The lost millions: miscarriage of census 2007

By Tibibir Yilemlim [March 6, 2009]

1. THE CENSUSES OF MAY 1984, OCTOBER 1994 AND MAY 2007

Ethiopia has had three population and housing censuses over the last 25 years: that of May
1984 was during the military regime; it was the least obsessed with divisive issues like ethnicity,
and government leaders were least involved.

Next came that of October 1994 , three years after TPLF guerrilla fighters grabbed power by
force, and took over Minilik Palace, all the time singing about ethnicity and intoxicated with
anti-Amara propaganda. To make matters worse, the census commission was chaired by Meles
Zenawi, the chief architect of ethnicity and divide-and-rule.

The last one came in May 2007, chaired by Deputy Chairman of EPRDF and Chairman of ANDM,
the ruler of Amara region over the last 17 years.

One may, therefore, say that the censuses of 1994 and 2007 were highly politicized, the census
secretariat taking orders from the all-powerful census commission chairmen. An opposition
that is always divided has not been able to do anything about it.

In any country and census, census populations have some common characteristics: almost fixed
sex and age structures, and ethnic compositions that change very little over several peaceful
decades, the length of the period depending on the speed of social and economic development.
If the country develops fast, as in the Tigers of south-east Asia, then some of these
characteristics may change slightly faster.

In Ethiopia’s ethnic regions, migration is effectively prohibited even within the same region; the
issue of in-migration from other lands is unthinkable, except possibly from our neighbors, and
even that is insignificant compared to the sizes of the ethnic groups that may have such in-
coming kith and kin.

If there were significant in-migration or out-migration after 1984, then its effect would be to
change the 1984 ethnic composition. However, Ethiopia has not had any significant in-
migration of Oromos or former residents of Oromia, or Sidamas or former residents of SNNP
from any neighboring states, or from other countries, and there has not been any mass exodus
of Amaras either. Hence, the dramatic increase in Oromos or Sidamas, or the equally dramatic
decrease in Amaras in the 2007 population census cannot be rationally explained..

Another possible explanation for the unusually sharp decline in the percent share of the Amara
is if only Amaras had died in huge numbers between 1994 and 2007, but that has not been
reported through formal or informal channels.

1
A scientific explanation for the vastly different rates of growth reported in the 2007 census is
one where there are significant differences in mortality and fertility, leading to definite
differences in the rates of natural increase; when significant in-migration is added on, the
increase of the in-migrating ethnic group can be larger. However, there is no evidence of such
developments either.

The last possibility for a dramatic change in ethnic composition is if there has been a systematic
process of intentionally miscounting or misreporting the Amara, or otherwise throwing away
census records before and during data processing; this may be the only possible explanation.
However, no amount of record review can undo such damage, and this is what Addisu Legesse’s
Census Commission wants to do to cover up work that is very badly done !

If none of the above scenarios had applied - and they did not - the ethnic composition of 1984
should have remained almost the same after the censuses of 1994 and 2007, and that would
still keep the Oromo as the largest ethnic group with about 29.1% of the population. Amaras
would be the second largest ethnic group with about 28.3% of the population, a difference of
only 0.8%, and not one of 7.6%, as given in the census of May 2007, shown in Table 1 below.

If the ethnic structure had remained about the same as that of 1984, which is normal in
censuses of countries like Ethiopia, the difference in population sizes between Oromo and
Amara should have been about 600,000 in the population census of May 2007, and not over
5,000,000. Likewise, the population of Amara region should have been larger by at least
3,000,000.

However, it is interesting to note that in the census of 1994, the population counts by mother
tongue gave those whose mother tongue as Amharic as the largest, and those with Oromigna
as a mother tongue as the second largest group. Hence, one wonders which one of the two
ethnic groups is the largest.

The Sidama’s percent share should have also remained around 2.96%, the level of 1984, but it,
too, has shot up to 4.02% in 2007, as shown in Table 1 below..

Table 1: Share of selected regions and ethnic groups in

Ethiopia’s census populations

Region/ethni May 1984 October 1994 May 2007 Annual growth

c groups Census census census Rates
(%)
(Millio (%) (Million (%) (million (%) 1984- 1994-
ns) s) s) 2007 2007
Ethiopia 42.62 100. 53.13 100. 73.92 100.0 2.394 2.64
0 0
REGIONS
Amara … …. 13.83 26.0 17.21 23.3 … 1.75

2
Oromia … … 18.47 34.8 27.16 36.7 … 3.08
SNNP … … 10.37 19.5 15.04 20.3 … 2.97
Tigray … … 3.14 5.9 4.31 5.8 … 2.53
Somali … … 3.14 5.9 4.44 6.0 … 2.77
Affar … … 1.05 2.0 1.41 1.9 … 2.36
Benishagul … … 0.46 0.9 0.67 0.9 … 3.01
Gambela … … 0.16 0.3 0.31 0.4 … 5.29
Harari … … 0.13 0.2 0.18 0.2 … 2.60

Amara 12.055 28.2 16.01 30.1 19.87 26.89 1.75 2.173

9 3
Oromo 12.388 29.0 17.08 32.1 25.49 34.48 3.08 3.139
7 5
Kembata 0.433 1.02 0.500 0.94 0.63 0.85 1.85 1.630
Haditya 0.644 1.51 0.93 1.75 1.28 1.73 2.99 2.986
Sidama 1.262 2.96 1.84 3.47 2.97 4.02 3.03 3.721
Welayita 1.093 2.56 1.27 2.39 1.71 2.31 1.22 1.954
Source: Population and Housing Reports of 1984, 1994 and 2007.

A brief look at the data for the Amara, Oromo and Sidama, and therespective regions in Table 1
shows the following strange features for the period 1994-2007:

The percent share of the Amara in the census population has gone down by 1.4%
from 28.3% in 1984 to 26.9% in 2007, whereas that for Oromo has gone up by 5.4%
from 29.1% in 1984 to 34.5% in 2007. Where did the additional 4,000,000 Oromos
come from in the census count of May 2007, and where did the Amara go ?
Likewise, the Sidama have consistently made huge jumps in 1994 and 2007.

Why do we have so much percent change in the ethnic structure of a country that
has been peaceful, does not encourage free movement from one ethnic region to
another, does not have significant in-migration of kith and kin from abroad, and has
no fertility and mortality experiences that vary significantly over different ethnic
groups or regions ?

The average annual population growth rate for the period 1994-2007 in Amara
region was reported as 1.75%, which is much lower than the national average, but
that for Oromia was 3.08%. The demographic, social and economic experience over
the period 1984 to 2007 indicates that the population growth rate of Oromia cannot
be larger than that for the Amara region, but the census counts appear to say that
Oromia grew almost twice as fast. This does not make sense.

SNNP grew annually at 2.97% , which is not too far from that of Oromia, and the
determinants for population growth in Ethiopia during 1994-2007 cannot justify

3
 such growth rates; this growth presumably happened in the Sidama group, but not
in the Kembata, Welayita and Hadiya ethnic groups, and that is also incredible.

After the census of 1994, the difference between the percent shares of the Amara
region and that of Oromia in the census populations was 8.8%, but that difference
jumped to 13.4% after the May 2007 census. How and why did the Amara region
contract so drastically, and how did Oromia explode so incredibly ?

TPLF/EPRDF came into Addis Ababa loaded with poisonous ethnic propaganda and designs: it
oversaw the redrawing of the political map of the regions, oversaw the dismemberment of
Ethiopia, acted it out during Amara mass murders at Arbagugu, Wotter, Eastern Wellega and
Assebot, supervised the delivery of a nation of 83 ethnic groups to the mercy of only 9 of them,
giving each absolute power over farmland, and even life and death ; it closely controlled the
censuses of 1994 and 2007, possibly “doctoring” census reports to give the abnormal ethnic
compositions in 1994 and 2007.

There is quite a lot of evidence on the Amara and the Oromia regions to show that the huge
growth rate for Oromia and the huge decline in Amara region are pure fabrications; they are
not supported by any mortality, fertility and migration experiences of both regions over the last
two and one-half decades.

Table 1 includes all regions and only 6 ethnic groups: Kembata, Hadiya, Sidama, Welayita,
Oromo and Amara. Tthe six ethnic groups accounted for 65% of the population of Ethiopia in
the 1984. census, but their collective share in the 2007 census population was over 70%, which
cannot be explained since it has gone up by some 5%.

Under normal conditions when the statistics are clean, population structure remains almost
constant, and that is borne out for Kembata, Hadiya, and Welayita in Table 1. Apparently, this
also suggests that the total census population count in May 2007 is, perhaps, not too far from
74 million.

Among other regions, Tigray had an artificial growth in 1994, compared to 1984, by annexing
sizable Amara populations in Alamata and Korem in North Wollo, and Simen in north-east
Begemidir and Simen. However, it did stabilize in 2007 since it had stopped grabbing new
fertile lands from neighboring regions. .

The erratic composition for Somali region is understandable; Afar, Benishangul, Gambela and
Harari have percent shares that are about the same in 1984, 1994 and 2007, as expected, in
spite of erratic developments in Amara, Oromo, Sidama, Amara region, SNNP and Oromia.

2. The Picture in Urban Ethiopia

Another problem area is urban Ethiopia where Addis Ababa is reported to have 2.74 million
inhabitants in May 2007. Table 2 below shows the reported growth of our urban population:

4
 Table 2. Growth of Urban Population, 1984-2007

May October May Growth Growth

1984 1994 2007 rate per rate per
Population year (%) year (%)
Sector (millions) (millions) (millions) 1984-2007 1994-2007

Ethiopia 4.87 7.32 11.96 3.90641 3.63672

Urban Addis
Ababa 1.42 2.11 2.74 2.85783 1.93533

Ethiopia less 3.45 5.21 9.22 4.27392 4.22811

Addis Ababa
Source: Population and Housing Reports of 1984, 1994 and 2007.

The increased urbanization of Ethiopia explains the increased growth in 1984-2007 and 1994-
2007. Table 2 also shows that Addis Ababa grew annually at 1.94 % between the last two
censuses, whereas other urban populations in Ethiopia grew annually at 4.23%. Using these
latter growth rates which are more compatible with the actual growth of Addis Ababa, the
population of Addis Ababa may have been under-reported by about 1,000,000.

Addis Ababa’s share of the urban census population went down by 5.5% in May 2007; its share
of the total population also went down. Both declines are inconceivable for Addis Ababa. A
possible explanation will be offered latter, but first to the fertility and mortality history of
Ethiopia since these are the principal determinants for population growth..

3. THE FERTILITY AND MORTALITY BACKGROUND

The fertility and mortality history over the last 15 years comes from three official sources, as
shown in Table 3 below: two of them are demographic and health surveys (DHSs) carried out by
the Central Statistical Authority and an American survey institution which carries out such
surveys all over the world; the other is the 1994 population and housing census. All three
official reports give information on the level of fertility and mortality for the various ethnic
regions of Ethiopia, as shown in Table 3 hereunder.

TABLE 3. ETHIOPIA’S FERTILITY AND MORTALITY EXPERIENCE

5
 1994
REGION POPULATION DHS 2000 DHS 2005
CENSUS
1Q 0 e0 1Q 0 TFR BIRTH RATE e0
(female)

Affar 118 50.3 129.2 4.9 37.3 50.8

Amara 116 50.8 112.4 5.9 37.3 56.0
Oromia 118 50.4 116.2 6.4 37.3 55.5
SNNP 128 48.6 113.4 5.9 37.3 53.5
Tigray 123 49.6 103.6 5.8 37.3 54.9
Somali 96 54.8 99.4 5.7 37.3 55.4
Benishagul 139 46.8 97.6 5.4 37.3 51.1
Gambela 99 54.2 122.6 4.5 35.7 58.5
Harari 113 51.4 118.3 4.4 35.7 54.7
Source: Population and Housing Reports of 1994,DHS 2000 and DHS 2005.

e 0 = the number of years a new-born may expect to live if the intensity of mortality
at all ages remained unchanged throughout the life of the new-born.
1Q0 = the number of deaths of children under one year old in a year per 1000 live
births during the same year.

The 1984 census report (See :The 1984 Population and Housing Census of Ethiopia: Analytical
report at national level, pp.238- 241), which was relatively free from the obsession of
ethnicity, gave the general pattern for the level of mortality as shown in Table 4 below:

Table 4: State of the Rate of Natural Increase

Region Level of Fertility Level of mortality Probable level of

population growth
Arssi High High Low
Bale High High Low
Illubabor Average Average Low
Wellega Low Low Low
Keffa High High Low
Harerghe High High Low

Gojjam High average Medium

Gonder Average Low Medium
Wello Low Average Low
Shewa Average Average Low

Gamu Goffa High Average Medium

Sidamo Average Low Medium

6
A high birth rate and a low mortality rate give rise to a high rate of population growth, and a
low fertility rate with a high morality rate will give a low rate of population growth, if migration
is negligible in both cases. Hence, from Table 4, it is not possible to conclude that today’s
Oromia or SNNP will have the highest population growths, as suggested to us by the
population census of May 2007.

In the 1994 census, fertility and mortality were not accurately measurable, as they have always
been, and the best that the Central Statistics Authority (CSA)could do to prepare population
projections for 1995-2000 was to use a simple approach.

It assumed age and sex structure to be fixed for all Ethiopia, and it also assumed that urban
Ethiopia will grow at 4.11% annually whereas rural Ethiopia would grow at 2.23 % annually;
these were the growth rates used to project populations for 1995-2000 in Volume I, Part V of
the Population and Housing Census abridged statistical reports for each of the 9 regions.
These assumptions had to be made after the census of October 1994 since CSA had no credible
evidence that showed that one region of Ethiopia was growing faster than another.
The only additional evidence on mortality and fertility that came up after the 1994 census was
that in the Demographic and Health Survey (DHS) 2000 and DHS 2005, and the relevant data
are given above in Table 3.

DHS 2005 is quoted by the Ministry of Health in its 2006/2007 annual report, entitled Health
and Health Indicator, on Page 6, as follows: Fertility is given as 37.3 births per 1000 in Tigray,
Afar, Amara, Oromia, Somali, Benishagul-Gumuz, SNNP, and Gambela, and 35.7 in Harari, Dire
Dawa and Addis Ababa. Life expectancy, which is a summary of mortality experience of all ages
of the entire population, is given as 53.4 years for Amara, 53.0 for Oromia, 51.4 for SNNP and
53.4 for Ethiopia, all for males; that for females was 56.0 for Amara, 55.5 for Oromia, 53.5 for
SNNP and 55.4 for Ethiopia.

Whereas fertility was the same for Amara and Oromo, there was some variation in mortality:
Amara men and women presumably lived longer than any of those in Ormia and SNNP; a larger
life expectancy meant a lower overall mortality level; this same picture was also given in the
1994 census analysis where life expectancy in the Amara region was again higher than those for
Oromia or SNNP.

Hence, on the basis of all available official information, it is possible to conclude that Amara
region cannot possibly have a lower population growth rate than Oromia or SNNP, and that is
also supported by the earlier data in Table 4.

However, the 2007 census gives an average annual growth rate of 1.75% for Amara region ,
3.08% for Oromia and 2.97% for SNNP; the average annual growth rate for the Oromo is given
as 3.21% and that for Amara as only 1.73% , both growth rates being as incredible as those for
the two regions. None of these growth rates is supported by any official data in Ethiopia’s
demographic history.

7
Amaras, the Amara region and Addis Ababa have been given highly reduced population sizes
that defy information and logic. Perhaps, the explanation may lie in EPRDF’s perception of
these people and regions after May 2005, and EPRDF’s possible chances in the elections of May
2010..

4. WHY DID THE CENSUS OF MAY 2007 GO WRONG AGAIN?

The continuing decline of the Amara, the populations of Addis Ababa and the Amara region
during the last two EPRDF-controlled censuses have engaged a variety of professionals at home
and abroad. Some uninformed observers have gone as far as saying that the Amara living in
non-Amara ethnic regions may have chosen to report themselves as either Oromo or Sidama or
any other ethnic group in October 1994 and May 2007.

On account of the immense political pressure exerted on the Amaras by the Federal
Government and local ethnic governments, this is very likely in order to have access to
education, health services, farmland and other basic government services in their areas of
residence. However, the number is too small to explain the huge decline in Amaras.

According to the equally questionable 1994 population census, the Amara population residing
in regions other than Amara region and Addis Ababa, where ethnicity has not been a serious
problem, was 2,373,000. Given the fact that migration from one ethnic region to another is
forbidden by our ethnic-based laws, if only that Amara population is allowed to grow annually
at the rate of 3.14%, which is as large as the fastest growing large ethnic group reported in the
May 2007 census, then, using the 1994 census as a baseline, the maximum possible Amara
population in all other ethnic regions would be only 3,602,000.

The population census of May 2007 has reported that there is an Amra population of 2,835,000
in the various other ethnic regions, excluding Amara region and Addis Ababa. Hence, the
maximum possible number of Amaras that may have reported their ethnicity as anything but
Amara is, at most, 3,602,000 – 2,835,000 = 767,000 only. However, the missing Amara
population, in relation to other major ethnic groups, is well over 5,000,000. Hence, the
argument that Amaras may have chosen to report themselves as anything else but Amara in
non-Amara regions cannot explain the abnormal and unique decline in the Amara population.

Furthermore, the argument of ethnic change due to local political pressure fails totally when we
look at the huge decline of over 3,000,000 in the Amara region and another of over 1,000,000
in Addis Ababa where over 47% of that missing population is estimated to be Amara.

Contrary to the provision of the Constitution, when Prime Miniser Meles announced the results
of the 1994 census in the media, his first remark was that he was surprised that the Amara
came to be so large. Evidently, that was a problem that he could not handle very well; he had
obviously ignored the 1984 census.

The current TPLF/EPRDF leaders are particularly strongly anti-Amara, and EPRDF has been doing
everything to weaken or destroy the Amara: Regional boundaries have been redrawn, and the

8
only loser has been the Amara region which has lost fertile parts of north Wollo and north-east
Begemidir and Simen to Tigray, west Gojjam and west Begemidir and Simen to Benishangul and
to Sudan, and large chunks of northern shoa to Oromia, thereby forcing Amaras to go different
ways to ensure survival in such a hostile environment.

Most Amaras, like most Ethiopians, have always opposed ethnicity since allowing 9 ethnic
groups to impose their will on 74 others is totally unjust and undemocratic. This is good
enough for EPRDF to ensure that Addis Ababa, Amaras and Amara region are weakened well
before 2010 by reducing their votes, and hence their political representation in the Ethiopian
Parliament in 2010.

The elections in May 2005 clearly demonstrated that Ethiopians would have dump ed EPRDF if
they could. Meles also knows that the youth that flooded Meskel Square and all adjoining
areas and streets on May 8, 2005, singing “Tilant lehodachin, zarie lagerachin” , may do
exactly the same in May 2010 to achieve a democratic Ethiopia and a brighter future for itself.

Hence, EPRDF is busy making preparations for May 2010: Agreements for negotiations reached
with imprisoned leaders of Kinijit through the elders have been ignored; census counts are
being manipulated to suit EPRDF’s designs; no external observers are allowed in the revised
election law; opposition leaders and members of the free press and independent artists are
being thrown into jail.

EPRDF is making sure that no strong civil society is allowed by the new law; the youth, women,
workers and farmers are all being urgently organized for EPRDF; no independent courts, no
neutral election board, no independent public mass media, no strong private press, no neutral
police and local militia so far; lobbyists are also busy in Europe and America to seek support for
another stolen election in 2010. It is all business as usual!

Do we need to go to Mandela to tell us how his successful coalition of 26 multi-colored

opposition parties (democrats, racists, communists, socialist, regionalists, nationalists, etc.) won
the elections in April 1994 ? Their trick was to get together during 1991-1994, and prepare
meticulously to ensure free and fair elections. They succeeded without any need for any form
of transition.

We can win in May 2010 if (i) we come together now, and ensure that the prerequisites for
free and fair elections are in place now, (ii) create a strong coalition that will avoid dividing the
votes of supporters on voting day in 2010, and (iii) ensure that all ballots are correctly counted
and reported, thereby avoiding the deadly errors of the opposition in May 2005.

9
Appendix: Census Counts

AMARA ONLY ALL ETHNIC GROUPS

Region/Kilil Oct. 1994 May 2007 Oct. 1994 May 2007 Growth rate
(%)
Amara 12,615,160 15,747,800 13,834,000 1`7,214,000 1.696
Oromia 1,684,128 1,961,277 18,474,000 27,158,000 3.008
SNNP 312,558 420,243 10,371,000 15,042,000 2.902
Tigray 81,297 70,334
Somali 20,951 29486
Afar 46,534 72,507
Benishangul 102,061 142,557
Harari 42,781 41,755
Gambela 12,566 25,856
Addis Ababa 1,019,729 1,288,306 2,113,000 2,738,000 2.0132
Dire Dawa 69,798 68,887
Amara in Ethiopia by urbanization
Rural 12,903,000 15,483,000
Urban 3,105,000 4,388,000
Total 16,008,000 19,871,000
Ethiopian Population by urbanization
Rural 45,816,000 61,962.000 2.349
Urban 7,316,000 11,956,000 3.8505
Total 53,132,000 73,918,000 2.8755

10