Estimating the Impact of Sectoral Minimum Wages in

South Africa

Haroon Bhorat and Benjamin StanwixDevelopment Policy Research Unit

School of Economics, University of Cape Town

Presentation to Portfolio Committee on Labour Wednesday 20th August 2014

Parliament of South Africa

Outline

• Employment & Wage shifts: What is the evidence of changes across sectors?

• Sectoral Minimum Wages: A Tale of Two Policy Shocks

• Enforcement: Weak enforcement with partial compliance

Where Have All the Jobs Gone?

Source: SARB & StatsSA (LFS 2001-2007 and QLFS 2008-2012), Author’s Calculations

What Should Have Happened…But Didn’t

Source: SARB , Quarterly Bulletin, Various issues and Authors’ Calculations

Wanted:Labour Intensive Growth…

Relatives & Absolutes DO Matter…

Growth (2001-2012) Employment SharesShare of Change

(ΔEj/ΔE)

Absolute Relative (%ΔEj/%ΔE)

2001 2012 (2001-2012)

Primary -719,232* -2.6 0.15 0.07 -0.28Agriculture -514,468* -2.7 0.10 0.04 -0.20Mining -204,764* -2.2 0.05 0.02 -0.08Secondary 537,376* 1.0 0.21 0.21 0.21Manufacturing 112,149 0.3 0.14 0.12 0.04Utilities 10,774 0.5 0.008 0.008 0.004Construction 414,453* 2.5 0.05 0.07 0.16Tertiary 2,720,821* 1.6 0.63 0.71 1.08Trade 513,572* 0.9 0.21 0.21 0.20Transport 288,364* 2.1 0.04 0.06 0.11Financial 782,108* 2.8 0.09 0.13 0.31Comm Serv 1,041,524* 2.1 0.17 0.22 0.42Priv Hholds 95,253 0.4 0.09 0.08 0.04Total 2,497,763* 1.0 1 1 1

The Winners and Losersfrom Employment…

Proportions Change in PropChange in

No2001 2012 2001-2012

Primary

High-Skilled 0.03 0.08 0.05 27,602Med-Skilled 0.54 0.37 -0.17 -571,229*Unskilled 0.43 0.56 0.13 -175,392*Total 1 1 -719,232*

Secondary

High-Skilled 0.14 0.18 0.04 188,518*Med-Skilled 0.70 0.62 -0.08 136,140Unskilled 0.16 0.20 0.04 214,002*Total 1 1 537,376*

Tertiary

High-Skilled 0.27 0.29 0.02 931,498*Med-Skilled 042 0.43 0.008 1,214,349*Unskilled 0.31 0.28 -0.03 576,288*Total 1 1 2,720,821*Source: StatsSA (LFS 2001 and QLFS 2012), Author’s Calculations

Wither: Employment Since Democracy

• Employment driven by 2001-2008 growth• Primary sector employment collapse

– Agriculture (Impact of Wm) and Mining together losing over 700 000 jobs• Both employers of least-skilled workers

• Lacklustre employment growth in Manufacturing• Growth within tertiary sectors such as financial

services and community services – Public sector as a growing source of employment– Financial Services & Temporary Empl. Service Providers

• Employment gains in high- and medium-skilled occupations

Real Average Wages By Sector

Source: Wittenberg (2014), ‘Analysis of employment, real wage, and productivity trends in South Africa since 1994’, ILO.

Real Average Wages By Sector. cont’d

Source: Wittenberg (2014), ‘Analysis of employment, real wage, and productivity trends in South Africa since 1994’, ILO.

Mainly Good, and One Bad Story:Sectoral Minimum Wages

• Five sectors non-agricultural sectors (2.2 million individuals in 2007, 17.2 % of employment).

• Area types A, B, C, etc., with A areas generally urban, B semi-urban, and C rural.

• Mapping of workers to minimum wages using sector, occupation, and area codes. Sector Occupation/Areas Year and month Schedules Within

Domestic Workers

A and B August 2002 4

Private Security A, B, C,D, E November 2001 5

Taxi Taxi Drivers and Fare Collectors

July 2005 2

Retail Managers, Clerks, Sales and Shop Assistants, Drivers, Cashiers, Forklift operators, Security Guards, Areas A, B,C

February 2003 24

ForestryAgriculture A and B

March 2006March 2003

12

Data

• The dataset used is a pooled dataset consisting of 15 waves of the South African Labour Force Survey (LFS), conducted between September 2000 and 2007.

• The LFS is a biannual national household survey conducted by Statistics South Africa. This is the last available dataset containing useable information on income in South Africa.

• Data pooled and treated as repeated cross sections across time.

Method Is Important:Identification of Control Groups

• Overall Control Group sample restricted to non-unionsed, non-Wm employed with less than Grade 12 schooling.

• For each Wm, sector and occupation codes used to identify similar occupations.

• Domestic workers: African and Coloured females; unskilled.

• Forestry: African and Coloured; unskilled.

• Retail: Individuals in semi-skilled occupations.

• Security and Taxi: African and Coloured males; medium-skilled.

• Agriculture: Unskilled individuals or in elementary occupations.

So is Specification1

Yijkt = β0 + β1 POSTt + β2Treatmentk + β3POSTt *Treatmentk + j + ijkt (1)

• where Yijkt is the outcome variable of interest (employment, wages, hours of work) for individual i in group k in period t.

• POSTt refers to the time dummy, capturing the period before (0) and after (1) the minimum wage is introduced.

• Treatmentk captures whether the individual is in the treatment (1) or control group (2).

• POSTt *Treatmentk is the difference-in-difference term, which captures the impact of the minimum wage as a consequence of being in the treatment group, during the treatment period.

• j are district council effects.

Just To Be Sure: Specification 11

A difference-in-differences model as well, but tests to see whether wages increased more in areas where workers wages were lower in the pre-law period (Gapjk ).

Yijkt =0+ 1Postt + 2 Gapjk + 3Postt *Gapjk + ijkt +

ijkt (II)

Where

•Gapjk is a constructed variable measuring the proportional increase in the pre-law wage wjk

(t-1) necessary to meet the initial introduced minimum wmjk

(t) (real terms).

Gapjk=[wmjk(t) -wjk

(t-1)]/wjk (t-1) where Gapjk≥0

• ijkt are controls for various worker characteristics.•Control group comparison, rules out changes in outcome variable due to non-minimum wage shifts over sample period.

Sector RetailDomestics Forestry Taxi Security

Post 0.004** 0.007*** -0.0001 -0.003*** 0.003**

Wage gap -0.0001 -0.014*** 0.005** -0.00002 -0.002*

Wage gap*Post -0.0001 -0.001 -0.00004 0.0001 0.0001

Constant 0.023*** 0.019** 0.020*** 0.003*** -0.003

Observations 493,809 493,809 267,108 455,266 464,046Individual Controls Yes Yes Yes Yes Yes

The Good News In Non-Agric Sectors: Employment

Notes: Robust standard errors used adjusting for clustering at the district council level (not shown). ***p<0.01, ** p<0.05, * p<0.1.

Good News…

Retail Domestics Forestry Taxi Security

I II I II I II I II I II

Post 0.069** -0.004 0.083* 0.086***

0.134**

* 0.132 0.032 -0.089 0.073** -0.076

Sector 0.101**

-0.360*** -0.132 -0.615*** -0.238***

Sector*Post -0.007 0.064 0.050 -0.037 -0.062

Wage gap -0.286***

-0.558*** -0.021 -0.027*** -0.362***

Wage gap*Post 0.067*** 0.048* 0.029 0.020*** 0.222***

Constant 1.801*** 0.671*** 1.122*** 0.856***

1.253**

* 0.889*** 1.856*** 1.127*** 1.832*** 1.129***

Individual

ControlsNo Yes No Yes No Yes No Yes No Yes

District

effects Yes No Yes No Yes No Yes No Yes No

Notes: Robust standard errors clustered by district council (not shown here). ***p<0.01, ** p<0.05,

*p<0.1

But the Market Does Respond: Hours of Work

Sector Retail Domestics Forestry Taxi Security

I II I II I II I II I II

Post -0.318**

-

1.463***

-0.986 -1.166-

1.372***

-1.148 0.481 0.734-

1.388***

-

2.051***

Sector 2.344*** -1.684* 1.330** 20.24*** 10.95***

Sector*Post -1.218*** -0.399 0.026 -0.802 -1.444*

Wage gap 2.720*** 3.214** -0.344** 0.342*** 3.542*

Wage

gap*Post -0.135 -0.095 -1.226 -0.775*** -2.591

Constant 47.79*** 52.60*** 49.08*** 44.26*** 48.17*** 50.29*** 47.94*** 59.73*** 49.29*** 61.25***

Individual

ControlsNo Yes No Yes No Yes No Yes No Yes

District

effects Yes No Yes No Yes No Yes No Yes No

Notes: Robust standard errors clustered by district council (not shown here). ***p<0.01, ** p<0.05,

*p<0.1

The Bad Story:The Minimum Wage in Agriculture

Average Characteristics of the Treatment Group (2000-2007)

2000 2001 2002 2003 2004 2005 2006 2007N 3 363 3 021 3 084 2 047 2 498 2 490 2 484 2 361Weighted 805 715 804 162 819 048 623 750 538 438 555 514 576 319 555 549Area A 0.20 0.24 0.24 0.33 0.31 0.27 0.30 0.32Age 35 36 36 36 36 36 35 36Education 5.45 5.35 5.38 5.40 5.34 5.78 6.05 6.21Male 0.57 0.66 0.63 0.72 0.70 0.71 0.69 0.69African 0.79 0.75 0.78 0.69 0.73 0.75 0.74 0.75Full-Time 0.69 0.75 0.77 0.94 0.93 0.90 0.87 0.91Hours per Week 39 43 42 49 49 48 46 47Full-time equivalent hours 1 194 774 1 101 992 1 134 201 744 331 742 379 615 397 619 609 598 223Nominal Monthly Wage 498 586 513 738 770 917 1 068 1 221Nominal Hourly Wage 4.03 4.52 4.65 5.14 5.23 5.69 6.16 6.27Fraction < Min. (Area B) 0.80 0.76 0.75 0.58 0.61 0.58 0.61 0.60Written Contract 0.31 0.29 0.30 0.47 0.51 0.49 0.49 0.52

Notes: Robust standard errors in parentheses. All regressions are weighted. *** p<0.01, ** p<0.05, * p<0.1. The dependent variable is whether the individual is employed as a farmworker (1) or not (0). The sample includes individuals of working age who are unemployed or searching for work who have no more than 12 years of education. POST = 1 after March 2003 and 0 otherwise. The Wage Gap is the district level difference between the log of median farmworker wages and the log of median wages for the control group.

VARIABLES (1) (2)

     

POST -0.1204*** -0.0877***  (0.0265) (0.0228)Wage Gap 0.1013*** 0.1523***  (0.0194) (0.0107)Wage Gap*POST -0.0353 -0.0563***  (0.0261) (0.0128)Controls NO YES     Constant 0.4011*** 0.9606***  (0.0221) (0.0272)     Observations 160 817 160 817R-Squared 0.049 0.126

The Bad News In Agriculture:Employment

Good News for the FewFarmworkers Left: Hourly Wages

VARIABLES (1) (2) (3)       POST 0.284*** 0.340*** 0.388***  (0.00840) (0.0624) (0.0530)Farmworker -0.548***      (0.0118)    Farmworker*POST 0.176***      (0.0157)    Wage Gap   -0.154* -0.1394*    (0.0811) (0.0708)Wage Gap*POST   0.221** 0.1751**    (0.101) (0.0907)Controls for Age, African, Education

  NO YES

Constant 1.338*** 0.871*** 0.687***  (0.00666) (0.0495) (0.0560)       Observations 90,986 33,892 33,575R-squared 0.063 0.068 0.228

Notes: Robust standard errors in parentheses. *** p<0.01, ** p<0.05, * p<0.1. All regressions are weighted. Regression 1 is run on the sample of farmworkers and the control group. Regressions 2 and 3 include only farmworkers. Regressions have the 'Log of Hourly Wages' as dependent variables. POST = 1 after March 2003 and 0 otherwise. The Wage Gap is the district level difference between the log of median farmworker wages and the log of median wages for the control group.

One Minor Positive Result:Contract Coverage

VARIABLES (1) (2) (3)

       POST 0.124*** 0.140*** 0.169***  (0.00475) (0.0145) (0.0144)Farmworker -0.170***      (0.00613)    Farmworker*POST 0.0561***      (0.00801)    Controls for Education, Age, African

  NO YES

Wage Gap   -0.178*** -0.132***    (0.0189) (0.0188)Wage Gap*POST   0.0876*** 0.0331Constant 0.496*** 0.421*** 0.443***  (0.00382) (0.0108) (0.0128)       Observations 69,743 31,218 31,017R-squared 0.040 0.038 0.064

Notes: Robust standard errors in parentheses. *** p<0.01, ** p<0.05, * p<0.1. All regressions are weighted. Regression 1 is run on the sample of farmworkers and the control group. Regressions 2 and 3 include only farmworkers. The dependent variable is whether the individual has a written employment contract (1) or not (0). POST = 1 after March 2003 and 0 otherwise. The Wage Gap is the district level difference between the log of median farmworker wages

and the log of median wages for the control group.

Notes: Robust standard errors in parentheses. *** p<0.01, ** p<0.05, * p<0.1. All regressions are weighted. Regression 1 is run on the sample of farmworkers and the control group. Regressions 2 and 3 include only farmworkers. The dependent variable is hours worked per week. POST = 1 after March 2003 and 0 otherwise. The Wage Gap is the district level difference between the log of median farmworker wages and the log of median wages for the control group.

Hours Worked: Move away from Non-Permanent workers

VARIABLES (1) (2) (3)       POST -0.350 -3.557*** -3.508***  (0.241) (1.063) (1.107)Farmworker -5.747***      (0.496)    Farmworker*POST 5.549***      (0.355)    Wage Gap   -12.36*** -14.213***    (0.5679) (0.626)Wage Gap*POST   7.225*** 9.058***    (1.424) (1.504)Controls YES NO YES       Constant 52.415*** 54.26*** 59.26***  (0.618) (0.564) (0.826)       Observations 78,451 39,126 38,803R-Squared 0.024 0.126 0.150

The Good and the Bad Stories:Revisited…

Non-Agriculture Minimum Wages1. Some evidence of significant increase in real hourly wages in the post-law period as a result of the introduction of a minimum wage in four out of the five Non-Agriculture sectors examined (notably the Retail, Domestic worker, Taxi and Security sectors).2. No significant employment effects of the new law in any of the sectors assessed.3. Some indication that for sectors where employment continued to rise in the post-law period, notably the Retail and Security sectors, the introduction of minimum wages may have been associated with a reduction in the usual number of weekly hours worked.

Agriculture Minimum Wage• Minimum wage law in Agriculture in South Africa has had significant labour market effects:

1. Farmworker Wages rose by approximately 17% as a result of the law.2. Increased contract coverage , as number of workers with a written employment

contract increased to reach 57% in 2007. 3. Adjustments at the intensive margin were observed as part-time workers lost

jobs.4. Employment fell significantly in response to the law.

Enforcement of the Minimum Wage

• A key consideration for minimum wage legislation.• ECC oversees Sectoral Determinations but not

Enforcement.• DoL responsible for enforcement through the

Inspection and Enforcement Service (IES) which is a provincial competence.

• 45% of covered workers paid wages below the legislated minimum in 2007 – Recent administrative data still suggest significant non-

compliance

• Probability of inspection low.• IES under-resourced.

Fines for Violation of Sectoral Determinations

Source: Schedule 2 of the BCEA, (1997)

No previous violation R100 per employeeNo previous violation in respect of the same provision of the Act

R200 per employee

A previous violation the same year or two violations in respect to the same provision during the past 3 years

R300 per employee

3 previous violations of the same provision within 3 years

R400 per employee

4 previous violations of the same provision within 3 years

R500 per employee

Maximum Permissible Fine involving underpayment

No previous violation

25 % of the underpayment, including any interest owing on the amount at the time of the order

A previous violation of the same provision during the past 3 years

50 % of the amount due including applicable interest

A previous violation of the same provision within a year, or 2 previous violations, or 2 previous violations of the same provision

75 % of the amount due, including applicable interest

3 previous violations of the same provision during the past 3 years

100 % of the amount due including applicable interest

3 previous violations of the same provision during the past 3 years

200 % of the amount due including applicable interest

Partial Compliance in Agriculture: 2001-2005

0.5

1

-2 -1 0 1 2lncpi_hourlywage

2001 2002 2003 2004 2005

Notes: The vertical red line represents the (‘urban’) Minimum Wage in 2003.

Partial Compliance in Agriculture, By Province: 2000-07

Notes: The horizontal red line represents the (‘urban’) Minimum Wage in 2003. The vertical black line represents the timing of the law.

Provinces coded as: 1-WC, 2-EC, 3-NC, 4-FS, 5-KZN, 6-NW, 7-GTG, 8-MPM, and 9-LMP.

Towards A National Minimum Wage

Some issues to consider:•Choosing the level of the Wage is Crucial•Deciding on single or split wage (e.g. Youth; Agriculture and Non-Agriculture)•Enforcement Efforts•A Phase-In Period?•ECC as governing structure or a new Minimum Wage Commission?

Thank you