Lockdowns have been followed by a surge in domestic violence. It is vital to understand the impact of unemployment, income loss and time spent at home to assess the remedial potential of cash transfers compared to initiatives such as shelters that can help women to escape.
Social distancing in the wake of the Covid-19 pandemic has led to millions of families the world over being locked down together in their homes and to widespread job and income losses. This has coincided with a substantial global surge in domestic violence, which the United Nations (UN) has described as a ‘shadow pandemic’ (UN Women, 2020).
How much has domestic violence increased in recent months?
Consider early statistics for the UK. In the first month after the initiation of lockdown on 23 March, the rate of homicide of women was more than twice the average of two women a week, and the highest rate in the last 11 years (The Guardian, 2020). At the same time, 80% of women’s frontline support services reported a reduced service because of less face-to-face contact, staff sickness and technical issues, including insufficient laptops to enable working from home (The Guardian, 2020).
The media in many countries have highlighted an increase in calls to domestic violence helplines since the onset of the pandemic. As domestic violence tends to exhibit seasonal variation, researchers have compared 2020 indicators with indicators for the same week or month in the preceding year or years.
For example, a study of calls to the Metropolitan Police Service in London shows an 11.4% increase in calls related to domestic abuse relative to the same weeks in the preceding year (Ivandic and Kirchmaier, 2020). The increase is entirely driven by calls from third parties (such as neighbours and other family members) rather than victims. This may in part be because neighbours are more likely to be at home, but it also squares with reports from around the world that women may find it harder to find the space to make a call when they are locked in with the perpetrator.
Data on actual cases (rather than calls) reveal an 8.5% increase in violence against current partners as well as a decrease of 9.4% in violence against ex-partners. This fits with mobility restrictions having made it harder to commit crimes outside the home.
Using variation in lockdown across the Indian districts, Ravindran and Shah (2020) show that domestic violence complaints increased more in districts with the strictest rules. But rape and sexual assault complaints fell, which is consistent with decreased mobility of women in public spaces.
Attitudes toward domestic violence play an important role in reporting. A study using Italian data finds a doubling of calls to the domestic violence helpline following the introduction of an anti-abuse campaign that encouraged reporting (Colagrossi et al, 2020). The analysis also shows smaller increases in areas with higher baseline rates of gender inequality, which the researchers attribute to differences in reporting.
Similarly, a study using data from 15 large US cities shows that the pandemic was associated with a 10.2% increase in domestic violence calls (Leslie and Wilson, 2020). The timing of the increase coincides with people spending more time at home, as evident from GPS tracking of mobile phones and data on seated restaurant customers. The increase in reported incidents is evident across demographic groups, and it appears to be driven by households without a history of domestic violence, both of which undermine the relevance of ‘structural’ causes.
Although, in general, domestic violence tends to spike on weekends, the pandemic-related increases in these US cities are most evident on weekdays, which is consistent with individuals being at home who in previous years would have been at work. But the evidence is still ambiguous. Another US study, of eight cities, finds no evidence that serious domestic assaults increased during lockdown (Ashby, 2020); and a study in Queensland finds no increase in breaches of domestic violence orders (Payne and Morgan, 2020).
Further research that identifies mechanisms and isolates incidence from reporting is needed. The studies of Italy and India cited above indicate that reporting is influenced by gender norms. It may also be influenced by unemployment (Bhalotra et al, 2020).
In the early weeks of the Covid-19 crisis, reporting was made harder as a result of support services being under-staffed, but in response to reports of rising domestic abuse, many countries have introduced special measures to encourage reporting. A careful analysis of the surge in domestic violence in the wake of the pandemic requires adjusting for both behavioural and policy-driven changes in reporting.
What does research on domestic violence tell us about its causes?
At first glance, an increase in reports of domestic violence since the onset of Covid-19 is consistent with what we know about the causes of domestic violence. Covid-19 has led to job loss (Adams-Prassl et al, 2020; Hupkau and Petrongolo, 2020; Alon et al, 2020), a deterioration in mental health associated with economic uncertainty and reduced contact with friends and support networks (Banks and Xu, 2020; Etheridge and Spantig, 2020). It has also led to the immediate family being forced to spend more time together at home.
In the rest of this section, we consider the predictive power of alternative approaches to understanding the drivers of domestic violence.
The lens of household bargaining
Research for the UK, which leverages differences in local area unemployment rates for men and women in a period that included the 2008/09 recession, shows that increases in men’s unemployment lower domestic violence, while increases in women’s unemployment raise it (Anderberg et al, 2016). A study using data on changes in the wages of women relative to men finds a similar result in the United States (Aizer, 2010).
These patterns can be rationalised with reference to economic analysis of household bargaining in which the power balance within the couple is influenced by their ‘outside options’. Deterioration of men’s earnings potential will attenuate their power (because it weakens their options outside the current partnership), and this will tend to tame them into less violent behaviour. Similarly, if the woman’s outside options deteriorate, she may be more likely to tolerate violence than if she could walk away with self-sufficiency from the partnership.
Women’s jobs have suffered more in the Covid-19 crisis than men’s jobs because, unlike in previous recessions, women happen to be over-represented in the hardest hit sectors (Adams-Prassl et al, 2020; Hupkau and Petrongolo, 2020; Harkness, 2020). This may explain the observed increase in domestic violence.
Related question: How will the response to coronavirus affect gender equality?
Male backlash and instrumental control
Since the shadow pandemic of domestic violence is global in nature, it is important to recognise that the behavioural patterns evident in the UK and the United States may not hold everywhere.
Alternative (sociological) analysis of ‘male backlash’ (Macmillan and Gartner, 1999) reverses the predictions of the household bargaining model. If male identity is closely tied to a breadwinner norm, men’s job loss will tend to prime male identity, create stress and trigger domestic violence. Alternatively, if men seek to control their partners by sabotaging their economic activity or if they seek to usurp resources in the hands of women (Anderberg and Rainer, 2011; Bloch and Rao 2002), then a decline in women’s earnings may lead to lower domestic violence.
Consistent with this is evidence that an improvement in the financial circumstances of women leads to higher domestic violence (Heath, 2014; Bhalotra et al, 2019; Tur-Prats, 2019; Estefan, 2019; Kotsadam and Villanger, 2020; Carr and Packham 2020).
Overall, domestic violence appears to be sensitive to how the pandemic alters women’s financial position relative to that of men. But the direction of the effect appears to depend on underlying gender equality, conditioned by social norms and women’s access to financial independence.
Household income losses
The models of behaviour discussed so far focus on the relative or absolute earnings of the woman. Another possibility is that violence is triggered by a decline in household income (Bhalotra et al, 2020).
Previous research establishes that job loss leads to heightened stress (Black et al, 2015; Schaller and Stevens 2015). It seems plausible that income constraints contribute to stress, which lowers the bar for conflict. For example, couples may need to renegotiate how their more limited resources are spent.
The Covid-19 surge in domestic violence is consistent with this explanation as many households have suffered a significant fall in income, and additional stress from uncertainty over the duration and extent of the decline.
If the total loss in household income is a driver of domestic abuse, then welfare payments may mitigate. It turns out that this depends first, on who receives the payments, and second, whether payments have unintended behavioural effects. Experiments conducted in Kenya (Haushofer et al, 2019) and Mali (Heath et al, 2020) show that cash transfers to men lead to lower rates of physical violence.
Results from studies of cash transfers to women are more mixed. In some cases, there is a decline in violence (Hidrobo et al, 2016; Heath et al, 2020), but in other cases, there is an increase that appears to arise from men using violence to extract resources from the woman (Angelucci, 2008; Carr and Packham, 2020).
Unemployment benefits appear not to mitigate the problem, which is because they lead to longer unemployment duration. While the cash benefits are effective in easing constraints on household finances, this is offset by longer unemployment duration, which increases ‘exposure’ or opportunities for violence (Bhalotra et al, 2020).
Time spent at home
The exposure model, emerging from criminology, emphasises the role of exposure or the time that couples spend together (Dugan et al, 2003). It is consistent with the stylised fact that domestic violence tends to escalate during national holidays, weekends and nights (Vazquez et al, 2005) and during periods of bad weather (RAINN in the United States) because in these cases families are at home together for longer.
Increases in unemployment or unemployment duration lead to more time at home. Lockdown has reinforced this by forcing families to be together for long periods of time. Media coverage of the Covid-19 spike in domestic violence in several countries refers to women being ‘stuck at home’, unable to ‘escape’ to relatives, or to call on their social network. For example, ‘It is thought that [domestic violence] cases have increased by 20% during the lockdown, as many people are trapped at home with their abuser’ (BBC News, 2020).
There appear to be no experimental estimates of the causal effects of exposure on domestic violence. But the fact that lockdown has forced many families on full pay to spend more time together at home provides a unique opportunity to investigate this.
What are the policy options for tackling domestic violence?
The scale of the problem was already large before Covid-19. Recent estimates from a major multi-country study reveal that, on average, one in three women report that they have experienced intimate partner violence at some point in their lives (Garcia-Moreno et al, 2006).
The focus of the discussion so far has been on intimate partner violence, but this often coincides with child abuse. US data indicate that about a quarter of all children subject to maltreatment at home in 2015 lived in households with reports of physical intimate partner violence (US Department of Health and Human Services, 2017). This strengthens the case for policy interventions.
Prior to Covid-19, policies addressing domestic violence have included women’s shelters, counselling, preventive orders, training programmes for perpetrators and skills training or job opportunities for women. Policy responses to the spike in domestic violence during the Covid-19 pandemic have taken the form of additional government funding, measures to encourage reporting, and measures that facilitate women moving out to a safer space.
For example, in the UK and elsewhere, hotels have been asked to open their emptied rooms to women seeking refuge, and women who need to leave abusive partners are now allowed to travel on trains without purchasing a ticket. Measures to facilitate reporting include allowing women to use a code on their phones when it is not safe to speak (the ‘silent solution’), to send a quiet WhatsApp message, or to report abuse to pharmacies, post offices, grocery stores or other routine contact points.
These are all useful and timely measures, but much less attention has been paid by policy-makers to implementing measures to address the actual incidence of domestic violence. Policies that compensate individuals for the loss in earnings or employment may indirectly act to lower domestic violence, but going forward, more fundamental research is needed to understand how best to design preventive measures. This requires an understanding of how people respond to incentives and constraints.
Over and above identifying causes, policy in this domain must be sensitive in designing solutions that can accommodate the privacy of the individuals involved, consider the children in the household, and recognise the fact that some women victims may want redress but may not want their partners incarcerated or fined, while other women may seek ways to leave the marital home.
Where can I find out more?
Covid-19 and ending violence against women and girls: Policy brief from UN Women.
Domestic abuse in times of quarantine: Crime economists Ria Ivandic and Tom Kirchmaier look at data on domestic abuse calls to the police in London.
WHO multi-country study on women’s health and domestic violence against women: 2006 report by the World Health Organization.
Layoff, benefits and domestic violence: first draft study by Sonia Bhalotra, Diogo Britto, Paolo Pinotti and Breno Sampaio – available on request by email.
Who are experts on this question?
- Sonia Bhalotra
- Anna Aizer
- Dan Anderberg
- Manuela Angelucci
- Jillian Carr
- Claudio Deiana
- Andrea Garcia
- Ludovica Giua
- Johannes Haushofer
- Rachel Heath
- Melissa Hidrobo
- Andreas Kotsadam
- Emily Leslie
- Analisa Packham
- Manisha Shah
- Ana Tur-Prats
- Helmut Rainer
- Riley Wilson