International Journal of Educational Development 113 (2025) 103228

Contents lists available at ScienceDirect

International Journal of Educational Development

journal homepage: www.elsevier.com/locate/ijedudev

The teachers who leave: Teacher attrition in Burkina Faso

Biniam Bedasso a,* , Amina Mendez Acosta b
a
Center for Global Development, 1 Abbey Gardens, Great College St, London SW1P 3SE, United Kingdom
b
Consultant for Center for Global Development, 2055 L St NW, Washington, DC 20036, United States

A R T I C L E I N F O A B S T R A C T

Keywords: High teacher attrition affects education systems through direct costs in replacing teachers who left the service,
Education and indirect costs in classroom disruption and loss in experience. Efforts to address teacher shortage must be
Teachers informed by which teachers leave and why. Using administrative data from Burkina Faso, we aimed to analyze
Attrition
the demographic and geographic correlates of teacher turnover. We find that early career teachers, female
Low- and middle-income countries
teachers, and teachers with tertiary education, are more likely to attrite. Teachers who hold higher position-
s—such as school principals—have better retention rates. In terms of school-level attrition, rural and remote
schools tend to lose teachers to other schools whereas schools in urban or more developed regions are more likely
to lose teachers to options outside of the teacher workforce. Finally, we discuss policy options in improving
teacher retention given these findings.

1. Introduction more remote schools, which already experience challenges in recruit-

ment and retention and where quality teachers are already in short
Government efforts to recruit teachers and improve the quality of supply (Evans and Mendez Acosta, 2023b).
teaching in low- and middle-income countries (LMICs) can be under- Improving teacher retention is particularly important in the context
mined by high teacher turnover. Almost half of the education budget in of the increasing demand for teachers expected in the coming years
LMICs go to teacher wages (Crawfurd and Pugatch, 2022). High teacher driven by population growth and improving enrollment rates. In Sub-
attrition represents significant loss in resources by way of high costs of Saharan Africa, countries would need to recruit 15 million teachers to
recruiting and training replacement teachers. Even when teachers are meet the Sustainable Development Goal 4, i.e. education for all at the
quickly recruited to replace those who exit the service, it might take primary and secondary level by 2030 according to a UNESCO report
considerable time from recruitment to deployment. This means class- (2024). From the same report, about one-third of this number (5.6
room activities and academic calendars are affected, which leads to loss million teachers) will be needed to fill vacant posts due to attrition while
of instructional time, or the new teachers might not have the same level the rest will be needed to fill new positions. This reflects the rapidly
of skills and experience as those who choose to leave (UNESCO, 2010). growing youth population in the region which is expected to grow by 1.5
In these scenarios, student learning may suffer. For example, a teacher times between 2022 and 2040, compared to 1.2 times in South Asia
resigning in Rwanda is associated with lower school test scores, poten- based on authors’ analysis using data from the UN Population Division
tially driven by an increase in class size and the high share of teachers (2022). In contrast, the population growth rates in other regions such as
who provide instruction outside of their subject matter expertise East Asia and the Pacific and Europe and Central Asia are expected to
(Zeitlin, 2021).1 These impacts are potentially bigger on poorer and remain stagnant for the same period (UN Population Division, 2022).

* Corresponding author.
E-mail address: [email protected] (B. Bedasso).
1
The negative impacts of attrition on learning would most likely depend on the quality of teachers who leave. An incentive program in Peru encouraged teachers
to move to hard-to-staff schools, improving retention by up to 5 percentage points. But while test scores in the schools with the incentive did not change, test scores in
neighboring schools with no incentive (and which experienced higher attrition) saw up to 0.30 standard deviation higher math and test scores (Castro and Esposito,
2022). This suggests that the teachers who decided to move schools are lower-skilled such that the non-incentive schools were left with higher average teacher
quality, if slightly fewer teachers. In many settings like Rwanda, teachers who leave because of better job opportunities are exactly those with desirable teacher
qualities (higher educational attainment, specialized skills in STEM, or better mastery of language such as English) such that teacher loss conceivably results in a drop
in test scores.

https://doi.org/10.1016/j.ijedudev.2025.103228
Received 27 September 2024; Received in revised form 19 December 2024; Accepted 30 January 2025
Available online 13 February 2025
0738-0593/© 2025 The Authors. Published by Elsevier Ltd. This is an open access article under the CC BY license (http://creativecommons.org/licenses/by/4.0/).
B. Bedasso and A. Mendez Acosta International Journal of Educational Development 113 (2025) 103228

Efforts to meet this teacher demand will require better teacher hiring to leave the workforce. Second, we summarize the existing evidence on
and deployment practices, and better retention of teachers through policies that improve teacher retention in LMICs and highlight the in-
better teacher working conditions. Understanding the profile of teachers terventions that are promising but do not have sufficient evidence.
who leave the workforce and where in their career trajectory they are Finally, we contribute to the literature using a novel data source (teacher
likely to leave can help shape policies to prevent early exit from the panel data) from a developing country in armed conflict facing multiple
teacher workforce. challenges in teacher retention (Burkina Faso) that has not been previ-
Teachers may exit the work force earlier than the retirement age due ously used to provide insights to teacher attrition.
to a combination of better opportunities outside of teaching (such as The remainder of the paper is organized as follows: Section 2 pro-
other civil service positions or jobs in the private sector), dissatisfaction vides the context on teacher attrition in LMICs and the teacher working
with their working conditions (poor pay, remote assignments, or limited conditions in Burkina Faso, Section 3 discusses the methodology and its
classroom resources), demographic and personal factors (such as illness limitations, Section 4 reports the results, Section 5 discusses some of the
or to raise a family), or combinations of such factors (for example, how a policy options, and we conclude in Section 6.
teacher’s race or caste fits the social and cultural dynamics of the school
they were assigned) (UNESCO, 2010; Nguyen et al., 2020). Governments 2. Global and local context
have sought to address these factors through different policies. One of
type these policies provides incentives to stay in the profession such as 2.1. Teacher retention in low- and middle-income countries
financial incentives implemented in Latin America (Cabrera and Web-
bink, 2020; Camelo and Ponczek, 2021; Hinze-Pifer and Méndez, 2016) Retaining teachers, especially quality teachers, is a challenge for
and in Sub-Saharan Africa (Chelwa et al., 2019; Pugatch and Schroeder many education systems and particularly so in Sub-Saharan Africa
2014, 2018; Swai 2013). Another set of policies are national recruitment (Mulkeen et al., 2007; Pitsoe and Machaisa, 2012) (UNESCO, 2023).
drives such as the ones recently implemented in Tanzania, Kenya, and Earlier work describes a “hemorrhage of teachers who leave the pro-
Zimbabwe (Byemelwa 2024; Moses 2024; New Ziana 2024). Yet another fession before retirement age” in a paper that reports on efforts to retain
type of policies targets specific groups of teachers such as female secondary level teachers in Sub-Saharan Africa (Mulkeen et al., 2007)
teachers (Gad, 2015; Kamere et al., 2019). while another paper refers to the “teacher attrition catastrophe” in the
In this paper, we attempt to examine the demographic profiles of region. A more recent work shows lower attrition levels but highlight
teachers who leave the profession and the predictors of school-level structural inefficiencies: an analysis of 30 countries in Sub-Saharan Af-
attrition in Burkina Faso. As a low-income country that has been rica region shows the average attrition rate for primary school in the
struggling to maintain access to education and improve quality amid mid-2010s was 4.8 percent (Bennell, 2023). Even with lower attrition,
chronic poverty and widespread conflict, Burkina Faso could serve as a the delays in replacing these teachers, which in some cases may take
good case study for a context in which teachers might be faced with years, drive the high vacancy rates and amplifies the impact of teacher
economic hardships, insecurity, and limited outside options in deciding attrition on actual availability of teachers ready to teach in classrooms.
to stay in their jobs. As such, gaining a better understanding of the in- The latest attrition rates show a higher average of 8 percent between
dividual characteristics of teachers who are more likely to leave and the 2018 and 2022 across thirteen countries (Appendix Figure A1). Some
type of schools that are more likely to experience higher attrition could countries in particular struggle with relatively high teacher attrition:
be useful to devise better targeted retention strategies. Rwanda, Sierra Leone, Mauritania, and Benin all have attrition rates
Examining teacher retention trends requires panel data on teacher above 10 percent. Burkina Faso is close to the median with an attrition
movement in and out of the workforce and between schools as well as rate of 5.4 percent for the period. The challenge of retaining teachers is
demographic and location-specific characteristics. We use the School not unique to low- and middle-income countries: the average teacher
Census Data from Burkina Faso’s Education Management Information attrition rate in 14 OECD countries in 2016 is at 6.6 percent (OECD,
System (EMIS) from 2014 to 2019 to build the attrition profiles of 2021). However, the stronger education systems in these countries allow
teachers with respect to various demographic and socioeconomic fac- for quicker recruitment and deployment such that effects on student
tors. We also compute the school-level rates of attrition consisting of learning and participation might be better managed and mitigated.
teachers who leave the profession and those who transfer to another Why do teachers leave? Teachers may voluntarily leave before
school. We find that teachers in the first five years of teaching, teachers reaching retirement age due to demographic and personal factors (such
with university degrees, and female teachers are more likely to leave the as illness or to raise a family), more attractive opportunities outside
teaching profession. This suggests that competing professional interests teaching either through better paid civil service positions or other non-
outside the teacher profession and personal milestones such as child- teaching jobs in the private sector, concerns on career development and
birth and domestic responsibilities may motivate attrition. progression, or other issues relating to teacher working conditions
At the school level, we find that there is a significant difference be- (UNESCO, 2010; Nguyen et al., 2020). In Cameroon, migration of new
tween factors that are associated with the number of teachers that leave teachers to other government departments is only next to retirement and
the profession and those who switch schools. Specifically, schools that death in top reasons cited for teacher attrition (World Bank 2018a).
are located in urban areas and more developed regions are likely to Similarly in Guinea-Bissau, many teachers leave to join other ministries
experience more attrition in terms of teachers leaving the profession for after the first year of teaching, and as many as 90 percent of teachers in
outside options whereas schools in rural areas and less developed re- public schools also work in the private sector as their second job (World
gions are more prone to teachers transferring to urban schools. We also Bank 2018b; 2019).
summarize the policy options and their available evidence in improving Not many studies attempt to profile the type of teachers who leave,
teacher retention such as providing monetary incentives and similar potentially because of the limitations on data availability. One study
benefits, enhancing opportunities for teacher professional development, that does so uses administrative records and teacher placement data in
improving overall job satisfaction, and addressing specific barriers that Rwanda to identify the profile of teachers who attrite (Zeitlin, 2021).
female teachers face. We highlight how the policies may be relevant to Male teachers, teachers of Math subjects, and those in their first five
the Burkina Faso context based on our findings on the correlates of years of tenure are significantly more likely to leave the profession,
attrition. potentially because of better job opportunities outside teaching avail-
We add to two bodies of literature and present research from a new able to young professionals with strong quantitative background.
data source. First, we contribute to the general understanding of the Teacher attrition in Rwanda does not seem to be driven by rural-urban
challenge of teacher attrition in the Sub-Saharan Africa region, espe- designation, but attrition is surprisingly negatively associated with
cially on the demographic and geographic determinants of the decision pupil-teacher ratio (such that teachers are less likely to leave schools

2
B. Bedasso and A. Mendez Acosta International Journal of Educational Development 113 (2025) 103228

with higher class size) which could be the result of both teachers and applicable to Burkina Faso than others. An analysis of teacher pay in 15
students sorting to better schools over time. countries in Sub-Saharan Africa shows that the monthly median earn-
Several other studies — albeit of more qualitative nature — show ings of teachers in Burkina Faso of 650 dollars per month (PPP in 2011
both personal factors and lack of resources as drivers of teacher attrition. dollars) is comparable to peers in the region where the average is 677
A survey of about 100 current and former secondary school teachers in dollars per month, ranging from 100 dollars in Democratic Republic of
Kenya shows that male teachers are more likely to attrite, and that un- the Congo to 2306 dollars in Namibia (Evans et al., 2022). Teacher
like in Rwanda, married teachers and teachers with work experience of salary in the country is almost three times the average of other wage
over 20 years are more likely to quit (Mabeya 2021). In South Africa, workers despite working up to 25 percent fewer hours than the same
qualitative surveys with a small sample of teachers cite unfavorable wage workers (40 hours versus 53 hours per week). However, as with
working conditions due to “disintegration of discipline,” limited facil- any other profession, opportunity cost increases as individuals get more
ities for teaching, inadequate incentives, and school overcrowding as qualified. Teachers with secondary education earn as much as other
detriments to teacher retention (Pitsoe 2013). Earlier studies suggest managers and other professionals, but teachers with post-secondary
poor teaching conditions in Senegal, Burkina Faso, and Mali, including education can earn twice more by switching sectors. Similar to many
classroom overcrowding, shortage of textbooks and teaching materials, countries, teachers with permanent contracts earn much more than
and subpar living accommodations lead to teachers quitting (Caillods teachers with temporary contracts; in Burkina Faso, civil service
2011; Mulkeen et al., 2007). teachers on average earn almost three times that of contract teachers.
There is also a rich literature on teacher attrition in high-income Teacher deployment is rigid with limited inter- and intra-region
contexts: a meta-analysis found 120 studies that report on associations mobility (UNESCO, 2020). Teachers declare the region they intend to
of personal and school-level correlates to teacher turnover (Nguyen work in during the application process to be certified teachers. Once
et al., 2020). Although none of the papers use data from low- or hired, they are deployed within the region they selected based on which
middle-income countries, the review presents a conceptual framework schools experience the highest teacher shortages. They may request to
for understanding teacher attrition that is still relevant to the context of transfer to a different school within the region after three years of ser-
developing countries. The framework combines three main drivers: (1) vice, and to transfer to a school outside of their region after six years.
personal factors such as demographics and teacher qualifications, (2) Majority of new teachers are often assigned to rural locations in favor of
school factors such as location, resources, and learners’ profiles, and (3) putting teachers with more seniority in more desirable locations (Sirois,
external policy factors such as pay and evaluation policies, presence of 2017).
unions and hiring practices. Similarly, there is a rich adjacent literature The workload of teachers as proxied by pupil-teacher ratios is 40:1 at
(including a systematic review of over 100 studies) on improving the primary level and 23:1 at the secondary level which are comparable
retention of health workers especially in rural and remote areas that to other countries in the Sub-Saharan Africa region and other countries
parallel many of the findings on why teachers leave the profession and of the same income group; see Appendix Figure A2 (World Bank, 2023).
what might work to retain them (World Health Organization, 2020). For Student completion rate is 64 percent at the primary level which is
these civil servants, isolated initiatives such as short-term financial in- comparable to the regional average and average for the same income
centives rarely work or only work while the incentives are in place, but group, and 38 percent at the secondary level which is slightly lower than
the effect quickly fades after the end of the program. Rather, monetary those of peer countries as reported in Appendix Figure A3 (World Bank,
and non-monetary incentives delivered with initiatives that improve 2023). At both primary and secondary levels, completion rates of girls
community and family engagement, wider health services reform. and exceed those of boys.
more sustainable service redesigns together have a higher likelihood of
affecting change. 2.3. Student learning in Burkina Faso
Countries experiencing active armed conflicts are particularly
vulnerable to teacher shortages. Human Rights Watch (2020) docu- Students in the country still have low levels of learning. In an anal-
mented 126 attacks on schools, students, and teachers in Burkina Faso ysis of test scores of primary school students in 14 Francophone coun-
between 2017 and 2020, and over 270 attacks in 2022–2023 (Global tries, only one in three early primary student in Burkina Faso has met
Coalition to Protect Education from Attack 2024). Aside from direct what is considered minimum proficiency level in the language of in-
threats to physical safety to students and education professionals, a struction — the level at which students have gained the skills that would
conflict-induced crisis also disrupts critical education infrastructures allow them to continue their schooling without difficulty (PASEC, 2020;
such as disbursement of salaries, professional development activities, Le Nestour, 2021). Only one in five students can count more than 80 (a
school construction, and supplies of teaching materials (Mendenhall skill usually gained by the second grade of schooling). In both language
et al., 2018). Protracted crises also affect longer-term supply of teachers and math, students in Burkina Faso performed slightly lower than the
(disrupted education will produce fewer and less-qualified teachers) and mean across the countries in the sample. Given that teacher quality is
education financing (limited national budgets competing against directly associated with better learning outcomes—where approxi-
rebuilding and rehabilitation, and short-cycled humanitarian funding mately one-third of the variability in test scores in the surveyed coun-
being incompatible to sustaining recurring teacher salaries costs) tries can be attributed to variations in teacher knowledge (Bietenbeck
(Nicolai and Hine 2015). Migration adds another layer of complexity et al., 2023) —the availability of teachers, especially knowledgeable and
when refugees settle in areas already facing teacher shortages and experienced teachers, is imperative to addressing these low learning
overcrowding, teachers lose their work permits when crossing borders, levels.
and teachers preemptively migrate away from conflict areas (UNESCO
2018). 3. Methodology
While demographic patterns might be influenced over time through
targeted hiring practices, governments and policy actors can also 3.1. Data
explore more immediate and sustainable ways to improve teacher
working conditions in general to encourage the best of teachers in the We use the annual school census data in Burkina Faso’s EMIS for the
workforce to stay. analysis in this paper. The school census collects quantitative data on a
range of school level indicators. We have data for six years, from 2014 to
2.2. Teacher working conditions in Burkina Faso 2019. The data covers public, private and religious primary schools
across the country. The total number of schools included in the census
Some of the factors to attrition mentioned above might be more ranges from 13,200 in 2014–16,482 in 2019. A notable feature of the

3
B. Bedasso and A. Mendez Acosta International Journal of Educational Development 113 (2025) 103228

Burkina Faso school census is that it has a teacher roster which is used to tracked in the census data (regardless of attrition status). There is little
collect demographic and employment data on all teaching staff in every systematic difference in the basic characteristics of teachers who were
school. The total number of teachers across all schools on whom data successfully tracked for the entire length of the panel compared to those
were collected ranges from 42,477 in 2014–65,651 in 2019. tracked only for parts of it. As such, we do not anticipate that the
A distinct feature that makes the data suitable to build individual measurement errors due to uneven tracking will affect our analysis of
attrition profiles for teachers is that there is a unique teacher identifi- the attrition profiles of various groups of teachers or the correlates of
cation (ID) that is, at least in theory, consistently used to identify teacher attrition at the school level.2
teachers across various census years. Therefore, as long as the IDs are Another issue with the data is that the total number of schools in the
accurately captured, it should be possible to track teachers over time. census has increased more dramatically over the six-year period than
The census captures data on all teachers employed at the school at the could be explained by a plausible trend in the construction of new
time of data collection, but it does not include any explicit data on schools. The most likely explanation is that data collection coverage
attrition such as the total number of teachers who have left the school expanded significantly over time. But there is no systematic difference in
during the census year. Therefore, we constructed a new variable to type of schools (rural versus urban) or teacher characteristics between
indicate the attrition status of each teacher in the roster between two the pool of schools across the years, as the descriptive statistics in
consecutive years. In other words, each teacher captured in the census Table 1 show. Hence, the rapid increase in the number of schools in our
data in year t is assigned one of three possible values for year t + 1: data is unlikely to affect the analysis in Section 4.

• “retained” if they remained in the same school in year t + 1,

• “switched school” if they are captured under a different school in 3.3. Description of the sample
year t + 1,
• “left the profession” if they are not captured under any school which The dataset we use for this analysis contains 54,000 teachers per year
is part of the census in year t +1 or any subsequent year until 2019. in 12,300 primary schools spread across the 13 administrative regions in
Burkina Faso (see Table 1 for the descriptive statistics of the sample).
We used the combination of teacher ID, surname and date of birth to The average teacher age is 35 years old which is in line with other an-
track teachers over census years. Once we have constructed attrition alyses of teacher demographics for the country (Evans and Mendez
variables for each teacher, we have then computed–for each school–, the Acosta, 2023a; World Bank, 2022). Almost all of the teachers have
aggregate retention, the number of teachers who switched schools, and completed at least some level of secondary education: over 80 percent of
the number of teachers who left the profession. Since we have measured the teacher population has completed lower secondary education, while
these indicators for all schools over multiple years, we are able to another 14 percent has completed upper-secondary education and a
construct a panel of school-level attrition data covering 5 years. smaller number (3 percent) has attained college degrees. In general,
teachers in Burkina Faso are considered competent—at least 88 percent
of primary school teachers in the country have the minimum required
3.2. Data limitations qualifications to teach in primary education in 2022 (UNESCO, 2023).
There are slightly fewer female teachers than male teachers (45 percent
Although the school census data is supposed to maintain a unique ID of the most recent count of teachers are female), and this is consistent for
for each teacher on the roster, there is extensive data entry error which is most regions except for the regions Cascades and Sud-oest where male
common in administrative data that is not especially cleaned with a teachers outnumber female teachers 3:2 (Appendix Table A1).
specific research objective in mind. Since the census data is not At the beginning of the period of our analysis (2014), only a fifth of
commonly used to construct teacher-level panel datasets, it appears that the teacher workforce were permanent members of the civil service
limited attention was given to the accuracy of the teacher ID data. Un- while the rest were on temporary contracts. At that time, contract
fortunately, the name and date of birth of teachers captured in the roster teachers either had to pass a competency test or undergo a pre-service
are also subject to the same data entry inaccuracies which makes them teacher training program to be eligible for a civil service position.
inadequate for tracking teachers on their own. Therefore, we had to use However, beginning in January 2016, all contract teachers were granted
teacher ID, surname and date of birth simultaneously to identify civil service status (UNESCO, 2020), which is reflected in the census
teachers and match their data across census years. data that shows almost zero contract teachers in the database from 2016
Both the individual-level and school-level attrition data are subject onwards. The change in status may have had anticipatory effects on
to some degree of measurement error due to the way the indicators were teachers’ attrition rate both for contract teachers (who might be more
constructed using the raw data in the teachers roster. Firstly, teachers likely to stay given the imminent promotion) and for civil service
who, for any reason, were not captured in subsequent survey rounds teachers (who might be inclined to leave if the new teachers are viewed
while still being in the workforce are deemed to have atritted. Secondly, as competition). For these reasons, we forego the analysis on teacher
too many observations may have failed to match across census years due retention rate by contract type.
to the stringent requirement of identifying details that is used to track Between 2014 and 2019, the number of schools in the survey have
teachers and the likelihood of data entry error in either one of the three grown by 3200 schools (an increase of 25 percent over six years).
identifying variables. Both of the above factors may have led to an Although there were several ad hoc programs of school construction (for
overestimation of aggregate attrition rates such that the rates we find example, a non-profit built 117 schools between 2014 and 2021
could be higher than other national estimates of teacher attrition. (buildOn, 2021)3), we find no evidence of a national school construction
However, there is no reason to believe that this measurement error is program that would explain this substantial increase in schools. The
correlated with any individual or school-level characteristics as it is
mostly driven by data entry errors. In Table A2 in the Appendix, we
compare the characteristics of teachers by the number of years they were 2
The measurement error seems higher for 2019 since the attrition rate for
that year is higher than the average for the earlier four years by 3.6 percentage
points. So, we tried to test the sensitivity of our findings in Section 4 to potential
measurement error by excluding 2019 from the data. There is no significant
difference in the results.
3
Another well documented program built over 120 girl-friendly schools over
10 years starting in 2006, but most of the school construction activities
happened before the time period of our analysis (Ingwersen et al. 2019).

4
B. Bedasso and A. Mendez Acosta International Journal of Educational Development 113 (2025) 103228

Table 1
Descriptive statistics for the sample.
Year

Average 2014 2015 2016 2017 2018 2019

Number of teachers 53,906 42,477 45,833 51,583 54,968 62,921 65,651

Average age 35.35 35.1 35.4 35.2 35 35.4 36
(standard deviation in parentheses) ​ (6.26) (6.21) (6.32) (6.4) (6.5) (6.5)
Education ​ ​ ​ ​ ​ ​ ​
% with junior high school 82 % 84 % 84 % 83 % 84 % 80 % 78 %
% with senior high school 14 % 12 % 13 % 13 % 14 % 15 % 17 %
% with tertiary education 3% 3% 3% 3% 3% 3% 4%
Gender ​ ​ ​ ​ ​ ​ ​
% female 46 % 42 % 44 % 47 % 48 % 49 % 49 %
Type of contract ​ ​ ​ ​ ​ ​ ​
% civil service 57 % 19 % 16 % 14 % 97 % 99 % 99 %
% temporary contract 41 % 80 % 83 % 85 % 1% 0% 0%
Number of schools 12,344 13,204 13,950 14,655 15,330 15,756 16,482
% rural 81 % 82 % 81 % 81 % 81 % 81 % 81 %
Number of violent conflict incidents by region* 5.9 ​ ​ ​ ​ ​ ​
(12.1)
Regional Human Development Index** 0.428 ​ ​ ​ ​ ​ ​
(0.085)

Source: Authors’ analysis of the underlying data as described in Section 3.

Note: The first column shows the average of the descriptive statistics from the six years in the dataset. The other columns represent the raw data from the surveys.
* Conflict data for the census years was compiled from Armed Conflict Location and Event Data (ALCED) dataset.
** Subnational HDI data for regions was obtained from the Global Data Lab.

dataset itself also reports only around 900 schools founded between
2015 and 2019. In addition, while gross primary completion rates have
slowly increased from 88 percent in 2014, up to 92 percent in 2018, and
back to 90 percent in 2019, we would expect a higher jump in enroll-
ment if there were that many new schools constructed over that time
period (World Bank, 2023). More likely, this increase reflects improved
response rate to the survey as more schools joined the annual census
data collection.
Our final dataset shows, between 2015 and 2019, about 12 percent of
teachers left the workforce (Fig. 1, Panel A). The first two years saw
slightly lower rates of attrition of around 11 percent, potentially coin-
ciding with the en masse promotion of teachers from contractual to civil
service posts. Almost all of these teachers are younger than the pre-
scribed retirement age of 60 years old when primary school teachers
start receiving their pension (Fig. 1, Panel B). The attrition rate we
calculate is similar to the 11 percent that Zeitlin (2021) finds in Rwanda,
the only other study we know of that uses EMIS data to estimate
teachers’ attrition in a Sub-Saharan Africa context. However, we note
that missing data and the difficulty of tracking teachers from year to year
in our dataset as outlined in Section 3.2 limit the representativeness of
these attrition rates such that actual teacher attrition at the national
level is likely to be lower than what we see from the dataset.

3.4. Empirical method

We examine the correlates of attrition both at the individual and

school levels. At the individual level, we seek to analyze how the pro-
pensity to attrite changes over the lifecycle of individuals for groups of
teachers with different demographic and socioeconomic characteristics.
To do this, we adopt the Age-Period-Cohort framework, a widely
recognized method for disentangling the intertwined effects of age, time
period, and birth cohort on a particular phenomenon. The method en-
ables us to describe the age profile of teachers’ attrition while also taking Fig. 1. Teacher attrition rates among primary school teachers in Burkina Faso
into account the potentially heterogeneous attrition profiles of different (2015–2019), Panel A: Annual attrition rate, Panel B: Attrition by whether the
cohorts at the same age. Moreover, it would also help us to disentangle teacher retires before or after pension age.
the time effects that are common to all age groups and birth cohorts and
are possibly caused by exogenous factors. More importantly, we would culty of identifying the age, cohort and period elements statistically
be able to describe the age, cohort and period patterns in the attrition of because the value of cohort is completely determined by age and period
various salient groups such as male and female, teachers with different (i.e. cohort = period − age). We employ the Age-Period-Cohort-
levels of education, and teaching staff with various roles. Interaction (APC-I) approach proposed by Luo and Hodges (2020) to
The longstanding challenge in implementing this model is the diffi-

5
B. Bedasso and A. Mendez Acosta International Journal of Educational Development 113 (2025) 103228

estimate the age and period effects of teachers’ attrition with the cohort Table 2
effects represented as age-period interaction. This means the cohort ef- Demographic and professional determinants of teacher retention.
fect can be identified because it is estimated indirectly without imposing (1) (2) (3) (5) (6)
additional restrictions to circumvent the collinearity issue. Specifically,
Base reference (Age: ​ ​ ​ ​ ​
we estimate the following model: 20–24)
Age: 25–29 1.44*** 1.44***
g(E(Yij )) = μ + αi + βj + αβij(k) (1)
​ ​ ​
​ (0.10) ​ ​ ​ (0.10)
Age: 30–34 1.45*** ​ ​ ​ 1.44***
Where E(Yij ) represents the expected value of teachers’ attrition for the ​ (0.10) ​ ​ ​ (0.10)
ith age group at time j while g is the link function. α denotes the age Age: 35–39 1.23*** ​ ​ ​ 1.19***
effect whereas β stands for the period effect. αβij(k) denotes the interac- ​ (0.08) ​ ​ ​ (0.08)
Age: 40–44 1.03 ​ ​ ​ 0.98
tion of the ith age-group and jth period, corresponding to the effect of ​ (0.07) ​ ​ ​ (0.07)
the kth cohort. Age: 45–49 0.88* ​ ​ ​ 0.83***
Since individual teacher attrition is a binary variable, the APC-I ​ (0.06) ​ ​ ​ (0.06)
Age: 50–54 0.79*** 0.74***
model in (1) can be estimated using a binary logit model given bellow: ​ ​ ​
​ (0.06) ​ ​ ​ (0.06)
( π ) Age: 55–59 0.53*** 0.50***
ln
​ ​ ​
= μ + αi + βj + αβij(k) (0.05) (0.05)
1− π ​ ​ ​ ​
Female ​ 0.89*** ​ ​ 0.92***
The second level of analysis consists of estimating the correlates of ​ ​ (0.01) ​ ​ (0.01)
school-level attrition. We employ standard panel data techniques to Base reference (Educational attainment: Junior ​ ​ ​
estimate the predictors of teacher attrition either in the form of high school)
Senior/technical high 1.09*** 1.05***
switching school or entirely leaving the profession. The underlying
​ ​ ​
school
specification takes the form of a standard panel data model with time- ​ ​ ​ (0.02) ​ (0.02)
invariant variables: Tertiary education ​ ​ 0.80*** ​ 0.83***
​ ​ ​ (0.03) ​ (0.03)
yit = Xʹit β + Zi γ + μi + εit (2) Base reference (Personnel type: ​ ​ ​ ​
Assistant teachers)
Where yit is teacher attrition rate at school i in year t, Xit is a vector of Regular teachers ​ ​ ​ 1.39*** 1.40***
time varying covariates and Zi is cross-sectional time-invariant vari- ​ ​ ​ ​ (0.03) (0.03)
School Principals 1.18*** 1.11***
ables. Estimating the coefficients in (2) using a standard fixed effects
​ ​ ​
​ ​ ​ ​ (0.02) (0.02)
estimator would remove any time-invariant characteristics of schools Constant 7.16*** 9.13*** 8.84*** 7.14*** 6.66***
which we expect to predict teacher attrition, such as geographic location ​ (0.48) (0.16) (0.14) (0.16) (0.47)
and public or private ownership. Therefore, we employ the Hausamn- Year fixed effects Yes Yes Yes Yes Yes
Observations 266,299 266,299 262,297 265,944 261,947
Taylor estimator which would enable us to recover the coefficients of
some of the time-invariant variables at the same time as accounting for Note: The table shows the odd ratios of a teacher staying in the workforce given
unobserved individual heterogeneity (Hausman and Taylor, 1981). several demographic and professional factors. All specifications include year
fixed effects. Robust standard errors (in exponentiated form) in parentheses. ***
4. Results p < 0.01, ** p < 0.05, * p < 0.1

4.1. Attrition rates by personal characteristics 0.01) (Table 2, column 2). When gender is plotted against age based on
the age-period-cohort estimates, we see that female teachers are more
Age. Younger teachers in the age group of 25–35 years are the most likely to leave at every age group, and the gender gap widens for
likely to remain in the workforce with a retention rate of 90 percent (44 teachers older than 40 years old (Fig. 3). Particularly, the widening gap
percent higher odds of staying than the reference age group 20–24, p- over the course of the 40 s might suggest that women are more likely to
value < 0.01, Table 2 column 1). Teachers above 39 years old see pro- drop out of the workforce than men as family responsibilities accumu-
gressively lower rates of retention (Fig. 2, Panel A). There is a decline in late. In Sub-Saharan Africa, women cite unpaid care work as the reason
retention from 83 percent for teachers age 50–54 to 77 percent for for not participating in the workforce (34 percent) (Addati et al., 2018).
teachers age 55–59, signaling early retirement for teachers in this For women aged 40 years and above, this unpaid care may take the form
cohort. The trend—where retention starts relatively low for the first five of caring for aging parents and young grandchildren rather than their
years in teaching, peaks for teachers with 5–15 years of experience and own children4 and may mean moving to part time work or the informal
then steadily declines henceforth—is true for any year between 2015 sector. This finding is opposite to what we see in Rwanda (Zeitlin, 2021)
and 2019 (Fig. 2, Panel B). The lower than average retention rates in and Kenya (Mabeya 2021), suggesting that the dynamics of gender affect
2019 could potentially be attributed to a combination of factors. Firstly, job choices differently for Burkina Faso.
the 2019 data does not benefit from the presence of data in subsequent Education. Teachers with university level education have 20 percent
years which could be used to verify whether missing data is due to actual lower odds of staying in the workforce on average (p-value < 0.01),
attrition or one-off data entry error in that round. Hence, attrition rates especially in the earlier years, compared to teachers with only junior
for 2019 are likely to be overstated. Secondly, the escalation of the high school (Table 2, column 3; and Fig. 4). Interacted with age, we find
conflict that had broken out first in 2016 and continued to escalate in this trend to hold true across the age groups except for teachers 50 years
later years may have contributed to higher attrition of teachers (Seibert, and older, when teachers with the highest education levels are most
2020). The higher exit rate for early career teachers is consistent with likely to be retained (Fig. 4).
what we see in Rwanda where 40 percent of teachers in their first years
resign their posts and attrition remains above 10 percent in the first five
years of teaching (Zeitlin, 2021), and opposite to what we see in Kenya
4
where older teachers (albeit, secondary school teachers) are more likely Median age of women in Burkina Faso on the birth of their first child is 20
to attrite (Mabeya 2021). years old (INSD and ICF. 2021). At 40 years old, women are more likely to bear
the burden of caring for both elderly parents and their grandchildren especially
Gender. Female teachers have 11 percent lower odds of staying in the
given the importance of grandparents’ role in childcare in the region (Hatch
public teacher workforce than male teachers on average (p-value <
and Posel, 2018; Mhaka-Mutepfa et al. 2017)

6
B. Bedasso and A. Mendez Acosta International Journal of Educational Development 113 (2025) 103228

Fig. 2. Teacher attrition rates among primary school teachers in Burkina Faso by age group, Panel A: Attrition by age group, combined for all years, Panel B: Attrition
by age group, by year, Note: The dashed line shows the average retention for the whole group. Age group 20–24 is the reference category.

Fig. 3. Teacher attrition rates among primary school teachers in Burkina Faso by gender.

7
B. Bedasso and A. Mendez Acosta International Journal of Educational Development 113 (2025) 103228

Fig. 4. Teacher attrition rates among primary school teachers in Burkina Faso by educational attainment.

Function. Assistant teachers who are below regular teachers in the 4.2. Attrition rates by school characteristics
career ladder are more likely to leave than the regular teachers (odds
ratio of regular teachers staying compared to assistant teachers = 1.39, We also report school-level correlates of the annual attrition rate
p-value < 0.01) except for the oldest age group (55 years and up). This is (share of teachers leaving the profession) and annual school-level
likely because main teachers leave the workforce to pensionable early turnover (share of teachers switching schools) (Table 3). We show
retirement while assistant teachers may find financial incentives to stay province level teacher attrition and switching rates in Fig. 6. For the
active in the workforce (Fig. 5). School principals enjoy the highest rate education system as a whole, the relevant metric is the proportion of
of retention across all ages which suggests that the usually higher salary teachers who leave the workforce entirely. However, for a particular
and benefits, elevated social standing, and the non-teaching tasks school, what really matters as a measure of attrition is the proportion of
associated with management positions make staying attractive. teachers who leave the school whether they are leaving the profession or
Taken together, we see that teachers in the first five years of their transferring to another school. Therefore, we estimate the predictors of
career and teachers with more advanced educational attainment—pre- both forms of attrition to obtain a complete picture of which types of
cisely the teachers that may have access to better paying opportunities schools are likely to be affected by teacher turnover.
either outside the education sector or outside the civil service—have a We find that teachers are more likely to leave the profession from
relatively high likelihood of leaving. Conditional on not leaving early in schools with higher provincial incidence of violent conflict.6 A one
their careers, teachers with tertiary degrees often have high retention in standard deviation change in conflict incidence is associated with an 18
their later years, which coincides with the higher retention of teachers in percent increase in the share of teachers who leave the workforce from a
higher management positions (positions that often require more given school. Similarly, teachers employed in urban schools are more
advanced degrees). This suggests that highly qualified teachers are more likely to leave the profession than teachers in rural schools. Specifically,
likely to stay given access to promotions. Female teachers in Burkina attrition in urban schools is 1.14 percentage points (0.09 standard de-
Faso are more likely to leave than male teachers contrary to the evidence viation) higher than attrition in rural schools. The coefficient of urban
from Rwanda (Zeitlin, 2021) and in the majority of LMICs (Le Nestour school becomes insignificant, however, when we include an indicator
and Moscoviz, 2020). for the level of regional development. Teachers based in more developed
One classic explanation to higher female teacher attrition is related provinces, as measured by the regional human development index, are
to family rearing when women leave the workforce to (temporarily) more likely to leave the profession than teachers in less advanced areas.
raise children.5 In our sample, male teachers are twice more likely than This suggests that urbanization may be standing in as a proxy for level of
female teachers to complete upper-secondary school and have tertiary development which drives attrition by increasing the availability of
degrees (both genders are equally likely to complete lower-secondary more lucrative labor market options other than teaching. Taken
education) and only 10 percent of school principals are women. This together, the results with respect to the conflict and regional develop-
points to limited career progression for female teachers in the teacher ment dimensions of school location indicate that teacher attrition is
workforce, such that they may be the parent of choice to stay at home. partly governed by an interplay of push and pull factors. The map in
However, regressing probability of retention with age, gender, educa- Panel A of Fig. 6 shows that the province hosting the capital, Ouaga-
tional attainment, and position shows that the gender influence is still dougou, is one of the regions characterized by the highest percentage of
statistically significant beyond these factors. We are unable to verify teachers leaving the workforce.
either the parental status of leaving teachers or to track teachers who The other factor that might predict the percentage of teachers who
leave and return to the service, either of which can indicate evidence for are leaving the workforce is the type of school ownership. Around 1 in 4
female teachers exiting the workforce to raise children. of the primary schools in our panel dataset is either a private or religious
school whereas the other 3 are public schools. The results in Table 3

6
The types of violence conflict include political violence, demonstrations
5
For example, a study in rural Pakistan reports that more than a quarter of and related strategic developments such as battles, explosions or remote
female teachers plan to retire from teaching after getting married and almost violence, violence against civilians with or without reported fatality, riots,
half plan to stop if their spouses ask them to stop teaching to care for their looting or property destruction, and related events as reported in the Armed
families (de Talancé 2017). Conflict Location & Event Data Project (ACLED, 2024).

8
B. Bedasso and A. Mendez Acosta International Journal of Educational Development 113 (2025) 103228

Fig. 5. Teacher attrition rates among primary school teachers in Burkina Faso by function.

Table 3
Determinants of school-level teacher attrition.
Percentage left Percentage left Percentage Percentage Percentage Percentage Percentage Percentage
profession profession switched switched switched school switched school switched school switched school
school school (Male) (Male) (Female) (Female)

Urban school 1.41*** − 0.348 − 12.1*** − 10.0*** − 10.11*** − 7.78*** − 1.35 − 2.11 * *
(0.406) (0.869) (0.716) (0.733) (0.914) (0.912) (0.93) (0.921)
Regional Human ​ 31.0** ​ − 147.3*** ​ − 125.0*** ​ − 50.0 * **
Development ​ (14.9) ​ (9.42) ​ (10.3) ​ (10.3)
Index
Incidence of violent 1.57*** 1.51*** − 3.17*** − 2.48*** − 4.07*** − 3.47*** − 3.19*** − 2.81
conflict in (0.179) (0.183) (0.297) (0.314) (0.326) (0.34) (0.324) (0.335)
province (log)
Public school − 10.7*** − 9.11*** 3.03*** − 1.64 1.78 3.7 31.1*** 32.2 * **
dummy (0.728) (0.936) (1.44) (1.46) (2.41) (2.36) (2.97) (2.87)
Controls School infrastructure, school age, school management committee, share of teachers over 40 years old
Estimation method Hausman-Taylor
Region dummies Yes No Yes No Yes No Yes No
Observations 42,072 42,072 39,355 39,355 35,987 35,987 26,967 26,967

Note: Robust standard errors in parentheses. *** p < 0.01, ** p < 0.05, * p < 0.1

show that teachers in public schools are almost 10 percent less likely to moving to small towns and larger cities as part of the internal transfer
leave the profession than teachers in private or religious schools. This process (Sirois, 2017). The map in Panel B of Fig. 6 shows that provinces
result is not surprising because teachers are more likely to stay in gov- in the northern part of the country tend to experience a higher rate of
ernment funded schools which are historically better endowed than teachers switching schools.
private institutions. In addition to endowment, public schools might be It is curious that schools situated in high-conflict regions exhibit
better managed which, in turn, attract higher skilled and more moti- lower rates of teacher-switching compared to those in relatively safer
vated teachers (Leaver et al., 2019). Except for a few elite private schools areas. One way of interpreting this finding is that teachers in conflict-
in large cities, private low-fee schools often have less favorable affected areas are more likely to leave the workforce and the respec-
employment conditions for teachers.7 tive area altogether, as confirmed by the previous results, than trying to
What type of schools do teachers transfer away from? Unsurpris- switch schools. It could also be the case that administrative structures
ingly, rural schools and schools in less developed regions all see statis- that would normally be responsible for teacher transfer processes are
tically significant higher rates of teacher switching. This means that affected by the conflict that switching schools through the internal
teachers tend to transfer away from rural and remote schools to urban transfer process is less frequent in conflict-affected areas. Public schools
schools in more developed regions before they leave the teaching are also more likely to see lower levels of teacher switching schools, but
workforce, presumably for greener pastures such locations may offer. the effect is not precise.
This trend of career progression is confirmed by a detailed study of the Male teachers are more sensitive to these school correlates than fe-
employment histories of a sample of teachers in Burkina Faso which male teachers: for example, male teachers are up to three times more
shows that entry-level teachers often start in rural areas before gradually likely to transfer away from schools in rural areas (-7.78 percentage
points decrease in probability to switch schools for male teachers given
that the school is in an urban location versus − 2.11 percentage points
for female teachers) and twice more likely to do so for schools with
7
In India, private school teachers earn less than a sixth of what civil service lower regional development index (10 percentage points versus 4 per-
teachers do (Muralidharan and Sundararaman, 2013), and in Kenya, teachers of centage points per standard deviation). The incidence of violence in the
the same qualifications earn twice as much working for the public service than
province is a strong predictor for switching schools for male teachers but
the private sector (Barton, Bold, and Sandefur, 2017).

9
B. Bedasso and A. Mendez Acosta International Journal of Educational Development 113 (2025) 103228

Panel A: Teacher attrition

Fig. 6. Teacher attrition and switching rates by province, Panel A: Teacher attrition, Panel B: Teacher switching rate, Note: The rates displayed in the figures are
simple averages of school-level attrition and switching rates for each administrative province.

10
B. Bedasso and A. Mendez Acosta International Journal of Educational Development 113 (2025) 103228

Table 4
Profiles of teachers most likely to attrite in Burkina Faso, potential factors that drive attrition, and suggested policy solutions.
Profile of teachers who are Type of teacher attrition driver drawing from earlier works on the Potential policy solutions
likely to attrite in Burkina conceptual framing of teacher attrition (UNESCO, 2010, Nguyen,
Faso 2020, See et al. 2020)

Early career teachers (first - Opportunities outside of teaching Policy recommendation #1: Financial and non-financial incentivesPolicy
five years of teaching) - Dissatisfaction with working conditions/professional growth recommendation #2: Opportunities for teacher professional development and
Teachers with tertiary improved teacher working conditionsPolicy recommendation #3:Building on
education motivations and improving overall job satisfaction
Teachers with lower
positions
Female teachers - Gender gap in working conditions Policy recommendation #4: Efforts to recruit female teachers
- Personal factors, social dynamics and expectations

not for female teachers. Opportunity rather than preference may explain Uruguay improved retention for teachers but only for those later in their
the gender gap in the decision to switch: male teachers are more likely to careers such that the average teacher tenure increased by one year for
have higher educational attainment and hold higher teaching positions, beneficiary schools (Cabrera and Webbink, 2020). A similar program
allowing them to more easily switch schools than their female peers. that provided a wage premium of between a quarter to a third of current
salaries improved teacher turnover in Brazil by 5 percentage points
5. Policies for improving teacher retention (Camelo and Ponczek, 2021). Another incentive program, although of
lesser monetary value and this time in Chile, did not change teacher
The descriptive nature of this study allows us to profile teachers most retention but in fact reduced hours worked by teachers (Hinze-Pifer and
likely to attrite according to several demographic factors (age, gender, Méndez, 2016). The evidence from the Sub-Saharan Africa region shows
tenure, and qualifications). In the absence of additional information on mixed results: a 20 percent rural hardship allowance in Zambia led to
their motivations, such as qualitative surveys, discrete choice experi- more teachers but no effect on average student scores (Chelwa et al.,
ments, or impact evaluations of interventions, we are not able to report 2019), and a similar hardship allowance of 30–40 percent in Gambia led
on the exact mechanism of decision-making for these teachers. We to 10 percentage point more (and better qualified) teachers in remote
therefore rely on the established literature that propose conceptual areas (Pugatch and Schroeder 2014), but with also no effect on average
frameworks to understand teacher attrition (UNESCO, 2010, Nguyen, test scores (Pugatch and Schroeder 2018).
2020, See et al., 2020) to draw potential links between the teacher As countries often have different teacher pay contexts (how teacher
attrition profiles and what might be driving these teachers to quit pay compares to other professions in the country and how that pay
(Table 4). The high attrition among young teachers early in their ca- adequately provides for an acceptable standard of living within the local
reers, teachers with tertiary education, and those in currently low po- context), financial incentives may have different mileage in changing
sitions suggest that factors such as compelling opportunities outside of exit behavior. Since Burkina Faso’s teachers are relatively well paid
teaching combined with potential unsatisfactory working conditions compared to other similarly educated professionals, at least for teachers
and professional growth may affect the decision to attrite. These drivers who have secondary education, financial incentives may work best
then can be linked to corresponding policies that improve teacher when targeting teachers either in the first five years of teaching (when
working conditions and benefits to be at par with other competing op- probability of attrition is high) or those with higher educational
portunities. Similarly, the higher likelihood of attrition for female attainment, potentially tied with opportunities for professional devel-
teachers suggests a gender-driven difference in working conditions opment or access to management positions.
and/or social and cultural expectations, which require policies that The current wage structure for teachers in Burkina Faso allows for
specifically overcome these barriers to retention. government-provided housing for teachers assigned to schools in rural
In this section, we discuss potential policy options in addressing areas, and a housing allowance to compensate if there are no appro-
teacher attrition and the scope of evidence behind these interventions. priate accommodations available, in addition to a “hardship allowance”,
Across these policy options, only the monetary incentives have a robust a special allowance based on the poverty zone of the school. All of these
evidence base as shown by a few systematic reviews in the topic (Evans bonuses can represent up to 25 percent of a new teacher’s salary (Sirois,
and Mendez Acosta, 2023b; See et al., 2020). The other policies, while 2017). However, teacher personal preferences may also differ (e.g.
promising in the contexts in which they were implemented, have more teachers from those rural regions may require less incentive to teach
limited evidentiary support in diverse contexts. Nevertheless, we believe there because of family ties) such that incentives might be more
there is still merit in expanding the tools policymakers may use to cost-effective if they are also designed to respond to teacher preferences
improve retention to include these measures (as may be appropriate in and not just on rigid school-level characteristics (Bobba et al., 2022).
their contexts), given the interaction of personal, financial, and struc-
tural factors affecting teacher attrition as discussed in Section 2.
5.2. Opportunities for teacher professional development and improved
teacher working conditions
5.1. Financial and non-financial incentives
There is far less evidence on improving retention through teacher
Providing financial benefits and related perks such as housing, professional development, and those that are available vary in out-
transportation, and allowances is the most straightforward to implement comes, but the particularly successful programs, at least in terms of
and also the most evaluated policy option in improving teacher reten- improving student outcomes, often link participation in professional
tion especially for particularly hard-to-staff schools or schools that see development programs to incentives such as promotions (Popova et al.,
higher turnover (Evans and Mendez Acosta, 2023b; See et al., 2020). 2022; Quota and Bhatia 2022; Bennell and Akyeampong 2007). A survey
Most of the evidence for financial incentives point to positive outcomes of over 130 public expenditure reports across low and middle income
on attracting and keeping teachers, but results vary depending on countries find one fifth of the reports mention in-service trainings, often
implementation, and the effectiveness of these programs may only last in the context of their limited availability and the gap in participation in
as long as the incentives are present. An incentive of up to 26 percent these trainings by female teachers and those from poorer communities
increase in salary for teachers working in poor neighborhoods in (Evans and Mendez Acosta, 2023c). As such, improving in-service

11
B. Bedasso and A. Mendez Acosta International Journal of Educational Development 113 (2025) 103228

trainings, through coaching and mentoring, structured pedagogy, and Recruiting teachers locally also has the advantage of potentially keeping
appropriate use of technology aids, may help improve the quality of female teachers in the workforce even as they raise families since they
existing teachers and make career promotions more accessible and are more likely to have strong social support in their families within the
equitable. local community. These policy interventions would particularly be
Given that teachers with management positions (such as school helpful in the case of Burkina Faso where female teachers have much
principals) are much more likely to be retained than regular teachers in lower retention rate than male teachers.
Burkina Faso, ensuring access to teacher professional development,
especially those tied to promotion, may help encourage retention for 6. Conclusion
qualified teachers especially those with tertiary education who are more
likely to attrite. Similarly, ensuring ample opportunities for promotion Teacher attrition can be a significant challenge particularly if it leads
and professional development for early career teachers may help miti- to the exit of effective teachers and in light of the increasing teacher
gate the high attrition rate for teachers in their first five years in the demand driven by growing school population in most low- and middle-
profession. income countries such as Burkina Faso. A nuanced understanding of the
characteristics of the teachers who are likely to leave and the schools
5.3. Building on motivations and improving overall job satisfaction that are more prone to suffer from attrition is an important step in
addressing teacher attrition. Surprisingly, there has been very limited
A synthesis of over 200 studies globally on who becomes a teacher effort so far to utilize administrative data, such as annual school census
and why documents the presence of intrinsic and altruistic motives to data, to study teacher attrition in regions such as Sub-Saharan Africa.
teaching in addition to extrinsic motivations (See et al., 2022). For the This paper has used six years of school census data to examine the de-
group that would have gone to teaching but decided not to, the social mographic and geographic correlates of teacher attrition in Burkina
status of the profession is rated as important, alongside the working Faso.
environment and salary (See et al., 2022). High-salary and social pres- In line with the literature on competing employment opportunities
tige do not necessarily go together: the Global Teacher Status Index outside of teaching and family priorities, we find that younger teachers
ranks countries based on the society’s respect to teachers, and there is no in their first five years of teaching, teachers with lower positions, and
clear indication that countries that pay high teacher wages respect them teachers who have completed tertiary education are more likely to quit
more (see Appendix Figure A4). the profession. We also find that female teachers are more likely to leave
Changing social perceptions is less straightforward than increasing than male teachers, suggesting gender as a crucial factor in job choices.
salaries, but there is some evidence that low-cost behavioral nudges may In terms of school characteristics, rural and remote schools are more
have an impact on tapping into teachers’ altruistic motives. For likely to lose teachers to other schools while schools in more developed
example, a nationwide teacher recruitment campaign in Peru encour- areas — ostensibly with more active industries and job markets — tend
aged teachers to apply for more disadvantaged/remote schools to lose teachers to options outside of teaching.
(Ajzenman et al., 2024). They sent out messages and pop-outs in the The findings suggest the need for teacher retention policies to
online application portal either highlighting the altruistic nature of especially target women and young teachers who are in the first few
teaching (i.e. “Thank you for being an agent of social change”) or pro- years of their careers. Additionally, the drivers of attrition and the
moting an existing monetary incentive when they apply to disadvan- destinations of teachers leaving schools in urban and more developed
taged schools. Both interventions improved applications to remote areas are likely to be different from those in rural and remote locations.
schools, especially for high-performing teachers that received the Accordingly, the policy options to address attrition may also need to be
“altruistic” messages. A similar program in Chile had the opposite effect differentiated based on the specific needs of various geographic areas
where high-performing students are less likely to be attracted to and settings. As localized conflict and fragility continue to compound
teaching when primed with “altruistic messages,” while low-performers the effects of existing disadvantages on teacher attrition in certain areas,
(ostensible from disadvantaged households) are more likely to respond it is even more urgent to devise policy solutions that take these de-
to messages promoting monetary rewards to teaching (Ajzenman et al., mographic and geographic factors into account.
2021).
Neither motivation to teach nor job satisfaction is included in our Funding
analysis as determinants of attrition (since our study is based on
administrative data), but these factors are likely candidates in influ- This work is made possible through funding from the Bill and
encing the decision to switch or leave the profession for future research. Melinda Gates Foundation.

5.4. Efforts to recruit female teachers CRediT authorship contribution statement

Female teachers encourage enrollment and improve student learning Amina Mendez Acosta: Writing – review & editing, Writing –
outcomes especially for girls by serving as role models and encouraging original draft, Visualization, Investigation, Data curation. Bedasso
higher aspirations (Muralidharan and Sheth, 2016; Haugen et al., 2014; Biniam: Writing – original draft, Supervision, Methodology, Data
Eble and Hu, 2020; Lim and Meer, 2020). However, female teachers are curation, Conceptualization.
historically difficult to attract and retain in remote schools where they
may be needed the most because of the lack of infrastructure such as
housing and transportation or concerns over their personal safety in Conflicts of interest statement
areas of conflict (UNESCO, 2019). In addition to policies that address the
general teacher working conditions that affect all teachers, efforts that The authors declare no competing interest.
specifically address the barriers female teachers face may help improve
teacher gender parity. Surveys of female teachers in Ghana and Kenya Acknowledgement
report that incentives such as study leave with pay, expedited promo-
tion, provision of housing, and hardship allowance may make remote We would like to thank Maimouna Konate for excellent research
schools more attractive for teaching (Gad, 2015; Kamere et al., 2019). assistance. All errors are our own.

12
B. Bedasso and A. Mendez Acosta International Journal of Educational Development 113 (2025) 103228

Appendix Tables and Figures

Appendix Table A1
Descriptive statistics by region

Number of Number of Proportion of Proportion of schools Teacher Teacher attrition Teacher attrition
schools teachers female teachers considered to be in a rural attrition rate rate (female rate (male teachers)
location teachers)

BOUCLE DU 1382 7076 41 % 92 % 8% 6% 6%

MOUHOU
CASCADES 658 3072 39 % 89 % 6% 5% 4%
CENTRE 1850 5810 51 % 28 % 13 % 10 % 9%
CENTRE-EST 1217 6443 51 % 82 % 8% 6% 6%
CENTRE-NORD 1375 5842 45 % 91 % 8% 6% 5%
CENTRE- 1283 8019 45 % 90 % 9% 6% 6%
OUEST
CENTRE-SUD 670 3875 49 % 83 % 10 % 7% 7%
EST 1267 6787 44 % 92 % 11 % 8% 7%
HAUTS- 1538 7099 43 % 68 % 10 % 8% 7%
BASSINS
NORD 1654 7562 45 % 91 % 9% 6% 6%
PLATEAU 967 4043 50 % 93 % 9% 6% 6%
CENTRAL
SAHEL 1037 4936 40 % 95 % 8% 6% 5%
SUD-OUEST 858 4533 39 % 90 % 11 % 7% 7%
ALL REGIONS 15,756 75,097 45 % 81 % 9% 7% 6%
Note: Attrition rates are regional averages of school-level attrition and represent attrition rates for five years (2014–2018).

Appendix Table A2
Descriptive statistics by tracking status

Number of appearances in matched dataset (number

of years a teacher was tracked)

1 2 3 4 5 Average

Number of teachers 29670 28731 14336 16044 25651 53,906

Average age 35.74 34.48 33.67 35.22 37.06 35.35
Education ​
% with junior high school 80.59 82.45 82.2 83.35 82.85 82
% with senior high school 15.97 14.27 14.68 13.72 13.88 14
% with tertiary education 3.45 3.28 3.12 2.94 3.27 3
Gender ​
% female 45.54 52.08 49.63 47.5 46.02 49
School location ​
% rural 77.39 81.27 85.26 83.04 78.79 81

13
B. Bedasso and A. Mendez Acosta International Journal of Educational Development 113 (2025) 103228

20%

15.7% 16.1%
14.5%
15%
10.8%
10% 8.7% 9.8%
7.9%

5.5% 6.0%
5.4%
5% 3.8% 4.4%
1.5%
0.2%
0%
Average

Niger

Seychelles

Rwanda
Tanzania

Togo

Djibouti

Côte d'Ivoire

Sierra Leone

Mauritania
Cabo Verde

Benin
Mauritius

Burkina Faso

Figure A1. Teacher attrition rates among primary school teachers in several countries in Sub-Saharan Africa (2018–2022), Source: UIS Statistics (UNESCO, 2020),
Note: These are the only countries from the region with teacher attrition rates reported in the database for the years 2018–2022

Burkina Faso 40
Primary

Sub-Saharan Africa (exc high income) 37

Low income 40

Burkina Faso 23
Secondary

Sub-Saharan Africa (exc high income) 22

Low income 22

- 5 10 15 20 25 30 35 40 45
Figure A2. Pupil-teacher ratios at the primary and secondary level for Burkina Faso, low- and middle- income countries in Sub-Saharan Africa and low-income
countries (2018 rates), Source: World Bank Development Indicators (World Bank, 2023)

14
B. Bedasso and A. Mendez Acosta International Journal of Educational Development 113 (2025) 103228

Panel A: Primary completion rates

80%
71% 73%
69% 69% 70%
67%
64% 64%
59%
60%

40%

20%

0%
Burkina Faso Sub-Saharan Africa (exc high Low income
income)

Primary completion rate Primary completion rate, female Primary completion rate, male

Panel B: Secondary completion rates

60%

45% 46%
43% 43%
41% 41%
38% 38%
40%
34%

20%

0%
Burkina Faso Sub-Saharan Africa (exc high Low income
income)

Secondary completion rate Secondary completion rate, female

Secondary completion rate, male
Figure A3. Completion rates for Burkina Faso, low- and middle- income countries in Sub-Saharan Africa and low-income countries (2021 rates), Panel A: Primary
completion rates., Panel B: Secondary completion rates, Source: World Bank Development Indicators (World Bank, 2023)

15
B. Bedasso and A. Mendez Acosta International Journal of Educational Development 113 (2025) 103228

100,000

Teacher salary ($USD, PPP adjusted)

80,000

60,000

40,000

20,000

Turkey

France

Netherlands
Taiwan
Russia

Korea

Singapore

Greece

Finland
Japan
Egypt

Germany
Chile
Portugal

Hungary
China

Indonesia

India

Canada

Switzerland
Panarna
United States

peru
Colombia
Spain

Czech Republic

Ghana
Italy
Israel
Brazil
Malaysia

Uganda

Argentina
New Zealand

United Kingdom
Figure A4. Index ranking of countries according to social prestige of teachers against teacher salary in that country., Source: Authors’ figure from data from Global
Teacher Status Index 2018 (Dolton et al., 2018), Note: Countries are listed from highest social prestige (left, China) to lowest social prestige (right, Brazil). The
trendline displayed is a linear estimation of the relationship between the two variables

References Crawfurd, Lee, and Todd Pugatch. 2022. Teacher Labor Markets in Developing Countries.
In The Routledge Handbook of the Economics of Education, by Brian P. McCall.
Routledge. 〈https://www.iza.org/publications/dp/12985/teacher-labor-markets-in-
ACLED. 2024. Armed Conflict Location & Events Data. ACLED. May 29, 2024. 〈https
developing-countries〉.
://acleddata.com/〉.
Dolton, Peter, Oscar Marcenaro, Robert De. Vries, and Po-Wen She. 2018. Global Teacher
Addati, Laura, Umberto Cattaneo, Valeria Esquivel, and Isabel Valarino. 2018. Care
Status Index 2018 | Unesco IIEP Learning Portal. 〈https://learningportal.iiep.unesco.
Work and Care Jobs for the Future of Decent Work | International Labour
org/en/library/global-teacher-status-index-2018〉.
Organization. 〈https://www.ilo.org/publications/major-publications/care-work-an
Eble, Alex, Hu, Feng, 2020. Child beliefs, societal beliefs, and teacher-student identity
d-care-jobs-future-decent-work〉.
match. Econ. Educ. Rev. 77 (August), 101994. https://doi.org/10.1016/j.
Ajzenman, Nicolás, Bertoni, Eleonora, Elacqua, Gregory, Marotta, Luana, Méndez
econedurev.2020.101994.
Vargas, Carolina, 2024. Altruism or Money? Reducing Teacher Sorting Using
Evans, David K., Mendez Acosta, Amina. 2023a. Africa’s Looming Recruitment Crisis:
Behavioral Strategies in Peru. Journal of Labor Economics 42 (4). https://doi.org/
Teacher Demand in the Coming Decades. Unpublished manuscript. Center for Global
10.1086/725166.
Development., and .
Ajzenman, Nicolás, Elacqua, Gregory, Hincapié, Diana, Jaimovich, Analía, López
Evans, David K., Mendez Acosta, Amina, 2023b. How to recruit teachers for hard-to-staff
Boo, Florencia, Paredes, Diana, Román, Alonso, 2021. Career Choice Motivation
schools: a systematic review of evidence from low- and middle-income countries.
Using Behavioral Strategies.. October Economics of Education Review 84, 102173.
Econ. Educ. Rev. 95 (August), 102430. https://doi.org/10.1016/j.
https://doi.org/10.1016/j.econedurev.2021.102173.
econedurev.2023.102430.
Barton, Nicholas, Bold, Tessa, Sandefur, Justin, 2017. Measuring rents from public
Evans, David K., Mendez Acosta, Amina. 2023c. Teacher Working Conditions in Low- and
employment: regression discontinuity evidence from Kenya. Working Paper.
Middle-Income Countries: Lessons from 130 Public Expenditure Surveys.
Rochester, NY. https://doi.org/10.2139/ssrn.3013390.
Unpublished manuscript. Center for Global Development..
Bennell, Paul, 2023. Dissatisfied but nowhere to go: teacher attrition in the context of
Evans, David K., Yuan, Fei, Filmer, Deon, 2022. Teacher pay in Africa: evidence from 15
mass unemployment in Sub-Saharan Africa. Glob. Soc. Educ. 1–12. https://doi.org/
countries. World Dev. 155 (July), 105893. https://doi.org/10.1016/j.
10.1080/14767724.2023.2233481.
worlddev.2022.105893.
Bennell, Paul, and Kwame Akyeampong. 2007. “Teacher Motivation in Sub-Saharan
Gad, Boadum Kusi. 2015. Recruitment and Retention of Public Sector Teachers in Ghana:
Africa and South Asia.” Researching the Issues No. 71. Department for International
A Discrete Choice Experiment. University of Ghana. 〈http://ugspace.ug.edu.
Development.
gh/bitstream/handle/123456789/8207/BOADUM%20KUSI%20GAD_RECRUITME
Bietenbeck, Jan, Irmert, Natalie, Sepahvand, Mohammad H., 2023. Teacher quality and
NT%20AND%20RETENTION%20OF%20PUBLIC%20SECTOR%20TEACHERS%
cross-country differences in learning in francophone Sub-Saharan Africa. Econ. Educ.
20IN%20GHANA%20A%20DISCRETE%20CHOICE%20EXPERIMENT_2015.pdf?se
Rev. 96 (October), 102437. https://doi.org/10.1016/j.econedurev.2023.102437.
quence= 1&isAllowed=y〉.
Bobba, Matteo, Tim Ederer, Gianmarco Leon-Ciliotta, Christopher Neilson, and Marco G.
Global Coalition to Protect Education from Attack. 2024. “Country Profile: Burkina
Nieddu. 2022. Teacher Compensation and Structural Inequality: Evidence from
Faso.” 〈https://protectingeducation.org/wp-content/uploads/eua_2024_burkinafa
Centralized Teacher School Choice in Peru. Working Paper 29068. 〈https://christ
so.pdf〉.
opherneilson.github.io/work/documents/TeacherSchoolChoice/BELNN_Feb2022.
Hatch, Michelle, Posel, Dorrit, 2018. Who cares for children? A quantitative study of
pdf〉.
childcare in South Africa. Dev. South. Afr. 35 (2), 267–282. https://doi.org/
buildOn. 2021. buildOn Burkina Faso. 〈https://www.buildon.org/wp-content/uploa
10.1080/0376835X.2018.1452716.
ds/2021/06/Burkina-Faso.pdf〉.
Haugen, Caitlin S., Klees, Steven J., Stromquist, Nelly P., Lin, Jing, Choti, Truphena,
Byemelwa, Adonis. 2024. “Tanzania’s Bold Move: 11,015 New Teaching Positions to
Corneilse, Carol, 2014. Increasing the number of female primary school teachers in
Tackle Nationwide Education Crisis - PAN AFRICAN VISIONS.” Pan African Visions,
African countries: effects, barriers and policies. Int. Rev. Educ. 60 (6), 753–776.
July 21, 2024, sec. Algeria. 〈https://panafricanvisions.com/2024/07/tanzanias-bo
https://doi.org/10.1007/s11159-014-9450-0.
ld-move-11015-new-teaching-positions-to-tackle-nationwide-education-crisis/〉.
Hausman, Jerry A., Taylor, William E., 1981. Panel data and unobservable individual
Cabrera, José María, Webbink, Dinand, 2020. Do higher salaries yield better teachers and
effects. Econometrica 49 (6), 1377–1398. https://doi.org/10.2307/1911406.
better student outcomes? J. Hum. Resour. 55 (4), 1222–1257. https://doi.org/
Hinze-Pifer, Rebecca, Méndez, Carolina, 2016. Evaluating the Impact of Supplemental
10.3368/jhr.55.4.0717-8911R3.
Funds on Disadvantaged Schools: Evidence from Chile. University of Chicago. 〈htt
Caillods, Françoise. 2011. “Financing Secondary Education in Selected Francophone
ps://aefpweb.org/sites/default/files/webform/41/AEFPversion2.pdf〉.
Countries of Africa: Issues and Perspectives - UNESCO Digital Library.” In Financing
Human Rights Watch. 2020. “‘Their War Against Education’: Armed Group Attacks on
Secondary Education in Developing Countries: Strategies for Sustainable Growth.
Teachers, Students, and Schools in Burkina Faso.” Human Rights Watch. 〈https://re
〈https://unesdoc.unesco.org/ark:/48223/pf0000125723〉.
source-centre.savethechildren.net/document/their-war-against-education-armed-
Camelo, Rafael, Ponczek, Vladimir, 2021. Teacher turnover and financial incentives in
group-attacks-teachers-students-and-schools-burkina-faso/〉.
underprivileged schools: evidence from a compensation policy in a developing
Ingwersen, Nicholas, Harounan Kazianga, Leigh L. Linden, Arif Mamun, Ali Protik, and
country. Econ. Educ. Rev. 80 (February), 102067. https://doi.org/10.1016/j.
Matthew Sloan. 2019. The Long-Term Impacts of Girl-Friendly Schools: Evidence
econedurev.2020.102067.
from the BRIGHT School Construction Program in Burkina Faso. Working Paper
Castro, Juan F., Esposito, Bruno, 2022. The effect of bonuses on teacher retention and
25994. Working Paper Series. National Bureau of Economic Research. 〈https://doi.
student learning in rural schools: a story of spillovers. Educ. Financ. Policy 17 (4),
org/10.3386/w25994〉.
693–718. https://doi.org/10.1162/edfp_a_00348.
INSD, and ICF 2021. Burkina Faso Demographic and Health Survey 2021 [Dataset].
Chelwa, Grieve, Pellicer, Miquel, Maboshe, Mashekwa, 2019. Teacher Pay and
〈https://dhsprogram.com/methodology/survey/survey-display-562.cfm〉.
Educational Outcomes: Evidence from the Rural Hardship Allowance in Zambia.
Kamere, Isabela, Makatiani, M.I., Nzau, Arthur Kalanza, 2019. Policy interventions for
South African Journal of Economics 87 (3). https://doi.org/10.1111/saje.12227.
attraction and retention of female teachers in rural secondary schools: perspectives

16
B. Bedasso and A. Mendez Acosta International Journal of Educational Development 113 (2025) 103228

of rural educators in Makueni County, Kenya. MSINGI J. Dep. Educ. Found. 1 (2), Pugatch, Todd, Schroeder, Elizabeth, 2014. Incentives for Teacher Relocation: Evidence
50–59. from the Gambian Hardship Allowance.. August Economics of Education Review 41.
Le Nestour, Alexis. 2021. New PASEC Results Show Modest Improvements in Student https://doi.org/10.1016/j.econedurev.2014.04.003.
Learning in Francophone Africa, but Inequalities Are Widening. CGD Blog Posts Pugatch, Todd, Schroeder, Elizabeth, 2018. Teacher Pay and Student Performance:
(blog). 2021. 〈https://www.cgdev.org/blog/new-pasec-results-show-modest-impro Evidence from the Gambian Hardship Allowance. Journal of Development
vements-student-learning-francophone-africa〉. Effectiveness 10 (2). https://doi.org/10.1080/19439342.2018.1452778.
Leaver, Clare, Renata Freitas Lemos, and Daniela Scur. 2019. Measuring and Explaining Quota, Manal, Bhatia, Jayanti, 2022. Designing TPD Programs to Overcome Motivational
Management in Schools: New Approaches Using Public Data. SSRN Scholarly Paper. Barriers: 10 Strategies. World Bank Blogs (blog) 2022. 〈https://blogs.worldbank.or
Rochester, NY. 〈https://papers.ssrn.com/abstract= 3485938〉. g/en/education/designing-tpd-programs-overcome-motivational-barriers-10-strate
Lim, Jaegeum, Meer, Jonathan, 2020. Persistent effects of teacher–student gender gies-0〉.
matches. J. Hum. Resour. 55 (3), 809–835. https://doi.org/10.3368/jhr.55.3.0218- See, Beng Huat, Morris, Rebecca, Gorard, Stephen, Kokotsaki, Dimitra, Abdi, Sophia,
9314R4. 2020. Teacher recruitment and retention: a critical review of international evidence
Luo, Liying, Hodges, James S., 2020. Constraints in random effects age-period-cohort of most promising interventions. Educ. Sci. 10 (10), 262.
models. Sociol. Methodol. 50 (1), 276–317. https://doi.org/10.1177/ See, Beng Huat, Munthe, Elaine, Ross, Sarah A., Hitt, Loraine, El Soufi, Nada, 2022. Who
0081175020903348. becomes a teacher and why? Rev. Educ. 10 (3), e3377. https://doi.org/10.1002/
Mabeya, Mary, 2021. Influence of Personal Characteristics on Teacher Attrition in Public rev3.3377.
Secondary in Kenya. SSRN Scholarly Paper. Social Science Research Network, Seibert, Lauren. 2020. Their War Against Education: Armed Group Attacks on Teachers,
Rochester, NY. 〈https://papers.ssrn.com/abstract=3796164〉. Students, and Schools in Burkina Faso. 〈https://www.hrw.org/report/2020/05/26/
Mendenhall, Mary, Sonia Gomez, and Emily Varni. 2018. “Teaching amidst Conflict and their-war-against-education/armed-group-attacks-teachers-students-and-schools〉.
Displacement: Persistent Challenges and Promising Practices for Refugee, Internally Sirois, Geneviève. 2017. Attirer, Recruter et Retenir Les Enseignants En Zones Rurales:
Displaced and National Teachers - UNESCO Digital Library.” 〈https://unesdoc.un Une Analyse Des Politiques Du Burkina Faso En La Matière. 〈https://papyrus.bib.
esco.org/ark:/48223/pf0000266060〉. umontreal.ca/xmlui/handle/1866/20059〉.
Mhaka-Mutepfa, Magen, Elias Mpofu, Ami Moore, and Stan Ingman. 2017. Carer Swai, Anna. 2013. “The Effects of Incentive Initiatives on Teacher Retention in Tanzania:
Grandparents of Sub-Saharan Africa: ‘Foster to Be Fostered.’ In Grandparents in A Case of the Rukwa Region.” Master’s Capstone Projects, January. 〈https://sc
Cultural Context. Routledge. holarworks.umass.edu/cie_capstones/35〉.
Moses, Golden. 2024. “TSC Announces Recruitment of 46,000 Teachers to Address the Talancé, Marine de, 2017. Better Teachers, Better Results? Evidence from Rural Pakistan.
National Shortage.” ODRi Media News, October 1, 2024. 〈https://www.odrimedia. J. Dev. Stud. 53 (10), 1697–1713. https://doi.org/10.1080/
co.ke/tsc-announces-recruitment-of-46000-teachers-to-address-the-national-sh 00220388.2016.1265944.
ortage/〉. UN Population Division. 2022. World Population Prospects 2022. 2022. 〈https://populat
Mulkeen, Aidan, David W. Chapman, Joan G. DeJaeghere, and Elizabeth Leu. 2007. ion.un.org/wpp/Download/Standard/MostUsed/〉.
Recruiting, Retaining, and Retraining Secondary School Teachers and Principals in UNESCO. 2010. Teacher Attrition in Sub-Saharan Africa: The Neglected Dimension of the
Sub-Saharan Africa. Washington, DC: World Bank. https://doi.org/10.1596/978-0- Teacher Supply Challenge; a Review of Literature. 〈https://unesdoc.unesco.
8213-7066-7. org/ark:/48223/pf0000188197〉.
Muralidharan, Karthik, Sheth, Ketki, 2016. Bridging education gender gaps in developing UNESCO. 2018. “Global Education Monitoring Report, 2019: Migration, Displacement
countries: the role of female teachers. J. Hum. Resour. 51 (2), 269–297. https://doi. and Education: Building Bridges, Not Walls.” 〈https://unesdoc.unesco.org/ark:/4
org/10.3368/jhr.51.2.0813-5901R1. 8223/pf0000265866〉.
Muralidharan, Karthik, and Venkatesh Sundararaman. 2013. The Aggregate Effect of UNESCO. 2019. Global Education Monitoring Report 2019: Gender Report: Building
School Choice: Evidence from a Two-Stage Experiment in India. Working Paper. Bridges for Gender Equality. 〈https://unesdoc.unesco.org/ark:/48223/pf0000
Working Paper Series. National Bureau of Economic Research. https://doi.org/ 368753〉.
10.3386/w19441. UNESCO. 2020. A Review of the Use of Contract Teachers in Sub-Saharan Africa.
Nestour, Alexis, and Laura Moscoviz. 2020. Six Things You Should Know about Female UNESCO. 〈https://teachertaskforce.org/sites/default/files/2020-07/Contract%20t
Teachers. Center for Global Development Blog Post (blog). 2020. 〈https://www.cgd eachers%20in%20sub-Saharan%20Africa_EN.pdf〉.
ev.org/blog/six-things-you-should-know-about-female-teachers〉. UNESCO. 2023. The Teachers We Need for the Education We Want: The Global
Nguyen, Tuan D., Pham, Lam D., Crouch, Michael, Springer, Matthew G., 2020. The Imperative to Reverse the Teacher Shortage. 〈https://unesdoc.unesco.org/ark:/4
correlates of teacher turnover: an updated and expanded meta-analysis of the 8223/pf0000387001〉.
literature. Educ. Res. Rev. 31 (ember), 100355. https://doi.org/10.1016/j. World Bank. 2018a. Cameroon Public Expenditure Review: Aligning Public Expenditures
edurev.2020.100355. with the Goals of Vision 2035. World Bank, Washington, DC. 〈http://hdl.handle.net/
Nicolai, Susan, and Sebastien Hine. 2015. “Investment for Education in Emergencies.” 10986/30979〉.
ODI Global. 〈https://odi.org/en/publications/investment-for-education-in-emerg World Bank. 2018b. Guinea-Bissau Public Expenditure Review: Seizing the Moment -
encies/〉. Managing Public Finance for Development. 〈https://openknowledge.worldbank.
OECD. 2021. Figure D7.1. Teacher Attrition Rates by Type of Institution (2016) in org/entities/publication/2ce08830-dcab-50f0-b275-f3a1d54e8510〉.
Education at a Glance 2021: OECD Indicators. 〈https://www.oecd-ilibrary.org/site World Bank. 2019. Guinea-Bissau Public Expenditure Review: Managing Public Finance
s/bcc79d57-en/index.html?itemId= /content/component/bcc79d57-en〉. for Development. 〈https://openknowledge.worldbank.org/entities/publication/2
PASEC. 2020. PASEC 2019 Quality of Education Systems in French-Speaking Sub- 6fc8a28-74d2-5590-b492-37f515cd2eac〉.
Saharan Africa: Teaching/Learning Performance And Environment in Primary World Bank. 2022. Burkina Faso: Enquête Harmonisée Sur Le Conditions de Vie Des
Education. Dakar, Senegal. 〈https://pasec.confemen.org/en/〉. Ménages 2018-2019. 〈https://microdata.worldbank.org/index.php/catalog/4290〉.
Pitsoe, Victor, 2013. Teacher Attrition in South Africa: Trends, Challenges and Prospects. World Bank. 2023. World Development Indicators. 2023. 〈https://databank.worldbank.
Journal Fo Social Sciences 36 (3). org/source/world-development-indicators#〉.
Pitsoe, Victor J., Machaisa, Petunia R., 2012. Teacher attrition catastrophe in Sub- World Health Organization. 2020. Retention of the Health Workforce in Rural and
Saharan Africa: a hurdle in the achievement of UPE, EFA policy goals and MDGs. Sci. Remote Areas: A Systematic Review. Human Resources for Health Observer Series
J. Sociol. Anthropol. 1 (2012), 1–7. No. 25. 〈https://www.who.int/publications-detail-redirect/9789240013865〉.
Popova, Anna, Evans, David K., Breeding, Mary E., Arancibia, Violeta, 2022. Teacher Zeitlin, Andrew, 2021. Teacher turnover in Rwanda. J. Afr. Econ. 30 (1), 81–102.
professional development around the world: the gap between evidence and practice. https://doi.org/10.1093/jae/ejaa013.
World Bank Res. Obs. 37 (1), 107–136. https://doi.org/10.1093/wbro/lkab006. New Ziana. 2024. “Zim Government in Massive Teacher Recruitment Drive.” New Ziana,
October 10, 2024. 〈http://newziana.co.zw/zim-government-in-massive-teacher-recr
uitment-drive/〉.

17