The impact of COVID-19 in South Africa during the first year of the crisis: Evidence from the NIDS-CRAM survey

ABSTRACT

This paper provides an introduction to this Special Issue of Development Southern Africa that evaluates the impact of COVID-19 in South Africa, one year into the pandemic. All of the papers use evidence from the National Income Dynamics Study – Coronavirus Rapid Mobile Survey (NIDS-CRAM), a five-wave longitudinal survey conducted from April 2020 to July 2021. As we write this article in June 2022, South Africa has just returned to the same level of GDP that it had at the end of 2019. This two-year period marks one of the most tumultuous in the country’s economic history. We showcase results pertaining to employment, income support, hunger, schooling, early childhood development, mental health, and vaccine hesitancy. We also reflect on the policy learnings that can be gleaned in each of these domains and draw on some of the international lessons learnt to point to the way forward.

KEYWORDS:

• COVID-19
• NIDS-CRAM
• socio-economic impact
• survey data

1. Introduction

The National Income Dynamics Study – Coronavirus Rapid Mobile Survey (NIDS-CRAM) was a five-wave longitudinal (panel) dataset of adults (age 18+) in South Africa, interviewed telephonically approximately every 3 months between May 2020 and May 2021. This period marked one of the most profound negative shocks in South Africa and the world, brought on by the spread of the COVID-19 pandemic. In response, governments around the world rallied to minimise the impact of the virus on their populations, using unprecedented policy tools associated with restricting the movement of people, goods and services that became known as lockdown regulations. In South Africa, a State of Disaster was declared on the 15th of March 2020, which lasted until the 15th of April, 2022 (Department of Cooperative Government, 2022), during which time a series of different lockdown levels were enforced.

It was in this context that the NIDS-CRAM survey was conceptualised. The project was spearheaded by Dr Nic Spaull (Stellenbosch University), who in March 2020, as the country was entering one of the strictest lockdowns in the world, brought together over thirty academics and researchers from institutions across the country to collaborate on what became one of the largest data collection projects in Africa during the first year of the crisis. The Southern Africa Labour and Development Research Unit (SALDRU) at the University of Cape Town became responsible for implementing NIDS-CRAM, commencing with securing ethical clearance for the survey (reciprocal ethics approval was later obtained from Stellenbosch University), managing the computer-assisted telephonic interview (CATI) process, and reporting to a multi-institution Steering Committee, tasked with evaluating the progress of each of the five waves of data collection.

The main purpose of the survey was to capture timely and frequent information on some of the most important socio-economic impacts of the crisis during the ongoing lockdown, particularly given that the collection and release of official data by Statistics South Africa would take some time. Indeed, this survey became an indispensable resource to both academics and policy-makers alike, and in addition to the over 70 working papers published by the academic community, the findings from the survey featured in countless webinars, policy-outreach sessions, and print and broadcast media outlets.

A key challenge for the NIDS-CRAM team was how to collect nationally representative data that could be used to reliably inform policy-making in the context of a strict lockdown and limitations on face-to-face contact. A decision was taken to draw a sample from the pre-existing NIDS survey which had last been conducted by SALDRU in 2017, and which was broadly representative of the South African population. In the first wave of data collection of NIDS-CRAM, which took place over May/June 2020, just over 7000 adults were interviewed in a 20-minute telephonic survey. Thereafter, these same individuals were interviewed another four times over the following year, providing a longitudinal picture of the early impacts of the pandemic (the sample was also refreshed in Wave 3 to ensure sufficient statistical power for analysis).¹ Information was collected on a range of themes, among them employment, earnings, income support, hunger, schooling, early childhood development, mental health, and vaccine acceptance. Of course there were many other important topics we would like to have tracked, but there were strict limitations on telephone interview length. Nonetheless, a wealth of information was collected on the impact of the various lockdown stages on the wellbeing of South Africans during the first year of the pandemic.

The purpose of this special issue is to showcase some of the most important findings on the socio-economic impacts of the crisis that emerged from the analysis of the five waves of the NIDS-CRAM survey. The collection of articles in this volume (adapted from the Wave 5 working papers by the authors) evaluate changes to the welfare of the population as lockdown regulations fluctuated between the most restrictive level (Level 5, implemented from 27 March 2020–30 April 2020) and the least restrictive level (an ‘adjusted’ Level 1², implemented from 1 March 2021–30 May 2021). As such, they represent a profound contribution to our knowledge of the immediate impact of these types of regulations on the population, for no other survey in South Africa can be mapped so accurately to the precise dates of each lockdown. Collectively then, these articles provide us with an opportunity to learn about the efficacy of such a policy response, perhaps with an eye to the future so that if ever we are to implement similar regulations again, we do not repeat the mistakes but instead amplify the successes.

This introductory paper to the volume is organised as follows. We start by describing the macroeconomic context in which the NIDS-CRAM survey was conducted, and then we highlight the main findings on the socio-economic impacts of the crisis as detailed in the collection of papers in this issue. In doing so, we reflect not only on the policy implications emerging from this body of research, but also the lessons learnt from the data collection process itself. We organise the discussion of the papers around some of the main themes captured in NIDS-CRAM: employment and income support; food security and mental health; and education and child wellbeing. We then give some consideration to the way forward and to the important policy lessons learnt during the crisis. The final section offers a brief conclusion.

2. The socio-economic impacts of the crisis: key findings from five waves of NIDS-CRAM

2.1. Macroeconomic context

In a recent statistical release by Statistics South Africa (SSA, 2022), it is apparent that South Africa only returned to its pre-pandemic GDP level in the first quarter of 2022. To get some understanding of the extent of the shock that the COVID-19 crisis inflicted on the economy, and how it impacted different sectors in the economy, Table 1 provides figures on GDP growth for the years directly preceding the pandemic as well as for 2020 and 2021.

Table 1. Annual change in industry value added and GDP, at constant 2015 prices.

% change year-on-year	2017	2018	2019	2020	2021
Agriculture, forestry and fishing	19,1	0,5	−6,4	14,9	8,8
Mining and quarrying	2,4	−0,8	−0,6	−11,8	12,0
Manufacturing	−0,2	1,6	−1,0	−12,5	6,5
Electricity, gas and water	0,3	0,9	−3,3	−5,9	2,2
Construction	−5,7	−1,4	−3,4	−18,5	−2,2
Trade, catering and accommodation	−1,3	1,0	−0,7	−12,4	6,4
Transport, storage and communication	1,0	2,8	−0,7	−15,4	4,7
Finance, real estate and business services	2,5	3,3	2,5	0,7	3,3
General government services	1,4	1,9	1,4	0,7	0,1
Personal services	1,3	0,5	1,4	−2,0	5,5
Total value added at basic prices	1,2	1,6	0,2	−5,9	4,7
Taxes less subsidies on products	1,1	1,0	0,9	−10,3	6,9
GDP percent change	1,2	1,5	0,3	−6,3	4,9

Source: Statistics South Africa, 2022

The table shows that the size of the economy contracted by a massive 6.3% in 2020, but then grew by 4.9% in 2021, on its way to a full recovery by the end of Q1-2022 (GDP growth was recorded at 1.9% for this quarter). The initial shock was felt across the economy, although the sectors that were hit the hardest in 2020 were construction; transport, storage and communication; manufacturing; trade, catering and accommodation; and mining and quarrying. However, agriculture, forestry and fishing had a bumper year, increasing by nearly 15%, driven by various factors, among them large harvests and strong commodity prices. It should be noted that the reduction in GDP does not capture the negative impact of the crisis on the care economy and is unlikely to capture the full impact on the informal economy, which was also substantial.

Despite two more waves of the virus during 2021, lockdown restrictions were milder in 2021 compared to the previous year, and the economy was able to move towards a recovery with almost all sectors recording growth. So much so, that from a year-on-year perspective, South Africa experienced the strongest growth rate for over a decade at 4.9%, but off the very low base created by COVID’s economic destruction (and following on from the lacklustre growth of the preceding decade). However, the construction sector – a leading indicator of pro-cyclical business cycle fluctuation – remained negative, indicating that uncertainty with respect to the economy’s recovery was still manifest.

Table 2 provides more detail on the quarterly changes that occurred over the period of the pandemic during which NIDS-CRAM was collected. The figures show that the most dramatic reduction in economic activity took place in the quarter ending June 2020, during which the strictest set of lockdown regulations were in place in South Africa (Levels 5 and 4). During this phase, almost all economic activity was suspended except for the production of essential goods and services; there were curfews in place limiting the movement of people, goods and services; schools and childcare facilities were shut down; borders were closed; and certain industries like alcohol and tobacco were prohibited from selling their goods. Thereafter, as lockdown levels were relaxed and economic activity started to resume, there was progressive recovery in the economy, although at varying rates across the sectors. In the second quarter of 2021 (ending June 2021), all sectors recorded growth.

Table 2. Quarterly change in industry value added and GDP, at market prices.

% change year-on-year	Dec 19	Mar 20	Jun 20	Sep 20	Dec 20	Mar 21	Jun 21
Agriculture, forestry and fishing	−1,8	49,6	18,8	21,2	30,0	11,6	52,7
Mining and quarrying	10,6	18,8	−8,6	17,0	18,6	34,3	82,9
Manufacturing	1,6	0,4	−25,9	−3,0	0,4	2,3	43,4
Electricity, gas and water	7,8	10,8	−6,2	4,8	9,7	−0,5	9,0
Construction	−0,8	−3,3	−31,4	−19,6	−14,4	−11,8	16,1
Trade, catering and accommodation	3,0	2,7	−24,6	−2,8	−1,1	3,8	47,3
Transport, storage and communication	1,2	0,4	−25,7	−12,3	−7,8	−7,4	34,5
Finance, real estate and business services	6,5	8,0	−5,5	−0,8	5,8	0,8	15,4
General government services	5,1	11,7	6,7	7,3	7,2	2,9	4,3
Personal services	5,6	6,2	−2,6	1,4	5,4	5,2	12,6
Total value added at basic prices	4,5	6,7	−11,2	0,1	3,6	3,6	27,2
Taxes less subsidies on products	5,6	4,3	−30,2	−9,1	3,9	2,4	59,2
GDP percent change	4,6	6,4	−13,0	−0,8	3,6	3,5	29,7

Source: Statistics South Africa, 2022

The NIDS-CRAM data were collected in five waves over the period spanning May 2020 to May 2021, and covered economic activity in six key months: February 2020 or the pre-pandemic baseline (collected retrospectively); April 2020 or Level 5 lockdown; June 2020 or Level 3 lockdown; October 2020 or Level 1 lockdown; January 2021 or ‘adjusted’ Level 3 lockdown; and March 2021 or ‘adjusted’ Level 1 lockdown. The data therefore provide us with unique insight into what was happening over the course of the first year of the pandemic, during which economic activity was most drastically curtailed and the fallout most severe, as shown in the tables above. It was also a period during which lockdown levels changed regularly. Thus NIDS-CRAM offers us a window into the impact of various lockdown regulations, as we are able to map these fluctuations to the socio-economic outcomes of the South African population.³ It is to this that we now turn in the following sections, which highlight some of the key findings from the collection of papers in this issue.

2.2. Employment and income support

The first paper in the series, by Daniels et al. (2022a), shows how devastating the initial shock of Level 5 lockdown was to the labour market: between February 2020 and April 2020, the employment-to-population ratio recorded in NIDS-CRAM fell sharply from 56.6% to 48.3% for the working-age population (18-64 year olds). Their results also highlight how employment changed with subsequent fluctuations in lockdown levels: the employment-to-population ratio was much the same in June 2020 when lockdown restrictions were still relatively severe (Level 3 lockdown), it recovered substantially in October 2020 when Level 1 lockdown was in place, only to fall again in January 2021 when regulations were tightened (to an ‘adjusted’ Level 3 lockdown) following the start of the second wave of the pandemic. The final wave of NIDS-CRAM suggests a subsequent recovery to pre-pandemic levels in March 2021, when the least restrictive ‘adjusted’ Level 1 lockdown was in place.

A key feature of the paper is that the authors conduct a comparative analysis between the five waves of NIDS-CRAM and Statistics South Africa’s Quarterly Labour Force Surveys (QLFS: 2020-Q1 to 2021-Q1). They find that while the largest fall in the employment-to-population ratio also occurred between Q1 and Q2 of 2020 in the QLFS, thereafter the ratio never recovered fully to pre-pandemic levels. Their investigation helps explain why the two surveys produce different estimates. In doing so they highlight that under conditions of massive uncertainty and change, as was the case during the first year of the pandemic, the definition of employment or what counts as ‘having a job’, and how this information is collected, becomes more complex and can affect the labour market estimates.

While the QLFS did not change the way the main labour market questions were asked relative to previous waves (as recommended by the ILO), NIDS-CRAM was not bound by the need for international (and temporal) comparability and was ‘designed from the ground-up to identify the impact of lockdowns’ (Daniels et al. 2022b). The QLFS captures whether the respondent was employed in the previous week, with interviews spanning the 3 months of the quarter. NIDS-CRAM collected information on employment in a specific month. Therefore while the QLFS smooths over some of the big swings in employment that occurred within the quarter as a result of changes to lockdown levels, NIDS-CRAM is able to capture the impact of different lockdown regulations. Instead of the two surveys being seen as contradictory, the authors make the point that they should be viewed as complementary as they serve different purposes. This paper is therefore an important contribution to the series, not only because it illustrates the impact of lockdown regulations in the labour market, but because it helps the academic and policy-making community understand why there are differences between NIDS-CRAM and the official estimates, and what each set of data can be used for.

The next paper in the series by Casale & Shepherd (2022) highlights that even though the aggregate employment estimates in NIDS-CRAM appear to have largely recovered to pre-pandemic levels in March 2021, there was substantial variation by gender. As has been found in many other countries around the world, women in South Africa suffered disproportionately from the crisis. Women’s employment fell by 22.5% between February and April 2020 as a result of the strict lockdown, while men’s employment fell by 9.8%. Although there was substantial recovery for both women and men thereafter, one year later (in March 2021) women’s employment was still down roughly 8.4% compared to its pre-COVID level, while men’s employment was back to its pre-pandemic level. A similar gendered pattern was found in the number of hours worked per week among the employed.

A key reason the authors identify for this uneven effect is that pre-COVID women were more likely to be employed in those parts of the economy that were hit hardest by the crisis, among them non-essential retail, personal care, tourism, hospitality, domestic work and childcare – jobs that cannot be performed at home and that require face-to-face contact (and are often more irregular in nature). A second important reason is that women were far more affected by the crisis in childcare that ensued as a result of school and childcare facility closures during the COVID-19 lockdown.⁴ A unique feature of the NIDS-CRAM survey is that unlike other surveys conducted in SA during the crisis, information on the time respondents spent on childcare was collected. These data show clearly that women took on more of the extra childcare than men as the first lockdown was implemented, and that women’s time was far more responsive than men’s to school closures and reopenings over the course of the first year of the pandemic. When asked directly, twice as many women as men said that having to perform childcare during lockdown affected their ability to work, search for work or work the same number of hours as before.

Despite these uneven effects, women benefited less from the government income support provided to unemployed and furloughed workers through the Unemployment Insurance Fund-Temporary Employer/Employee Relief Scheme (UIF-TERS) and the special COVID-19 Social Relief of Distress Grant (SRDG). In the case of the former, this was because fewer women were employed and registered on the formal UIF system, and in the case of the latter, unemployed women collecting the Child Support Grant on behalf of children (a substantial number) were initially excluded from also receiving the SRDG because of the grant’s conditions. The authors’ work highlights the importance of analysing both economic outcomes and policy responses through a gendered lens, taking into account women’s and men’s differential roles and responsibilities in the labour market and in the home.

While the UIF-TERS scheme may have benefited women less, it nonetheless played a fundamental role in the government’s attempt to limit mass retrenchments and support employers and employees unable to go to work or to work the same number of hours as in pre-COVID times because of lockdown restrictions. The scheme was introduced in April 2020, and while there were payment delays to beneficiaries, official data suggest that by the end of March 2021, nearly R59 billion had been paid out to 5.4 million employed individuals. The work by Köhler & Hill (2022) in this issue provides a detailed analysis of this important wage subsidy retention scheme. In doing so, they exploit another unique feature of the NIDS-CRAM survey, namely that questions on individual UIF-TERS receipt were asked of both employers and employees in each wave following the introduction of the programme (i.e. waves 2-5), information not available even in the QLFS. This allows the authors to show how many of the employed benefited in any one month, who among the employed was more likely to benefit, and whether UIF-TERS receipt resulted in job retention.

Their findings come out in strong support of the programme. Millions of workers benefited from the scheme, particularly during the strictest lockdown periods in 2020 when economic activity was highly restricted. To identify who benefited most from the programme, the authors compare relative shares in overall employment to shares of UIF-TERS receipt for a range of demographic groups. They find that men, the semi-skilled, those with written contracts, and those in the secondary sector benefited relatively more from the programme. However, over time as the economy started reopening and fewer workers were receiving UIF-TERS, the beneficiaries were increasingly from the more vulnerable groups: Black South Africans, those with very low levels of education, and those unable to work from home.

Perhaps one of the most important findings from the paper however derives from the longitudinal analysis of the NIDS-CRAM data on job retention. The authors find that UIF-TERS receipt was strongly correlated with job retention during the strictest lockdown period in 2020, suggesting that in addition to providing valuable income support to a large number of workers and their households, the scheme also achieved its core objective of attenuating job loss. Nonetheless, as the authors themselves recommend, additional work is required using other data sources and alternative methodologies to corroborate this key finding.

The final paper covering labour market outcomes in this series is the contribution by Espi-Sanchis, Leibbrandt & Ranchhod (2022). In this work the authors disaggregate labour market outcomes over the period by age group, and as in Casale & Shepherd (2022), they show differential impacts. While the youth have the lowest employment rate, younger workers did better over the period than older workers. The youth category (age 18-24) recorded an increase in the employment-to-population ratio between February 2020 and March 2021 (from 33% to 35%), while the oldest cohort (age 55-64) experienced a decline (from 45% to 41%).

An important feature of this paper is that it also employs the panel nature of the data to investigate transitions between waves. While the analysis at the cross-section finds (as in Daniels et al., 2022b) that the employment-to-population ratio had largely recovered to pre-pandemic levels by March 2021, the analysis of employment dynamics uncovers substantial churning in the labour market over this first year of the crisis, and more so than during pre-pandemic times. By March 2021, 23% of those who were employed in February 2020 had lost work, and 30% of those who were not employed in February 2020 had found work. Within this large restructuring of the labour market, young workers benefited more than older workers. Young workers were more likely to find work, while older workers were less likely to find work, and more likely to exit the labour market.

The authors also illustrate an important limitation of the NIDS-CRAM survey, namely that due to the relatively small sample size, there is insufficient statistical power for highly disaggregated analysis. This affects the age group analysis, but is particularly evident when trying to identify patterns in occupation and sector of employment.⁵ Future work using the larger samples in the national labour force surveys should investigate these age patterns further to uncover whether they persisted into the second year of the pandemic, and what kinds of jobs were lost and created during this period of massive disruption to the economy.

2.3. Food security and mental health

One of the main objectives of the NIDS-CRAM survey was to try and capture the impact of the crisis and the ensuing fallout in the labour market on people’s socio-economic wellbeing. Under ideal circumstances, one would like to have collected information on household income to be able to measure poverty rates over the period. However, collecting data of this nature is complex and time-consuming, and a roughly 20-minute telephonic interview was not the appropriate mechanism. Instead, a decision was taken to ask questions related to food security and hunger as a proxy for wellbeing.

The work by van der Berg et al. (2022) investigates these data and the picture they paint is a deeply worrying one. Almost half of all respondents said their household had run out of money to buy food at the height of the strict lockdown in April 2020, one in four respondents said someone in their household had gone hungry in the week preceding the interview, and roughly one in seven respondents said a child in the household had gone hungry. Although not strictly comparable to the information collected in the pre-COVID rounds of the General Household Survey (GHS), these indicators of food insecurity and hunger appear to have increased drastically compared to pre-pandemic times.⁶

The authors also suggest that social welfare interventions, such as the temporary top-up of the pre-existing grants between May and October 2020 and the introduction of the SRDG and UIF-TERS, were crucial in helping mitigate the impact in households. These interventions, together with a gradual resumption of economic activity, meant that food insecurity and hunger declined by Wave 2. However, after that, they stabilised at higher levels than in pre-pandemic times. More specifically, the percentage of respondents who said their household had run out of money to buy food fell from 48% in April 2020 to 38% in June 2020, but then remained at similar levels (of between 35% and 40%) in October 2020, January 2021 and March 2021. Similarly, household hunger declined from 23% in Wave 1 to 16% in Wave 2, but then fluctuated between 16% and 18% in Waves 3, 4 and 5. These results highlight the importance of more frequent data collection during times of crisis and high volatility. Information on food insecurity and hunger collected once a year through the GHS would not have been able to pick up these big shifts during the course of the year.

Another important finding from the data is that child hunger has proven much harder to reduce. In Wave 1, 15% of respondents said a child had gone hungry in their household, this declined to 11% in June 2020, but then increased again and remained at between 14% and 16% in Waves 3–5. While adults appear to have been ‘shielding’⁷ children from hunger to some degree, clearly this was not possible in all households and the consequences of this are dire; poor nutrition and stunted growth in childhood have negative consequences for children’s development that span well into the future. The authors argue that social support to vulnerable households, and especially those with children, should remain a priority. This message cannot be overstated, especially in light of the below-inflation increases to the social grants over the last years, the rapidly rising cost of living, the ongoing administrative challenges around paying out the SRDG, and the continued uncertainty around the extension of the grant.

Visagie & Turok (2022) probe the data further and provide a spatial analysis of the socio-economic impacts of the crisis. They disaggregate the urban sample into ‘suburbs’ (or formal residential areas), ‘townships’, ‘shack-dwellers’ (or informal settlements) and ‘peri-urban areas’ (including farms, small-holdings and traditional areas). Their findings reflect pre-existing inequalities; shack-dwellers were affected most severely by the crisis, while people living in the suburbs were least affected.

Shack-dwellers experienced a more precipitous decline in employment than residents of other areas following the implementation of the strict lockdown in March 2020. And one year into the pandemic, their employment rate remained depressed (by at least 10 percentage points), while in other areas employment rates were almost fully recovered according to the NIDS-CRAM data. There was also far more churning among residents of shack settlements, townships and peri-urban areas, whereas the suburbs had the highest rate of stable employment (people employed in all five waves of the survey). Despite being the ‘main shock-absorbers’ of the crisis (Visagie & Turok, 2022), the authors find that the percentage of shack-dwellers who received the SRDG was generally lower than in townships and peri-urban areas, and shack-dwellers were no more likely to receive the UIF-TERS than residents of other areas. These factors help explain why shack-dwellers reported the highest rates of food insecurity and hunger during the pandemic. The percentage of shack-dwellers whose households had run out of money to buy food at the height of the strict lockdown was about twice that of the suburbs, while rates of hunger were almost three times as high (with the other areas in between). A year into the pandemic, almost half of shack-dwellers were still reporting that their households were food insecure and one in four were still reporting hunger.

The vulnerability of informal settlements to the fallout from the crisis is not surprising. These areas are densely populated, they lack clean water, infrastructure and other basic facilities, and many of their residents rely on informal and casual work in the labour market. This combination of characteristics left these areas particularly open to infection and to the devastating impacts of the lockdown (which initially prohibited the informal trade that so many rely on). The paper by Visagie & Turok (2022) highlights how communities that live in informal housing and rely on informal activities for their livelihoods, and who therefore lack many of the legal and financial protections available to those in better-off areas, need special attention during times of crisis.

In addition to measures of economic wellbeing, the NIDS-CRAM survey attempted to collect information on psychological wellbeing or mental health. Again, in a short telephonic interview, it was not possible to include the kinds of questions that would allow for more accurate measurement of this sensitive issue. Nonetheless, the information collected on depressive symptoms in Waves 2, 3 and 5 at the least can be used to identify people at risk of mental health problems, and has been invaluable in drawing attention to this growing social problem.⁸

Although not in this volume, the Wave 5 working paper by Hunt et al. (2021) showed that the prevalence of depressed mood was 24% by the middle of 2020 (interviews for Wave 2 took place in July and early August) and increased further to 29% of the adult population by Wave 3 (November/December 2020) where it remained in Wave 5 (April/May 2021). While it is not possible to identify by how much the prevalence had risen compared to pre-COVID times because no baseline measure is available in NIDS-CRAM, that between one third and one fourth of the population was suffering from depressive symptoms at any one point in time during this first year of the crisis is alarming.

Another interesting finding in their working paper is that when tracking people over time, it becomes evident that it is not always the same people experiencing depressed mood. People moved in and out of the state between waves; indeed, just over half of the population interviewed (52%) reported depressed mood at least once over the course of the three waves, with approximately 7% reporting depressed mood in all three waves. The authors suggest that this substantial ‘churning’ implies environmental factors likely play a key role in determining depressed mood.

Of course the factors driving poor mental health are varied and complex, and they will likely interact in numerous ways with the devastating consequences (economic, physical and social) of the COVID-19 crisis. The NIDS-CRAM survey was not designed to identify these pathways. Nonetheless, a simple analysis of correlations can point to potential environmental causes and important avenues for future research. In this regard, Hunt et al. (2021) identify a strong association between hunger in the household and depressive symptoms. In doing so, they draw attention to the suffering that economic instability inflicts not only on physical health but also emotional wellbeing, lending further impetus to the call by van der Berg et al. (2022) to prioritise the eradication of food insecurity and hunger through social welfare policy.

2.4. Education and child wellbeing

Unlike in previous economic crises, the COVID-19 crisis involved the complete shutdown of all schools, early childhood care and education (ECCE) centres, and other childcare facilities from 18 March 2020. In addition, the roughly one million domestic workers and childminders employed pre-crisis were unable to return to work until 1 June 2020 (when Level 3 lockdown was implemented). Schools and ECCE centres eventually re-opened in a staggered fashion from June/July 2020 onwards, but many weeks of learning were lost in 2020, and when the second wave of the pandemic occurred at the end of that year, a decision was taken to delay the reopening of schools in 2021 to the middle of February. Even when schools were open in 2020 and 2021, social distancing requirements necessitated rotational schedules in many schools, which involved children receiving contact learning on alternate days or weeks. With only around 10% of households reporting access to the internet in the GHS 2019 (Shepherd & Mohohlwane, 2022), this meant that the majority of learners in South Africa were unable to learn for extended periods during the course of 2020/2021.

This had a number of severe knock-on effects. The impacts in the labour market and on care work in the home have already been discussed (Casale & Shepherd, 2022; Daniels et al., 2022a). In this section, we provide an overview of the key findings in Shepherd & Mohohlwane (2022) and Wills & Kika-Misty (2022), whose in-depth analyses cover the effects of the school and ECCE centre closures respectively. Indeed, capturing information on education disruptions and child wellbeing was one of the main objectives of the NIDS-CRAM survey.

An important contribution of Shepherd & Mohohlwane (2022) is their attempt to quantify the effect of school closures on absenteeism and drop-out. Their estimates of non-attendance rely on adult reporting in the NIDS-CRAM surveys on the recent attendance and return to school of learners living in their households (as NIDS-CRAM interviewed adults only and was unable to capture information on each child in a household). Nonetheless, based on a series of plausible assumptions and the usual qualifications, the authors estimate that the number of learners not attending school in April/May 2021 was four times larger than in pre-pandemic years, amounting to roughly 725 000 additional learners out of school compared to estimates from other recent pre-pandemic surveys (namely Wave 5 of NIDS conducted in 2017 and the GHSs collected from 2002 on). The authors make the startling claim that this is the highest number of children out of school in at least two decades. Particularly worrying is the finding that increases in non-attendance were highest at the primary level, a phase that usually sees almost universal return of learners each year. Coupled with the substantial learning losses measured through other COVID-era surveys among children returning to school (Ardington et al., 2021), this crisis has further weakened an already fragile education system in South Africa.

Another terrible consequence of school closures was the suspension of the National School Nutrition Programme (NSNP) in South Africa, which was estimated to have provided meals to approximately 82% of the school population pre-COVID. This school-feeding scheme, a lifeline to many poor households with children, was initially suspended for four months until civil society organisations took the government to court in mid-2020 to force its reinstatement. As the authors write, ‘[t]his is significant, as the right to nutrition was previously seen as a supplementary education function rather than an educational right’ (Shepherd & Mohohlwane, 2022). Despite this legal victory, the provision of school meals in the context of school closures and rotational timetabling proved challenging, and evidence from NIDS-CRAM suggests that it never fully recovered to pre-COVID levels over the course of the following year. Compared to the 65% of adults who reported at least one learner in their household had received a school meal at least weekly in the GHS 2019, only 26% reported that a child in their household had received a school meal in the preceding week in July 2020. This percentage increased to 49% in November/December 2020 and to 56% in April/May 2021, although it is highly likely these children were not receiving school meals every day or as often as before the pandemic. Administrative data suggest that 1.5 million or 14% of targeted learners were still not receiving meals at school by the middle of 2021.

The interruptions to the school-feeding scheme no doubt contributed to the sharp increases in reported food insecurity and hunger that we describe above, and may help explain why child hunger has proven particularly difficult to decrease. Shepherd & Mohohlwane (2022) also highlight the implications for caregiver mental health. They show that the prevalence of depressed mood among adults living with children rose between Waves 2 and 5 (from 23% in July 2020 to 29-30% in November 2020 and April 2021), and was significantly higher among those who reported household hunger compared to those living in households with no reports of hunger (40% vs 26% in April 2021). Similarly large differences are found when comparing depressive symptoms among adults living in food insecure households in which learners had not received a school meal compared to adults living in food insecure households where at least one child had been able to access a school meal in the preceding week (51% vs 36% in April 2021). Although the NIDS-CRAM survey couldn’t collect information on children’s mental health, long absences from school and isolation from friends, hunger, and more general COVID worries around caregiver job loss, death and illness, would no doubt have had a deleterious effect on children’s mental health, an area that requires far more urgent attention in our education and public health systems.

The final paper in the series provides empirical evidence on attendance levels in the early childhood care and education (ECCE) sector, an understudied area in low-to-middle income countries given the nature of the sector (consisting of many informal, private providers⁹) and the complexity of data collection during the pandemic. NIDS-CRAM provides the only available national-level data to track the impacts of the lockdown in the sector, although the same caveat as with the schooling data applies - NIDS-CRAM collects information from adults on whether any children in their household attended ECCE programmes, and it is not possible to accurately measure how many children were affected.

Nonetheless, with these limitations in mind, Wills & Kika-Mistry (2022) show that the sector took a substantial knock during the pandemic. About 39% of adult respondents living with children aged 0–6 years indicated that at least one of these children had attended an ECCE programme pre-COVID (i.e. in February 2020). This figure fell markedly to 7% in July/August 2020 as a result of the lockdown and government-mandated closures. A High Court judgment on 6 July 2020 ruled that ECCE programmes could reopen subject to meeting certain safety standards, but substantial volatility was still recorded in the sector in the months that followed, with (household) attendance estimated at 28% in November/December 2020, 7% in early February 2021 (when schools were still closed), and 36% in April/May 2021. Reasons given for non-return were initially dominated by supply-side constraints, particularly that ECCE centres remained closed, but then demand-side issues became increasingly prominent, specifically concerns about children contracting COVID at the centre and the inability to afford the centre fees.

The significant recovery reported in the final wave of data collection is reassuring, and additional analysis by the authors suggests that a large part of it was related to households’ higher perceived ability to afford ECCE programme fees. This may have had to do with lower-fee ECCE programmes reopening in anticipation of government financial support (relief to the sector was announced in October 2020, applications opened in February 2021, but substantial payment delays were reported as late as May 2021). The authors also note the contribution of private and non-profit organisations in supporting ECCE facilities to reopen, another recurring theme throughout the pandemic when government systems often failed to act timeously.

Providing adequate financial and administrative support to this sector should be a policy priority going forward, not just because of the importance of (quality) ECCE for children’s development and performance at school, but because of the additional spin-offs in addressing gender inequality in the labour market. The sector is labour-intensive and employs predominantly women, and the availability of affordable childcare also helps free up women’s time to join the paid economy. The NIDS-CRAM data clearly showed that in households where children had not returned to an ECCE programme, the burden of care fell to a woman in the household in the vast majority of cases (Casale & Shepherd, 2022).

Another important policy imperative highlighted by Wills & Kika-Mistry (2022) is the need for more frequent collection of data on the sector to be able to track its progress. Indeed, this call applies to a number of the phenomena tracked by the NIDS-CRAM survey. An important feature of the collection of papers in this issue, is that in interrogating the NIDS-CRAM data and its limitations, the authors also highlight where future or official data collection initiatives should be focussed and where they could be improved. While the health impacts of the COVID-19 crisis may be abating, the socio-economic effects are ongoing and will be felt for some time. Continued efforts to measure and document these are crucial.

3. Reflecting on the way forward

The overwhelming experience of South Africa during the first year of the pandemic was one of profound loss and the rise of precarity. One year into the pandemic, vaccine access and uptake became the single most important focus of public policy to transition out of the crisis, with South Africa’s mass vaccination campaign starting in May 2021. NIDS-CRAM provided the first nationally representative information about vaccine use, hesitancy and intentions. Important papers (not in this volume) by Kollamparambil, Oyenubi & Nwosu (2021) and Burger et al. (2021) showed the extent of vaccine hesitancy and its determinants. In the region of 24-29% of respondents in Waves 4 and 5 reported being vaccine hesitant¹⁰, with concerns around vaccine efficacy, safety and side effects; distrust of government; and misinformation all contributing factors (Burger et al., 2021). Subsequent to this, SALDRU implemented the COVID-19 Vaccine Survey in late 2021 and early 2022 (Eyal et al., 2022), from which we have learnt more reasons why stated willingness to be vaccinated has not translated into vaccine uptake, and South Africa’s fully vaccinated rate in June 2022 remained at a mere 31.9% (Source: ‘Our world in data’). The spread of false information via social media and other channels has been particularly destructive, and much effort needs to be expended on correcting this.

At the same time, more effort needs to be directed to how best South Africa as a nation can learn from this crisis and develop a more agile governance approach to future crises. During the beginning of the COVID crisis, the government often failed to act fast enough, administrative systems were bulky and flawed at times, and the private and NGO/NPO sector had to step in to assist communities before catastrophic outcomes ensued. The ongoing issues with the timely payment of the SRDG is more evidence that this problem persists. Augmenting social protection via policies like universal basic income grants can pay tremendous dividends, but finding the policy levers to stimulate economic growth will always be among the most important objectives, for it is the most sustainable way to reduce the debt to GDP ratio.

It is important to note that the international community has been very active in developing policy recommendations to aid governments and civil society to make decisions about how best to approach future crises. Adam (2020), Caduff (2020) and Manzo (2020) very quickly pointed out the dangers of responding to the pandemic on the basis of epidemiological information alone, and stressed the need for complex social networks to filter into the thinking of the policy response. For South Africa, that means not going into extreme forms of lockdown associated with the beginning of the COVID-19 crisis, but rather nurturing an agile and differentiated approach to regulating social and economic activity, and only when absolutely necessary.

The Commission on Global Economic Transformation (CGET, 2021) was established by the Institute for New Economic Thinking (INET) in 2021, bringing together major macroeconomic thinkers to develop proposals on how the world can better deal with the negative shocks of COVID-19. Three key policy areas emerged, namely (1) mass distribution of vaccines, (2) globally subsidised expansionary fiscal policy, and (3) globally coordinated debt service restructuring. CGET (2022) also focussed specifically on the impact of COVID in Africa, conducting qualitative interviews with representatives from many countries on the continent about their experience of the crisis in their countries.

Most recently, the International Science Council (ISC, 2022) created policy recommendations that bridge both the local and the global, in a report entitled: ‘Unprecedented & unfinished: COVID-19 and implications for national and global policy’. They’ve developed recommendations for policy that cover global equity; understanding risks; trust and public mobilisation; science diplomacy; capacity development for science advice and resilience building; multilateral system reform; and investment in policy learnings (see the Appendix for a synopsis of key recommendations in each of these domains). There is much to be learnt from these various papers and policy documents, and South African policy-makers should take advantage of some of the major lessons emerging from the crisis.

4. Conclusion

There is no doubt that South Africa and the world have witnessed a profound crisis with COVID-19, on a magnitude not seen for several generations. It is abundantly clear from the evidence available for the first year of the pandemic that the true impact was not just on physical health, but also reflected in jobs lost, education learning losses, child care and nutrition deficits, and poorer mental health outcomes, among others. NIDS-CRAM has been one of the most powerful informational instruments to assist the country understand the consequences of the decisions made and the policy approaches taken to deal with COVID-19.

As the preceding sections have highlighted, the process of collecting and analysing the data itself, especially under such difficult circumstances, also offered many important lessons for future data collection efforts in times of crisis. Key among these was that it is crucial to produce reliable high-frequency data (and with a quick turnaround) during periods of substantial upheaval and volatility, so that it can be used by policy-makers to inform real-time decision-making. Being able to design a purpose-driven survey to deal with the pressing issues of the time, rather than being bound by the need for international or temporal comparability, proved invaluable. At the same time, however, it is important to recognise that there are limits to the kind of information that can be accurately collected in a short, telephonic survey of this kind with a relatively small sample size of adults, and that the continued collection of large nationally-representative data remains essential.

What the NIDS-CRAM survey has shown us is that the overwhelming outcome of the crisis has been a reinforcement of inequality in the country, already one of the most unequal in the world in terms of income, wealth and implicit opportunity. Those with access to formal sector employment have generally been better off than those without; those with access to private education have generally been better off than those without; those with access to private health care have generally been better off than those without; those living in formal areas have been better off than those in informal shack settlements and other poor neighbourhoods; those with access to the internet have generally been better off than those without; women have borne a disproportionately negative impact, and with that the children of this country.

However, there are some important learnings and successes to take out of our collective experience. The private and NGO sectors stepped up to support government relief efforts in a way that hasn’t been seen before in the post-apartheid era. The rapid mobilisation and expansion by government of social protection was a notable success, even if delays in the implementation of specific instruments reduced their immediate efficacy. The bold fiscal approach to the crisis was in line with international best practices, but we will need to be much bolder still in our efforts to reduce hunger and destitution and manage the consequent debt that will arise. The opportunity for a great reset is indeed now fully present. Let us seize the moment to alter South Africa’s socio-economic trajectory to one more equitable and less capricious.

Acknowledgements

The authors would like to acknowledge the massive contribution of Nic Spaull (Stellenbosch University), founder and Principle Investigator of the NIDS-CRAM project. Nic conceptualised the project, brought together the team of academics and data scientists, and worked tirelessly to ensure that each subsequent wave of data was released to policy-makers, the media and the broader public as quickly as possible. A project of this size would have been impossible to pull off without the support of many other parties. These include the excellent data team at SALDRU at the University of Cape Town consisting of Reza Daniels (also one of the Principal Investigators), Tim Brophy and Kim Ingle; the team leads; members of the steering committee; members of the working groups on sampling, data collection and quality, and the various subject themes; the data collection company AskAfrica; all the authors on the over 70 working papers and technical reports; the many reviewers and discussants; partners in government; and, importantly, the survey participants themselves. Without the hard work, dedication and collegiality from all of these stakeholders, the NIDS-CRAM project would not have been realised. Last but not least, the authors would like to gratefully acknowledge the generous funding from the Allan & Gill Gray Philanthropy Fund, the Federated Employers Mutual Education Fund, and the Michael & Susan Dell Foundation.

Disclosure statement

No potential conflict of interest was reported by the author(s).

Notes

1 For more details of the survey, including the sampling design and weighting approach, see Ingle et al (2021). Access to the data, working papers, and links for the webinars associated with each release, is publicly available at the survey website: https://cramsurvey.org/

2 Note that Levels 1–5 denoted progressively more restrictive lockdown regulations. ‘Adjusted’ lockdown levels denote the relaxation of some of the original regulations at each lockdown level.

3 See Appendix Table A.1 in Casale and Shepherd (2022) in this issue for a detailed discussion of which lockdown regulations were associated with each wave of NIDS-CRAM.

4 See Casale and Shepherd (2022), Appendix Table A.1, in this issue for details on school and childcare facility closures and reopenings during the various NIDS-CRAM waves.

5 In addition to insufficient statistical power, another limitation of the NIDS-CRAM survey is that due to interview length constraints, information on occupation and sector was not collected for the job held in February 2020. This means there is no pre-pandemic baseline against which to compare later occupational and sectoral distributions.

6 For example, the authors note that twice as many households reported running out of money to buy food in a single month, i.e. April 2020, as did in the entire year of 2017.

7 Shielding is measured as occurring when no children are reported to have gone hungry in the week preceding the interview, even though an adult in the same household went hungry during that week.

8 Depressed mood was measured using a two-question version of the Patient Health Questionnaire (PHQ-2), which asks respondents about how often in the previous two weeks they have ‘had little interest or pleasure in doing things’ and how often they ‘have been feeling down, depressed or hopeless’. These questions can be used to screen for depression and identify people at risk who require further evaluation (Hunt et al 2021).

9 Unlike with Grades R-12, in South Africa the ECCE sector largely relies on fees collected from caregivers to stay open, with only a small percentage of operators benefiting from state subsidies which are paid to registered providers on a child attending per day basis. This means that during lockdowns and government-enforced closures, when children couldn’t attend, most operators would have lost their primary source of income. Also, the majority of workers in the sector are not registered for UIF and initially would have been unable to claim from the UIF-TERS (Wills & Kika-Mistry, 2022). When they were allowed to reopen, the financial feasibility of many ECCE operators remained tenuous given costly safety protocols and child capacity limits.

10 In NIDS-CRAM Waves 4 and 5, respondents were asked to what extent they agreed or disagreed with the statement ‘If a vaccine for COVID-19 were available, I would get it’. The four response options were: ‘Strongly agree, somewhat agree, somewhat disagree, and strongly disagree’. Vaccine acceptance was defined to include both those who ‘strongly’ or ‘somewhat’ agree with the statement, while vaccine reluctance or hesitancy was defined as those who ‘strongly’ or ‘somewhat’ disagreed, as well as those who said that they did not know (Burger et al 2021).

Appendix:

Policy recommendations to mitigate long-term impacts of COVID-19 and prepare for future crises (ISC, 2022)

Table

POLICY / ACTION	MAIN RECOMMENDATION		ACTIONS
Global equity	Focus on reducing growing inequalities affecting societies’ capacity to cope with COVID-19	1.	Improve health system access, capacity and adaptability
		2.	Support the recovery of education
		3.	Support the care economy
Understanding risks	Review and reframe the way risk is assessed and integrate it more formally into policy development	1.	Focus on consequences, not just specific risks
		2.	Reframe long-term risks around actionable measures to address acute needs
		3.	Remain vigilant to other existential risks,
		4.	Don’t deprioritize sustainability
Trust and public mobilisation	Build trust through coherent decision-making and reliable information	1.	Ensure trust through transparent decision-making and information flows
		2.	Mitigate misinformation and disinformation with planning and trust
		3.	Adapt the crisis management approach as the situation changes
		4.	Change definitions of success rather than focusing just on case numbers and deaths
Science and science diplomacy	Prioritize science collaboration and diplomacy in times of peace, so that research can be promptly and equitably mobilised in a crisis	1.	Invest in R&D and sharing of benefits for the public good
		2.	Ensure recognition of the work of scientists in lower- and middle-income countries
		3.	Support further developments in open science to ensure equitable access
		4.	Enhance science advisory mechanisms at both national and multilateral levels
Capacity development for science advice and resilience building	Ensure broad capacities in science advice, evidence synthesis and brokerage	1.	Invest in an effective array of evidence synthesis and brokerage capabilities
		2.	Build capacity and support advice sharing with low-income countries
		3.	Plan for science communication, science literacy and ‘risk listening’
		4.	Establish a UN Science Advisory Board
Multilateral system reform	Reform the multilateral system to enhance international cooperation and regional responsiveness before and during crises	1.	Address structural weaknesses in the WHO and other multilateral stakeholders
		2.	Establish multiple coordinated processes to mobilise the wider national, regional and international community at earlier stages of a crisis
		3.	Support the role of civil society and the private sector in reducing social deficits
		4.	Improve international agreements in view of COVID-19 responses
Investing in learning	Increase capacity for policy learning at local, regional, national and international levels	1.	Analyse the role of people and their mobility in zoonotic outbreaks to understand how human actions can exacerbate effects
		2.	Analyse what went wrong in events, focusing on developing understanding and mechanisms to address future risks
		3.	Seek out multiple kinds of data and knowledge to learn the right lessons
		4.	Improve understanding internationally through cooperation in the multilateral system