Policy Research Working Paper 10126
Labor Market Transitions in Egypt
Post-Arab Spring
Jingyuan Deng
Nelly Elmallakh
Luca Flabbi
Roberta Gatti
Middle East and North Africa Region
Office of the Chief Economist
July 2022
Policy Research Working Paper 10126
Abstract
This paper examines the Arab Republic of Egypt’s labor market, which instead shows a large degree of dynamism
market transition dynamics post–Arab Spring based on regardless of individual initial labor market states at baseline.
the two most recent rounds of the Egypt Labor Market Auxiliary regression analyses focusing on transitions to and
Panel Survey conducted in 2012 and 2018. In addition to from the dominant absorbing labor market states in Egypt
providing disaggregated-level analysis by examining labor —public sector employment for both genders, nonpartici-
market transitions by gender, education, and age groups, pation for women, and the informal sector for men—show
the paper provides a cross-country, cross-regional perspec- that having a post-secondary education is associated with
tive by comparing Egypt’s labor market transitions with a lower probability of remaining out of the labor force for
Mexico’s, relying on data from the Encuesta Nacional de women who were already out of the labor force at baseline,
Ocupación y Empleo. To match the span of Mexico’s tran- while being married at baseline is found to be a significant
sitions (which are measured over a one-year period) and predictor for women to stay out of the labor force if they
Egypt’s (which are measured over six years), the analysis uses were already so. Among men, the better educated are found
Monte Carlo simulations of repeated discrete-time Markov to be more likely to secure formal employment, be it in the
chains. Based on these results, the Egyptian labor market public or private sector, and are more likely to keep their
appears to be highly rigid compared to the Mexican labor public formal jobs once they secure them.
This paper is a product of the Office of the Chief Economist, Middle East and North Africa Region. It is part of a larger effort
by the World Bank to provide open access to its research and make a contribution to development policy discussions around
the world. Policy Research Working Papers are also posted on the Web at http://www.worldbank.org/prwp. The authors
may be contacted at jdeng1@worldbank.org, nelmallakh@worldbank.org, lflabbi@email.unc.edu, rgatti@worldbank.org.
The Policy Research Working Paper Series disseminates the findings of work in progress to encourage the exchange of ideas about development
issues. An objective of the series is to get the findings out quickly, even if the presentations are less than fully polished. The papers carry the
names of the authors and should be cited accordingly. The findings, interpretations, and conclusions expressed in this paper are entirely those
of the authors. They do not necessarily represent the views of the International Bank for Reconstruction and Development/World Bank and
its affiliated organizations, or those of the Executive Directors of the World Bank or the governments they represent.
Produced by the Research Support Team
Labor Market Transitions in Egypt Post-Arab Spring
Jingyuan Deng, Nelly Elmallakh, Luca Flabbi and Roberta Gatti1
Keywords: employment, informality, transitions, labor market, Egypt.
JEL codes: J10, J21, J22, J45, J46.
1
Deng: World Bank’s Office of the Chief Economist for the Middle East and North Africa (jdeng1@worldbank.org);
Elmallakh: World Bank’s Office of the Chief Economist for the Middle East and North Africa
(nelmallakh@worldbank.org); Flabbi: University of North Carolina at Chapel Hill (lflabbi@email.unc.edu); Gatti:
World Bank’s Office of the Chief Economist for the Middle East and North Africa (rgatti@worldbank.org).
1. Introduction
With its rich history and large population, the Arab Republic of Egypt’s economy plays a pivotal
role in the Middle East and North Africa region. Unsurprisingly the functioning of Egypt’s labor
markets has garnered a lot of attention from academics and practitioners alike, especially since
jobs creation for Egypt’s young population is seen as a key element of economic development and
social cohesion.
Important work has been done to understand the Egyptian labor market. Assaad (2002),
Assaad (2009), Assaad and Krafft (2015) and Assaad and Krafft (2022) provide a characterization
of the evolution of the Egyptian labor market focusing on the youth bulge in the 1990s, reforms
and global slowdown in the 2000s, and political upheavals in the 2010s. This literature has
highlighted key trends in Egypt’s labor market including the long-term decline in female labor
force participation combined with rising levels of education, the large gender gap in the
unemployment rate, and the more recent decline in male labor force participation.
A complementary set of contributions has highlighted the lack of dynamism of Egypt’s
labor market (Amer 2015; Assaad 2014; Assaad and Krafft 2016; Hertog 2020; Yassine 2015).
Both transitions between labor market states and between sectors are infrequent. Even job-to-job
accession and separation are rare (Yassine 2015). Private sector workers rarely move to the public
sector, and vice versa (Assaad 2014; Yassine 2013). Women persistently – and indeed increasingly
– stay out of the labor force (Assaad et al. 2020; Krafft, Assaad and Keo 2019) and unemployed
youths still struggle to find jobs after finishing school (Amer 2015; Assaad and Krafft 2016).
While these papers outline the key contexts in which labor market transitions occur, the
literature examining the actual transitions is considerably sparser and tends to focus on specific
groups or specific stages of the life cycle (an exception is the earlier contribution of Gatti et al,
2012). Yassine (2015) takes an aggregate view of labor market transitions in Egypt, computing
flow rates that characterize the level of dynamism in the labor market, primarily with the 2012
round of the ELMPS. The paper also provides an excellent vantage point to view the extent of
rigidity in Egypt’s labor market relying on descriptive survival analyses over how long individuals
would stay in one job. Meanwhile, Amer (2015) studies labor market insertion, focusing in
particular on youth entering the labor market, and the time needed for insertion. Therefore, by
construction Amer (2015) only studies transitions to the first job. The paper affirms Yassine’s
(2015) finding on rigidity and adds further insights from a gender perspective, highlighting that
women who obtain a formal job upon entry into the labor market end up staying longer in the labor
market as opposed to women who obtain other types of employment.
Building on the existing rich compendium of notable work examining the employment
profile in Egypt and labor market transitions for specific groups, our paper contributes to the
literature in at least two ways. First, our paper provides a detailed discussion of Egypt’s labor
market transition dynamics post-Arab Spring, focusing on the two most recent rounds of the Egypt
Labor Market Panel Survey. Thanks to the panel structure of the data, we examine labor market
transitions in Egypt between 2012 and 2018, tracking individual employment status and job
trajectories between the two waves. While important work has been done to characterize the
2
evolution of the Egyptian labor market over the past years, the literature tends to address these
questions from a cross-sectional perspective, thus falling short on tracking individual trajectories
between survey rounds. Tansel and Ozdemir (2019) is the only paper thus far using transition
matrices to study the labor market in Egypt, which provides us with a methodological guide, as
well as an understanding of transitions from 2006 to 2012. This paper updates their work with
more recent data from 2018 and provides further insights with more disaggregated analyses by
education and age groups. Second, our paper provides a cross-country and cross-regional
perspective on such transitions by comparing Mexico’s labor market transitions with Egypt’s. Such
cross-country or cross-regional perspectives remain wanting in the existing literature, while they
are of utmost importance to better understand the functioning of the Egyptian labor market and to
assess its relative rigidity with respect to a relevant comparator country.
Our paper relies on the two most recent rounds of the Egypt Labor Market Panel Survey
(ELMPS), conducted in 2012 and 2018, and the 2005 round of its Mexican counterpart, Encuesta
Nacional de Ocupación y Empleo (ENOE). For an overview of the Egyptian labor market, we use
cross-sectional data to summarize individuals’ employment statuses and employment profiles
across sectors in both years. To track individual transitions between the two rounds in Egypt, we
take advantage of the panel structure of the ELMPS, as well as the detailed information on labor
market status, employment, and individual characteristics. We also rely on the panel structure of
ENOE to simulate labor market transitions in Mexico over the same time span. Relying on the
Mexican data, we observe individual transitions over at most one year in Mexico. To compute
transitions commensurable with those in Egypt, we simulate 6-yearly transitions in Mexico with
Monte Carlo simulations of repeated discrete-time Markov chains. Next, this paper provides
regression analysis of the potential determinants of labor market transitions in Egypt, with a focus
on transitions to and from the dominant absorbing labor market states in Egypt for both women
and men—public sector employment for both genders, non-participation for women, and informal
sector employment for men.
The remainder of this paper is structured as follows. Section 2 describes the data we use
for Egypt and Mexico, followed by Section 3 discussing the background and context of the
Egyptian labor market. Section 4 describes the methodology employed to compute the transition
matrices in Egypt and that to simulate their Mexican counterparts. Section 5 presents the results
from Egypt’s transition matrices, puts the Egyptian matrices in context by way of comparing with
those of Mexico, and regression analysis output on the determinants of transitions in Egypt.
Section 6 presents some concluding remarks.
2. Data
2.1 The Egypt Labor Market Panel Survey (ELMPS)
We use data from the ELMPS (Egypt) and ENOE (Mexico). The ELMPS is a nationally
representative survey conducted by the Economic Research Forum (ERF) in collaboration with
Egypt’s Central Agency for Mobilization and Statistics (CAPMAS). In this paper, we rely on the
3
two most recent rounds of the ELMPS data conducted in 2012 and 2018 to characterize the
Egyptian labor market post-Arab Spring. As in a typical labor market survey, the ELMPS covers
topics such as employment, unemployment, job dynamics, and earnings. It also provides very rich
information on education, household demographics, migration, and socio-economic characteristics.
We rely on the cross-sectional dimension of the ELMPS surveys in 2012 and 2018 and also
exploit the panel dimension of the data when examining labor market transitions between 2012
and 2018. The 2018 survey tracks households and individuals who were previously interviewed in
2012 and also includes a refresher sample of 2,000 households. By design, the refresher sample
oversampled rural communities that were among the “1,000 poorest villages” of Egypt (see Krafft,
Assaad and Rahman (2019) for more information on the data and sample design).
In our analysis, we make use of the employment module of the ELMPS, which provides
information on individual work status (employed, unemployed, or out of the labor force) and rely
on the market definition of work status with reference one week. The market definition of
employment considers as employed only individuals engaged in market economic activities, while
it excludes individuals who engage in subsistence economic activities. Following the ILO, we also
rely on a standard unemployment definition, which requires individuals to be actively searching
for a job to be considered as unemployed. We also distinguish between employed individuals
depending on their sector of employment (public versus private) and on the formality status.
Therefore, we rely on the following tripartite classification: public formal, private formal, and
private informal sectors. An individual is considered to be employed in the formal sector if the
individual has a work contract and a social security for the primary job, while an individual is
considered to be informally employed if the individual did not have a work contract, social security,
or neither. Public sector jobs include the following: public and government, while private sector
jobs include the following: private, investment, international, and other.
Whenever relying on the cross-sectional rounds of ELMPS, we focus on individuals aged
between 20 and 59 years old in each round. On the other hand, when we rely on the panel
dimension, we focus on those aged at least 20 in 2012 and at most 59 in 2018 (hence 53 in 2012)
to account for aging between the two surveys. Table A.1 in the Appendix reports descriptive
statistics on the sample of men and women as well as discussions on the sampling bias resulting
from focusing on the panel sample. We contrast raw data from each round of the ELMPS survey
against our cross-sectional and panel samples, to which the age restrictions described above are
applied.
2.2 Encuesta Nacional de Ocupación y Empleo (ENOE)
The ENOE is Mexico’s labor force survey, nationally representative of the population older than
14. As does the ELMPS, the ENOE also includes comparable labor market characteristics of labor
market states and dynamics, as well as demographic information on age, gender, education, among
others. Notably, the ENOE is conducted at a much higher frequency than the ELMPS. In each
4
round of ENOE, respondents are interviewed every quarter for a total of five quarters, 2 with a fifth
of the sample being refreshed every quarter.
In line with Bobba, Flabbi, and Levy (2022), we rely on the 2005 round. This is due to two
reasons: first, informality rates remained relatively stable over this time period and second, the
institutional background was likewise stable, with no reform taking place in the period. Similar to
our treatment of ELMPS, we restrict our sample in ENOE to individuals aged between 20 to 53 at
the beginning of the survey whenever relying on the panel dimension of the data. As in the ELMPS,
we choose not to impose restrictions on sector, rurality, work schedules or education levels to
capture a more comprehensive picture of the two labor markets, and thus avoid overrepresenting
certain sectors or labor market states over others.
The definitions of labor market states used in ENOE are consistent with those in ELMPS.
Unemployment requires job search effort and absence of a job or market activity in a one-week
reference period. Being out of the labor force requires an individual to not have conducted search
in the reference week. 3 The ENOE contains a detailed discussion on the definition of formality
and informality as part of its documentation, 4 consisting of 14 categories for the former and 9 for
the latter. Similar to the ELMPS, all workers without social security benefits are classified as
informal. One category that has been discussed at length is that of the self-employed. Both ENOE
and ELMPS count them as part of the informal sector.
3. Employment Profiles in Egypt
We start by presenting the respective employment profiles for men and women in Figure 1. The
vast majority of men are employed. While male unemployment remained roughly the same
between 2012 and 2018, employment fell and the share of men who are out of the labor force (OLF)
increased. Instead, in both rounds, around three-quarters of women (20-59 years) are out of the
labor force, and only one-fifth of women are employed. Female employment remained constant
between the two years, while the share of women being OLF increased and female unemployment
declined. Labor market attachment decreased for both genders, with a corresponding decline in
employment for men and in unemployment for women.
2
Individuals are interviewed for four quarters in a given year and in the first quarter of the following year.
3
Information on the definition of out of the labor force can be accessed through:
https://en.www.inegi.org.mx/app/glosario/default.html?p=ENOE15
4
Information on the definition of informality can be accessed through:
https://en.www.inegi.org.mx/app/biblioteca/ficha.html?upc=702825060459
5
Figure 1: Employment status among those aged 20-59, by sex
Men Women
3.83
9.18
20.64
2012
6.50
72.87
86.99
Employed Unemployed Out of Labor Force Employed Unemployed Out of Labor Force
Men Women
3.97 12.63
20.03
4.91
2018
75.06
83.40
Employed Unemployed Out of Labor Force Employed Unemployed Out of Labor Force
Notes. The analysis relies on cross-sectional data from the Egypt Labor Market Panel Survey (ELMPS) in 2012
and 2018. This figure presents the employment status of individuals aged between 20 and 59 years old in each
survey round. We rely on the market definition of work status, search required (reference 1 week). Expansion
weights are used.
We now turn to the sectoral make-up of the Egyptian labor market. We adopt a tripartite
classification of private informal, private formal, and public formal sectors. Egypt’s large informal
sector is similar in size to those of Mexico and Colombia, each having around 60% of workers
employed informally. 5 However, Egypt had a substantially larger public sector, which accounted
for over one-quarter of employment. Instead, in 2020, the public sector accounted for 13% of
Mexico’s employment 6 and just over 4% 7 of Colombia’s. Other countries at similar level of GDP
5
Lanau, S, Rodríguez-Delgado, D., and Toscani, F. 2018. ‘Colombia Selected Issues’. IMF. p 17.
Gurría, Á. 2019. Presentation of the 2019 Economic Survey of Mexico. OECD.
6
Source: ILO and Encuesta Nacional de Ocupación y Empleo. Data on the year of 2020. Retrieved Jan 2022.
7
Source: ILO and Gran Encuesta Integrada de Hogares. Data on the year of 2020. Retrieved Jan 2022.
6
per capita, such as Indonesia and Malaysia, also had much smaller public sectors, accounting for
8% and 15% of total employment, respectively. 8
The size of Egypt’s public sector is more in line with other Arab states, where public sectors
tend to account for between 20% to 30% of employment. 9 Figure 2 shows a breakdown of
employment in Egypt. Most notable is the very large informal sector, which had grown over the
six years, mostly at the expense of the public sector. Also notable is the small size of the private
formal sector, approximately a quarter of the informal private sector.
Figure 2: Share of employment in Egypt by sector, 2012 and 2018
2012 2018
25.21
30.58
54.14
14.18 60.61
15.27
Private informal Private formal Public formal Private informal Private formal Public formal
Notes. The analysis relies on cross-sectional data from the Egypt Labor Market Panel Survey (ELMPS) in 2012 and
2018. The analysis is restricted to individuals aged between 20 and 59 years old in each survey round. A job is defined
as being formal if the individual had both a work contract and social security in the primary job. A job is defined as
informal if the individual did not have a work contract, a social security, or neither. Expansion weights are used.
4. Labor Market Transitions: Methodology
Employment profiles provide a useful snapshot of individuals' labor market positions in a given
point in time. But labor market positions are dynamic, i.e. they change and evolve over time. Such
evolution is as important as an accurate description of a specific point in time. For example, it is
very useful to record the unemployment rate in a given month and year but it is very different if
most of those unemployed in that month and year have been in the state for a long or a short period
of time.
To describe labor market dynamics, we construct transition matrices over the relevant labor
market states. Transition matrices link the labor market state of a given individual at a given date
with the labor market state of the same individual at a future date. As a result, transition matrices
8
Source: ILO, Malaysia Labour Force Survey, and Indonesia National Labour Force Survey. Data on the year of
2020. Retrieved Jan 2022.
9
Source: ILO and national statistical offices. Data on the year of 2020. Retrieved Jan 2022.
7
require panel data and that is why the ELMPS for Egypt and the ENOE for Mexico are appropriate
data sets for the analysis. All transition matrices describe transitions among five different labor
market states: public sector employment, private formal sector employment, private informal
sector employment, unemployment, and out of labor force. The computation of transition matrices
for each individual country follows standard procedures. By construction, the sample is restricted
to individuals interviewed in both rounds. The transition matrices are disaggregated by sex. Panel
weights are used.
A frequency gap exists between the six-yearly ELMPS and quarterly ENOE. Even with the
panel structure of the ENOE, one can at most compute annual transitions. Six-yearly transitions
from Egypt and annual transitions from Mexico are clearly not comparable, as individuals not only
have far more time to transition but also age to different extents. In the absence of data that allow
for directly comparable transition, harmonization is in order.
In principle, we can either infer continuous-time transition intensity matrices for both labor
markets or simulate in discrete time six-yearly transition matrices in Mexico based on annual ones.
However, the former approach is riddled with issues that make it less reliable and can only be
imperfectly addressed by existing techniques. 10 We therefore opt for simulation. In the simulations,
we assume labor market transitions in each period to be discrete-time Markov processes. That is,
the transition probability for each worker from one state to another depends on and only on her
initial state. Hence, Markov processes are “memoryless,” as each individual’s history is irrelevant.
This is unlikely to be entirely accurate, as the probability of one’s transition would depend on
whether, or to where, one has recently transitioned but delivers a feasible solution to the problem
of comparability.
Further, we assume the transition probability matrix in each period to always be the
empirical annual transition matrix observed from 2005 Q1 to 2006 Q1. This assumption is unlikely
to be entirely accurate as well. Individuals age and change transition behavior over time. The
underlying macroeconomy will have changed as well, leading to cyclical changes in employment
and favoring some sectors over others. The assumption is that a six-year horizon is not long enough
to generate significant departures from the underlying steady state equilibrium. See Appendix A.2
for a more complete discussion of the limitations of our simulation method.
To simulate the transitions, we firstly model the individual labor market transition process.
Starting from each initial state, we draw a possible transition outcome in accordance with the
transition probability matrix, representing one transition. Then the same procedure is repeated with
the new initial state changed to the result of the previous draw. This process is done six times in
total, modeling transitions after six years.
We then apply the Monte Carlo method to calculate the six-yearly transition probability
matrix ������������������������������������� �. For each possible initial state ������������ , the aforementioned procedure is repeated 100,000
10
These issues of estimation are of two types: aliasing problems and embeddability problems. The former refers to
the fact that the estimated transition intensity matrix is not necessarily unique. The latter refers to the fact that there
could be no solution whose estimated parameters satisfy the requirements of probability distributions. See Bosch and
Maloney (2007) or Fougère and Kamionka (2005) for a more complete discussion.
8
times, with each time producing a destination state ������������ . Therefore, the Monte-Carlo-estimated
transition probability from state ������������ to ������������ after six periods is simply the number of times where state ������������
is reached (������������������������������������ ) divided by the number of simulations beginning from state ������������ (������������������������ . , which is set to
be 100,000). Thereby, for each entry of the matrix we have
������������������������������������
������������������������������������ =
������������������������ .
It is worth noting that transition simulations of this kind have a tendency to move towards
a steady state where the transition probability distributions converge for all initial states, as one
increases the number of transitions each individual can have. More generally, as long as for all ������������
and ������������ there is no such ������������������������������������ = 1, then as the number of transitions becomes arbitrarily large, the
transition probability distributions would move towards convergence regardless of the initial states.
5. Labor Market Transitions: Results
5.1 Transition Matrices in Egypt
The transition matrices below show individuals’ transition across sectors and employment statuses
between 2012 and 2018 in Egypt. Each row shows the respective percentages of individuals
moving to each employment status/sector in 2018, by their employment status/sector in 2012. The
percentages on the far right show the shares of individuals in each employment status/sector in
2012, with percentages in 2018 in brackets. Bold texts indicate the shares of individuals who did
not make a transition.
In Figure 3, we focus on men and women of working age (aged at least 20 years old in
2012 and at most 59 years old in 2018). We find that men who were unemployed or OLF at baseline
tended to become employed, while unemployed or OLF women in 2012 tended to either drop out
or remain out of the labor force. Indeed, three-quarters of women who were unemployed in 2012
dropped out of the labor force and 84% of already OLF women remained OLF in 2018. Transitions
within the private sector, between its formal and informal sectors, are much more frequent than
those from the private to the public sector. Conversely, those starting in the public sector rarely
transition to either private sector, though, when they do, they are more likely to move to the
informal than the formal sector.
Figure 3: Transition matrices by employment status, formality and sectors between 2012 and 2018
Private informal 76 10 5 4 5 53% (55%)
Private formal 43 39 12 3 3 14% (14%)
Public formal 8 8 82 2 21% (23%)
Men
Unemployed 56 17 11 8 8 4% (3%)
Out of labor force 56 14 11 5 14 9% (5%)
Private informal Private formal Public formal Un-employed Out of labor force
9
Private informal 32 222 62 9% (11%)
Private formal 18 28 17 3 34 1% (2%)
Public formal 32 84 1 10 10% (10%)
Women
Unemployed 8 24 13 73 7% (5%)
Out of labor force 9 1
15 84 73% (73%)
Private informal Private formal Public formal Un-employed Out of labor force
Notes. This figure relies on panel data from the Egypt Labor Market Panel Survey in 2012 and 2018. We restrict our
analysis to individuals who were interviewed in both rounds. We focus on those aged at least 20 years old in 2012
and at most 59 years old in 2018. We use transition matrices by sex to examine individuals’ transition between their
work status/sector in 2012 and their work status/sector in 2018. We rely on the market definition of work status,
search required (reference 1 week). The percentages reported at the end of each row show the shares of individuals
in each employment status/sector in 2012, with the shares in 2018 reported in brackets. Panel weights are used.
Figure 4 and Figure 5 show the transition matrices across employment statuses/sectors, by
education, for men and women, respectively. Across all education groups, unemployed and OLF
men, at baseline, tended to find jobs (mostly in the informal private sector). Though transitions to
the informal private sector remain the most likely transition among men who were unemployed or
OLF in 2012, regardless of their educational attainment, we note that the highest the level of
education, the greater the share transitioning to the formal sector, be it in the private or public
sector. Similarly, we also find that men’s education also matters for within-sector transitions, and
particularly for transitioning out of the private informal sector. Indeed, a third of men with above
secondary education, who were employed in the private informal sector at baseline, successfully
managed to transition to the formal sector (private or public) versus only 7% of informally
employed men in 2012 with no formal education qualification.
As for women, we consistently find that those who were unemployed or OLF at baseline
tended to drop out or remain out of the labor force, regardless of their education. However, women
with above secondary education are much more likely to become employed or to remain
unemployed—instead of dropping OLF—relative to women with lower levels of education. Public
sector employment is highest among highly educated women (those with above secondary
education), with a third of educated women being employed in the public sector in 2018 versus
less than 1% of less educated women (with primary education or less). Similarly to men, we also
find that the higher the educational level, the more likely it is to retain a public sector job between
the two survey rounds. Indeed, 84% of women with above secondary education who had a public
sector job in 2012 retained in 2018, while only a third of women with no education qualification
did.
Figure 6 and Figure 7 show the transition matrices across employment statuses/sectors, by
age groups, for men and women, respectively. Among men, those unemployed or out of labor force
moved into employment between 2012 and 2018, especially among the younger cohorts. Men
across all age groups exhibited high transitions from the formal private to the informal private
sector between 2012 and 2018. Such transitions from the formal to the informal private sector are
10
more pronounced among younger age groups. Approximately half of the men aged 20-29 years
and formally employed in the private sector in 2012 transitioned to the informal private sector in
2018, versus 43% of those aged between 50-59 years old.
Women, however, increasingly withdrew from the labor force between 2012 and 2018,
regardless of their age but particularly if they were in a relatively vulnerable situation at baseline,
that is informally employed, unemployed or OLF. Nevertheless, around a third of prime-aged
women (20-39 years old) who were formally employed in the private sector in 2012 still dropped
OLF by 2018. Across all age groups, more than 80% of women who were employed in public
sector employment in 2012 retained their jobs in 2018. Inter-sectoral transitions between the
formal private sector and the informal private sector are generally low and when they do occur,
female transitions from the formal private sector to the informal private sector are more likely than
vice versa.
11
Figure 4: Employment transition matrices for men by employment status, formality and sectors between
2012 and 2018, by level of education
Men
Private informal 84 5 24 5 75% (75%)
Private formal 57 28 10 3 2 9% (7%)
No formal Public formal 15 8 73 4 10% (10%)
education Unemployed 76 2 6 2 14 2% (3%)
Out of labor force 74 151 19 5% (5%)
Private informal Private formal Public formal Un-employed OLF
Private informal 75 13 25 5 64% (62%)
Private formal 51 36 42 7 13% (15%)
Public formal 7 11 78 13 13% (13%)
Primary
Unemployed 72 4 5 9 10 2% (4%)
Out of labor force 61 13 7 2 17 8% (6%)
Private informal Private formal Public formal Un-employed OLF
Private informal 75 11 6 4 4 49% (55%)
Private formal 45 38 12 3 2 15% (16%)
Public formal 8 8 81 3 19% (22%)
Secondary
Unemployed 60 15 10 7 8 4% (3%)
Out of labor force 54 17 9 6 14 13% (5%)
Private informal Private formal Public formal Un-employed OLF
Private informal 57 20 14 4 5 26% (28%)
Private formal 31 48 16 3 2 21% (22%)
Post- Public formal 6 7 84 03 41% (44%)
secondary Unemployed 40 31 14 8 7 7% (3%)
Out of labor force 36 22 22 10 10 4% (4%)
Private informal Private formal Public formal Un-employed OLF
Notes. This figure relies on panel data from the Egypt Labor Market Panel Survey in 2012 and 2018. We restrict our
analysis to men who were interviewed in both rounds. We focus on those aged at least 20 years old in 2012 and at
most 59 years old in 2018. We use transition matrices to examine individuals’ transition between their work
status/sector in 2012 and their work status/sector in 2018, by educational levels. We rely on the market definition of
work status, search required (reference 1 week). The percentages reported at the end of each row show the shares of
individuals in each employment status/sector in 2012, with the shares in 2018 reported in brackets. Panel weights are
used.
12
Figure 5: Employment transition matrices for women by employment status, formality and sectors between
2012 and 2018, by level of education
Women
Private informal 37 1 1 61 15% (17%)
Private formal 44 10 46 0% (0%)
No formal Public formal 5 5 33 57 1% (0%)
education Unemployed 29 71 0% (1%)
Out of labor force 14 1 85 84% (81%)
Private informal Private formal Public formal Un-employed OLF
Private informal 28 1 1 5 65 8% (11%)
Private formal 19 9 72 1% (1%)
Public formal 8 66 26 1% (1%)
Primary
Unemployed 14 8 7 71 1% (2%)
Out of labor force 10 12 87 89% (85%)
Private informal Private formal Public formal Un-employed OLF
Private informal 21 5 4 5 65 5% (7%)
Private formal 10 16 21 7 46 1% (3%)
Public formal 41 86 9 10% (11%)
Secondary
Unemployed 10 12 10 77 11% (6%)
Out of labor force 6 4 2 7 81 72% (72%)
Private informal Private formal Public formal Un-employed OLF
Private informal 29 13 15 2 41 3% (5%)
Private formal 15 44 14 27 5% (5%)
Post- Public formal 22 84 1 11 33% (32%)
secondary
Unemployed 6 4 7 24 59 16% (10%)
Out of labor force 4 3 4 12 77 43% (49%)
Private informal Private formal Public formal Un-employed OLF
Notes. This figure relies on panel data from the Egypt Labor Market Panel Survey in 2012 and 2018. We restrict our
analysis to women who were interviewed in both rounds. We focus on those aged at least 20 years old in 2012 and
at most 59 years old in 2018. We use transition matrices to examine individuals’ transition between their work
status/sector in 2012 and their work status/sector in 2018, by educational levels. We rely on the market definition of
work status, search required (reference 1 week). The percentages reported at the end of each row show the shares of
individuals in each employment status/sector in 2012, with the shares in 2018 reported in brackets. Panel weights are
used.
13
Figure 6: Employment transition matrices for men by employment status, formality and sectors between
2012 and 2018, by age groups
Men
Private informal 76 10 6 5 3 56% (63%)
Private formal 48 41 7 31 11% (16%)
Public formal 14 14 70 1 1 7% (11%)
20-29
Unemployed 50 22 9 8 11 6% (5%)
Out of labor force 58 16 10 6 10 19% (5%)
Private informal Private formal Public formal Un-employed OLF
Private informal 77 10 6 4 3 54% (53%)
Private formal 39 40 16 3 2 18% (16%)
Public formal 7 11 80 2 24% (26%)
30-39
Unemployed 67 8 18 5 2 3% (3%)
Out of labor force 46 13 14 5 22 1% (2%)
Private informal Private formal Public formal Un-employed OLF
Private informal 73 14 33 7 45% (43%)
Private formal 43 39 11 3 4 16% (15%)
Public formal 7 5 86 2 36% (34%)
40-49
Unemployed 56 29 5 10 1% (2%)
Out of labor force 36 6 11 5 42 2% (5%)
Private informal Private formal Public formal Un-employed OLF
Private informal 71 10 31 15 33% (34%)
Private formal 43 31 9 1 16 14% (9%)
Public formal 5 2 84 1 8 45% (40%)
50-59
Unemployed 41 25 34 1% (1%)
Out of labor force 30 5 8 57 7% (15%)
Private informal Private formal Public formal Un-employed OLF
Notes. This figure relies on panel data from the Egypt Labor Market Panel Survey in 2012 and 2018. We restrict our
analysis to men who were interviewed in both rounds. We focus on those aged at least 20 years old in 2012 and at
most 59 years old in 2018. We use transition matrices to examine individuals’ transition between their work
status/sector in 2012 and their work status/sector in 2018, by age groups. We rely on the market definition of work
status, search required (reference 1 week). The percentages reported at the end of each row show the shares of
individuals in each employment status/sector in 2012, with the shares in 2018 reported in brackets. Panel weights are
used.
14
Figure 7: Employment transition matrices for women by employment status, formality and sectors between
2012 and 2018, by age groups
Women
Private informal 27 25 4 62 6% (10%)
Private formal 16 33 12 3 36 2% (2%)
Public formal 21 81 4 12 6% (7%)
20-29
Unemployed 7 25 18 68 11% (7%)
Out of labor force 9 12 7 81 75% (74%)
Private informal Private formal Public formal Un-employed OLF
Private informal 32 5 33 57 8% (12%)
Private formal 20 22 25 33 1% (4%)
Public formal 4 4 81 11 11% (11%)
30-39
Unemployed 12 3 3 11 71 8% (5%)
Out of labor force 10 4 1 5 80 71% (68%)
Private informal Private formal Public formal Un-employed OLF
Private informal 37 21 60 11% (11%)
Private formal 7 29 9 3 52 2% (1%)
Public formal 3 86 11 17% (15%)
40-49
Unemployed 323 8 84 2% (1%)
Out of labor force 9 1
1 89 68% (71%)
Private informal Private formal Public formal Un-employed OLF
Private informal 22 78 11% (8%)
Private formal 14 57 13 16 1% (1%)
Public formal 41 86 9 18% (16%)
50-59
Unemployed NA 0% (0%)
Out of labor force 6 1 93 70% (76%)
Private informal Private formal Public formal Un-employed OLF
Notes. This figure relies on panel data from the Egypt Labor Market Panel Survey in 2012 and 2018. We restrict our
analysis to women who were interviewed in both rounds. We focus on those aged at least 20 years old in 2012 and
at most 59 years old in 2018. We use transition matrices to examine individuals’ transition between their work
status/sector in 2012 and their work status/sector in 2018, by age groups. We rely on the market definition of work
status, search required (reference 1 week). The percentages reported at the end of each row show the shares of
individuals in each employment status/sector in 2012, with the shares in 2018 reported in brackets Panel weights are
used.
15
5.2 Transition Matrices in Egypt in Comparison with Mexico
Mexico and Egypt are both developing countries with comparable levels of GDP per capita at
purchasing power parity (as of 2020, $18,444 in Mexico and $12,607 in Egypt). Importantly,
notable similarities exist between both labor markets. Both have large informal sectors that make
up just under two-thirds of total employment (Wahba and Assaad 2017). Both have a young labor
force, but also face low labor force participation rates, especially among women (OECD 2018).
These underlying similarities make for an informative comparison of labor market
transitions in the two countries, which also differ in significant respects. Egypt has a considerably
larger public sector segmented from the rest of the labor market and a much smaller private formal
sector. Workers in Mexico also regularly transition from the informal sector to the formal sector
as they secure better employment (Gong, van Soest, and Villagomez 2004).
As described in Section 4, we simulate six-yearly labor market transitions in Mexico based
on observed annual ones. Figure 8 below firstly shows the empirical annual transition for men
aged 20 to 53 in Mexico over one year, on which we base our simulations.
Figure 8: Transition matrices by employment status, formality and sectors in Mexico, 2005 Q1 – 2006 Q1
Private informal 57 29 43 7 30% (35%)
Private formal 23 66 23 5 36% (43%)
Public formal 27 31 35 5 6% (5%)
Men
Unemployed 33 38 3 13 13 3% (4%)
Out of labor force 28 28 3 6 34 24% (13%)
Private Informal Private Formal Public Unemployed OLF
Private informal 39 16 52 39 21% (21%)
Private formal 17 49 52 28 19% (21%)
Public formal 12 15 45 26 7% (7%)
Women
Unemployed 23 26 4 5 42 2% (2%)
Out of labor force 17 13 42 64 50% (48%)
Private Informal Private Formal Public Unemployed OLF
Notes. This figure relies on panel data from the ENOE in 2005 Q1 and 2006 Q1. We restrict our analysis to individuals
aged 20 to 53 who were interviewed in both rounds. We use transition matrices to examine individuals’ transition of
their work status/sector between the two dates. We rely on the market definition of work status, search required
(reference 1 week). The percentages reported at the end of each row show the shares of individuals in each
employment status/sector in 2005 Q1, with the shares in 2006 Q1 reported in brackets. Panel weights are used.
Even on an annual basis, transitions are more frequent in Mexico than in Egypt on a six-
yearly basis. And even the stickiest private formal sector only retains two-thirds of workers after
one year. Those out of the labor force also find employment rapidly. The two largest stocks of
16
Official Use
states both have large flows in and out of them. Additionally, unlike Egypt, the public sector sees
the most transitions in and out of it. This might be due to its small size, employing only around 5%
of the workforce in this sample; small movements in and out of the sector would therefore have a
large effect on transitions. Figure 9 shows the simulation results among men aged 20-53 in 2005
Q1. The lower panel reports the comparable transition matrices for Egypt (focusing on men in the
same age group observed empirically between 2012 and 2018). The lower panel was previously
presented in Figure 3.
Figure 9: Six-yearly transition matrices for males aged 20-53 in Mexico and Egypt
Private informal 37 45 7 3 7 30% (37%)
Private formal 37 46 7 3 7 36% (45%)
Public formal 36 45 8 3 7 6% (7%)
Mexico
Unemployed 37 46 7 3 7 3% (3%)
Out of labor force 37 45 7 3 7 24% (7%)
Private Informal Private Formal Public Unemployed OLF
Private informal 76 10 5 4 5 53% (55%)
Private formal 43 39 12 3 3 14% (14%)
Public formal 8 8 82 2 21% (23%)
Egypt
Unemployed 56 17 11 8 8 4% (3%)
Out of labor force 56 14 11 5 14 9% (5%)
Private informal Private formal Public formal Un-employed Out of labor force
Notes. The upper panel is simulated by Monte Carlo Method described in Section 4 Methodology. 100,000
simulations of transitions after six years were performed for each initial state, where the transition probability matrix
per period is taken from the empirical annual transition matrix in Figure 8. The annual transition matrix relies on
panel data from the ENOE in 2005 Q1 and 2006 Q1. We restrict our analysis to men aged 20 to 53 who were
interviewed in both rounds. We use transition matrices to examine individuals’ transition of their work status/sector
between the two dates. We rely on the market definition of work status, search required (reference 1 week). The
percentages reported at the end of each row show the shares of individuals in each employment status/sector in 2005
Q1, with the shares in 2006 Q1 reported in brackets. Panel weights are used. The lower panel relies on panel data
from the Egypt Labor Market Panel Survey in 2012 and 2018. We restrict our analysis to individuals who were
interviewed in both rounds. We focus on those aged at least 20 years old in 2012 and at most 59 years old in 2018.
We use transition matrices by sex to examine individuals’ transition between their work status/sector in 2012 and
their work status/sector in 2018. We rely on the market definition of work status, search required (reference 1 week).
The percentages reported at the end of each row show the shares of individuals in each employment status/sector in
2012, with the shares in 2018 reported in brackets. Panel weights are used.
17
Official Use
Figure 10: Six-yearly transition matrices for females aged 20-53 in Mexico and Egypt
Private informal 7 22 21 2 47 21% (21%)
Private formal 7 22 21 2 47 19% (22%)
Public formal 7 22 21 2 48 7% (7%)
Mexico
Unemployed 7 22 21 2 48 2% (2%)
Out of labor force 7 22 21 2 48 50% (48%)
Private Informal Private Formal Public Unemployed OLF
Private informal 32 222 62 9% (11%)
Private formal 18 28 17 3 34 1% (2%)
Public formal 32 84 1 10 10% (10%)
Egypt
Unemployed 8 24 13 73 7% (5%)
Out of labor force 9 1
15 84 73% (73%)
Private informal Private formal Public formal Un-employed Out of labor force
Notes. The upper panel is simulated by Monte Carlo Method described in Section 4 Methodology. 100,000
simulations of transitions after six years were performed for each initial state, where the transition probability matrix
per period is taken from the empirical annual transition matrix in Figure 8. The annual transition matrix relies on
panel data from the ENOE in 2005 Q1 and 2006 Q1. We restrict our analysis to women aged 20 to 53 who were
interviewed in both rounds. We use transition matrices to examine individuals’ transition of their work status/sector
between the two dates. We rely on the market definition of work status, search required (reference 1 week). The
percentages reported at the end of each row show the shares of individuals in each employment status/sector in 2005
Q1, with the shares in 2006 Q1 reported in brackets. Panel weights are used. The lower panel relies on panel data
from the Egypt Labor Market Panel Survey in 2012 and 2018. We restrict our analysis to individuals who were
interviewed in both rounds. We focus on those aged at least 20 years old in 2012 and at most 59 years old in 2018.
We use transition matrices by sex to examine individuals’ transition between their work status/sector in 2012 and
their work status/sector in 2018. We rely on the market definition of work status, search required (reference 1 week).
The percentages reported at the end of each row show the shares of individuals in each employment status/sector in
2012, with the shares in 2018 reported in brackets. Panel weights are used.
The convergence seen in the upper panels of Figure 9 and Figure 10 shows that there is no
labor market persistence in Mexico after six years – regardless of the initial state one finds oneself
at the beginning, the transition probability distributions are almost identical. This result is subject
to certain limitations. As the Markov chains assume memoryless-ness and hence a constant rate of
transition, the simulations will tend to over-state dynamism in the labor market. Such caveats are
discussed in further detail in Appendix A.2. However, as Appendix A.2 shows, this over-statement
is not unreasonably large. The simulations therefore still provide evidence in support of the
remarkable dynamism of Mexico’s labor market. Further details on how the transitions take place
in Mexico can be found in the per-period simulations in Appendix A.3, which shows that
persistence in Mexico ends after 3 years for men and 4 years for women.
18
Official Use
This result stands in stark contrast to what we empirically observe in Egypt, where only a
small proportion of individuals have left their original sectors. One should note that since the
transition is observed only at the two ends of a six-year period, it is entirely possible for individuals
to have moved to another (or several other) different sectors, only to move back to the original
starting point. However, existing research instead points to workers not changing jobs at all as
primary reason for this phenomenon (Yassine 2015). As also noted by Amer (2015), job tenures
tend to be very long in Egypt. 11
The resulting labor market composition after six years also points to the rigidity of Egypt’s
labor market. Figure 9 indicates substantial labor market upgrading among Mexican men, but
almost no compositional changes among Egyptian men. In Mexico, only 7% of men remain out of
the labor force after six years; the share of men in the private formal sector increases from 36% to
45%, while the corresponding increase in the informal sector goes from 30% to 37%. Whereas
among men in Egypt, the only compositional changes come from a small reduction in non-
participation, which translates into small increases in the shares of employment in the informal
and public sectors. However, Mexico’s labor market upgrading does not apply to Mexican women,
who fare as poorly as their Egyptian counterparts. Figure 10 shows that neither labor market has
led to substantive compositional changes for women over the six-year window of analysis.
5.3 Regression Analysis
Table 1 and Table 2 present results from a linear probability model on employment status and
formality-sector transitions. In each column, the sample is restricted to those individuals with the
relevant initial status in 2012, and the dependent variable takes the value one if the individual
reached the corresponding destination status in 2018. The reference category for the educational
dummies is having no formal education qualification.
Table 1 shows the transition regression outputs from the two initial baselines women are
most likely to be in, either out of the labor force or employed in the public sector. We find that
younger women are more likely to transition from OLF to employment than older women. Aging
by one year makes a young woman more likely to enter employment, with the highest likelihood
at age 33. We also find that having primary, secondary or post-secondary schooling (relative to
having no formal education) reduces the probability of transitioning to employment from OLF for
women. Among women, we also find that having post-secondary education is associated with a
lower probability of remaining OLF for those already OLF at baseline. Being married at baseline
11
We have attempted to explore labor market states’ persistence in further detail in ELMPS, but substantial
measurement errors preclude a useful analysis. We tracked individual labor market status in 2012 based on
retrospective information provided in the 2018 survey and cross-checked the accuracy of this information based on a
sample of panel observations that were also interviewed in 2012. The data analysis revealed significant discrepancies
between what individuals report in the current section of the 2012 round and what they report in the retrospective
section of the 2018 round. Of those who retrospectively reported in 2018 that they were not employed in 2012, 44%
were found to be employed in 2012. On the other hand, of those who retrospectively reported being employed in 2012,
20% were found to be unemployed.
19
Official Use
is another significant predictor for women to stay out of the labor force if they were already so.
On the other hand, holding any educational qualification also significantly reduces the probability
that a publicly employed woman drops out of the labor force. Younger, unmarried women, and
women living in smaller households are more likely to move from the public to the private formal
sector.
Table 1: Determinants of employment status transitions between 2012 and 2018
Women
Initial status 2012 Public Formal Out of labor force (OLF)
Destination status Public Private Private Public Any
2018 Formal Formal Informal OLF Formal Employment OLF
Age -0.018 0.011* 0.008 -0.012 -0.021* 0.023*** -0.023***
(0.016) (0.010) (0.011) (0.010) (0.011) (0.004) (0.003)
Age^2 0.001 -0.003* -0.003 0.004 0.003* -0.002*** 0.002***
(0.002) (0.002) (0.002) (0.003) (0.001) (0.001) (0.000)
Primary school 0.209 -0.088 -0.036 -0.408** 0.034 -0.063*** 0.048***
(0.142) (0.062) (0.083) (0.163) (0.031) (0.012) (0.016)
Secondary school 0.239 -0.067 -0.087 -0.502*** 0.158*** -0.050*** -0.003
(0.132) (0.059) (0.064) (0.152) (0.031) (0.014) (0.013)
Post-Secondary 0.238 -0.075 -0.085 -0.499*** 0.373*** -0.072*** -0.057**
(0.127) (0.048) (0.062) (0.148) (0.042) (0.017) (0.024)
Married 0.051 -0.049*** 0.005 0.032 -0.137*** -0.039** 0.108***
(0.029) (0.009) (0.017) (0.031) (0.044) (0.011) (0.024)
Household size 0.010 -0.013* -0.000 0.072 -0.004 0.000 -0.002
(0.013) (0.004) (0.001) (0.011) (0.008) (0.002) (0.001)
Region dummies ✓ ✓ ✓ ✓ ✓ ✓ ✓
Occupation dummies ✓ ✓ ✓ ✓ ☓ ☓ ☓
Observations 811 812 812 908 790 6,618 6,618
Notes. *** p<0.01, **<0.05, *<0.1. A linear probability model’s coefficient estimates and standard errors are reported. This
table uses panel data from the Egypt Labor Market Panel Survey in 2012 and 2018. The sample is restricted to those aged at
least 20 years old in 2012 and at most 59 years old in 2018. All control variables refer to 2012. Panel weights are used.
Table 2 shows the regression output on formality-sector transitions for men. We consider
men who were employed either in the public sector or informally—which we define as the two
absorbing states—at baseline and examine the transition probability to the public formal sector,
private formal sector and private informal sector in 2018. Predictably, those better educated are
found to be much more likely to secure formal employment, be it in the public or in the private
sector. Having a secondary or a post-secondary level of education is strongly associated with
moving out of the private informal sector. The better educated are also much more likely to keep
their public formal jobs once they secure them. They are much less likely to move to the private,
20
Official Use
especially private informal, sector than their less educated counterparts. Married men are far more
likely to stay put in the public sector, though being married at the baseline does little to help an
informally employed man to secure a public sector job.
Table 2: Determinants of formality-sector transitions between 2012 and 2018
Men
Initial status in 2012 Public Formal Private Informal
Public Private Private Public Private Private
Destination status in 2018 Formal Formal Informal Formal Formal Informal
Age -0.002 0.002 -0.008 0.007* 0.004 -0.005
(0.010) (0.017) (0.008) (0.004) (0.003) (0.007)
Age^2 0.003 -0.004 0.002 -0.004* 0.001 0.001
(0.004) (0.003) (0.003) (0.002) (0.001) (0.003)
Primary school 0.075 -0.002 -0.066* 0.004 0.065*** -0.074***
(0.039) (0.033) (0.025) (0.006) (0.010) (0.017)
Secondary school 0.117** -0.036 -0.082** 0.043*** 0.073*** -0.103***
(0.042) (0.032) (0.034) (0.012) (0.014) (0.012)
Post-Secondary 0.190*** -0.058 -0.133*** 0.109*** 0.118*** -0.235***
(0.036) (0.036) (0.028) (0.008) (0.020) (0.018)
Married 0.150*** -0.082** -0.078** -0.002 0.005 -0.009
(0.024) (0.034) (0.031) (0.005) (0.008) (0.015)
Household size 0.009 -0.006* 0.003 0.002 -0.000 0.004
(0.007) (0.003) (0.002) (0.003) (0.001) (0.003)
Region dummies ✓ ✓ ✓ ✓ ✓ ✓
Occupation dummies ✓ ✓ ✓ ✓ ✓ ✓
Observations 1,713 1,713 1,713 4,305 4,305 4,305
Notes. *** p<0.01, **<0.05, *<0.1. A linear probability model’s coefficient estimates and standard errors are reported. This
table uses panel data from the Egypt Labor Market Panel Survey in 2012 and 2018. The sample is restricted to those aged at
least 20 years old in 2012 and at most 59 years old in 2018. All control variables refer to 2012. Panel weights are used.
6. Concluding Remarks
The persistence in and the rigidity of labor market states were always key characteristics of the
Egyptian labor market. This paper revisited these questions relying on transition matrices to
examine the dynamics of labor market transitions post-Arab Spring. The analysis relies on the two
most recent rounds of the Egypt Labor Market Panel Surveys (ELMPS) and exploits the panel
structure of the data to track individual employment status and job trajectories between the two
rounds.
21
Official Use
To offer a cross-country and cross-regional perspective, we compare the Egyptian labor
market transition matrices with those of Mexico. Mexico provides an interesting comparator
country given its similar level of GDP per capita in purchasing power parity, its large private
informal sector, its young labor force, and low female labor force participation—all of which are
common characteristics with Egypt’s labor market.
Our transition matrices showcase particular dominant absorbing labor market states among
women and men. Specifically, we find a large degree of persistence in being out of labor force
among women. On the other hand, we find very low female labor market engagement in the private
sector, be it formal or informal, while we observe an important female presence in the public sector
and in particular, a persistence in female public sector employment among the most educated. As
for men, we find suggestive evidence of high informality—in particular among the less educated—
and a lack of dynamism to transition to the private formal sector or the public formal sector. Such
transitions are only likely to occur among the most educated men.
The stickiness of these dominant absorbing states is further confirmed when contrasting
Egypt’s transition matrices with those of Mexico. Relying on data from Mexico’s labor force
survey, Encuesta Nacional de Ocupación y Empleo, and simulating labor market transitions over
a six-year window using Monte Carlo simulations of repeated discrete-time Markov chains, our
results highlight very divergent levels of dynamism between the two labor markets. Mexico’s labor
market shows a large degree of dynamism over the six-year window regardless of individual initial
labor market states at baseline, while Egypt’s labor market shows great persistence in labor market
states and far fewer transitions between the two rounds.
22
Official Use
Bibliography
Amer, Mona. 2015. “Patterns of Labor Market Insertion in Egypt.” In The Egyptian Labor Market
in an Era of Revolution, eds. Ragui Assaad and Caroline Krafft. Oxford University Press, 70–89.
https://oxford.universitypressscholarship.com/view/10.1093/acprof:oso/9780198737254.001.000
1/acprof-9780198737254-chapter-4 (November 5, 2021).
Assaad, Ragui, ed. 2002. The Egyptian Labor Market in an Era of Reform. Cairo, Egypt: American
University in Cairo Press.
Assaad, Ragui. 2014. “Making Sense of Arab Labor Markets: The Enduring Legacy of Dualism.”
IZA Journal of Labor & Development 3(1): 6.
Assaad, Ragui, and Krafft, Caroline, ed. 2015. The Egyptian Labor Market in an Era of
Revolution. Oxford, UK: Oxford University Press.
Assaad, Ragui, and Caroline Krafft. 2016. Working Papers Labor Market Dynamics and Youth
Unemployment in the Middle East and North Africa: Evidence from Egypt, Jordan and Tunisia.
Economic Research Forum. https://ideas.repec.org/p/erg/wpaper/993.html (November 5, 2021).
Assaad, Ragui, Rana Hendy, Moundir Lassassi, and Shaimaa Yassin. 2020. “Explaining the
MENA Paradox: Rising Educational Attainment yet Stagnant Female Labor Force Participation.”
Demographic Research 43(28): 817–50.
Bobba, Matteo, Luca Flabbi, and Santiago Levy. 2022. “Labor Market Search, Informality, and
Schooling Investment.” International Economic Review 63(1): 211–59.
Bosch, Mariano and Maloney, William. 2019. “Comparative Analysis of Labor Market Dynamics
Using Markov Processes: An Application to Informality”. World Bank Policy Research Working
Paper 4429.
Elsayed, Ahmed and Wahba, Jackline. 2016. Informalization Dynamics and Gains: Why Want a Job
Contract? Working Papers 1001, Economic Research Forum, revised May 2016.
Fougere, D. and Kamionka, T., 2008. Econometrics of individual labor market transitions. In The
Econometrics of Panel Data (pp. 865-905). Springer, Berlin, Heidelberg.
Gatti, Roberta, Angel-Urdinola, Diego, Joana Silva, Jonas, and Bodor, Andras. 2012. “Striving
for Better Jobs: The Challenge of Informality in the Middle East and North Africa”. MENA
Knowledge and Learning Quick Notes Series; No. 49. World Bank, Washington, DC.
Gong, Xiaodong, Arthur van Soest, and Elizabeth Villagomez. 2004. “Mobility in the Urban Labor
Market: A Panel Data Analysis for Mexico.” Economic Development and Cultural Change 53(1):
1–36.
Hertog, Steffen. 2020. “Segmented Market Economies in the Arab World: The Political Economy
of Insider–Outsider Divisions.” Socio-Economic Review.
23
Official Use
Krafft, Caroline; Assaad, Ragui; and Keo, Caitlyn, "The Evolution of Labor Supply in Egypt from
1988-2018: A Gendered Analysis" (2019). Economics & Political Science Faculty Scholarship. 102.
Krafft, C., Assaad, R. and Rahman, K.W., 2019. Introducing the Egypt Labor Market Panel Survey
2018. Economic Research Forum (ERF).
ILO, 2014. Informal employment in Mexico: Current situation, policies and challenges. ILO.
https://www.ilo.org/americas/sala-de-prensa/WCMS_245889/lang--en/index.htm
OECD. 2018. Good Jobs for All in a Changing World of Work: The OECD Jobs Strategy. OECD.
https://www.oecd-ilibrary.org/social-issues-migration-health/good-jobs-for-all-in-a-changing-
world-of-work_9789264308817-en (April 29, 2022).
Wahba, Jackline, and Ragui Assaad. 2017. “Flexible Labor Regulations and Informality in Egypt.”
Review of Development Economics 21(4): 962–84.
Yassine, C., 2013. Structural labor market transitions and wage dispersion in Egypt and Jordan.
Economic Research Forum Working paper No. 753.
Yassine, Chaimaa. 2015. “Job Accession, Separation, and Mobility in the Egyptian Labor Market
over the Past Decade.” In The Egyptian Labor Market in an Era of Revolution, eds. Ragui Assaad
and Caroline Krafft. Oxford University Press, 218–40.
https://oxford.universitypressscholarship.com/view/10.1093/acprof:oso/9780198737254.001.000
1/acprof-9780198737254-chapter-12 (April 28, 2022).
24
Official Use
Appendix A.1: Sample Selection Comparing Raw, Cross-Sectional, and Panel Samples
Table A.1: Descriptive statistics on the 2012 and 2018 estimation samples versus raw data
Men Women
2012 2018 2012 2018
Mean Raw Sample Panel Raw Sample Panel Raw Sample Panel Raw Sample Panel
(St. Dev.)
Age 25.817 34.957 33.211 25.924 35.653 39.284 26.795 34.768 33.138 26.969 35.503 39.187
(19.656) (10.856) (9.236) (19.805) (10.604) (9.275) (20.192) (11.248) (9.638) (20.426) (10.87) (9.609)
Rural 0.567 0.546 0.59 0.642 0.624 0.589 0.560 0.540 0.596 0.636 0.621 0.593
(0.495) (0.498) (0.492) (0.479) (0.484) (0.492) (0.496) (0.498) (0.491) (0.481) (0.485) (0.491)
Household wealth 0.381 0.386 0.384 0.256 0.261 0.263 0.382 0.390 0.388 0.258 0.266 0.266
(0.129) (0.128) (0.125) (0.137) (0.14) (0.14) (0.134) (0.134) (0.129) (0.141) (0.144) (0.145)
Household size 5.013 4.813 4.902 4.767 4.538 4.576 4.945 4.698 4.851 4.687 4.463 4.538
(2.167) (2.141) (2.174) (1.835) (1.793) (1.702) (2.237) (2.139) (2.139) (1.961) (1.844) (1.763)
Below primary 0.283 0.216 0.207 0.296 0.223 0.246 0.397 0.345 0.338 0.399 0.336 0.37
(0.45) (0.411) (0.405) (0.456) (0.417) (0.431) (0.489) (0.476) (0.473) (0.49) (0.472) (0.483)
Primary 0.251 0.165 0.169 0.232 0.136 0.129 0.212 0.123 0.124 0.213 0.125 0.107
(0.433) (0.371) (0.375) (0.422) (0.343) (0.336) (0.409) (0.328) (0.33) (0.41) (0.331) (0.309)
Secondary 0.303 0.399 0.423 0.322 0.437 0.398 0.257 0.339 0.362 0.259 0.352 0.327
(0.459) (0.49) (0.494) (0.467) (0.496) (0.49) (0.437) (0.474) (0.481) (0.438) (0.478) (0.469)
Post-secondary 0.164 0.221 0.201 0.150 0.204 0.226 0.134 0.192 0.175 0.129 0.187 0.195
(0.37) (0.415) (0.4) (0.357) (0.403) (0.418) (0.34) (0.394) (0.38) (0.335) (0.39) (0.396)
Cairo 0.105 0.111 0.081 0.072 0.078 0.081 0.112 0.118 0.083 0.077 0.08 0.084
(0.307) (0.314) (0.273) (0.259) (0.269) (0.273) (0.315) (0.323) (0.276) (0.267) (0.271) (0.277)
Alexandria 0.081 0.087 0.07 0.059 0.063 0.07 0.079 0.088 0.066 0.056 0.062 0.067
(0.273) (0.282) (0.254) (0.235) (0.242) (0.256) (0.27) (0.283) (0.248) (0.23) (0.24) (0.249)
Urban Lower Egypt 0.110 0.115 0.113 0.099 0.101 0.113 0.111 0.117 0.115 0.101 0.106 0.116
(0.313) (0.319) (0.316) (0.299) (0.302) (0.317) (0.314) (0.321) (0.319) (0.301) (0.308) (0.32)
Urban Upper Egypt 0.138 0.142 0.147 0.128 0.133 0.147 0.140 0.138 0.140 0.130 0.132 0.140
(0.345) (0.349) (0.354) (0.334) (0.34) (0.354) (0.347) (0.345) (0.347) (0.337) (0.338) (0.347)
Rural Lower Egypt 0.284 0.287 0.314 0.291 0.295 0.313 0.277 0.277 0.307 0.286 0.289 0.305
(0.451) (0.453) (0.464) (0.454) (0.456) (0.464) (0.447) (0.448) (0.461) (0.452) (0.453) (0.461)
Rural Upper Egypt 0.282 0.258 0.276 0.351 0.329 0.275 0.282 0.262 0.289 0.349 0.332 0.288
(0.45) (0.437) (0.447) (0.477) (0.47) (0.447) (0.45) (0.439) (0.453) (0.477) (0.471) (0.453)
Employed 0.615 0.872 0.878 0.572 0.844 0.915 0.141 0.204 0.199 0.125 0.184 0.222
(0.487) (0.334) (0.328) (0.495) (0.363) (0.279) (0.348) (0.403) (0.399) (0.331) (0.387) (0.416)
Unemployed 0.025 0.036 0.035 0.029 0.041 0.035 0.045 0.068 0.074 0.032 0.05 0.047
(0.156) (0.187) (0.184) (0.166) (0.198) (0.184) (0.208) (0.251) (0.262) (0.177) (0.219) (0.211)
Out of labor force 0.360 0.092 0.087 0.400 0.115 0.050 0.814 0.728 0.727 0.843 0.766 0.731
(0.48) (0.288) (0.283) (0.49) (0.319) (0.218) (0.389) (0.445) (0.446) (0.364) (0.423) (0.443)
Notes. This table relies on data from the Egypt Labor Market Panel Survey (ELMPS) 2012 and 2018.
25
Official Use
By construction, the raw data differs with respect to the sample in 2012 and 2018, as well as with
respect to the panel data along the age dimension. By excluding all individuals below 20 years old,
individuals in our sample and panel data are older than those in the raw data. Overall, we find that
men and women distribution across Egyptian regions seem to be comparable in the raw, sample,
and panel data. However, the sample data seems to slightly underrepresent individuals from rural
residency, with smaller household size, and lower wealth score relative to the raw data. Both the
sample and panel data seem to overrepresent individuals with higher levels of educational
attainment (secondary education and above secondary education) and to slightly underrepresent
individuals with lower levels of education (those with primary education or less). In terms of work
status, by definition, we find higher incidence of employment, among both men and women, in
the sample and panel data relative to the raw data. Unemployment is likewise higher in the sample
and panel data relative to the raw data, while the incidence of being out of the labor force (OLF)
is higher in the latter relative to the former.
While the descriptive analysis shows some differences across multiple dimensions, several
of these are to be expected, by definition, given the age restrictions employed. This applies to the
age differences, the educational differences, as well as the differences in terms of work status.
Educational differences—in terms of slight overrepresentation of the highly educated in the sample
and the panel—is likely driven by the exclusion of the youth (below 20 years old) with lower levels
of education. Likewise, the overrepresentation of the active population (employed and unemployed)
and the underrepresentation of the OLF in the sample and panel data relative to the raw data once
again merely reflects our focus on prime working age individuals.
26
Official Use
Appendix A.2: Discussion on the Limitation of Monte Carlo Simulations of
Markov Chains in Labor Markets
As noted in the Methodology Section 4, the repeated Markov process being simulated assumes
memoryless-ness and hence also constancy over time. In reality, when conditioning on history,
probabilities of transitioning to other states are likely to be lower than the unconditional
probabilities. Individuals who have recently made job moves are unlikely to do so as much as those
who did not. Therefore, by assuming memoryless-ness, the simulations have a tendency of over-
stating dynamism in the labor market, but not to an extent that makes the results unreasonable. As
an example, the robustness check below shows the simulated outcome of four quarterly transitions
(based on transitions from 2005 Q1 to 2005 Q2) with the empirical annual transition matrix
obtained from ENOE. Though this does qualify our simulations as noted in the paper, our main
argument nevertheless holds, as it is evident that the degree of dynamism of the labor market is
highly divergent between Mexico and Egypt.
Table A.2: Simulated annual transition with quarterly data and empirical annual transition of Mexican men
Simulated Empirical
employed
employed
Informal
Informal
Formal
Formal
Private
Private
Private
Private
Public
Public
OLF
OLF
Un-
Un-
Public 33% 32% 26% 3% 7% 35% 31% 27% 2% 5%
Private Formal 3% 57% 29% 3% 7% 2% 66% 23% 3% 5%
Private Informal 3% 35% 45% 4% 13% 4% 29% 57% 3% 7%
Unemployed 4% 37% 39% 4% 16% 3% 38% 33% 13% 13%
OLF 3% 28% 37% 4% 27% 3% 28% 28% 6% 34%
Notes. The left panel is simulated by Monte Carlo Method described in Section 4 Methodology. 100,000 simulations
of transitions after 4 quarters were performed for each initial state, where the transition probability matrix per period
is taken from the empirical quarterly transition matrix computed from ENOE 2005 Q1 and Q2. We restrict our
analysis to men aged 20 to 53 who were interviewed in both rounds. We use transition matrices to examine
individuals’ transition of their work status/sector between the two dates. We rely on the market definition of work
status, search required (reference 1 week). Panel weights are used. The right panel relies on panel data from the
ENOE in 2005 Q1 and 2006 Q1. We restrict our analysis to individuals aged 20 to 53 who were interviewed in both
rounds. We use transition matrices to examine individuals’ transition of their work status/sector between the two
dates. We rely on the market definition of work status, search required (reference 1 week). Panel weights are used.
27
Official Use
Appendix A.3: Per-Period Labor Market Transition Simulations in Mexico
Appendix A.3 shows the simulated transition matrices in Mexico for each period of transition,
where one can examine the degree of labor market rigidity and dynamism in Mexico in further
detail. Figure A.1 confirms the high degree of dynamism among men in the Mexican labor market,
with persistence largely disappearing after just three years. Transitions into the labor force occur
especially quickly. By contrast, women in Mexico need 4 years of transitions to reach their steady
state.
Figure A.1: Simulated transition matrices by employment status, formality and sectors in Mexico per
period of transition, among men aged 20-53
Years of transition Men
OLF 34 6 28 28 3
Unemployed 13 13 33 38 3
Private Informal 7 3 57 29 4
1 year Private Formal 5 3 23 66 2
Public 5 2 27 31 35
OLF Unemployed Private Informal Private Formal Public
OLF 16 5 35 40 4
Unemployed 10 5 36 45 4
Private Informal 9 4 43 40 4
2 years
Private Formal 7 4 32 54 4
Public 7 3 34 42 14
OLF Unemployed Private Informal Private Formal Public
OLF 11 4 36 44 4
Unemployed 9 4 37 46 4
Private Informal 9 4 38 45 5
3 years
Private Formal 8 4 35 50 4
Public 8 4 36 45 8
OLF Unemployed Private Informal Private Formal Public
28
Official Use
OLF 9 4 37 46 4
Unemployed 9 4 36 47 4
Private Informal 9 4 37 46 5
4 years
Private Formal 8 4 36 48 4
Public 8 4 36 46 5
OLF Unemployed Private Informal Private Formal Public
OLF 9 4 37 46 4
Unemployed 8 4 37 47 4
Private Informal 8 4 37 47 5
5 years
Private Formal 8 4 36 47 4
Public 8 4 36 47 5
OLF Unemployed Private Informal Private Formal Public
OLF 9 4 36 47 4
Unemployed 8 4 36 47 4
Private Informal 8 4 37 47 4
6 years
Private Formal 8 4 36 47 4
Public 8 4 36 47 5
OLF Unemployed Private Informal Private Formal Public
Notes. Each panel is the simulated result of transitions after a given period of time by Monte Carlo Method described
in Section 4 Methodology. 100,000 simulations of transitions after six years were performed for each initial state,
where the transition probability matrix per period is taken from the empirical annual transition matrix in Figure 8.
The annual transition matrix relies on panel data from the ENOE in 2005 Q1 and 2006 Q1. We restrict our analysis
to men aged 20 to 53 who were interviewed in both rounds. We use transition matrices to examine individuals’
transition of their work status/sector between the two dates. We rely on the market definition of work status, search
required (reference 1 week). The percentages reported at the end of each row show the shares of individuals in each
employment status/sector in 2005 Q1, with the shares in 2006 Q1 reported in brackets. Panel weights are used.
29
Official Use
Figure A.2: Simulated transition matrices by employment status, formality and sectors in Mexico per
period of transition, among women aged 20-53
Years of transition Women
OLF 64 24 13 17
Unemployed 42 5 4 26 23
Private Informal 26 2 45 15 12
1 year
Private Formal 28 2 5 48 17
Public 39 2 5 16 39
OLF Unemployed Private Informal Private Formal Public
OLF 53 2 6 18 21
Unemployed 46 2 6 24 22
Private Informal 38 2 23 20 17
2 years
Private Formal 40 2 6 31 20
Public 46 2 6 20 25
OLF Unemployed Private Informal Private Formal Public
OLF 49 2 6 21 21
Unemployed 47 2 7 23 22
Private Informal 43 2 14 21 20
3 years
Private Formal 45 2 7 25 21
Public 48 2 6 22 22
OLF Unemployed Private Informal Private Formal Public
OLF 48 2 7 21 21
Unemployed 47 2 7 22 22
Private Informal 46 2 10 22 21
4 years
Private Formal 47 2 7 23 21
Public 48 2 7 22 22
OLF Unemployed Private Informal Private Formal Public
30
Official Use
OLF 48 2 7 22 21
Unemployed 48 2 7 22 21
Private Informal 47 2 7 22 21
5 years
Private Formal 47 2 7 22 21
Public 48 2 7 22 21
OLF Unemployed Private Informal Private Formal Public
OLF 48 2 7 22 21
Unemployed 48 2 7 22 21
Private Informal 47 2 7 22 21
6 years
Private Formal 47 2 7 22 21
Public 48 2 7 22 21
OLF Unemployed Private Informal Private Formal Public
Notes. Each panel is the simulated result of transitions after a given period of time by Monte Carlo Method described
in Section 4 Methodology. 100,000 simulations of transitions after six years were performed for each initial state,
where the transition probability matrix per period is taken from the empirical annual transition matrix in Figure 8.
The annual transition matrix relies on panel data from the ENOE in 2005 Q1 and 2006 Q1. We restrict our analysis
to women aged 20 to 53 who were interviewed in both rounds. We use transition matrices to examine individuals’
transition of their work status/sector between the two dates. We rely on the market definition of work status, search
required (reference 1 week). The percentages reported at the end of each row show the shares of individuals in each
employment status/sector in 2005 Q1, with the shares in 2006 Q1 reported in brackets. Panel weights are used.
31
Official Use