Secondary datasets are available on how the pandemic has impacted sub-Saharan Africa countries, how governments have responded and how urban transport systems have been impacted. Secondary health data on COVID-19 infections, fatalities, vaccinations and excess deaths were sourced from the University of Oxford-affiliated Our World in Data, and the Johns Hopkins University-affiliated Centre for Systems Science and Engineering and Worldometer. Secondary data on the severity of lockdown restrictions were sourced from the University of Oxford-affiliated Our World in Data. Secondary data on changing activity participation and travel patterns were sourced from Google COVID-19 Community Mobility Reports. Calendar date data records were used to fuse datasets.

The two-wave panel survey of sub-Saharan African urban transport planning practitioners and academics sought both retrospective and prospective perspectives on the impacts of COVID-19. A Delphi survey was identified as a suitable method for this objective. The Delphi technique was originally developed in the 1950s by the Rand Corporation in the United States for technology forecasting (Dalkey 1969). The method involves a sequence of two or more surveys used to gather the informed opinions of a panel of experts. The method is especially useful in forecasting exercises in which expert opinions are the only available source of information. Expert panel interaction is structured around rounds of response and feedback, providing opportunities for individuals to modify their earlier views and for groups to shift towards a consensus. In the later rounds, panellists are provided with the results of the earlier round so that they can revise their original opinion if they are influenced by the opinions of others.

For the Delphi method to be effective, the anonymity of panellists is essential. Advocates of the method hold that direct confrontation can either lead to the irrational defence of a position once it is taken (to avoid losing face) or to a predisposition to be swayed by persuasively stated opinions of others without sufficient independent thought (eds. Linstone & Turoff 2002). Anonymity is argued to promote independent and unbiased thought and to avoid ‘bandwagon’ and ‘halo’ effects.

The panel of experts was recruited from a network formed through the Volvo Research and Educational Foundations’ Mobility and Access in African Cities Programme. Of the 37 experts invited to participate in the study (14 women and 23 men from 16 countries), 15 (from 10 countries) agreed to participate. The panel was comprised of university researchers (10), nongovernmental organisation practitioners (2), private consultants (2) and a government official (1). The home cities of the Delphi panel included Accra, Blantyre, Cape Town, Dar es Salaam, Freetown, Harare, Kampala, Kigali, Kumasi, Lagos and Nairobi. Fifteen experts completed the first wave (circulated on 29 July 2021) and 14 (93%) completed the second wave (circulated on 31 August 2021).

The purpose of the first-wave survey was to search for common positions on the current and future transport impacts of COVID-19 in different countries and regions. Through an online questionnaire, it was intended to provide insights into the observed impacts of COVID-19 on travel behaviour and trip substitution, operations, livelihoods and poverty and possible long-term disruptive effects on future public transport service supply, travel behaviour and how impacts might vary across socioeconomic and demographic groups. The second-wave survey was comprised of summary feedback on first-wave responses and an accompanying further online questionnaire. The purpose of the second-wave survey was to seek consensus on propositions regarding the future impact of COVID-19 on urban transport and to explore policy implications for aiding public transport recovery and mitigating social exclusion. Research ethics approval (20210708223855) was obtained from the Ethics in Research Committee in the University of Cape Town’s Faculty of Engineering and the Built Environment.

Google has assembled comparative quantitative data on the impacts of lockdown restrictions on trip-making from the location tracking functionality of smartphones (and other mobile devices), with trip purposes imputed from land use geography information overlays. The shortcoming of these data is that they are limited to a subset of the population that uses smartphones. In countries with lower household incomes, this means that the data are skewed towards wealthier socioeconomic groups. Whilst many, if not most, people in the selected sub-Saharan countries own mobile phones, smartphones are owned by a minority. For example, in Rwanda, the latest comparative data indicate that there were 76 mobile phones per 100 people (in 2019), but only 9% of the population owned a smartphone (in 2017) (Onkokame, Schoentgen & Gillward 2018; Silver & Johnson 2018).

With the above caveat in mind, changes in trips to workplaces compared on a percentage basis before and after lockdown restrictions were imposed are illustrated in Figure 2(a). The before-lockdown baseline (i.e., the 0 value on the vertical axis) was set as the median value for each day of the week, calculated from daily measurements over a 5-week period between 03 January and 06 February 2020. The data suggest that, in the short term, reduction in trips to workplaces varied in scale across countries. South Africa and Zimbabwe, for instance, saw a reduction of around 60%, whereas in Mozambique and Tanzania, the maximum reduction was considerably lower (at around 25%). The reduction was short-lived in some countries (e.g., Mozambique and Zimbabwe returned to pre-lockdown levels within 3 or 4 months), whilst others remained below the baseline for up to 20 months (e.g., South Africa). The spiked reductions of Uganda, Rwanda and South Africa mirror the arrival of second, third and fourth infection waves (see Figure 1).

Changing frequencies for two categories of shopping trips are tracked in Figure 2(b, c). The figure reveals that shopping trips for essential items returned to baseline frequencies much sooner than trips for nonessential items. Figure 2(b) plots trips to obtain food and medication essential for household survival. Short-term reductions in essential shopping trips ranged between 25% and 60%, with a return to, or a surpassing of, baseline levels occurring within 3 and 13 months. After 18 months, all of the selected countries had exceeded baseline trip frequencies by between 10% and 80%. Figure 2(c) plots trips to purchase nonessential items and to participate in other discretionary recreational activities. The figure reveals short-term reductions in nonessential shopping (and recreation) trips ranging between 25% and 70%, with a return to baseline levels occurring later than essential shopping trips (but slightly earlier than trips to workplaces) for most of the countries observed. For all but Uganda and South Africa, baseline trip-making had been exceeded after 18 months.

Figure 2(d) presents the corollary of parts (a – c). It illustrates that as trips to out-of-home activities were foregone (and to some extent substituted for remote work, education, social or entertainment activities), the amount of time people spent at home increased. Apart from Angola, Mozambique and Zimbabwe, all countries continued to show higher-than-baseline time spent at home, i.e., 22 months from when lockdown restrictions were first introduced.

Figure 3(a, b, c) presents the perceptions of the Delphi panellists provided in the first-wave survey, regarding the impact of lockdown restrictions on travel patterns. Panellists from countries with the least stringent regulations (Tanzania and Malawi) reported the least impact (see part a). Half of the panellists (47%) were undecided on whether the segment of individualised private mode use had increased (see part b), but half (47%) perceived the segment of walking and cycling to have increased (see part c), which is consistent with other observations in sub-Saharan Africa (Bird, Kriticos & Tsivanidis 2020). These perceptions refer to the relative segments of modal splits, not to an increase in trip generation, which is likely to have declined (Barbieri et al. 2020, 2021).

The most common public transport restrictions in the 11 home cities of the Delphi panellists were reported to be reduced vehicle occupancy (73%) and reduced frequency, if not cancellation, of services (60%). The impact of these lockdown restrictions can be observed in comparative secondary data on the trips to public transport stations presented in Figure 4. Apart from Uganda, Mozambique and South Africa, all the selected sub-Saharan African countries had surpassed baseline levels after 22 months. Using the same dataset, Brondum et al. (2021) showed that low-income and lower-middle-income countries returned to baseline frequencies earlier than higher-income countries (in December 2020).

The decline in number of passengers indicated in Figure 4 had an impact on farebox revenue and service viability and operator compliance with lockdown restrictions. Brondum et al. (2021) note, as do Venter, Hayes and Van Zyl (2020), that the data may be skewed in countries where the origins of public transport trips are often not formal public transport interchanges, ranks, stations or stops. As a result, it is difficult to distinguish between the loss of passengers experienced by formal and informal operators.

Calnek-Sugin and Heeckt (2020) contend that COVID-19 undermined the business model for unsubsidised informal paratransit operators that are reliant upon farebox revenue for viability. Half of the Delphi panellists (47%) reported in the first-wave survey that sub-Saharan public transport operations became either less or much less viable with the other half indicating either no change (20%) or insufficient evidence to judge (13%). For South Africa, Fobosi (2020), Jennings (2020), Luke (2020) and McLachlan (2020) all report that informal paratransit operators incurred significant losses. This led to various forms of protest and widespread non-compliance with a 70% vehicle occupancy restriction, and the national government eventually removing the vehicle occupancy restriction for informal paratransit operators. Venter et al. (2020) suggest that removing the restriction allowed passenger ridership in informal paratransit services to recover more quickly because the restriction largely affected a captive market. By contrast, formal bus and train services, often serving markets where passengers have greater choice, experienced an almost tenfold reduction in ridership. Luke (2020) and Baffi, Boudet and Mene (2021) note that sedan and motorcycle electronic hailing service providers experienced viability constraints as well.

Figure 3(d, e, f) presents perceived enforcement and compliance in the 11 home cities of the Delphi survey panellists. The figure illustrates that some governments took the opportunity presented by lockdown conditions to increase their regulatory control over public transport operations (see part d), but that this was not necessarily matched with enforcement capacity (see part e). Most panellists (53%) agreed that compliance with lockdown restrictions amongst informal paratransit operators was low (with 27% indicating they had insufficient evidence to judge).

Emerging fragments of socioeconomic data indicate that the unintended impacts of lockdown restrictions on economic livelihoods have been devastating. Many economies have shed jobs in large numbers and entered into recession, which will have impacted activity participation and mobility as well as the incomes and expenses (and societal inclusion) of households in complex ways. Figure 3(g) presents the perceptions of the Delphi panellist on the impacts of the pandemic on poverty. There was near consensus that the proportion of the urban population experiencing acute poverty has increased. Drawing from several secondary data sources, Teachout and Zipfel (2020) estimated that an additional 9% of the sub-Saharan African population fell into extreme poverty as a result of the pandemic and associated lockdown restrictions. They estimated that, at the end of the first 8 weeks of lockdown, 19% of the population could no longer afford their pre-COVID-19 level of food consumption, and 4% of the population was severely food deprived.

Most of the available data on the economic welfare impacts on sub-Saharan African households are focused on South Africa. Insights into the extent of job losses can be derived from several metrics. Spaull et al.’s (2020) analysis of the National Income Dynamics Study Coronavirus Rapid Mobile (NIDS-CRAM) survey data estimated that between 2.5 million and 3.6 million South Africans lost their jobs between February and April 2020. Adding a spatial dimension, Turok and Visagie (2021) found that the percentage of adults in paid employment living in ‘suburbs’ dropped from 58% in February 2020 to 52% in June 2020 and that in ‘townships’ dropped from 51% to 42% over the same period. Carlitz and Makhura’s (2021) analysis of Statistics South Africa survey data found that 2% of permanently employed respondents reported losing their jobs in the first 6 weeks of lockdown, whilst 5% reported having to close their businesses. For respondents who remained employed during this period, 21% reported reduced income.

Perhaps the most tangible impact on daily household activity schedules was the requirement to stay at home. In a (n = 1056) consumer panel survey, undertaken as part of an international comparative study (Balbontin et al. 2021), Venter, Van Zyl and Cheure (2021) found that in November of 2020, 40% of employees were working 5 days a week from home. However, as confirmed by the Delphi panellists (see Figure 3[h]), remote working during COVID-19 was only a possibility for a wealthy minority. Using the same residential location categories as Turok and Visagie (2021), Carlitz and Makhura’s (2021) analysis of Statistics South Africa survey data found that in ‘suburbs’ 88% of respondents were able to work from home, whereas in ‘townships’ only 5% were able to do so. As a result, low-income households were less able to comply with lockdown restrictions (Carlitz & Makhura 2021; Pedan & Kobusingye 2020). Aside from the need to obtain food and water, in a (n = 20990) telephonic survey of 20 African Union member countries, PERC (2020) found that the most commonly cited barriers to stay-at-home compliance were fears of running out of money and losing a job.

Figure 5(a) presents the perceptions of the Delphi panellists regarding which socioeconomic and demographic groups experienced the greatest impact on their welfare. Low-income households and women top the ranking. Numerous authors have argued that women were more likely than men to have lost their jobs, and that this hardship was compounded by increased demands to provide family care (Casale & Posel 2021; Jennings, Allen & Arogundade 2020; Jennings & Arogundade 2021; Porter et al. 2021). Analysing the NIDS-CRAM data from a gender perspective, Casale and Posel (2021) found that employment of adult South African women declined by 10 percentage points between February and April 2020 (from 46% to 36%), whereas adult male employment dropped by 5 percentage points (from 59% to 54%).

The Delphi panellists highlighted reduced accessibility (particularly to work, health care and education activities) as a feature of the impact of lockdown regulations on economic welfare (see Figure 5[b]). Disproportionality in impacts on accessibility is illustrated by Bird et al. (2020), who estimated that lockdown restrictions in Nairobi resulted in five times fewer job opportunities within 1 h for people reliant on nonmotorised modes than people with access to cars. The lockdown period may therefore have lasting economic effects that make vulnerable and excluded segments of the population more vulnerable and excluded.

Bem’s ‘self-perception theory’ provides a useful lens through which to consider the long-term impacts of temporarily imposed lockdown restrictions on patterns of trip generation and trip substitution (Bem 1972). Applying this theory, the imposed (or at least encouraged) increase in remote activity participation during more stringent lockdown periods – provided the experience is good – can make attitudes towards remote working, schooling and shopping more positive and subsequently lead to enduring trip substitution after activity destinations are fully opened and restrictions on movement are removed.

Based on first-wave survey perceptions, a proposition was formulated to test this theoretical perspective in the second wave. The first-wave survey did not establish a majority view on whether travel behaviour patterns would return to what they were before the pandemic (50% yes, 36% no and 14% undecided; see Figure 6[a]). In contrast to studies elsewhere (Barbieri et al. 2021), there was also uncertainty on whether individualised private motorised transportation would increase (21% yes, 28% no and 50% undecided; see Figure 6(b)) and on whether there would be greater online shopping and personal business transactions after COVID-19 has passed (34% yes, 33% no and 33% undecided; see Figure 6[c]).

The first-wave survey nevertheless established a consensus view that a relatively small portion of a city’s population would have the resources to engage productively with remote activity participation platforms (see Figure 3[h]) and a majority view that the activities most likely to be participated in remotely would include work and business meetings (identified by the most panellists), followed by shopping and tertiary education. A proposition was formulated as follows:

There was no disagreement with this proposition: 12 (86%) panellists agreed (43% agreed strongly). The two panellists who abstained were based in Nairobi. The accompanying explanation indicated that there are confounding contextual dynamics in Nairobi that make it difficult to predict the causes of future changes, even if it is evident that remote activity participation has increased and congestion levels have returned to base conditions:

Alonso’s ‘land rent theory’ provides another theoretical perspective from which to consider long-term disruption to trip patterns (Alonso 1964). Applying this theory, the emergence of large-scale ‘hybrid’ work practices and increased online retailing, entertainment and dining as a result of COVID-19 stay-at-home restrictions might decrease the sensitivities of some business and commercial land uses to the cost of physical access. This would enable them to survive on less accessible, cheaper land (or at least to reduce their land area requirement) and thereby disrupt the ‘bid rents’ of competing land use activities in city centres. This might lead to new land uses emerging in the place of some shops, offices, theatres, restaurants and parking garages, which would in turn change the distribution of trip attractions and productions across cities and alter travel patterns.

Based on first-wave survey perceptions, a proposition was formulated to test this theoretical perspective in the second-wave survey. The first-wave survey did not establish a large majority view on whether businesses would maintain greater online or virtual business operations after the pandemic (53% yes, 40% no and 7% undecided) and whether this would impact city centre land rents and land use distributions. The two diverging perspectives are illustrated as follows:

As indicated earlier, the first-wave survey did nevertheless establish a consensus view that a portion of a city’s population will have the resources to engage productively with remote activity participation platforms and that economic impacts have been severe. A proposition was formulated as follows:

There was consensus on this proposition: all panellists agreed (29% agreed strongly). One panellist did, however, offer the following caveat:

Gwilliam’s ‘regulatory cycle’ provides a useful lens through which to consider the long-term disruptive effects of lockdown regulations on public transport operator attrition and changes in service mix (Gwilliam 2008). In terms of this conceptualisation, the reduced financial viability resulting from reduced ridership and farebox revenue during the pandemic, if unmatched by sufficient compensatory state subsidy, might nudge a change in cycle phase and lead to a growth or decline in informal operators.

Based on first-wave survey perceptions, a proposition was formulated to test this theoretical perspective in the second wave. The first-wave survey did not establish a majority view on whether public transport operations will recover to prepandemic levels (47% yes, 40% no and 13% undecided; see Figure 6(e)). It did, however, establish a (weak) majority view that, despite efforts to assert greater control during the pandemic, the regulatory and enforcement capacity of public transport authorities would not be strengthened in the long term (53% agree, 20% disagree and 27% undecided; see Figure 6[d]) and that there would not be a thinning of illegal or unregulated public transport operators (53% agree, 20% disagree and 27% undecided; see Figure 6[f]). Common perspectives on paratransit recovery are illustrated as follows:

A proposition was formulated as follows:

There was no disagreement with this proposition: 10 panellists (72%) agreed (43% agreed strongly). However, there were three panellists who were undecided, and one panellist abstained. The accompanying explanations indicate that the reason for doubt was an absence of formal public transport undertakings in the city to compare paratransit with:

Hägerstrand’s ‘time geography’ provides a useful lens through which to consider reduced mobility and access and associated impacts on livelihoods and economic and societal inclusion (Hägerstrand 1987). Applying this theory, the increased ‘capability constraints’ amongst the most vulnerable and marginalised households (i.e., the reduced ability to pay for private and public mobility), resulting from job losses and reduced incomes stemming from the pandemic’s impact on the local economy, might reduce activity participation ‘domains’ and increase social exclusion.

Based on perceptions expressed in the first-wave survey, a proposition was formulated to test this theoretical perspective in the second-wave survey. The first-wave survey did not establish a majority view on whether access and mobility (and associated livelihood) impacts would be disproportionate across socioeconomic groups (36% yes, 29% no and 36% undecided). The two diverging perspectives are illustrated as follows:

The first-wave survey established a majority view that the proportion of the urban population experiencing acute poverty has increased (73% yes, 14% no and 13% undecided) and that women have been more adversely impacted by COVID-19 with respect to mobility and access than men (60% yes, 14% no and 27% undecided). A proposition was formulated as follows:

There was majority agreement with this proposition: 10 panellists (72%) agreed (22% agreed strongly). Two panellists disagreed. The accompanying explanations suggest that the basis for disagreement was limited disruptions during less stringent lockdowns:

A further two panellists abstained. The accompanying explanations suggest that households developed coping strategies to deal with transport system disruptions: