Content uploaded by Ivan Turok
Author content
All content in this area was uploaded by Ivan Turok on Sep 30, 2020
Content may be subject to copyright.
National Income Dynamics
Study (NIDS) – Coronavirus
Rapid Mobile Survey (CRAM)
WAVE 2
14
30 September 2020
Justin Visagie - Human Sciences Research Council
Ivan Turok - Human Sciences Research Council
The Uneven
Geography of the
COVID-19 Crisis
Embargoed until 30 September 12:00 (noon). Do not quote, cite or share.
The Uneven Geography of the
COVID-19 Crisis
Justin Visagie, Research Specialist, Human Sciences Research Council1
Ivan Turok, Distinguished Research Fellow, HSRC, and NRF Research Professor, University of the Free State2
30 September 2020
Abstract
This paper analyses the impact of the pandemic on different parts of South Africa, bearing in mind
their contrasting vulnerability and resilience. It compares the severity of the initial COVID-19 shock
(February-April 2020) and the subsequent trajectory (April-June) of the metros, smaller cities/towns
and rural areas. It also considers the different impacts within cities – between suburbs, townships,
shack areas and peri-urban areas. A key question is whether COVID-19 has aggravated pre-existing
spatial disparities? A second question is whether government social support has helped to mitigate
these gaps in income and well-being? The paper reveals that the pandemic has magnified the
existing economic and social divides (i) between cities and rural areas, and (ii) between suburbs
and townships/informal settlements within cities. Government grants have helped to offset the large
economic disparities between places, but the incidence of hunger is still much higher in informal
settlements, townships and rural areas than in suburbs. There is a strong case for more targeted
efforts to boost jobs and livelihoods in lagging urban and rural areas. Pre-existing conditions were
bad enough, but now there is further ground to make up, and the withdrawal of temporary relief grants
could be a serious setback for poor communities and groups reliant on cash transfers.
1 jpvisagie@hsrc.ac.za
2 iturok@hsrc.ac.za
Embargoed until 30 September 12:00 (noon). Do not quote, cite or share.
1 | The Uneven Geography of the COVID-19 Crisis
Executive summary
This paper analyses the impact of the pandemic on different parts of South Africa, bearing in mind
their contrasting vulnerability and resilience. It compares the severity of the initial COVID-19 shock
and the subsequent trajectory of the metros, smaller cities/towns and rural areas. It also considers
the different impacts within cities – between suburbs, townships, shack areas (informal settlements
and backyarders) and peri-urban areas (small-holdings, farms or tribal land on the urban fringe).
A key question is whether COVID-19 has aggravated pre-existing spatial disparities? A second
question is whether government support has helped to mitigate these gaps in income and well-
being?
The metros proved more resilient than rural areas and cities/towns. They started out in February
in a much stronger position with 57% of adults (over 18) in paid employment, compared with 46%
in smaller cities/towns and 42% in rural areas. All regions lost about a fifth of their jobs between
February-April. However between April-June metros and smaller cities/towns had already begun
their recovery whilst rural areas continued to lose jobs. The net result was that rural unemployment
in June was 52% compared with 43% in cities/towns and 35% in the metros.
The suburbs resisted the shock of the lockdown better than townships and informal
settlements. They were in a strong position in February with 58% of adults in paid employment,
then lost one in seven of their jobs (14%) by April, compared with one in four in the townships (24%)
and peri-urban areas (23%) and more than a third of jobs (36%) in shack areas! Shack dwellers were
extremely vulnerable to the lockdown and restrictions on informal enterprise and related activities.
There were signs of a recovery in shack areas between April-June although partly because
furloughed workers had been brought back onto the payroll. Overall, the economic crisis has hit
poor urban communities much harder than the suburbs, resulting in a rate of unemployment
of 42-43% in townships and informal settlements compared with 24% in the suburbs.
In summary, the pandemic has magnified pre -existing economic divides (i) between cities
and rural areas, and (ii) between suburbs and townships/informal settlements within cities.
Turning to the provision of social support, rural communities have been much bigger beneficiaries
of government grants than the metros and smaller cities/towns. Nearly three out of five rural
respondents (59%) lived in households receiving social grants in June 2020, compared with less
than half in cities/towns (47%) and one in three in the metros (32%). This was because rural residents
were far less likely to be in paid employment. Government grants have clearly helped to protect rural
livelihoods and compensate these areas for their weak local economies and lack of jobs. However,
this poses a risk to these communities when the temporary relief is withdrawn.
Similar points apply to the differences within cities, where more than half of peri-urban
respondents (54%) lived in households receiving social grants, compared with less than half of
township residents (45%), two in five shack dwellers (40%) and one in four suburban residents
(26%). The implication is that government grants have helped to offset unemployment and
poverty in townships and informal settlements. The premature withdrawal of social programmes
could aggravate conditions in poor urban communities.
In terms of special relief from the crisis, one in three rural residents (33%) said that someone
in their household had received the COVID-19 grant, compared with one in four in cities/
towns (24%) and one in five in the metros (21%). These differences are smaller than for other
grants, suggesting that the COVID-19 grant is benefiting people who did not qualify for government
support before, such as unemployed men. Among urban residents, 29% of peri-urban residents
said their households had received the COVID-19 grant, compared with 27% in townships, 18% of
shack dwellers and 16% in suburban areas. The proportion of shack dwellers receiving these and
Embargoed until 30 September 12:00 (noon). Do not quote, cite or share.
2 | The Uneven Geography of the COVID-19 Crisis
other grants is surprisingly low considering their levels of poverty and distress. Further research is
required to explain this.
The proportion of respondents who said their household had run out of money to buy food in
April was 44% in the metros, 48% in the cities/towns and 52% in the rural areas. These figures
were very high everywhere. By June 2020, these proportions had fallen to 35% in the metros, 37%
in the cities/towns and 40% in the rural areas.
The proportion of respondents who said that someone in their household had gone hungry in
the last seven days (in May/June) was 17% in the metros, 24% in the cities/towns and 29% in
the rural areas. By July these proportions had fallen to 13% in the metros, 16% in the cities/towns
and 20% in the rural areas. In other words, hunger had fallen everywhere, but was still worse in the
rural areas.
Turning to the differences within cities, the proportion of respondents who said their household
had run out of money to buy food in April was 31% in the suburbs, 48% in the townships and
61% in the shack areas. Shack-dwellers were noticeably worse off than rural respondents. This
adds to the concern that far fewer shack-dwellers receive government grants. By June 2020, these
proportions had fallen to 24% in the suburbs, 40% in the townships and 50% in the shack areas.
Everywhere improved, although the gap between the shack-dwellers and other groups was still
large. Shack-dwellers also continued to be worse off than rural residents, and with less social relief.
The proportion of urban respondents who said that someone in their household had gone
hungry in the last seven days (in May/June) was 11% in the suburbs, 22% in the townships
and 32% in the shack areas. By July/August these proportions had fallen to 7% in the suburbs,
16% in the townships and 22% in the shack areas. The differences between urban neighbourhoods
clearly remained very large.
Summing up, government social grants have helped to offset the large economic gaps
between places, but the incidence of hunger is still much higher in informal settlements,
townships and rural areas than in suburbs. There is a case for more targeted efforts to boost jobs
and livelihoods in lagging urban and rural areas. Pre-existing conditions were bad enough, but now
there is further ground to make up, and the withdrawal of temporary relief grants could be a serious
setback for poor communities reliant on cash transfers, especially for groups who did not qualify for
grants before, such as unemployed young men.
Embargoed until 30 September 12:00 (noon). Do not quote, cite or share.
3 | The Uneven Geography of the COVID-19 Crisis
1. Introduction
The geography of the country’s worst public health crisis in a century has been neglected to date,
despite major contrasts in the vulnerability and resilience of different places. South Africa is one of
the most unequally developed countries in the world (Makgetla, 2018; Turok, 2018), so one would
expect the pandemic to have uneven spatial impacts. The geography of the economy matters
because some places are better resourced than others to withstand and recover from shocks, having
more diverse industries and stronger local institutions. The spatial distribution of the population also
matters because the risks and hazards facing different communities vary greatly, with different
levels of education and employment, and different local resources to fall back on in times of stress.
The focus of the disease analysis and response has been at the national and provincial levels, yet
the coronavirus spreads locally through human contact and interaction. Large cities became the
infection hotspots and experienced higher mortality rates than towns and rural areas, reflecting
their relatively high population densities and their strong connections to external regions and
nations. Assessing the impact of COVID-19 on the welfare of people living in townships, informal
settlements and rural areas is vital because of the precarious nature of jobs and livelihoods in these
communities. Many of these places also have inhospitable living environments and weaker social
infrastructure and safety nets than suburban areas (Turok, 2014a, 2016; Seeliger and Turok, 2014;
Visagie and Turok, 2020).
The impact of the socio-economic shock and the nature of the subsequent recovery are bound to
differ between localities and regions. Spatial disparities in South Africa are usually reduced to a
simple urban-rural divide (e.g. World Bank, 2018). This duality is far too limited because it ignores
the exchange of resources, trade in goods and services, and other interactions between urban and
rural areas (Turok, 2018). It also obscures the enormous economic and social variations across
different kinds of urban area. For example, the economies of big cities have little in common with
those of small cities and towns (Motlanthe, 2017; Todes and Turok, 2018; Turok, 2016). Put simply,
different places within the country face different challenges and opportunities. These need to be
taken more seriously for the country as a whole to prosper.
The main objective of this paper is to assess the impact of the COVID-19 crisis on different types
of locality and region. This is important for government responses to be targeted more carefully
on the places that have suffered the worst effects or are struggling the most to recover. A blanket
approach to the provision of support that treats places equally will not narrow the gaps between
them. The analysis is novel and exploratory because the first phase of the NIDS-CRAM study
focused on individual characteristics (race, gender, education, occupation, earnings, etc.) and paid
little attention to spatial considerations.
Two particular locational typologies are employed for this analysis of spatial patterns and trends. The
first is concerned with the disparities between three different types of area - large cities (‘metros’),
smaller cities and towns (‘cities/towns’) and the countryside (‘rural areas’). The rationale for this is
explained below. The second focuses on the differences within cities between relatively rich and
poor neighbourhoods. A four-fold classification based upon residents’ own perceptions is used to
distinguish between suburbs, townships, shack dwellers (informal settlements and backyarders)
and peri-urban areas (which include small-holdings, farms or tribal land on the urban fringe). The
rationale for this typology is also explained below.
Looking at the first locational typology, the scale and composition of local economies should
influence their ability to bounce back from the difficult conditions experienced during the first few
months of the lockdown. In particular, large cities have more productive and diversified economies
than towns and rural areas, with stronger public and private institutions and human capabilities
(Martin, 2018; Turok, 2018). Many of their firms would have had larger reserves and other assets to
rely on. They would have been better placed to diversify into producing goods and services for new
markets (e.g. for personal protective equipment and other medical supplies).
In contrast, rural areas tend to have narrower and more fragile economies. Household earnings are
Embargoed until 30 September 12:00 (noon). Do not quote, cite or share.
4 | The Uneven Geography of the COVID-19 Crisis
lower, with greater dependence on remittances from elsewhere (Makgetla, 2018; Turok, 2014b). It
is more difficult for firms and families in these places to replace lost income with other sources.
Consequently, one would expect cities to be better positioned than towns and rural areas to resist
the economic consequences of the pandemic. Yet, this advantage may have been offset by their
greater openness to external trade and consequent susceptibility to the closure of national borders,
airports and seaports. Tourism has been an obvious casualty of the lockdown, affecting cities as
well as towns and rural areas.
Turning to the second theme of differences within cities, the marked contrasts between the
infrastructure and quality of life of suburbs, townships and informal settlements are bound to
influence the challenges people face in relation to their livelihoods and living conditions. One would
expect suburban residents to be more resilient to the lockdown because their jobs tend to be more
secure and higher paid. Professionals and white-collar workers found it easier to work from home
than manual workers and had larger savings to protect them in the event of being laid-off. Suburban
residents have higher levels of education and training, so they are more employable and have more
options available in the event of job loss. Car ownership is higher, so they are also more mobile than
residents of poorer neighbourhoods dependent on public transport.
In contrast, people living in townships and informal settlements experience higher population
densities and more crowded living conditions (Turok, 2014a; Turok and Borel-Saladin, 2016). They
suffered greater burdens from restrictions on movement and rules that they stay at home. Their
health services, schools, transport and childcare facilities are inferior and poorly equipped to cope
with pandemics, the protocols of social distancing and other sources of adversity. Residents have
higher levels of debt and fewer fixed assets to cushion them from setbacks. Temporary workers and
informal enterprises are particularly vulnerable to stoppages in trade which can lead to indefinite
layoffs and business closures. Households also tend to rely more on state support in the form of
social grants, free basic services and free school meals, which were stopped in most provinces
when the schools closed. Consequently, one would expect these residents to experience greater
misery and hunger during economic downturns and restrictions on public spending. They are more
likely to need relief in the form of food parcels and top-ups to social grants.
The paper focuses on three particular dimensions of the COVID-19 crisis: the labour market,
household incomes and the incidence of hunger. These phenomena are closely connected. The
chain of causation runs from the labour market to household incomes and onto hunger. The logic
is that changes in employment (such as job loss) are transmitted to households through a loss
of earnings, which in turn affects whether people go hungry. The scale and severity of the shock
are moderated by provision of social protection from the government in the form of social grants
and food parcels. A special COVID-19 social relief of distress grant worth R350 per month was
introduced in June 2020. The causal chain also works when conditions improve. The stronger the
recovery, the bigger the gain in employment and earnings, and the fewer people who go hungry.
The next section discusses the methods employed in the paper. The following section considers
the changes in employment and unemployment. The subsequent section assesses the changes in
household incomes and social assistance, followed by the incidence of hunger. Each section begins
by considering the contrast between metros, cities/towns and rural areas. It then examines the
differences within cities between suburbs, townships, informal settlements and peri-urban areas.
2. Methods and definitions
This paper draws primarily upon survey data from waves 1 and 2 of the National Income Dynamics
Study: Coronavirus Rapid Mobile Survey (NIDS-CRAM). The NIDS-CRAM was designed as a
‘barometer’ for assessing the socio-economic impact of the COVID-19 pandemic on South African
individuals and households (Spaull et al, 2020). The survey was based upon a sample of adults who
were previously surveyed as part of Wave 5 of the National Income Dynamics Study (NIDS) in 2017.
Hence, the NIDS-CRAM provides another two rounds of socio-economic data for a subsample of
Embargoed until 30 September 12:00 (noon). Do not quote, cite or share.
5 | The Uneven Geography of the COVID-19 Crisis
individuals (aged 18 years and older) from the NIDS:W5 who were re-interviewed in May/June 2020
(NIDS-CRAM: wave 1) and again in July/August (NIDS-CRAM: wave 2). At times we make use of the
NIDS:W5 as a baseline to give a sense of conditions before the onset of the pandemic.
Our approach is to describe the impact of COVID-19 on a broadly representative sample of individuals
living in different localities and regions. This is the first attempt to compile empirical evidence about
the uneven geography of the crisis. This interpretation of the trends and dynamics is a first step.
Further research is required to corroborate the analysis and help to explain some of the striking
findings, using different data sources and methods.
A natural concern of these findings is the extent to which the NIDS-CRAM might under- or over-
sample segments of the population from particular regions. Both the NIDS and NIDS-CRAM
apply weights which were calibrated to improve representivity between sample and population
and explicitly include spatial controls (see Kerr et al, 2020 and Branson and Wittenberg, 2019).
Notwithstanding post-stratification adjustments, our results should still be treated with caution for
the following reasons: firstly, the NIDS-CRAM sample is small even at the national level, which leads
to fairly large standard errors3. We take care to report on the margin of error in all our estimates.
Secondly, the original NIDS 2008 sample (which the NIDS-CRAM is based upon) is limited in its
design for sub-national analysis.4 Therefore we take care to focus our analysis on larger geographic
aggregations and avoid reporting at a provincial level.
We construct two different locational typologies based upon two different levels of spatial analysis
– inequalities between cities and rural areas, and within urban areas. The first typology divides the
country into cities and rural areas using three mutually exclusive types of location based upon an
individual’s reported sub-place5:
• ‘Metros’: the eight largest urban agglomerations in South Africa. These are defined as sub-
places that fall within metropolitan municipalities that are also classified as urban according
to StatsSA. The eight metros are Johannesburg, Cape Town, eThekwini, Ekurhuleni, Tshwane,
Nelson Mandela Bay, Buffalo City and Mangaung.
• ‘Cities/towns’: smaller cities and towns. These are technically defined as sub-places that fall
within urban areas according to StatsSA but excluding those within metropolitan municipalities.
• ‘Rural areas’: rural and either commercial farms or land governed by traditional authorities. These
cover the rest of the country and are areas classified as ‘rural’ according to StatsSA.
Figure 1 shows the distribution of the population in these categories as applied to the Community
Survey (CS) 2016, NIDS W5 and NIDS-CRAM W1 and W2.6 The CS 2016 provides a useful baseline
with which to appraise the representivity of the estimates in the remaining surveys as a much larger
household survey. The CS2016 and NIDS W5 are very closely aligned which gives confidence to the
original NIDS design. Further to this, age, gender and demographic splits between these surveys
are also a very close match (see Appendix B).
However, both the NIDS-CRAM W1 and W2 appear to significantly underestimate the percentage of
individuals living in rural areas whilst overestimating those living in smaller cities/towns. Only 17.6%
of the adult population was living in rural areas in NIDS-CRAM W1 compared to more than 30% in
the NIDS W5 and CS2016. The difference is found in the proportion of individuals living in cities/
towns which is estimated at 45% of the adult population in NIDS-CRAM W1 compared with 27% of
the population in both the NIDS W5 and CS2016.
3 7,073 and 5,676 individuals were re-interviewed in waves 1 and 2 of the NIDS-CRAM respectively.
4 The NIDS 2008 sampling frame was limited to only 400 clusters nationally as opposed to more than 3,000 for similar sized surveys such
as the QLFS which was explicitly intended to be representative of provinces and metropolitan municipalities.
5 TheNIDS-CRAMisatelephonicsurveywhichmeantthatlocationinformationwasderivedfromaperson’sself-reportedplaceof
residence and linked back to StatsSA classification of sub-places. The NIDS-CRAM:W2 had 155 missing location responses however
this was reduced to 39 by imputing the location of the place of residence from NIDS:W5 on the condition that the individual reported
that they were living in the same residence as in NIDS.
6 Weconstructoursampleof‘locationaltypes’asasetof‘cross-sections’ratherthanasa‘balancedpanel’acrosstheNIDS:W5,NIDS-
CRAM:W1 and NIDS-CRAM:W2 surveys. We do this in order to maximize on our sample size (a large number of observations are
dropped when constructing a balanced panel) which increases the precision of our estimates.
Embargoed until 30 September 12:00 (noon). Do not quote, cite or share.
6 | The Uneven Geography of the COVID-19 Crisis
Despite the large mismatch, the demographic profiles for each spatial category across surveys
are a surprisingly close match which might suggest that the larger (smaller) population sizes in the
NIDS-CRAM in cities/towns (rural areas) had less impact on who was sampled within each area (see
Appendix B). In light of these concerns, we refrain from reporting any of our findings using absolute
totals and instead focus on the proportions or percentages between waves which are not sensitive
to fluctuations in the size of the population. We reiterate that our findings are exploratory and any
hard facts would need to be backed up with evidence from other sources.
Figure 1: Location type: metros, cities/towns and rural areas
0,0 0,8
40,8 39,9 37,3 36,4
27,0 27,4
45,0 38,7
32,3 32,7
17,6 24,2
0
10
20
30
40
50
60
70
80
90
100
CS2016 NIDS W5 NIDS-CRAM W1 NIDS-CRAM W2
Percent (%)
Missing Metros Cities/towns Rural
Source: Communit y Survey 2016, NIDS W5, NIDS-CRAM W1 and 2
Notes: CS 2016 and NIDS:W5 estimates are restricted to the adult population to be comparable with the NIDS-CRAM. See table A1 for
sample size and 95% confidence intervals. N = 5676 for NID S-CRAM W2. The data are weighted.
The second locational typology focuses on differences within urban areas. This is based upon a
respondent’s perception of their neighbourhood type and limited to a sub-sample of individuals
who were already located in urban areas according to their reported place of residence. Data on
household location in face-to-face interviews is usually based upon geo-coordinates which are
captured directly at the time of the interview. In light of the fact that the NIDS-CRAM was a telephonic
survey, respondents were also asked about how they perceived their area type which we manipulate
to produce a four-fold classification7:
• ‘Suburbs’: which residents identified as “formal residential” areas. This category could include a
variety of urban neighbourhood types including individuals living in apartment blocks through to
affluent households in low-rise suburbia.
• ‘Townships’: which residents identified as “townships”. Former black townships have been slow
to transform and many experience inadequate infrastructure and low levels of formal economic
activity.
• ‘Shack-dwellers’: which residents identified as “informal settlements”. We augment this category
to include residents who recorded living in “an informal house like a shack” and hence explicitly
include other forms of informality such as backyard shacks.
• ‘Peri-urban’: which residents identified as a range of low-density categories including “farm”,
“small holding” or “traditional”. We have imposed the condition that a person’s sub-place was
classified as urban by StatsSA and hence these neighbourhoods would be in or around the urban
fringe.
It should be noted that our urban typologies are not technically representative of different urban
areas themselves, but of urban resident’s perceptions of neighbourhood types. There is bound to
7 The NIDS-CRAM questionnaire asks respondents to answer the following: “Which of the following best describes the area you live in
now: Traditional, informal settlement, township, formal residential, farm, or small holding?”
Embargoed until 30 September 12:00 (noon). Do not quote, cite or share.
7 | The Uneven Geography of the COVID-19 Crisis
be some discrepancy between how individuals perceive their neighbourhood and how this would be
defined through geo-referencing. There is no way of independently checking the level of correlation.
Nonetheless, it is still interesting to consider how socio-economic outcomes map against these self-
identified urban types.
Figure 2: Urban type: Suburbs, townships, shack-dwellers and peri-urban
1,7
31,5
37,6
13,4
15,9
0
10
20
30
40
50
60
70
80
90
100
NIDS-CRAM: W2
Percent (%)
Peri-urban
Shack-dwellers
Townships
Suburbs
Missing
Source: NIDS-CRAM W2
Notes: See table A 2 for sample sizes and 95% confidence intervals. N = 3,8 51. The data are weighted.
Figure 2 shows the distribution of urban types derived from the NIDS-CRAM W2. We are not able
to recreate these categories for previous waves of the survey because respondents were only
asked about their perception of their location type in the NIDS-CRAM W2. Instead, we create an
urban panel incorporating socio-economic data from previous waves based upon self-identified
neighbourhood type from W2 – even if some people had changed location between waves.8
3. The impact of COVID-19 on different places
3.1. Labour Market
This section considers the impact of the pandemic on employment conditions in different parts
of the country, starting with the contrasts between cities and rural areas, and then examining the
differences within cities.
Metros vs Cities/towns vs Rural areas
An important finding from the NIDS-CRAM W1 was the large fall in the employment-to-population
ratio (which we refer to as ‘total employment’ in the rest of the paper) of approximately 15% between
February-April 2020, which amounted to 3 million jobs. If workers who were absent from work (or
reported earning zero income) are not counted among the employed, then the reduction in total
employment was even more dramatic, falling by 33% (Ranchhod and Daniels, 2020; Spaull et al,
2020).
8 The large majority of respondents did not change their place of residence between waves: 82% of NIDS-CRAM respondents in
W2 reported that they were in the same dwelling as in W1, while 73% of respondents reported living in the same dwelling as when
interviewed in 2017 (i.e. NIDS W5).
Embargoed until 30 September 12:00 (noon). Do not quote, cite or share.
8 | The Uneven Geography of the COVID-19 Crisis
Figure 3 shows that these national figures mask important spatial differences. First, employment
levels were already much lower in rural areas than in cities, as measured in February before the
crisis hit. The labour markets of the metros were much stronger with 60% of all adults (aged 18 years
and older) holding a job, compared with 43% in rural areas.
Second, the impact of the crisis was severe across both urban and rural areas. All regions initially
experienced a strong decline in total jobs between February-April of roughly 15%. This reduction
was even larger if furloughed workers (i.e. those who were employed but received no income) are
excluded at approximately 20%.
Third, there was little recovery in employment after the hard lockdown between April-June, as
restrictions on activity were eased. It appears that there was a slight upturn in the number of people
in paid employment in the metros and cities/towns, but this seems to have been mostly as a result of
furloughed workers being brought back onto the payroll or returning to self-employment. The slight
reduction in rural areas was not statistically significant.
Figure 3: Geographic type: Percentage employed or furloughed (adults 18 years +)
57,2
46,0 48,7 46,3
35,8 40,1 41,5
33,7 31,8
3,0
5,0 3,0 3,0
5,1
2,9 1,7
4,3 3,1
0
10
20
30
40
50
60
70
Feb April June Feb April June Feb April June
Metros Cities/towns Rural
Percent (%)
Employed Furloughed (no pay)
Source: NIDS-CRAM W1 and W2
Notes: The sample is adults aged 18 years and older. Furloughed workers had a job but reported zero earnings. See table A3 for 90%
confidence intervals. The data are weighted.
The net employment losses in figure 3 conceal the extent of job churn that occurred as people moved
into and out of employment. The graph on the left of figure 4 suggests that as many as 22%, 27%
and 29% of the employed lost their jobs in metros, cities/towns and rural areas respectively between
February-April. This was the most stringent period of the lockdown and government relief was still
being planned. The period April to June has seen these losses begin to moderate in both metros
and cities/towns (declining by about 4-6 percentage points). However, rural areas still experienced
job losses of a similar intensity (a slight recorded reduction was not statistically significant).
The graph on the right of figure 4 shows that some people did manage to get jobs during the severe
lockdown, although they were far fewer than those who lost jobs. Job gains were similar across
all types of location, with roughly 10% of adults who were previously not employed (i.e. either
unemployed or not economically active) finding employment between February-April. However, a
disparity began to emerge between locations in the following period (April-June), where the rate
of hiring in the metros increased by 20%, by 16% in cities/towns, but did not change much in rural
areas. Overall, rural areas experienced both a larger fall in jobs and a greater lag in recovery than
the cities.
Embargoed until 30 September 12:00 (noon). Do not quote, cite or share.
9 | The Uneven Geography of the COVID-19 Crisis
Figure 4: Urban type: Labour market churn
0
5
10
15
20
25
30
35
Metros Cities/towns Rural
Percent (%)
% employed who lost a job
Feb-Apr Apr-Jun
0
5
10
15
20
25
30
35
Metros Cities/towns Rural
Percent (%)
% not employed who gained a job
Feb-Apr Apr-Jun
Source: NIDS-CRAM W1 and W2
Notes: The sample is adults aged 18 years and older. See table A4 for 90% confidence inter vals. Self-identified neighbourhood type is
defined in W2. The data are weighted.
Figure 5 shows the ‘headline’ rate of unemployment across the metros, cities/towns and rural areas.
The base period is taken from the NIDS:W5 in 2017 where rates of unemployment were surprisingly
low.9 This probably mean that the unemployment rate is somewhat understated in the NIDS compared
with nationally representative surveys like the Labour Force Surveys over the same time period
(Ardington, 2020; Ranchhod & Daniels, 2020). Nevertheless, the trend over time is very striking as
are the differences in joblessness between cities and rural areas. In 2017, rural unemployment was
at least 10 percentage points higher than in the cities.
These disparities seem to have widened since the onset of COVID-19. The rate of unemployment in
rural areas shot up to 48% in April 2020 and 52% in June. Unemployment in both cities/towns and
metros also increased to 45% and 43% in cities/towns and 37% and 35% in metros in April and June
respectively. The gap in the unemployment rate between metros and rural areas has therefore risen
from 10 to 18 percentage points over the period. The rate of unemployment in the cities/towns falls
between the metros and rural areas.
Figure 5: Geographic type: Rate of unemployment
0
10
20
30
40
50
60
2017: NIDS W5 April 2020: NIDS-CRAM W1 June 2020: NIDS-CRAM W2
Percent (%)
Metros Cities/ t owns Rural
Source: NIDS W5, NIDS- CRAM W1 and W2
Notes: Expanded rate of unemployment (i.e. includes the non-searching unemployed). The sample is adults aged 18 years and older.
See table A5 for 90% confidence intervals. The data are weighted.
9 There are bound to be some changes in labour market status in the years between 2017 and February 2020 – which would have been
the appropriate pre-crisis baseline. Nevertheless, we do not suspect that changes between 2017 and 2020 were dramatic in light of
what is reported about changes to the rate of unemployment for this period in the QLFS.
Embargoed until 30 September 12:00 (noon). Do not quote, cite or share.
10 | The Uneven Geography of the COVID-19 Crisis
Suburbs vs Townships vs Shack dwellers vs Peri-urban areas
South African cities also contain striking disparities which may be amplified by the COVID-19 crisis.
People living in different neighbourhoods experience very different types of vulnerability, opportunity
and resilience, which warrant careful scrutiny.
Figure 6 shows how rates of employment differed across people living in cities in NIDS-CRAM W2
according to their neighbourhood type. Adults who perceived themselves to be living in the suburbs
had the highest total employment levels in February at close to 60%, followed by townships (55%)
and peri-urban areas (49%). Employment levels are difficult to estimate for shack-dwellers (due to
the size of measurement error), but they may have been as high as the suburbs before the crisis
(probably reflecting many informal livelihoods).
All areas experienced a major shock to employment between February-April. However, the
reduction in jobs was largest among shack-dwellers (falling by 27%), followed by peri-urban areas
(20%), townships (15%) and the suburbs (12%). These differences are even bigger if furloughed
workers are removed, with employment falling by as much as 36% among shack dwellers, 23-24%
in townships and peri-urban areas, but only 14% in the suburbs. The extreme fall in employment in
shack settlements probably reflects their dependence on informal jobs shut down during level 5 of
the lockdown.
Shack dwellers experienced some apparent recovery in employment by June (partly through
furloughed workers going back onto the payroll) as lockdown restrictions eased. Yet, total employment
was still 10 percentage points lower than in February. Suburban residents also showed some signs
of improvement, while conditions in the townships and peri-urban areas did not. (Note that none of
these changes were large enough to be statistically significant)
Figure 6: Urban type: Percentage employed or furloughed (adults 18 years and older)
57,5
49,4 51,8 51,3
39,2 42,1
59,2
38,1
49,2 44,7
34,6 32,7
1,9
3,1 3,6 3,4
7,2 2,7
2,7
7,1
2,2
3,8
4,4 2,9
0
10
20
30
40
50
60
70
Feb April June Feb April June Feb April June Feb April June
Suburbs Townships Shack dwellers Peri urban
Percent (%)
Employed Furloughed (no pay)
Source: NIDS-CRAM W1 and W2
Notes: The sample is adults aged 18 years and older. Furloughed workers had a job but reported zero earnings. See table A3 for 90%
confidence intervals. Self-identified neighbourhood type is defined in W2. The data are weighted.
The total number of households impacted by employment shifts was even larger when considering
job losses and job gains (i.e. the extent of labour market churn). Figure 7 suggests that roughly 1 in 4
people in townships or peri-urban areas who were employed in February lost their job by April. This
was as high as 1 in 3 among shack-dwellers. Jobs in the suburbs contracted by only about 17%. The
corresponding job gains between February-April among those previously without employment (i.e.
either unemployed or not economically active) were consistently low at roughly 10% in all places.
Embargoed until 30 September 12:00 (noon). Do not quote, cite or share.
11 | The Uneven Geography of the COVID-19 Crisis
In the subsequent period (April-June) job losses appeared to moderate across all areas (with the
exception of peri-urban) although they were still as high as 1 in 5 people in townships, shack-
dwellers and peri-urban areas. The job gains between April-June were largest among shack
dwellers followed next by the suburbs. The overall message is that a large proportion of individuals
experienced the stress of job churn (both losing and gaining employment) during the COVID-19
crisis. Shack dwellers were the most vulnerable (with huge losses and gains), while suburban
residents experienced the greatest stability.
Figure 7: Urban type: Labour market churn
0
5
10
15
20
25
30
35
40
45
Suburbs Townships Shack dwellers Peri urban
Percent (%)
% employed who lost a job
Feb-Apr Apr-Jun
0
5
10
15
20
25
30
35
40
45
Suburbs Townships Shack dwellers Peri urban
Percent (%)
% not employed who gained a job
Feb-Apr Apr-Jun
Source: NIDS-CRAM W1 and W2
Notes: The sample is adults aged 18 years and older. See table A4 for 90% confidence inter vals. Self-identified neighbourhood type is
defined in W2. The data are weighted.
The impact of the crisis on the labour market of cities can be seen in changes to the rate of
unemployment in all neighbourhoods (figure 8). Rising unemployment rates were far larger for
people living in peri-urban areas (up by 30 percentage points), compared with townships (by 24
percentage points), shack-dwellers (by 20 percentage points) and the suburbs (13 percentage
points). The direction of the trends and differences in rates between location types clearly illustrate
how suburban residents were less affected by the crisis. Shack-dwellers also showed some resilience
ending the period with an apparent lower rate of unemployment than peri-urban areas (and the
same as townships). However, shack dwellers also had to contend with considerable volatility as
unemployment soared to 50% in April before declining again in June. The situation is still dire
everywhere, with an unemployment rate of 24% in the suburbs, 42% among shack-dwellers, 43% in
the townships and 52% in the peri-urban areas in June.
Figure 8: Urban type: Rate of unemployment
0
10
20
30
40
50
60
2017: NIDS W5 April 2020: NIDS-CRAM W1 June 2020: NIDS-CRAM W2
Percent (%)
Suburbs Townships Shack dwellers Peri urban
Source: NIDS W5, NIDS- CRAM W1 and W2
Notes: Expanded rate of unemployment (i.e. includes the non-searching unemployed). The sample is adults aged 18 years and older.
See table A5 for 90% confidence intervals. Self-identified neighbourhood type is defined in W2. The data are weighted.
Embargoed until 30 September 12:00 (noon). Do not quote, cite or share.
12 | The Uneven Geography of the COVID-19 Crisis
3.2. Social support
This section considers the welfare safety net provided by government grants and examines the
extent to which it cushioned households in different places from the economic shock of COVID-19.
Metros vs Cities/towns vs Rural areas
It is well known that rural communities in South Africa are generally far more reliant on government
grants than urban communities (Makgetla, 2018; Turok, 2014b, 2018). This is because rural
economies are much weaker, unemployment is much higher and people are generally poorer. The
NIDS-CRAM Wave 2 asked respondents about their various sources of household income. Figure
9 shows that nearly three out of five rural respondents (59%) received social grants in June 2020,
compared with less than one in two residents in cities/towns (47%) and less than one in three metro
residents (32%). In other words, grants were the main source of livelihood protection in rural areas,
where the proportion of residents receiving them was nearly double the proportion in the metros.
These figures reflect the sum of ‘grants only’ and ‘earnings plus grants’.
At the same time, rural respondents were far less likely to be in paid employment. Less than one
in three rural residents (32%) had work-related earnings, compared with less than half of residents
in cities/towns (44%) and more than half of metro residents (54%). These figures reflect the sum
of ‘earnings only’ and ‘earnings plus grants’. Therefore, government grants have clearly helped to
compensate rural areas for their fragile local economies and the shortfall in employment. Without this
form of income support, the economic gap between cities and rural areas would have been much
larger. However, increasing reliance on grants is also a source of vulnerability for these communities
if one or more of these cash transfers is withdrawn.
Figure 9: Geographic type: Sources of household income, June 2020
23,2
35,9
48,4
10,1
9,4
7,7
44,6
32,6
21,3
9,1
11,4
10,2
7,5
5,8
6,4
5,5
5,0
6,0
0% 10%20%30%40%50%60%70%80%90%100%
Metros
Cities/towns
Rural
Grants only Other only Earnings only Earnings & grants Other combination No income
Source: NIDS-CRAM: W2
Notes: See table A6 for 90% confidence intervals. The data are weighted.
The government introduced the special COVID-19 distress relief grant in June specifically to target
adults who had no other source of income, such as working-age unemployed men. Over 5 million
people currently benefit from a grant of R350 per month. The NIDS-CRAM wave 2 survey asked
respondents about their receipt of this grant. Figure 10 shows that one in three adults in rural areas
(33%) reported that someone in their household had received a COVID-19 grant, compared with less
than one in four in cities/towns (24%) and just over one in five in the metros (21%). These proportions
are lower than for other grants, and the differences between cities and rural areas are narrower,
suggesting that the COVID-19 grant is benefiting groups that have not qualified for government
support before. The higher proportion of rural beneficiaries is consistent with the higher rate of
unemployment in the countryside. However, this poses risks for poor communities and groups such
as unemployed young men because the COVID-19 grant was only envisaged to provide temporary
relief and is due to be withdrawn at the end of October.
Embargoed until 30 September 12:00 (noon). Do not quote, cite or share.
13 | The Uneven Geography of the COVID-19 Crisis
Figure 10: Geographic type: Percentage adults reporting their household received a COVID-19 SRD Grant
0
5
10
15
20
25
30
35
40
Metros Cities/towns Rural
Percent (%)
Source: NIDS-CRAM: W2
Notes: See table A7 for 9 0% confidence intervals. The data are weighted.
Suburbs vs Townships vs Shack dwellers vs Peri-urban areas
Turning to the differences within cities, peri-urban areas were more likely to benefit from government
grants than people in the townships, shack areas and suburbs (f i g u r e 11 ). More than half of peri-
urban respondents (54%) lived in a household which received social grants in June 2020, compared
with less than half of township respondents (45%), two in five shack dwellers (40%) and one in four
suburban residents (26%). In other words, the proportion of peri-urban residents receiving grants
was more than double the proportion in the suburbs. These intra-urban disparities are wider than
between rural areas and metros (see figure 9). These estimates reflect the sum of ‘grants only’ and
‘earnings plus grants’.
At the same time, peri-urban residents were far less likely to be in employment. Just over one in
three peri-urban residents (35%) lived in households with work-related earnings, compared with
less than half of shack dwellers (45%), about half of township respondents (49%) and nearly three
in five suburban residents (57%). The implication is that government grants have helped to offset
unemployment and poverty in townships, shack areas and peri-urban areas.
Figure 11: Urban type: Sources of household income, June 2020
21,8
30,7
26,4
43,5
12,7
7,8
9,5
7,9
52,7
35,3
31,4
25,0
4,5
14,0
14,0
10,4
4,3
7,7
8,7
7,6
4,1
4,5
10,1
5,6
0% 10%20%30%40%50%60%70%80%90%100%
Suburbs
Townships
Shack dwel lers
Peri urban
Grants only Other only Earnings only Earnings & grants Other combination No income
Source: NIDS-CRAM: W2
Notes: See table A6 for 90% confidence intervals. The data are weighted.
Embargoed until 30 September 12:00 (noon). Do not quote, cite or share.
14 | The Uneven Geography of the COVID-19 Crisis
Figure 12 shows that among urban residents, 29% of peri-urban adults lived in a household where
someone had received the COVID-19 grant. This was 27% for residents in townships, 18% among
shack dwellers and 16% for residents living in suburban households. The proportion of shack
dwellers receiving the COVID-19 relief and other social grants is surprisingly low considering the
level of poverty and distress in these areas. Further research is required to explain the reasons for
this. It may, for example, reflect the fact that many shack dwellers do not have a proper address
(street name and house number) for claiming grants. It could also reflect the disproportionate
number of foreign nationals living in shacks.
Figure 12: Urban type: Percentage adults reporting their household received a COVID-19 SRD Grant
0
5
10
15
20
25
30
35
Suburbs Townships Shack dwellers Peri urban
Percent (%)
Source: NIDS-CRAM: W2
Notes: See table A7 for 9 0% confidence intervals. The data are weighted.
To summarise this section, government grants have clearly helped to compensate rural areas,
townships and informal settlements for their relatively weak economic situation compared with
metros, and especially the suburban areas of cities. The level of these grants is generally low,
so they are not a substitute for productive employment. They help to alleviate poverty rather than
providing a pathway to lift people out of poverty. Temporary grants also create vulnerabilities in poor
communities if they are withdrawn prematurely.
3.3. Food poverty
This section considers the proportion of households in different places that have experienced
financial hardship and food insecurity (hunger).
Metros vs Cities/towns vs Rural areas
The NIDS-CRAM Wave 1 data showed that 47% of adults throughout the country reported that their
household had run out of money to buy food in April 2020. The graph on the left of figure 13
shows the differences between metros (44%), cities/towns (48%) and rural areas (52%). The figures
are very high everywhere, but rural households were clearly finding it rather more difficult than their
counterparts in the metros. Figure 13 also shows that the problem was much worse in April 2020
than it was in 2016, when the last Community Survey was conducted.10 In 2016, the breakdown was
metros (16%), cities/towns (21%) and rural areas (28%). By April 2020, the gap between metros
and rural areas may have narrowed, but roughly twice as many respondents were experiencing this
hardship as in 2016.
10 The Community Survey 2016 asked individuals whether their household had run out of money to buy food in the past 12 months
whereas the NIDS-CRAM only asked about the past month. The much wider timeframe in the Community Survey implies that our 2016
baseline would be even lower if individuals had been asked to report about the past month.
Embargoed until 30 September 12:00 (noon). Do not quote, cite or share.
15 | The Uneven Geography of the COVID-19 Crisis
Figure 13: Geographic type: Hunger and food poverty
0
10
20
30
40
50
60
Metros Cities/towns Rural
Percent (%)
% ran out of money to buy food
CS2016* W1 : April 2020 W2: June 2020
0
5
10
15
20
25
30
35
Metros Cities/towns Rural
Percent (%)
% hungry in past 7 days
W1: May/June 2020 W2: July/August 2020
Source: NIDS-CRAM W1 and W2, Community Survey 2016
Notes: *The CS2016 asks individuals if their household had run out of money to buy food in past 12 months. See table A8 for 90%
confidence intervals. The data are weighted.
The NIDS-CRAM Wave 2 data shows that the proportion of respondents that said they had run out
of money to buy food in June 2020 had fallen to 35% in the metros, 37% in the cities/towns and 40%
in the rural areas. In other words, the numbers of respondents that had run out of money to buy food
had fallen by about a fifth in all areas compared with April. This is a noticeable improvement, and is
likely to reflect the extra government support through social grants. Nevertheless, more than a third
of respondents were still struggling to buy food at some point during the month of June 2020. This
is significantly higher than in 2016, indicating a persistent problem.
The NIDS-CRAM Wave 1 data also showed that one in five respondents (21%) reported that someon e
in their household had gone hungry in the last seven days. This was referring to when the survey
was done in May/June – after many households had started receiving the government grants. The
graph on the right of figure 13 shows the breakdown between metros (17%), cities/towns (24%) and
rural areas (29%). There is a large difference between the big cities and rural areas, indicating the
much higher incidence of food poverty in the countryside. Nearly one in three rural respondents
said someone had gone hungry in May/June, compared with one in six metro respondents.
The Wave 2 data shows that the proportion of respondents saying that someone in their household
had gone hungry in the last seven days (in July/August) had fallen to 13% in the metros, 16% in the
cities/towns and 20% in the rural areas (Figure 13). This was a significant fall in hunger of close to
10 percentage points between May/June and July/August for rural areas. This may reflect the fact
that a higher proportion of rural households benefit from social grants, so rural areas would benefit
disproportionately from a top-up to these grants. The metros also contain more foreign migrants
than rural areas, who do not qualify for government grants.
Suburbs vs Townships vs Shack dwellers vs Peri-urban areas
The graph on the left of figu re 14 shows the proportion of respondents that said that their household
had run out of money to buy food in April 2020, broken down between suburbs (31%), townships
(48%), shack areas (61%) and peri-urban areas (46%). The difference between shack-dwellers and
suburban residents was much larger than between metros and rural areas (shown in figure 13).
Indeed, shack-dwellers were noticeably worse off than rural respondents (notwithstanding fairly
large measurement errors for shack areas). The problem is linked to the fact that far fewer shack
dwellers received government grants than rural residents, while being more dependent on precarious
forms of employment, so they were disproportionately harmed by the lockdown and restrictions on
economic activity.
Embargoed until 30 September 12:00 (noon). Do not quote, cite or share.
16 | The Uneven Geography of the COVID-19 Crisis
Figure 14: Urban type: Hunger and food poverty
0
10
20
30
40
50
60
70
80
Suburbs Townships Shack dwellers Peri urban
Percent (%)
% ran out of money to buy food
W1: April 2020 W2: June 2020
0
5
10
15
20
25
30
35
40
Suburbs Townships Shack dwellers Peri urban
Percent (%)
% hungry in past 7 days
W1: May/June 2020 W2: July/August 2020
Source: NIDS-CRAM W1 and W2, Community Survey 2016
Notes: Self-identified neighbourhood type is defined in W2. The data are weighted.
The Wave 2 data shows that the proportion of respondents whose household had run out of money
to buy food in June 2020 had fallen to 24% in the suburbs, 40% in the townships, 50% in the shack
areas and 38% in the peri-urban areas. These reductions compared with April are broadly similar in
size across the different locations. This was a notable improvement, although the gap between the
shack-dwellers and suburban residents was still extremely large. Shack-dwellers also continued to
be worse off than rural residents.
The graph on the right of figure 14 shows the proportion of respondents who said that someone
in their household had gone hungry in the last seven days (in May/June), broken down between
suburbs (11%), townships (22%), shack areas (32%) and peri-urban areas (25%). The difference
between shack-dwellers (one in three) and suburban residents (one in nine) is very striking, and
larger than the gap between metros and rural areas. This indicates the high incidence of food
poverty among shack dwellers.
The Wave 2 data shows that the proportion of respondents saying that someone in their household
had gone hungry in the last seven days (in July/August) had fallen to 7% in the suburbs, 16% in
the townships, 22% in the shack areas and 19% in the peri-urban areas. There was a reduction all
round, although the gap between the shack-dwellers and suburban residents remained extremely
large.
4. Conclusion
South Africa is one of the most unevenly developed countries in the world with stark differences
in life chances between locations. COVID-19 has exposed the unequal living conditions and
vulnerabilities of different communities very visibly. Overcrowded and under-serviced settlements
have been particularly at risk from the spread of the coronavirus and could suffer again from any
resurgence. They have also been ravaged by the economic effects of enforced physical distancing
through hard lockdowns and restrictions on travel to work.
Evidence from the NIDS-CRAM surveys indicates that COVID-19 has magnified pre-existing
social and economic divisions both (i) between cities and rural areas, and (ii) between suburbs
and townships/informal settlements within cities. The metros proved to be more resilient than rural
areas and smaller cities/towns, perhaps because of their stronger institutions and higher levels of
human capital. The chasm within cities between suburbs and informal settlements has proved to be
even larger. Shack dwellers were extremely vulnerable to the shutdown and constraints on informal
enterprise.
Government social grants have helped to offset the economic disparities between places and
Embargoed until 30 September 12:00 (noon). Do not quote, cite or share.
17 | The Uneven Geography of the COVID-19 Crisis
compensate the residents of rural areas and townships for their high levels of unemployment and
poverty. However, only a minority of households have benefited from the special COVID-19 relief
grant. As a result, the incidence of hunger is still much higher in informal settlements, townships and
rural areas than in suburban areas. Conditions in these places were bad enough before the crisis,
but now there is considerable further ground to make up. The imminent withdrawal of temporary
cash transfers could be a serious setback for poor and vulnerable communities reliant on social
support, especially for groups who did not qualify for grants before, such as unemployed young
men.
Looking ahead, there are several implications for the government’s response to the crisis.
First, it is important to recognise that different parts of the country face different challenges. Treating
unequal places in the same manner won’t narrow the gap between them. Blanket national policies
and actions are insensitive to these variations and can have unintended consequences in amplifying
inequalities. National programmes need complementary efforts to boost jobs and livelihoods in
lagging urban and rural areas. This means targeting places as well as people in tackling poverty
and unemployment.
Second, the findings suggest that a special focus on informal settlements and backyarders is
warranted because they have been hardest hit by the crisis and face the most uncertain prospects
of recovery. Pre-existing conditions were miserable enough, but now the task of upgrading is that
much more urgent. Density and overcrowding are symptoms of poverty, rather than fundamental
causes of contagion. Congested settlements need to be de-risked by converting population density
into more resilient ‘economic density’. Investment in buildings and infrastructure should lie at the
heart of settlement upgrading, accompanied by the development of skills, jobs, incomes and
more functional environments for enterprise and economic development to thrive. De-densification
through relocation is a distraction when resources and energy should be focused on improving
basic services and reconfiguring layouts.
Third, wide disparities between urban and rural areas will continue to spur migration out of the
countryside and into cities in search of a better life. The process cannot be suppressed. Rather,
local authorities should work with national and provincial governments in recognising people’s
constitutional right to freedom of movement and support the provision of basic services and shelter
in the cities. Preparing land for human settlement in anticipation of urbanisation is more cost-effective
than trying to retrofit infrastructure into dense informal settlements and fend off unauthorised land
occupations from disaffected backyarders.
Lastly, more effort is required to improve the quality of information and intelligence on local
economic and health conditions. The focus of the COVID-19 analysis and response has been at
the national and provincial levels, yet the transmission mechanisms are essentially local, and the
public health, economic and social impacts have also been highly localised. Stronger evidence and
research would improve understanding of these dynamics and help to empower local institutions
and partnerships to develop constructive responses. This would go some way to help kick-start the
recovery and realise the potential of all places.
Embargoed until 30 September 12:00 (noon). Do not quote, cite or share.
18 | The Uneven Geography of the COVID-19 Crisis
REFERENCES
Ardington, C. (2020) NIDS-CRAM Wave 1 Data Quality. NIDS-CRAM Working Paper.
Branson, N and Wittenberg, M. (2019) Longitudinal and Cross sectional Weights in the NIDS data
1-5. National Income Dynamics Study Technical Paper no 9. Cape Town: Southern African Labour
and Development Research Unit.
Kerr, A., Ardington, C. and Burger, R. (2020). Sample design and weighting in the NIDS-CRAM
survey. NIDS-CRAM Working Paper.
Makgetla, N. (2018) The systemic underpinnings of inequality in South Africa. In Smith, M. N. (ed)
Confronting Inequality: The South African Crisis. Jacana Media. 73-105.
Martin, R. L. (2018). Shocking Aspects of Regional Development: Towards an Economic Geography
of Resilience. In G. L. Clark, M. P. Feldman, M. S. Gertler, & D. Wójcik, (Eds.) (Vol. 1). Oxford
University Press. http://doi.org/10.1093/oxfordhb/9780198755609.013.43
Motlanthe High Level Panel (2017) Report on the assessment of key legislation and the acceleration
of fundamental change, Executive Summary, available from: https://www.parliament.gov.za/high-
level-panel
Ranchhod, V. and Daniels, R. (2020) Labour market dynamics in South Africa in the time of COVID-19:
Evidence from wave 1 of the NIDS-CRAM survey, NIDS-CRAM Working Paper, Southern Africa
Labour and Development Research Unit, University of Cape Town.
Schotte, S., Zizzamia, R. and Leibbrandt, M. (2017) Social stratification, life chances and vulnerability
to poverty in South Africa, SALDRU Working Paper 208, University of Cape Town.
Seeliger, L and Turok, I. (2014) ‘Averting a downward spiral: Building resilience in informal urban
settlements through adaptive governance, Environment and Urbanisation, 26(1), pp.184-199.
Spaull et al. (2020) NIDS-CRAM Wave 1 Synthesis Report: Overview and Findings. NIDS-CRAM
Working Paper.
Statistics South Africa (2017) Poverty trends in South Africa: An examination of absolute poverty
between 2006 and 2015.
Todes, A. and Turok, I. (2018) ‘Spatial Inequalities and Policies in South Africa: Place-based or
People-centred?’, Progress in Planning. http://dx.doi.org/10.1016/j.progress.2017.03.001
Turok, I. (2014a) ‘South Africa’s tortured urbanisation and the complications of reconstruction’, in
Martine, G. and McGranahan, G. (eds) Urban Growth in Emerging Economies: Lessons from the
BRICS, London: Routledge.
Turok, I. (2014b) ‘The resilience of South African cities a decade after local democracy’, Environment
and Planning A, 4 6 , pp.74 9 -76 9.
Turok, I. (2016) ‘South Africa’s new urban agenda: Transformation of compensation?’, Local Economy,
31(1), pp.9-27.
Turok, I. (2018) “Worlds apart: spatial inequalities in South Africa”. In: Smith, M.N. (ed). Confronting
Inequality: The South African Crisis. Johannesburg: Jacana Media. 129-151
Turok, I. and Borel-Saladin, J. (2016) ‘Backyard shacks, informality and the urban housing crisis in
South Africa: Stopgap or prototype solution?’, Housing Studies. 31(4), pp.384-40 9.
Embargoed until 30 September 12:00 (noon). Do not quote, cite or share.
19 | The Uneven Geography of the COVID-19 Crisis
Visagie, J. and Turok, I. (2020) ‘Getting density to work in informal settlements in Africa’, Environment
and Urbanisation, online. doi/10.1177/0956247820907808
World Bank. (2018). “An Incomplete Transition: Overcoming the Legacy of Exclusion in South Africa.”
Washington DC: World Bank. http://doi.org/10.1596/29793
Embargoed until 30 September 12:00 (noon). Do not quote, cite or share.
20 | The Uneven Geography of the COVID-19 Crisis
Appendix A
Table A1: Geographic type: Metros, cities/towns and rural
CS2016 (18+) NIDS W5 (18+) NIDS-CRAM W1 NIDS-CRAM W2
Estiamte Lower Upper N Estiamte Lower Upper N Estiamte Lower Upper N Estiamte Lower Upper N
Missing 0.04 0.01 0.18 3 0.78 0.41 1.49 39
Metro 40.77 40.7 40.85 765236 39.92 34.64 45.44 5990 37.33 33.39 41.45 1627 36.39 32.1 40.91 1223
Other
urban 26.96 26.89 27.02 581066 27.39 23.18 32.05 9046 45.01 41.6 48.48 3836 38.65 35.16 42.26 2628
Rural 32.27 32.2 32.34 819318 32.69 28.4 37.29 11986 17.61 15.61 19.81 1607 24.18 21.3 27.32 1786
Total 100 2165620 100 27022 100 7073 100 5676
Source: Communit y Survey 2016, NIDS W5, NIDS-CRAM W1 and W2; 95% confidence inter vals
Table A2: Urban type: Suburbs, townships, shack-dwellers and peri-urban
Estimate Lower Upper N
Missing 1.72 1.07 2.75 76
Formal residential 31.5 28.1 35.1 949
Township 37.57 33.58 41.74 1392
Shack-dweller 13.35 11.09 15.99 487
Peri-urban 15.86 13.53 18.51 947
Total 100 3851
Source: Communit y Survey 2016, NIDS W5, NIDS-CRAM W1 and W2; 95% confidence inter vals
Embargoed until 30 September 12:00 (noon). Do not quote, cite or share.
21 | The Uneven Geography of the COVID-19 Crisis
Table A3: Employment status
Feb April June
Not
Employed Employed Furloughed Total Not
Employed Employed Furloughed Total Not
Employed Employed Furloughed Total
Geo type
Metro % 39.89 57.17 2.95 100 49.06 45.97 4.98 100 48.29 48.70 3.01 100
CI [36.88,42.97] [54.27,60.02] [2.01,4.30] 100 [45.95,52.17] [43.04,48.93] [3.88,6.36] 100 [44.17,52.44] [44.50,52.93] [2.00,4.51] 100
Other Urban % 50.74 46.29 2.97 100 59.05 35.82 5.13 100 57.00 40.10 2.90 100
CI [48.49,52.99] [44.15,48.44] [2.19,4.01] 100 [56.80,61.27] [33.68,38.01] [4.25,6.17] 100 [54.15,59.80] [37.42,42.84] [2.06,4.06] 100
Rural % 56.77 41.51 1.72 100 61.99 33.69 4.32 100 65.11 31.82 3.07 100
CI [53.21,60.25] [38.06,45.05] [1.24,2.40] 100 [58.90,64.98] [30.81,36.71] [3.05,6.09] 100 [62.08,68.03] [28.88,34.91] [2.30,4.08] 100
Urban Panel
Formal
residential % 40.62 57.51 1.87 100 47.51 49.43 3.06 100 44.63 51.76 3.61 100
CI [36.52,44.86] [53.15,61.75] [0.97,3.56] 100 [42.97,52.10] [44.96,53.90] [2.02,4.60] 100 [40.59,48.73] [47.65,55.85] [2.23,5.79] 100
Township % 45.32 51.27 3.41 100 53.64 39.16 7.20 100 55.17 42.11 2.72 100
CI [42.02,48.67] [48.19,54.34] [2.10,5.48] 100 [50.46,56.79] [35.79,42.64] [5.41,9.53] 100 [51.22,59.05] [38.29,46.04] [1.85,3.98] 100
Shack dweller % 38.13 59.17 2.70 100 54.78 38.08 7.14 100 48.59 49.18 2.22 100
CI [32.93,43.62] [53.13,64.94] [1.34,5.36] 100 [48.72,60.71] [32.63,43.84] [4.57,11.00] 100 [43.42,53.79] [43.68,54.71] [1.23,3.98] 100
Peri urban % 51.44 44.72 3.84 100 61.02 34.58 4.40 100 64.38 32.74 2.89 100
CI [47.04,55.81] [40.65,48.86] [2.10,6.92] 100 [56.69,65.18] [30.43,38.98] [3.13,6.15] 100 [60.13,68.41] [28.77,36.97] [1.46,5.61] 100
Source: NIDS-CRAM W1 and W2; 90% confidence intervals
Embargoed until 30 September 12:00 (noon). Do not quote, cite or share.
22 | The Uneven Geography of the COVID-19 Crisis
Table A4: Job churn
Feb - April April - June
Job loss Job gain Job loss Job gain
Geo type
Metro % 21.59 10.18 18.3 19.81
CI [18.50,25.03] [7.83,13.13] [14.43,22.93] [16.44,23.68]
Cities/towns % 27.02 9.91 21.11 16.04
CI [24.64,29.53] [8.20,11.92] [18.46,24.04] [14.14,18.15]
Rural % 28.84 13.29 27.37 12.76
CI [24.95,33.06] [10.53,16.63] [21.82,33.72] [10.35,15.63]
Urban Panel
Suburbs % 16.92 7.77 11.54 18.67
CI [13.21,21.41] [5.16,11.55] [8.32,15.79] [14.78,23.30]
Townships % 23.9 11.14 20.07 14.43
CI [20.38,27.81] [8.46,14.53] [15.87,25.05] [11.38,18.14]
Shack dwellers % 33.55 11.33 21.26 28.72
CI [26.42,41.52] [7.47,16.82] [14.91,29.37] [23.08,35.10]
Peri urban % 26.98 7.77 27.57 13.68
CI [22.44,32.05] [5.42,11.02] [21.47,34.65] [10.85,17.11]
Source: NIDS-CRAM W1 and W2; 90% confidence intervals
Embargoed until 30 September 12:00 (noon). Do not quote, cite or share.
23 | The Uneven Geography of the COVID-19 Crisis
Table A5: Rate of unemployment
NIDS W5 (18+) CRAM W1 (April) CRAM W2 (June)
Geo type
Metros % 17.9 36.89 34.89
CI [15.96,20.03] [33.45,40.46] [30.51,39.54]
Cities/towns % 17.88 45.24 42.86
CI [16.28,19.59] [42.77,47.74] [39.65,46.14]
Rural % 28.19 48.48 52.48
CI [26.43,30.01] [44.87,52.10] [48.76,56.16]
Urban Panel
Suburbs % 11.44 29.47 24.16
CI [8.42,15.35] [25.27,34.05] [20.13,28.70]
Townships % 19.36 40.74 43.36
CI [15.98,23.26] [37.20,44.38] [39.19,47.63]
Shack
dwellers % 23.08 49.31 42.88
CI [17.43,29.89] [42.90,55.73] [37.56,48.37]
Peri urban % 21.12 49.92 51.57
CI [16.98,25.95] [45.07,54.78] [46.38,56.73]
Source: NIDS W5, NIDS- CRAM W1 and W2; 90% confidence intervals
Embargoed until 30 September 12:00 (noon). Do not quote, cite or share.
24 | The Uneven Geography of the COVID-19 Crisis
Table A6: Sources of household income, June 2020
Earnings only Grants only Other only Earnings &
grants
Other
combination
No income Total
Geo type
Metros % 44.58 23.21 10.1 9.093 7.518 5.509 100
CI [39.45,49.82] [20.09,26.64] [8.073,12.56] [7.296,11.28] [5.755,9.764] [3.957,7.623]
Cities/towns % 32.61 35.93 9.368 11.35 5.758 4.987 100
CI [30,35.34] [33.36,38.58] [7.865,11.12] [9.683,13.25] [4.92,6.728] [3.907,6.346]
Rural % 21.3 48.4 7.704 10.16 6.389 6.039 100
CI [18.89,23.92] [45.28,51.54] [5.91,9.986] [8.607,11.97] [4.995,8.137] [4.816,7.549]
Urban Panel
Suburbs % 52.69 21.78 12.69 4.466 4.274 4.1 100
CI [48.08,57.26] [18.7,25.2] [10.05,15.9] [3.255,6.099] [2.94,6.176] [2.781,6.006]
Townships % 35.34 30.67 7.817 14.03 7.677 4.46 100
CI [31.69,39.18] [27.56,33.96] [6.009,10.11] [11.42,17.13] [6.265,9.377] [3.171,6.24]
Shack dwellers % 31.44 26.41 9.467 13.96 8.667 10.06 100
CI [25.63,37.88] [22.2,31.1] [7.13,12.47] [10.46,18.39] [5.387,13.66] [6.123,16.09]
Peri urban % 25.02 43.49 7.884 10.39 7.592 5.623 100
CI [20.75,29.84] [39.21,47.86] [5.497,11.18] [7.983,13.43] [5.966,9.617] [3.996,7.86]
Source: NIDS-CRAM W2; 90% confidence intervals
Embargoed until 30 September 12:00 (noon). Do not quote, cite or share.
25 | The Uneven Geography of the COVID-19 Crisis
Table A7: Percentage adults reporting their household received a COVID-19 SRD Grant
No Yes Total
Geo type
Metros % 79.47 20.53 100
CI [75.15,83.21] [16.79,24.85]
Cities/towns % 76.47 23.53 100
CI [73.97,78.81] [21.19,26.03]
Rural % 66.86 33.14 100
CI [63.26,70.28] [29.72,36.74]
Urban Panel
Suburbs % 84.04 15.96 100
CI [79.02,88.03] [11.97,20.98]
Townships % 73.36 26.64 100
CI [68.99,77.33] [22.67,31.01]
Shack dwellers % 81.6 18.4 100
CI [76.24,85.98] [14.02,23.76]
Peri urban % 71.39 28.61 100
CI [66.89,75.51] [24.49,33.11]
Sourc es: NIDS-CRAM W2; 9 0% confidence intervals
Embargoed until 30 September 12:00 (noon). Do not quote, cite or share.
26 | The Uneven Geography of the COVID-19 Crisis
Table A8: Hunger
Anyone gone hungry in last week Child gone hungry in last week Ran out of money for food
W1 W2 W1 W2 CS2016 (year)* W1: (in April) W2: (in June)
Geo type
Metros % 17.43 13.33 10.98 8.743 16.22 43.6 34.93
CI [15.28,19.81] [11.03,16.02] [8.653,13.83] [6.322,11.97] [16.14,16.29] [40.11,47.15] [30.62,39.49]
Cities/towns % 23.79 15.82 16.85 11.81 20.77 47.89 37.1
CI [21.79,25.92] [13.98,17.84] [14.64,19.31] [10.09,13.78] [20.67,20.86] [45.78,50.02] [34.54,39.74]
Rural % 28.79 19.69 17.94 13.13 28.17 51.64 39.85
CI [25.89,31.88] [17.04,22.64] [15.22,21.01] [10.36,16.51] [28.08,28.26] [48.13,55.14] [36.9,42.87]
Urban Panel
Suburbs % 11.36 7.314 6.883 7.704 31.28 23.8
CI [8.887,14.41] [5.444,9.76] [4.947,9.5] [5.226,11.22] [27.48,35.35] [20.09,27.94]
Townships % 21.9 16.4 14.4 11.48 47.82 39.65
CI [19.25,24.8] [14.06,19.04] [11.72,17.56] [8.89,14.7] [44.34,51.33] [36.09,43.33]
Shack dwellers % 32.39 22.51 20.14 9.289 61.28 51.04
CI [27.61,37.58] [18.19,27.51] [14.69,26.97] [6.02,14.07] [55.06,67.15] [43.88,58.16]
Peri urban % 25.38 19.22 14.65 13.41 45.75 38.2
CI [21.75,29.39] [16.19,22.67] [11.73,18.15] [10.58,16.86] [42.16,49.39] [33.98,42.61]
Source: Communit y Survet 2016; NIDS-CRAM W1 and W2; 90% confidence intervals
Embargoed until 30 September 12:00 (noon). Do not quote, cite or share.
27 | The Uneven Geography of the COVID-19 Crisis
Appendix B
0,2 0,0 0,0 0,0
48,3 46,9 46,9 48,4
51,6 53,1 53,1 51,6
0%
10%
20%
30%
40%
50%
60%
70%
80%
90%
100%
NIDS:W5 NIDS-CRAM:W1 NIDS-CRAM:W2 C S2016
National
Don't know Male Female
0,2 0,0 0,0 0,0
78,6 78,4 78,6 78,4
9,0 9,6 9,2 9,2
2,8 2,4 2,4 2,8
9,3 9,6 9,7 9,6
0%
10%
20%
30%
40%
50%
60%
70%
80%
90%
100%
NIDS:W5 NIDS-CRAM:W1 NIDS-CRAM:W2 C S2016
National
Don't know African Coloured I ndian White
18,4 15,5 15,6 20,5
29,3
25,7 26,1
27,2
20,6
23,2 22,7
19,9
13,9 14,9 15,3 14,3
9,7 10,5 10,4 9,9
5,4 7,0 6,7 5,6
2,7 3,2 3,3 2,7
0%
10%
20%
30%
40%
50%
60%
70%
80%
90%
100%
NIDS:W5 NIDS-CRAM:W1 NIDS-CRAM:W2 CS2016
National
15-24 25-34 35-44 45-54 55-64 65-74 75+
17,1 13,3 13,4 16,9
30,9
25,6 27,7 25,9
21,3
27,5 24,3 23,1
14,3 14,9 15,4 15,7
9,3 9,6 8,9 10,3
4,6 6,4 6,3 5,8
2,5 2,7 4,0 2,4
0%
10%
20%
30%
40%
50%
60%
70%
80%
90%
100%
NIDS:W5 NIDS-CRAM:W1 NIDS-CRAM:W2 CS2016
Metros
15-24 25-34 35-44 45-54 55-64 65-74 75+
0,3 0,0 0,0 0,0
49,1 46,4 49,5 49,5
50,6 53,6 50,5 50,5
0%
10%
20%
30%
40%
50%
60%
70%
80%
90%
100%
NIDS:W5 NIDS-CRAM:W1 NIDS-CRAM:W2 C S2016
Metros
Don't know Male Female
0,5 0,0 0,0 0,0
67,2 66,5 64,8 68,4
11,6 13,2 12,3 11,8
6,0 4,8 5,4 5,4
14,7 15,5 17,6 14,4
0%
10%
20%
30%
40%
50%
60%
70%
80%
90%
100%
NIDS:W5 NIDS-CRAM:W1 NIDS-CRAM:W2 C S2016
Metros
Don't know African Coloured I ndian White
Embargoed until 30 September 12:00 (noon). Do not quote, cite or share.
28 | The Uneven Geography of the COVID-19 Crisis
17,2 16,8 16,6 20,4
30,0 24,8 23,1
28,5
20,7
22,2 23,3
19,5
14,5 15,6 15,9
14,6
10,6 10,2 11,1 9,7
4,9 7,3 7,2 5,0
2,3 3,2 2,9 2,3
0%
10%
20%
30%
40%
50%
60%
70%
80%
90%
100%
NIDS:W5 NIDS-CRAM:W1 NIDS-CRAM:W2 CS2016
Cities/towns
15-24 25-34 35-44 45-54 55-64 65-74 75+
0,0 0,0 0,0 0,0
49,9 47,3 45,7 49,3
50,0 52,7 54,3 50,7
0%
10%
20%
30%
40%
50%
60%
70%
80%
90%
100%
NIDS:W5 NIDS-CRAM:W1 NIDS-CRAM:W2 C S2016
Cities/towns
Don't know Male Female
0,1 0,0 0,0 0,0
73,6
82,3 78,9 72,1
13,6
9,3 12,3
14,5
0,7
1,0 1,0 2,2
12,0 7,4 7,8 11,2
0%
10%
20%
30%
40%
50%
60%
70%
80%
90%
100%
NIDS:W5 NIDS-CRAM:W1 NIDS-CRAM:W2 C S2016
Cities/towns
Don't know African Coloured I ndian White
21,2 16,9 17,0
25,1
26,6 28,5 28,7
27,7
19,6 16,6 19,6
16,1
12,9
13,1 13,8 12,3
9,5 13,1 11,5 9,6
6,9 7,6 6,7 6,0
3,3 4,3 2,8 3,2
0%
10%
20%
30%
40%
50%
60%
70%
80%
90%
100%
NIDS:W5 NIDS-CRAM:W1 NIDS-CRAM:W2 CS2016
Rural
15-24 25-34 35-44 45-54 55-64 65-74 75+
0,1 0,0 0,0 0,0
45,9 47,2 45,0 46,3
54,0 52,8 55,0 53,7
0%
10%
20%
30%
40%
50%
60%
70%
80%
90%
100%
NIDS:W5 NIDS-CRAM:W1 NIDS-CRAM:W2 C S2016
Rural
Don't know Male Female
0,0 0,0 0,0 0,0
96,7 93,8 98,2 95,4
2,1 2,7 0,2 1,8
0,8 1,1 0,3 0,2
0,5 2,5 1,3 2,7
0%
10%
20%
30%
40%
50%
60%
70%
80%
90%
100%
NIDS:W5 NIDS-CRAM:W1 NIDS-CRAM:W2 C S2016
Rural
Don't know African Coloured I ndian White
Embargoed until 30 September 12:00 (noon). Do not quote, cite or share.
For further information please see cramsurvey .org
Embargoed until 30 September 12:00 (noon). Do not quote, cite or share.
