 2012-2013

Swinnen, J. 2007. Global supply chains, standards and the poor. Wallingford, UK: CABI Publishing.

Tefera, A., and T. Tefera. 2013. Ethiopia: Coffee Annual Report, Gain report ET-1302, USDA

Teuber, R. 2010. “Geographical Indications of origin as a tool of product differentiation: The case of
coffee”. Journal of International Food and Agribusiness Marketing 22: 277-298

Teuber, R., and, R. Herrmann. 2012. “Towards a differentiated modeling of origin effects in hedonic
analysis: An application to auction prices of specialty coffee”. Food Policy 37: 732-740

33  
  
World Bank. 2014. 'Economic Update III: Strengthening Export Performance through Improved Competi-
About the AuthorsWashington D.C.
tiveness',
Bart Minten is a Program Leader of the Ethiopia Strategy Support Program, and a Senior Research Fellow in thethDevelopment Strat-
Yewondwossen, M. 2014. Two major coffee traders suspended from ECX, Capital, February 10 2014
egy and Governance Division of the International Food Policy Research Institute (IFPRI). Seneshaw Tamru is a PhD Candidate of
accessed
LICOS, Center at
for Institutions SUMMARY  
and Economic Performance of the University of Leuven. Tedesse Kuma is a Senior Researcher in |  APRIL  201
Agirulture http://www.capitalethiopia.com/index.php?option=com_content&view=article&id=4028:two-major-
and Rural Development of the Ethiopian Development Research Institute (EDRI). Yaw Nyarko is a Professor of Economics
at New Yorkcoffee-traders-suspended-from-ecx&catid=35:capital&Itemid=27
Universirty, and Director in Center for Technology and Economic Development.

ETHIOPIAN DEVELOPMENT RESEARCH INSTITUTE

P.O. Box 2479, Addis Ababa, Ethiopia | T: +251.11.550.6066 ; +251.11.553.8633 | F: +251.11.550.5588 |
[email protected] | www.edri-eth.org
ACKNOWLEDGMENTS
The authors benefited from comments of participants at seminars at the Ethiopian Development Research Institute (EDRI) and
the Ministry of Agriculture (MoA) in Addis Ababa. They would like to thank John Mellor and Carl Cervone for useful comments
on earlier drafts of the paper, Ibrahim Worku for the analysis on consumption patterns, and Helina Tilahun for making the
maps. However, the authors are solely responsible for content.

34