By Asefa Belachew (firstname.lastname@example.org)
Individuals and their political, civic, and religious Diaspora organizations provide different numbers on the size and composition of the population of Ethiopia with respect to their religion and ethnic background. Their discourses are often made on the basis of their impression or wishes without reference to any data source or they distort the numbers to fit their conclusion. At least one organization does not want to use the existing numbers and dissuades others from making any reference to the numbers.
The primary purpose of this essay is therefore to share some descriptive data based on the three censuses of Ethiopia to foster a common understanding and muster a common knowledge on these issues. There have been previous attempts to provide Census data and analysis. For instance, Abate reviews the Census of 1984 with a view to providing numeric foundations for our discussions. In a like manner, Berhanu provides a profile of the population based on the Census of 1994. However, the popular discussion fails to make references to these analysis and data.
This essay aims to up-date those studies and provides a description of the characteristic of the population based on the three Censuses of 1984, 1994 and 2007. It will in particular delve deeper into the Census of 2007. Census data are often contentious and the essay will point out the contentions regarding the 2007 Census.
The Census of 1984 was obviously carried out during the Derg regime, while the latter two were conducted during the EPRDF period. Several changes have occurred during this time that impacted Census-taking. First, Ethiopia moved from the Derg-era provincial administrative structure to the Ethnic-based regional structure during the EPRDF regime. As a result, the geographic boundaries have been re-configured and the land and population sizes reshuffled. The former provinces have been discarded, and only one that remains by name is Tigray. Second, small towns have become full-fledged region or Special Region (Harar and Dire Dawa, respectively), while several population groups with over a million population each are lumped under one region (e.g. Sidama, Hadiya, Wolaita, etc). Third, population groups that were reported separately (e.g. Borena and Guji) under the 1984 Census are rightfully reported under Oromia in 1994 and 2007 Censuses. These and similar factors make comparison across the three censuses a difficult task, while it is relatively easier to compare the 1994 and 2007 Census results. With these and similar caveat some important lessons could be gleaned by examining the output of the three censuses together.
- Trend Analysis of the Three Censuses
Ethiopia has conducted three censuses since the mid-1980s. These have been supplemented by a number of national or urban family and fertility surveys as well as three full-fledged Demographic and Health Surveys (DHS) and a mini-DHS in 2014. Prior to 1984, the year the first census was carried out, estimates of the population was done on the basis of several sample surveys.
|Table 1: Comparison of Population in the Three Censuses – by Region|
|Population in 000||Growth Rate||Projection||2007|
|1984*||1994||2007||1984-94||1994-2007||2007||Proj. – Actual||Dif as % Actual|
|* Data for 1984 are provided by provinces as the current regional division did not exist.|
Ethiopia’s population was estimated at 42.6 million in 1984 and increased by about 2.3 percent per year to reach 53.4 million in 1994. It again increased by 2.5 percent per year to reach 73.8 million in 2007. When the 1984 figure was released, many in the development community in Ethiopia were alarmed as the population estimate used for all economic and social planning purposes until then was 33 million – about 13 million or about 40 percent less than the census number.
The regional data across censuses are not comparable for reasons mentioned above. Leaving the data for 1984 for the moment, it is observed that between 1994 and 2007 the population of Oromia region increased at an implicit growth rate of 3.6 percent, while the population of all other regions increased by less than the average implicit national growth of 2.5. The slow reported implicit growth of Amhara region and Addis Ababa are particularly notable and are the focus of much debate. Based on these figures, the share of Amhara region declined from about 26 percent in 1994 to 23.7 percent in 2007 that of Oromia region increased from 32 percent in 1994 to 36.6 percent in 2007. Somalia, Tigray and Addis Ababa experienced some decline in their share.
Table 2 below provides the trend in population by the religions practiced in the country. Christians of various denominations accounted for over 60 percent, with their share in 2007 reaching about 63 percent. The share of Orthodox Christians decreased from about 54 percent in 1984 to 43.5 percent in 2007, while the share of Protestant Christians increased from 5.5 percent to 18.5 during the same period. The latter increased primarily at the cost of the Orthodox Christians. The share of Moslems remained about constant at 33—34 percent of the population. A major decline is also noted in the traditional religion and other small sects practiced in the country.
|Table 2: Comparison of the Three Censuses – by Religion|
|Traditional and Others||2527.1||3106.4||2.1||2427.3||-1.9||6.6||5.8||3.3|
Table 3 provides the trend by self-reported ethnic group. The Table itemizes those ethnic groups (seven groups) with a population of 1 million in 1984. In 1984, 95 ethnic groups were identified. Of these 5 were either of “mixed parentage”, “naturalized Ethiopians” or “foreigners”. In 1984, 12.4 million people (or 29.1 percent) reported to be Oromo. This is an underestimate as Borena and Guji, who are essentially Oromo, are reported separately. Even without this correction, Oromos remain the largest Ethnic group in the country. Oromo are closely followed by Amharas, who numbered 12.1 million (28.3 percent of the population).
The 1994 Census identified 93 Ethnic groups. Of these, 8 included a new category for “other Ethiopians” and a more detailed list of the “foreigner’s category” as well as those with “mixed parentage”. Thus the number of ethnic groups adds up to only 85 against 90 in the 1984 Census. Even within the 85, there have been reclassifications that increased the population of some and reduced the population of others. For instance, Borena and Guru are added to Oromo; Garage is divided into Garage and Silted; and Tarawa is dropped altogether.
In the Census of 2007, the number of ethnic groups drops to 90; 8 of which are the same categories mentioned above. This leaves 82 ethnic groups as against 85 in the 1994 census. In 2007, there have been similar re-classifications as in the case of the 1994 census. For instance, Gurage is sub-divided into 3, namely Silte, Sebatbet and Sodo.
In 2007, there were 10 ethnic groups with a population of one million people. Oromo, with a population of 25.4 million, accounted for over a third of the population. The Amhara numbered close to 20 million and accounted for about 27 percent.
It is interesting to note that the number of those who self-report to be of “mixed parentage” and declared “Ethiopian” as their ethnic group is very small. In 1984, 47,797 declared “mixed parentage”, the number declined to 26,797 in 1994 and rose to 36,794 in 2007. The number in 2007 fell far short of the number in 1984. The “Ethiopian” classification came into existence in 1994. That year, 108,208 reported “Ethiopian” as their ethnic group, while the number increased to 145,361 in 2007.
|Table 3: Comparison of Population in the Three Censuses –|
|by Ethnic Group|
|Growth Rate||Percentage Share|
|*Sum of Silte, Sodo Guragi and Sebatbet; **The 1984 Census provides data separately for Tigrawai (4.2 million) and Tigre (0.63 million). The 1994 and 2007 Census report data only for Tigre. *** Figure for 1984 seems to be an undercount.|
To summarize, taking the three censuses together, we learn that the Oromo population account for slightly over a third of the population, while Amharas account for a quarter. As will be shown below these ball mark numbers do not change much if the region or ethnic group or linguistic definition are used. In terms of religion, Christians of various denomination account for over 60 percent, while Muslims account for about a third of the population. The share of Orthodox Christian has declined substantially while that of the Protestant Churches has increased markedly.
- A closer Examination of the Census of 2007
Below we will take a closer look at the Census of 2007 and discuss the data in terms of the regional, ethnic, and linguistic and religious distribution of the population.
Distribution by Region and Ethnic Affiliation
The Pie Chart below shows the inter-regional distribution of the total population in 2007. As mentioned above, the Census of 2007 puts the population of Ethiopia during the year at 73.8 million. Of these, almost 27 million or 36.6 percent lived in Oromia Region. The second largest was Amhara Region. The Region provided residence to 17.2 million people and accounted for 23.4 percent. The third was Southern Nations, Nationalities and peoples (SNNP) Region with a population of 14.9 million or a share of 20.2 percent. Somali Region, with a population of 4.4 million or 6.0 percent of the population, is a distant fourth. Tigray stands fifth and occupies 4.3 million people or 4.9 percent of the population.
In each region the Census questionnaire asked ethnic group, mother tongue, religious affiliation and marital status of the household members. It says, “In the 2007 Census, “Ethnic identity” of respondents was obtained through the question “What is (NAME’S) ethnic group?” Ethnic group (nation/nationality) of a person is traced through his/her national or tribal origin. A detailed list of ethnic groups in the country was obtained from the House of Federation.”
The Regions are multi-ethnic, multi-religious, and multi-lingual entities. To arrive at the national population by ethnic affiliation requires some adjustment to the regional population. For instance, taking the Amhara ethnic group, to arrive at the total population who reported to be Amhara one has to deduct the non-Amhara people resident in the Amhara Region and add the Amhara population living in the Non-Amhara Regions. The computation for the major four ethnic groups is provided below.
Table 4: Distribution of Population by Ethnic Group – Census of 2007
|Amhara Region||Oromo Region||Somali Region||Tigray Region||SNNP Region|
|Residents from other regions||1469.0||3285.2||266.4||149.2||266.4|
|Residents of the region living outside the region||4125.2||1655.0||171.1||220.0||?|
Table 4 requires some explanation. The row referred to as Regional Resident refers to the total number of people residing in the particular Region. Due to a lack of a better term, I use residents from other regions (in-migrants) and residents of the region living outside the region (out-migrants) to refer to those people in a region who report ethnicity other than the one attached to the region. For instance, in the Amhara Region, there were about 1.5 million non-Amharas residents in the region (about 8.5 percent of the Regional population). At the same time however about 4.1 million Amharas resided in the non-Amhara Regions (equivalent to 25 percent of the regional population). The former, I called In-Migrants although the people may have been born and lived in that particular Region for generations. The latter, I called Out-Migrants although they may have lived in the other regions for many years. This adjustment shows that the total number of Amharas (close to 20 million) exceeds the number of Amhras resident in Amhara Region (17.2 million). A similar computation for ethnic Oromos shows that a total of close to 27 million Oromos lived in Oromo Region. An additional close to 1.7 million Oromos lived outside of the Oromo Region (equivalent to 6.3 percent of the Regional population). At the same time, Oromo Region housed close to 3.3 million non-Oromos (equivalent to 12.2 percent of the Regional Population). This puts the total national Oromo population at close to 25.4 million people, lower than the residents of the Oromo Region.
After making the above adjustments, it is clear that Oromos still stand as the largest ethnic group with a share of 34.4 percent of the total population. With a share of 27 percent, Amharas are the second largest. The share of Oromo ethnic group falls a little short (by 2.2 percent) of the share of residents in the Oromo Region, while the share of Amhara rises by 3.6 percent compared to the residents of the Amhara Region. The share of Somalis makes a little gain, while share of Tigrie has a net out-migration of about 70,000 people – about 1.6 percent of the regional population.
Distribution by Mother Tongue
The Census identifies 84 distinct Ethiopian languages. “Mother Tongue” of respondents was identified by asking the question “What is (NAME’S) mother tongue?” Mother tongue is the language used by the respondent for communication with his/her family members or guardians during his/her childhood… A detailed list of the names & codes of the country’s languages were provided to the enumerators.” The results of the Census are provided in Table 5. The table provides the number of people and shares in total population for those languages that have a minimum of half a million speakers in 2007. There are 16 languages spoken by at least half a million people.
|Table 5: Distribution by other Tongue – Census of 2007|
The most widely spoken mother language is Oromigna as 33.8 percent of the population of Ethiopia (or 24.9 million people) report Oromigna as their mother tongue. Since there are only 25.4 million Ethnic Oromos and 24.9 million people whose mother tongue is Oromo, it appears that about half a million Ethnic Oromos have a mother tongue other than Oromigna. The second place is held by people with Amarigna as their mother tongue. Amarigna speakers number 21.6 million and accounted for 29.3 percent of the population. Since there are only 20 million Amharas and 21.6 million people who report Amarigna as their mother tongue, it shows that about 1.6 million non-Amharas start life speaking in Amharigna. With 6.2 percent of the population (or 4.6 million people) Somali speaking people stand third in mother tongue classification. Comparing the population in Tables 5 and 6, there are 300,000 more people whose mother tongue is Somali than there are Somali people. The number of Tigrigna speaking people is at par with the number of Tigrie ethnic group.
It is important to mention that these numbers relate to the mother tongue (home language or 1st language). But people could speak a second language in addition to the mother tongue. The numbers above therefore tend to understate the total number of people who speak a particular language. 
Distribution by Religious Group
Table 6 provides the national and intra-regional and inter-regional distribution of the major religious groups in Ethiopia in 2007.
Table 6: Intra- and Inter-Regional Distribution by Religion
A = Orthodox; B = Protestant; C = Catholic; I = Islam; T = Traditional; O = Others; T = Total
The Table shows that 43.5 percent of the Ethiopian population reported Orthodox Christian as their religious affiliation. Together with followers of the Protestant and Catholic churches, the share of Christians in the total population stands at 62.7 percent. Islam is practiced by 33.9 percent of the population. Of the rest, 2.7 percent are adherents of traditional religions.
In terms of the intra-regional distribution, 44.3 percent of Orthodox Christian are located in Amhara Region; another 25.6 percent are in Oromia. A very large majority of Protestants 60.6 percent of all Protestants in the country live in SNNP and another 35 percent live in Oromia Region. The two regions, taken together, account for 95.6 percent of Protestants. Looking at each region separately from a different angle, Tigray is 96.1 percent Christian – primarily Orthodox Christian. The share of Moslems in the regional population is 4 percent. In Amhara Region, 82.5 percent are Orthodox Christians, while 17.2 are Moslem. There is hardly any non-Orthodox Christian denomination in Amhara and Tigray regions. In Oromia, 47.5 percent of the regional population is Moslem. There are 30.4 percent Orthodox Christians. Including the people that practice Protestant and Catholic faiths, the share of Christians of all denominations rises to 48.6 percent – slightly higher than Moslems. In SNNP Region, with 55.5 percent, the share of Protestants is by far the highest. Afar, Somali, Harari, and Dire Dawa are predominantly Moslem.
- Contentions on Census of 2007
As I mentioned above, the census-taking process and data obtained through them are often contentious. Since the United States started taking census in 1790, the numbers and the approaches have always received partisan reception. Nigeria is another example. Nigeria, just like Ethiopia, is a multi-religious, multi-ethnic and multi-lingual country. Since the first census of the 1950s, Nigeria’s follow-up census or attempted census (four or five of them) have been marred by controversies. In both the US and Nigeria, the census data are used for re-defining the size of electoral districts (thereby Congressional or Parliamentary seats) and federal allocation of funds to the States and Districts, among other uses. As a result, census very much politicized.
In Ethiopia, the size of the regional population has both political and economic ramifications. Politically, at least theoretically, population size is an important variable on the bases of which the number of electoral districts as well as representation in the House of Federation is determined. Economically, census data carries a big weight in resource allocation. For instance, over 60 percent of the weight of the index on which the unconditional block grant destined to regions is computed is based on population size.
Because of these reasons the census of 2007 had been shrouded with serous controversy. At least three contentious areas had emerged. Two of them had been recognized and admitted by the Population Census Commission of Ethiopia, the Agencies that oversaw the administration of the census. First, the Commission reporting to the Ethiopian Parliament admitted that errors were noted in the count of the Affar and Somali Regions that prompted immediate actions. As a result, the population of the two Regions were recounted and the errors corrected. The second was the issue of the missing 2-3 million Amharas. The Commission admits the errors, but no corrective action was taken. It states that the undercount was across the board and that the Commission saw no inconsistencies in the key ratios. Likewise, the undercount of the population of the City of Addis Ababa was recognized but no action was taken.
How do 2-3 million people miss? Since conducting a census is a very involved and expensive feat and population trends do not change over a short time, it is usually undertaken every 10 years. Once the numbers for the Census year are ascertained, a projection is undertaken to determine the likely evolution of the population into the future – usually for each of the next 10 years until the next census is undertaken. As a normal practice, upon the completion of the Census of 1994, the Population Census Commission projected the population of Ethiopia to reach 77.1 million in 2007. For example, the population of Amhara Region, Oromia Region and the City of Addis Ababa were estimated at 19.6 million, 27.3 million, and 3.1 million, respectively. Compared to the actual population of 2007 based on the Census, it appears that the projection figures were invariably higher for all regions. As shown in Table 1, however, the difference is 13.9 percent for Amhara Region (or 2.4 million people), 1.1 percent (0.3 million) for Oromo Region, and 11.7 percent (0.3 million) for City of Addis Ababa. The large margins of error of the Amhara Region and the City of Addis Ababa are the causes for the controversy.
Considering the large differences, no correction was done; neither was a recount undertaken. Indeed, as a recount of the entire population in Amhara Region would have been expensive, the short cut and less expensive corrective measure would have been possible. One proper approach to correct the error would have been to take 10 or 20 percent randomly selected Woredas and perform a recount of their population. This approach would have helped to determine the average size of the under-count and compute a correction factor on the bases of which the population of the other woderas would have been inflated.
The third contention is regarding the population of Oromia Region. Feyisa writing about the Census of 1984 states that “the 1984 census was undertaken with a deliberate objective to ‘prove’ the Oromos are less numerous than Amharas. Enumerators were instructed to register any person who spoke Amharic in Oromia as Amhara.” This assertion is however questionable considering that the Regional leadership was represented in the Regional Census Commission and that local people, including teachers, were deployed as enumerators. Regardless, these assertions make the Census of 2007 equally debatable.
Summary and Conclusion
A closer examination of the census of 2007 reinforces the finding of the previous two censuses. Additionally, a closer look reveals some important information is revealed. It becomes clear that the regions are not homogenous. For instance, about 9 percent of the population of the Amhara region are not Amharas. Likewise, about 12 percent of the population of Oromo region are non-Oromos. On the other side, about 20 percent of Amharas live outside of the Amhara region. Taking the net-migration, there are more Amharas and Tigres living outside of the respective regions. In terms of the languages, there are about half a million Oromos whose mother tongue is other than Oromigna (Oromiffa), whereas about 1.6 million non-Amharas start life with Amharigna as their mother tongue. In terms of religion, Orthodox Christianity is the dominant religion in Tigre and Amhara regions. Most Protestants live in SNNP and Oromia. The number of Christians and Moslems in Oromia is about equal.
Census data are potentially a strong tool to allocate resource in a just and fair manner. Fairness and justice suggests that important resources such as senior Government positions, employment, scholarships and like be distributed in a manner that is proportional to the population. That means about a third of all such resources should go to Oromos, another quarter to Amharas, and down the row in a like manner. Indeed, some would be saved for the very small ethnic groups in terms of number. This kind of principle based allocation will avoid conflict and promote harmony.
Both for political and economic reasons, the technical accuracy of the processes and the credibility of the outcome of census data are important. The Central Statistical Agency is planning to undertake the next census in 2017 . Therefore, the preparatory work should start in advance to avert potential sources of challenge in due course.
Table 6: Intra- and Inter-Regional Distribution by Religion
 Berhanu Abegaz (Ph. D.), Ethiopia: A Model Nation of Minorities, http://www.ethiomedia.com/newpress/census_portrait.pdf, June 1, 2005
 This included the national sample survey in 1964/67 (first round) and 1969/71 (second round), the 1978 manpower and housing survey of major towns, and the rural demographic survey of 1980/81 and 1981/82. Office of the Population and Housing Census Commission, The 1984 Population and Housing Census of Ethiopia, Analytical Report at the National Level, Addis Ababa, December 1991
 Each of the three Censuses has asked questions about the respondents’ ethnic origin, religion, and mother tongue. The 1984 Census Report presents data for those ethnic groups that number over 20,000. There were 59 of them. Of these, Guji and Borena, who happen to be cousins and speak the same Oromo language, are reported separately. If the sum of the two (0.64 million) is added to the reported Oromo population, the sum for Oromos would rise by 5.2 percent from 12.4 million to 13 million. For those whose number falls below the threshold of 20,000, the numbers are added together. The number of ethnic groups in the 1994 census is 71 but after further re-definitions, fragmentations and re-districting, the number increased to 93 in the 2007 Census. For instance, the following 16 ethnic groups appear in the census of 2007 that do not exist in the census of 1994. These are Irob, Kefficho, Karo, Komo, Konso, Koyego, Kusumie, Mashola, Me’enite, Mejengerm Murle, Shekocho, Surma, Upo, and Yem. On the other side, Kimant and Guagu do not appear in the census of 2007 although data is given for 1994. The data for Guragie in 1994 includes Silte, while the two are provided separately in 2007.
 There is no consolidated report showing the characteristics of the population for 1994. Instead, there are upto 3-4 reports for each region addressing different issues from which one has to construct a national aggregate. The alternative is IMIS data file (found here http://imis.csa.gov.et/bineth/RpWebEngine.exe/Portal?BASE=MYAPP3&lang=eng). The IMIS source under-reports the size of Moslem in Somali and Afar Regions. In order to correct for this under reporting, the data in the published Somali and Afar Report is taken.
 As noted above the total population for 1984 was 42.62 million. Since the table relates to the groups whose number exceeds 20,000, there is a difference of 4.41 million between Tables 1 and 2. As a result, the shares and growth rate could change depending which number is used for the total population. In addition, the sum for 1984 does not include the rural population of rural Tigay and Eritrea.
 Feyisa provides an interesting review of the growth of the Oromo people since 1850. He also reconstructs the census of 1984 and puts the Oromo population at 20.9 million as against 12.4 million in the Census. Part of his reconstruction is legitimate as Borena and Guji, who are essentially; Oromo are classified separately in the Census. According to Feyisa about 50 percent of the population in 1984 and 1994 were of Oromo decent. Feyisa Demie, Population Growth and Sustainable Development: The Case Of Oromia in the Horn of Africa, The Journal of Oromo Studies, Vol. 4, Number 1 & 2, July 1997. And also Feyisa Demie, The Oromo Population and the Politics of Numbers in Ethiopia, The Oromo Commentary, Vol. 6, No. 1, 1996.
 Central Statistical Agency, Federal Democratic Republic of Ethiopia, Population Projection of Ethiopia for All Regions, At Wereda Level from 2014 – 2017, August 2013
 Central Statistical Agency, Population and Housing Census Report-Country – 2007
 There are 56 ethnic groups in SNNP Region http://www.snnprs.gov.et/about.html. Going through each of the Regional Tables in the Census and assembling the data is a tedious work.
 Ethnic differences may also be observed from the great variety of languages spoken in the country, of which there are an astonishing eighty-three, with 200 dialects. These can be broken into four main groups: Semitic, Cushitic, Omotic, and Nilo-Saharan., http://www.ethiopia.gov.et/web/Pages/People
 This does not however mean that Oromigna is the most widely spoken language. One would have to add the number of those people who speak Oromigna or Amarigna as a second language to determine which of the two is the most widely spoken language.
 Amy Sullivan, Why the 2010 Census Stirs Up Partisan Politics, February. 15, 2009 http://content.time.com/time/nation/article/0,8599,1879667,00.html
 Helen Chapin Metz, ed., Nigeria: A Country Study, Library of Congress, Federal Research Division, 5th Edition, June 1991;  Sandra Yin, Objections Surface Over Nigerian Census Results, Population Reference Bureau, April 2007, http://www.prb.org/Publications/Articles/2007/ObjectionsOverNigerianCensus.aspx
 Abu Girma Moges, An Economic Analysis of Fiscal Federalism in Ethiopia, paper presented at the Ethiopian American Foundation International Symposium on Contemporary Development Issues in Ethiopia, Addis Ababa, 2003
 Refer to Berhanu Abegaz for a critical evaluation. Berhanu Abegaz , Three Million Amara are Missing: An Analysis based on the 1994 and the 2007 Ethiopian Population Censuses, March 23, 2015, http://ethiomedia.com/101facts/4513
 Feyisa, Ibid.
 Central Statistical Agency, National Statistical Development Strategy, No Date