Again a report hit the front pages of both Business Day and Business Report (May 7 2012). "South Africa cannot afford South Africans" was prepared on behalf of the trade union UASA by Mike Schüssler from economist.co.za.

The report recommends draconic cuts in the lowest wages and says that the country needs more inequality. The proposals are immoral and would further weaken the low economic demand for goods and services inherited by colonialism and apartheid. The proposed measures would add fuel to the unemployment crisis. But before such discussions there are things like data sources and the evaluation of data to be considered.

[Important note: Please note the claim made in the following four paragraphs (and table) - that the UASA report used a "defunct" data series to calculate changes in Nominal and Real Unit Labour costs - has been retracted by Mr Forslund. His full retraction and apology can be read here. The Editor (June 22 2012)]

The data over labour costs development that is used by the SA Reserve Bank (SARB) are published in the national accounts on page S-150. UASA and Schüssler have used a defunct index series. There exist two accounts for productivity and unit labour cost in the national accounts. One of the accounts is crazy. This should now be common knowledge among journalists and economists who frequent the media.

The details of this affair were described by AIDC last year in a September press statement after contact with the SARB and Statistics SA. In relation to this, a public debate on labour productivity is available on PoliticsWeb or SA Labour News.

Table 1: Left table are the changes in unit labour costs (ULC) according to "Labour in the non-agricultural sectors" used by the SARB in its policy discussions. Right table is derived from a defunct index series in the table called "Labour: Labour costs in the non-agricultural sector".

If data says that labour costs per unit produced suddenly increased by 24.7 percent in 2002, and again by 18.7 percent in 2003, the researcher, but not the propagandist, asks: "What on earth happened in those years?" Exceptional data oblige the researcher to take a look at lived history, on earth. In this case, nothing extreme happened in 2002 and 2003. Nominal wage increases were below 12 percent in 2002 and below 9 percent in 2003. Wage settlements averaged below 8 percent (Andrew Levy). No extra pay-roll tax was introduced, and so on.

Based on the crazy data series in the table, the UASA report then says that unit labour costs adjusted for inflation increased by 10 percent during the 2000's, but in the rich OECD countries they fell. We should realize that this is wrong. It doesn't fit with the decline in wages' share of the national income (GDP) since 1998.

The rectified data series used in all public discussions by the SA Reserve Bank show that real unit labour costs (RULC) fell by 10 percent from 2000 to 2011, just as they did in the OECD. On the average, real wages have been increasing at about one percentage point lower than the 3 percent average yearly growth in labour productivity (Diagram 1).

Such a process leads to a falling wage share of the national income. All this broadly fits in with phenomena like growing indebtedness among households or record hoarding of R520 billion on bank accounts by SA big business, reported by the banks in January. The UASA report's account for labour costs development is built on ignorance supported by bad judgment.

Diagram 1: Left hand scale: Percent change in real wages (quarter by quarter, year on year) and average year on year increases in labour productivity during the period. Right hand scale: Real unit labour costs (RULC), year on year in third quarters, 3^rd quarter 2000 = index 100 (Source: The SA Reserve Bank table "Labour in the non-agricultural sectors". SARB's account for nominal ULC adjusted for inflation with the GDP-deflator by the author).

As for the actual situation, the report concludes: "The Working Class can now be called the comfortable class". Prof. Chris Malikane and Tim Cohen from Business Day (May 21) have both intervened in the discussion. Because Schüssler's report is inexact on this point, they both believe that the UASA report quote wage data from Stats SA. This is not so. 

Schüssler informed me by email that he has mainly used pay roll data collected by 21^st Century Pay Solutions, which I now have seen. State owned enterprises and private companies contribute with the pay rolls for about 700 000 of all formal employees in 2010. Wages for temporary employees and part time employees are not included. The companies providing data buy the data and use it for comparison. Those who are in the sample have actively chosen to be there, which isn't randomly selected. What this could lead to is not discussed in the UASA report.

"The average worker who gets about R13 200 per month at present", states the report. This seems to refer to the Stats SA QES report for November 2011, as does Tim Cohen. All comparisons in the report are however for 2010. "R13 200" is more than thousand rand above the average reported in QES for 2010. QES don't report median wages. This is the wage of the person in the middle or the typical wage. In the UASA sample the median wage is about R11000 per month. This coincides quite accurately with wage statistics for blue collar workers at Eskom. It would be good for employment in SA if workers had such wages or higher. But the material used by UASA is not representative of workers in formal employment.

The study "Monthly earnings of South Africans, 2010" from Statistics SA concentrates on the typical wage, the median. The salaries of 686 000 employers, usually paying high or very high to themselves (median is R7000), are excluded from the report. The majority of 1.2 million "Own-account workers" has very low wages (median is R1820). They are also excluded. After those crucial exclusions 11 058 000 workers are represented by the samples. It is estimated that 8 571 000 of them are formally employed.

On request, Stats SA gave me data over the average wage from last quarter 2009 to first quarter of 2011. In the study, the average monthly wage among the close to 8.6 million formally employed workers is R6621 during 2010. The median wage is R3683: Half of the formal employees in SA earn from R3700 per month and less. The median wage of informal and formal workers weighted together has not moved much since the National Planning Commission reported in the Diagnostic Report that is was R2500 per month for 2008. After two years of inflation it is R2800. The buying power of the typical SA worker is stagnating or declining.

Tim Cohen tries to undermine the Stats SA reports by writing "trust me when I say that"..."10% of the South African's earned more than that", referring to R13000 per month in 2010. Well, that is what it means to say that the tenth decile starts at R13000. The inequality problem is about where this decile ends. Secondly, his reference to the R140000 taxable income median rather proves the opposite of what Schüssler argue.

The median among about 4.6 million persons obliged to declare their income for 2010 are not representative of 9.1 million formally employed workers. Less than 4 million of them had to pay personal income tax. The rest fell below the first tax bracket. This means that at least about 5 million formally employed workers earned R4750 per month or less in 2010. Is this hard to understand?

Finally, Cohen tries to show that the Stats SA survey is incompatible with the wage sum in the national accounts (R1202 billion in 2010). His calculator is broken. An average of R6 621 times 12 month times 8 571 000 formally employed gives R681 billion. If adding the pension contributions, the much higher average salaries of 686 000 employers and 1.2 million "own account workers", the reported average of R6 621 per month in 2010 for workers makes perfect sense.

With data biased towards high income earners as their foundation, Schüssler and UASA argue that South Africa has a more equal labour market than USA, UK, Australia or NZ. What a sensation! UASA says that the top 10 percent of formal employees have salaries that are only 5.8 times more than the wages of the 10 percent lowest paid.

The generally high salaries of 686 000 employers, as well as the generally low earnings of own-account workers, have been excluded in the solid survey from Stats SA. Even after doing this, we can deduct from it that the top 10 percent of formal employees earned 11.6 times more than the bottom 10 percent in 2010. South African inequality lands outside the A4-paper in Schüssler's comparative Power Point presentation. The inequality-scale for all his other countries only runs to ten.

Dick Forslund

Dick Forslund is an economist and researcher at the Alternative Information and Development Centre in Cape Town.

Retraction and apology

Note by Dick Forslund, June 22 2012: The part of this article saying that the two completely different accounts for real unit labour costs made by the SA Reserve Bank and the Uasa report depend on the Uasa report using the structural breaks series published by Stats SA unrevised as it is (displayed in the table's right column) has been retracted together with an apology. See full text here.

Click here to sign up to receive our free daily headline email newsletter