South African labour market transitions during the global financial and economic crisis:

oppositemincedΔιαχείριση

28 Οκτ 2013 (πριν από 3 χρόνια και 9 μήνες)

64 εμφανίσεις

South African labour market transitions during
the global financial and economic crisis:


Micro
-
level evidence from the NIDS panel

Dennis Essers

Institute of Development Management and Policy (IOB)
University of Antwerp

Presentation at
the Arnoldshain Seminar XI “
Migration, Development, and Demographic Change:

Problems, Consequences and Solutions”

University of Antwerp, 27 June 2013, Session 3B (12:30


14:30)

Contents


Introduction


NIDS data description


Empirical model set
-
up and main results


Further probing


Concluding remarks

27/06/2013

2

Introduction


Many studies have documented macro
-
level impacts of
2008
-
2009 global crisis on developing and EM
economies: private capital flows, trade, remittances,
etc. (IMF 2009, 2010; ODI 2010; World Bank 2009)


South Africa was well
-
integrated into the world
economy and did not escape the crisis; entered
recession in 2008Q4, driven by decline in
manufacturing, mining, wholesale/retail trade and
financial/real estate/business services


Recovery has not been spectacular and punctuated by
renewed global economic slowdown


27/06/2013

3

27/06/2013

4

Annualised growth of (seasonally
-
adjusted) quarterly GDP


at constant prices (%)

6.5

3.1

5.0

6.0

3.0

4.4

1.8

-
1.7

-
6.3

-
2.7

1.7

3.5

4.4

3.1

3.6

4.4

4.8

1.9

1.9

3.3

2.5

3.4

1.2

2.1

0.9

-8.0
-6.0
-4.0
-2.0
0.0
2.0
4.0
6.0
8.0
Introduction (2)


Adverse macro
-
economic trajectory has not been without
consequences for South Africans (e.g.
Ngandu

et al.

2010)


Focus here on labour market transitions:


Official Quarterly Labour Force Survey (QLFS) figures indicate net
employment loss of about 1 million individuals over 2008Q4
-
2010Q3


Labour market status is critical determinant of household and
individual well
-
being (World Bank, 2012), also in SA (
Leibbrandt

et al.
2012)


(Pre
-
crisis) high and structural unemployment and segmented labour
markets described as SA’s “Achilles’ heel” (
Kingdon

& Knight 2009)


Complement to earlier crisis impact studies, which use repeated
cross
-
sections

of QLFS (Leung
et al.
2009;
Verick

2010, 2012)


Research question: which household
-
level, individual and
job
-
specific characteristics are associated with staying
employed, or not, in SA during the global crisis?


27/06/2013

5

27/06/2013

6

Total number of employed individuals aged 15
-
64


(in thousands)

14,027

12,975

13,621

12,750
13,000
13,250
13,500
13,750
14,000
14,250
Net employment loss
of +/
-

1 million

Net employment gain
of +/
-

650 thousand

Data description


National Income Dynamics Study (NIDS)
is SA’s first nationally
representative panel data survey


So far 2 NIDS ‘waves’ have been conducted, resulting in panel of
21,098 individuals appearing both in wave 1 (Jan
-
Dec2008) and
wave 2 (May2010
-
Sep2011)


NIDS combines household and individual questionnaires on various
topics: expenditure, demographics, health, education, labour
market participation etc.


Analysis of NIDS is a useful complement to existing studies on SA
labour markets during the crisis:


Convenient timing: before height of the global crisis and during timid recovery


Longitudinal character enables analysis of
gross

changes/transitions in labour
market participation


Labour market section contains detailed information on job history,
occupation/industry, hours worked, earnings and benefits, contract types,
unionisation, job search strategies, labour market expectations, etc.





27/06/2013

7

Data description (2)


Analysis here restricted to
‘balanced panel’ adults aged 20
-
55

in 2008


Four mutually exclusive groups/labour market statuses:


Employed

(regular wage/self
-
/casual/subsistence agriculture/assistance with
others’ business)


Searching unemployed


Discouraged unemployed


Not economically active

(NEA)


Cross
-
sectional

analysis of NIDS and comparison with QLFS suggests some
misclassification between different categories of the non
-
employed during
wave 2 fieldwork (SALDRU 2012)


NIDS data best
-
suited for
longitudinal

study of individual labour market
transitions; simplest representation by means of transition matrix for
different labour market statuses (
Cichello

et al.
2012)

27/06/2013

8

Employment status in 2010/11

Employment status in 2008

50.6

12.0

5.0

32.4

Employed

Unemployed,
search.

Unemployed,
disc.

NEA

53.0

Employed

71.6

6.7

3.2

18.5

18.5

Unemployed,
search.

32.3

21.6

6.5

39.7

6.3

Unemployed,
disc.

28.0

18.1

10.8

43.1

22.1

NEA

22.1

15.0

6.1

56.8

Transition matrix for employment
status

2008
-
2010/11: row proportions (%)

Transition matrix for employment status and type 2008
-
2010/11: row proportions (%)

Employment status
/type in
2010/11

Employment
status/type

in
2008

39.8

6.0

4.7

12.0

5.0

32.5

Reg. wage
empl oyment

Sel f
-
employment

Casual

and
other

empl oyment

Unemployed,
search.

Unemployed,

di sc
.

NEA

37.1

Reg. wage
empl oyment

76.4

3.2

3.2

5.3

2.7

9.3

7.4

Sel f
-
employment

16.6

34.0

5.3

7.8

2.6

33.8

8.6

Casual

and
other

empl oyment

24.1

6.4

6.1

12.1

6.1

45.3

18.5

Unemployed
,

search.

21.7

3.9

6.5

21.6

6.5

39.8

6.3

Unemployed,
di sc.

18.0

3.2

6.8

18.1

10.8

43.1

22.2

NEA

14.0

3.8

4.4

15.0

6.1

56.8

Mobility (%)

Overall: 51.4

Upward: 12.6

Downward: 15.1

Within non
-
empl
.: 17.1

Within
empl
.: 6.6

Mobility (%)

Overall: 44.8

Upward: 12.6

Downward: 15.1

Within non
-
empl
.: 17.1

27/06/2013

9

Model set
-
up


S
imple (survey
-
weighted) binary
probit

model:


Pr(y=1|
X
,
Z
) =
Φ
(
X

β

+
Z

δ
)


Two kinds of
probits
:

1)
y equals
1

if individual employed in 2008 and again in 2010/11;


0

if no longer employed in 2010/11

2)
y equals
1

if individual in regular wage employment in 2008 and again in
2010/11;


0

if no longer in regular wage employment in 2010/11


X

is vector of individual and household
-
level demographic and
location variables for 2008: age cohort, education, race,
household size, rural/urban, province dummies, etc.


Z

is vector of job
-
specific variables for 2008: occupation and
industry types, union membership, contract type/duration,
months in wage employment, take
-
home pay


Estimation separate for men and women


27/06/2013

10

(
1a)

(1b)

(2a)

(2b)

(3a)

(3b)

(4a)

(4b)

Male

Female

Male

Female

Male

Female

Male

Female

Omitted: age 20
-
25

Age 26
-
35

0.0751*

0.0494

0.0414

0.0356

0.0652*

0.0558

0.0723*

0.0502

Age 36
-
45

0.1298***

0.0975*

0.0833*

0.0678

0.1123**

0.1036**

0.1198***

0.0949*

Age 46
-
55

0.0777

0.0494

0.0221

0.0124

0.0703

0.0465

0.0630

0.0363

Omitted: no education

Primary education

0.0217

0.0481

0.0227

0.0521

0.0328

0.0403

-
0.0110

0.0110

Secondary education

0.1367***

0.1620***

0.1358***

0.1618***

0.1455***

0.1578***

0.0841**

0.0785

Tertiary education

0.1881***

0.3032***

0.1870***

0.3075***

0.1880***

0.2943***

0.1153**

0.1990***

Omitted: Black/African

Coloured

0.1071**

-
0.0461

0.1182***

-
0.0443

0.1057***

-
0.0425

0.1039**

-
0.0732

Asian/Indian

0.1467***

0.0984

0.1613***

0.1054

0.1673***

0.0963

0.1151*

-
0.0122

White

0.1149**

0.0548

0.1141**

0.0622

0.1282***

0.0584

0.0668

-
0.0363

Married

0.0639*

-
0.0273

0.0437

0.0065

0.0632*

-
0.0210

0.0488

-
0.0488

Household size

-
0.0170***

-
0.0123**

-
0.0102**

-
0.0081

-
0.0105**

-
0.0156**

-
0.0092**

-
0.0061

Rural

0.0246

-
0.1367***

0.0239

-
0.1326***

0.0225

-
0.1415***

0.0429

-
0.1151***

Household head

0.1024***

0.0806**

Omitted: No other workers in household

1 other worker

-
0.0016

-
0.0356

2 or more other workers

-
0.1562***

0.0670

Household per capita
income

(log)

0.0572***

0.0767***

Observations

1576

1933

1572

1918

1576

1933

1576

1933

Probit

estimates for employment
transitions
2008
-
2010/11 (baseline and extra
household variables): average marginal effects

27/06/2013

11

(
1a)

(1b)

(2a)

(2b)

(3a)

(3b)

(4a)

(4b)

Male

Female

Male

Female

Male

Female

Male

Female

Omitted: age 20
-
25

Age 26
-
35

0.0550

0.0467

0.0258

0.0608

0.0627

0.0643

0.0488

0.0510

Age 36
-
45

0.1335*

0.0827*

0.0985

0.0989*

0.1423**

0.1054**

0.1245*

0.0816*

Age 46
-
55

0.0855

0.0414

0.0439

0.0418

0.0935

0.0567

0.0718

0.0267

Omitted: no education

Primary education

-
0.0976**

0.0050

-
0.0940**

0.0147

-
0.0980**

-
0.0036

-
0.1035**

-
0.0433

Secondary education

0.0084

0.1621***

0.0093

0.1588***

0.0095

0.1544***

-
0.0156

0.0544

Tertiary education

0.0228

0.2621***

0.0272

0.2634***

0.0241

0.2549***

-
0.0199

0.1246**

Omitted: Black/African

Coloured

0.0352

-
0.0389

0.0467

-
0.0423

0.0386

-
0.0321

0.0401

-
0.0694

Asian/Indian

-
0.0311

0.0450

-
0.0202

0.0399

-
0.0408

0.0445

-
0.0615

-
0.1140

White

-
0.0367

0.0489

-
0.0397

0.0392

-
0.0400

0.0436

-
0.0741

-
0.0647

Married

0.0989**

0.0510

0.0807**

0.0522

0.1012**

0.0407

0.0903**

0.0142

Household size

-
0.0154***

-
0.0106

-
0.0093

-
0.0082

-
0.0176***

-
0.0155**

-
0.0085

-
0.0018

Rural

-
0.0471

-
0.1486***

-
0.0485

-
0.1483***

-
0.0487

-
0.1483***

-
0.0275

-
0.1194***

Household head

0.0865*

0.0247

Omitted: No other
regular wage workers
in household

1 other
regular wage worker

-
0.0067

0.0260

2 or more other
regular wage workers

0.0649

0.1159***

Household per capita
income

(log)

0.0415*

0.1057***

Observations

1122

1199

1118

1189

1122

1199

1122

1199

Probit

estimates for regular wage employment
transitions
2008
-
2010/11 (baseline
and extra household variables): average marginal effects

27/06/2013

12

(
1a)

(1b)

(2a)

(2b)

(3a)

(3b)

(4a)

(4b)

(5a)

(5b)

(6a)

(6b)

(7a)

(7b)

Male

Female

Male

Female

Male

Female

Male

Female

Male

Female

Male

Female

Male

Female

Baseline regressors (not shown)

…….

…….

…….

…….

…….

…….

…….

…….

…….

…….

…….

…….

…….

…….

…….

…….

…….

…….

…….

…….

…….

…….

…….

…….

…….

…….

Omitted: elementary occupation

Semi
-
skilled

-
0.0311

0.1014**

Manag
./professional

-
0.0495

0.1081**

Omitted:
agriculture,
hunting,
forestry,

fishing

Mining and quarrying

-
0.0899

0.1725***

Manufacturing

-
0.0285

-
0.0869

Utilities

0.1200***

Construction

-
0.2723***

-
0.0392

Wholesale and retail trade

-
0.1678**

-
0.0181

Transport, storage and communication

-
0.0814

-
0.1041

Fin.
intermed
., real estate
and
bus. services

-
0.0854

-
0.0146

Community, social and personal services

-
0.0491

-
0.0225

Union member

0.0548

0.0981***

Written contract

0.0710*

0.0341

Omitted: limited contract duration

Unspecified contract duration

0.0499

0.0157

Permanent contract

0.1609**

0.1010

Months

in
wage

employment

(log)

0.0381***

0.0556***

Monthly

take
-
home

pay

(log)

0.0812***

0.1011***

Observations

1096

1183

995

891

1092

1179

1110

1192

1117

1190

954

1023

1122

1199

Probit

estimates for regular wage employment
transitions
2008
-
2010/11 (extra job variables):
average marginal effects

27/06/2013

13

Some further probing


Some of the employment transitions may reflect ‘free choices’
rather than influence of external factors (such as economic
climate)


NIDS wave 1 and 2 include questions on
subjective well
-
being

from which we can construct following variables:


Change in self
-
reported life satisfaction (
-
/0/+)


Change in self
-
reported economic status of household (
-
/0/+)


Difference between self
-
reported economic status of household in
2010/11 and economic status anticipated in 2008 (
-
/0/+)


Do these measures differ between those that remain
employed between 2008 and 2010/11 and those that leave
employment over the same period?

27/06/2013

14

Changes in subjective well
-
being, by gender and transition outcome 2010/11:
proportions (%)

Male

Female

Male

Female

Not
empl
.

Empl
.

F
-
stat.

Not
empl
.

Empl
.

F
-
stat.

Not
wage
empl
.

Wage
empl
.

F
-
stat.

Not
wage
empl
.

Wage
empl
.

F
-
stat.

Change in life satisfaction

-

0.5939

0.5335

0.94

0.5300

0.4638

2.86

*

0.5441

0.5273

0.28

0.5909

0.4819

2.63

*

0

0.1141

0.1244

0.1201

0.1846

0.1531

0.1330

0.1440

0.1715

+

0.2920

0.3421

0.3498

0.3516

0.3027

0.3398

0.2651

0.3467

Change in

economic
status

-

0.3638

0.2830

2.60

*

0.3625

0.3298

1.71

0.3942

0.2775

4.73

***

0.3527

0.3229

0.26

0

0.3389

0.3340

0.2753

0.3406

0.3613

0.3330

0.3185

0.3469

+

0.2974

0.3830

0.3622

0.3296

0.2446

0.3895

0.3288

0.3301

Difference between
actual and anticipated
economic status


-

0.7132

0.5899

6.07

***

0.6468

0.6390

0.11

0.7801

0.5645

11.17

***

0.6932

0.6461

0.70

0

0.1487

0.2676

0.2102

0.2050

0.1112

0.2749

0.1770

0.2180

+

0.1380

0.1425

0.1430

0.1560

0.1087

0.1606

0.1298

0.1359

27/06/2013

15

Conclusions


Main findings:


There was
considerable mobility

(movements in and out of jobs) in SA
labour markets over 2008
-
2010/11 (cf. other periods, see e.g.
Banerjee

et al.
2008;
Ranchod

&
Dinkelman

2008)


Transitions may be, to some extent, explained by ‘individual choice’,
but there seem to be
certain types of workers
with a significantly
lower probability of retaining (broadly defined) employment:


Young (20
-
35) and older (46
-
55) workers


Workers with less than secondary education


… and a significantly lower probability of retaining regular wage
employment:


Female wage workers with less than secondary education


Female wage workers in elementary occupations


Male wage workers in construction and wholesale/retail trade


Male wage workers with a non
-
permanent contract


(Wage workers with a shorter job history or a lower take
-
home pay)


27/06/2013

16

Conclusions (2)


Further analysis indicates that changes in self
-
perceived life satisfaction
and economic status differ significantly between those that remain
employed and those that do not


Avenues for future research:


On the NIDS data:


More detailed occupation/sector information (not publicly available)


Incorporating NIDS wave 3 (available soon), to check whether labour market
transitions are different between wave 2 and 3


NIDS data on hours worked and wage earnings is patchy


On the QLFS data:


Using algorithm similar to that of
Ranchod

&
Dinkelman

(2008) to match
individuals from wave
t

to wave
t+1
for QLFS data 2008Q1
-
2012Q4 (rotating
panel of dwellings); cf. Verick 2012


Any inference from these matched panels needs to take into account that false
matches cannot be ruled out and probability of matching individuals is non
-
random



27/06/2013

17

Thank you for your attention

Mail
:
dennis.essers@ua.ac.be

Matching algorithm


for QLFS (cf. R&D 2008)

1)
Pool all cross
-
sections/‘waves’ and match households using identifiers

2)
Drop households present in only one wave

3)
Within each wave, drop individuals that belong to the same household
and have the same race, gender and age (or age difference of 1 year)

4)
Match remaining individuals across wave
t

and wave
t+1
on

household
identifier, gender, race and
age
t

= age
t+1

5)
Match also individuals across wave
t

and wave
t+1
on

household
identifier, gender, race and
age
t

+1 = age
t+1

6)
Take matched individuals of steps 4 and 5 together to form ‘expanded
match panels’

7)
Apply extra consistency checks to ‘expanded match panels’ to form
‘strict match panels’, dropping:


Individuals whose level of education is non
-
missing and differs between waves


Individuals whose status changes from ‘married’/‘divorced’/‘widowed’ to ‘never
married’

27/06/2013

19