Content uploaded by Daniel Mügge
Author content
All content in this area was uploaded by Daniel Mügge on May 17, 2019
Content may be subject to copyright.
The Lure of Ill-Fitting Unemployment Statistics: How South Africa’s
Discouraged Work Seekers Disappeared From the Unemployment
Rate
Juliette Alenda-Demoutiez and Daniel Mügge
Department of Political Sciences, University of Amsterdam, Amsterdam, The Netherlands
ABSTRACT
Unemployment refuses unambiguous definition. Its statistical
representation is always open to contestation, especially where labour
markets differ from the Western-industrial norm. Why do countries
adopt international standards even if they may fit local conditions
poorly? South Africa is an exemplary case to answer this question. When
Apartheid ended in the early 1990s, South African statisticians embraced
the new emancipatory spirit. Their broad unemployment indicator
defied international conventions but did justice to the marginalised
Black population, and to Black women in particular. Since then, however,
South Africa has fallen in line with the much narrower definition of the
International Labour Organization (ILO), in spite of widespread criticism.
Why? We find that ILO standards were not forced upon South Africa.
Instead, South African statisticians themselves embraced international
standards to repel charges of arbitrary or politically motivated numbers.
Counterintuitively, international standards become alluring precisely
when doubts about statistics’fit with local conditions are the greatest.
ARTICLE HISTORY
Received 10 December 2018
Accepted 26 April 2019
KEYWORDS
International standards;
political economy; Sociology
of quantification; South
Africa; unemployment
Introduction
For people who want to work, the inability to find employment can be a source of enormous
hardship –economically, socially and personally. For societies at large, widespread unemployment
is a fundamental political challenge. When labour market conditions are particularly dire, unemploy-
ment may be the central economic problem to be tackled. Effective policy and informed public
debate, in turn, hinge on an accurate understanding of the size and shape of the issue. Statistics
about unemployment are central to the fight against it.
Unemployment is not a natural category (Salais et al.1986). Nineteenth century labourers pushed
for censuses of the nascent working class to reveal its plight (Desrosières 1998). At their inception,
unemployment statistics were a weapon for class struggle. Since then, just how we should concep-
tualise joblessness has been contested: who is included in the figures and who is not, how much do
you have to work to fall in one or the other category, and so on (Baxandall 2004, Zimmermann 2006).
These measurement choices are highly consequential: they highlight or obscure changes in labour
markets and people’s working lives, and the statistics based on them guide policymakers and the
public’s understanding (Gautié 2002, Hoskyns and Rai 2007). Unemployment statistics, in short, are
deeply political: their definitions create winners and losers, and they lead us to ask who writes
them in the first place (Desrosières 1998).
© 2019 Informa UK Limited, trading as Taylor & Francis Group
CONTACT Juliette Alenda-Demoutiez juliettealenda@hotmail.fr
NEW POLITICAL ECONOMY
https://doi.org/10.1080/13563467.2019.1613355
We know little about such indicator politics outside the OECD world. That is surprising. Labour
market conditions in many developing countries are strained, and in the absence of strong
welfare states, joblessness threatens people existentially.
1
Yet it is not obvious that the essentially
Western concept of unemployment (Garraty 1979, Topalov 1994) is well-suited to developing
country labour markets. Subsistence agriculture is widespread outside large cities. Labour markets
are highly segmented, often along racial or ethnic lines. And large parts of the population may
find themselves in informal, precarious employment, below the statistical radar. These features com-
plicate labour market statistics, and they enlarge the scope for political fights about them. At the
same time, poorer countries often have asymmetrical and fraught relationships with international
organisations, whose push for harmonised statistical standards may meet little enthusiasm on the
ground. It is not clear why developing countries would stick to international statistical standards at
odds with their socio-economic realities.
This article sets out to map and explain such unemployment indicator politics in South Africa. Job-
lessness has been an enormously political challenge in the country (Kingdon and Knight 2001).
Unemployment there is one of the highest in the world: different measurement approaches put it
somewhere between 26 and 38 percent in the first quarter of 2018.
2
The central difference
between these estimates lies in the treatment of ‘discouraged work seekers’, people who do not
count as unemployed following the ILO definition but would still like to work.
Over the past two decades, South Africa has increasingly embraced the narrow ILO definition of
unemployment and privileged it in its unemployment statistics –even if it remains politically dis-
puted and arguably ill-suited to the country. Our central question is what has pushed South
African unemployment statistics in this surprising direction. We can break it down into a descrip-
tive-empirical as well as a more theoretical question: what have been the main political fights over
unemployment statistics in South Africa since the mid-1990s? And which factors explain the
choices that have been made?
Counterintuitively, South Africa has embraced international standards not in spite of their limited
fit with domestic conditions, but precisely because of these difficulties. Defining and measuring
unemployment in South Africa is such a fraught endeavour that every practical solution has immedi-
ately invited plausible criticism. Politicians quickly cast doubt on official figures: political incumbents –
in particular the African National Congress –have criticised figures as too high; those in political
opposition or advocating for the labour movement have embraced the opposite position. These criti-
cisms have gnawed at the credibility of Statistics South Africa (Stats SA) –the official office for
national statistics –even when the problem has not been the agency’s competence and determi-
nation, but rather the fundamental mismatch between the structure of the South African society
and economy, and unemployment as a concept.
The ultimate choice for a narrow official definition has been driven by the wish to conform to inter-
national standards. The latter have not been forced on South Africa. Instead, they have been
embraced bottom-up to buttress South Africa’s credibility with the international community as
well as that of national statisticians vis-à-vis the political class, and to allow at least a superficial (if
ultimately misleading) comparability between South Africa’s labour market conditions and those
in the rest of the world.
This article builds on 25 interviews, conducted in 2018, with present and past South African stat-
isticians, labour and business representatives, researchers, politicians and consultants, as well as on a
range of primary documents about statistical debates and development there. It is structured as
follows: the next section outlines the main insights social scientists have gathered about the political
nature of unemployment indicators in Europe and North America and asks how well we should
expect them to travel to developing countries. In the body of this article we then detail the two
main phases in which South African unemployment statistics have been embattled and demonstrate
how the quest for legitimacy has driven Stats SA to embrace an unemployment definition that
matches local circumstances poorly. Our conclusions outline the implications of the South African
2J. ALENDA-DEMOUTIEZ AND D. MÜGGE
experience for our thinking about the politics of economic statistics more generally and in develop-
ing countries in particular.
The Politics of Unemployment Indicators
Ubiquitous as internationally harmonised economic indicators are these days, they had been devised
for domestic purposes: to evidence the plight of the nineteenth century working class in case of
unemployment (Salais et al.1986) and cost of living indicators (Stapleford 2009); to facilitate macro-
economic management in the case of gross domestic product (GDP), whose predecessors were
developed in the 1930s and 1940s (Lepenies 2013). Economic statistics, in other words, were
tailor-made for Western industrialised nations, and they had clear political purposes.
Their global spread only gathered pace after the Second World War (Ward 2004). International
organisations such as the International Labour Organisation (ILO), the United Nations, and later
the International Monetary Fund, the World Bank and the Organisation for Economic Co-operatio-
nand Development proselytised for their use and the developmentalist ideas underlying them
(Masood 2016). Such governance by numbers has extended further with Millennium Development
Goals and Sustainable Development Goals, which champion indicators as tools for economic and
social development (Taylor 2016). Indeed, global benchmarking has emerged as a prominent all-
purpose mode of transnational governance (Desrosières 1998, Davis et al.2012, Jany-Catrice 2012,
Broome and Quirk 2015, Cooley and Snyder 2015, Kelley 2017), with indices covering everything
from gender equality and business climates to financial opacity and good governance.
Globally proliferating quantification can suffer from shortcomings that parallel those we find in
domestic governance by numbers (Porter 1995): statistics’air of objectivity and accuracy can hide
shoddy methodologies (Broome et al.2018), poor data (Jerven 2013, Linsi and Mügge 2019), out-
dated policy priorities (Fioramonti 2013), and biases in the data (Mügge 2019). These below-the-
radar politics of economic statistics motivate us to investigate political fights over South African
unemployment measures and the roles that domestic actors and international organisations play
in them.
In public and political discourse, the official unemployment rate often functions as the de facto
thermometer for national labour market conditions. Across history and societies, quite different
phenomena have been collected under the heading ‘unemployment’(Zimmermann 2006). On the
one hand, unemployment as a label, and the statistics later built on it, consciously recognised that
joblessness was not necessarily a sign of individual failure but could be caused by structural econ-
omic factors beyond personal reach. This framing of joblessness attenuated the moral opprobrium
attached to it and shifted part of the responsibility for the plight of the unemployed onto govern-
ments (Gautié 2002).
Unemployment statistics as they solidified after the Second World War in Europe and the USA
increasingly coalesced around a clear prototype:
An able-bodied, prime-age male industrial breadwinner with plant specific skills who [had] been laid offfrom full-
time formal work as the result of a plant closing in a declining industry. (Baxandall 2004, p. 212)
On that basis, employment statisticians typically distinguish three categories of people: the
employed, the unemployed, and the economically inactive. People in both the second and third
groups have no jobs. However, the unemployed want a job, are available for one and are actively
searching; the economically inactive normally fail the last criterion (Green 2000). The ‘active
search’-criterion functions as a litmus test to gauge whether joblessness is voluntary or not.
This approach has shortcomings, as statisticians themselves concede (for example Sengenberger
2011 from the ILO). It excludes people who want a job but do not look for one because they are dis-
couraged or deem their qualifications insufficient (Kingdon and Knight 2001). In many Western
countries, controversies about unemployment statistics have focused on such involuntary economic
inactivity (Baxandall 2004). Indeed, in 1982 the International Conference of Labour Statisticians
NEW POLITICAL ECONOMY 3
included two distinct definitions in its guidelines: a narrow one, excluding the jobless who wanted
work but did not search actively, and a broad one, which did include this group. In practice, most
countries around the world have settled on the narrow definition as the official headline figure.
This choice matters because it shapes the political priority attached to reducing unemployment of
one or the other kind. Governments are likely to develop very different policies via-a-vis the
people who fall between the two definitions, not least the welfare measures benefitting them.
A second contentious aspect of the ILO definition gains particular significance in developing
countries: the unemployed have to be available for work. In Western contexts, that means being
ready to begin work more or less right away –impossible only under special circumstances, for
example due to impending childbirth or medical constraints. By implication, the unemployed are
those people for whom the main barrier to a job is that the right one simply cannot be found.
In poor countries, the reasons not to be available may be very different. Rural subsistence farmers,
for example, might long for formal employment but be unable simply to take up a paid job. Their
whole life situation, as that of their dependents, may forbid that, and the employment sought
may not be anywhere nearby. Even if someone is unable to walk away from her living situation –
not ‘available’and hence not unemployed –there may still be a serious labour market problem
(Kingdon and Knight 2001, Posel et al.2014).
Because of their monetary bias (Mügge 2019), economic statistics systematically sideline unre-
munerated (reproductive) labour, much of which is done by women within households (Hoskyns
and Rai 2007). Such gender bias also feeds into unemployment statistics, and the gap between
narrow and wide definitions: because a disproportionate share of household and care responsibilities
lands on their shoulders, women are often not available for paid employment, even if –under a more
equal division of tasks between men and women –they might be happy to seek employment, not
least to buttress their economic independence. A narrow unemployment rate can thus hide the
specific socio-economic difficulties women face.
High South African unemployment rates have repeatedly been doubted with the argument that, if
joblessness really were that sky-high, we should witness large-scale riots. Yet in fact, socio-economic
deprivation may be institutionalised:
The elaborate mechanisms of proletarianization, repression, and discrimination not only impoverished indigen-
ous people physically, but probably did even more psychological damage. As soon as family and other social
structures were disrupted, the disciplinary and civilising effects of those traditional structures were undermined.
In this way a subculture –or syndrome –of poverty was institutionalised among poorer Africans and coloureds.
(Terreblanche 2002, p. 40)
Hence, if there is a specific form of discouragement that is a legacy of apartheid and racism, the offi-
cial unemployment definition may obscure that in particularly pernicious ways.
The question is how, among all the imperfect and potentially ill-fitting options, that single
leading labour market indicator is constructed –which in turn dominates policy, public debate
and external perceptions (cf. Khan et al.2015). The literature offers several hunches, which we
take as inspirations as we investigate the indicator politics in South Africa. Most obviously, poli-
ticians may tinker with definitions to embellish their economic achievements or, if in the opposi-
tion, detract from those of the ruling government (Moon and Richardson 1985). Second,
international organisations such as the ILO and the United Nations have promulgated standards
such as the System of National Accounts since the end of the Second World War (Ward 2004,
Clegg 2010). It is unclear, however, how these internationalisation effects have played out in devel-
oping countries, where the gulf between international standards and the local situation may be
particularly wide. Third, measurement approaches can become path-dependent once specific
policy commitments are attached to them (Baxandall 2004,p.216),suchasworkers’rights to
financial assistance. The interests that congeal around particular definitions may serve to entrench
them. To what degree do these hunches help us make sense of the politics surrounding unemploy-
ment indicators in South Africa?
4J. ALENDA-DEMOUTIEZ AND D. MÜGGE
Unemployment Measures and the end of Apartheid
South African statistics have always been tightly linked to the country’s idiosyncratic politics and
followed a strict racial logic. Its Current Population Survey (CPS) measured the Whites, the Colour-
eds and the Indians (as the categories went) until 1990 (Standing, Sender and Weeks 1996). The
Black Africans, in contrast, were excluded. For the rest, the Central Statistical Service (CSS)
mainly concentrated on Whites (Lehohla 2002). Nevertheless, already in the 1970s and 1980s,
South African debates about unemployment statistics asked how much of it was in fact voluntary
(Standing et al.1996). Some argued that many rural-dwellers chose to be unemployed, content
with subsistence agriculture (Kantor 1980,Gerson1981); others pushed back (Simkins 1982,
Kingdon and Knight 2001).
The apartheid legacy is fundamental to understand the specificities of South African labour
markets and why narrow definitions may be particularly ill-fitting. Rural unemployment in the
country is higher than urban unemployment because apartheid had severely restricted blacks’
mobility. Black homelands were rural areas with poor land and little formal employment.
3
For
people living there, finding paid work often meant waiting for formal-sector job opportunities
to arise far away, outside the homelands. The geographical and racial mismatch between
where people live and where employment is to be found mars South African labour markets to
this day.
Apartheid legacies continue to shape present-day labour markets in other ways, as well: through
highly unequal access to education (du Toit and Neves 2014) as well as through a large (and again
racialized) informal economy (Rogerson 1992, Chandra et al., 2002). Given the history of unequal
access to paid employment, unpaid and highly gendered household labour continues to play a
central role especially for poor South African families (du Toit and Neves 2014, p. 834, Cousins
et al.2018).
In addition, Apartheid-age repression of black labour had meant that it was available cheaply, not
least for the labour-intensive resources extraction and agricultural sector. International boycotts and
foreign companies’reluctance to invest in Apartheid South Africa had held productivity growth back.
Post-apartheid economic modernisation then meant that catch-up productivity growth hurt demand
for labour. 500.000 jobs were lost in the first five years of democratic government, while an additional
Figure 1. Labour Market Evolution since 1994 (%). Source: authors, based on Stats SA databases and statistics publications.
NEW POLITICAL ECONOMY 5
450.000 young people entered the formal labour market. By whichever measure, unemployment sky-
rocketed (Cling 1999), as Figure 1 shows.
In 1994, the democratic movement in South Africa released the Reconstruction and Development
Programme (RDP), which focused on redistribution following a Keynesian paradigm (Adelzadeh 1996,
Koelble 2004). Two years later, the government shifted to an orthodox economic reform programme,
encouraged not least by the major conglomerates of the country and the wish to attract foreign
direct investment (Carmody 2002, Hamilton 2014). This international orientation, as we will argue
below, eventually bolstered the case for adoption of international standards, including in economic
statistics.
4
The government opted for ‘regulated flexibility’of the post-1994 labour market: minimum wages,
combined with a recognition of a two-tier labour market of permanent, protected workers, as well
as temporarily employed and less protected ones. Overoptimistically, the government had banked
on a nine-fold increase in foreign direct investment (FDI) to meet employment targets. In the
event, South Africa registered a net FDI outflow of around $1.6 billion between 1994 and 1999 as
domestic companies internationalised. Old cleavages in South Africa’s labour market therefore
have persisted, as does stifling unemployment and rampant inequality (the Gini coefficient varied
between 0.66 and 0.70 between 1993–2012, see Isaacs, 2016).
The abolition of apartheid legislation in 1991 and the wider economic and political reversals in its
wake brought challenges for economic statistics, too. The October Household Survey (OHS) of 1993
was the first one that aspired to include the entire population. The mission was to transform the CSS,
formerly part of an oppressive apartheid state, into a democratic institution. In 1994, the government
set up a task force to craft what would eventually become the Statistics Act. With the assistance of the
Swedish, Australian and Canadian statistical agencies, it published a widely discussed policy paper in
1997; two years later, Stats SA was established and enshrined as the only institution producing official
statistics (Stats SA 1999).
With the apartheid approach discredited, statistical standards were up in the air. Using this
opening, the CSS defied international practice and adopted the expanded unemployment definition
in the first 1993 OHS (Bangane 1999).
5
It stuck to this definition for the subsequent years. An official
report on the 1995 OHS defined unemployment as follows:
The proportion of people in the economically active population who are not in paid employment or self-employ-
ment at a given point in time, but who are available for work or for other income-generation activities, and who
want to be employed or self-employed. (CSS 1996, p. 15)
This ‘want’element is the key characteristic of the expanded definition; people are not obliged to
actively ‘look for’a job to be included in the statistics.
This approach was a conscious policy decision. As Mark Orkin and Ros Hirschowitz
6
explained in
the same report:
It has been widely recognised that the strict definition is too limited in the present South African context, where
employment opportunities are extremely limited, and many unemployed people have ceased to seek work
actively […] This applies mainly to women, particularly those in rural areas, where employment or income-gen-
erating activities are scarce, and transport is expensive. The unemployment rate is consequently defined by the
CSS in terms of the expanded definition. (CSS 1996,p.14–15)
Even with apartheid officially abolished, the South African labour market remained highly fragmen-
ted and extremely unequal. In the spirit of wanting to help redress these imbalances by highlighting
them, the 1996 CSS report disaggregated the gap between expanded and strict rates by race and
gender (Figure 2).
Three features are particularly striking: the unemployment rates vary strongly in line with the
different racial categories, unemployment rates are substantially higher for women across all racial
groups, and –most important for our purposes –the gap between the strict and expanded
definitions is particularly high for Black South Africans and for women. In effect, economic inactivity
that might potentially count as unemployment is Black, and it is heavily female. That gives the
6J. ALENDA-DEMOUTIEZ AND D. MÜGGE
differing definitions a clearly gendered and racialized charge. In early post-Apartheid South Africa,
CSS officials consciously defied international conventions and used unemployment definitions that
would avoid such biases.
The Creeping Rise of an ‘Official’Definition
No single indicator –narrow or broad –can answer all questions about labour markets. A struggle
over indicator definitions therefore is a struggle over which perspective is highlighted and which
one is sidelined. When several definitions and time series exist side by side, it still matters which
one is designated as the ‘official’set of figures, relegating competing data to a secondary status.
During the second half of the 1990s, the narrow unemployment definition creepingly emerged as
the ‘official’one –setting the stage for the increasing marginalisation of the broad definition in
the two decades to follow.
In South Africa, this definitional struggle has played out between the politician who, as minister,
was in charge of economic statistics, and the head of statistics himself. Trevor Manuel has been min-
ister in South African governments from 1994 to 2014, covering a range of portfolios. During the tran-
sition from CSS to Stats SA, he was in charge of statistics as Minister of Trade and Industry. Realizing
how crucial statistics were for economic and political development of South Africa, Manuel fought
hard for their reform once apartheid had ended (Green 2008). The debate was deeply politicised,
as Ravi Naidoo, director of the National Labour and Economic Development Institute in the 1990s
and early 2000s and part of the Statistics Council remembers:
It was very contentious because the [trade] union was keen to say that unemployment was a much bigger
problem. And that therefore the government should be proactive in the economy. Whereas business, at that
time, was happy to say we don’t need much intervention […] The fight was really to make the government
more interventionist, because we had a very conservative economic team then in government. (Interview with
Ravi Naidoo, Johannesburg, 2018)
A‘wage-citizenship nexus’took a central role in the post-apartheid South Africa (Barchiesi 2011). A
strong focus on ‘jobs’normalised paid work at the centre of the liberation of South Africans but, by
implication, excluded non-wage workers. This narrative combined economic modernisation and for-
malisation with a catch-up to neoliberal, efficiency- and productivity-driven economic policies fash-
ionable elsewhere in the world at the time (cf. Ferguson 2015).
Figure 2. Unemployment Rates by Race and Gender (%). Source: October Household Survey, 1995.
NEW POLITICAL ECONOMY 7
After an initial phase in which statistical development was mainly inspired by the wish to shed
Apartheid-legacies, public statistics professionalised. Statistics SA gained increasing autonomy
through the Statistical Act of 1999. According to Peter Buwembo, chief director of the Quarterly
Labour Force Survey, this autonomy became crucial for Stats SA:
That is the good thing, they give us enough space, good space. They don’t tell us at all what to do, they have to
accept what we say. It is a story from a long time ago, when the minister [Manuel] said ‘don’t give me what I want
to hear but what I need, what I need to understand, because what I cannot measure, I cannot manage it.’Nobody
asked for figures before. They get to know them at the same time as everybody, in the media. We have a strong
Act; it helped us. (Interview with Peter Buwembo, Pretoria, January 2018)
This changed status also meant that methodological and inferential considerations –rather than
purely political ones –increasingly gained weight. Hence, the first challenge to the broad definition –
which asked whether people ‘wanted’to work rather than whether they were ‘looking for a job’–was
rooted in a statistical argument about robustness. A 1998 Stats SA report critically noted that
the expanded unemployment rate does, however, introduce more subjectivity into the measure of the unemploy-
ment rate, and instability in tracking trends, as it is more difficult to distinguish what constitutes ‘wanting’a job
than to say whether someone has engaged in definite actions to find one. (Stats SA 1998, p. 63)
The technical demands on statistics might come to trump the appropriateness of the underlying
definition.
It was clear to all involved that people’s labour market situations come in many shades of grey.
Nevertheless, translating those nuances into hard and fast categories presented difficulties of its
own. One option was a ‘very expanded’definition of unemployment, including even the jobless
who professed no desire for employment.
7
The other one was to classify the ‘not looking’simply
as economically inactive. Table 1 shows the difference this categorisation makes.
The gaps between the figures were vast. The eventual compromise between Orkin and Manuel
was to publish the narrow and the expanded numbers, with all attendant detail (age, sex, region,
and so on.), but to designate the narrow yardstick as the official one (Green 2008).
As a former sociologist, who’s worked for [the Congress of South African Trade Unions] and for Jay Naidoo [its
leader until 1993], we wouldn’t gain by ceasing to report the expanded definition, so now we report both. (Inter-
view with Mark Orkin, Johannesburg, 2018)
Table 2 reproduce how Statistics SA decided to present unemployment trends in its 1998 report.
8
The second driver behind the narrow measure came from outside South Africa in the form of ILO
standards. To be sure, the ILO did not impose its definitions in any way. Yet in the years after the
immediate post-Apartheid enthusiasm, South African politicians felt a need to build international
credibility, including by adherence to international technical norms and standards. The desire
seemed to justify privileging a strict measure, and an increasing use of international consultants
and rising regard for social development indicators further pushed South Africa in that direction.
The ILO definition granted countries some leeway in the treatment of the jobless who were not
looking for a work, depending on the labour market structure and social constraints for job-searchers
and non-searchers (Hussmanns et al.1990, p. 107–108). That said, a 1996 ILO report stressed that
including the non-searching unemployed might exaggerate unemployment (Standing et al.1996).
The ILO would tolerate the broad definition, but clearly not encourage it, and indeed three quarters
of countries around the world ignored the jobless who were not actively looking for work from their
unemployment statistics (Posel et al.2014, Stats SA 1998).
Table 1. October Household survey, 1994–1997: unemployment rates.
Rates of unemployment (%) 1994 1995 1996 1997
Very expanded unemployment rate 38.4 37.4 41.7 42.4
Expanded unemployment rate 30.9 29.1 35.6 37.8
Official unemployment rate 19.2 16.9 21.0 22.9
Source: Statistics South Africa, 1998.
8J. ALENDA-DEMOUTIEZ AND D. MÜGGE
On top, Orkin, and then Hirschowitz, argued that the broad definition would dent the country’s
investment ratings through an excessively gloomy picture of economic conditions and disadvantage
it in the international use made of comparative statistics (Green 2008; Interviews). Hence, South Africa
published its new official unemployment rate following ‘widely-accepted international practice’(Stats
SA 1998, p. 11).
9
The ILO supported the effort and assisted South Africa in tackling the many practical
problems they confronted in building new statistics, remembers Neva Makgetla of the Trade & Indus-
trial Policy Strategies Institute:
The ILO provided a lot of help to set up the system. Before 1994, even in the census Africans were not counted at
all if they were in the former so-called ‘homelands’–at that time close to half the population –and they only
counted 10 percent of Africans living in the nominally ‘White’areas. The then Central Statistical Office had no
idea how to manage a survey of any kind that included Africans fully. They were themselves all White, for a
start …(Interview with Neva Makgetla
10
, Johannesburg, 2018)
The move towards international standards was buttressed by international consultants who sup-
ported Orkin and his team. The Swedish programme for example assisted not only with strategic
management systems and the development of provincial offices and census planning; it also
helped to refine household survey methodologies and to improve South Africa’s national accounts
(Stats SA 1999). Transnational expert networks helped diffuse de facto international standards.
The 2000s: Discouraged Work Seekers or an Expanded Definition?
If the 1990s had earned the narrow unemployment indicator the ‘official’label, the 2000s solidified
this position. The September 2004 Labour Force Survey (LFS) was the last to detail the expanded
unemployment as much as the strict, official one. After that, the broad unemployment indicator
became an occasional, ancillary shadow statistic –even if debate has refused to die down about
the ‘discouraged work seekers’-category and the misfit of international standards with South
African conditions.
Early in the 2000s, statisticians were still content defending the legitimacy of both unemployment
measures. In a 2002 briefing Hirschowitz, then Deputy Director-General for Quality and Integration,
was ask to respond to criticism of the official (ie, narrow) unemployment figures from the Congress
of South African Trade Unions (COSATU), the largest trade union confederation in the country. She
replied that
there is an official and an expanded definition of unemployment. Countries are given the discretion to use either
definition depending on the circumstances […] both definitions are valid in South Africa and therefore Stats SA
used both of them. (Finance Standing Committee 2002)
Table 2 Official and Expanded unemployment rates measured by OHS 1994–97, and corollaries.
(ii): Official unemployment rate measured by OHS 1994–1997, and corollaries
1994 1995 1996 1997
d Unemployed measured by OHS: official definition (000s) 1,988 1,644 2,019 2,238
e = b+d Economically active (000s) 9,959 9,713 9,609 9,787
f = a-e Not economically active (000s) 10,907 11,612 12,206 12,507
g = 100*d/e Official unemployment rate ( % ) 20.0 16.9 21.0 22.9
h = 100*e/a Labour force participation rate ( % ) 47.7 45.5 44.0 43.9
(iii): Expanded unemployment rate measured by OHS 1994–97, and corollaries
1994 1995 1996 1997
i Unemployed measured by OHS: expanded definition: (000s) 3,672 3,321 4,197 4,551
j = b+i Economically active (000s) 11,643 11,390 11,787 12,100
k = a-j Not economically active (000s) 9,223 9,934 10,028 10,195
l = 100*i/j Expanded unemployment rate (%) 31.5 29.2 35.6 37.6
m = 100*j/a Labour force participation rate (%) 55.8 53.4 54.0 54.3
Source: Stats SA (1998).
NEW POLITICAL ECONOMY 9
Yet in the mid-2000s, South African labour statistics shifted further towards the narrow definition. The
LFS of the early 2000s still featured both the official and the expanded unemployment numbers, even
in the highlights; the annexes contained detailed information about the latter. Nevertheless, the Ten
Year Review (a decade after the end of apartheid) of government programmes, issued by the Policy
Co-ordination and Advisory Services, clearly supported the strict definition (PCAS 2003), and the
mood shifted further against the broad one.
By 2005 the broad unemployment rate had reached roughly 40 percent. Finance minister Manuel –earlier a sup-
porter of a flexible approach –was incredulous: ‘If 40 percent of South Africans were really unemployed, there’d
be a revolution.’He warned that unemployment figures should not simply be ‘bandied about’.‘Sure, unemploy-
ment is a problem,’he said, ‘but that figure is wrong.’(IOL 2005)
Thabo Mbeki, the president himself, towed a similar line: in an ANC Today column, he observed that, if
one were to believe the figures,
in March 2004 there were at least 4 million South Africans walking about in our villages, our towns and cities
‘actively looking for work’. This is such a large number of people that nobody could possibly have missed the
millions that would be in the streets and village paths ‘actively looking for work’in all likely places of employment.
It, therefore, seems quite unlikely that the Stats SA figure is correct, if indeed it used the standard international ILO
definition to determine the unemployment rate. (Mbeki 2005)
In the LFS from September 2004, expanded unemployment numbers had been relegated to
the annexes. Half a year later, the March 2005 LFS replaced the ‘expanded unemployment’category
with a separate entry for ‘discouraged work seekers’–the jobless who wanted a job and were
available but had not sought work because no jobswereavailableinthearea,becausethey
were unable to find work requiring their skills, or because they had lost hope of finding any kind
of work (Stats SA 2009). The statistics increasingly walled offthe formally unemployed from
otherpeopleinwantofwork.
This shift attracted political attention. The opposition decried the narrowness of the indicator. In
October 2005, the Finance Standing Committee questioned dropping the broad definition of unem-
ployment ‘as the strict definition did not accommodate the large informal and self-employed sector,
which needed to be measured’(Finance Portfolio Committee 2005). The South African Reserve Bank
(SARB) also raised doubts, given the importance of unemployment for its monetary policy. Hirscho-
witz offered three defences: first, a separate ‘discouraged work seekers’category would allow better
identification of this group’s characteristics. Second, the new approach followed recommendations
by the IMF, which had reviewed South African labour statistics. Third, the old presentation might
confuse people who would ‘[compare] South Africa’s broad definition with other countries’strict
definitions’. Ian Davidson of the opposition Democratic Alliance remained unimpressed, fearing a
growing disconnect between the real and the statistical world. The new approach would not
provide the necessary information about the difference between the broad and strict unemployment
definitions. ‘The broad category of unemployment must be retained and captured as it would reflect
the real jobless rate’(Ibid).
Spats continued between observers who found the figures too high or too low, and they further
discredited the home-grown measures. Pali Lehohla, Statistician General until 2017 and Orkin’s
deputy before 2000, conceded that Stats SA ‘had previously suffered negative publicity regarding
the accuracy of its statistics pertaining to its community survey, after irresponsible reporting by a
certain journalist [..] and after allegations’(Finance Standing Committee 2008). However, the
‘narrow definition was adopted for international comparability and hence was the official definition
for unemployment’, he avowed (Ibid).
To regain credibility, Stats SA asked the World Bank to review its statistical approach (interview
with Peter Buwembo, Pretoria, January 2018). Heeding the Bank’s advice, Stats SA completely over-
hauled its survey design and shifted from the bi-annual LFS to a Quarterly Labour Force Survey (QLFS)
(Yu 2009).
10 J. ALENDA-DEMOUTIEZ AND D. MÜGGE
The report introducing the new statistical tools highlights conformity with ‘internationally
acclaimed practices’(Stats SA 2009, p. 19) no less than six times. To avoid being ground up
between politically opposing domestic parties, national statisticians sought refuge among inter-
national statistical experts and their standards. That spirit lingers. In the words of Rashad Cassim,
head of the SARB research department, former Deputy Director-General at Stats SA and member
of the Statistics Council:
We really invest in making sure that what we do is keeping international practices. So we follow very strictly the
ILO convention around what is considered an informal worker, what is considered a discouraged worker. (Inter-
view with Rashad Cassim, Pretoria, February 2018)
Through this full embrace of ILO definitions, the ambiguity about where unemployment ends and
genuine inactivity begins in South African statistics disappeared: the QLFS unequivocally files discour-
aged work seekers under ‘not economically active’. It has remained that way since (see Table 3, taken
from the first QLFS 2018).
The Evolution and Debates Until Today
In specialist circles, the debate about these categories continues. As always with statistical categor-
isation, the devil is in the detail. Reviewing the first QLFS of 2008, Meth observed that
it would seem that in the past, those who said that they lacked the money to pay for transport to seek work, or
who said that there was no transport available, were classified as discouraged. The new definition no longer
includes such folk, an important change, and one which deserves to be widely debated, especially in view of
apartheid’s horrible distortions of South Africa’s spatial economy. (Meth 2009, p. 84)
Even disregarding such seeming technicalities, many discouraged work seekers are arguably still part
of the labour force. Posel et al.(2014) show that many do not search actively because of the literal and
figurative costs of job search and the low chances of success; instead, they rely on their social net-
works for sustenance (cf. Lloyd and Leibbrandt 2013, Merten 2016). Faldie Esau, member of the
South African Statistics Council, confirmed that argument:
Table 3. Key labour market indicators, as presented in the first QFLS 2018.
Jan-Mar
2017
Oct-Dec
2017
Jan-Mar
2018
Qtr-to-qtr
change
Year-on-year
change
Qtr-to-qtr
change
Year-on-year
change
Thousand Per cent
Population 15–64 yrs 37,061 37,525 37,678 153 618 0.4 1.7
Labour Force 22,426 22,051 22,358 307 −68 1.4 −0.3
Employed 16,212 16,171 16,378 206 165 1.3 1.0
Formal sector (Non-
agricultural)
11,337 11,244 11,355 111 18 1.0 0.2
Informal sector (Non-
agricultural)
2,681 2,808 2,901 93 220 3.3 8.2
Agriculture 875 849 847 −3−28 −0.3 −3.3
Private households 1,319 1,270 1,275 5 −45 0.4 −3.4
Unemployed 6,214 5,880 5,980 100 −234 1.7 −3.8
Not economically active 14,634 15,474 15,320 −154 686 −1.0 4.7
Discouraged work-
seekers
2,277 2,538 2,787 249 510 9.8 22.4
Other (not economically
active)
12,357 12,936 12,533 −403 176 −3.1 1.4
Rates (%)
Unemployment rate 27.7 26.7 26.7 0.0 −1.0
Employed / population
ratio (Absorption)
43.7 43.1 43.5 0.4 −0.2
Labour force participation
rate
60.5 58.8 59.3 0.5 −1.2
Source: Stats SA (2018).
NEW POLITICAL ECONOMY 11
The challenge that you have is the tools [are] not designed for lower levels. There are certainly pockets of unem-
ployment in certain provinces for a lot of reasons. I will give you some examples. It’s called Murraysburg and the
closest town, a big town, is Graff-Reinet. So the challenge for those people is that they don’t have the money to go
to Graff-Reinet and register. So they may do for a few months and after they stop doing that. (Interview with
Faldie Esau, Cape Town, May 2018)
These arguments frequently run along political lines. In the words of Peter Buwembo:
Some people prefer to use the broader [definition], because they have some interest, especially the unions. (Inter-
view with Peter Buwembo, Pretoria, January 2018)
Neva Makgetla doubted whether these political motivations would not distort the unions’under-
standing of the statistics:
The important thing, of course, is what you are trying to reflect –the ratio of people seeking work actively to the
employed, or the ratio of those who want work, even if they’ve given up looking for it. [..] People who don’t actu-
ally work with the numbers often fetishize them –there were some people in COSATU who insisted on the broad
figure mostly because they wanted to get government to prioritise unemployment more. (Interview with Neva
Makgetla, Johannesburg, May 2018)
Thus, in South Africa, finding employment is widely seen as the key to escaping poverty. Political
debates hence revolve around unemployment and poverty, but sideline the large number of working
poor, who do not fit the jobs vs. poverty dichotomy. In 2012, more than a fifth of workers lived in
households unable to meet basic needs, and 58 percent of poor South Africans lived in a household
with an employed person (Rogan and Reynolds 2015). Indeed, Scully (2016) calculated that 42
percent of South Africa’s employed labour force can count as precarious, such that unremunerated
labour in the household and beyond remains essential to people’s survival strategies. Dominant
views of unemployment effectively hide such problems.
Where criticism from the unions tends to cast unemployment figures as too low, criticism from the
business sector has veered in the opposite direction. In 2011 the Adcorp work agency and consul-
tancy avowed that the actual unemployment rate was only around 11.3 percent (Harding 2014), as
opposed to the official 24.8 percent. Adcorp economist Loane Sharp observed that this latter
number is ‘simply incredible, because we should have expected civil disobedience and disorder on
a grand scale if this were true’(Harding 2014). Whence this discrepancy between the real and the
official numbers? Stats SA vastly underestimated informal employment, Adcorp argued, and
should add more than six million people to its employment figures. The Adcorp methodology
immediately drew academic fire (Wittenberg and Kerr 2012), but also fuelled doubts about the
reliability and usefulness of official statistics (van der Berg 2013).
On the back of such incessant debate, the Finance Standing Committee debated the unemploy-
ment definition yet again in May 2013. Speaking for the Democratic Alliance, Tim Harris suggested
that ‘Stats SA be asked to compile data on the broader definition to enable the Committee to get
a better picture of the situation’(Finance Standing Committee 2013). The former Deputy Director-
General of Stats SA agrees:
I do think that saying that the narrow unemployment rate is 27 percent is misleading. Because at the end of the
day you’re saying that discouraged workers are technically not part of your unemployment, because they tech-
nically stopped looking for a job. So because they stopped looking for a job, they are not unemployed anymore.
But they’ve stopped looking because they couldn’tfind [one]. So my view is that although, in terms of compar-
ability, we give the narrow definition, I do think as a country we constantly have to give two together […] And we
should be constantly monitoring why these discouraged workers are discouraged. (Interview with Rashad Cassim,
Pretoria, February 2018)
Based on these arguments, Stats SA has carefully reintroduced some of this information by listing
the expanded definition in the QLFS annexes detailing unemployment by province in 2010. In
addition, three years later, it is inserted it in the last section of the principal results, through a map
summarising the QLFS (see the example in Figure 3). Since then, the expanded rate has operated
as a kind of shadow statistic. The figures are available to those who really want to know and look
12 J. ALENDA-DEMOUTIEZ AND D. MÜGGE
for them; for the rest, the narrow definition of unemployment remains the official one to be used in
political discourse and the media.
Conclusion
Unemployment has no obvious demarcations, and indicators to capture it are therefore fundamen-
tally ambiguous and potentially vulnerable to contestation. This ambiguity becomes particularly clear
in countries such as South Africa, in which broad and narrow definitions produce such widely dispa-
rate numbers.
This ambiguity has been hard to sustain, certainly for a fledging statistical office such as the South
African one. During the past decades, unemployment measures in the country have undergone a
dual movement. After the end of Apartheid, the Central Statistical Service initially embraced a
broad unemployment definition to capture the socio-economic realities particularly of Black commu-
nities. Yet over the course of the 1990s and the 2000s, international standards increasingly dominated
South African statistics, narrowing the unemployment indicator evermore.
Definitional quandaries do not only affect South African statistics. Labour markets in rich countries
also change rapidly. Automation affects which kind of labour is in demand and which one is no
longer. Entrepreneurial forms of self-employment are on the rise, as are part-time work and combi-
nations of several jobs to make ends meet. Female labour force participation rates have risen substan-
tially over the past decades, as well. The male factory worker as unemployment statistics lodestar is
less and less useful –all around the world.
South Africa had tried to move beyond established unemployment standards in the early years
of democratisation, and current Stats SA publications offer useful, nuanced detail about the
countries labour markets. Nevertheless, the narrow unemployment rate is now unambiguously
the official series. Whenever a single number for South African labour market conditions is being
sought –whether by international investors, for global comparisons or by academics using
Figure 3. Summary of labour market measures at a glance, Q1: 2018. Source: Stats SA (2018).
NEW POLITICAL ECONOMY 13
large-n data sets in their research –they will use one that has been fought over for decades, and
which remains contested as a reflection of South African labour markets. ‘As far as I know, the
debate about broad versus narrow is no longer particularly important,’Neva Makgetla told us.
Thus, as Desrosières (1998) put it, the statistical measure of unemployment in South Africa has
now become an established convention –‘information’–insofar as it has become reliable, even
if it offers a skewed representation of reality.
What has driven this dynamic? It is worth noting first what we did not find. As pointed out above,
alternative unemployment definitions and measures clearly have a racial dimension. An expanded
rate emphasises the poor labour market conditions of Black South Africans much more clearly
than a narrow definition does. Critics could have levelled the charge of (potentially inadvertent)
racism at narrow unemployment measures.
In our interviews and document research, however, we found that actual debate in South Africa
has accorded surprisingly little attention to this issue. Presumably this relative silence is explained by
the ANC’s grip on political power since the end of Apartheid. With its roots in the Black South African
community, the party might have an incentive to highlight racial biases in political institutions such as
official statistics. However, it also has a political track record to defend, and for that, the expanded rate
is counterproductive. The charge of racial bias might have been much more prominent if other pol-
itical actors would have wielded power in South Africa.
Instead, our analysis highlights the counterintuitive role of international statistical harmonisation.
The high speed of statistical (and concurrent democratic) construction in the country since the end of
apartheid has engendered a direct need to legitimize the new ways of measuring South Africa’s social
and economic conditions. In line with our expectations, this legitimacy was first sought internation-
ally. Compliance with international standards increasingly functioned as a seal of approval and
quality, insulating figures against claims of political bias.
This finding is not as obvious as it may seem. We have argued above that unemployment indi-
cators are fundamentally ambiguous; there never is one obviously correct measurement, and the
merits of alternative standards vary across countries. That may seem to make international standards
both harder to achieve and less attractive, because no single size fits all. Instead, we find ambiguity to
have the opposite effect: it incentivizes national statistical agencies to sign up to international stan-
dards as buffers against domestic criticism, which itself feeds on the indicator ambiguity. Inter-
national harmonisation may thus be alluring not in spite, but because of the indeterminacy of
statistical measures and their inevitable political weight.
Notes
1. The label ‘developing countries’carries regrettable connotations, for example differences in countries’advance-
ment along a single ‘development’path or a materialist conception of national progress. We don’t endorse those
connotations. Lacking better alternatives, we simply use it, reluctantly, to designate relatively poor, non-Western
countries.
2. Early in 2018, there were approximately 5.9 million officially unemployed in South Africa (actively searching for a
job) as well as 2.7 million discouraged work-seekers out of a population of somewhat below 60 million (Stats SA,
2018).
3. 10 years after the establishment of the Representation of Native Act in 1936, the ethnic homelands (or Bantu-
stans) were created to assign black Africans.
4. At the time, the four largest of those conglomerates controlled 83 per cent of the companies listed on the Johan-
nesburg Stock Exchange before apartheid ended. Their close links to the state made for a highly coordinated
‘national’approach to economic policy.
5. Two central characters in the evolution of South African statistics have been Mark Orkin, then the Head of CSS and
later the first Statistician-General of Statistics SA, and Trevor Manuel, the first post-1994 Minister of Finance.
6. One of the designers of the OHS, which initially underpinned unemployment figures, and Statistician General for a
short period after Mark Orkin and before Pali Lehohla at Stats SA.
7. Linguistic difference can complicate things further. In some official languages, ‘interviewers would have trans-
lated ‘unemployed’as ‘looking for work, [..] others [..] simply as not working’(Stats SA, 1998, 64).
14 J. ALENDA-DEMOUTIEZ AND D. MÜGGE
8. All the reports (OHS, LFS, QFLS) have three parts: the highlighted results, which are a table summarising the prin-
ciple results of the study; some principal results, regarding employment, unemployment, and other specific infor-
mation; the annexes, were all the rest of the information is delivered in tables.
9. Henceforth, the unemployed were those people, within the economically active population, who did not work
during the seven days prior to the interview, want to work and are available to start work within a week of
the interview, and have taken active steps to look for work or to start some form of self-employment in the
four weeks prior to the interview.
10. Senior Economist at the Trade & Industrial Policy Strategies institute. She has been head of the COSATU Policy
Unit, and has been involved at the Development Bank of Southern Africa and at the Economic Development
Department.
Acknowledgements
This research is part of the FICKLEFORMS project at the University of Amsterdam. We are grateful to the team members
for their support and helpful comments.
Disclosure statement
No potential conflict of interest was reported by the authors.
Funding
This work was supported by Nederlandse Organisatie voor Wetenschappelijk Onderzoek [Vidi grant 016.145.395]; H2020
European Research Council [Grant Number 637683].
Notes on contributors
After a PhD in France, Juliette Alenda-Demoutiez is now a Post-doctoral Fellow at the University of Amsterdam, working
on the history of macroeconomics indicators in South Africa. Her research interests are, besides in political economy of
statistics, in development and social protection, still in Sub-Saharan Africa.
Daniel Mügge is Professor of Political Arithmetic at the University of Amsterdam. Together with a team of researchers, he
studies the political roots of macroeconomic indicators and their political baggage.
References
Adelzadeh, A., 1996. From the RDP to GEAR: the gradual embracing of neo-liberalism in economic policy. Transformation,
31, 66–95.
Bangane, T., 1999.The unemployment problem in South Africa with specific reference to the LVTMA. Johannesburg: Magister
Dissertation.
Barchiesi, F., 2011.Precarious liberation: workers, the state, and contestes social citizenship in post-apartheid South Africa.
Albany, NY: SUNY Press.
Baxandall, P., 2004.Constructing unemployment: the politics of joblessness in east and west. Hoboken, NJ: John Wiley &
Sons.
Broome, A., Homolar, A., and Kranke, M., 2018. Bad science: international organizations and the indirect power of global
benchmarking. European journal of international relations, 24 (3), 514–539.
Broome, A. and Quirk, J., 2015. Governing the world at a distance: The practice of global benchmarking. Review of inter-
national studies, 41 (5), 819–841.
Carmody, P., 2002. Between globalisation and (post) apartheid: The political economy of restructuring in South Africa.
Journal of Southern African studies, 28, 255–275.
Chandra, V., Nganou, J., and Noel, C., 2002.South Africa - constraints to growth in Johannesburg’s black informal sector -
evidence from the 1999 informal sector survey. Washington, DC: World Bank.
Clegg, L., 2010. Our dream is a world full of poverty indicators: the US, the world bank, and the power of numbers. New
political economy, 15 (4), 473–492.
Cling, J.-P., 1999. La politique économique sud-africaine face à la mondialisation. Politique africaine, 73 (73), 146–158.
Cooley, A. and Snyder, J. eds., 2015.Ranking the world. Grading states as a tool of global governance. New York: Cambridge
University Press.
NEW POLITICAL ECONOMY 15
Cousins, B., et al., 2018. Social reproduction of ‘classes of labour’in the rural areas of South Africa: contradictions and
contestations. The Journal of Peasant studies,45(5–6), 1060–1085.
CSS, 1996.Living in South Africa, selected findings of the 1995 October household survey. Pretoria: CSS.
Davis, K., et al., 2012.Governance by indicators: global power through quantification and rankings. New York: Oxford
University Press.
Desrosières, A., 1998.The politics of large numbers: A history of statistical reasoning. Cambridge, MA: Harvard University
Press.
Du Toit, A. and Neves, D., 2014. The government of poverty and the arts of survival: mobile and recombinant strategies at
the margins of the South African economy. Journal of peasant studies, 41 (5), 833–853.
Ferguson, J., 2015.Give a Man a Fish: reflections on the new politics of distribution. Durham, NC: Duke University Press
Books.
Finance Standing Committee, 2002. minutes can be found on the archives of the website of the Parliamentary
Monitoring Group. Available from: https://pmg.org.za/committee-meetings/.
Finance Standing Committee, 2005. Minutes can be found on the archives of the website of the Parliamentary Monitoring
Group. Available from: https://pmg.org.za/committee-meetings/.
Finance Standing Committee, 2008. Minutes can be found on the archives of the website of the Parliamentary Monitoring
Group. Available from: https://pmg.org.za/committee-meetings/.
Finance Standing Committee, 2013. Minutes can be found on the archives of the website of the Parliamentary Monitoring
Group. Available from: https://pmg.org.za/committee-meetings/.
Fioramonti, L., 2013.Gross domestic problem. The politics behind the world’s most powerful number. London: Zed Books.
Garraty, J.A., 1979.Unemployment in history. Toronto: Fitzhenry and Whitheside Limited.
Gautié, J., 2002.Del’invention du chômage à sa deconstruction. Genèses, 46 (46), 60–76.
Gerson, J., 1981.The question of structural unemployment in South Africa. South African journal of economics, 49 (1), 7–15.
Green, A., 2000. Problems of measuring participation in the labour market. In: D. Dorling, and S. Simpson, ed. Statistics in
society: the Arithmetic of politics. London: Arnold, 312–323.
Green, P., 2008.Choice, not fate. The life and times of trevor manuel. Johannesburg: Penguin books.
Hamilton, L., 2014.Are South Africans Free? London: Bloomsbury Academic.
Harding, A., 2014. South African unemployment: Lies and statistics. BBC, 14 November. Available from: http://www.bbc.
com/news/world-africa-20324189.
Hoskyns, C. and Rai, S., 2007. Recasting the global political economy: couting women’s unpaid work. New political
economy, 12 (3), 297–317.
Hussmanns, R., Mehran, F., and Verma, V., 1990.Surveys of economically active population, employment, unemployment
and underemployment: An ILO manual on concepts and methods. Geneva: ILO.
IOL, 2005.Manuel is too glib about unemployment. Available from: https://www.iol.co.za/business-report/opinion/manuel-
is-too-glib-about-unemployment-749442.
Isaacs, G., 2016.A national minimum wage for South Africa. Johannesburg: NMW-RI.
Jany-Catrice, F., 2012.La performance totale: nouvel esprit du capitalisme? Villeneuve d’Ascq: Presses universitaires du
Septentrion.
Jerven, M., 2013.Poor numbers: how we are misled by African development statistics and what to do about it. Ithaca: Cornell
University Press.
Kantor, B., 1980. Blacks: Is there unemployment? Businessmen’s law, 9 (104-7), 143–144.
Kelley, J.G., 2017.Scorecard diplomacy. Grading states to Influence their reputation and behavior. Cambridge: Cambridge
University Press.
Khan, A., Wales, J., and Stuart, E., 2015.Country priorities for data development: what does history tell us? London: ODI.
Kingdon, G. and Knight, J., 2001. What have we learnt about unemployment from microdatasets in South Africa? Social
Dynamics,27 (1), 79–95. 10.1080/02533950108458705.
Koelble, T.A., 2004. Economic policy in the post-colony: South Africa between Keynesian remedies and neoliberal pain.
New political economy, 9 (1), 57–78.
Lehohla, P., 2002.The National Statistics System: Our Challenge. Presentation at Paris21.
Lepenies, P., 2013.Die Macht der einen Zahl: eine politische Geschichte des Bruttoinlandsprodukts. Frankfurt: Suhrkamp
Verlag.
Linsi, L. and Mügge, D., 2019. Globalization and the growing defects of international economic statistics. Review of inter-
national political economy,DOI:10.1080/09692290.2018.1560353.
Lloyd, N. and Leibbrandt, M. 2013. How do the non-searching unemployed feel about their situation? On the definition of
unemployment. Econ3(3, 25 June. Available from: http://www.econ3(3.org/article/how-do-non-searching-
unemployed-feel-about-their-situation-definition-unemployment.
Masood, E., 2016.The great Invention. The story of GDP and the making and unmaking of the modern world. New York:
Pegasus Books.
Mbeki, T., 2005. Letter from the president. ANC today, 5 (20), 20–26. May.
16 J. ALENDA-DEMOUTIEZ AND D. MÜGGE
Merten, M., 2016. SA Unemployment and the community survey: it’s complicated. Daily Maverick, 24 March.
Available from: https://www.dailymaverick.co.za/article/2016-03-24-sa-unemployment-and-the-community-survey-
its-complicated/.
Meth, C., 2009. Why waste money on quarterly labour force Surveys? Waste it on youth development instead!.
Transformation critical perspectives on Southern Africa, 71 (1), 76–102.
Moon, J. and Richardson, J., 1985.Unemployment in the UK: politics and policies. London: Gower.
Mügge, D., 2019. International economic statistics: biased Arbiters in global affairs? Fudan journal of humanities and social
sciences,https://doi.org/10.1007/s40647-019-00255-5.
PCAS, 2003.Towards a ten year review, synthesis report on implementation of government programmes. Pretoria: Office of
the Presidency.
Porter, T., 1995.Trust in numbers. The Pursuit of objectivity in Science and public life. Princeton: Princeton University Press.
Posel, D., Casale, D., and Vermaak, C., 2014. Job search and the measurement of unemployment in South Africa. South
African journal of economics, 82, 66–80.
Rogan, M. and Reynolds, J., 2015.The working poor in South Africa, 1997–2012. Grahamstown: Institute of Social and
Economic Research, Rhodes University.
Rogerson, C.M., 1992. The absorptive capacity of the informal sector in the South African City. In: D. M. Smith, ed. The
apartheid city and beyond. London: Routledge, 161–172.
Salais, R., Baverez, N., and Reynaud, B., 1986.L’invention du chômage. Paris: Puf.
Scully, B., 2016. From the shop floor to the kitchen table: the shifting centre of precarious workers’politics in South Africa.
Review of African political economy, 43 (148), 295–311.
Sengenberger, W., 2011.Beyond the measurement of unemployment and underemployment. Geneva: ILO.
Simkins, C., 1982.Structural unemployment revisited. SALDRU Fact Sheet No. 1, University of Cape Town.
Standing, G., Sender, J., and Weeks, J., 1996.Restructuring the labour market, The South African challenge. Geneva: ILO.
Stapleford, T., 2009.The cost of living in America: A political history of economic statistics, 1880–2000. Cambridge:
Cambridge University Press.
Stats SA, 1998.Unemployment and employment in South Africa. Pretoria.
Stats SA, 1999.Annual report. Pretoria.
Stats SA, 2009.Annual report. Pretoria.
Stats SA, 2018.Quarterly labour force Survey, 1st quarter. Pretoria.
Taylor, M., 2016.The Political Economy of Statistical Capacity: A Theoretical Approach (Discussion Paper IDB-DP-471).
Washington, DC: Inter-American Development Bank.
Terreblanche, S., 2002.A history of inequality in South Africa, 1652–2002. Scottsville: University of Natal Press, KMM Review.
Topalov, C., 1994.Naissance du chômeur, 1880–1910. Paris: Albin Michel.
Van der Berg, S., 2013. Adcorp’s employment and unemployment figures are not taken seriously by researchers –yet they
can do much harm. Econ3(3, 12 February. Available from: http://www.econ3(3.org/article/adcorp%E2%80%99s-
employment-and-unemployment-figures-are-not-taken-seriously-researchers-%E2%80%93-yet-they.
Ward, M., 2004.Quantifying the world: UN ideas and statistics. New York: Indiana University Press.
Wittenberg, M. and Kerr, A. 2012.Criticisms of the adcorp employment index.DataFirst, University of Cape Town, February.
Available from: http://nationalminimumwage.co.za/wp-content/uploads/2015/08/0033-Criticisms-of-the-Adcorp-ind
ex.pdf.
Yu, D., 2009. The comparability of labour force survey and quarterly labour force survey. Department of economics, Bureau
for economic research, University of Stellenbosch.
Zimmermann, B., 2006.Arbeitslosigkeit in Deutschland. Zur Enstehung Einer Sozialen Kategorie. Frankfurt am Main: Campus
Verlag.
NEW POLITICAL ECONOMY 17
