UNDP SOMALIA

UNITED NATIONS DEVELOPMENT PROGRAMME

SOMALIA Somalia

UNDOS

Population Statistics of Somalia

1997
This report was prepared by K. E. Vaidyanathan, the UNFPA Consultant.
Comments and suggestions may be forwarded to KNS Nair, at
[email protected]
 REPORT OF THE UNFPA CONSULTANT ON
POPULATION STATISTICS OF SOMALIA
TABLE OF CONTENTS
INTRODUCTION 1
TERMS OF REFERENCE OF THE CONSULTANT 2
ACTIVITIES OF THE CONSULTANT 3

REVIEW OF UNDOS METHODOLOGY OF ESTIMATING POPULATION 4

REVIEW OF THE METHODOLOGY OF TIMOTHY FOWLER 7

REVISED ESTIMATES OF POPULATION OF SOMALIA 8

• Population estimates of Somalia by sex 1995 – 2015 (Table 1) 10
• Population estimates of Somalia by region 1995 – 2015 (Table 2) 11-13

DEMOGRAPHIC DATA COLLECTION THROUGH UNDOS SURVEYS 14

PROPOSAL FOR A SENTINEL REPORTING SYSTEM 15
RECOMMENDATIONS FOR FUTURE ACTION 15
ACKNOWLEDGEMENTS 17

ANNEX 1: ESTIMATION OF POPULATION OF SOMALIA BY REGIONS, SEX & AGE GROUPS, 1995 – 2015

• Introduction 18
• Sources of Data and their Limitations 18
• Population Size of Somalia 19
• Review of UNDOS Methodology of estimating Population 21
• Review of the Methodology of Timothy Fowler – Fertility, Mortality & Migration 23-26
• Estimation of Population by Region 26
• Assumptions regarding Fertility, Mortality & Migration 26-28
• Methodology of estimation up to year 2015 28
• Results of the Projections 28
• Limitations of the Estimates & Projections 29
• Population estimates of Somalia by sex 1995 - 2015 (Table 1) 30
• Population estimates of Somalia by region 1995 - 2015 (Table 2) 31-33

ANNEX 2: DESIGN OF A DEMOGRAPHIC SURVEY OF SOMALIA

• Introduction 34
• Sample Design 34
• Questionnaire: 36
• Relationship 36
• Marital Status 36
• Place of Birth & Place of Previous Residence 37
• Education 37
• Economic characteristics 37
• Paternal/Maternal Orphanhood 38
• Age at first Marriage/Duration of Marriage 38
• Children ever born & Children surviving 38
• Date of Birth of Most Recent Livebirth 39
• Part 2, 3 & 4 of the questionnaire 39-40
• Recommended Tabulations 40
• Concluding Remarks 41
• Recommended analysis regarding Fertility, Mortality & Migration 42-43
• Demographic and Social Characteristics 44
 2

ANNEX 3: PROPOSAL FOR A SENTINEL REPORTING SYSTEM FOR SOMALIA

• Introduction 45
• Objectives of a Sentinel Reporting System 46
• Method of Data Collection 47
• Household Register 48
• Birth Report 48
• Death Report 49
• Processing of Data 49
• Output of the Sentinel Reporting System 49
• Future Development of the System 50

LIST OF PERSONS MET DURING THE MISSION 52

 3

REPORT OF THE UNFPA CONSULTANT ON

POPULATION STATISTICS OF SOMALIA
15 August - 8 Oct 1997
K.E.Vaidyanathan
UNFPA Consultant at UNDOS, Nairobi

INTRODUCTION
Somalia, situated in the Horn of Africa has undergone a major civil war during 1990-92 leading
to the fall of the government of Siad Barre, and the subsequent balkanization of the country.
Although all Somalis belong to one religion and speak the same Somali language, the Somali
society is divided into numerous clans and sub-clans and each sub-clan is sensitive about its
economic interests and this is compounded by the vested interests of the clan/sub-clan leaders.
The situation is even now unstable with frequent inter-clan fighting erupting in one area or other
for the flimsiest reason. This situation has resulted in far-reaching demographic changes
including refugee migration to neighbouring countries, considerable internal displacement of
population, and presumably a fall in fertility and an increase in mortality as a result of the
difficult economic situation. There has been no Central administration since the collapse of the
Siad Barre government in 1991.Consequently, there is no statistical machinery to collect and
compile data on a systematic basis, so as to be useful to local administration, NGOs and
international agencies.
The United Nations Development Office for Somalia (UNDOS), a project of UNDP and
executed by the Office of Project Services (OPS) of UNDP has attempted to collect and compile
data on key areas such as number of settlements, population, household income and
expenditure, prices of essential commodities, agriculture, education and health in areas that
could be reached. In addition several international NGOs engaged in humanitarian activities
collect data pertaining to the areas of their interest, particularly education and health. UNDOS
is making efforts to compile, evaluate and analyze these data to provide a picture of the
demographic, economic and social situation in the different regions of the country. Under the
auspices of UNDOS a Statistical Working Group has been established to harmonize the
concepts and procedures of data collection, and to establish a data base on the current economic,
social and demographic situation and trends in Somalia. UNDOS has prepared the statistical
profiles of five regions (Bay, Middle Shabelle, Bari, Nugal and Mudug). Using the settlement
statistics on the number of houses and the estimates of average population per house obtained
from the family budget survey, UNDOS has prepared population estimates for these regions.
These activities are extremely important in the absence of any official machinery for data
collection and dissemination.
Demographic data are scanty and incomplete for Somalia, although two censuses have been
carried out during the prewar period. The first census carried out in 1975 was not published,
and only an analytical report based on the census results was brought out in 1984. A national
demographic survey was carried out in 1980-81, but the data were not processed, barring a few
hand-tabulations. Another census was carried out in 1985-86, and once again the census got
bogged down in doubts about its accuracy, and was not published. There are varying accounts
of the census count of population, one giving a figure of 5.8 million and another 8.5 million.
The only published set of demographic data comes from the 1980-81 POPLAB survey of three
regions – Banadir, Bay and Lower Shabelle – carried out in collaboration with the University of
North Carolina.
Data deficiencies are particularly severe for the nomadic population, who constitutes about half
the population of Somalia. Data on fertility and mortality were collected in the 1975 census but
for only the sedentary population. As a consequence of the civil war during 1990-92, there has
been great disruption of population through out the country. It is believed that about a million
 4

people left the country during the war, and about 750000 are said to have perished during the
war. There are no independent sources of information to confirm these estimates. Even today
there are close to half million internally displaced persons (IDPs) inside Somalia, and about half
million refugees outside the country, principally in Kenya and Ethiopia but also in far off places
like Canada, Europe and Australia. Other human costs of the civil war have been equally
enormous. There are many abandoned women, and there are large numbers of men and women
whose marriages have been disrupted. Apart from the deaths caused by the war, there has been
an increase in mortality as a result of the economic crisis and malnutrition and disruption of
health services. Unfortunately information is lacking to document these human costs.
Following a mission by Mrs. Mehri Hekmati, Director of Arab States and Europe Division in
UNFPA and a request by UNDP, Somalia, UNFPA decided to field this consultancy mission to
Nairobi to review the estimates of population for Somalia and to make recommendations for the
design and implementation of a system of collection of demographic and socio-economic data in
a phased manner.

TERMS OF REFERENCE OF THE CONSULTANT

The terms of reference of the Consultant were as follows:
1. Review the methodology used by the US Bureau of Census Demographer to reconcile
the population and other demographic statistics compiled by the different agencies and
produce a time-series data that are acceptable to both UN Organizations and donors.
2. Develop the tools and procedures to conduct censuses/surveys with appropriate
periodicity to estimate at a fairly desegregate level:
(i) total population;
(ii) sex-wise distribution;
(iii) age-wise distribution;
(iv) Spatial distribution (urban, rural and nomadic); and
(vi) reproductive health statistics.
3. Design appropriate surveys to collect and compile data and to derive estimates of vital
statistics;
4. Develop methodologies for estimation of emigration, immigration and internally
displaced population.

ACTIVITIES OF THE CONSULTANT

The Consultant arrived in Nairobi on 15 August 1997, and he was briefed by Dr. K. N. S. Nair,
Officer-in-charge of UNDOS and Head of the Economic Planning Unit about the activities of
UNDOS, and the settlement and other surveys carried out by UNDOS in five regions of Somalia
(Bay, Middle Shabelle, Bari, Mudug and Nugal). He also informed the Consultant of the future
plans of UNDOS to carry out similar surveys in a phased manner. The Consultant was provided
the records of the UNDOS surveys and also introduced to the professionals engaged in the
analysis of UNDOS survey data.
Following this briefing, the Consultant met with the professionals of international agencies and
NGOs engaged in related areas of health, education and humanitarian assistance. The objectives
of these consultations were: (i) to ascertain what demographic, reproductive health and other
data are required by them; (ii) what demographic and other related data are compiled by them
during the course of their work; and (iii) to what extent they could assist in the collection and
compilation of such data.
 5

He also met with the Statistical Working Group and the working group on Health Information
System and briefed them on the aims of the Consultant's mission and how different agencies
could harmonize their data collection efforts in a way that could add to our understanding of the
Somalia situation. He discussed with them the proposed demographic survey and the sentinel
reporting system for reproductive health and vital statistics, and elicited their views on these. A
list of persons met during the mission is attached herewith.
The Consultant undertook a review of the available demographic data. This was far from easy
since the UNDOS Documentation Unit did not have the pre-war census and survey publications
of Somalia, since most of these were lost during the civil war prior to the establishment of
UNDOS. None of these publications were available in the UNEP, UNDP, UNESCO or World
Bank libraries. Even the most basic information such as sex-age distribution of population of
Somalia by region from the 1975 Census and the 1980-81 and 1990 surveys and the estimates of
fertility and mortality derived therefrom were not available. Considerable time was spent in
writing to different international agencies for obtaining such information.
There are a number of questions on which there are no clear-cut answers. For example, how
many Somali refugees and emigrants have returned home after the civil war, and what is their
sex-age composition? Are there substantial differences between regions in their fertility and
mortality? With one-half of the population nomadic it is important to know their fertility and
mortality and how they differ from those of the settled population. Such statistics are not
available. There has been a tremendous exchange of population between different regions due
to inter-clan quarrels, resulting in internal migration between regions, however no information is
available regarding these movements. Because of these constraints the Consultant had to make
his estimates on the basis of the limited information available concerning several population
parameters.
In spite of these shortcomings the Consultant accomplished all the tasks assigned in the TOR.
The available estimates of population were reviewed and a report on the demographic situation
in Somalia has been prepared. In addition, projections of population by sex and age have been
prepared according to region. The Consultant has provided a design for a demographic survey
to be carried out in a phased manner. This survey is intended to yield estimates of fertility and
mortality at the level of the regions and for the urban, rural settled and nomadic populations,
and to provide key reproductive health statistics. He has reviewed the procedures and
questionnaires of UNDOS for collection of socioeconomic data and recommended
improvements. The Consultant has proposed a Sentinel Reporting System for Vital statistics
which could be implemented in a phased manner through the participation of NGOs. Last but
not the least important, the Consultant has contributed to the training of Somali professionals in
survey methodology through elaborate discussions on the "know-how" and "know-why" of the
procedures.
REVIEW OF UNDOS METHODOLOGY OF ESTIMATING POPULATION
Since its inception in 1994, UNDOS has been playing a key role in the collection and
compilation of demographic and socioeconomic data required for decision making in a number
of areas such as population, health, education and socioeconomic data required for humanitarian
assistance and rehabilitation. These data are being used by the international agencies,
international and Somali NGOs and even the local authorities where they are functioning.
UNDOS has developed a unique methodology for estimating the population at the micro level
through settlement surveys and these data are aggregated to provide estimates at the district,
regional and national levels. A settlement consists of a main village and a network of satellite
villages surrounding the main village. The typical main village had a market, school and
mosque. In the livestock producing areas, most of the beels had only the main village and no
satellite villages.
Under the UNDOS procedure, the supervisors and enumerators contacted the village elders and
other knowledgeable persons to gather information on the number of houses, population and
socioeconomic characteristics of the main village and four large satellite villages utilizing the
 6

settlement questionnaire. While this is going on, one of the enumerators goes round the villages
and makes a physical count of the number of houses in these villages. The main village and
satellite villages are then plotted on a map utilizing the geographical positioning system (GPS),
which provided the coordinates (longitudes and latitudes) for these villages. In some of these
villages a family budget survey is carried out in a sample of households, and the average
population per house is calculated from the family budget survey data. UNDOS estimated the
population by multiplying the number of houses in each village (from the actual count) by the
average population per house derived from the family budget survey. For the smaller satellite
villages, the number of houses was based on actual count or through information gathered from
village elders, and population of these villages was estimated by multiplying the number of
houses by the average population per house. This methodology has the merit of simplicity, and
is appropriate for situations like that of Somalia, where carrying out a nationwide census is not
feasible because of the security situation. Another advantage of this methodology is its
flexibility, since unlike in a census we do not have to cover all the areas on a fixed reference
date, and savings are effected on cost and time through the combined use of the GPS and
sampling.
The population estimates based on this method for the five regions appear realistic from the time
series data given below:

Population Estimates

Regions 1975 (Census) 1986 (Census) 1996 (UNDOS)

Bari 194,368 222,287 229,000

Nugal 64,790 74,774 114,000

Mudug 269,464 311,230 324,000

Middle Shabelle 309,221 352,040 419,000

There are however two possible sources of underestimation. First, there is a universal tendency
in censuses and surveys to miss children, and this could have happened in the family budget
survey which provided the estimate of average population per house. Evidence of this is found
in the percentage of population in the childhood age groups:

Age group Bari Nugal Mudug M. Shabelle Combined

M F M F M F M F M F
Under 1 1.07 1.48 1.45 1.19 0.73 0.72 0.29 0.53 0.84 0.99
1-4 12.60 13.21 13.12 11.37 11.89 11.55 12.61 13.53 12.53 12.49
5-9 18.05 15.58 19.28 16.19 19.25 15.87 18.67 19.27 18.74 16.73

Under stable conditions one would expect the percentages in the three age groups to be 4, 13
and 15 percent respectively. The percentages of population less than one year is obviously low,
indicating a serious omission of children or a fall in fertility. The children under one year of age
are survivors of those born one year earlier, namely 1995, whereas those in the age group 1-4
are those born during 1991-94 which include the years of turmoil. Therefore one would expect
the percentages to be much lower than expected in the 1-4 age group, and not in the under 1 age
group. This would suggest that there has been an under count in the under one age group. On
the assumption that the 1-4 age group is correct, we may apply a survival factor to obtain an
estimate of the population under 1. This would indicate that the population under 1 may be an
underestimated by 76 percent, but this will result in an undercount of only 3 percent of the total
population.
 7

Another puzzling feature of the age data for the four regions is the shortfall in the number of
females below 20 and above 40, while in the ages 20-40, the number of women consistently
exceed the number of men. The deficit of men in the age range 20-40 can be explained by
selective emigration of men. The 1975 census and 1980 survey also have more males than
females in the younger ages, but females exceed males in the ages above 40. There is therefore a
suspicion that more females than males may have been omitted in the ages above 40 or fewer
women are surviving to older ages. The proportion of population above 60 years of age in the
1975 population census was 5.29 percent-5.20 percent for males and 5.37 percent for females.
The settlement survey of UNDOS gives only 3.86 percent in this age group- 5.12 percent for
males and 2.48 percent for females. To some extent shortfalls can occur due to high mortality at
older ages, but the difference observed for females is too large to be accounted by this factor
alone. If we assume that the sex ratios found in the 1975 census for ages beyond 40 is
acceptable, this would indicate a 34 percent under-enumeration of women beyond 40 years of
age. In terms of total population, this will amount to an under count of 3.1 percent. These two
factors, namely omission of children under 1 and the women above 40 years of age will thus
account for an undercount of 6 percent of the population. Even an undercount of this order is a
remarkable achievement under the circumstances prevalent in Somalia.
There is however scope for improvement of the UNDOS Settlement Survey. It appears that the
UNDOS survey was designed at a time when there was not much appreciation of the kind of
data collected. Therefore, the questions on the demographic part of the UNDOS survey do not
go beyond sex, age, marital status, settlement pattern and out-migration. However, to estimate
demographic parameters and vital statistics the scope of the survey and design of the schedules
need to be broadened by adding a module on reproductive health statistics. It is important that
this task is entrusted to UNDOS as they are the only agency involved in conducting
comprehensive surveys in Somalia on a regional basis and also have the capacity to undertake
the analysis and dissemination of the data. The capabilities of UNDOS can be further enhanced
through technical consultancies in the relevant areas.
The scope of the UNDOS survey could be expanded by including standard questions that are
generally asked in demographic surveys which would enable the estimation of demographic
parameters, such as fertility, mortality and migration. One of the improvements proposed is the
inclusion of such questions in future surveys. Instead of tagging on a few demographic
questions to the family budget survey, it is recommended that a Demographic Survey
questionnaire is added to the Settlement Survey, and this could include a module on
reproductive statistics. The methodology is described in a separate note, and need not be
elaborated here. It should however be pointed out UNDOS is greatly benefiting the international
community, the NGOs and local authorities by such surveys. If UNDOS had not been collecting
such data and providing the data base for the decision makers, either the decision making will
be done without such data, or the international agencies and NGOs will be spending
considerable time, funds and efforts for collecting such data, an effort more urgently required
for humanitarian relief. Moreover if peace returns to Somalia, and a government is established,
the administration will continue to depend upon UNDOS data base for the information they
require for decision making. Therefore in the foreseeable there is no alternative but the
continuation of the data collection by UNDOS through their settlement and socioeconomic
surveys. However, UNDOS will require assistance from UNFPA in the form of funds and
technical inputs, for carrying out the Demographic Survey and the Module or Reproductive
statistics mentioned earlier.
 8

REVIEW OF THE METHODOLOGY OF TIMOTHY FOWLER

In March 1997, USAID Somalia Office mounted a mission by Timothy Fowler of US Bureau of
Census to examine the available data for Somalia and make population projections at the
national level. Fowler adopted the cohort-component method of projections using the RUP
software developed by the US Bureau of Census. He started with the base population from the
1975 census, and based on the available information on fertility, mortality and migration
incorporated assumptions regarding these parameters for successive years. His assumptions are
briefly summarized below:
Fertility
The total fertility rate (TFR) is assumed to be constant at 7.25 from 1975 up to 1990, declined to
5.5 in 1992 due to famine and civil unrest and return to the 1990 levels by 1999.It is further
assumed that the TFR will decline by 10 percent by 2009, and to 5 by the year 2025, and to 3 by
the year 2050.
Mortality
Taking into consideration the deterioration of the economic situation in Somalia during the
1980s Fowler assumed a lower expectancy of life at birth (44.7 and 47.8 respectively for males
and females) than that indicated by the official life tables (49.2 and 50.0 for males and females
respectively). Mortality levels were assumed to be constant for the period 1989-91, and allowed
for excess deaths due to famine and civil unrest of the order of 250,000 during 1992-1993. He
then assumed that mortality levels will return to the 1989 levels by 1995 leading to an
improvement in expectancy of life at birth from 16.9 years for males and 17.1 years for females
in 1992 to 21.7 years for males and 22.2 years for females in 1995. He further assumed that
there will be no change in the expectancy of life upto 2000, and this will be followed by an
improvement in life expectancy upto 2050 following a logistic model.
Migration
Migration is a major factor in population change in Somalia and is also difficult to predict. With
practically open borders on all sides, and a tradition of nomadism which does not respect
international frontiers, it is difficult to obtain reliable data on international migration. This is
compounded by the frequent refugee movements as a result of political turmoil in Somalia itself
and in neighbouring Ethiopia and Yemen. Fowler has pieced together bits and pieces of
information on the refugee movements in and out of Somalia and arrived at the assumptions
incorporated in his projections. The estimates of number of migrants up to 1995 is based on the
number of refugees repatriated by UNHCR, and for subsequent years it is based on the
repatriation plan of UNHCR. He assumed that 6000 will be repatriated during 1997, 75000 will
be repatriated in 1998 and another 25000 in 1999. It is further assumed that a residual group of
75000 remaining in Ethiopia at the end of 2000 will be repatriated by 2005.
Fowler's approach to population projections is a sound one and provides an alternative estimate
of population in addition to the one provided by UNDOS on the basis of the settlement survey.
He has taken enormous pains to assemble mortality and migration estimates for successive
years affected by famine, civil strife and difficult economic conditions. However, his base
population figures are on the high side, if we compare his estimate up to 1990 with those
obtained earlier by other experts:
Population in millions

Year Fowler Seetharam World Bank PPPB/DTCD

1975 4.13 4.08 4.12 4.13
1980 5.79 4.65 4.67 4.71
1985 6.45 5.38 5.40 5.41
1990 6.68 6.28 6.28 6.25
 9

The estimates presented here are pre-1986 estimates and are not influenced by the doubtful
figures of the 1986 Census. Like Fowler they have made use of the 1975 census data and
subsequent surveys, but had not foreseen the traumatic events of 1990-92. In particular,
Seetharam was the ILO Expert in the Ministry of Planning in 1986, and his estimate for 1996 is
based on an in-depth analysis of the Somalia situation until that time.
Secondly, it appears that Fowler's assumption of a TFR of 7.25 from 1975 up to 1990 appears to
be on the high side. The estimates of Mohammed Afzal, the CTA of the 1985-86 Census for the
Banadir region is particularly relevant:

Source & Method TFR

1986-86 Census - Brass Method 6.62

6.60
" " - Trussell Method 6.22
6.21
1982 LF Survey - Brass Method 6.85
" " - Trussell Method
1975 Census - Brass Method

REVISED ESTIMATES OF POPULATION OF SOMALIA

Revised estimates of population of Somalia for 1995 have been worked out on the basis of the
following assumptions:
1. The population of Somalia in 1990 was 6.28 million (Seetharam’s estimate) instead of
6.68 million estimated by Fowler.
2. Based on the sensitivity analysis by the Consultant the total fertility rate (TFR) for the
period 1990-95 is 5.0 and not 7.07 as assumed by Fowler.
3. The estimates of international migration adopted by Fowler upto 1995 is acceptable in
the absence of additional information for this period.
On the basis of these assumptions Fowler’s estimate of population for 1995 was scaled down by
applying suitable adjustment factors. If we apply a correction for a smaller base population in
1990, the estimate of population in 1995 comes down to 5.89 million. If we make a further
correction for the lower fertility during the period 1990-95, the population of Somalia in 1995
works out 5.52 million. This estimate is only slightly higher than the estimate of population
derived by UNDOS (5.44 million). It must be pointed out that this estimate should be taken
only as an approximate indication of the population of Somalia, since it is based on rather
questionable data and assumptions concerning the population situation in Somalia.
The estimated total population in 1995 (5.52 million) was distributed pro-rata among the
regions on the basis of the distribution of population by region worked out by UNDOS. The
regional population estimates were further divided according to sex and age groups according to
the sex and age composition of the population of the regions for which data are available from
UNDOS surveys. The socio-economic conditions and migration situation in each region was
examined and an appropriate ‘model’ sex-age distribution was applied. The sex-age distribution
of population for each region for 1995 thus obtained formed the basis for the projections of
population for the period 1995-2015. The cohort-component method of population projections
was applied for each region and aggregated to get the projections of population for Somalia as a
whole. The methods and assumptions adopted are explained in separate not given in the annex,
and the estimates of population by sex and age groups for each region are provided in the
diskette. The estimates for Somalia and for the 18 regions are summarized in Tables 1 and 2.
 10

Table 1
POPULATION ESTIMATES OF SOMALIA BY SEX
1995 - 2015
(figures in hundreds)

Year Both sexes Males Females

1995 55200 29042 26158

1996 56315 29543 26772
1997 57559 30123 27436
1998 58928 30775 28153
1999 60399 31484 28915
2000 62015 32283 29732

2001 63670 33076 30594

2002 65563 34011 31552
2003 67495 34958 32536
2004 69461 35917 33544
2005 71420 36859 34561

2006 73696 37973 35723

2007 75861 39027 36834
2008 77972 40058 37914
2009 80047 41075 38972
2010 82132 42105 40027

2011 84252 43160 41092

2012 86480 44259 42221
2013 88792 45397 43395
2014 91170 46570 44600
2015 93602 47774 45828
Table 2
POPULATION ESTIMATES OF SOMALIA BY REGION
1995 - 2015
(figures in hundreds)

Year Awdal Waalbaad Togdeer Sanaag Sool Bari Nugaal Mudug Galgadud Hiraan Bakool Bay Middle Lower Mogadishu Gado Middle Lower
Shabelle Shabelle Juba Juba

1995 1380 3170 2320 2210 610 2320 1160 3290 1880 2260 2480 6520 4250 6250 6979 3350 1940 2830
1996 1408 3227 2368 2255 624 2370 1184 3353 1918 2300 2530 6654 4329 6369 7130 3425 1976 2884
1997 1439 3291 2423 2305 638 2421 1210 3426 1961 2344 2586 6811 4417 6504 7311 3502 2015 2942
1998 1473 3361 2484 2360 653 2474 1239 3506 2008 2390 2648 6986 4513 6655 7520 3584 2059 3002
1999 1509 3436 2555 2418 669 2528 1271 3591 2060 2438 2716 7178 4615 6817 7759 3667 2104 3064
2000 1548 3517 2632 2482 684 2583 1306 3633 2117 2490 2790 7391 4725 6995 8033 3754 2154 3129

2001 1589 3599 2708 2549 702 2642 1341 3793 2175 2543 2866 7616 4837 7175 8297 3844 2204 3197
2002 1635 3695 2798 2625 723 2711 1383 3903 2244 2608 2956 7861 4969 7381 8582 3951 2263 3279
2003 1683 3794 2891 2702 745 2784 1426 4013 2315 2675 3049 8108 5105 7592 8866 4061 2324 3366
2004 1731 3896 2985 2780 767 2858 1471 4123 2387 2743 3145 8356 5245 7805 9150 4173 2357 3456
2005 1780 3996 3081 2858 790 2934 1517 4235 2460 2813 3242 8599 5384 8019 9428 4287 2450 3549

2006 1835 4117 3187 2948 813 3019 1568 4357 2543 2895 3352 8881 5549 8270 9754 4416 2524 3656
2007 1888 4230 3290 3033 836 3099 1616 4479 2621 2973 3455 9148 5704 8510 10076 4536 2595 3757
2008 1939 4341 3388 3115 859 3175 1662 4601 2697 3047 3553 9408 5856 8744 10401 4652 2663 3855
2009 1989 4449 3485 3196 882 3248 1707 4723 2770 3119 3648 9670 6004 8976 10730 4763 2729 3949
2010 2039 4558 3582 3276 907 3320 1752 4845 2842 3192 3742 9931 6153 9210 11070 4874 2796 4044

2011 2090 4665 3681 3360 931 3393 1798 4969 2919 3255 3842 10195 6284 9444 11404 4974 2855 4128
2012 2143 4784 3783 3446 955 3472 1846 5091 2997 3340 3944 10463 6456 9697 11757 5101 2933 4240
2013 2198 4910 3886 3535 979 3556 1894 5215 3076 3427 4047 10735 6633 9965 12125 5231 3013 4355
2014 2255 5042 3991 3625 1003 3642 1943 5339 3155 3517 4153 11009 6814 10243 12504 5363 3095 4474
2015 2312 5177 4097 3717 1026 3730 1993 5465 3236 3608 4260 11285 7000 10531 12893 5497 3179 4595
Fertility in other urban areas is likely to be higher than that of the Banadir region (which has the
capital city, Mogadishu), while the fertility of nomadic population is likely to be lower. The
World Bank in its Population, Health and Nutrition Sector Review (September 1985) has
adopted a TFR of 6.7 for Somalia for 1980s. Since the population projections are greatly
affected by the assumptions on TFR, the assumption of a TFR of 7.25 instead of 6.7 could result
in an overestimation of population for the projection period.
A third source of overestimation in Fowler's estimate could arise from the handling of refugee
migration. He has accepted UNHCR's repatriation plans as the basis for migration assumptions.
The ground realities appear to be different. If two refugees are sent back by UNHCR, there is at
least one leaving the country. For instance, the UNHCR is sending back 20000 refugees a
month, but the number in the refugee camps in Kenya is not decreasing, it has in fact increased
during the past several months. The UNHCR camps in Kenya have close to 180000 refugees
and the UNHCR camps in Ethiopia have 480000 refugees. In addition to the refugees in the
UNHCR camps, there are at least thrice that number in private households in Kenya and
Ethiopia, although there are no statistical data to substantiate this. In view of the presence of
large Somali communities in these countries, and the extended family ties governing their
relationships, it is easy for a Somali to find refuge in the home of a friend or relative in Kenya
and Ethiopia. So long as conditions in Somalia remain tenuous as at present, there is little
incentive for Somalis to return to their country for permanent settlement. They may practice
some form of "international nomadism", whereby they go to Somalia and return to wherever
they are. A more realistic assumption for population projections at this point of time is to
assume zero net migration beyond 1995.

DEMOGRAPHIC DATA COLLECTION THROUGH UNDOS SURVEYS

In the past UNDOS was collecting some basic demographic data through the family budget
survey, but these are not adequate for assessing demographic characteristics of the population or
estimating demographic parameters such as fertility, mortality and migration. The Consultant
held discussions with UNDOS staff on the feasibility of conducting a Demographic Survey
alongside the family budget survey and other socioeconomic surveys to collect the data required
for the estimation of demographic parameters. He has proposed a cluster sample design
covering 600 urban, 600 rural and 300 nomadic households for each region (with the exception
of Mogadishu where a slightly larger sample of 2000 households may be covered. The
Demographic survey should include a module on reproductive statistics, to obtain information
on pregnancy history and age at marriage, age at first child birth, duration of fertile union, and
other key parameters required by policy makers. A draft questionnaire of the Demographic
Survey and Reproductive Statistics module were discussed thoroughly with UNDOS
professional staff and Somali professionals, and these questionnaires will be protested in
Mogadishu and Jowhar before the surveys are launched. Unlike a population census which has
to be carried out at a fixed reference date, these surveys could be carried out in a phased manner
in regions where the security situation will permit the undertaking of the survey. A detailed
tabulation plan and the analytical procedures have been prepared by the Consultant, which if
implemented will provide the estimates of the key demographic parameters required for
planning and policy making. See the details in the Annex.

PROPOSAL FOR A SENTINEL REPORTING SYSTEM

While the above mentioned Demographic Survey and the Reproductive Statistics Module
attached to it will provide estimates of demographic parameters at the macro level, we will still
need data on vital statistics at the micro level of small areas. Such data can be collected only
through a vital registration system at the level of village or settlement (beel). Unfortunately
there is no administrative structure in Somalia to initiate and sustain such a system. The entire
health sector is dominated by international and local NGOs who are active in providing essential
 2

drugs, immunization and MCH services. Some NGOs such as Adventist Development and
Relief Agency (ADRA) and World Concern have an elaborate system of data collection through
the traditional birth attendants (TBAs) and Community Health Workers (CHWs) in the areas
where they are having health interventions. Even the other NGOs have some kind of data
gathering to monitor their programmes, and to plan health interventions.
The experience of the NGOs like ADRA and World Concern shows that if there is committed
staff and good supervision it is feasible to collect vital statistics. The Consultant found that
several NGOs are receptive to the idea of instituting a system of data collection in a few areas,
which could be designated as sentinel areas, where a common format for data collection could
be adopted. The choice of the sentinel areas is very important. These areas should have strong
NGO presence and supervision of the NGO's Medical Officer and Social Worker. It should also
have committed and well trained TBA and CHW. The Consultant has prepared the formats of
the Household Register and the Birth and Death Report forms. These forms could be discussed
by the NGOs and international agencies and modified, if necessary. The principle of the
Sentinel Reporting System is to choose a few areas where the ingredients of success are present
and implement the system. The system could be expanded gradually in terms of space and
content, so as to provide the key demographic parameters for the different parts of Somalia.
Apart from using the data for their internal monitoring and evaluations, the NGOs could provide
the data to the UNDOS for compilation and dissemination to various NGOs and international
agencies. The details of the Sentinel Reporting System are given in the Annex

RECOMMENDATIONS FOR FUTURE ACTION

1. Conducting a census in Somalia in the present circumstances is an impossible task.
There is no administrative machinery for statistical data collection, moreover the
security situation in the country, and the frequent factional fights in the areas under the
control of different factions makes it difficult to carry out a census at a fixed reference
date. Moreover carrying out a census is an expensive and time consuming task and the
efforts can be better utilized for more urgently required humanitarian relief.
Consequently, efforts should be focused on the collection of data through the present
system of Settlement Surveys carried out by UNDOS.
2. While the UNDOS settlement surveys provide the estimates of population at the level of
the settlements, further aggregated to that of the districts and regions, they are not
designed to provide the estimates of the demographic parameters. At present some
demographic data are collected through the family budget survey. It is recommended
that a separate demographic survey be implemented covering the different regions in a
phased manner. A cluster sample design is recommended whereby 50 clusters of 30
household may be selected at random from each region. Even here, because of the
security situation strictly random sampling procedures may not be applicable, and in
such a situation efforts should be directed to ensure a wider spread of the sample to
represent all the cross sections of the population. Likewise, it is impossible to locate a
random sample of nomadic households in the semi-arid expanse of Somalia. Nomadic
families should be covered wherever they are found, their frequent movement being
treated as a random process.
3. UNDOS should pre-test the questionnaires of the Demographic Survey and other
Socioeconomic Surveys in the light of the changes made by the Consultant, and
implement them in a phased manner in the different regions, taking into account the
security situation, availability of qualified enumerators, considerations of accessibility
etc. UNDOS is planning a Household Income and Expenditure survey in some regions.
This would provide an opportunity to pre-test the methodologies and questionnaires
proposed.
 3

4. UNFPA should consider supporting UNDOS on a two or three years project for
Demographic Data Collection in a phased manner. This project could include local
training of Somali statistical and survey professionals in the collection and processing
of data, publication of survey results and the preparation of an analytical report. It is
essential to strengthen UNDOS by the addition of a Computer Specialist for Processing
Survey Data. The provision of a demographer in the project can greatly contribute to
the analysis of the demographic and socioeconomic data collected through the surveys
carried out by UNDOS.
5. The NGOs working in different parts of Somalia can greatly contribute to our
knowledge of Somalia, if they could adopt a common format of data collection for
health information, vital statistics and reproductive statistics. Besides using such
information for their internal evaluation of their activities as done at present they could
provide these data to UNDOS for compilation and dissemination. While it is not
feasible to collect such information for all parts of the country, they could be collected
in areas where there is strong NGO supervision and committed and trained staff.
6. UNDOS is at present processing the data from their surveys using the spreadsheet and
data base management application software. It may be advantageous to install the
specialized data processing software such as the Integrated Microcomputer Processing
System (IMPS) software developed by the US Bureau of Census, Washington D.C. It is
also advisable if a senior staff member of UNDOS is given training at the US Bureau of
Census on the use of this software. This person could then train the Somali
professionals in its use. It may be stated the IMPS are most suited for the processing of
diverse data collected by UNDOS.

ACKNOWLEDGEMENTS
The Consultant has greatly benefited from the support and encouragement of UNDOS, in
particular Dr.K.N.S.Nair, Officer-in-Charge and Dr.Robert Hagan, WHO Representative for
Somalia. Mr.Timothy Fowler of US Bureau of Census and Mr. Phil Steffen of USAID/FEWS
provided the Consultant the information available in their respective organizations.
 4

ANNEX 1

ESTIMATION OF POPULATION OF SOMALIA

BY REGION, SEX AND AGE GROUPS, 1995-2015
by
K.E. Vaidyanathan
UNFPA Consultant on Population Statistics

Introduction
Somalia has gone through a civil war during 1990-92, and this has greatly affected the
population distribution and the sex and age composition of the population. Somalia is also
unique in other respects as well. The population movements take place with the beginning of the
dry season, and the reverse movements take place with the arrival of the wet season. With about
one half of the population nomadic, there is frequent change in the population size depending
upon the movements of the nomadic population. The frequent quarrels and fights between the
different clans and sub-clans has also resulted in movement of population to safer areas, and
their return to their previous habitats when the situation cools down. International migration,
both forced and spontaneous, is important in Somalia. With open borders on all sides, there are
frequent movements of refugees across the borders to Ethiopia and Kenya. The direction of
these movements has changed according to the political situation in Somalia and the changes in
her relationship with her neighbours. In other words, Somalia's population is in a state of flux,
and therefore estimates of population in such a situation is subject to great uncertainties.
Moreover, the population figures can be different depending upon whether it is defacto or dejure
population. What has been attempted here is to look at the various sketchy information is
available for Somalia, and put together estimates of population on the basis of the available
information and assumptions where the information is lacking. This note indicates the methods
adopted and the assumptions made, to enable the users of these estimates to make adjustments
of these estimates in the light of any further information that may become available.

Sources of Data and their Limitations

Demographic Data available for Somalia are very limited, and even these are greatly suspect.
The first Census of Somalia was carried out in 1975, but the complete census results were not
published. Moreover the 1975 census collected data on fertility and mortality only for the
settled population. Only an analytical report of the Census appeared in 1984 with limited
information. A Demographic Survey carried out in 1980-81 was never processed. Another
survey was carried out in 1980-81 in three regions (Banadir, Bay and Lower Shabelle) under the
POPLAB project and even until this day this survey has remained the only source of
information on fertility and mortality. In 1986-87 another census was carried out, but the results
were not published. Information on nomadic population which constitute about one-half of the
population are greatly deficient. Moreover, data on international migration are lacking except
for those living in refugee camps in Ethiopia and Kenya and their repatriation by UNHCR.
Even here there is no information available about number of returnees and the areas where they
were resettled. It is difficult to define and measure internal migration in a situation where a
significant proportion of the population is nomadic.

Population Size of Somalia

The population Census held in February 1975 enumerated a population of 3.25 million. On the
basis of a post enumeration survey and other various analytical methods it was estimated that
the undercount in the census was 20.4 percent. Thus the adjusted population in 1975 was
worked out as 4.089 million. On the basis of this figure and estimates of other parameters, the
 5

Ministry of National Planning of Somalia estimated the population in mid-1985 to be 5.4

million. This estimate did not include the 700,000 refugees who moved into Somalia from
Ethiopia as a consequence of the war with Ethiopia in 1977 and the subsequent draught. On the
other hand this estimate included the 100,000 Somali workers in the Gulf region since they were
regarded as temporary migrants. In other words, the defacto population of Somalia in 1986
could be 6.0 million, while the dejure population was 5.4 million.
There is great controversy about the population of Somalia according to the 1986-87 Census,
and the figure range between 5.1 and 8.5 million. Mohammed Afzal who was the CTA for the
Census project mentions the adjusted population for 1986-87 as 7.1 million. It is not clear
whether this is for mid the year of 1986 or 1987. If the two estimates for 1976 and 1986 were
accepted, this would imply a growth rate of 4.6 percent, which is clearly untenable. Moreover
this is not consistent with the Ministry of National Planning estimate of 5.4 million for 1985.
Another indication of inflation comes from the fact that the percentage of nomadic population in
1986-87 Census was reported to be 59.4 percent against 46 percent in the 1975 census and 42
percent for 1985 estimated by Seetharam, CTA in the Ministry of National Planning at that
time. Obviously, the villain of the piece is the nomadic population, which has been grossly
exaggerated in the 1986-87 Census. Based on the data available prior to the 1986-87 census
Seetharam has projected the population of Somalia upto 1990 and came up with a figure of 6.28
million for mid 1990. This estimate was made prior to the 1986-87 Census and is therefore not
influenced by the doubtful figures of this census. This figure is credible as a base population for
our exercise in population estimation for 1995.
For making estimates of population in 1995, we need some estimates of fertility, mortality and
migration during the period 1990-95. Unfortunately, consequent upon the civil war during
1990-91 and the resulting movements of the population, the entire demographic composition of
the population had changed. The settled populations of several areas moved out, while the
nomadic population moved to fill their places. There were great casualties during the civil war
and equal or even greater casualties due to starvation, but no information is available. Members
of different clans moved to the areas where they felt more secure. Because of the chaotic
conditions of those days' mortality increased while fertility may have declined although there
are no evidences to substantiate this. This makes estimation of population difficult without
additional data on these parameters.
What is attempted here is a kind of sensitivity analysis of what might happen under varying
conditions of fertility, mortality and migration. The assumptions are made in terms of
alternative values of Total Fertility Rate (TFR), expectancy of life at birth and net international
migration. The 1990 population was the same for all the variants attempted. We have made use
of two slightly different patterns of age specific fertility rates given below:

Age Specific Fertility Rates

Age group Pattern I Pattern II

15-19 .0701 .1674
20-24 .2859 .2434
25-29 .3329 .2593
30-34 .2709 .2494
35-39 .2236 .2101
40-44 .1486 .1237
45-49 .0960 .0691
Sources: Pattern I. POPLAB Survey of Banadir, Bay & Lower Shabelle, 1980
Pattern II. Estimates by Mohammed Afzal for Banadir region based on 1986-87 Census.
 6

These two sets of age specific fertility rates have been taken to represent two possible patterns
rather than exact levels of fertility, the levels being represented by the assumed TFR values. The
results of these test calculations (designated as SOMTEST) is given below:

SOMTEST Results

TFR 6.61 6.61 5.00 5.00 4.5

Pattern I I I I I
Expectancy
Males 40.0 40.0 40.0 40.0 40.0
Females 42.0 42.0 42.0 40.0 40.0
Net Migration (000)
Males – -550 -550 -550 -550
Females – -250 -250 -250 -250
Est. Pop 1995 (000) 7047 6171 5909 5886 5805
Sex Ratio 100.7 91.4 90.9 91.5 91.3
Growth Rate(%) 2.29 -0.36 -1.23 -1.31 -1.58

TFR 6.61 6.00 5.00 5.00 4.50

Pattern II II II II II
Expectancy
Males 40.0 40.0 40.0 40.0 40.0
Females 42.0 42.0 42.0 42.0 42.0
Net Migration (000)
Males – – – -550 -550
Females – – – -250 -250
Est. Pop 1995 (000) 7099 6988 6806 5947 5838
Sex Ratio 100.7 100.6 100.6 90.9 91.4
Growth Rate(%) 2.44 2.13 1.60 -1.60 -1.47

These calculations show that the population estimate for Somalia varies between 5,805,000 and
7,099,000 depending upon the pattern of fertility, and the average values of TFR, expectancy of
life at birth and net migration assumed for the period 1990-95. The higher values of population
can be dismissed as unrealistic as a negative growth rate may be expected in the Somalia
situation. This would indicate that the TFR could be 5.0 or even 4.5 during the period 1990-95
instead of the usually assumed values ranging between 6.7 and 7.4. The assumed pattern of
fertility is based on the pre-war situation, and may not be valid for the period 1990-95. The
average expectancy of life at birth for the period 1990-95 in Fowler's calculation works out to
31.5 for males and 33.2 for females. If these figures are assumed that an expectancy of life as at
birth is 40 years, 180,000 will further reduce the population, i.e. to 5,705,700. The figures of
net out migration are again an assumption, although based on the number of refugees repatriated
by UNHCR. The actual figures could be higher or lower, but we have no information about the
number of returnees among those living in private households. In sum, at the end of this
exercise, we are nowhere near the ends of the tunnel in our search for a unique population
estimate for Somalia.

Review of UNDOS Methodology of Estimating Population

Since its inception in 1994, UNDOS has been playing a key role in the collection and
compilation of demographic and socioeconomic data required for decision making in a number
of areas such as population, health, education and socioeconomic data required for humanitarian
assistance and rehabilitation. The international agencies, international and Somali NGOs and
 7

even the local authorities are using these data where they are functioning. UNDOS has
developed a unique methodology for estimating the population at the micro level through
settlement surveys and these data are aggregated to provide estimates at the district, regional and
national levels. A settlement consists of a main village and a network of satellite villages
surrounding the main village. The typical main village had a market, school and mosque. In the
livestock producing areas, most of the beels had only the main village and no satellite villages.
Under the UNDOS procedure, the supervisors and enumerators contacted the village elders and
other knowledgeable persons to gather information on the number of houses, population and
socioeconomic characteristics of the main village and four large satellite villages utilizing the
settlement questionnaire. While this is going on, one of the enumerators goes round the villages
and makes a physical count of the number of houses in these villages. The main village and
satellite villages are then plotted on a map utilizing the geographical positioning system (GPS),
which provided the coordinates (longitudes and latitudes) for these villages. In some of these
villages a family budget survey is carried out in a sample of households, and the average
population per house is calculated from the family budget survey data. UNDOS estimated the
population by multiplying the number of houses in each village (from the actual count) by the
average population per house derived from the family budget survey. For the smaller satellite
villages, the number of houses was based on actual count or through information gathered from
village elders, and population of these villages was estimated by multiplying the number of
houses by the average population per house. This methodology has the merit of simplicity, and
is appropriate for situations like that of Somalia, where carrying out a nationwide census is not
feasible because of the security situation. Another advantage of this methodology is its
flexibility, since unlike in a census we do not have to cover all the areas on a fixed reference
date, and savings are effected on cost and time through the combined use of the GPS and
sampling.
The population estimates based on this method for the five regions appear realistic from the time
series data given below:
Population Estimates

Regions 1975 (Census) 1986 (Census) 1996 (UNDOS)

Bari 194,368 222,287 229,000

Nugal 64,790 74,774 114,000

Mudug 269,464 311,230 324,000

Middle Shabelle 309,221 352,040 419,000

There are however two possible sources of underestimation. First, there is a universal tendency
in censuses and surveys to miss children, and this could have happened in the family budget
survey which provided the estimate of average population per house. Evidence of this is found
in the percentage of population in the childhood age groups:

Age group Bari Nugal Mudug M. Shabelle Combined

M F M F M F M F M F
Under 1 1.07 1.48 1.45 1.19 0.73 0.72 0.29 0.53 0.84 0.99
1-4 12.60 13.21 13.12 11.37 11.89 11.55 12.61 13.53 12.53 12.49
5-9 18.05 15.58 19.28 16.19 19.25 15.87 18.67 19.27 18.74 16.73

Under normal conditions one would expect the percentages in the three age groups to be 4, 13
and 15 percents respectively. The percentages of population less than one year is obviously
low, indicating omission of children or a fall in fertility. The children under one year of age are
 8

survivors of those born one year earlier, namely 1995, whereas those in the age group 1-4 are
those born during 1991-94 which include the years of turmoil. Therefore one would expect the
percentages to be much lower than expected in the 1-4 age group, and not in the under-one age
group. This would suggest that there has been an under count in the under-one age group. On
the assumption that the 1-4 age group is correct, we may apply a survival factor to obtain an
estimate of the population under one. This would indicate that the population under one may be
an underestimate by 76 percent, but this will result in an undercount of only 3 percent of the
total population.
Another puzzling feature of the age data for the four regions is the shortfall in the number of
females below 20 and above 40, while in the ages 20-40, the number of women consistently
exceed the number of men. The deficit of men in the age range 20-40 can be explained by
selective emigration of men, as well as higher death rates due to the civil war. The 1975 census
and 1980 survey also have more males than females in the younger ages, but females exceed
males in the ages above 40. There is therefore a suspicion that more females than males may
have been omitted in the ages above 40, or fewer women are surviving to older ages. The
proportion of population above 60 years of age in the 1975 population census was 5.29 percent
– 5.20 percent for males and 5.37 percent for females. The settlement survey of UNDOS gives
only 3.86 percent in this age group – 5.12 percent for males and 2.48 percent for females. To
some extent shortfalls can occur due to high mortality at older ages, but the difference observed
for females is too large to be accounted by this factor alone. If we assume that the sex ratios
found in the 1975 census for ages beyond 40 is acceptable, this would indicate a 34 percent
under-enumeration of women beyond 40 years of age. In terms of total population, this will
amount to an under count of 3.1 percent. These two factors, namely omission of children under
one and the women above 40 years of age will thus account for an undercount of 6 percent of
the population. Even an undercount of this order is a remarkable achievement under the
circumstances prevalent in Somalia.

Review of the Methodology of Timothy Fowler

In March 1997, USAID Somalia Office mounted a mission by Timothy Fowler of US Bureau of
Census to examine the available data for Somalia and make population projections at the
national level. Fowler adopted the cohort-component method of projections using the RUP
software developed by the US Bureau of Census. He started with the base population from the
1975 census, and based on the available information on fertility, mortality and migration
incorporated assumptions regarding these parameters for successive years. His assumptions are
briefly summarized below:
Fertility
The total fertility rate (TFR) is assumed to be constant at 7.25 from 1975 up to 1990, declined to
5.5 in 1992 due to famine and civil unrest and return to the 1990 levels by 1999. It is further
assumed that the TFR will decline by 10 percent by 2009, and to 5 by the year 2025, and to 3 by
the year 2050.
Mortality
Taking into consideration the deterioration of the economic situation in Somalia during the
1980s Fowler assumed a lower expectancy of life at birth (44.7 and 47.8 respectively for males
and females) than that indicated by the official life tables (49.2 and 50.0 for males and females
respectively). Mortality levels were assumed to be constant for the period 1989-91, and allowed
for excess deaths due to famine and civil unrest of the order of 250,000 during 1992-1993. He
then assumed that mortality levels will return to the 1989 levels by 1995 leading to an
improvement in expectancy of life at birth from 16.9 years for males and 17.1 years for females
in 1992 to 21.7 years for males and 22.2 years for females in 1995. He further assumed that
there will be no change in the expectancy of life up to 2000, and this will be followed by an
improvement in life expectancy up to 2050 following a logistic model.
 9

Migration
Migration is a major factor in population change in Somalia and is also difficult to predict. With
practically open borders on all sides, and a tradition of nomadism which does not respect
international frontiers, it is difficult to obtain reliable data on international migration. This is
compounded by the frequent refugee movements as a result of political turmoil in Somalia itself
and in neighbouring Ethiopia and Yemen. Fowler has pieced together bits and pieces of
information on the refugee movements in and out of Somalia and arrived at the assumptions
incorporated in his projections. The estimates of number of migrants up to 1995 is based on the
number of refugees repatriated by UNHCR, and for subsequent years it is based on the
repatriation plan of UNHCR. He assumed that 6,000 will be repatriated during 1997, 75,000
will be repatriated in 1998 and another 25,000 in 1999. It is further assumed that 2005 will
repatriate a residual group of 75,000 remaining in Ethiopia at the end of 2000.
Fowler's approach to population projections is a sound one and provides an alternative estimate
of population in addition to the one provided by UNDOS on the basis of the settlement survey.
He has taken enormous pains to assemble mortality and migration estimates for successive years
affected by famine, civil strife and difficult economic conditions. However, his base population
figures are on the high side, if we compare his estimate up to 1990 with those obtained earlier
by other experts:
Population in millions

Year Fowler Seetharam World Bank PPPB/DTCD

1975 4.13 4.08 4.12 4.13
1980 5.79 4.65 4.67 4.71
1985 6.45 5.38 5.40 5.41
1990 6.68 6.28 6.28 6.25

The estimates presented here are pre-1986 estimates and are not influenced by the doubtful
figures of the 1986 Census. Like Fowler they have made use of the 1975 census data and
subsequent surveys, but had not foreseen the traumatic events of 1990-92. In particular,
Seetharam was the ILO Expert in the Ministry of Planning in 1986, and his estimate for 1996 is
based on an in-depth analysis of the Somalia situation until that time.
Secondly, it appears that Fowler's assumption of a TFR of 7.25 from 1975 up to 1990 appears to
be on the high side. The estimates of Mohammed Afzal, the CTA of the 1985-86 Census for the
Banadir region is particularly relevant:

Source & Method TFR

1986-86 Census - Brass Method 6.62

6.60
" " - Trussell Method 6.22
6.21
1982 LF Survey - Brass Method 6.85
" " - Trussell Method
1975 Census - Brass Method

Fertility in other urban areas is likely to be higher than that of the Banadir region (which has the
capital city, Mogadishu), while the fertility of nomadic population is likely to be lower if the
experience of countries like Sudan and Mauritania is valid for Somalia. We have however no
basis for confirming this assumption until we are able to collect data on fertility of the nomadic
and settled population and compare them. The World Bank in its Population, Health and
Nutrition Sector Review (September 1985) has adopted a TFR of 6.7 for Somalia for 1980s.
The fertility in a war situation is invariably lower than under normal times as the experience of
Europe during the Second World War has shown. The values of TFR assumed by Timothy
 10

Fowler for the period 1990-95 averages to 7.07, while a more realistic value would be 5.0 even
4.5. Again we have no basis to confirm this figure except the sensitivity analysis performed
earlier. Since the population projections are greatly affected by the assumptions on fertility, the
assumption of a TFR of 7.07 instead of 5.0 or 4.5 could result in an overestimation of
population for 1995. The overestimation could be of the order of 6.3 percent and 9.2 percent if
the TFR value is reduced to 5.0 and 4.5 respectively. We need to adjust Fowler's population
estimate for 1995 first for the larger base population in 1990, and further for the higher value of
TFR assumed by him. If we apply the correction for the higher base population his estimate
comes down to 5.89 million, and when a further correction is carried out for the fall in fertility
during 1990-95, the estimate gets further reduced to 5.52 million.
A third source of overestimation in Fowler's estimate could arise from the handling of refugee
migration. He has accepted UNHCR's repatriation plans as the basis for migration assumptions.
The ground realities appear to be different. If two refugees are sent back by UNHCR, there is at
least one leaving the country. For instance, the UNHCR plans to send back 20,000 refugees a
month, but the number in the refugee camps in Kenya is not decreasing, it has in fact increased
during the past several months. At this time (October 1997) the UNHCR camps in Kenya have
close to 180,000 refugees and the UNHCR camps in Ethiopia have 480,000 refugees. In
addition to the refugees in the UNHCR camps, there are at least thrice that number in private
households in Kenya and Ethiopia, although there are no statistical data to substantiate this. In
view of the presence of large Somali communities in these countries, and the extended family
ties governing their relationships, it is easy for a Somali to find refuge in the home of a friend or
relative in Kenya and Ethiopia. So long as conditions in Somalia remain tenuous as at present,
there is little incentive for Somalis to return to their country for permanent settlement. They may
practice some form of "international nomadism", whereby they go to Somalia and return to
wherever they are. However we have no basis for adjusting Fowler's estimate for possible over
or under estimation of net migration during 1990-95.
To sum up, the UNDOS population figure of 5.44 million in 1995 is very close to this estimate
of 5.52 million for 1995, the difference being only 1.5 percent of the adjusted population of 5.52
million. The assumptions on which this estimate is based are indicated above, and the estimate
can be revised on the basis of the fresh data that will be made available from the UNDOS
surveys. Indeed there is need to fresh look at these estimates after about 6 months when the
Demographic Survey and Reproductive Statistics Module would have yielded data on fertility
and mortality and to a limited extent on migration.

Estimation of Population by Region

UNDOS has provided the break-up of the population of Somalia by region for 1995. In the
absence of other information the estimated population of 5.52 million for 1995 is distributed
between the regions on a pro-rata basis. Comparison of these estimates for four regions with
those available from other sources such as 1975 census show that they are realistic, although no
one could be sure about the exact figure for each region. However, these estimates could be
accepted as a working basis until further data become available. These estimates are available
for both sexes only and not for males and females. Our first task is to split the population of
both sexes into males and female population. Various alternatives were explored and it was felt
that the post-war data of UNDOS would be more realistic rather than those of the pre-war
period. The different regions were grouped into 5 categories on the basis of the social and
economic situation prevailing in each region and the sex age composition of the regions for
which data are available from UNDOS surveys was attributed. When UNDOS surveys are
carried out in the different regions, the appropriate sex-age composition could be adopted.
 11

Assumptions Regarding Fertility, Mortality and Migration

Fertility
Fertility data are limited to the pre-war period, and are not strictly valid for present day Somalia.
The only fertility estimates available are those derived from the 1975 census and the POPLAB
survey of 1980-81, The 1975 census results are of poor quality and the only acceptable
estimates are those of the POPLAB survey for the urban and rural settled population. No data
are available for the nomadic population of Somalia. The World Bank had estimated the TFR
of Somalia in the 1980s as 6.7 on the assumption that the TFR of nomads is 6.0, and the TFR of
Mogadishu, other urban and rural settled population are 7.2, 7.5 and 7.1 respectively. Due to
the difficult economic conditions following the civil war and increase in the number of widowed
and abandoned women, one should expect a decline in fertility in Somalia. The only available
information is the child woman ratios calculated from the 1975 adjusted data and the data from
the UNDOS surveys for the four regions during 1996, given below:
Child Woman Ratios

Census UNDOS Survey

1975 1996
Child Woman Ratio 1 0.847 0.614
Child Woman Ratio 2 0.902 1.017
Percent under 10 33.1 31.2

It must be pointed out that these are not pure measures of fertility and are subject to the errors of
age mis-reporting and omissions of population, and age differentials in migration. However it is
interesting to note that while Child Woman Ratio 2 is nearly the same for the two periods, Child
Woman Ratio 1 shows a marked decline. This perhaps reflects a falling fertility during 1991-96
in comparison to the levels during 1986-91. It is however difficult to quantify the decline in
terms of the fall in TFR, since there is no one to one relationship between the two. As a rough
approximation however one may infer that the drop in fertility might be of the order of one-
fourth since the 1980s. If the TFR in the 1980s is assumed to be 6.7 (World Bank estimate), the
TFR in the 1990s could be around 5.0. A country like Somalia with universal marriage, absence
of contraception, and a high value attached to children (they are the blessings of ALLA), cannot
sustain a low level of fertility for a long time. As soon as peace returned, and families reunite
we should expect a "baby boom" as it happened in the western societies after the war. Indeed
this seems to be happening among the Somalis in Kenya (or at least in UNDOS).
The baseline estimates of TFR for each region was derived as the weighted average of the TFRs
for the urban, rural and nomadic population for each region, the weights being the population of
each of these groups in each region. The TFRs assumed for urban, rural and nomadic population
is 7.1, 7.4 and 6.0 respectively. These are taken as the base figures for the 1980s. The TFR for
Somalia as a whole is assumed to be 6.0 for the period 1995-2000 increase to 6.5 for the period
2000-2005, and then decline to 6.0 in 2005-10, and 5.5 for 2010-15. The TFRs for each of the
regions were prorated to match these figures for Somalia as a whole.
Mortality
Data on mortality in Somalia are even more limited than the data on fertility. The only reliable
source of mortality estimates is the POPLAB survey of 1980-81, and until this day these
estimates are the basis for all the analysis and projections for Somalia. Life expectancy in the
late 1970s is estimated at 42 years (41.1 for males and 42.3 for females) in the rural settled
population. and at 48 years (47.0 years for males and 48.8 years for females) in the urban
population. There are no reliable data to estimate nomad mortality. On the assumption that
nomad mortality is similar to rural settled mortality in Somalia, national life expectancy in 1985
was estimated at 47 years (45.7 for males and 47.6 for females) with a national IMR of around
 12

145 and about 250 children dying within their first five years of their life for every 1000
livebirths.
During 1990-95 mortality rates increased considerably as a result of the civil war, and the
economic crisis which followed, and the expectancy of life at birth was in the range of 30-35
years. With the return of peace in much of Somalia, one should expect that the expectancy life
should have returned to the conditions of the late eighties during 1995-2000. We may expect
the expectancy of life at birth to remain at this level until 2005, and increase at the rate of half
year per year during the subsequent years. The base line expectancies of life for each region
were estimated by weighting the expectancies of life for the urban, rural and nomadic
populations by their populations and the changes for the subsequent periods were assumed to
follow the pattern assumed for Somalia as a whole. These assumptions will need to be reviewed
when fresh data become available from the demographic surveys planned by UNDOS.
Migration
Both international and internal migration is important components of regional population
change, however we do not have any reliable data to estimate the net migration for each region.
The existence of nomadic and seasonal movements makes it more difficult to estimate internal
migration. Moreover the direction of migration changes from year to year in response to
economic conditions, draught, as well as policies of neighbouring countries to Somalia refugees.
In view of the unpredictable nature of migration and the lack of data we have assumed that
migration is negligible during the coming years up to 2015. This assumption will need revision
in the light of any fresh data that may become available through UNDOS surveys.
Methodology of estimation up to year 2015
The methodology adopted for the projections is the standard cohort component method,
whereby the survivors of the population in each age group is estimated by application of
survival ratio corresponding to the assumed expectancies of life. The Coale-Demeny South
Model Life Tables have been adopted for deriving these survival ratios. The estimates of
children in the age group 0-4 were derived by applying the age specific fertility rates
corresponding to the assumed values of TFR to the estimated number of women in the
reproductive age group for each projection period. The PEOPLE software developed by UN-
ESCAP, Bangkok was utilized for the projection exercise.
Results of the Projections
The estimates of population by sex and age groups for each region and for Somalia as a whole
are available in a diskette, and a summary of these estimates are presented in Tables 1 and 2.
Limitations of the Estimates and Projections
The estimates presented here are based on scanty and often doubtful information on most of the
parameters of the projections-sex-age distribution of population, fertility, mortality and
migration. Assumptions are assumptions and are not facts. When facts become available,
assumptions should be replaced by facts. It is here UNDOS has a major role in the coming
years to collect the data required for a more scientific approach to population estimation. It is
expected that UNDOS surveys will provide the data for the indirect estimation of fertility and
mortality, and some direct and indirect evidences about migration. UNDOS surveys will also
provide the sex-age distribution of population which is vital for any demographic analysis, and
even more so for population projections. When these data become available the estimates and
the projections presented here should perforce be revised and updated.
 13

Table 1
POPULATION ESTIMATES OF SOMALIA BY SEX
1995 - 2015
(figures in hundreds)

Year Both sexes Males Females

1995 55200 29042 26158

1996 56315 29543 26772
1997 57559 30123 27436
1998 58928 30775 28153
1999 60399 31484 28915
2000 62015 32283 29732

2001 63670 33076 30594

2002 65563 34011 31552
2003 67495 34958 32536
2004 69461 35917 33544
2005 71420 36859 34561

2006 73696 37973 35723

2007 75861 39027 36834
2008 77972 40058 37914
2009 80047 41075 38972
2010 82132 42105 40027

2011 84252 43160 41092

2012 86480 44259 42221
2013 88792 45397 43395
2014 91170 46570 44600
2015 93602 47774 45828
Table 2
POPULATION ESTIMATES OF SOMALIA BY REGION
1995 - 2015
(figures in hundreds)

Year Awdal Waalbaad Togdeer Sanaag Sool Bari Nugaal Mudug Galgadud Hiraan Bakool Bay Middle Lower Mogadishu Gado Middle Lower
Shabelle Shabelle Juba Juba

1995 1380 3170 2320 2210 610 2320 1160 3290 1880 2260 2480 6520 4250 6250 6979 3350 1940 2830
1996 1408 3227 2368 2255 624 2370 1184 3353 1918 2300 2530 6654 4329 6369 7130 3425 1976 2884
1997 1439 3291 2423 2305 638 2421 1210 3426 1961 2344 2586 6811 4417 6504 7311 3502 2015 2942
1998 1473 3361 2484 2360 653 2474 1239 3506 2008 2390 2648 6986 4513 6655 7520 3584 2059 3002
1999 1509 3436 2555 2418 669 2528 1271 3591 2060 2438 2716 7178 4615 6817 7759 3667 2104 3064
2000 1548 3517 2632 2482 684 2583 1306 3633 2117 2490 2790 7391 4725 6995 8033 3754 2154 3129

2001 1589 3599 2708 2549 702 2642 1341 3793 2175 2543 2866 7616 4837 7175 8297 3844 2204 3197
2002 1635 3695 2798 2625 723 2711 1383 3903 2244 2608 2956 7861 4969 7381 8582 3951 2263 3279
2003 1683 3794 2891 2702 745 2784 1426 4013 2315 2675 3049 8108 5105 7592 8866 4061 2324 3366
2004 1731 3896 2985 2780 767 2858 1471 4123 2387 2743 3145 8356 5245 7805 9150 4173 2357 3456
2005 1780 3996 3081 2858 790 2934 1517 4235 2460 2813 3242 8599 5384 8019 9428 4287 2450 3549

2006 1835 4117 3187 2948 813 3019 1568 4357 2543 2895 3352 8881 5549 8270 9754 4416 2524 3656
2007 1888 4230 3290 3033 836 3099 1616 4479 2621 2973 3455 9148 5704 8510 10076 4536 2595 3757
2008 1939 4341 3388 3115 859 3175 1662 4601 2697 3047 3553 9408 5856 8744 10401 4652 2663 3855
2009 1989 4449 3485 3196 882 3248 1707 4723 2770 3119 3648 9670 6004 8976 10730 4763 2729 3949
2010 2039 4558 3582 3276 907 3320 1752 4845 2842 3192 3742 9931 6153 9210 11070 4874 2796 4044

2011 2090 4665 3681 3360 931 3393 1798 4969 2919 3255 3842 10195 6284 9444 11404 4974 2855 4128
2012 2143 4784 3783 3446 955 3472 1846 5091 2997 3340 3944 10463 6456 9697 11757 5101 2933 4240
2013 2198 4910 3886 3535 979 3556 1894 5215 3076 3427 4047 10735 6633 9965 12125 5231 3013 4355
2014 2255 5042 3991 3625 1003 3642 1943 5339 3155 3517 4153 11009 6814 10243 12504 5363 3095 4474
2015 2312 5177 4097 3717 1026 3730 1993 5465 3236 3608 4260 11285 7000 10531 12893 5497 3179 4595
 ANNEX 2

DESIGN OF A DEMOGRAPHIC SURVEY OF SOMALIA

by K.E.Vaidyanathan
UNFPA Consultant on Population Statistics

Introduction
Demographic data for Somalia have been scanty and unreliable even before the eruption of
political crisis in 1990, and the civil war has only aggravated the situation. Somalia had carried
out a census with UNFPA assistance in 1975, but only a brief analytical report appears to have
been published. Ten years later Somalia carried out another census in 1985-86, but the results
were not released, since the government felt that the census had underestimated the population.
The results were never evaluated to throw light on the extent of errors in the census.
Consequently even the population count based on the census is subject to considerable doubt. If
census taking had been jinxed in Somalia, sample surveys fared no better. A national
demographic survey was carried out in 1980-81, but the data were never fully processed and
only a few hand tabulations were released. Another survey of three regions (Banadir, Bay and
Lower Shabelle) was carried out in 1981 in collaboration with the University of North Carolina,
Chapel Hill, and this survey has remained the only source of demographic parameters.
The traditional approaches to the estimation of demographic parameters are vital registration,
censuses and surveys. Vital registration is non-existent in Somalia and as noted earlier census
taking has not been successful in the past. The absence of a central authority for the collection
and compilation of data precludes the undertaking of a Census in the present circumstances. The
frequent eruption of fighting complicates the situation further. Therefore the only feasible
approach in the Somalia context is the undertaking of demographic surveys in a phased manner,
starting with the areas where the security and logistics will permit the undertaking of such
surveys. Sample surveys have the merit that they can be better monitored and supervised than
censuses resulting in better quality of data. A whole range of survey methods and techniques of
analysis are presently available for estimation of demographic parameters by indirect methods.
Some priority areas for inclusion in a survey of this nature have been identified through
discussions with professionals of UNDOS, various international agencies and NGOs. Every
question included in a survey has a cost in terms of the money spent in collecting and
processing the data, and also in terms of the quality of data collected, the longer the
questionnaire, the poorer the quality. The availability of new data processing softwares such as
IMPS and EISA has made the processing of survey data lot faster and easier and the survey
results could be brought out within six months after the completion of the survey.

Sample Design
Before the civil war Somalia was divided into 18 regions (gobols), further subdivided into 102
are in turn divided into departments (waaxdas). In large cities such as Mogadishu, the waaxdas
are further subdivided into neighbourhoods (tabeelas).
In principle, a two-stage cluster sampling is recommended for the survey with probability
proportional to size (PPS) selection at the first stage, the waahdas or tabeelas being used as the
primary sampling unit. In practice, this is going to be difficult since we do not have the number
of households in each primary sampling unit. Where data are available from earlier censuses or
surveys these could be utilized, but these are likely to be incorrect because of the major
displacement of population that has taken place. In Mogadishu, the tabeelas could be used as
the first stage sampling units, while in the other urban areas the waahdas could be the first stage
sampling units. The tabeelas are supposed to have approximately 100 households each. Where
the tabeelas are too large, they could be split into two or three units denoted by a, b, c... Where
 2

a tabeela is too small it could be combined with a neighbouring one. A listing of these units
will constitute the sampling frame of PSUs. According to a survey carried out in 1984, there
were 884 tabeelas in Mogadishu. At present the number of PSUs after the adjustments
mentioned above could be 1000 units. Out of these simple random sampling or systematic
sampling with a random start could select a sample of 50 tabeelas. From each of these selected
PSUs a cluster of 30 households could be selected at random. This will yield a sample of 1500
households for Mogadishu.
For other regions, it would suffice to have 20 clusters of 30 households for the urban areas of
each region, yielding a sample of 600 urban households for each region.
For rural areas of each region, the settlements identified through the settlement survey could be
grouped into three broad groupings, namely, small, medium and large villages according to their
approximate population size on the basis of a priority information or judgement. From each of
these strata 10 villages could be selected at random as the first stage units, and from each of the
selected villages a cluster of 20 households could be selected at random. This would yield a
sample of 600 rural households for each region.
For the nomadic population, the most practical approach will be to include them whenever they
are found as a family, since nomadic men have the tendency to leave their families, and move
with their camels to the water points. If only the women are found in their tents, inquiry should
be made about their menfolk and they should be included in the households even if they were
not present at the time of the visit. Previous studies have shown that censuses taken at
waterpoints result in undercounts of women and children. It should be ensured that at least 300
nomadic households are included for each region. If the number is not reached, further efforts
should be made to locate their whereabouts, and include them.
To sum up, the above design will result in a sample of 1500 households for each region
comprising 600 urban households and 600 rural settled households and 300 nomadic
households. In the case of Mogadishu the entire sample of 1500 households will be urban
households. Under normal circumstances the sample will include 1000 ever-married women in
the reproductive age groups in each region, 500 in urban and 500 in rural settled population.
However some of the earlier surveys in Somalia have shown that ever-married women tend to
be under reported. In this case the number of households covered should be suitably increased
to ensure that the sample includes the required number of ever-married women in the rural and
urban samples. For nomadic population, the aim should be to include at least 200 ever married
women for each region.
Under normal conditions a team of ten enumerators and two supervisors could complete the
survey of 600 households in two weeks. If the same team is utilized for rural and urban areas,
the survey could be completed in about a month, provided transport and other logistics are taken
care of. The supervisors and enumerators should be given a week's training, including both
theoretical and practical lessons on the conduct of the survey, including mock interviews and
enumeration of two or three households who will not be included in the sample. The
questionnaires filled during the training period should be reviewed thoroughly to find out if
women and children may have been omitted, whether any of the questions have been missed,
and whether the responses are consistent. Various validity checks should be provided in order
to reduce the non-sampling errors. A list of historical events in each area should be provided in
order to compute age of the household members. In the event of non-response from a
household, it is preferable for the enumerator to proceed to the next household until the required
number of households is completed.
Questionnaire
Sex and age: Sex and age are the key variables in all demographic surveys, since population by
sex and age constitute the denominators in the calculation of rates and are required for various
analysis, including gender analysis. Classification of data by sex and age enables the evaluation
of the data through comparisons between the sexes in key parameters. Other uses of sex-age
 3

classifications include application of stable or quasi stable population models for the estimation
of demographic parameters, estimation of child mortality and fertility patterns through reverse
survival and "own children" methods, estimation of adult mortality through the growth balance
method and for detecting the extent of omissions in adult death registration. However, age is
the most difficult variable to get accurately in any survey in developing countries, and the
enumerators should be trained well to get the correct age. Since most respondents do not know
their dates of birth, age should be asked and recorded in completed years. When the respondent
is unable to state his/her age, he/she should be helped to work out the age in relation to a known
historical event or in relation to the age of another person whose age is known. If the person
has children, asking the age of the parent when the children were born, and the current age of
the parent could be estimated could arrive at the age of the parents. The enumerator should aim
at getting the age as close to reality as possible.
Relationship: A question on relationship is included in the survey for three reasons. First, by
asking this question for each person, it is possible to detect if any member of the household may
have been omitted, or forgotten. Secondly, by relating the children to their parents living in the
same household, we are able to estimate fertility patterns by the "own children" method.
Thirdly, the information on relationship could be utilized to classify households into nuclear,
extended and other type of households. The enumerator should start with the head of
household, spouse of head, sons/daughters of the head, brothers/ sisters of the head or spouse of
head, parents of head/spouse of head, grandchildren of head/spouse of head, other relatives,
non-relatives in that order.
Marital Status: Current marital status is important for the analysis of nuptiality and fertility. The
Singulate Mean Age at Marriage (SMAM) could be calculated from the percentage of single
persons in successive age groups, and proportion widowed could be utilized to estimate adult
mortality under certain conditions. Marital status is also a control variable in fertility models.
In Somalia, marriage and divorce are easier than in most developing societies. Polygamy is
widely prevalent among men. About one third of marriages end in divorce, and over 60 percent
of women in the age group 45-49 have been married more than once. Another recent
phenomenon is "abandonment" of the wives by their husbands, the magnitude of which is
unknown. Marital status should be recorded in terms of the following categories – "never
married", "currently married", "widowed", "divorced" and "abandoned" respectively.
Place of Birth and Place of Previous Residence: A question on place of birth is included in the
survey to elicit information about lifetime migration through a classification of the region of
birth with the region of enumeration. The respondent will be asked the place of birth of the
members of the household, and the region of birth will be recorded if the person was born
within Somalia, and the country of birth if the person was born outside Somalia. This question
will not tell about the timing of migration or the number of moves involved. Nevertheless, this
question would enable us to get a picture of the major migration flows in the past.
The question on previous residence and duration of residence will provide the data for
estimating recent migration.
Education: Education is a key variable influencing fertility, mortality and migration, besides
educational characteristics of the population are an important indicator of the socioeconomic
status of the population. Koranic schools play an important role in education in Somalia. A
question is included in the questionnaire to find out if the person is illiterate or literate without
any education whether he/she is attending a koranic or modern school, and the highest grade
completed. Three questions are included in the questionnaire pertaining to education. The
enumerator should ask whether the person could read and write, whether the person has
attended or attending school, whether it is a koranic or modern school that the person has
attended or attending, and what level and grade the person has attended or attending. He will
also ask about the highest grade completed by the members of the household. The enumerator
should receive sufficient training to ask these questions in proper sequence and then record the
information in the questionnaire.
 4

Economic characteristics: Like education, economic activity is cross-classification variable for

examining the differentials in fertility and mortality, besides providing information on the
importance of different occupations in the society. Three questions have been included to
classify the population according to employed, unemployed and those not in the labour force.
The employed will be further classified according to the most important occupations. As in the
case of education, the enumerator will have to ask more than one question before recording the
answer in the questionnaire. Every person irrespective of sex should be first asked whether
he/she is working for pay, profit or family gain. If the person is not working for pay, profit or
family gain, he/she should be asked whether he/she is looking for work, or accepts work if
offered. Those answering in the affirmative should be classified as unemployed. Those who
are not classified as employed, nor unemployed should be further asked whether they are
students, housewives not working for family gain, retired, disabled, retired or other
economically inactive category. For the employed persons, occupation and employment status
should be ascertained.
Paternal/Maternal Orphanhood: Two questions on whether the person's father is alive and
mother is alive have been included in order to provide the basis for estimation of adult male and
female mortality. These questions are fairly simple to ask and have yielded good results in
many developing countries. Technical procedures developed by the Hill and Trussell makes it
possible to estimate survival probabilities from age 25, or to estimate survival probabilities from
birth by including an estimate of child mortality.
Age at first Marriage/ Duration of Marriage: In many developing countries, duration since the
first marriage has proved to be a better classificatory variable than age for indirect estimation of
fertility and mortality. This is particularly appropriate for societies such as that of Somalia
where there is little cohabitation outside marriage. However, there are difficulties in obtaining
correct information on age at first marriage and duration of marriage. These two items should
be protested in some areas before a decision is taken about the choice. Asking a direct question
on duration since first marriage is preferable to asking the age at first marriage and taking the
difference between current age and age at first marriage. Experience in many countries shows
that the duration of marriage estimated in this manner is likely to be vitiated by errors in the
reporting of current age as well the age at first marriage, and it is difficult to foresee the
direction of this error. The high rates of divorce and remarriage in the Somali society makes it
extremely difficult to obtain correct information on duration since first marriage as well as age
at first marriage. Enumerators should be trained how to get the duration since first marriage by
referring to some historical events around the time of first marriage, or by reference to current
age of the first child, and how long after the marriage the child was born, or by reference to the
duration of marriage of another person who may have married about the same time.
Children ever born and Children surviving: These questions have been included in the
questionnaire to provide the basis for indirect estimation of fertility as well as infant and child
mortality. Experience in most developing countries has shown that these questions are fairly
simple to ask and also yield good results. The estimates of fertility and child mortality for
Somalia available to us are based on these questions in the surveys carried out during 1980-81
and 1984. In Somali society, as elsewhere, a grandmother or other relative may bring up some
children who was not living in the household at the time of the survey. Therefore, questions
should be asked about the children born alive to her, children who are still living with her, and
the children living elsewhere. The sex of the children should also be asked in order to classify
the number of children born alive and living children according to sex, since mortality estimates
are made separately for males and females. Fertility can be estimated by applying a fertility
model to the average number of children ever born per woman, classified according to age
groups or duration of marriage. Similarly, the proportion dead among the children born to
mothers classified according to age groups or duration of marriage can be utilized for the
estimation of infant and child mortality. The enumerators should be trained to apply some
consistency checks to the information obtained, and to detect errors in the data, so that the data
could be corrected in the field. For example, children ever born should be greater than the
 5

number of children who are living at the time of the survey. Male and female children should
add up to the total children, and children living in the household and the children living
elsewhere should add up to total number of living children.
Date of Birth of Most Recent Livebirth: A question on date of birth of the most recent livebirth
is included in order to obtain information on the pattern of current fertility. What is of interest
to us is the number of births during the 12 months preceding the survey. Experience from
surveys around the world has shown that the question on the last livebirth is preferable to a
question on the births during the preceding 12 months. Another way the question can be asked
is "How many births have occurred in the household since the last Ramadhan or Idd? In this
case we may not get the number of births during the preceding 12 months, but for a shorter or
longer duration. When the question is asked about the most recent livebirth, the births during
the preceding 12 months could be utilized, and the births that occurred earlier than the 12
months prior to the survey could be ignored. If the number of births occurring to mothers of
different age groups is understated to the same extent, the pattern of fertility will be still close to
reality. Techniques developed by William Brass enables the adjustment of the age pattern of
fertility by comparison with the average number of children ever born by age group of mothers
reported in the survey. Another approach is to utilize the average parity by duration of marriage
to derive the adjustment factor. Since births are likely to be omitted in surveys, the enumerator
will have to probe in order to get the missing events. Especially children born alive who die
within a short time, are not likely to be reported. Such cases should be identified and recorded
as livebirths as well as infant deaths, failure to do so will result in the underestimation of both
fertility rate as well as infant death rate.
Part 2 of the Questionnaire: This section aims to classify the households according to the type
of house, ownership of the house, type of settler and type of settlement.
Part 3 of the Questionnaire: This section seeks information on members of the household who
may have migrated, their destination and the reason for migration. Likewise, information is
sought on the persons who may have moved into the household, their previous location, date
and reason for migration. These questions can be utilized to derive estimates of in- and out-
migration. Unlike the question on birthplace which provides estimates of lifetime migration,
this question is expected to provide estimates of migration in the recent past.
Deaths in the household during the 12 months preceding the survey will also be recorded, along
with the sex and age of the deceased person. Generally there are greater omissions of deaths
than births in surveys, moreover respondents have difficulty remembering the timing of deaths.
However, "the Growth Balance Method" of William Brass provides a means to check the extent
of completeness of deaths by comparison of the distribution of deaths by age with the
distribution of population by age. The question on deaths during the previous 12 months
generally yields good estimates of mortality after childhood, and it is hoped that this will be the
case for Somalia as well.
Part 4 of the Questionnaire: Module on Reproductive Health Statistics: A module on
Reproductive Health will be canvassed in a sub-sample of the households covered by the main
survey. This module will include questions on pregnancy history, open and closed birth
intervals, antenatal and postnatal care of mothers, duration of breast feeding, duration of
lactation amenorrhoea, prevalence of contraception, and immunization coverage of children
under 5 years of age.
Recommended Tabulations
The following minimum tabulations of the survey data are recommended:
1. Population by five year age groups (0-4, 5-9, 10-14, 15-19,...70-74, 75-79 and 80 years
and over). If the number of persons of 70-74, 75-79 and 80+ are small, then 70+ may
be made the terminal age group.
2. Population by single years of age and sex up to 79 and 80+.
 6

3. Population by five-year age groups and sex according to marital status.

4. Population by five-year age groups, sex and educational status.
5. Population by five-year age groups, sex and occupational status.
6. Ever married female population 15 years and over by five-year age groups, total
children ever born and total children surviving by sex of the children.
7. Ever married female population 15 years and over by five year duration group (0-4,
5-9,...20-24 and 25+), total children ever born and total children surviving by sex of
the children.
8. Ever married female population 15 years and over by five-year age group and number
of children ever born (0,1,2,...11,12+) by sex of the children.
9. Ever married female population 15 years and over by five-year age group and number
of children surviving (0,1,2,...11, 12+) by sex of the children.
10. Ever married female population 15 years and over by five-year duration group (0-4,
5-9,...20-24, 25+) and number of children ever born (0,1,2,…11, 12+) by sex of
children.
11. Ever married female population 15 years and over by five-year duration group (0-4,
5-9,...20-24,25+) and number of surviving children (0,1,2,...11,12+) by sex of children.
12. Ever married female population 15 years and over by five-year age group, number of
children ever born (0,1,2,…11,12+) and number of surviving children (0,1,2...11, 12+)
by sex of children.
13. Ever married female population 15 years and over by duration group and number of
children ever born (0,1,2,...11.12+) and the number of surviving children (0,1,2,...11,
12+) by sex of children.
14. Ever married female population 15 years and over by five-year age group and five-year
duration group.
15. Population by five-year age group, sex, region of enumeration and region or country of
birth (lifetime migration).
16. Population by five-year age group, sex, region of enumeration and region or country of
previous residence (recent migration).
17. "Own children" as identified from the relationship column by single years of age
(0,1,2,...15) by single years of age of mother (15,16,...49).
18. Number of livebirths in the 12 months preceding the survey and the deaths among them
by sex and five-year age groups of mothers (15-19, 20-24,...45-49).
19. Number of livebirths in the 12 months preceding the survey and the deaths among them
by sex and five year duration of marriage of mothers (0-4, 5-9,...20-24 and 25+).
20. Infant Deaths among the livebirths in the 12 months preceding the survey by age at
death in months (less than one month, 1-2 months, 2-3 months,...11-12 months) by age
group of mother (15-19, 20-24,...45-49) and sex of the child.
21. Population by five-year age group, sex, and whether father is alive or not.
22. Population by five-year age group, sex, and whether mother is alive or not.
23. Deaths in the 12 months before the survey by five-year age groups (0, 1-4,
5-9,...65-69, 70+) and sex of the deceased.
24. Deaths in the 12 months before the survey by broad causes of death and sex.
25. Disabled population by sex and five-year age group and nature of disability.
 7

26. Number of mothers according to the interval since last livebirth (in months).
27. Number of mothers according to the interval between the last livebirth and the previous
one (in months).
28. Number of mothers classified by age groups and the number who received antenatal,
postnatal care and TT.
29. Number of children under 5 years of age who have received immunization by type of
immunization.
30. Number of mothers by age group who knew about contraception, ever used
contraception and currently using contraception.

Concluding Remarks
The Demographic Survey of Somalia proposed here opens the possibilities of application of
several new tools of demographic estimation. At the same time it offers great flexibility in
survey design and implementation. Unlike a census which is a snapshot of population at a given
point of time, this survey can be carried out in a phased manner in the different regions taking
into account the security situation, availability of transport, manpower, etc. It is preferable if
the survey could be carried out within a short time (two to three weeks) in a given region by
employing adequate number of enumerators and supervisors. While the enumeration of
sedentary population may not pose any problems, special efforts may have been devoted for
locating nomadic households, and enumerating them.
The choice of a pre-coded questionnaire makes the recording of information easier. However,
there is the risk that the enumerator may tick the answers without a proper inquiry. It is the duty
of the supervisors to check every response and make sure that they are valid responses. The
supervisors should be trained to detect errors through internal consistency checks. He should
discuss these errors with the enumerator and correct them. Also it would be helpful if the
supervisor could contact the elders in the area and find out if any births or deaths have taken
place in the households covered by the survey during the preceding 12 months. This
information could be utilized to detect omissions of births and deaths in these households and
include them.
New data processing software such as Integrated Microcomputer Processing System (IMPS)
developed by the US Bureau of Census makes the data processing lot easier and faster. This
software has several modules for data entry, editing, quick tabulation and a more elaborate
tabulation. Also it has the facility that data entry screen can be in English or in Somali
language. A consultant may be hired for two weeks to train the data entry staff in the use of this
software. Once the tabulations for a region are completed, the data should be published
immediately, without waiting for a detailed analysis to be carried out.
The following analyses are recommended:
Fertility
1. A fertility model can be fitted to the average number of children ever born for the most
reliable age groups of women (normally 20-24, 25-29, and 30-34) and the conventional age
specific fertility rates can be derived from the model. This is based on the assumption that
there has been no major change in fertility. This assumption may be valid for Somalia.
2. A fertility model can be fitted to the average number of children ever born by five-year
duration groups of women and the model can be used to estimate the age specific fertility
rates. This model is applicable when there is no childbirth outside marriage and there is no
fertility control (as in the case of Somalia). However, in Somalia there is a great risk of
duration of marriage being misreported because of the high frequency of divorce and
remarriage (about 60 percent of women 45-49 have been married more than once).
 8

3. If children in the household can be assigned to their mothers by relationship information

and an estimate of child mortality is available, "the own children method" can be applied to
the data on the number of children under age 15 to derive estimates of births and fertility
patterns year by year in the recent past.
4. Age pattern of fertility can be calculated from the births during the 12 months preceding the
survey, and adjusted with reference to the average parity by age group of mother or by
duration of marriage of mothers. This will enable the estimation of the extent of under
reporting of livebirths during the preceding 12 months in the survey.
5. Cross tabulation of births in the preceding 12 months by age of mother and by number of
children ever born allows the study of birth order specific fertility rates.
6. Although this is a questionable assumption in the Somali situation, stable or quasi stable
population models can be applied to the sex and age structure of the population at the
national level from the survey to obtain estimates of fertility (and mortality) and compare
these estimates with those obtained by other methods.
Mortality
1. The level of infant and child mortality can be estimated by indirect method utilizing the data
on the proportion dead among the children ever born by age group of mother, and the age
pattern of fertility derived earlier.
2. The level of infant and child mortality can be estimated from the proportion of children ever
born by duration of marriage group of mother on the assumption that there is no
cohabitation outside marriage, and there is little fertility control and the data on duration are
fairly correct.
3. The age distribution of population along with the age distribution of deaths obtained from
the survey could be utilized to estimate the extent of completeness in the reporting of deaths
in the survey, and the number of deaths above age 5 can be adjusted by the "growth balance
method". This method generally gives good estimates of mortality after childhood.
4. The proportions of population whose father/ mother is alive can be utilized to estimate
mortality among males/ females above age 25, or by including an estimate of child mortality
the male/ female survival probabilities from birth can be estimated. It has been observed
that the estimates of female mortality derived from maternal orphanhood are more reliable
than the estimate of male mortality derived from male orphanhood. This needs to be tested
in the case of Somalia.
5. As in the case of fertility analysis, stable or quasi-stable population models can be applied
to the age and sex structure of the population from the survey to estimate the levels of
mortality, although the assumption of stability may not hold good for Somalia.
6. A life table may be constructed by linking the infant, child and adult mortality estimates
derived by the above methods for the urban, rural and nomadic populations and the various
life table functions such as expectancy of life at birth can be read off from the life table.
Migration
1. The place of birth data can be utilized to estimate the number of lifetime in-migrants to a
given region according to the regions and countries of their origin. When the survey is
carried out in all regions, it will be possible to estimate lifetime out-migration from each
region to other destination regions. Likewise, the net migration to each region during the
preceding 5 years could be estimated from the data on previous residence and duration of
residence.
From the question on the members of the household who had moved out, it is possible to
estimate recent out-migration, and from the persons found in the household who are not usual
members of the household, the number of recent in-migrants can be estimated.
 9

Demographic and Social Characteristics

1. The survey will provide estimates of age distribution by sex which is vital for most
demographic analysis.
2. Estimates of distribution of population according to marital status can be used in the
analysis of nuptiality patterns, and the application of nuptiality models, besides yielding
estimates of parameters such as mean age at marriage, mean duration of child bearing etc.
3. The survey will provide the sex-age distribution of population by educational status which
is important in the analysis of differentials in fertility, mortality and migration, and in
projections of school enrollment etc.
4. The survey will provide the sex-age distribution of population by occupation, which is
again important in the analysis of fertility, mortality and migration, and in labour supply
projections.
5. The survey is expected to provide regionwise estimates of demographic parameters for the
urban, rural settled and nomadic populations and their demographic characteristics. Similar
estimates will be produced for other special groups such as internally displaced persons
(IDPs) and refugees in each region.
The above analysis calls for specialized skills. Therefore, the services of an experienced
demographer are needed to carry out the analysis, derive estimates of demographic parameters
and prepare an analytical report.
 10

Annex 3

PROPOSAL FOR A SENTINEL REPORTING SYSTEM

FOR REPRODUCTIVE HEALTH AND VITAL STATISTICS FOR SOMALIA
by K.E.Vaidyanathan
UNFPA Consultant of Population Statistics

Introduction
The traditional approach to the collection of data on vital events is through a statutory vital
registration system covering the whole country. Unfortunately this is not possible in a country
which has been plagued by a civil war, and consequent disruption of administrative machinery.
Alternatively, vital rates could be estimated by indirect methods through censuses and surveys,
but these will not provide current data on fertility and mortality on a continuing basis. In the
best of times, censuses are taken once in ten years, and demographic surveys are carried out
once every four or five years. Moreover, censuses and surveys provide estimates at the
aggregate (national or regional) levels and not at the level of a village or local area. For many
purposes we need data at the sub-regional levels, in order to make an assessment of health
problems, plan health interventions and to evaluate the impact of programmes. Such data are
required on a continuing basis in order to monitor the programme implementation and changes
in the health situation. Regrettably, it is not feasible to introduce a data gathering system for the
entire Somalia because of the absence of the necessary infrastructure to carry out such an
operation. The least we can do is to introduce such a system in a few areas where the necessary
infrastructure exists in one form or another. Such areas may be regarded as Sentinel areas,
which can provide an indication of the health conditions in the surrounding hinterland, if not the
entire region where such areas are located. This proposal gives an outline of such a system,
which could be attempted in a few areas in the beginning, evaluated and improved upon and
gradually expanded to other areas. This proposal takes into account the specific conditions
prevalent in Somalia.
In Somali society, the elders (variously known as odayal, Guurti or Ddayasha in different
regions) have great influence in the village community. They represent the village or group of
villages (beels) in political decision making, settle disputes between individuals and sub-clans,
and they are the persons to reckon with in any projects involving international agencies. In
addition, most if not all villages has a village council and a village health committee. Every
village has at least one community health worker (CHW) and one traditional birth attendant
(TBA). The larger villages have 2 or 3 CHWs and more than one TBA. The CHWs and TBAs
are both voluntary workers who get remunerated by the community and play a key role in the
health services in Somalia. Where the public health services are functioning (as in the case of
the North West Zone, known as Somaliland), there is an official machinery to supervise the
CHWs and TBAs. At the regional level there is the Regional Health Officer (RHO) assisted by
MCH staff (who are qualified nurses). The 6 regions (previously 5) of Somaliland have 18
districts, and each district has one or more MCH Centres with a PHC supervisor. In
Somaliland, there are 40 MCH Centres and 125 Health Posts (HPs), each with one CHW and
one or more TBAs. This was the pattern of health administration in entire Somalia before the
war. However in most other regions of Somalia, the public health administration which was
disrupted during the civil war has not been restored. Public health services in most parts of
Somalia are presently run by international and local NGOs, although the institution of CHWs
and TBAs continues to exist sustained by the community.
In the Somaliland area, the CHWs are literate persons with education up to 6th standard, and all
have received CHW training. The TBAs are not literate, but have knowledge of mother and
childcare from experience and from training. Over 90 percent of the TBAs have been trained
through the assistance of UNICEF and NGOs. The TBAs are provided a form prescribed by
 11

WHO, and the CHW has been provided a form which includes the information provided by the
TBA. The TBA form has pictures representing eight items such as pregnancies, deliveries,
referrals etc, and is easy to fill.
The CHW includes these and the other information required from her and sends physically takes
the information to the PHC supervisor, who in turn sends it to the RHO, and onward to the HIS
unit in the Ministry of Health in Hargeisa. The Consultant found the forms well-filled and neat.
However, the forms for some regions like Sool and Sanag are not received on time due to
communication problems. These forms are not consolidated and compiled to draw conclusions
on the state of health in the region. The Director General of Health in Somaliland agreed that
the data are not compiled because of lack of staff and computer facilities, also the staff of the
HIS (only 2) lacked the training.
In the remaining regions where the NGOs are providing assistance to health services, more or
less similar systems of recording exist with varying degrees of efficiency. ADRA Somalia and
World Vision have more elaborate forms for collecting health information, and these are
received by the respective organizations. Probably the NGOs exercise better supervision of the
CHWs and TBAs than the health officials in the NW Region, because of the better logistical
support they enjoy. Therefore, any system of data collection at the micro level can be
implemented only through the administrative machinery in the NW region and through the
NGOs in the other regions. The Sentinel Reporting System proposed here involves the selection
of a few sentinel areas in each region where the efforts could be concentrated to ensure accuracy
and completeness of data through more frequent visits of supervisors for checking. The areas
selected should have capable and motivated CHWs, functioning health committees and
supervisory staff either from the official machinery or NGOs. In such areas it is feasible to
establish a system of data collection which is cost-effective and will provide the data on vital
statistics required for decision making.

Objectives of a Sentinel Reporting System

1. To provide data on population, births, deaths and health conditions on a continuing basis for
a few sentinel areas where there is a programme run by NGOs.
2. To provide a field site for the NGOs to carry out an assessment of the acceptability and
effectiveness of their health interventions (e.g. nutrition, immunization, health education,
oral therapy, contraception etc), leading to more effective programme strategies.
3. In the long run, to provide the planners, international agencies and the NGOs an indication
of the changes in demographic parameters and health conditions in the different sentinel
areas of Somalia.

Method of Data Collection

The collection of data under the SRS will be the responsibility of the Community Health
Workers (CHWs) with the assistance of the TBAs in their areas. They will be supervised by the
PHC supervisors in the North West Zone and by the Medical Officers or Social Workers of the
NGOs in the areas where the NGOs are having a strong presence in the other regions. The
CHWs will maintain a Household Register where they will record details of the members of the
household, as well as any events – births, deaths, marriages, divorces, and changes in the
membership of the household during the course of the year. In addition to the Household
Register, she will be expected to complete the Birth Report for each individual birth and Death
Report for each death as and when they occur. While the listing of household members is done
only once a year, the recording of births and death and changes in the households needs to be
done every month. The CHWs should make frequent enquiries and check the Household
Register and the Birth Report and Death Report and ensure that no person or event is missed. A
CHW can cover between 300 and 400 households in a month, therefore the area assigned to a
 12

CHW should not exceed 400 households. If the area has more than 400 households, it should be
assigned to two CHWs with clearly demarcated jurisdiction. While the TBA will not be able to
fill the forms, they are capable of providing the information to the CHWs especially about
pregnancies, livebirths and stillbirths. Since each CHW supervises one or two TBAs, she
should meet the TBA at least once a week and discuss the work of the TBA, review the
completeness of the registration and to record the vital events in the prescribed forms. The
CHW should establish contact with the village elders, school teachers, village midwifes, imams
or sheiks or other persons in the area to find out if any births, deaths, marriages and divorces
have taken place during the month. A list of "informants" in the area should be identified at the
beginning of the project and their cooperation enlisted to give information on vital events. This
should be done in an informal manner, without a questionnaire or reporting forms. When she
learns about any birth or death which was not already recorded she should visit the household,
make enquiries from the family, gather the relevant information and include them in the
household register and birth and death reports. The PHC Supervisor in the North West Zone
and the doctor or social worker of the NGO in the other regions should visit the area at least
once a month, and further check the Household Register and the Birth and Death Reports.
He/she should also meet with the village health committee and the elders and discuss the health
situation in the area and anything that might have been overlooked by the CHW. These reports
could be collected by the PHC Supervisor or Medical Officer of the NGO and passed on to the
RHO in the North West Zone and to the NGO headquarters each month. These reports will be
compiled and reviewed by the RHO/ NGO for an evaluation of their activities, and inclusion in
their quarterly and annual reports.
There are two ways UNDOS would help to compile the data and bring out an analysis of the
health situation. One way will be to produce two copies of the CHW's forms for the sentinel
areas and one copy could be collected by the RHO or NGO and sent to UNDOS. The other
approach could that the RHO or NGO compiles the information for their sentinel areas, and
these compiled data could be sent to UNDOS for consolidation. The third approach is for
UNDOS to send a person to each of these areas to collect the forms. In fact all the three
approaches could be tried, and for different areas different approaches could be adopted.

Household Register
The Household Register will record information on the name and occupation of head of
household, address of the household, names of members of the household, their sex, age, marital
status. There will be provision in the Household Register to record any changes in the
household such as births, deaths, marriages, divorces, in and out migration. The Household
Register should be updated at least once each year, preferably during the last week of December
or first week of January so that the register will reflect the population as of 1 January each year.
Since there are considerable variations in population between the dry season and wet season, it
will be preferable that the Household Register is updated twice a year, but to start with this
should be done at least once a year. The Household Register will provide the numerators
(number of births and deaths) and denominators (population) for the calculation of vital rates.
The Household Register will be maintained by the CHW in the health post or in her home,
whichever is agreed upon by her supervisor, namely the PHC Supervisor or NGO Medical
Officer.

Birth Report
The details of every birth recorded in the Household Register should be included in the Birth
Report, and further details recorded. The items to be included in the Birth Report are the name
of the mother, her age, date of occurrence of the birth, type of birth, sex of the child and
attendance at birth. In the past birth ceremonies were great events in the Somali villages, so the
TBAs and CHWs will not miss it. However due to the difficult circumstances, these ceremonies
 13

are becoming more private, and mothers can give birth to babies with anyone taking note of it.
Therefore it will be advantageous to include an incentive for registration of the births.
Whenever a mother comes to register the birth, the TBA could give the mother vitamin and folic
acid tablets, some sugar and salt to make to make ORS in the event of diarrhea, or a packet of
maize meal. The distribution of these could be linked to the immunization of the child, so that
there is no misuse of these incentives. In the difficult circumstances facing Somalia, there is a
great potential for misuse of incentives, therefore the PHC Supervisor and the NGO Medical
Officer should check if the immunization records and the birth records match with the
distribution of incentives.

Death Report
The details of every death recorded in the Household Register should be included in the Death
Report along with other additional information. The items included in the Death Report are date
of death, name of the deceased, sex, age, marital status of the deceased, cause of death and
whether the deceased received any medical attention prior to the death. Death ceremonies are
still common in Somali society, and are usually performed by an imam or sheikh. Therefore the
CHW could obtain the information from the imam or sheikh and follow up this lead with a visit
to the household to obtain the required information. Information on causes of deaths is bound to
be incorrect in the conditions of Somalia, however the inaccuracies could be minimized by
training the CHW in the identification of the causes by their symptoms. In any case they should
be able to identify pregnancy-related deaths among women. Since we have no other source of
information on maternal mortality, the Sentinel Reporting System could be a major source of
data for estimating maternal mortality rate.

Processing of Data
While the CHWs will be responsible for the collection and recording of the data, HIS Unit of
the Ministry of Health in Somaliland could be made responsible for editing, verification, coding
and tabulation of the data for the NorthWest region (Somaliland). The Director General for
Health in Somaliland expressed that they could do this if some financial and technical assistance
is provided either by WHO or UNDOS. In the other regions, the data may be compiled by the
NGOs and included in their quarterly and annual reports. The data provided by the different
NGOs will be compiled and published by UNDOS with the collaboration of WHO Somalia. A
staff member of WHO Somalia could participate in the training of the TBAs and CHWs, and
also make periodic visits to see if the records are maintained properly, and give the feedback to
the NGOs concerned. If all the NGOs adopt a standard format for the Household Register, Birth
Reports and Death Reports and follow some standard tabulation, their consolidation by UNDOS
will be easier.

Output of the Sentinel Reporting System

If the sentinel areas are chosen to represent different regions and ethnic groups, we will have
estimates of vital rates and population change for each of the chosen areas. If the system works
for a few years, we will also have time series data on vital rates for these areas. Experience all
over the world indicates that intra-regional variations are far less than inter-regional variations.
The data could be cautiously interpreted to represent the conditions in different regions.
Another important output will be population size and their characteristics. In particular sex-age
composition of population is very important for policy making. The data will yield the
proportion of population in the childhood, young adolescent, working ages and elderly
population, and the proportion of women in the reproductive age groups.
 14

The data from SRS will also provide information on the extent of antenatal care and the
proportion of births and deaths taking place in institutions and those attended by trained
persons.
While the cause of death data from SRS may not be accurate, it could provide an indication of
proportion of deaths in different age groups, and about the extent of maternal deaths. If the data
are gathered for successive years we could also compute infant mortality rates from the
livebirths and the deaths occurring among them during a twelve month period.

Future Development of the System

There are three issues that are bound to arise, namely, coverage, continuity and sustainability.
To begin with the system should be tried in a few areas where the conditions are favourable to
its implementation and sufficient inputs are available from the authorities or NGOs
implementing the system. An important precondition is that the authorities and NGOs should
have the motivation to implement the system. We must also select areas which are reasonably
secure. If there were some quarrel once in a while this would not affect the system, so long as
the CHWs and TBAs remain in the area, and the doctors supervising the system are able to visit
the area once a month. The incentives for the establishment and continuation will have to be
worked out by UNDOS in consultation with WHO and the concerned NGOs. UNFPA may be
approached for limited funding for logistic support, training of CHWs, and the processing of
data.
Experience around the world shows that any system will work well only if there is feedback
from the user of the system. The principal user of the system in this case is the NGO concerned.
The NGO Medical Officer and Social Worker should discuss the Household Register and the
Birth and Death Registers with the TBA, the CHW and any other persons concerned with
health. Examples of such discussions are: what is the population of the village, how many
births and deaths occurred in the village in the last month, how do these compare with the
previous month, are there any stillbirths or abortions, what are the causes, how many of the
infants are keeping good health, how many died, and for what reasons this was allowed to
happen, what are the major causes of mortality in the area and how these could be prevented.
Somalis irrespective of their social status are highly articulate and communicative, and surely
the medical officer and social worker can learn as much from these discussions, as he could
provide the TBA and CHW by way of health education.
At the next level the users are the international agencies interested in the population and vital
rates for various purposes. At this stage UNDOS has the major role of compiling the
information, interpreting the data and disseminating them. WHO and UNICEF will be able to
use the information for planning interventions in the areas of reproductive health, nutrition, and
immunization.
The Household Register and Birth and Death Report forms should be kept very simple at the
beginning, and can be improved upon gradually with the experience gained during the coming
years. At the same time, the system can be expanded by the addition of new sentinel areas. The
selection of the areas for the sentinel reporting system is very important. We must choose the
areas where the CHWs and TBAs are well motivated, even if less qualified, and the NGO is able
to exercise the required supervision.
Formats of the Household Register and Birth and Death Report forms are given in the Annex.
 15

Annex 4
LIST OF PERSONS MET DURING THE MISSION

UNDP Somalia
1.Mr. Dominik Langenbacher, Resident Representative
2.Ms. Kristi Ragan, Senior Deputy Resident Representative
3.Mr. Vineet Bhatia, Assistant Resident Representative
4.Mr. Ali Yusuf Ahmed, Programme Officer for UNFPA projects
United Nations Development Office for Somalia (UNDOS)
1.Mr. K. N. S. Nair, Officer-in-Charge and Head, Economic Planning Unit
2.Mr. Gian Paolo Aloi, Head of LAS Unit
3.Mr. Amar Setia, Specialist in Administration
4.Mr. Giorgio Sartori, GIS Specialist
5.Ms. Mariam Alwi, Hussein, Economic Planner
6.Ms. Farah Dar, Health and Nutrition Specialist
7.Ms. Sza-Sza Zelleke, Documentation Specialist
WHO Somalia
1.Dr. Robert Hagan, WHO Representative
2.Dr. Sanaa Mustafa, Medical Coordinator, UNFPA Project
UNICEF Somalia
1.Mr. John Spring, Senior Programme Officer
2.Ms. Marianne Lindner, Monitoring and Evaluation Officer
3.Ms. Pirkko Heinonen, Health and Nutrition Officer
USAID
1.Mr. John Bierke, Director
2.Mr. Phil Steffen, FEWS/ Somalia Representative
WFP Somalia
1.Ms. Christine Cambrazy, Nutritionist
2.Mr. Ermino Sacco
European Community, Somalia Unit
1.Mr. Theo Kaspers
2.Ms. T. Paffenholz
3.Ms. Elizabeth Kiarie
UNHCR
1.Mr. E. Hounsouko, Cross Border Operations
UNESCO Somalia
1.Mr. Mark Richmond, Director
2.Mr. M. Devadoss, Programme Coordinator
ADRA-Somalia
1.Mr. John Ludlow
World Vision
1.Mr. Enzo Vacchio
Library of Congress, Nairobi
1.Mrs. Thomas, Director
Development Solutions Private Limited, Nairobi
1.Dr. Ron Schwarz, Director
Central Bureau of Statistics, Kenya
1.Mrs. Margaret Chemnge'tich, Director
2.Mr. Francis Munene, Demographer