When the city of Philadelphia introduced an emergency eviction ban in the summer of 2020, the local landlords’ association, HAPCO, responded with a lawsuit questioning the broad societal interest of the ban. In a cynical twist of phrase, they argued that this legislation protects the few, not the many. But there was a consequence they didn’t expect – the claim motivated a group of scientists in the United States to investigate the effects of eviction bans on the course of the pandemic.

Their recent publication in Nature Communications makes a strong argument ‘that policies to stem evictions are a warranted and important component of Covid-19 control.’ They built their model in the late summer and early autumn of 2020, based on data from the first Covid-19 wave in the US.

Epidemiological modelling is commonly done using SEIR (susceptible, exposed, infectious, and recovered) models. Everyone in the population belongs to one of the four groups in the acronym. The interaction between people represented by a network of their contacts keeps the epidemic going and changes the number of people in each category. On top of a complex network (corresponding to a population of one million), the researchers added the possibility of turning on and off and varying the effects of different control measures, lockdowns, and eviction bans.

Backed by previous extensive studies on evictions and homelessness, the authors assume that most evictions result, at least in the short term, in ‘doubling up’: the evicted household moving in with another household, usually family or friends. The rest end up homeless straight away. These cases lead to an increase in prolonged contacts and the creation of pandemic hotbeds, be it in the new, enlarged household, or a homeless shelter. Unlike ‘fused households’, which are voluntary and may help mitigate the effects of the pandemic, there’s nothing voluntary in ‘doubling up’, and it happens at the worst time for those involved.

To see the impact of eviction bans on an American city (using the example of Philadelphia in the paper), the researchers set up different scenarios of interest. Who gets evicted? How many evictions happen in a given time? Where do evicted people go?

In the most basic model, evictions could be taken to occur at random, and the household taking the evicted family in could be again randomly selected. The eviction rates could be taken from the historical data, ranging from 0.1 to 2 percent of households per month in large US cities. Even though these settings largely ignore the entrenched inequality in the cities, the results show an excess in Covid-19 cases just by virtue of evictions happening. No matter how few happen, the effect is noticeable. The risk of contracting the virus increases substantially for the evicted family and their hosts – but a noticeable increase in the risk is seen for those ‘unaffected by evictions’ as well.

In the next iteration of modelling, the divisive reality is taken into account. Both evictions and Covid-19 disproportionally hit poor, minority communities. Even though the economic division runs deep and minimises interaction between the rich and the poor, as witnessed by the mobility data, the simulation once again has shown that evictions that mostly happen in the working-class parts of Philadelphia—where opportunities for social distancing in the entire community are low due to nature of workplaces and housing—would still hike the risk of infection for the inhabitants in the wealthy neighbourhoods.

The landlords’ premise that eviction bans protect the few (i.e. the tenants endangered by evictions) is demonstrably false: stopping the evictions is a protective measure for everyone. The federal judge delivering the opinion in the case of HAPCO against the city of Philadelphia cited preliminary findings of the model presented in this paper as a part of the argumentation.

Finally, it’s not realistic to expect the rate of evictions to stay the same in a pandemic where the income of the working class is gravely jeopardised, both by the closing of businesses and by the long-term health hazard. A significant increase in evictions (a consequence of the lifting of the ban in countries like Ireland that tenants and charities are bracing for right now) would result in a multifold increase in all the effects flagged so far.

It’s not hard to continue the list of factors that would aggravate the situation: from deterioration of the overall health condition of people under threat of eviction, to the inability for the evicted and/or homeless to access health services in case of Covid-19 infection. The magnitude at which evictions make the pandemic deadlier can hardly be overestimated.

In this article, I have avoided spelling out concrete numbers. Does one eviction result in two excess cases of Covid-19 on average? Does the risk for the privileged increase by 1 or 10 percent?

Instead, it’s the trends that are described. One of the reasons not to report figures from the paper is because the authors ‘caution against using the precise numerical estimates from [the] model to infer the effects of eviction moratoria in particular locations,’ given that there are still many uncertainties in our knowledge of Covid-19 dynamics, but also of the social interactions and demographics affected by the pandemic and housing insecurity alike.

The other reason is to avoid the bargain we have seen too much of already – how many cases, lives, jobs could ‘we’ afford losing, and what the sum of money or percentage of GDP they could be equated to. The phenomena seen in the model are hard to dispute: no matter how you vary the parameters of it, evictions threaten everyone.