In a biting critique on contemporary India, Ranjani Iyer Mohanty draws from the epic fable of a blind King Dhritarashtra to argue that Indian authorities are ignoring problems and pretending they don’t exist.
“In 2010, when the Commonwealth Games were held in New Delhi, rather than using them as an opportunity to improve conditions, they temporarily moved the homeless out of the city and put up advertisement hoardings around the slums to hide them… instead of fighting poverty, the planning commission lowered the poverty line and announced a reduction in the number of poor.” (Financial Times, April 3)
The criticism may be exaggerated and unfair.
But it does strike a chord in relation to the current debate in SA on employment statistics. The spectre of denialism has elicited howls of outrage and incredulity about the Adcorp Employment Index (AEI), which places the rate of unemployment at about 10 percent compared to the official 24 percent.
This debate has raged for some years, but until recently mostly below the public radar. Faced with overly technical and contradictory data, bemused media have resorted to recycling briefings from Statistics SA and Adcorp as they are released. Society is left none the wiser.
This, some may argue, is in the nature of democracy. There should be no standard truth but a variety of views – and people can decide what they wish to believe.
But the levels of employment and unemployment are foundational indicators of the nation’s well-being. They are a major consideration in crafting economic and social policies. Society cannot form an accurate view of its own humanity, let alone plan for the future, if it does not have sufficient consensus on the extent to which its citizens are included in or marginalised from economic activity.
It is for this reason that, on a global scale, there are concerted efforts to standardise methodologies and agree on the rigour, integrity and ethics of official data collection and analysis. Employment statistics are among the most cardinal in this regard.
In an attempt to shed light on the public debate on this issue, the Mapungubwe Institute for Strategic Reflection (Mistra) organised a round table of experts on labour statistics at the beginning of the month. Among the participants were Statistician-General Pali Lehohla and Kefiloe Masiteng, Stats SA’s deputy director-general for population and social statistics; Adcorp’s Loane Sharp; Professor Haroon Bhorat (UCT); Dr Andrew Kerr and Professor Martin Wittenberg of Data First, also from UCT; Dr Miriam Altman of the Human Sciences Research Council; Michael Mwasikakata of the International Labour Organisation mission in SA; Dr Azar Jammine of Econometrix; and Abrahams Mutedi of the Labour Department’s division on labour market information and statistics.
Shorn of statistical jargon, what are the areas of convergence and divergence, and how can these issues be taken forward?
There is agreement that Stats SA has made tremendous progress in developing a new base of information, worlds apart in inclusivity and accuracy from the fragmented data of the apartheid era. It has improved the Quarterly Employment Survey (QES, measuring formal employment), and introduced the new Quarterly Labour Force Survey (QLFS).
Using methodologies applied on a global scale, the QLFS is a household survey, based on a sample of the population, from which generalised conclusions are drawn. It is universally accepted that survey data provide the best possible estimates, rather than precise figures on employment and unemployment.
To measure the quality of the data, response rates and degrees of precision are used. On this score, Lehohla asserts that the Stats SA data surpass the minimum globally-accepted quality measures. In turn, Sharp agrees that, barring small deviations, Stats SA and Adcorp are in agreement on the level of formal employment.
Where, then, are the areas of divergence?
Adcorp argues, correctly, that because surveys offer estimates, there is a possibility of undercounting as respondents may not declare everything about their economic activities. It is thus necessary to do a “sense-check”. This entails, among other things, weighing the data from these surveys against other information on economic activity such as tax returns, the amount of money in circulation, studies on the informal sector and estimations of unrecorded economic activity, including that of illegal immigrants.
There is no question that in respect of VAT receipts, number of tax submissions and size of the small business sector, the figures indicate a slightly higher number of people involved in economic activity than suggested by the employment surveys. But, as Wittenberg and others argue, the challenge arises when attempts are made to extrapolate from this information into the estimation of employment numbers.
For instance, tax submissions may reflect intersecting sets of owners and activities. While money in circulation may suggest higher levels of economic activity than provided for in other data, the currency demand method does have its own pitfalls. How do we calculate the velocity of money circulation (for example, the number of times the same R10 note is used by different people over a week), especially in the informal sector, and what assumptions are required to do this?
The existence of unrecorded economic activity cannot be questioned – and this includes illicit and laundered resources of the criminal underworld. The currency demand method may point to this phenomenon. But is it of such a magnitude that the number of employed people would be millions more than official data suggest?
The most substantive area of divergence is from estimations of unrecorded, informal sector employment. These are unregistered economic actors, either self-employed or working for others. Using the survey done by Finscope on behalf of the Business Trust, the Department of Trade and Industry and others, Adcorp points to a much larger figure for small businesses than is estimated in official statistics: at 5.6 million, about double the number in official figures.
The difference in these tallies, Adcorp argues, may account for a large part of the unrecorded economy. After making assumptions about labour intensity in this sector, and factoring in activities by illegal immigrants, Adcorp comes up with a higher number of employed and self-employed people.
To assert this level of authority, the facts need to be checked and rechecked in a transparent peer review processes. Has the Finscope survey undergone such review? What about the methodology in calculating informal and unrecorded sector employment? Is it a given that, if surveyed for the QLFS, people in this sector would conceal their status? What about assumptions about labour intensity in this sector? It is around these issues that there were persistent requests from the other participants at the Mistra round table, for Adcorp explicitly to outline its methodology and assumptions.
Posing these questions about Adcorp’s assertions does not and should not suggest that there are no improvements required in labour data. Altman, for instance, makes the observation that overruns in VAT receipts have been under discussion for some time and sector investigations are under way. Further, Stats SA’s business register was updated some five years ago and it may be necessary to do this again.
Similarly, Sharp has raised a few critical questions that should help further clarify the rigour of the official employment statistics. These relate to such issues as definitions; “atypical employment” – in terms of temporary and part-time workers, some of whom are employed through labour brokers – which Sharp argues “now accounts for more than 30 percent of total employment”; the decline in Stats SA’s informal sector employment figures from 3.6 million in 2000 to 2.1 million last year, which he attributes to “official undercount of illegal immigrants”; and technical issues on sampling variabilities.
But, as others retort, how does Adcorp calculate its variabilities; how does it estimate the number and activity of illegal immigrants; would measuring “atypical employment” separately change how people respond to surveys?
What these exchanges point to, primarily, is the need for maximum transparency in outlining how data is collected and processed – not only on the part of the official statisticians but also by those who are sceptical about the official data. As Masiteng underlined, it is in Stats SA’s profound self-interest that users of its data interrogate the rigour of its work, the better to improve the official statistics and society’s understanding of itself.
It has therefore been agreed that statisticians from Stats SA and Adcorp and other experts will interact to clarify the issues that have been raised by either side.
Masiteng reports that, since the Mistra round table, Stats SA “has had a meeting with Loane… where the team answered his questions. Ultimately we were able to appreciate the underlying cause of his misunderstanding of our data.” Further interactions will be held, and work on “sense-checks” such as VAT overruns and the size of the small business sector will continue.
In this process, the statisticians should strive to meet the injunction of Thomas Gradgrind in Charles Dickens’s Hard Times: “Now what I want is, Facts… Facts alone are wanted in life. Plant nothing else, and root out everything else.”
But as Bhorat has warned, a narrow focus on these issues can distract us from debating critical employment trends about how SA can move on to a higher growth path and absorb the mass of people who do not have jobs. These trends include the fact that, even as we are striving to emerge from the downturn, our growth model is still consumption-driven. Our manufacturing sector is weak and uncompetitive; the primary sector is not creating jobs. Yet labour-broker employment is booming.
We cannot afford to behave like Mohanty’s King Dhritarashtra, blinded to the corrosive trends that have trapped the country in a low-growth, low-employment trap. But we must also grapple to get the detailed facts right.
n Netshitenzhe is the executive director of the Mapungubwe Institute