Monitoring the socio-economic impacts of COVID-19
on Djiboutian households in Djibouti
Results from fourth wave of survey
(collected March 11-April 25)
October 2021
The report was prepared by Bilal Malaeb, Anne Duplantier and Romeo Jacky Gansey, from The World Bank; Sekou Tidani Konate and Omar Abdoulkader from the
Institute of Statistics of Djibouti (INSTAD).
The team acknowledges the efforts of the Director General of INSTAD, Mr Idriss Ali Soultan, and the team of the INSTAD in undertaking the data collection, and
would like to thank Djibouti’s Ministry of Social Affairs and Solidarity (MASS) for sharing the social registry data with INSTAD.
Monitoring the socio-economic impact of
Executive Summary
The fourth round of data collection on monitoring of socio-economic impacts of the COVID-19 pandemic in Djibouti
took place between March 11 and April 25, 2021, and was implemented over phone by the Institute of Statistics of
Djibouti (INSTAD). The fourth wave sample consists of 1,561 respondents, 1,122 of which are panel households
interviewed in wave 3, and 439 replacement households. Seven themes are tackled during this wave to understand
the trend of impacts of the COVID-19 crisis: economic activities, livelihoods, safety nets, access to basic goods, access
to services, food insecurity and gender.
Djiboutian breadwinners continue to return back to work, showing a consistent trend in employment relative to the
previous rounds of surveys. Indeed, 85 percent of households reported their breadwinner working the week before
the survey, compared to 58, 77, and 83 percent in the first, second, and third wave of data collection. Notably, 83
percent of breadwinners who worked before the survey reported working as usual, compared to 77 percent in the
third wave. In terms of sources of income, an increase in waged work and family business as a source of income is
observed, together with a drop in assistance from government as the COVID-19 social assistance program came to an
end in March 2021.
While Djibouti shows clear signs of recovery, some groups of the population may be lagging behind. For instance,
informal workers exhibit signs of precarity, as they have a higher propensity to work less than usual, and among those
a higher proportion receive no pay. Similarly among the female breadwinners who report working less than usual or
not at all (6 and 7 percent respectively), 66 percent report not receiving any pay. In fact, compared to the private
sector, public sector employees had a higher propensity to declare working as usual and receiving full pay. Indeed, 56
percent of households whose breadwinner works in the public sector reported having enough resources for the
following 30 days, compared to 32 percent among those whose breadwinner is in the private sector. Poor households,
as identified in the social registry, are also more likely to declare not having enough resources, compared to non-poor
household (41 and 30 percent respectively).
In a positive sign, nearly all households report having access to basic goods, including food and basic medicines, and
having access to healthcare when needed. Compared to the third wave, more households appear to have an adequate
food consumption score, reflecting adequate food frequency intake and dietary diversity. Furthermore, around 98
percent of both boys and girls attend schools. However, differences are observed among children who require school
catch-up activities. While 34 and 29 percent of girls and boys, respectively, are declared to need catch-up scholarly
activities, only 70 percent of girls participate in them when needed compared to 92 percent of boys.
In the fourth wave of this survey, a new module was added on gender, intra-household decision making, and time-use.
When household decisions are taken by a single household member, women tend to participate more than men in
decisions related to everyday purchases and healthcare of household members. Where more than one household
member is involved in making the decision, women participate in the decisions jointly with men in most of the cases,
but it is often less likely that only women make the decision. On time-use, women are more likely to spend time on
grocery shopping, domestic work, children’s studies, healthcare and leisure activities, than they are on income-
generating activities. A year after the onset of the COVID-19 in Djibouti, most respondents declared that COVID-19 had
not changed their life. This is particularly the case when it is one member who spends the most time undertaking the
activity. With regards to public safety, men report a higher likelihood of being a victim of crime and not feeling safe in
public spaces than women do, but women report a higher likelihood of experiencing domestic conflict.
As Djibouti had experienced an increase in the COVID-19 cases in March 2021, this survey also elicited respondents’
attitudes towards vaccines. Most respondents reported that they would accept to take an approved and free COVID-
19 vaccine. The main reasons for refusing a COVID-19 vaccine are worries about undesirable effects (for 31 percent of
the respondents) and the fact that respondents do not trust vaccines in general (23 percent). Around 10 percent of
the respondents would not accept to take the COVID-19 vaccine but would be more likely to take it if someone, such
as family, friends, religious leaders, recommends it. Respondents from poor households report a lower propensity to
accept the vaccine, but a higher likelihood to change their mind if someone recommended it.
2
Monitoring the socio-economic impact of
More than a year after the onset of the COVID-19 pandemic, Djibouti faced a new wave of COVID-19 between
INTRODUCTION
March and May 2021, with the highest number of cases registered ever in the country during the week of March
29 (1,260 confirmed cases, according to WHO). As of July 13th, 2021, 11,621 confirmed cases and 155 deaths of
COVID-19 were registered in Djibouti (WHO). Moreover, the vaccination campaign started with 26,796 vaccine
doses administered as of June 23th, 2021 (WHO). Since the end of the first wave and the lift of most of the
restrictive measures by the end of May 2020, the country has not initiated any further measures of
confinement. Nevertheless, the potential negative effects of the pandemic may have persisted. Indeed, the first
two waves of this survey revealed some of the negative effects the pandemic and its ensuing restrictive
measures may have had on households’ welfare, in particular in terms of breadwinners’ employment and
access to good and services. Despite an economic recovery that had been observed since the first wave, the
third wave highlighted the precarity of some households that may have been left behind.
The fourth wave of the this COVID-19 survey aimed to follow the households that had been previously
interviewed in the first three rounds, as well as a replacement sample. Seven themes are tackled during this
wave to understand the trend of impacts of the COVID-19 crisis: economic activities, livelihoods, safety nets,
access to basic goods, access to services, food insecurity and gender.
THE PHONE The fourth round of data collection on monitoring of socio-economic impacts of the COVID-19 pandemic took
place between March 11 and April 25, 2021, and was implemented over phone by the Institute of Statistics of
SURVEY Djibouti (INSTAD). This wave aimed to follow the households from the national sample that had been
interviewed in the first three rounds of data collection, as well as a replacement sub-sample. Information on
the households and breadwinners is provided by an adult respondent, randomly chosen household head and
spouse and distributed equally between male and female across households, allowing gender decomposition
of relevant personal data of the respondents by gender. The objective of this study is to identify trends in
economic activities and livelihoods, access to basic goods and services, food insecurity and safety nets. New to
this wave, a module on gender issues such as intra-household decision-making and time-use has been added.
As in the three previous waves, the sample of national households is drawn from the Ministry of Social Affairs
and Solidarity’s social registry (see Box 1 for information about the sampling strategy and the sampling
weights). The results are representative of the country’s urban population (except the top wealth quintile) and
can be disaggregated by location and poverty status.
The fourth wave sample consists of 1,561 respondents, 1,122 of which are panel households interviewed in
wave 31, and 439 replacement households (see Box 2 for the analysis of attrition and the composition of the
sample by panel status). The response rate of the whole sample stands at 71.8 percent (Table 2.1), with
variations across location and replacement status.
Table 2.1: Response rate to the survey
Number of Successful Interviews Response Rate (%)
Whole Sample 1,561 71.8
By Replacement Status
Panel (wave 3 to 4) 1,122 81.1
Replacement 439 55.6
By Location
Balbala 539 75.1
Rest of Djibouti City 527 71.6
Other Urban Areas 495 68.8
Source: Djibouti COVID-19 phone survey, 4th wave.
1 In previous waves of data collection, households who had not responded to any previous wave were not considered in subsequent
samples. In the fourth wave, households who were not reachable in wave 3 but were part of the first two waves were considered as part
of the sampling frame and are accounted for in the category “replacement” sub-sample of Table 2.1 to allow comparison of the panel sub-
sample with other waves.
3
Monitoring the socio-economic impact of
Around 59 percent of the respondents are female, 60 percent are the head of the household, and 47.5
percent are aged between 35 and 49 years old (Table 2.2). Around 81 percent of the households have a
breadwinner who is a member of the household and for 41 percent of them, the respondent is the breadwinner.
For those who are household member, breadwinners tend to be mainly male and household head (57 percent),
and the majority of them have 35 to 49 years old. For more details on weighting, see Box 1.
Table 2.2: Characteristics of respondents and breadwinners (%)
Respondent Breadwinner
Gender
Male 40.6 57.4
Female 59.4 24.3
Not a household member - 18.5
Age group
18-34 17.9 15.8
35-49 47.5 41.7
50-64 26.1 19.6
65+ 8.5 4.4
Not a household member - 18.5
Status in the household
Household head 60.5 56.9
Spouse 38.7 11.6
Other 0.8 13.2
Not a household member - 18.5
Observations 1,561 1,561
Source: Djibouti COVID-19 phone survey, 4th wave.
ECONOMIC More than a year after the onset of COVID-19, a large majority of households (85 percent) had their
breadwinner2 working the week before the 4th wave of the survey (Figure 3.1). Compared to previous waves,
ACTIVITIES this proportion continued to increase albeit at a slowing rate over time, showing a steady recovery of jobs in
Djibouti. The proportion of breadwinners who stopped working since COVID decreased from 22 percent in wave
1, in June 2020, to 5 percent in wave 4. Moreover, there is a small percentage of households whose
breadwinner exhibits dynamic changes in their working status across the waves, working in some waves but
not others. For example, in the 4th wave, 5 percent of the households had a breadwinner who was not working
anymore the week before the survey whereas they reported their breadwinner as working in a previous wave.
Figure 3.1: Working status of breadwinners (%)
2For 44 percent of the households, the breadwinner’s income represents all the household’s income. Moreover, households with a female
breadwinner are less likely to rely totally on her income than households with a male breadwinner (37 percent compared to 46 percent).
4
Monitoring the socio-economic impact of
100 5
7 6
90 20 7 6
10 4 5
80 6
70 22
60
50
40 83 85
77
30 58
20
10
0
Wave 1 Wave 2 Wave 3 Wave 4
Worked in any previous wave but not the week before the survey
Not working before covid and before survey
Working before covid but not before survey
Worked the week before the survey
Source: Authors’ calculation based on Djibouti COVID-19 phone survey, 1st, 2nd, 3rd and 4th waves.
Notes: Statistics are based on cross-sectional proportions and not only the longitudinal sample. Breadwinners are divided into four
categories: 1) those working in the week before the survey, 2) those working before COVID-19 but were not working in the week before
the survey, 3) those that were neither working before COVID-19 nor in the week before the survey, 4) those who worked in any previous
wave but were not working the week before the survey.
Half of the breadwinners work in the informal sector, and most of them work in small businesses, as daily
laborers or employees (Figure 3.2). However, there are important differences according to the gender of the
breadwinner. Female breadwinners are much more likely to work in the informal sector than male
breadwinners (73 percent and 42 percent, respectively). Women are also more likely to work in small
businesses (62 percent compared to 37 percent) or to be self-employed (53 percent and 13 percent,
respectively).
Figure 3.2: Employment characteristics of breadwinners who worked before the survey (%)
a. Sector b. Firm type c. Employment category
100 100 100
11 10 11
18
30 34 80 22
80 9 80 36 42
37
43 21
60 19 60 62 60
25
40 40 19 40 36
73 14 43
6 5
51 53
20 42 20 8 20
25 29 5 24
13 13
0 0 0
All Male Female All Male Female All Male Female
Public Public administration Public firm Employer Self-employed
Private formal Large private firm Small business Daily laborer Employee
Private informal Household Other/Don't know Other/Don't know
Source: Authors’ calculation based on Djibouti COVID-19 phone survey, 4th wave.
Notes: A small business is a sole proprietorship or cooperative; public firms are state owned enterprises. The category “female” refers to
households with a female breadwinner while “male” refers to households with a male breadwinner.
When asked about their change in workload, most breadwinners who worked the week before the survey
reported working as usual (Figure 3.3). The proportion of breadwinners who reported working as usual
increased from 53 percent in wave 1 to 83 percent in wave 4. Only slight differences are observed according to
5
Monitoring the socio-economic impact of
the gender of the breadwinner. Female breadwinners are more likely to work as usual than their male
counterparts (85 and 83 percent, respectively), and are slightly less likely than men to have not worked at all
(7 percent and 10 percent, respectively). There are still some differences according to the sector of employment
of the breadwinner. Breadwinners working in the public sector are more likely to have worked as usual (88
percent) than breadwinners from the private formal sector (84 percent) and from the private informal sector
(80 percent). The main reason for not working as usual is the reduction of working hours due to lack of activity
(cited by 40 percent of those who worked less than usual or not all).
Figure 3.3: Reported change of workload of breadwinners who worked the week before the survey (%)
100
6 9 8 8 7 9 10 7
90 11 7 1 6
9 6 8 6
80 19
70 31
60
50
83 80 84 88 83 85
40 77
73
30 53
20
10
0
Wave 1 Wave 2 Wave 3 Wave 4 Private Private Public Male Female
informal formal
More than usual As usual Less than usual Did not work Don't know
Source: Authors’ calculation based on Djibouti COVID-19 phone survey, 1st, 2nd, 3rd and 4th waves.
Notes: Statistics are based on cross-sectional proportions and not only the longitudinal sample. The distinction by sectors of employment
(public, formal, informal) concerns all the households whose breadwinner was working before the survey. The category “female” refers to
households with a female breadwinner while “male” refers to households with a male breadwinner.
A similar proportion (53 percent) of breadwinners who worked less or not at all the week before the survey
and did not receive any pay is observed in the third and fourth waves of the survey (Figure 3.4). When working
less than usual, female breadwinners are much more likely to receive no pay than male breadwinners (66
percent and 50 percent, respectively). By contrast,, men are twice as likely as women to receive partial payment
(30 percent versus 14 percent), while both have roughly the same probability to receive a full payment. Working
in the public sector offers a protection in terms of labor income as 36 percent of the breadwinners who worked
less than usual received a full payment, while they are 24 percent among the breadwinners from the private
formal sector and 7 percent among the ones from the private informal sector. Breadwinners working in the
informal sector are much more exposed to the risk of receiving no pay when working less than usual than others
(62 percent). Thus, in addition of being already vulnerable, breadwinners in the informal sector may be
suffering negative impacts of the COVID-19 crisis for longer, both in terms of their employment status and their
wages.
6
Monitoring the socio-economic impact of
Figure 3.4: Reported change in labor income among breadwinners who worked less than usual or not at all
the week before the survey (%)
100 6 6 5
9
90
80 36
44
70 47 35 53 50
53 66
60 62
50 28
40 31
30 50 27 30
36 27
14
20 25 36
10 24 18
11 15 16 17
0 5 7
Wave 1 Wave 2 Wave 3 Wave 4 Private Private Public Male Female
informal formal
Full payment Partial payment Received no pay Don't know/Refusal
Source: Authors’ calculation based on Djibouti COVID-19 phone survey, 1st, 2nd, 3rd and 4th waves.
Notes: Statistics are based on cross-sectional proportions and not only the longitudinal sample. The distinction by sectors of employment
(public, formal, informal) concerns all the households whose breadwinner was working before the survey. The category “female” refers to
households with a female breadwinner while “male” refers to households with a male breadwinner. The sample size for the employment
sector is 153 for informal, 49 for formal and 44 for public. The sample size for the gender is 159 for men and 48 for women.
LIVELIHOODS Following the wave 3 trend and consistent with the working status outcomes, more households continue to
report income source from family business and waged work (Figure 4.1). Indeed, 79 percent of households
reported it as an income source in wave 4 compared to 22 percent in wave 1. However, the proportion of
households receiving assistance from the government has decreased compared to previous waves, as well as
the proportion of households receiving remittances and assistance from family and friends compared to the 3rd
wave.
Figure 4.1: Reported sources of household’s income for the last 12 months (%)
100
90 79
76
80
70
60
50 43 44 41
40 30 27 24
30 22 19
20 10 10 7 13 17
7 7 9
10 2 4 5 4 4 2
0
Wave 1 Wave 2 Wave 3 Wave 4
Family business and waged work Assistance from government
Pension Remittances and assistance from family/friends
Assistance from INGO Other
Source: Authors’ calculation based on Djibouti COVID-19 phone survey, 1st, 2nd, 3rd and 4th waves.
Note: Statistics are based on cross-sectional proportions and not only the longitudinal sample.
In wave 4, around 40 percent of the households declare having enough resources for the following 30 days
(Figure 4.2). This is an increase of 10 percentage points compared to wave 3. However, important differences
are observed according to the characteristics of the breadwinner and the poverty status of the household.
7
Monitoring the socio-economic impact of
Households whose breadwinner was not working the week before the survey are much less likely than others
to declare having enough resources for the next month (28 percent versus 41 percent). Moreover, while 56
percent of households with a breadwinner working in the public sector report having enough resources for the
following month, it is the case for only a third (32 percent) of the households whose breadwinner is working
either in the informal or formal sectors. Consistently, poor households are less likely to report having enough
resources for to meet their needs for the next 30 days, compared to non-poor households (30 percent and 41
percent, respectively).
Figure 4.2: Proportion of households who declared having enough resources for the following 30 days (%)
100
90
80
70
56
60
50 41 41
39
40 30 33 32 32 30
29 28
30
20
10
0
Wave 1 Wave 2 Wave 3 Wave 4 Private Private Public Working Not Non Poor
informal formal working poor
Source: Authors’ calculation based on Djibouti COVID-19 phone survey, 1st, 2nd, 3rd and 4th waves.
Notes: Statistics are based on cross-sectional proportions and not only the longitudinal sample. The distinction by sectors of employment
(public, formal, informal) concerns all the households whose breadwinner was working before the survey. The category “working” refers
to households whose breadwinner was working the week before the survey while “not working” refers to the households whose
breadwinners did not work the week before the survey.
SAFETY NETS Compared to previous waves, less households received assistance in the form of food stamp, but more
households received cash transfers and food assistance (Figure 5.1). Indeed, the proportion of households
receiving cash transfers and food assistance has more than doubled between waves 3 and 4, while those
receiving food stamps have decreased from 31 percent in wave 1 to 17 percent in wave 4. Consistent with the
previous waves, there is a decreasing trend in the proportion of households who received assistance from
government: from 93 percent in wave 1 to 59 percent in wave 4. Around 16 percent of the households received
help from INGOs, which is twice as in wave 3 but comparable with the first wave. Moreover, households are
still counting on help from social networks as 34 percent reported assistance from family and friends.
8
Monitoring the socio-economic impact of
Figure 5.1: Assistance received and source of assistance in the last 30 days before the survey (%)
a. Households that received assistance b. Source of assistance when received it
100 100 93
90 90 85
80 80 75
70 70 59
60 60
50 50 42
40 31 40 34
27
30 23 2017 30 21
16 15 16
20 11 10 20 8
6 9 7
10 1 4 2 4 4 2 10
0 0
Wave 1 Wave 2 Wave 3 Wave 4 Wave 1 Wave 2 Wave 3 Wave 4
Cash transfer Food assistance Government INGO Family/friends
Food stamp Other assistance in kind
Source: Authors’ calculation based on Djibouti COVID-19 phone survey, 1st, 2nd, 3rd and 4th waves.
Notes: Statistics are based on cross-sectional proportions and not only the longitudinal sample. Data of figure 5.1.b come from income
source questions whereas in previous waves, the source of assistance were computed from a different question.
ACCESS TO Almost all the households reported having access to basic goods the week before the survey (Figure 6.1).
Indeed, nearly all the households reported having access to wheat flour, rice, cooking oil and hand soap, and
BASIC GOODS
compared to previous waves, the trend continues to rise for all the goods. Access to basic medicines is still
slightly lower than for other basic goods, though the gap with other goods is closing (the difference with hand
soap was 10 percentage points in wave 3 compared to 6 percentage points in wave 4).
Figure 6.1: Access to basic goods in the last 7 days (%)
100 95 96 96 94 96 96 96 96 94 100 99 99 98 98
92
87 84
90
76 77
80
66 64
70 61 64
60
50
40
30
18
20
10
0
Wave 1 Wave 2 Wave 3 Wave 4
Wheat flour Rice Cooking oil Vegetables Hand soap Basic medicines
Source: Authors’ calculation based on Djibouti COVID-19 phone survey, 1st, 2nd, 3rd and 4th waves.
Note: Statistics are based on cross-sectional proportions and not only the longitudinal sample.
Depending on the goods (except for cooking oil), between 9 and 16 percent of the households reported a
price increase of main basic goods, similar to the findings in wave 3 (Figure 6.2). In wave 4, less households
declared a price increase of rice, vegetables, hand soap and basic medicines than compared to all the previous
waves. Notably, however, 39 percent of households reported experiencing an increase in cooking oil prices in
the last 7 days, an increase of 27 percentage points since wave 3.
9
Monitoring the socio-economic impact of
Figure 6.2: Increase in price in the last 7 days reported by households (%)
100
87
90
80
70
60
50 43 42 40
39
40
30 23 24 22 24
18 18 19 19 15 16
20 14 12 12 12 15 13 10
14
9
10
0
Wave 1 Wave 2 Wave 3 Wave 4
Wheat flour Rice Cooking oil Vegetables Hand soap Basic medicines
Source: Authors’ calculation based on Djibouti COVID-19 phone survey, 1st, 2nd, 3rd and 4th waves.
Note: Statistics are based on cross-sectional proportions and not only the longitudinal sample.
As in wave 3, almost half of the households reported a need of healthcare (Figure 7.1), and nearly all of those
ACCESS TO who needed healthcare had access to it. The proportion of households reporting having access to healthcare
SERVICES when needed increased from 60 percent in wave 1 to 96 percent in wave 4. Households with a female
breadwinner are less likely than those with a male breadwinner to have access to healthcare when needed (83
percent compared to 96 percent, respectively). Moreover, households whose breadwinner was not working
before the survey were slightly less likely to have access to healthcare when needed than those with a working
breadwinner (94 percent and 96 percent, respectively). While poor households are more likely to report being
in need of healthcare than non-poor ones (48 percent compared to 46 percent, respectively), the least well off
are less likely to report having access to healthcare when needed than the non-poor population. Indeed, 93
percent of the poor households reported being able to receive healthare when needed compared to 96 percent
of the non-poor households. These differences being in general quite small, the results highlight a very good
and almost universal access of the national population to healthcare services when needed.
Figure 7.1: Need of and access to healthcare during the last 30 days (%)
a. Percentage of households that need healthcare b. Access to healthcare among those needed it
100 100 96
90
90 90 85
80 80
70 70 60
60 52 60
47
50 50
40 36 40
30 30
17
20 20
10 10
0 0
Wave 1 Wave 2 Wave 3 Wave 4 Wave 1 Wave 2 Wave 3 Wave 4
Source: Authors’ calculation based on Djibouti COVID-19 phone survey, 1st, 2nd, 3rd and 4th waves.
Note: Statistics are based on cross-sectional proportions and not only the longitudinal sample.
10
Monitoring the socio-economic impact of
On education, very few households reported having a child who does not attend school every day (2 percent)
and there is no difference between boys and girls3 (Figure 7.2). However, children from a household with a
female breadwinner are slightly less likely to attend school every day than children from a household with a
male breadwinner (96 percent and 99 percent, respectively). Moreover, when girls do not attend school every
day, they are more likely than boys to miss school the whole week, while boys are more likely to miss school
occasionally.
Figure 7.2: Frequency of attendance of school (%)
100
90
80
70
60
50 98 99 96 98 98
40
30
20
10
0
All Male Female Boy Girl
breadwinner breadwinner
Every day Between 1-4 days None
Source: Authors’ calculation based on Djibouti COVID-19 phone survey, 4th wave.
Notes: The category “female breadwinner” refers to households with a female breadwinner while “male breadwinner” refers to househo lds
with a male breadwinner. The category “boy” refers to households with a randomly selected male child to question about education while
“girl” refers to households with a randomly selected female child.
Around a third (32 %) of households reported their child being in need of catch up activities (Figure 7.3).
While girls are slightly more likely than male children to need catch up activities, they are less likely to
participate in these activities when needed than boys (70 percent compared to 92 percent).
Figure 7.3: Proportion of children who needed and participated in catch up activities (%)
a. Need of catch up activities b. Participation in catch up activities when needed
100 100 92
90 90 80
80 80 70
70 70
60 60
50 50
40 32 34 40
29
30 30
20 20
10 10
0 0
All Boy Girl All Boy Girl
Source: Authors’ calculation based on Djibouti COVID-19 phone survey, 4th wave.
Note: The category “boy” refers to households with a randomly selected male child to question about education while “ girl” refers to
households with a randomly selected female child.
3 This question was asked about a randomly chosen child, distributed equally between boys and girls across households. Among the 1,047
households who have at least one school-age child (between 6 and 15 years old), a boy was picked in 526 households and a girl was chosen
in 521 households.
11
Monitoring the socio-economic impact of
FOOD Few respondents report experiencing food insecurity based on the three indicators captured in this survey
(Figure 8.1). Indeed, less than 10 percent of the respondents reported eating less than 3 meals a day the week
INSECURITY
before the survey, skipped a meal the last month, or went to bed hungry during the last month. Compared to
wave 34, the food security indicators have improved or stayed similar (indicators varied between 15 and 8
percent in wave 3). Despite female respondents being less likely to have skipped a meal than male, they are
more likely to have eaten less than 3 meals per day and have gone to bed hungry. Moreover, respondents from
a household whose breadwinner was not working the week before the survey are much more likely to suffer
from food insecurity than a respondent from a household with a working breadwinner.
Figure 8.1: Food security indicators by characteristics of the respondent (%)
100
90
80
70
60
50
40
30 20
20 14 15
8 6 8 7 5 8 8 7 7 6 5 5
10
0
All Male respondent Female respondent Working Not working
breadwinner breadwinner
Ate less than 3 meals per day the last 7 days
Went to bed hungry the last 30 days
Skipped a meal the last 30 days
Source: Authors’ calculation based on Djibouti COVID-19 phone survey, 4th wave.
Notes: The category “female respondent” refers to households with a female respondent while “male respondent” refers to households
with a male respondent. The category “working breadwinner” refers to households whose breadwinner was working the week before the
survey while “not working breadwinner” refers to the households whose breadwinners did not work the week before the survey.
Compared to wave 3, food security as expressed by the food consumption score5 has improved (Figure 8.2).
As in wave 3, a food consumption score is computed to capture issues related to food frequency and dietary
diversity. The proportion of households with an adequate food consumption has increased from 82 percent in
wave 3 to 87 percent in wave 4. Moreover, only 3 percent of the households record a poor food consumption
compared to 7 percent in wave 3. However, differences according to the characteristics of the breadwinner are
observed. Households with a female breadwinner are slightly less likely to have an adequate food consumption
than households with a male breadwinner (82 percent and 89 percent, respectively). Moreover, only 75 percent
of the households whose breadwinner was not working the week before the survey had an adequate food
consumption score (compared to 89 percent for the households with a working breadwinner). Thus, households
with a female breadwinner or a non-working breadwinner are more likely to have a poor food consumption.
4 However, both waves results are not perfectly comparable as in wave 3 the question was asked about any child in the household, while
in wave 4 the question focused on the respondent herself.
5 Following the World Food Program’s approach, the food consumption of a household is calculated using the frequency of consumption
of different food groups on a 7 days recall period. The food consumption is considered poor if the score is inferior or equal to 28, borderline
for a score ranging from 28.5 and 42, and adequate for a score between 42.01 and 160.
12
Monitoring the socio-economic impact of
Figure 8.2: Distribution of households by food consumption groups (%)
100 3
7 6 9
90 10 9 9
11 12
80 17
70
60
50
82 87 89 89
40 82
75
30
20
10
0
Wave 3 all Wave 4 all Male Female Working Not working
breadwinner breadwinner breadwinner breadwinner
Adequate FC Boderline FC Poor FC
Source: Authors’ calculation based on Djibouti COVID-19 phone survey, 4th wave.
Notes: The category “female breadwinner” refers to households with a female breadwinner while “male breadwinner” refers to households
with a male breadwinner. The working status “working” refers to households whose breadwinner was working the week before the survey
while “not working” refers to the households whose breadwinners did not work the week before the survey.
Households who have a poor food consumption score are also characterized by an unbalanced diet composed
mainly to staples (Figure 8.3). Indeed, milk and animal protein enter the diet of the households who have at
least a borderline food consumption. Consumption of vegetables is present for all the levels of food
consumption score but is more frequent for households with an adequate food consumption score, while fruits
are absent of the consumption of households with a poor or borderline consumption score. Even at high levels
of consumption score, fruits represent a very small part of the households’ diet . Compared to wave 3, fruits
enter the diet of households at lower food consumption scores (75 in wave 3 compared to food consumption
score equals to 45 in wave 4), and the same is observed for milk and animal protein.
Figure 8.3: Stacked food frequency of main food groups (median)
120
Consumption frequency (median)
100
80
60
40
20
0
10 20 25 30 35 40 45 50 55 60 65 70 75 80 85 90 95 100 105 110 115
Food consumption score
Staple Pulses Vegetables Fruits Animal protein Milk Sugar Oil
Source: Authors’ calculation based on Djibouti COVID-19 phone survey, 4th wave.
13
Monitoring the socio-economic impact of
GENDER In the fourth wave of this survey, a new module was added on gender, intra-household decision making, and
time-use. Respondents were asked who makes decisions within the household on a variety of issues: everyday
purchases, equipment purchases, substantial purchases, and healthcare of household members. Figure 9.1
reflects the distribution of decision makers by gender. Where multiple decision makers are involved, the second
bar reflects women’s participation in the decision-making process. In general, the main decisions of households
are taken by one member rather than several, but this varies by type of decision the household takes. For
example, 70 percent of the households have only one member involved in the decision making regarding
everyday purchases, whereas it is 55 percent of the households for decisions related to healthcare of household
members. When there is only one decision maker in the household, women are more likely than men to make
the decision around everyday purchases and healthcare of household members, whereas men are more likely
to be responsible for the decisions on equipment and substantial purchases. In the case of a decision-making
with several household members, both men and women are most of the time involved for all types of decision.
For example, the decision regarding substantial and equipment purchases is made by both men and women in
90 percent and 89 percent of the households, respectively.
Figure 9.1: Gender distribution among the decision makers of household’s main decisions (%)
100
90
28 32
80
51
70 59
60 73
89 90 86
50
40
72 68
30
4 49
20 41
10 23 4 3
3 11
0 7 7
One mbr More than One mbr More than One mbr More than One mbr More than
(70%) one (30%) (57%) one (43%) (59%) one (41%) (55%) one (45%)
Everyday purchases Equipment purchases Substantial purchases Healthcare of household
members
Women only Men only Both men and women
Source: Authors’ calculation based on Djibouti COVID-19 phone survey, 4th wave.
Notes: Only households with both adult men and women are included. Percentages in parentheses represent the proportion of households
who declared having one member making each type of decisions, and those who declared multiple members.
When it comes to time-use, in most cases, households declare that only one member spends the most time
undertaking the certain tasks, which could suggest specialization in certain tasks by gender (Figure 9.2). For
example, in 68 percent of the households, one member devotes the most time to domestic work, meaning that
this task is shared between several household members in 32 percent of the households. For most of the basic
household chores, such as domestic work, grocery shopping and healthcare of household members, women
dedicate the most time to the task (in more than 80 percent of the households in which there is only one
member spending the most time). Among activities that men spend most time on, income-generating activities
are mostly undertaken by men, but it is also help with children’s studies (compared to other tasks) that men
are more likely to devote time to compared to other activities (35 percent of the households in which only one
member spending the most time helping children report that a man spends the most time in helping children
with studies). Even when there is more than one member devoting time to domestic work, women undertake
this activity in 59 percent of the households. However, both men and women devote the most time to main
tasks in the majority of the households where several members share the burden of the time spent (between
64 percent to 82 percent of the households depending on the task). Moreover, men are more likely to spend
the most time on income generating activities than women (in 68 percent of the households).
14
Monitoring the socio-economic impact of
Figure 9.2: Gender repartition among those who spend the most time on main tasks since COVID (%)
100 6
90 18 19
39 35 30
80
70 64 68
70 78
60 2 82
50 94
40 82 81
5 65 70
30 59 10
20 2 5 32
31
10 20 16 17
0
One mbr More than One mbr More than One mbr More than One mbr More than One mbr More than One mbr
(63%) one (37%) (68%) one (32%) (60%) one (40%) (53%) one (47%) (58%) one (42%)
Grocery shopping Domestic work Help children with Healthcare of Social/leisure Income
studies household members activities generating
activities
Women only Men only Both men and women
Source: Authors’ calculation based on Djibouti COVID-19 phone survey, 4th wave.
Notes: Only households with both adult men and women are included. Percentages in parentheses represent the proportion of households
who declared having one member making each type of decisions, and those who declared multiple members.
The 4th wave survey asked respondents6 to estimate the amount of time devoted usually to main tasks (Figure
9.3). Male respondents spend on average more time than female respondents on income generating activities
(27 percent of daytime for men compared to 24 percent for women), as well as on social or leisure activities
and helping children with studies. In contrast, women are likely to dedicate more time than men to tasks such
as grocery shopping, domestic work and healthcare of household members. It is worth noting that the
distribution of daytime use by the respondent may not be representative of the time-use of women in Djibouti.
The respondents are in most of the cases the household head or spouse of the household head, which means
they are more likely than others household members to work. For this reason, this result cannot be directly
compared to Figure 9.2, especially as the latter refers to the person who spends “most” time on tasks, including
income generating activities.
Figure 9.3: Percentage of daytime usually spent by the respondent on main tasks (%)
a. Male b. Female
6 6
27 24
30 30
5 7
5 10 4 13
4 3 5 4
5 3 6 4
6During this 4th wave, the respondent was randomly chosen among the household head and the spouse to allow an even distribution
between male and female respondents across households.
15
Monitoring the socio-economic impact of
Source: Authors’ calculation based on Djibouti COVID-19 phone survey, 4th wave.
For most of the male and female respondents, the time spent by the respondent on the main tasks of the
day has not changed since COVID (Figure 9.4). However, female respondents are more likely than male
respondents to have increased their time dedicated to income generating activities (11 and 9 percent,
respectively for women and men), grocery shopping and domestic work since COVID. That said, twice as many
women respondents (9 percent) also reported decreasing time spent on income-generating activities compared
to men (4 percent). Thus, income generating activities is the task where gender differences in variation of time
spend since COVID are the highest compared to other tasks. In contrast, men are more likely to have increased
their time spent on helping children with studies and social/leisure activities. Regarding the reduction of
daytime devoted to certain tasks, men and women have on average the same likelihood to have decreased it
on most of the daily tasks, except on income generating activities.
Figure 9.4: Variation of time spent by the respondent on main tasks per day since COVID (%)
100 4 9 7 8
90 15 14 15 14 18 18
80
70
60
50 87 81 86 83
78 78 76 79
40 76 78
30
20
10
9 11 6 8 7 9 9 7 5 4
0
Male Female Male Female Male Female Male Female Male Female
Income generating Grocery shopping Domestic work Help children with Social/leisure
activities studies activities
Increased Stayed the same Decreased
Source: Authors’ calculation based on Djibouti COVID-19 phone survey, 4th wave.
Djiboutian women are in general less likely to work than men (Figure 9.5). Around 27 percent of all the
interviewed women aged between 15 and 64 years old had an income generating activity the week before the
survey, against 39 percent of men. No difference is observed compared to the period before COVID. Among
households that have both men and women adults, half have no working age women with an income
generating activity the last 7 days to the survey. In 15 percent of the households, all the adult women were
working.
16
Monitoring the socio-economic impact of
Figure 9.5: Women who had an income generating activity the last 7 days (%)
a. In the household b. In the population
100 100
90 15 90
2
80 12 80
70 70
60 17 60
50 50
39
40 40 33
30 27
53 30
20 20
10 10
0 0
None Less than half 1
Half More than half All All Men Women
Source: Authors’ calculation based on Djibouti COVID-19 phone survey, 4th wave.
Note: Only households with both adult men and women are included in the Figure 9.5.a.
Male respondents are on average more likely than female ones to report having been victim of a crime in the
last 14 days (Figure 9.6), as well as reporting not feeling totally safe in public space (18 percent of men and 13
percent of women). However, female respondents are more likely to declare experiencing more fights or
conflicts in the household during the last 14 days than male respondents.
Figure 9.6: Respondents who experienced some safety issues in the last 14 days (%)
100
90
80
70
60
50
40
30
15 18
20 13
8 7 9
10 4 5 3
0
All Male Female
Victim of crime Not feeling totally safe in public space More fights/conflicts in the HH
Source: Authors’ calculation based on Djibouti COVID-19 phone survey, 4th wave.
VIEWS ON Less than a third of the sample (29 percent) declared having ever been tested for COVID-19 at least once
since the beginning of the pandemic (Figure 10.1). Differences in likelihood to get tested are reported by
VACCINATION
poverty and working status. Respondents from a household with a non-working breadwinner are less likely to
have taken a COVID-19 test than others (13 percent compared to 32 percent). And among respondents from
households with a working breadwinner, those whose breadwinner works in the public sector are more likely
to have already been tested for COVID-19 than others. This points out a risk of unequally access to COVID-19
tests in the population or a difference of interest to get tested. Once they did a COVID-19 test, only 4 percent
of the respondents report a positive result.
17
Monitoring the socio-economic impact of
Figure 10.1: Prevalence of COVID-19 tests (%)
100
90
80
70
60
50 39
40 29 32 30
25 27
30 22
20 13
10
0
All Informal Formal Public Working Not Non- Poor
working poor
Source: Authors’ calculation based on Djibouti COVID-19 phone survey, 4th wave.
Notes: The distinction by sectors of employment (public, formal, informal) concerns all the households whose breadwinner was working
before the survey. The category “working” refers to households whose breadwinner was working the week before the survey while “not
working” refers to households whose breadwinner was not working the week before the survey.
Most respondents would accept to take an approved and free COVID-19 vaccine (Figure 10.2). Compared to
the previous wave, this proportion is slightly higher (75 percent in wave 4 compared to 73 percent in wave 3).
The acceptance of a COVID-19 vaccine is higher among respondents from households whose breadwinner
works in the informal sector (77 percent) compared to those with a breadwinner working in the formal and
public sectors (73 and 72 percent, respectively). The main reasons for refusing a COVID-19 vaccine are worries
about undesirable effects (for 31 percent of the respondents who are reluctant to take it) and the fact that
respondents do not trust vaccines in general (23 percent). Around 10 percent of the respondents would not
accept to take the COVID-19 vaccine but would be more likely to take it if someone, such as family, friends,
religious leaders, recommends it. Indeed, among the respondents who would not accept to take a COVID-19
vaccine, 24 percent declared that they would be more likely to receive the COVID-19 vaccine if their family and
friends receive or recommend it and 20 percent would do it if religious leaders receive or recommend it.
However, 58 percent of those who would refuse the vaccine reported that nothing would change their decision,
and these represent 14 percent of all the respondents. Respondents from poor households appear to be more
likely than others to change their mind if someone recommends the vaccine, while respondents from a
household with a non-working breadwinner are more likely than those with a working breadwinner to not
change their mind.
18
Monitoring the socio-economic impact of
Figure 10.2: Acceptance of a COVID-19 vaccine among respondents (%)
100
90 14 11 17 15 13 16 14 14
80 10 11 11 7 10 13
9 10
70
60
50
40 75 77 75 76 75
73 72 73
30
20
10
0
All Informal Formal Public Working Not Non-poor Poor
working
Yes No, but would change my mind No, nothing would change my mind Don't know
Source: Authors’ calculation based on Djibouti COVID-19 phone survey, 4th wave.
Notes: The distinction by sectors of employment (public, formal, informal) concerns all the households whose breadwinner was working
before the survey. The category “working” refers to households whose breadwinner was working the week before the survey while “not
working” refers to households whose breadwinner was not working the week before the survey.
This report aimed to provide an update on the monitoring of the impacts of COVID-19 in Djibouti based on a
CONCLUSION fourth wave of the COVID-19 survey, carried out between March and April 2021, which followed households
since June 2020. The findings suggest that the economy is very much on a path to recovery. Indeed, 85
percent of households reported their breadwinner working the week before the survey, compared to 58, 77,
and 83 percent in the first, second, and third wave of data collection. Additionally, more breadwinners in the
fourth wave declared working as usual than in previous waves. The data reveals that certains groups of the
population may be more adversely affected than others, notably informal workers and female breadwinners.
Informal workers exhibit a higher propensity to work less than usual, and among those a higher proportion
receive no pay. Similarly among the female breadwinners who report working less than usual or not at all (6
and 7 percent respectively), 66 percent report not receiving any pay.
Access to goods and services seems to have improved across the board for Djiboutian households: access to
basic food and medicines, healthcare when needed, and education. Gender differences are however observed
among children who need supplementary school activities, or catch-up activities. For instance, while 34 and 29
percent of girls and boys, respectively, are declared to need catch-up scholarly activities, only 70 percent of
girls participate in them when needed compared to 92 percent of boys.
This wave also explores gender differences in decision making and time-use. Women tend to participate more
than men in decisions related to everyday purchases and healthcare of household members, especially when
household decisions are taken by a single household member. Where more than one household member is
involved in making the decision, women participate in the decisions jointly with men in most of the cases. When
decisions are made by one household member, it is typically made by men . On time-use, women are more
likely to spend time on grocery shopping, domestic work, children’s studies, healthcare and leisure activities,
than they are on income-generating activities. This is particularly the case when it is one member who spends
the most time undertaking the activity. With regards to public safety, men report a higher likelihood of being a
victim of crime and not feeling safe in public spaces than women do, but women report a higher likelihood of
experiencing domestic conflict.
19
Monitoring the socio-economic impact of
As Djibouti had experienced an increase in the COVID-19 cases in March 2021, this survey also elicited
respondents’ attitudes towards vaccines. Most respondents (75 percent) reported that they would accept to
take an approved and free COVID-19 vaccine. The main reasons for refusing a COVID-19 vaccine are worries
about undesirable effects (for 31 percent of the respondents who are reluctant) and the fact that respondents
do not trust vaccines in general (23 percent). Around 10 percent of the respondents would not accept to take
the COVID-19 vaccine but would be more likely to take it if someone, such as family, friends, religious leaders,
recommends it. Respondents from poor households report a lower propensity to accept the vaccine, but a
higher likelihood to change their mind if someone recommended it.
Box 1. Sampling strategy and sampling weights in wave 4
Data from the national social registry of the Ministry of Social Affairs, restricted to urban households having
at least one phone number and interviewed after July 1, 2017 (to increase the response rates), serves as the
sampling frame for the Djiboutian sample of this survey. The social registry is an official database of
households in Djibouti that may benefit from poverty alleviation efforts including as targets from public
transfers. This data has been collected since 2014 and consists of about 70,000 households, with majority of
the fieldwork conducted from 2017 onwards. Even though this database over-represents the poor, it provides
an up-to-date sampling frame. The social registry collects a wealth of socioeconomic characteristics of
households along with working phone numbers of household heads or spouses of household heads. The use
of biometric information to record household level data negates the possibility of having duplicate entries.
This wave’s sample combined a panel of households interviewed during the first three waves, to which was
added a replacement sample to compensate for attrition. But unlike the three preceding waves, to keep
consistency with the approach used for the refugee sample, households that were lost to follow up in the third
wave were included in the sample. The data set consisted of 1,561 interviewed households with complete
information that were representative of the urban population, out of which 932 households entered the
survey since the first wave and 629 were added as replacement households in either wave 2, 3, or 4. The
sampling strategy allows for disaggregation by poverty status7 and by three survey domains, being Balbala
(539 households), rest of Djibouti city (527 households) and urban areas outside Djibouti city (495
households). Table A1 presents the breakdown of the sample of Djibouti nationals by survey domain.
Table A1: Sample of Djibouti nationals broken down by survey domain
Share of urban population Sample size
Survey domain (Household budget survey - Panel Replacement Total
EDAM, 2017) (%) (# households) (# households) (# households)
Balbala 54.1 310 229 539
Rest of Djibouti City 35.5 327 200 527
Other urban areas 10.4 295 200 495
Total 100.0 932 629 1,561
Both cross-sectional and panel weights are designed to adjust for differences in selection probability due to
either design or non-response. In addition, further adjustments in sampling weights were made to ensure that
indicators produced are representative of the country’s population, by poverty status and by location. The
sampling frame of the Djibouti nationals, the social registry of the Ministry of Social Affairs, over-represents
the poor and has an incomplete coverage of the upper distribution of income. To correct for these biases, we
rely on a post-calibration approach, using the household budget survey of 2017 (EDAM 2017) as the reference
data source. This is because EDAM 2017 survey was representative of the country’s population by poverty
status and survey domains. However, EDAM 2017 survey is restricted to the first four consumption quintiles
to ensure sufficient overlap of the universes covered by both surveys.
7Poverty status variable in the social registry database is based on consumption per capita, which is imputed for each household by the
Ministry of Social Affairs and Solidarity (MASS) based on observable characteristics and using the Proxy Means test formula using household
budget survey of 2013.
20
Monitoring the socio-economic impact of
Box 2: Attrition between wave 1 and wave 4
Table A2.1: Composition of the wave 4 sample and panel status
Panel status Frequency Percentage
Households interviewed in waves 1, 2, 3 and 4 802 51.4
Households interviewed in waves 2, 3 and 4 149 9.5
Households interviewed in waves 1, 2 and 4 130 8.3
Households interviewed in waves 2 and 4 35 2.2
Households interviewed in waves 3 and 4 171 11.0
Households interviewed in wave 4 only 274 17.6
Observations 1,561 100
Source: Djibouti COVID-19 phone survey, 1st, 2nd, 3rd, and 4th waves.
Regressing a variable indicating whether households dropped out of the survey on household characteristics shows
that there is no statistically significant correlation between attrition and observables characteristics, with the
exception of the replacement status in wave 1 where households from the replacement sample are more likely to
attrit in wave 4.
Table A2.2: Log-odds ratios of regressing an indicator of attrition on household characteristics
1(Drop out)
[Base=Balbala]
Other urban areas 0.10 0.09 0.08 0.06 0.06 0.06 0.06
[0.177] [0.178] [0.179] [0.179] [0.179] [0.179] [0.180]
Rest of Djibouti-Ville -0.15 -0.16 -0.16 -0.19 -0.19 -0.19 -0.19
[0.184] [0.184] [0.184] [0.187] [0.188] [0.188] [0.189]
Replacement in wave 1 (Yes=1) -0.045** -0.044** -0.044** -0.044** -0.044** -0.043**
[0.021] [0.021] [0.021] [0.021] [0.021] [0.021]
Log-household size -0.08 -0.05 -0.05 -0.06 -0.05
[0.119] [0.122] [0.123] [0.131] [0.131]
Sex of household head 0.20 0.20 0.20 0.19
[0.156] [0.157] [0.157] [0.157]
Age of household head 0.00 0.00 0.00
[0.005] [0.005] [0.005]
Poverty status (Poor=1) 0.05 0.05
[0.139] [0.139]
[Base=Did not work the week before the survey]
Worked week before survey 0.02
[0.154]
Worked week before survey (Don't know) 0.56
[0.604]
Constant -0.502*** -0.347** -0.22 -0.50 -0.49 -0.49 -0.50
[0.126] [0.145] [0.246] [0.347] [0.395] [0.394] [0.403]
Observations 1,486 1,486 1,486 1,486 1,486 1,486 1,486
Robust standard errors in brackets
*** p<0.01, ** p<0.05, * p<0.1
Source: Djibouti COVID-19 phone survey, 1st and 4th waves.
21
Monitoring the socio-economic impact of
Box 3: Output of a principal-components factoring analysis on food consumption score
A principal-components factoring analysis is used to validate consistency in the data based on eight food
groups recommended by the WFP (excluding condiments). It indicates that food consumption can be
regrouped along two main dimensions explaining approximately 51 percent of the variance in consumption
frequency. Staple, vegetables, milk, sugar, and oil represent the main dimension of food consumption
(explained variance = 29 percent), while pulses, fruits, and animal proteins define the second component of
food consumption (explained variance = 22 percent). Examination of these two components suggests no
redundant grouping of food items, as most food groups have high unique contribution to the explained
variance.
Number of obs = 1,561
Factor Eigenvalue
Factor1 2.33
Factor2 1.78
Factor3 0.96
Factor4 0.85
Factor5 0.72
Factor6 0.59
Factor7 0.43
Factor8 0.34
Factor Variance Difference Proportion Cumulative
Factor1 2.27 0.43 0.28 0.28
Factor2 1.84 . 0.23 0.51
LR test: independent vs. saturated: chi2(28) = 2431.05 Prob>chi2 = 0.0000
Pattern matrix and unique variances
Variable Factor1 Factor2 Uniqueness
Staple 0.50 0.20 0.72
Pulses 0.14 0.50 0.73
Vegetables 0.68 0.07 0.53
Fruits -0.34 0.68 0.43
Animal protein 0.00 0.83 0.31
Milk 0.39 0.62 0.47
Sugar 0.72 0.06 0.48
Oil 0.87 -0.10 0.24
Source: Djibouti COVID-19 phone survey, 4th wave.
22