The National Income Dynamics Study’s Coronavirus Rapid Mobile Survey - The Labour Market In The First And Second Quarters Of 2020

Hard on the heels of the release of the second quarter 2020 (“Q2”) Quarterly Labour Force Survey (“QLFS”) has come the release of the second wave data from the National Income Dynamics Study’s (“NIDS”) Coronavirus Rapid Mobile Survey (“CRAM”).

Introduction

Summary reports from the first and second wave have been published[1], as well as analyses by social scientists associated with the project[2]. There is information on a number of topics in the CRAM data set. This brief will be concerned only with labour market data. Particular attention will be paid to the relation between CRAM data and QLFS data, and how the CRAM data supplements the QLFS data.

What is the National Income Dynamics Study?

NIDS is the first national household panel study in South Africa. It seeks to follow the same households across time in order to track changes in them, a goal not achievable by successive independently sampled surveys. The study started in 2008 with a nationally representative sample of over 28 000 individuals in 7 300 households across the country. The core survey was repeated with these same household members every two years to three years, with the latest interview round (or “fifth wave”) being conducted in 2017. The five waves of NIDS were implemented by the Southern Africa Labour and Development Research Unit, based at the University of Cape Town’s School of Economics.

What is the Coronavirus Rapid Mobile Survey?

NIDS-CRAM is a special follow up of a subsample of adults (age 15 years and over) from households in Wave 5 of NIDS. NIDS-CRAM uses a much shorter questionnaire than the NIDS core survey, with a focus on the coronavirus pandemic and the national lockdown. Survey data were collected by Computer Assisted Telephone Interviewing (“CATI”). Six waves of NIDS-CRAM are planned. The results from the first two waves have been published. The first two waves contained labour market information for February and April 2020, and the second added data from June.

The NIDS-CRAM subsample

The subsample size of the first wave of NIDS-CRAM was 17 568 individuals, of whom 7 074 responded, implying a response rate of 40.3%. 48% of the sample was classified as a non-contact, i.e. the fieldwork team was not able to speak to the individual, either on the telephone number provided or on a number of alternative contact numbers (including the phone numbers of co-resident household members in 2017). A further 8% of the selected respondents were classified as a refusal. They were contacted but refused to be interviewed.

In order to compare results from NIDS-CRAM, the analysis is confined to those age 15 to 64. The sample size for this age range is 6 378. By contrast QLFS is a survey of households that collects information from approximately 30 000 dwelling units, so the QLFS sample is much larger.

NIDS-CRAM is a panel. The weighted sample is not designed to match the South African population in 2020. Theoretically, one should think of the weighted NIDS-CRAM survey data as reflecting the outcomes for a broadly representative sample of those 15 years and older in 2017 who were followed up three years later. However, for reasons outlined in Annexure 1, the weighted sample can be regarded as an approximation to a sample of the population age 15 to 64, excluding people who have immigrated to South Africa between the time of the NIDS 5 wave and the CRAM sample.

There was a 20% attrition of the subsample between the first and second waves leading to a second wave sample size of 5 105. Sample weights were adjusted to compensate for this attrition.

Analysis

The analysis here regards the February, April and June data as three cross-sectional data sets, which means that the panel features of the NIDS-CRAM waves are not considered[3].

Annexure 2 sets out results from the NIDS-CRAM surveys in February, June and April. Where corresponding results from the QLFS are available, they are shown as well.

Six themes emerge from Annexure 2:

  1. The NIDS-CRAM data indicate greater employment than the QLFS. A reason for this is that the reference period for employment is longer in the NIDS-CRAM study than in the QLFS, raising the probability that a respondent worked for at least an hour during it. Partly as a result, NIDS-CRAM may be finding more people with a marginal attachment to work than the QLFS. The considerably higher number of informal sector workers found in NIDS-CRAM lends some support to this hypothesis.
  2. The NIDS-CRAM data indicate that, in April, 23% of the employed were not working, but had a paid activity to return to. This proportion dropped to 11% in June. Much of the change will have resulted from return to work following reduction in lockdown restrictions between the two months.
  3. The NIDS_CRAM data indicate that, in both April and June, about 90% of the people who had a paid activity to return to were not working for reasons associated with the COVID epidemic. Of the people who were not working for COVID-related reasons, just under a half continued to be paid in April and there was only a small decline in the proportion between April and June. Some of both the employed who are working and those who were not experienced reduction in earnings and profits. The QLFS indicated that about one fifth of all the paid had earnings cuts.
  4. There is divergence in the trend of official unemployment between NIDS_CRAM and the QLFS between Q1 and Q2, but closer correspondence between expanded unemployment between the two surveys. The QLFS shows a marked dip in official unemployment between Q1 and Q2 and a marked rise in non-searching unemployment. This could be explained by the hypothesis that people who would ordinarily be seeking employment ceased to do so, either because they were prevented by lockdown restrictions or because they saw no point in searching during a period of substantial lockdown. But the NIDS-CRAM data show no corresponding dip and rise. On the other hand, both surveys show that more than ten million people who were not working in both Q1 and Q2 were willing and able to start work within seven days.
  5. Average hours worked per week in both quarters were lower and the decline between Q1 and Q2 was greater in the NIDS-CRAM data than in the QLFS.
  6. Relative to an index of 100.0 in February/Q1, the NIDS-CRAM survey found a Q2 index of 83.6 in Q2, while the index calculated from the QLFS was 86.4. Since the index of value added at basic prices was found to be 86.8 in Q2, compared with 100.0 in Q1[4], employment was roughly proportional to value added. The NIDS-CRAM data indicate that the index dropped to 79.8 in April, the month of the harshest restrictions.

Conclusion

If the proportionality between employment and output is maintained, the prospects for increased employment in the third quarter are good. Value added in sectors covered by monthly production statistics increased again between June and July, and these gains are likely to have been consolidated, if not improved, in August and September. The risk to both production and employment in the next few months will be a resurgence in the epidemic but, unless it is very severe, we are not likely to see the extent of the shutdown of the economy experienced in April.

This brief is still not the last word on the evolution of the labour market. Quarterly employment statistics are yet to come.

Charles Simkins
Head of Research
Helen Suzman Foundation

Annexure 1

Reconciliation of NIDS-CRAM Survey with mid-year 2020 population estimates and the second quarter QLFS

Size of the population age 15-64

One has to make two adjustments to reconcile the NIDS-CRAM data with the Mid-Year Population Estimates, 2020[5]:

· Immigration. Statistics South Africa projects that net immigration into South Africa will be 826 261 between 2016 and 2021. Assuming a constant rate, this implies 496 000 net immigrants between NIDS 5 in 2017 and NIDS-CRAM in 2020. What one needs for the adjustment is gross immigration, for which one needs an assumption about how many immigrants re-emigrate. 50% might be a reasonable guess, so that gross immigration would be 992 000 over the three year period.

· The NIDS-CRAM adult subsample consists of people of age 15 and over at the time of the NIDS 4 survey in 2017. This means that the NIDS-CRAM sample should not have people under 18 in it. (In fact, it has a few 17 year olds.) So an adjustment need to be made to the NIDS-CRAM data to make the population age range compatible with the QLFS age range. This is done using the number of 12 to 14 year olds in the NIDS 5 child sample. Very few of people in the 15-17 age range are in the labour market. The simplifying assumption in the analysis is that none are.

Table 1

  Thousands
Mid year population estimate, 2020 38 942
Less Immigrants 2017-2020 992
Population estimate net of immigrants 37 950
NIDS-CRAM population – Wave 1 32 403
Add 15-17 population 2 915
Less 17 year olds in NIDS-CRAM 4
Adjusted NIDS-CRAM population 35 314


The adjusted NIDS-CRAM population is 7% below the mid-year population estimate. However, no adjustment to the NIDS_CRAM survey weights are made. This means that absolute estimates of employment and other magnitudes made on the basis of the NIDS-CRAM Survey are slight under-estimates.

Composition of the population by age, population group and gender

Table 2 sets out the composition of the population by age, population group and gender for the mid-year population estimate, the QLFS estimate and the adjusted NIDS-CRAM estimate.

Table 2

Age

Mid year

QLFS Q2

CRAM

15-19

12.3%

13.1%

12.7%

20-24

12.4%

13.2%

12.8%

25-29

13.9%

12.8%

13.0%

30-34

14.5%

13.3%

13.1%

35-39

12.3%

12.1%

12.4%

40-44

9.6%

9.4%

10.3%

45-49

8.1%

8.6%

8.5%

50-54

6.6%

7.0%

6.3%

55-59

5.7%

5.7%

6.1%

60-64

4.6%

4.7%

4.8%

       

Total

100.0%

100.0%

100.0%

Black/African

80.6%

80.9%

80.4%

Coloured

9.0%

9.0%

9.5%

Indian/Asian

2.8%

2.6%

2.2%

White

7.7%

7.5%

7.8%

Total

100.0%

100.0%

100.0%

Gender

     

Women

50.9%

50.5%

52.2%

Men

49.1%

49.5%

47.8%

Total

100.0%

100.0%

100.0%


The distributions are similar. The most noticeable is the slight bias towards women in the NIDS_CRAM sample.

Annexure 2

 

NIDS-CRAM

QLFS

 

February/Q1

April

June

Q2

Q1

Q2

Employed

           

Job

14,638,234

9,226,374

11,231,505

10,228,940

   

Business, no job

968,318

549,464

1,069,314

809,389

   

No business, no job, active

1,702,302

901,625

1,130,791

1,016,208

   

No job, no business, not active, paid activity to return to

 

3,142,901

1,679,364

2,411,133

   

Employed

17,308,854

13,820,364

15,110,974

14,465,669

16,383,000

14,148,000

Population age 15-64

35,314,030

35,314,030

35,314,030

35,314,030

38,874,000

39,021,000

Absorption rate

49.0%

39.1%

42.8%

41.0%

42.1%

36.3%

Employment index (February/Q1=100)

100.0

79.8

87.3

83.6

100.0

86.4

Value added index (February/Q1=100)

100.0

   

86.8

 

86.8

Missing employment status

200,746

1,298,852

39,645

     

Missing employment status as percent of total

0.6%

3.7%

0.2%

     

Unemployed

           

Official

 

4,332,927

5,357,315

4,845,121

7,070,000

4,295,000

Did not search - discouraged

       

2,918,000

2,471,000

Did not search - other

       

809,000

3,493,000

Did not search - total

 

6,582,034

5,812,256

6,197,145

3,727,000

5,964,000

Expanded

 

10,914,961

11,169,571

11,042,266

10,797,000

10,259,000

Official rate

 

23.9%

26.2%

25.1%

30.1%

23.3%

Expanded rate

 

44.1%

42.5%

43.3%

39.7%

42.0%

Labour force participation rate (official)

 

57.8%

59.2%

58.5%

51.7%

51.5%

Labour force participation rate (expanded)

 

70.0%

74.4%

72.2%

69.9%

62.5%

Analysis of employment
Hours worked per week

           

Average hours per week - job

38.6

35.5

38.8

37.2

   

Average hours per week -business

28.2

15.3

26.7

21.0

   

Average hours per week, no job, no business, active

25.5

21.8

34.0

27.9

   

Average hours all

36.7

25.7

33.3

29.5

42.9

41.9

Annexure 2 - continued

 

NIDS-CRAM

QLFS

 

February/Q1

April

June

Q2

Q1

Q2

No job, no business, not active, paid activity to return to

           

Known reasons

 

3,134,818

1,661,375

2,398,097

   

COVID reasons

 

2,815,664

1,563,719

2,189,692

   

Per cent reasons COVID reasons

 

8,083

17,989

13,036

   

Percentage COVID reasons

 

89.8%

94.1%

91.3%

   

Payments - daily paid, COVID reasons

 

46,928

33,130

40,029

   

Payments - weekly paid, COVID reasons

 

284,182

45,697

164,940

   

Payments - fortnightly paid, COVID reasons

 

84,411

58,151

71,281

   

Payments - monthly paid, COVID reasons

 

883,488

541,663

712,576

   

Total payments to people absent because of COVID

 

1,299,009

678,641

988,825

   

Payment as a per cent of all people absent because of COVID

 

46.1%

43.4%

45.2%

   

Payments to all the employed

 

11,976,472

14,110,251

13,043,362

   

Payments to all the employed as a per cent of all employed

 

86.7%

93.4%

90.2%

 

81.3%

Formal and informal employment

           

Formal (non-agricultural)

   

8,021,418

   

10,064,000

Informal (non-agricultural)

   

4,183,723

   

2,280,000

Agriculture

   

716,594

   

799,000

Private services

   

61,946

   

1,005,000

No job, no business, not active, paid activity to return to

   

1,679,364

     

Missing data

   

447,929

     

Total

   

15,110,974

   

14,148,000


Notes

1. Recall the QLFS definitions:

Employed persons are those aged 15–64 years who, during the reference week, did any work for at least one hour, or had a job or business but were not at work (i.e. were temporarily absent).

Unemployed persons according to the official definition are those (aged 15–64 years) who:

· Were not employed in the reference week; and

· Actively looked for work or tried to start a business in the four weeks preceding the survey interview; and

· Were available for work, i.e. would have been able to start work or a business in the reference week;

· Had not actively looked for work in the past four weeks, but had a job or business to start at a definite date in the future and were available.

Unemployed persons according to the expanded definition are those (aged 15–64 years) who:

· Were unemployed on the official definition, plus those who were available to work but are discouraged work-seekers or have other reasons for not searching.

2. The reference period is a month rather than a week in the NIDS-CRAM survey. The longer reference period increases the estimate of employment and decreases the estimates of unemployed persons on both definitions.

3. The NIDS-CRAM data set includes people who worked for at least an hour in the preceding month, but who were not recorded as having a job or running a business. These people are described as “No job, no business, active”. It also separately identifies people who did no work in April and June, but had a paid activity to return to.

4. The absorption rate is the number of employed people divided by the population.

5. The value added index is taken from Charles Simkins and Charles Collocott, July production statistics: an indication of a V-shaped recovery? Helen Suzman Foundation Brief, 28 September 2020

6. The labour force consists of the employed plus unemployed. The labour force participation ratio is the labour force divided by the population.

7. Days worked per week were truncated to a maximum of six and hours worked per day to a maximum of ten, to avoid outliers. In the NIDS-CRAM data the number of hours was given for each person working. In the QLFS data, hours worked per week were published in range categories and the average number of hours worked in each category was assumed to be the mid-point of the category range.

8. In the NIDS_CRAM data, COVID reasons for not working, but having a job to return to were lockdown, self-isolating, business temporarily closed, illegal to go to work, enforced leave, no transport and (in the case of June only) retrenched.

9. The definition of informal employment in QLFS is when there is one or more of the following circumstances present: (a) no written contract, (b) no benefits such as pension or medical aid contributions from their employer, or (c) work in the informal sector. The informal sector has the following two components: (a) employers, own-account workers and persons helping unpaid in their household business who are not registered for either income tax or value-added tax,

and (b) employees working in establishments that employ fewer than five employees, who do not deduct income tax from their salaries/wages. In the NIDS_CRAM data, only information on (a) in either case is captured, and that information is used for making the distinction between formal and informal workers.


[1] See Nic Spaull et al, Overview and findings: NIDS-CRAM Synthesis Report Wave 1, 15 July 2020 and Nic Spaull et al, Synthesis Report: NIDS-CRAM Wave 2, 30 September 2020

[2] The synthesis reports and the analytical reports for both waves can be found at cramsurvey.org/reports

[3] For analysis which follows individuals over time, see Vimal Ranchhod and Reza Daniels, Labour market dynamics in South Africa in the time of COVID-19: Evidence from wave 1 of the NIDS-CRAM survey, SALDRU Working Paper No. 265, July 2020, and Vimal Ranchhod and Reza Daniels, Labour market dynamics in South Africa in the time of COVID-19: Evidence from Waves 1 and 2 of the NIDS-CRAM survey, NIDS_CRAM report 13, September 2020. Both can be downloaded from cramsurvey.org/reports

[4] See Charles Simkins and Charles Collocott, July production statistics: an indication of a V-shaped recovery? Helen Suzman Foundation Brief, 28 September 2020.

[5] Statistics South Africa, Mid-year Population Estimates, 2020, Statistical Release P0302, 9 July 2020