Research Brief

I COVID-19
nformal 1
Economy and Employment

Research Brief1

Date: December 2020

The impact of the pandemic on employment in Turkey:

What would have happened without COVID-19?

1- Summary
How many people would have been working in Turkey had not the pandemic arrived is a
simple question to ask but a hard one to answer. This research brief proposes the creation
of a counterfactual (a hypothetical scenario representing Turkey without COVID19) using
an econometric model. This model uses the month of the year as well as historical data on
several macroeconomic indicators to predict the prevailing employment level had not the
pandemic hit Turkey in 2020. The difference between actual employment levels and the
ones forecasted in the scenario without pandemic are then calculated. This type of analysis
is carried out for 29 groups of workers a monthly basis from February to August using
monthly data from Turkstat.

The idea behind analyzing 29 different groups of workers is to go beyond overall employment
levels and find which workers were hit harder by the pandemic. This brief finds that formality
(registration with the Social Security Institute) plays a key role in explaining employment
losses. On top of that, young people and especially young women seem to be among the
ones more at risk of suffering long-term employment losses. Last but not least, economic
activities tied to high informality levels or face to face interactions such as the construction
sector, hospitality or low skilled personal services have also suffered significant employment
losses between February to August.

2- Background
The social distancing, stay at home policies and lockdown measures applied by governments
in an attempt to control the COVID-19 pandemic have also brought severe disruptions to
most economies throughout the world.2 Turkey was no exception and soon found itself
closing down schools and cafes, restricting the mobility of children and the elderly and even
the mobility of entire cities during weekends.

These restrictions as well as the fear to the virus also affected the labour market; lack of
demand forced many companies to close down their businesses, informal workers and
those under precarious working conditions also faced a higher risk of losing their job and
workers in the hospitality sector may take a long time until they can recover their jobs due
to the sector’s high vulnerability, as pointed out by the research carried out by Şeker (2020).

1
Research brief prepared by Luis Pinedo Caro, for more information contact pinedo@ilo.org.
2
See ILO (2020).
2 Research Brief
The impact of the pandemic on employment in Turkey:
What would have happened without COVID-19?

However, even though the coronavirus does not distinguish borders, cities, or neighbourhoods
not everyone is equally affected. Those in less secured jobs or in sectors where face to face
interactions are an integral part of the economic activity, young people and especially young
women are at risk of receiving a much more severe and durable impact than other workers.
Understanding the extent to which vulnerable groups have been hit by the pandemic is the
aim of this research brief.

3- Impact of COVID-19 on employment

Whenever someone says “the impact of COVID-19 on employment” or “the employment
losses induced by COVID-19” that person has inevitable and, perhaps even unconsciously
built a counterfactual in her mind. In this context a counterfactual is meant to be something
along the lines of “what would have happened if COVID-19 had not existed”. That is, we are
intuitively comparing actual employment levels with the employment levels that would have
occurred if the pandemic had never existed.

Counterfactual analysis have already been used in Turkey for the calculation of employment
losses due to COVID-19. For instance, the estimations provided by DISK (2020) implicitly
assume employment levels would have been as they were back in 2019. A different set of
assumptions are followed by UN (2020) with de-seasonalised data; in this case, the author
seems to have assumed that employment levels without the pandemic would have behaved
as the de-seasonalized employment levels of December 2019.

Even though the results provided by the just-mentioned pieces of research may provide a
useful approximation to the number of jobs lost during the pandemic they also face some
limitations. First of all neither of them uses recent macroeconomic information to judge
whether the economic outlook and, thus, employment levels were going to improve or
not during 2020. Indeed, the Turkish economy saw its GDP grow by 6.4% and 4.4% during,
respectively, the 4th quarter of 2019 and the first quarter of 2020. This hints an economic
recovery could have been on its way after a terrible year in terms of labour market performance.
It can, thus, be argued that using pre-pandemic employment levels would underestimate
the impact of COVID-19 on employment by creating counterfactual employment levels that
fall short of what they would have been in the absence of the pandemic. On top of that,
using past employment figures as a counterfactual also incurs the risk of finding worse
employment outcomes without COVID-19 than with COVID-19, which is not reasonable. This
happens, for instance, across men in industry when we compare the employment levels
showcased in March 2019 (4.07 million) and in March 2020 (4.18 million).

3.1 Methodology

In order to create a counterfactual that takes the economic outlook of Turkey into account
we rely on autoregressive distributed lag models. This type of econometric model is fed
with historical data on the variable that needs to be forecasted (say the number of men in
the construction sector), past data on monthly or quarterly macroeconomic indicators3 as
well as the month of the year -to predict potential seasonality patterns. The macroeconomic
indicators that are fed into the econometric model are gathered from Turkstat and include
the following items:

3
Macroeconomic indicators are collected up to January 2020. No indicator is used beyond that date since they were
affected by the pandemic too.
 Research Brief
The impact of the pandemic on employment in Turkey:
What would have happened without COVID-19?
3
1. Consumers’ confidence index 6. Exports’ index
2. Retail confidence index 7. Food exports index
3. Construction confidence index 8. Number of construction permits
4. Industry output index 9. Construction cost index
5. Inflation rate 10. Exchange rate USD/Turkish Lira

All of these indicators are produced on a monthly basis with the exception of the number
of building permits, which is released quarterly. Thanks to this information the econometric
model is able to relate changes in the macroeconomic indicators from the February
2014-January 2020 period with changes in employment levels occurred during the same time
AllAllofof
span. these
theseindicators
These indicatorsare
relationships areproduced
between
produced ononamacroeconomic
amonthly
monthlybasisbasiswith the
theexception
indicators
with and
exception ofofemployment
the
thenumber
numberofofbuilding
levels
building are
permits,
then used which
to is released
forecast what quarterly.
would Thanks
have to this
happened information
between the econometric
February
permits, which is released quarterly. Thanks to this information the econometric model is able to 2020model
to is
Augustable to
2020
relate
had notchanges
relate the ininthe
pandemic
changes themacroeconomic
arrived.
macroeconomic indicators
indicatorsfromfromthetheFebruary
February2014-January
2014-January2020 2020period
periodwith with
changes
changes in employment levels occurred during the same time span. These relationshipsbetween
in employment levels occurred during the same time span. These relationships between
For this brief 25 models are
macroeconomic estimated using datathen
from February 2014 to January 2020.
macroeconomicindicators indicatorsand andemployment
employmentlevelslevelsare are thenusedusedtotoforecast
forecastwhat
whatwouldwouldhave have
Thanks
happened to these models 29 4
employment time-series are forecasted between February 2020
happenedbetween betweenFebruary
February20202020totoAugust
August2020
2020had hadnot
notthe
thepandemic
pandemicarrived.
arrived.
to August 2020. The employment levels forecasted refer to economic activities (11), status 5
For
Forthis
thisbrief
brief25256models
models areareestimated
estimated using data
datafrom
fromFebruary 2014 to January
January2020. Thanks
Thanksto tothese
in employment by formality 7
in theusing
non-agricultural February
sector2014
(7),toage-group 2020.
(youth 15-24 theseand
models
models 29 4
29 4employment time-series are forecasted between February 2020 to August 2020. The
employment time-series are forecasted between February 2020 to August 2020. The
adults 25+) by gender (4), by formality for all activities (2)5 and for the whole economy (1).7
employment
employmentlevels levelsforecasted
forecastedreferrefertotoeconomic
economicactivities
activities(11),
(11),status
5
statusininemployment
employmentby
6
6 formality 7
by formality
ininthethenon-agricultural
non-agriculturalsector sector(7),
(7),age-group
age-group(youth
(youth15-24
15-24andandadults
adults25+)
25+)bybygender
gender(4),
(4),bybyformality
formality
for
for all
allactivities
activities (2)
(2)and
and for
forthe
thewhole
whole economy
economy (1).
(1).
3.2 Overall impact
3.2
3.2Overall
Overallimpact
An example ofimpact
what forecasted employment levels look like is provided in Figure 1.a for total
AnAn example ofwhat
example
employment. of whatforecasted
According to the
forecasted employment
forecast, levels
employment levelslook
without like
likeisisprovided
COVID-19
look ininFigure
Turkey’s
provided 1.a
total
Figure for
fortotal
1.aemployment
total was
employment.
expected to have
employment. According
increased
According to the
to the forecast,
inforecast,without
comparison
without COVID-19
toCOVID-19 Turkey’s
what happened total employment
in 2019
Turkey’s total although
employment was expected
wasit expected
would not
totohave
have haveincreased
reached 2018
increased ininlevels.
comparison
comparisontotowhat
whathappened
happenedinin2019
2019although
althoughititwould
wouldnot nothave
havereached
reached
2018
2018levels.
levels.
Figure
XFigure 1. Impact
1. Impact of COVID-19
of COVID-19 on overall
on overall employment
employment
Figure 1. Impact of COVID-19 on overall employment
a)a)Forecasted
a) Forecasted andand
Forecasted andactual
actual employment
employmentlevels
employment
actual levels
levels b)b)Employment
b)Employment
Employmentlosses
lossesininin
losses %%(February-August)
%(February-August)
(February-August)

3030
ua y
br ar

M y

gu s t
ar h
r
Fe bru

M arc

Au ugu
st
Ap ril

Ju ne
ch

M ay

Ju ly
ril
Ap

ne
ay
Fe

Ju

Ju
Employment (millions)

ly

2929
A

00
Employment (millions)

2828 -2-2
2727 -4-4
-2,8
-2,8
2626 -6-6 -4,7
-5,8 -4,7
2525 -8-8 -5,8
-6,9
-6,9
2424 -10
-10 -8,1
-8,1
-12
-12 -10,5 -10,2
-10,2
2323 -10,5
-14
Jan
Jan

Jan

-14
Jul
Oct

Jul
Oct

Jul
Apr
Apr

Apr
Jan

Jan

Jan
Jul
Oct

Jul
Oct

Jul
Apr

Apr

Apr

2018 2019 2020 Source: Turkstat statistics and ILO’s own calcula-
2018 2019 2020
tions. Notes: The figures provide employment loss-
Real data (with COVID-19)
Real data (with COVID-19) es caused by COVID-19 as a percentage of the base-
Counter-factual
Counter-factual(without
(withoutCOVID-19)
COVID-19) line (the counterfactual). Said baseline represents a
scenario without COVID-19.
Source:
Source:Turkstat
Turkstatstatistics
statisticsand
andILO’s
ILO’sown calculations.Notes:
owncalculations. Notes:The
Thefigures
figuresprovide
provideemployment
employmentlosses
lossescaused
causedbybyCOVID-19
COVID-19
asasa apercentage of the baseline (the counterfactual). Said baseline represents a scenario without COVID-19.
percentage of the baseline (the counterfactual). Said baseline represents a scenario without COVID-19.
4
There are more variables forecasted than models estimated because some variables are put together. We forecast
the employment level of young women and young men separately, but then we may also create the variable
employment for young people as the sum of these two variables. Same goes for adults. Results for each gender are
aggregated
4
There areinmore
a similar fashion.
variables forecasted than models estimated because some variables are put together. We
4
There are more variables forecasted than models estimated because some variables are put together. We
5 forecast
NACE the
Rev.2
forecast employment
employmentlevel
codes.
the levelofofyoung
youngwomen
womenand
andyoung
youngseparately,
separately,but
butthen
thenwe
wemay
mayalso
alsocreate
createthe
the
6 variable
variable
Status employment
in employment
employment for
foryoung
youngpeople
follows people
the asasthe
ICSE-93 thesum
sumofofthese
thesetwo
classification. twovariables.
Four variables.Same
statuses Same
are goes
goesfor
foradults.
identified, adults.Results
Results
employees, for
for
own-account
gender
workers are
areaggregated
(self-employed
gender inina asimilar
without
aggregated fashion.
employees),
similar fashion.employers (self-employed with employees) and unpaid family workers.
7
5
A
NACE Rev.2 codes.
5
NACE Rev.2
worker codes.as formally employed if he or she is registered with the Social Security Institute.
is deemed
6
Status
6
Statusininemployment
employmentfollows
followsthe
theICSE-93
ICSE-93classification.
classification.Four
Fourstatuses
statusesare
areidentified,
identified,employees,
employees,own-
own-
account
accountworkers
workers(self-employed
(self-employedwithout
withoutemployees),
employees),employers
employers(self-employed
(self-employedwith
withemployees)
employees)and
andunpaid
unpaid
family
familyworkers.
workers.
4 Research Brief
The impact of the pandemic on employment in Turkey:
What would have happened without COVID-19?

The average8 decline in employment for the February-August period is estimated at 7.0
per cent although there is significant variability throughout the period. As it can be seen
in Figure 1.b the impact of the pandemic reached its peak in the months of April and May
where slightly more than 10 per cent of Turkey’s expected employment was lost. Since May
employment levels are moving towards their expected levels without pandemic, especially
since some of the restrictions were lifted back in June. However, as of August 2020, 4.7 per
cent of the jobs are still missing.

3.3 Groups at risk

Even though the COVID-19 disease does not make distinctions in terms of whom to infect the
economic effects of the pandemic do not apply equally to everyone. Workers with less secure
labour contracts, those who are not registered with the Social Security Institute or those who
are particularly attached to economic activities where human interaction and social distance
are hard to keep will most likely be more affected by the pandemic.

As such, we can identify informal workers and young people, especially young women,
to be some of the groups that endured heavier employment losses since February 2020.
There also exist overall gender differences (see Figure 2a) in terms of employment losses,
however, as it can be seen in Figure 2d gender differences are mostly due to the heavier
impact sustained by younger women. This group of women are not only shouldering one of
the greatest employment losses due to the coronavirus crisis (25.5 per cent jobs lost at the
peak of the pandemic) but are also more likely than young men not to ever come back to
the labour market. For example, male NEETs aged 15-24 have a 50.6 per cent probability of
exiting the NEET status within a year9 while that probability is just 23.5 per cent for young
female NEETs.

Major differences in terms of employment losses can also be found between formal and
informal workers, see Figure 2.b. The cause behind this formal/informal dichotomy is rooted
in the lack of protection and the vulnerability of informal employed workers. This lack of
protection have been exacerbated by new regulations that de facto banned the dismissal of
formal employees. The scope of these changes in the law is twofold; on the one hand they
provide a range of options for employers to keep their workers under different schemes,
on the other hand, it bans contract termination altogether. With respect to the former, the
government started using mechanisms like the reduced hours support (kısa çalışma desteği)
which already existed in the law under a simplified application procedure and by gradually
extending its duration. In addition to encouraging applications to the reduced hours support
scheme, employers have also been encouraged to implement teleworking whenever feasible
using the regulations stated in Article 14 of the Labour Law (amendments added in 2016).
And even if these two measures could not be applied due to business closure for COVID-19
related reasons employers still have the option of grating unpaid leave to their employees.
With regards to the prohibition of contract termination the government enacted the ban on
17 April 2020 for three months, ban which later was extended.

The average impact is defined as 100 (∑ August Lp- ∑ February Lf) where Lp denotes employment levels during the
8 February August

(∑ August
February L )
f

pandemic and Lf denotes forecasted employment levels in the absence of the pandemic.
9
The probabilities make reference to the recall module of the Household Labour Force Survey 2019. They refer to
the share of young people who were a NEET one year before the survey.
10
With some exceptions.
 Research Brief
The impact of the pandemic on employment in Turkey:
What would have happened without COVID-19?
5

X Figure
Figure
Figure 2.
2. Employment
2. Employment
Employment losses’
losses’
losses’ evolution,
evolution,
evolution, February-August
February-August
February-August
Figure 2. Employment
Figure 2. Employment losses’ evolution,
losses’ February-August
evolution, February-August
a) By sex a) Bya)sexBy sex b) By
b) Byb) formality
By formality
formality ofofwork
of work
work arrangements
arrangements
arrangements
FebruaryFebruary
a)
March March
By sex
a)
April April
By sex
May MayJune June
July August
July
b)
August
By formality
b) By
FebruaryFebruary
March March
of
formality work
April April
of arrangements
work
May May
arrangements
June June
July August
July August
0 0
FebruaryFebruary
March March
April May June July August
April May June July -4,4FebruaryFebruary
-4,4 August March March
April May
April June
May July
June August
July
-4,4 August
-4,4
0 0 -6,6 -6,6
-5 -5 -4,4 -4,4 -4,4 -4,4
(%) (%)

(%) (%)

-9,7 -9,7 -6,6 -6,6

-5 -5 -5,5 -5,5
change

change

-10 -10 -9,7 -9,7

-5,5 -5,5 -8,6 -8,6
change

-10
change

-10
-15 -15 -11,9 -11,9 -8,6 -8,6
Percentage

Percentage

-15 -15 -11,9 -11,9 -21,7 -21,7

Percentage

-20 -20
Percentage

-21,7 -21,7
-20 -20
-25 -25
-25 -25 Women
Women Men Men FormalFormal Informal
Informal
-30 -30 Women
Women Men Men FormalFormal Informal
Informal
-30 -30
c) By c)ageBygroup
age group d) Byd)sexByand
sexageandgroup
age group
c) By age group d) By sex and age group
FebruaryFebruaryc) By
March March age
c) Bygroup
Mayage
April April group
May
June June
July August
July August d) By
FebruaryFebruary sex
d) By
March March and
sexage
April April and
May Maygroup
age group
June June
July August
July August
0 0 -3,4 August
-3,4FebruaryFebruary
FebruaryFebruary
March March
April May
April June
May July
June August
July March March
April May
April June
May July
June August
July August
0 0 -3,4 -3,4
-5 -5 -8,1 -8,1
(%) (%)

(%) (%)

-3,9 -3,9
-5 -5 -8,1 -8,1
change

change

-10 -10 -3,9 -3,9

-14,8 -14,8 -8,9 -8,9
change

change

-10 -10
-15 -15 -14,8 -14,8 -8,9 -8,9
Percentage

Percentage

-15 -15 -21,3 -21,3

-20 -20
Percentage

Percentage

-17,7 -17,7
-21,3 -21,3 -25,5 -25,5
-20 -20 -17,7 -17,7
-25 -25 -25,5 -25,5
-25 Young Young Adult Adult Young Young
men men YoungYoung
women women
-25
-30
Young Young Adult Adult -30
Young men men
Adult men
Adult
Young men Young women
Adult women
Adult
Young women
women
-30 -30
Source:
Source:
Turkstat
Turkstat
statistics
statistics
and ILO’s
andown
ILO’scalculations. Notes:Notes:
own calculations. The figures Adult
The figures
provide men
Adult
provide men lossesAdult
employment
employment women
Adult
losses
caused women
caused
by COVID-19
by COVID-19
as as
a percentage
aSource:
Source: percentage
ofTurkstat
Turkstatthe of
baseline
the
statistics baseline
and(the counterfactual).
ILO’s(the
own counterfactual).
Said
calculations. baseline
Said
Notes: baseline
Therepresents
represents
figures a scenario
provide a scenario
without
employment without
COVID-19.
losses COVID-19.
caused by COVID-19 as
statistics and ILO’s own calculations. Notes: The figures provide employment losses caused by COVID-19 as
a percentage of the of
a percentage baseline (the counterfactual).
the baseline Said baseline
(the counterfactual). represents
Said baseline a scenario
represents without
a scenario COVID-19.
without COVID-19.
Source: Turkstat statistics and ILO’s own calculations. Notes: The figures provide employment losses caused by
COVID-19
In sum, as a percentage
In sum,
the lack
the of of the
lacksocial
of baseline
social
protection (the
protection counterfactual).
brought
brought
by informal Said
by informalbaseline
work work represents
arrangements acombined
arrangements scenario without
combined COVID-19.
with with
the the
measures
InIn measures
sum,
In
sum, the
sum,taken
the lack
the taken
lackbyof the
lack byofgovernment
social
of the
social government
protection
social in terms
protection
protection in terms
broughtof protecting
byofinformal
brought
brought protecting
by by jobs results
work
informal jobs
informal results
inwork
marked
in marked
arrangements
work arrangementsdifferences
differences
combined
combined
arrangements between between
withcombined
the the
with
the share
the share
measures of jobs
taken
measures ofbylost
jobs
taken thebylost
formal
bygovernment
formal
government
the and informal
and
in informal
terms
in employees.
of
terms employees.
protecting
of On the
jobs
protecting On positive
theresults
results
jobs positive
in side,
marked
in side,
thedifferences
damage
marked thedifferences
damagebrought
betweenbrought
by by
between
with the measures taken by the government in terms of protecting jobs results in marked
pandemic
the share pandemic
of jobs
the share to informal
oflost
jobstoby informal
workers
formal
lost by workers
and
formal seem
and seem
informalto be totemporary
be
employees.
informal temporary
Onasthe
employees. these
asthe
these
workers
Onpositive workers
side,
positive are
therecovering
side, are
therecovering
damage damage jobsbrought
brought at
jobs
a atby
by a
differences between the share of jobs lost by formal and informal employees. On the positive
fast pandemic
the pace.
fast pandemic
the pace.to informal to informalworkers seemseem
workers to betotemporary
be temporaryas these workers
as these are recovering
workers are recovering jobs at a at a
jobs
side, the damage brought by the pandemic to informal workers seem to be temporary as
fast pace.
fast pace.
3.4 The
these 3.4role
Theof
workers role informality
are of informality
recovering jobs at a fast pace.
3.4 The
3.4role
Theof role informality
of informality
Regulations
Regulationsbanning banningdismissals
dismissals
only only
concern
concernemployees,
employees,not self-employed
not self-employed workersworkersand thus, and thus,
differences
differences
Regulations inbanning
Regulations theinbanning
impact
thedismissals
impact
ofdismissals
COVID-19
of only
COVID-19
are expected
are expected
concern
only concern to
employees,betolarger
employees,be
notlarger
among
not amongsalaried
self-employed salaried
self-employed workers.
workers workers.
workersIndeed,
and Indeed,
thus,
and it thus,
it
3.4
canThe
be role
canseen
differences
differences of
beinseen
in the informality
Figure
inin Figure
impact
the 3impact
that
of3COVID-19
at
that
ofthe atpeak
the
arepeak
COVID-19 of theof
expected
are pandemic
thetopandemic
expected the
be differences
betolarger the differences
among
larger in salaried
salaried
among thein share
theworkers.
workers. share
ofIndeed,
jobs ofIndeed,
lost
jobs
it lost
it
between
can between
be formal
seen in formal
can be seen that
The regulations and
Figure informal
and
3
in Figure informal
that
banned atworkers
the workers
peak
3 thatdismissalsareof 28.1,
are
the
at the peakonly 28.1,
12.0
pandemic
of the 12.0
and 9.7
and
the
pandemic
concern percentage
9.7 percentage
differences
the differences
employees, points
in the points
for,
share respectively,
for,
of respectively,
jobs
in the share of jobs
not self-employed lost lost
workers
employees,
employees,
between
between employers,
formal employers,
and informal
formal andandinformal
own-account
andworkers
own-account
workersworkers.
are workers.
28.1,
are 12.0 12.0
28.1, and 9.7and percentage
9.7 percentagepointspoints
for, respectively,
for, respectively,
and thus, differences in the impact of COVID-19 are expected to be larger among salaried
employees, employers,
employees, employers, and own-account
and own-account workers.
workers.
workers.
Even Even Indeed,
though though
smaller, it canthebe
smaller, seen
relationship
the in Figure
relationship
between 3 that
between at the and
informality peakemployment
informality of the
and pandemic
employment isthe
losseslosses stillisdifferences
still carried in
carried
forward
Even
the forward
share
Evento
thoughofself-employed
to
jobs
though self-employed
smaller, lost the
smaller, individuals.
individuals.
relationship
between formal
the relationshipAn explanation
An
between
and explanation
for
informality
informal
between this
for phenomenon
this phenomenon
and employment
workers
informality andare may may
be related
losseslosses
28.1, 12.0
employment be related
is still
and to
is carried
9.7 the
to the
percentage
still carried
forward
points to respectively,
forward
for, self-employed
to self-employed individuals. Anemployers,
individuals.
employees, explanation
An explanation for this
and for phenomenon
this phenomenon
own-account may may
workers. be related
be related to theto the

Even though smaller, the relationship between informality and employment losses is still
carried forward to self-employed individuals. An explanation for this phenomenon may be
related to the higher costs of closing and re-opening a formal business; they tend to have
more structure and a higher number of workers. But it might also be related to the support
measures enacted by the government in an attempt to ease the impact of the pandemic. In
fact, the 21 points of the economic stability protection package (ekonomik istikrar kalkanı)
included loan payment deferrals, the deferral of social security premiums for selected
6 Research Brief
higher
The
What
impact
higher
costs
higher
higher
would
ofpandemic
costs
of the
number
closing
of
number
higher costscosts
have of
and
of closing
on
workers.
re-opening
and re-opening
employment
But
workers.
of closing
of closing
happened it
But it
and re-opening
and re-opening
without COVID-19?
a formal
in Turkey:
might also
might be
business;
a formal
also be
a formal
a formal
they they
business;
related to
relatedthe
business;
business; to
tend tend
the
they they
to have
support
tend tend
moremore
to have
measures
support
to have
to have
structure
enacted
measures
moremore
and aand a
structure
by
andthe
enacted
structure
structure by thea
aand
government
highergovernment
higher
number in an
ofinattempt
number an toBut
ofattempt
workers.
workers. ease the
toitBut
easeitimpact
might the
might
also of
impactthe
also
be pandemic.
ofbethe
related Insupport
pandemic.
related
to theto fact, themeasures
21
In support
the fact, thepoints of theof
21 points
measures economic
the
byeconomic
enacted
enacted by the
the
stability
governmentprotection
government an in package
stabilityinprotection
an
attemptattempt
to (ekonomik
package to
ease (ekonomik
ease
the istikrar
the
impact kalkanı)
istikrar
impact
of theof theincluded
kalkanı)pandemic.
pandemic. In loan
included In
fact,payment
loan
fact,
the the
21 deferrals,
payment21 points
points of the
deferrals,
theof deferral
theeconomic
the
economic deferral
industries and the cancelation of the accommodation tax until November among other
of social security
of social
stability
stability premiums
security
protection
protection premiums
package for
package selected
for
(ekonomik
(ekonomik industries
selected and
industries
istikrar
istikrar the
and
kalkanı)
kalkanı) cancelation
the cancelation
included
included of the
loan
loan payment ofaccommodation
the
payment accommodation
deferrals,
deferrals, tax
theuntil
tax until
deferral
the deferral
measures which obviously only benefit formal companies.
November
of of
social among
November
social other
among
security
security measures
other
premiums
premiums forwhich
formeasures selected
selected obviously
which only
obviously
industries
industries and benefit
only
and
the formal
thebenefit ofcompanies.
formal
cancelation
cancelation companies.
theofaccommodation
the accommodation tax until
tax until
November
November among among
otherother measures
measures which which obviously
obviously only benefit
only benefit formal formal companies.
companies.
X Figure
FigureFigure 3.
3. Employment
3. Employment
Employment losses losses’
losses evolution,
evolution, February-August
non-agricultural
evolution, non-agriculturalworkers, February-August
workers, February-August
FigureFigure 3. Employment
3. Employment losses losses evolution,
evolution, non-agricultural
non-agricultural workers,
workers, February-August
February-August
a) Employees
a) Employees a)
(formal/informal)
Employees
(formal/informal) (formal/informal) b) Employers
b) Employers
b) Employers
(formal/informal)
(formal/informal)
(formal/informal)
a) Employees
a) Employees (formal/informal)
(formal/informal) b) Employers
b) Employers (formal/informal)
(formal/informal)
February March March
February April May
April June
May July
June August
July February
August March March
February April May
April June
May July
June August August
July
0 0
FebruaryFebruary
March March
April April
May May
June June
July July
August August
FebruaryFebruary
March March
April April
May May
June June
July July
August August
0 0
(%) (%)

(%) (%)

-4,8 -4,8 -11,3 -11,3

-10 -10 -14,9 -14,9
-7,3 -7,3
change

change

-4,8 -4,8 -11,3 -11,3

-10 -10 -11,4 -11,4 -14,9 -14,9
-7,3 -7,3 -12,5 -12,5
change
change

-20 -20 -11,4 -11,4 -12,5 -12,5

Percentage

Percentage

-20 -20
Percentage
Percentage

-30 -30 -35,4 -35,4 -26,9 -26,9

-30 -30
-35,4 -35,4 -26,9 -26,9
FormalFormal
employers
employers
-40 FormalFormal
employees
employees
-40
Informal
Formal employees
Informal
Formal employees
employees
employees Informal
Formal employers
Informal
Formal employers
employers
employers
-40 -40
Informal
Informal employees
employees Informal
Informal employers
employers
c) OAW (formal/informal)
c) OAW (formal/informal) d) Unpaid family
d) Unpaid workers
family workers
c) OAW
c) OAW (formal/informal)
c) OAW (formal/informal)
(formal/informal) d) Unpaid d) Unpaid
family
d) Unpaid family
workers
family workers
workers
February March March
February April May
April June
May July
June August August
July February March March
February April May
April June
May July
June August August
July
5 5
FebruaryFebruary
March March
April April
May May
June June
July -0,8 August
July
August -0,8
FebruaryFebruary
March March
April April
May May
June June
July July
August August
50 0
5 -5,8 -5,8 -0,8 -0,8 Unpaid familyfamily
Unpaid workers
workers
-5
0 -50
(%) (%)

(%) (%)

-5,8 -5,8
-10
-5 -10
-5 UnpaidUnpaid
family family workers
workers
-6,5 -6,5
change

change

-15
-10 -15
-10 -6,5 -6,5
change
change

-20
-15 -20
-15 -15,5 -15,5
Percentage

Percentage

-25
-20 -25
-20 -15,5 -15,5
Percentage
Percentage

-30
-25 -30
-25
-35 FormalFormal
OAW OAW
-30 -35
-30
-40 -40 Informal
Formal OAWOAW
Formal
Informal
OAW OAW -42,1 -42,1
-35 -35
-45
-40 -45
-40 Informal
Informal OAW OAW -42,1 -42,1
-45 -45
Source: Turkstat
Source: statistics
Turkstat and ILO’s
statistics and own
ILO’scalculations. Notes:Notes:
own calculations. The figures provide
The figures employment
provide losseslosses
employment causedcaused
by by
COVID-19 as aTurkstat
COVID-19 percentage
as of theand
a percentage baseline
of the (the counterfactual).
baseline SaidThe
(the counterfactual). baseline
Said represents
baseline a scenario
represents alosseswithout
scenario COVID-19.
without
Source:
Source: Turkstat statistics
statistics and ILO’s ILO’s
own own
calculations. Notes:Notes:
calculations. The figuresfigures provide
provide employment
employment losses
caused by COVID-19.
caused by
Source: Turkstat
COVID-19 as astatistics
COVID-19 percentageand ILO’s
as a percentage
of the of own
the calculations.
baseline
baseline (the Notes:SaidThe
(the counterfactual).
counterfactual). Saidfigures provide
baseline
baseline employment
represents
represents a scenario
a scenario lossesCOVID-19.
without
without caused by
COVID-19.
In spite
COVID-19
of the
In spite ofsupport
as a percentage
measures
the support
of themeasures
announced
announced
baseline (the
by the
bygovernment,
the Said
counterfactual). government,the number
the number
baseline represents
of aformal employers
of formal
scenario
in in
employers
without COVID-19.
thespite
In non-agricultural
thespite
In ofnon-agricultural
theofsupport sector
the support (see
sector Figure
(see
measures
measures 3b) kept
Figure 3b)
announced
announced going
bykept down
going until until
down
bygovernment,
the the July,
government, tothe
July,
the numberjust
tothen
just
number see
of aformal
then
of formal timid
see signal
a timid of of
signal
employers
employers in in
Inthespite
recovery of
recoveryin the
August.
in
non-agricultural support
The
August.
the non-agricultural The measures
decrease
sectorsectordecreasein thein
(see Figure
(see Figure announced
number
the of
number
3b) going
3b) kept formal
of
kept going by the
employers
formal
downdown government,
outside
employers
until to
until July, July, of
outside
justtothen the
theof
just then thenumber
agricultural
see asee of
sector
agricultural
a timid
timid formal
sector
signalsignal
of of
employers
is justisthe
just
recovery
recovery inincontinuation
the the
incontinuation
August. non-agricultural
August.
The of a trend
The of inthat
a trend
decrease
decrease thein sector
started
that
the inof
started
number number (see
April
in of Figure
2018.
April
formal At that
2018.
formal 3b)
At that
employers
employers kept
point there
point
outside going
were
there
outside
of down
1.07
were 1.07
theofagricultural
the until
million such
million
agricultural
sector July,
suchto
sector
employers
just employers
then
is justis the in
see Turkey
aintimid
justcontinuation These
Turkey
the continuation employers
These
ofsignalof aemployers
a trend of
trend
that are
recoverythe
that are ones
inthe
started
started in with
ones
August. more
with
in April
April 2018. assets
more
AtThe
2018.
that at
assetsstake
Atdecrease
that point
point there and
at stake
in
there a
the
were pandemic
andwere
1.07anumber is
pandemic by
1.07 million
million far
ofis formal
such by far
such
one ofone
employers
employers theofin
employers worst
the inscenarios
worst
Turkey
outside These toagricultural
ofscenarios
Turkey These
the setto upseta are
employers
employers new
up company.
athe
new
are company.
the
ones
sector ones
withYetmore
is just understanding
Yet
with
the understanding
more
assets assets the
at stake
continuation motives
the
at stake
and and behind
ofamotives business
behind
a pandemic
apandemic
trend business
is by farin
is bystarted
that far
closures
one
April of thedoes
closures
one
2018. ofworst
the
At not
does
worst
that make
not itto
make less
scenarios
scenarios
point set
there painful
it to
lessset
upwere upfor
apainful
new the
a company.
new
1.07 country
formillion
company. since
the country
Yetsuch formal
Yetsince businesses
formal
understanding
understanding
employers thebusinesses
the
motives
in arebehind
motives
Turkey. both,
are the
both,
behind
These ones
the
business ones
business
employers
creating
closures most
creating
closures
does formal
most
does
not jobs
formal
not
make itand
jobs
make less the
itand
lessones
the
painful withthe
ones
painful
for showcasing
with
for showcasing
the
country higher
country
since levels
higher
since
formal of productivity.
levels
formal of productivity.
businesses
businesses are are
both, both,
the the
ones ones
are the ones with more assets at stake and a pandemic is by far one of the worst scenarios
creating most formal jobs and the ones with showcasing
creating most formal jobs and the ones with showcasing higher levels of productivity. higher levels of productivity.
to set up a new company. Yet understanding the motives behind business closures does not
make it less painful for the country since formal businesses are both, the ones creating most
formal jobs and the ones with showcasing higher levels of productivity.

Last but not least, the pandemic took away up to 42.1 per cent of the jobs held by unpaid
family workers -all of whom are considered to be working informally. However, these jobs
are being recovered with as much speed as the one with which they were lost, see Figure
3.d. It is not surprising, though, given the fact that they are not registered and the cost of
dismissing them is close to nothing.

The dramatic employment losses faced by informal employees and unpaid family workers
may also be taken as a better measure of the actual impact of the pandemic on the economic
activity. This is because formal employees could hardly be dismissed yet many were not
 Research Brief
The impact of the pandemic on employment in Turkey:
What would have happened without COVID-19?
7
Last but not least, the pandemic took away up to 42.1 per cent of the jobs held by unpaid family
working or were
workers -all of working
whom are less hourstothan
considered before.
be working In the However,
informally. absence these
of statistics on hours
jobs are being
recovered
worked with asto
the extent much
whichspeed as the one with
dismissible which they
workers arewere lost, see
sacked fromFigure 3.d.companies
their It is not surprising,
hints that
though,of
the impact given
thethe fact that they
pandemic mayare be
not much
registered and the
larger costthe
than of dismissing
mere 7.3 them
peris cent
close to nothing.
found among
formal employees.
The dramatic employment losses faced by informal employees and unpaid family workers may also
be taken as a better measure of the actual impact of the pandemic on the economic activity. This is
because employees could not be dismissed yet many were not working or were working less hours
than before.
3.5 What sectors In the
areabsence of statistics on hours worked the extent to which dismissible workers are
in trouble?
sacked from their companies hints that the impact of the pandemic may be much larger than the mere
Not all
7.3economic activities
per cent found amonghave
formalbeen equally affected by the pandemic. Some, like education,
employees.
public administration, and business activities can, to a great extent, be executed online -even
if the quality of such services is somewhat compromised. Others like human health activities
do not 3.5have
What yetsectors
theare in trouble?
ability to be carried out online. However the ease with which the new
coronavirus is spread, the
Not all economic activities havefactbeen
thatequally
muchaffected
of thebysector’s staffSome,
the pandemic. are either civil servants
like education, public or
formally hired employees
administration, and business and the ban
activities imposed
can, to by the
a great extent, government
be executed online not
-evenallowing staff to
if the quality
leave oftheir
suchposts
servicesduring the beginning
is somewhat compromised. of the
Others pandemic
like human ishealth
translated into
activities do anot
very
have mild sectoral
yet the
impact.ability to beactivities
Some carried outlike
online. However
trade the ease with
or transport havewhich the new coronavirus
showcased is spread,employment
slightly higher the fact
that much of the sector’s staff are either civil servants or formally hired
losses at 6.7 and 6.0 per cent. However the effects of the pandemic has been particularly employees and the ban felt
imposed by the government not allowing staff to leave their posts during the beginning of the
by the construction sector, hospitality activities and the “other” services group, which include
pandemic is translated in a very mild sectoral impact.11 Some activities like trade or transport has
activities like repair of computers and personal
showcased slightly higher employment losses at 6.7 and 6.0 and household goods,the
per cent. However beauty
effects salons
of the and
textilepandemic
washing hasservices amongfelt
been particularly other.
by the construction sector, hospitality activities and the “other”
services group, which include activities like repair of computers and personal and household goods,
X beauty 4.
Figure salons and textile
Average impact washing services among
on employment other.
(February-August) by economic activity
Figure 4. Average impact on employment (February-August) by economic activity

Agriculture Industry Construction Trade Transport Hospitality Business act.Government Education Health Other serv.
0

-2,5
Percentage change (%) in

-5 -2,8
-5,5 -5,8 -6,0 -5,1 -5,5
-6,7
employment

-10

-15
-14,1
-15,5
-20

-21,0
-25

Source: Turkstat statistics and ILO’s own calculations. Notes: The figures provide the average employment losses caused by
COVID-19 between February to August 2020 as a percentage of the baseline (the counterfactual). Said baseline represents a
Source:scenario
Turkstat statistics
without andActivities
COVID-19. ILO’s own calculations.
are defined Notes:
using NACE The figures
rev.2 codes. provide
Agriculture refers tothe average
section employment
A, industry to Sections losses
causedB,by COVID-19
C, D between
and E, construction to February to August
section F, trade 2020
to section as a percentage
G, transport to section H,of the baseline
hospitality (the
to section counterfactual).
I, business activities Said
baseline
to sections J, K, L, M and N, government to section O, education to section P, health to section Q and other services to section refers
represents a scenario without COVID-19. Activities are defined using NACE rev.2 codes. Agriculture
to section
R, S, A, industry
T and U. to Sections B, C, D and E, construction to section F, trade to section G, transport to section H,
hospitality to section I, business activities to sections J, K, L, M and N, government to section O, education to section
Factors
P, health that Q
to section explain the strong
and other servicesnegative impact
to section R, S, experienced
T and U. by restaurants, bars and hotels are the
measures taken by the government in April 2020, the relatively high informality rate showcased by
Factors that explain the strong negative impact experienced by restaurants, bars and hotels
are the measures taken by the government in April 2020, the relatively high informality rate
11
And the little impact recorded most likely refers to social work services, not to human health.
showcased by the sector, social distancing as well as the fear induced by the coronavirus. In
fact, at the peak of the pandemic (May) employment losses faced by the hospitality sector
reached 30.6 per cent of the employment that would have existed without COVID-19. Similar
explanations apply for the activities included under the “other services” label, since many of
them are face to face activities and some of them even require close contact with customers.

The case of the construction sector is slightly different; even though it ranks third in terms
of employment losses, most of them are derived from jobs that were never created. That
is, the hypothetical scenario without COVID-19 forecasted by the econometric model was

11
And the little impact recorded most likely refers to social work services, not to human health.
8
the Brief
Research sector and social distancing as well as the fear induced by the coronavirus. In fact, at the peak of
the pandemic
The impact (May)onemployment
of the pandemic losses
employment in faced by the hospitality sector reached 30.6 per cent of the
Turkey:
What would have happened without COVID-19?
employment that would have existed without COVID-19. Similar explanations apply for the activities
included under the “other services” label, since many of them are face to face activities and some of
them even
expecting require close
a significant contact
rise in thewith customers.
number of workers employed in this sector. The positive
outlook
The forecasted
case of the for the sector
construction is based
sector ondifferent;
is slightly the risingevenlevels
thoughofitconfidence
ranks third ininterms
the sector
of
in December 2019 and January 2020 and on the increase in the number
employment losses, most of them are derived from jobs that were never created. That is, thebuilding permits
requested during
hypothetical the 4th
scenario quarter
without of 2019.
COVID-19 Both measures
forecasted seem to,model
by the econometric and was
can expecting
be argued a to,
significant
lead and predictrisechanges
in the number of workers
in future employed inlevels
employment this sector.
in theThe positive outlook forecasted for
sector.
the sector is based on the rising levels of confidence12 in December 2019 and January 2020 and the
Leaving individual
increase in the cases
numberaside, informality
building can explain
permits requested duringa the
great
4th deal
quarterof of
variation
2019. Bothin employment
measures
destruction across sectors. Figure 5 offers a visualization of this phenomenon
seem to, and can be argued to, lead and predict changes in future employment levels in the sector. putting
together the sectors’ losses of employment (%) with their respective shares of informal
Leaving individual cases aside, informality can explain a great deal of variation in employment
workers. The existing relationship has been approximated with a linear regression line;
destruction among sectors. Figure 5 offers a visualization of this phenomenon putting together the
according tolosses
sectors’ this simple approximation
of employment (%) with the
theirshare of informal
respective shares ofworkers
informal explains
workers. The55 per cent of
existing
the variation in employment losses. Individual characteristics still matter,
relationship has been approximated with a linear regression line; according to this simple though. It can be
seen approximation
in the graph thethat theof
share hospitality sectorexplains
informal workers faced 55 higher employment
per cent losses
of the variation than it would
in employment
have losses.
been expected by its share still
Individual characteristics of informal workers.
matter, though. It canInbeother words,
seen in if were
the graph to only
that the consider
hospitality
sector faced higher employment losses than it would have been expected
informality we would have forecasted employment losses to be at 15 per cent instead of at by its share of informal
21 perworkers. In other
cent. This words, may
difference if were to only takebyaccount
be explained informality
the lockdown we wouldand
measures have
theforecasted
businesses’
employment losses to be at 15 per cent instead of at 21 per cent. This difference may be explained by
closures that were imposed at the beginning of the pandemic.
the lockdown measures and the businesses closures that were imposed at the beginning of the
pandemic.
X Figure 5. Informal employment and employment losses, by economic activity
Figure 5. Informal employment and employment losses, by economic activity

50
45
Share of informal workers (%)

Other serv.
40
Construction Trade
Hospitality Health 35
30
Transport 25
Industry 20
15
Business act.
y = -1,6252x + 10,416 Teaching 10
R² = 0,5578 5
Government
0
-25 -20 -15 -10 -5 0
Employment loss (%)
Source:
Source: HLFS 2019 4th quarter, Turkstat statistics and ILO’s own calculations. Notes: The figure shows economic activity-
Source:specific pairsHLFS
Source: of employment
2019 4thlosses due toTurkstat
quarter, COVID-19statistics
and pre-pandemic informality
and ILO’s rates. The red dotted
own calculations. Notes:lineThe
refers to the shows
figure
results
economic from a linear regression,
activity-specific pairs of the equation andlosses
employment the R2due
are also provided. The
to COVID-19 andagricultural sector has
pre-pandemic been excluded;
informality it isThe red
rates.
dottedregarded as an
line refers tooutlier.
the results from a linear regression, the equation and the R2 are also provided. The agricultural
sector has been excluded; it is regarded as an outlier.
The existence of other factors behind employment losses other than informality can be attested in
Table 1. In principle,
The existence of other all factors
disfavoured groups employment
behind showcased a higher informality
losses otherrate,
thanforinformality
example 41.2 per
can be
attested in Table 1. In principle, all disfavoured groups showcase a higher informality rate,
for example 41.2 per cent of young workers are informally hired in comparison with only
12
Turkstat Construction Sector Confidence Index.
33.3 per cent of adults. A similar story can be extracted for men (30.6 per cent) and women
(42.2 per cent). However, this empirical regularity is broken when we compare young (15-24)
and adult (25+) women, with respectively 36.0 per cent and 43.3 per cent informality rates.
That is, young women experienced stronger employment losses in spite of their contractual
arrangements being, on average, more secure.

12
Turkstat Construction Sector Confidence Index.
 Research Brief
The impact of the pandemic on employment in Turkey:
What would have happened without COVID-19?
9

X Table 1. Share of informal workers by gender and age-group, 4th quarter 2019

Group Informal (%) Group Informal (%)

Young people 41.2 Young women 36.0
Adults 33.3 Adult women 43.3
Women 42.3 Young men 43.6
Men 30.6 Adult men 28.7

Source: HLFS 2019 and ILO’s own calculations. Notes: Age-group and gender-specific rates of informality for
workers. Young refers to those aged 15-24 while adults are 25+ individuals.

In the case on younger vs. adult women we need to look at other factors such as tenure,
permanency of contracts and the type of economic activities to explain the observed
differences in employment losses. According to the 4th quarter of the HLFS 2019, it can be
seen that young women tend to be more involved in the hospitality sector (highly impacted)
while adult women work more often in the agricultural sector (mild impact). Moreover, young
female employees hold more temporary contracts (23.8 per cent) than the adult ones (8.2 per
cent). At last, it is also the case that adult women have more tenure than younger workers in
the sense of having spent more time at the same workplace; this is also associated to higher
job security levels.

4- Concluding remarks
The pandemic brought by COVID-19 has caused a significant disruption of the economic
activity in Turkey and has subsequently affected its labour market. Some of the effects on
the labour market include employment losses, reductions in hours worked and potentially
reduced wages. This research brief focuses on the former by measuring the impact of the
pandemic on overall employment as well as on the employment levels of several subgroups.
This exercise is carried out with the help of an econometric model that predicts what these
employment levels would have looked like had not the pandemic arrived in early 2020. The
comparison between the forecasts and the actual employment figures allows us to calculate
the causal impact of COVID-19 on the different groups of workers that are present in the
Turkish labour market.

According to the results the pandemic destroyed, at its peak back in May, nearly 3 million
jobs. However, job destruction was not shared equally among all types of workers. The
measures taken by the government with the intention of protecting employment worked
well in preventing massive lay-offs among formal workers. The same cannot be said about
those working without registration in the SSI; more than 1 in 5 informal jobs were lost in
May although the recovery process was fast once the lockdown measures were eased. In
fact, informality can be said to have shaped the impact of COVID-19 on employment. In
general, all areas of the labour market associated with high informality rates took a greater
impact. As such, women, young people, and those working in hospitality, construction and
low-skilled services lost a significant portion of their respective jobs.

In spite of the great importance behind the registration of workers in terms of employment
protection there exists other precarious forms of work that increase workers’ vulnerability.
This can be seen among young women (15-24) who have experienced heavier job losses than
10 Research Brief
The impact of the pandemic on employment in Turkey:
What would have happened without COVID-19?

adult women (25+) even though their formality rate is higher. In this case, other elements such
as temporary contracts seem to also be playing a role in reducing their protection. Moreover,
the very partial jobs’ recovery experienced by young women calls for a close monitoring of
their situation. Young women who are not in employment nor in education (NEET) have a
slim chance of finding a job in the near future and these chances keep decreasing the longer
they are away from the labour market.

Last but not least, the crisis brought by the novel coronavirus has delivered yet another
setback to Turkish formal employers. This group was totalling 1.074 million in April 2018,
went down to 0.832 million by October 2019 and when it started to recover (0.905 million
February 2020) the pandemic brought its number further down to 0.808 million in June 2020.
In light of the importance of formal employers in the creation of decent work further support
measures are due; otherwise there is a risk for an incomplete recovery both in qualitative
(contract quality) and quantitative terms (lower employment levels).

Even though this research brief sheds some light on the effects of the pandemic on the
Turkish labour market it only offers a partial view. Many workers have seen their hours
worked reduced or have been sent on unpaid leave. This analysis should also be carried out
in the future so as to have a more complete picture of the full impact of the pandemic.

X References

United Nations, 2020. Economic update, Office of the Resident Coordinator in Turkey.

DISK 2020. İşsizlik ve istihdamın görünümü raporu.

Şeker, S.D. & Özen, E.N. & Erdoğan, A.A. (2020). Jobs at risk in Turkey: Identifying
the impact of COVID-19. Social protection and jobs, discussion paper, No. 2004, July
2020.

ILO (2020). ILO monitor: COVID-19 and the world of work. Sixth edition.

X Contact

ILO Office for Turkey

Ferit Recai Ertuğrul Caddesi No. 4

06450 Oran, Ankara