Was Lockdown Racist?

2.1 Defining “Lockdown”

Soon after C-19 emerged in early 2020, many nations, including many in Africa, adopted similar packages of responses including, as a common theme, prohibitions on leaving the home except for specific reasons.³ They were initially implemented as part of a “suppression” strategy, being an effort to bring the reproductive number R below 1.⁴ The contrast was with “mitigation” strategies which aim to reduce R but not below 1. This sharp initial contrast quickly evaporated, however, because actual lockdowns often did not have the dramatic effect that was originally hoped for, at least in many parts of Africa where lockdowns had been implemented extremely hard and early.⁵ (Sceptics might say they did not work; proponents might say they were late, bungled, or disobeyed. For present purposes, it does not matter.) It was soon clear that R < 1 is not a magic threshold,⁶ and the central policy distinction altered somewhat, into a contrast between policies that were at once severe and population-wide, and those that were either less severe but still applied across a population (for example, limits on occupancy of restaurants) or were applied to targeted groups (for example, limiting visitors’ access to hospitals).

Lockdown was characterized by severe restrictions on conditions under which it was permitted to leave the home, as well as what one was permitted to do when out, but the nature of these conditions varied widely. So did compliance. It is therefore difficult to articulate an abstract definition of lockdown as a well-specified public health intervention. It is also undesirable for present purposes, since to do so would strip lockdown of its historical context. Lockdown is not like immunization: a familiar and recognizable public health intervention, oft-repeated, with well-defined characteristics and goals alongside varying local characteristics. C-19 lockdowns were novel, their goals were defined in multiple ways, their mechanism of action was understood in different ways, implementation was extremely various, and success criteria were unclear and disputed. Nothing like the C-19 lockdown had ever happened before; possibly, never will again. It is unreasonable to contest that the public health responses to C-19 that are labeled “lockdowns” were unprecedented in geographical scope and duration. They were also unusual in being prompted not locally, by community members falling obviously ill, but remotely, by scientific advice, guidance from international organisations, and events in other nations.

For these reasons—the wide variety of forms they took, and their historical uniqueness—it is appropriate for present purposes to consider the lockdowns of C-19 as parts of a larger episode. In keeping with this historical perspective, when “lockdown” is used in the abstract, it should be understood, not as any specific package of measures (for there is none), but as a policy trope with a particular historical context, without specifying the regulatory content of “lockdown” too exactly. To object to the present argument on the basis that “lockdown” is ill-defined would be akin to objecting to a discussion of life in the trenches of The Great War on the grounds that we lack a definition of “trench”; alternatively, to rejecting to the claim that nineteenth-century slavery was racist because slavery laws were differently enacted in different places. From our perspective, the two important features of lockdown are: that they are regulations (as opposed to recommendations) imposing major restrictions on leaving the home and carrying on activities outside the home (as opposed to minor restrictions); and that they were applied across entire populations of large areas such as nations (as opposed to being targeted at particular segments or subpopulations of the population of a large area). It is these features of lockdown that our argument relies upon.

2.2 Defining “Racist”

In 2002, race scholar Lawrence Blum argued at book length against what he perceived as a growing tendency for all kinds of actions and institutions to be categorized as racist merely because they contacted race or racial differences in some way.⁷ He argued that “racism” should be understood as arising only when two ingredients are present: inferiorization of one group due to supposed biological inferiority, and antipathy, which is a set of attitudes: “bigotry, hatred, and hostility.” He argues that to call “lesser racial ills and infractions” racist is to blur an important moral distinction between truly “heinous” acts and those that arise from ignorance, or a lack of critical awareness, or other attitudes that are surely not ideal but are less morally blameworthy. Others have expressed related views.⁸

A cynical response sees the effort to draw such distinctions as an exercise in apologetics. According to the cynic, articulate academics in elite institutions devise these distinctions so that the crude utterances and behaviors of poor and ill-educated racists can be deplored, while their own actions and advantages may be excused. Yet social elites, many race scholars included, benefit far more greatly from racist social structures, and do far more to preserve them, than abusive and violent, but essentially impotent, thugs. So goes the critique, much to the horror of scholars in the liberal tradition, who find themselves deplored as much as or even more than the open racists from whom, on the skeptical view, they have devoted their careers to distinguishing themselves.

On a more radical perspective, racism is not seen as requiring intent, knowledge, or any other kind of awareness of the nature or consequences of action;⁹ nor is it seen as requiring positive actions at all. Even ignorance can be racist, on the radical view.¹⁰ Racism is ascribed to rules and entire institutions even when they do not mention or address race explicitly.

The idea that ignorance itself can be racist is stronger than we need for our argument. What we require is a notion of negligent racism. The term “negligent” is used analogously to its deployment in English and American tort law, where it indicates that a duty to take care exists, that the required standard of care has not been met and thus the duty was breached, and that this has caused harm.¹¹ We call a policy, practice, institution, or other entity negligently racist when:

The legal analogy invoked by the term “negligent” is an analogy only: it is not our intention to motivate an actual development of tort law. The point is to stress that non-intentional actions may be subject to justice in developed legal systems. Moreover, the form this justice takes is often not retributive punishment of the perpetrator, but compensation of the victim; yet it is still a matter of justice. In negligence, as in many other legal contexts, compensation depends on the material harm caused to the victim and not on the mental state of the perpetrator. In particular, compensation depends on the quantity of harm and on whether a reasonable person in the perpetrator’s position could have foreseen it. Thus negligence is objective. You may not have intended to drive your car into someone else’s while you glanced at the fuel gauge, but you did. You may not have foreseen damage, but you reasonably could have foreseen it: reasonable foreseeability is an objective standard in law and requires no mental state. You caused a certain amount of damage. You are thus liable to compensate the victim for the damage, no matter how sorry you may be or how much of an accident it was. The point of this kind of law is not to judge you, but to compensate the victim.

It may matter greatly to you that you were merely negligent because your conscience, your reputation, your relationships, your livelihood, and even your liberty may depend upon it. However, none of this is materially relevant to the victim (although they may also care about retributive justice as well as material loss). Thus to apply the idea of negligence to racism is not to trivialize racism, but to acknowledge the interests of the victim, which include the harm sustained, and are not confined to the perpetrator’s mental states. Whereas to obsess about distinctions between attitudes of the perpetrator is to privilege the interests of the perpetrator over the material interests of the victim, since those attitudes are relevant to the moral evaluation of the perpetrator but make no difference to the material harm suffered by the victim. Even if the victim cares about the attitudes of the perpetrator, the scope of the victim’s interests is wider, and includes the material harm. If we extend this line of reasoning, then, where the nature of an injustice is inextricably racial in that it inextricably involves a racially distributed harm, it should at least be eligible as a candidate for the label “racism.” Otherwise, the material interests of the victim are not acknowledged. The material interests of the victim matter to the victim; the fact that they are not features of the perpetrator is at the root of resistance to this move, yet this resistance serves the interests of the perpetrator and not those of the victim, for whom it matters not only what kind of moral wrong was done, but what harm was caused and whether it could have been avoided. Distinguishing the material and moral harms of racism for purposes of redress and punishment is useful, but distinguishing them as two distinct entities—one racism, the other something else entirely—is probably impossible, certainly useless, and possibly ill-motivated.

2.3 The Meaning of “Lockdown was Racist”

Lockdown was racist because it was negligently racist. Lockdown was negligently racist in that it satisfies our definition of negligent racism above. It satisfies the foreseeability, non-necessity, and race conditions respectively as follows.

• (i) 

  Lockdown reasonably foreseeably harmed a large number of people of certain races (foreseeability, which we demonstrate in §3);

• (ii) 

  The harm was unnecessary because viable alternatives existed and were epistemically accessible at the time (non-necessity, which we demonstrate in § 4); and

• (iii) 

  At least some of the harm arose because of race and not merely due to a correlation with other factors (race, which we demonstrate in §5).

It would also be reasonable to state the argument in terms of unfair burdens rather than harms. One might take any unfair burden to be a sort of harm. Alternatively one might require that a harm be materially detrimental (in a way that a mere burden is not) rather than simply unfair. We operate with the latter, stronger, assumption, because we believe that it can and thus should be established that lockdown was racist in a stronger sense of harming people by race, and significantly so, rather than merely burdening them unfairly.

3.1 The Negative Effects of Lockdown in Africa

Many countries in Africa are poor; much poorer than those countries in the Global North in which the majority of White people live, and certainly those countries that are the usual reference points for Anglophone academic philosophy. For example, in Malawi, GDP per capita according to World Bank data is $357 at time of writing; in the UK it is $39,532.¹⁵ The difference is by a factor of around 110. The population of Sub-Saharan Africa living in extreme poverty (less than $1.90 per day) was 40% in 2018, and the region accounts for two thirds of extreme poverty globally.¹⁶ Although poverty in the region is falling as a proportion, the number of people living in poverty is rising due to population growth.¹⁷ Over 80% of employment in Africa is informal.¹⁸

58% of the population of Sub-Saharan Africa live in rural areas,¹⁹ often very remote, and often engaged in agriculture, including some subsistence. Of the urban population, around half live in slums where overcrowding may be severe: for example, in Accra, Ghana, 53% of households live in a single room and even more depend on public toilets.²⁰ The majority of both rural and slum dwellers do not have piped water in the home.

It is obvious in advance that lockdown would be difficult to implement in many of these circumstances. Many African states lacked the resources and logistics to mitigate the economic effects of lockdown either on businesses or individuals. Economics cannot be separated from health in Africa (if they can anywhere). Poverty means that hunger, if not already present, follows immediately if income stops, and informal employment means that income stops immediately if work stops. The manual nature of most work means that it cannot be done from home. In agricultural areas, livestock and crops will perish if not attended to. In urban areas, food security was threatened due to supply chain problems and reliance on informal trading. Hunger did indeed increase globally, and the biggest increase was seen in Africa, with 21% undernourished in 2020.²¹

In conditions of overcrowding, especially in urban slums, staying in the home was also exceedingly difficult due to intolerably cramped and often hot conditions, and brought attendant dangers of domestic abuse.²² Even if the letter of a lockdown order were complied with in all these circumstances, there would still be numerous essential trips outside the home for water and sanitation purposes.

Health services across Africa were seriously disrupted by lockdown. Evidence continues to emerge to this effect and, unsurprisingly, points in one direction. Here are a handful of examples from a much wider pool.

• One study found that “minor consideration was given to preserving and promoting health service access and utilization for mothers and children, especially in historically underserved areas in Africa.”²³ (Neonatal conditions are a leading cause of death in the region.)²⁴

• In South Africa, “ART [(Anti-Retroviral)] provision was generally maintained during the 2020 C-19 lockdown, but HIV testing and ART initiations were heavily impacted” with an estimated 47.6% reduction in HIV testing in April 2020²⁵ when South Africa was in lockdown.

• In Addis Ababa, Ethiopia, total tuberculosis (TB) detection fell 11%; TB treatment success rate fell 17%; latent TB infection treatment fell 44.7%; community health workers’ engagement in TB detection fell 77.2%; and rifampicin resistance increased 27.7% between April-March 2019-20 and 2020-21.²⁶ TB Notifications in African region fell from 1.4m in 2019 to 1.37m in 2020, then went back up to 1.48m in 2021,²⁷ making Africa the only region with a higher level of TB notifications in 2021 than 2019.

• On 15 July 2022, WHO/UNICEF reported the “largest continued backslide in vaccinations in three decades” due to C-19. In 2022, measles cases in Africa are surged due to vaccination disruption.²⁸ This was anticipated by some African scientists in 2020, some of whom argued that the deaths prevented by continuing childhood vaccination outweighed risks posed by vaccination clinic visits;²⁹ such arguments did not win the day.

Regional evidence for the effect of lockdown on education in Africa is hard to find. A Malawi study estimates 2 years learning loss for 7 months school closure.³⁰ It seems fair to suppose that the circumstances of education in Africa mean that school closures would have a much greater effect there than in Europe and America.

There are many other respects in which lockdown had negative effects in Africa. Some of these are hard to evidence, notably negative effects on women, which are plausible but often unreported, as well as community violence and discrimination more generally.³¹ Others are complex, such as the differing political circumstances on the continent. Often, enforcement was unacceptably harsh: South Africa and Uganda both went through periods where recorded C-19 deaths were lower than the number killed by the security forces.³² Coups have proliferated on the continent starting in 2021³³; while it would be wrong to attribute these too easily to lockdown, they do illustrate the existence of delicate situations, in which the suspending rights and placing power in the hands of security forces ought to carry significant weight even in a pandemic context. There is far more to say, but this is not an empirical paper, and the point is overridingly plausible. Henceforth, we take it as established that lockdown in Africa had many severe negative effects on many of the people living there.

3.2 The Positive Effects of Lockdown in Africa

To pronounce lockdown harmful requires a consideration of its benefits. Heart bypass surgery causes significant harm: skin is cut, the respiratory system repressed, blood vessels taken from the legs and attached to the heart, and so forth; but to pronounce the procedure harmful is at best misleading if it saves the patient’s life. Likewise, lockdown caused harms in a narrow sense: people lost jobs, health services were disrupted, and so forth; but to pronounce them harmful overall would be misleading at best if they saved many lives.

Evidence that lockdown had not been a highly effective policy in Africa was available early on, but was cautiously expressed. For example, a 2020 study of the impact of lockdown on C-19 transmission in nine African studies muses that “The graphs show no obvious pattern.” The authors continue:

Drawing conclusions about the impact of lockdown measures on COVID-19 transmission is difficult for several reasons… Nonetheless, logic and evidence from elsewhere indicates that lockdown measures will have lowered the effective reproduction rate of the virus in these nine countries… However, it is plausible that lockdown measures could have helped increase COVID-19 transmission in the large and dense informal or semi-formal settlements of Africa.³⁴

The paper provides a balanced and comprehensive summary of the collateral effects of lockdown in Africa, positive and negative, and has justifiably been well-cited. But in a different context, the same results could have been presented quite differently. Here is an alternative summary: no positive effect was found to support the hypothesis of effectiveness, with some evidence of significant adverse effects, including an increase in the transmission of C-19 in some instances. This would be the default summary in more normal circumstances—and certainly if this were a study of a pharmaceutical intervention. No doubt the authors were aware of all this, but writing in the first half of 2020, it was not open to them to do more than venture bemused remarks. The summary box includes no reference to the negative impacts that form the subject of the bulk of discussion.

Since then, numerous papers and policy discussions have called for efforts to devise more appropriate measures for dealing with epidemics in Africa. Most remain deferential to the early lockdowns, which we are presently engaged in arguing were racist. Yet this is not because the evidential picture has changed.

There are good reasons to doubt that it will change. One reason is that the majority of the continent’s population comprises children and teenagers: the median age in Africa is 19.7,³⁵ while the median age of death in the UK due to C-19 was 83 (mean: 80.4) in January 2021.³⁶ Even if prevailing morbidities in Africa make younger people more vulnerable, the age association is extremely strong, and the protective effect (and thus benefit) of lockdown would be less in proportion to any residual association with age.

There are also reasons to doubt the scale of benefit in terms of disease transmission. In urban areas, overcrowding is a reason already mentioned to doubt the effectiveness of locking down.³⁷ In rural areas, on the other hand, sparse population and remoteness of communities will limit speed of transmission even without lockdown, providing a natural brake on transmission and thus reducing the scale of any impact.³⁸ Compliance levels were almost certainly low due to the extreme difficulty or impossibility of compliance, as well as difficulties in enforcement. Finally, some were lifted out of necessity while cases were still rising (e.g.: in South Africa).³⁹ For all these reasons, it is fair to say that lockdown offered minimal benefits to the majority of Africa’s people.

3.3 Lockdown in Africa Shows that Lockdown was Racist

We have shown that the policy of lockdown was harmful in Africa, but how does that amount to show the general proposition that lockdown was racist?

It was foreseeable that lockdown would be harmful in many low-income settings and that this would not be offset by benefits.⁴⁰ All the circumstances that were likely to cause negative effects of lockdown, and stymie positive ones, were known in advance. The negative effects of lockdown are not offset by corresponding positive effects and may be fairly said to have foreseeably harmed very many people in Africa.

It follows that lockdown foreseeably harmed very many Black people. Most likely it is a large majority, although we do not need this in order to conclude that condition (i) of pronouncing lockdown racist is satisfied.

Is this enough to establish a general proposition that lockdown was racist? Would it not be more appropriate to say that it was racist in Africa, and remain agnostic about elsewhere? Was it not the case, in fact, that Black people suffered much greater C-19 mortality than White neighbors in wealthy countries, where these arguments will not apply—and thus might lockdown there have been positively anti-racist, or at least have offered significantly more benefit and caused fewer harms to Black people there?

We have three responses. First, the response is mere speculation. One might just as well speculate about the racial profile of delivery drivers in wealthy countries, or home occupancy by race, or a host of other factors that might increase the harms and reduce the benefits of lockdowns. None of this gets us anywhere, however, without evidence. But systemic racism should make us skeptical about the prospects of discovering that lockdown was non-racist in wealthy countries. Agnosticism is reasonable; presumption in favour of lockdown is not.

Second, we invite the reader to consider whether the concern would arise if the argument had focused solely on the plight of Black people in the US or the UK. If, instead, we had argued that lockdown was racist by comparing Black and White experiences in America, would anyone object that we could not generalize because we had not considered Africa? We have never seen this objection raised to parallel arguments in those contexts. And yet there are over a billion Black people in Africa, many more than elsewhere, and indeed than there are White people globally. Africa matters. To doubt that one can establish a general proposition by reference only to Africa is, we submit, a form of prejudice.

Third, we have already explained that we treat particular lockdowns as part of a historical episode. In that episode, a certain pattern of response to C-19 was prominent. This episode involved clear directions of flow of information and policy, out from Europe and the UK in particular, to Africa. The World Health Organization was particularly influential, and misguided, in releasing a Joint Mission Report with China, which heaped praise upon the Chinese response and urged it globally.⁴¹ Against this background, the political feasibility of not implementing a lockdown was limited, since Africa is dependent on both China and the West in so many ways, and because the local scientific capacity to develop counterarguments to this policy or to develop a different one are so limited. This is not to undermine the public health operations in Africa. It is simply to say that their resources are dwarfed by, and often dependent upon, those in China and the West.

The most influential scientific voices early in the pandemic were from Europe. Who had not heard of the Imperial College London models? Who, on the other hand, can name a single model developed on the African continent? We cannot here trace the patterns of influence and knowledge production that led to lockdown happening in Africa. But there is no reasonable objection to the claim that the implementation of lockdown in the first half of 2020 was a global phenomenon. It is this phenomenon that we refer to with the general proposition “Lockdown was racist,” which we justify by showing how its instances harmed Africa.

4.1 Alternatives to Locking Down in Africa

We have characterized lockdown as significant population-wide regulatory restriction on activities outside the home. Alternatives exist to both elements: the degree and kind of restrictions; and their application population-wide.

Regarding degree and kind of restriction, it is obvious that a very wide range of options exist, as evidenced by the wide variation in actual regulatory packages remarked in §2. These are not all equally effective; they do not all impose the same degree of human cost; and they do not distribute their burdens in the same way.

Geographic containment, travel restrictions and the prohibition of large gatherings are likely to have had the largest effect on community transmission in many African countries.⁴² Their effectiveness is not limited by considerations relating to overcrowding or lack of amenities in the home, and, while they may impact livelihoods, the impact is not so complete as a requirement to remain inside the home. It is still possible to tend crops, feed cattle, work on a construction site, trade within geographical boundaries, and so forth. Schools may remain open. Hospitals may continue to operate. Travel restrictions may be open to essential food supplies. The banning of religious events may come with a significant emotional cost but it will not lead to cholera, illiteracy, or starvation.

The move from this package to a lockdown adds relatively little to the benefit of these measures for many people living in contexts of poverty, while greatly compounding the human costs.⁴³ A viable alternative, therefore, to lockdown in many parts of Africa would have been to implement those measures that were most effective, partly in virtue of being the least burdensome to people living in contexts of poverty.

Every C-19 control measure imposed a human cost, and it is unsurprising if measures did so unequally to some extent: after all, privilege is in part the ability to avoid misfortunes of this kind. However, not all imposed burdens unequally to the same degree. School closure, for example, is a paradigm case of a measure having unequal impact. School closures hurt young people more than people who have already received their education, working parents or other child-carers more than those who do not need to work, parents and others who have to work outside the home more than those who can work from home, families in smaller dwellings more than in larger, families whose schools are less well-resourced more than wealthier schools, and so forth. Children themselves are also less at risk of serious or fatal Covid, and to this extent, the benefit they derive is lower, although of course if they transmit it to members of their family upon whom they depend then they will suffer. School closure is one of the paradigm cases of a measure that, when applied equally to all, harms people unequally.

On the other hand, travel restrictions clearly harm wealthy people who need to travel for business, have expensive vacations booked, have homes in the countryside, and so forth. They harm poor people too, and perhaps more so: a trader who cannot access a customary market will be in trouble. But the balance is more even, arguably, than in the case of school closures, where wealth is directly correlated with the ability to mitigate the negative impact. The details matter, too: being unable to travel 5km in any direction is different from being prohibited from crossing a provincial border. There is no doubt that travel restrictions can be more or less intelligent. But this means the measures can be tweaked. In addition, the effect may be mitigated by individuals finding ways to adapt to the new circumstances. Closing schools is an absolute and blanket measure, by contrast, and does not admit of tweaking; and the only real adaptation available is home schooling, which is essentially impossible in many low-resource settings (it is difficult enough in high-resource settings, as many parents discovered). Thus travel restrictions tend to distribute their harms less unequally than school closures, even if both distribute their harms unequally.

Therefore, a package of control measures involving geographical containment, travel restrictions, and a ban on large gatherings amounts to a viable alternative. These measures had the largest impact on C-19 transmission in many African contexts and imposed much lower burdens than lockdown. (It quickly became clear that schools were not drivers of C-19 pandemic.⁴⁴) We have already adduced evidence that there was little additional benefit of locking down in many African contexts. We conclude that the aforementioned package consisted of a viable alternative set of population-wide measures.

Besides kind and degree of restriction, the other dimension in which lockdown can be varied is by targeting vulnerable populations as a means to reduce mortality. Opportunities for doing this in African contexts were not explored because there was a near-total lack of consultation in the devising of C-19 control measures. The dominant discourse concerned expertise. The idea of asking the inhabitants of remote African villages what to do about a global pandemic would have been derided in some quarters. There are also practical challenges of consultation, especially in a short timeframe. However, the lack of any apparent efforts is striking. There are many public health workers on the ground across Africa and many of these have deep connections in local communities. These could have been consulted by national and international authorities. The makers of a short documentary had no trouble contacting a village in Malawi, explaining what was known about C-19 transmission, and asking for their suggestions.⁴⁵ The village consisted in well-spaced huts, in which generations lived together; the leaders suggested placing the older community members together on one side of the village and the younger ones in others. Family ties were important; the leaders suggested one-one or small meetings outdoors with distances maintained. These inputs were easy enough to gain and possess an obvious logic in the context. They were publicly available when the film was made in March 2020. Yet they did not form part of national Malawian recommendations at any stage, probably because they had no applicability in the Global North settings that informed the lockdown policy.

We conclude that there were viable alternatives to locking down in Africa. We now turn to the other element of (ii), that the alternative was epistemically accessible.

4.2 Precaution

One might admit that there was a viable alternative but deny that this could be known at the time. One might then contend that, based on what was known at the time, and on the level of uncertainty, it was appropriate to implement costly measures as precautions, and that lockdown was justified as such. This argument receives a good treatment in the context of the United Kingdom with regards to liberty rights,⁴⁶ but that discussion tells us little about African contexts. Moreover, applying a precautionary justification to the infringement of liberty rights does not tell us much about whether precautions are justified when they cause detriment to welfare (except insofar as infringements of liberty amount to welfare detriments). Even if precautionary lockdowns were a justifiable infringement of liberty in the United Kingdom, as White et al. (2022) argue, this does not show that the human costs of precautionary lockdown in Africa were justified. We will now argue that both the epistemic situation justifying precautionary reasoning, and the negative outcomes that precautionary reasoning is supposed to excuse, were different in Africa.

The epistemic differences between African contexts and high-income contexts concern not the effect of C-19 itself but of lockdown. We argued in §3.2 that it is very doubtful that lockdown was sufficiently effective in many parts of Africa to justify itself, and even suggested that in certain contexts they might have increased C-19 transmission. In §3.3 we argued that this was foreseeable. These facts preclude lockdown from qualifying as a precautionary measure in Africa. It is not that more was known about C-19 in Africa than in the UK, but that enough was known about the context to make the measure inappropriate for consideration as a precaution. Even if precautionary reasoning was justified, it did not justify lockdown in Africa.

It is also important to note that a precautionary argument for the infringement of liberty rights is inadequate for the African context. Infringements of liberties do matter, especially in contexts of political instability, authoritarian government, corruption, and so forth. However, lockdown in Africa was also likely to cause very significant human suffering, much of it translating into mortality due to one or another cause, as discussed in §2. A precaution that has costs of these kinds needs to be justified accordingly, and where the precaution is supposed to reduce human suffering and mortality, there must be some effort to weigh up the human cost of the precaution against the human cost of lesser measures. By the argument of §4.1 this weighing up does not result in a favorable outcome for lockdown in African contexts, which foreseeably incurs excessive human costs relative to welfare gains in those contexts.

Whatever the merits of a precautionary argument against libertarian objections in the Global North, they do not justify lockdown against welfare-based considerations in Africa.

5.1 Objection: Lockdown is Anti-Poor but not Racist

This objection says that lockdown was not racist per se but anti-poor, with the correlation between race and poverty accounting for the negative effects of lockdown on certain racial groups. Sally Haslanger (2012) points out that any non-intent-based notion of any species of oppression will face the challenge of distinguishing correlates, such as poverty, which might be advanced as alternative explanations for whatever unfair outcome is in question. While we have not characterized negligent racism as oppression, the same point applies here. Haslanger (2012) requires that there be a “non-accidental” correlation between the membership of the oppressed group and the injustice of the oppression, which she characterizes as one that supports counterfactuals. Charles Mills requires that race must play a crucial causal role (2017: 49-71). Regarding lockdown, poverty obviously affects the way the burdens of lockdown are distributed, and it thus provides an obvious reason to suspect that the correlation between harms of lockdown and race is accidental (failing Haslanger’s test) and non-causal (failing Mills’).

It bears noting that this objection concedes much that we are keen to accept: in particular, that lockdown is inegalitarian and socially unjust. Both these conclusions follow from our intended conclusion that lockdown is racist, assuming racism is a form of inegalitarianism and of social injustice. Therefore one who presses this objection is in large part our ally, accepting our critiques of lockdown in Africa, and our strategy of showing a general claim by focusing on a region where the relevant properties are instantiated. But this ally sees the relevant properties as poverty or its aspects, and resists taking the further step of applying the label “racism.”

Assuming such a person has accepted the notion of negligent racism already motivated, then this must be because they reason that poverty is the relevant explanation of negative effects. An actual-world correlation exists globally between poor people of all races and negative effects of lockdown. A counterfactual relationship is plausible between poverty and lockdown—if people in Africa had been rich not poor then the effects would have been different. But there is a lack of plausible counterfactual relationship in respect of race: if the inhabitants of Africa had been White not Black then the effects would have been the same.

However, this objection relies on the mistaken assumption that non-accidental correlations necessarily support counterfactual dependence. Haslanger gives the objector too much when she says:

a correlation counts as nonaccidental because it supports certain kinds of counterfactuals; the idea is that the group’s being a group of Fs is causally relevant to the injustice. (2012: 328)

Effects of a common cause are non-accidentally correlated but do not (typically) support counterfactuals, and effects do not counterfactually depend on causes that pre-empt other would-be causes.⁴⁷ Moreover, a counterfactual test is particularly inapplicable for causal effects of race because, as epidemiologists have discussed at length, race counterfactuals are often nonsensical.⁴⁸ If this Black person had been White, would they have still had Black parents? Does that even make sense? Would they have grown up in a Black neighborhood, or have spoken in the same way, or gone to the same school? And even if these questions could be answered, which of them relates to what we want to know? Race is all these things bundled up; or at least, it is not clearly reducible to any one or ones of them.

When one considers it, similar remarks apply too to poverty: would a person simply have a higher income but live in the same house, would they have gone to the same school, and so forth? It is for this reason that social epidemiology is resistant to counterfactual frameworks for thinking about causation: the causes it is interested in just do not fit that framework.⁴⁹

The objection relies on separating the effects of race from the effects of poverty. Counterfactual tests for non-accidental correlation may be inapplicable here, but they do tell us something useful: that it is not possible to separate race and poverty in the way the objection requires. Of course there are many poor people who are not Black and many Black people who are not poor. However, where Black people are poor, their race is often part of the causal explanation. That is not to say, nonsensically, “If these people had not been Black, they would not have been poor.” It is to point to the causal history of their poverty. Blackness is part of the causal history of poverty for most poor Black people. The objector is correct that poverty is a mediating variable, to adopt that terminology, on the causal path between lockdown and harm. However, in the case of many Black people, Blackness is a cause for poverty. The situation is not an either/or one: there is no contradiction between saying that lockdown harms poor people and that lockdown is racist. Both are true. The relationship between race and the harms of lockdown is not a by-product of the role poverty plays; on the contrary, poverty is a product of, among other things, race, in those cases where race is in play.

An example will help to illustrate the situation. Figure 1 shows the first and second epidemic waves in wealthy suburbs near Cape Town, South Africa. The first wave was markedly lower than the second wave, consistent with a protective effect of lockdown. Figure 2 shows the same period for nearby townships (slums). The two waves are far more comparable, consistent with a limited effect of lockdown. This is consistent with lockdown having a much greater protective effect in suburbs than townships. Assuming this is so (and we have not, of course, presented a detailed causal analysis), can a legitimate explanation fail to mention race? The geography of apartheid persists in South Africa: almost no White people live in townships and White people are relatively over-represented in wealthy suburbs, as a direct consequence of racist policies and ongoing racial injustices. If poverty explains the differences (and a causal interpretation might of course be disputed) then race is part of the explanation too, since race is part of the explanation for the poverty.

Figure 1: cases in suburbs near Cape Town, South Africa.

Figure 2: cases in townships near Cape Town, South Africa.

The objector might resist on the grounds that even if causation is transitive, explanation is not. We reply: even if this is so, neither explanation nor causation is intransitive (Broadbent 2012), and causes that are not immediately proximate may be explanatorily relevant in some contexts, including some that involve moral evaluation. Evolutionary facts may be invoked to explain contemporary events despite their temporal distance, for example. The racial effect of lockdown is another such case, especially (but not only) because the causal relationship between race and poverty is not merely historical but ongoing, and this gives rise to moral duties to attend to it.

Finally, if poverty causes racism in the sense of providing tools to justify exploitation, then one can succeed in explaining only cases where the negative effects of lockdown were explicitly blamed on the supposed inferiority of certain races. There are none such in mainstream academic literatures or media. In any case, even if there were, this defense would be highly eccentric. It would be like defending racist beliefs among Conquistadors on the basis that they were a mere cover for the necessity of slaughtering indigenous peoples in order to carry away their gold. That is racism, of the kind specified by Blum (2002), since it involves inferiorization and antipathy.

5.2 Objection: Lockdown Shows Pre-Existing Inequalities but is not Racist

One popular interpretation of the relatively worse impact of lockdown along all kinds of social hierarchy is that the pandemic highlighted pre-existing inequalities,⁵⁰ including along racial lines. This, of course, is true. However, it has received disproportionate attention, if the argument of this paper is correct. The choices that were made for dealing with the pandemic exacerbated pre-existing inequalities. Why ignore the compounding effects of policy choices on these pre-existing inequalities, while emphasizing their pre-existence? A cynic would say that it is in order to deflect scrutiny from the many scientists, other academics, politicians, and commentators—including many considering themselves progressive, egalitarian, and anti-racist—who were strongly in favor of lockdown, and strongly critical of nations like Tanzania which did not implement them. Social scientists who emphasize the pre-existing inequalities of lockdowns while ignoring those created by lockdown⁵¹ are apt to sound like universities bemoaning the inequalities of the school education system while ignoring their own admissions policies. Even if that is unfair, there is certainly an important distinction to be drawn within “the effects of the pandemic.” Some of these effects were controllable, and some not. Particular lockdowns were clearly controllable, since they were policy decisions. It is therefore important to analyze their effects on inequalities of all kinds.

As regards race, it is clear that pre-existing racism is compatible with lockdown being racist, and not an alternative explanation. Indeed, they are natural companions, since a racist decision to lock down would be much harder to explain in the absence of a racist context, and conversely a decision to implement a racially fair strategy in a racist context would be surprising. The only way in which the existence of pre-existing inequalities becomes an objection to the hypothesis that lockdown was racist is if there was no viable alternative. We have already disposed of this claim: in Africa, there were viable alternatives to locking down (without prejudice to the situation elsewhere). It follows that, although the pandemic surely does highlight pre-existing inequalities, this is no objection to, and indeed renders more plausible, the claim that lockdown was racist.