How come domestic violence became more frequent in Sweden when COVID-19 hit? Maria Perrotta Berlin investigates possible explanations, and finds that social distancing, home schooling, lack of recreation activities and unemployment are factors during the pandemic that can explain the increase in so called “intimate partner violence”.
As often is the case with crises, the COVID-19 pandemic has laid bare many weaknesses and problems in our societies that were simmering just under the surface, and has given them a new sharper edge. The attention given to the very young and the old, or the lack thereof – inequalities in education and shortcomings in elderly care management – are two of the areas that gained new dramatic salience in Sweden. Reports from other countries suggest potential increases in domestic violence. Are there reasons to expect this increase in Sweden as well?
Prevalence of domestic violence
Violence within the home is the most common form of interpersonal violence for women. While children and men are also victims of abuse of various kind within the family, intimate partner violence (IPV) committed by men against women is generally the most common form of domestic violence. Globally, a third of women that have ever been in a relationship report that they have experienced some form of physical or sexual violence at the hands of their partner. Almost 40% of murders of women are committed by an intimate partner (WHO, 2013) and an additional 25% by another family member (UNODC, 2019).
Sweden is unfortunately no exception to these patterns, despite the strong culture of gender equality. In 2019, a fifth of all reported violent assaults happened within an intimate relation, and an additional 15% was committed by another close relation. In most of these cases, the victim is a woman, while men are more often subject to assaults by an unknown perpetrator.
At the moment of writing, hard data that might help us form a precise picture of what is happening during the current crisis are not yet available other than anecdotal reports in the press. The Swedish National Council for Crime Prevention (Brottsförebyggande rådet or Brå for short) publishes preliminary statistics on crimes reported to the police each month, up to April 2020. Figure 1 shows the reports of violence in intimate relations, separately for women and men, in the Stockholm region, which is the most affected at the early stages of the pandemic. Even if we cannot fully interpret these numbers, we see a clear increase outside the bounds of normal seasonal variation. Although these are preliminary statistics, and probably a fraction of these cases will be dismissed in later stages of investigation, they may represent an underestimation as well as an overestimation of the actual cases, since fewer victims might be able to contact the police or social services in our current time of crisis.
In order to understand the connection between the COVID-19 crisis and patterns in domestic violence, and why such a pattern was largely expected, we can look for explanations in the theories of violence.
The determinants of IPV
Despite being a common phenomenon, IPV is not fully understood, partly due to the challenges in obtaining solid data and empirical evidence, and partly to its complex nature. Cultural, social and psychological factors, many of which change and evolve over time, concur to explain its occurrence. The search for and analysis of these factors follows one of two main directions. One body of social science theories interprets violence as the result of power imbalance within the household. This leads to investigating the role of women’s bargaining power, outside options and status, abstract concepts that are often embodied and approximated by observable factors such as education, employment and income. An alternative interpretation of IPV stresses instead the emotional and irrational nature of it. In this case, particular events or negative emotional shocks, such as the unexpected outcome of a football match, or a lay-off, are seen as triggers of the violent reaction.
Neither of these strands of theories attempts to justify the occurrence of violence by blaming it on external causes, nor to offer exclusive and exhaustive explanations. The two strands likely overlap in most cases, and importantly interact with underlying psychological factors, but also interact with legislation and social norms. Some individuals are more likely than others to resort to violence, all else being equal; some, arguably a very small fraction, even enjoy it. Violence within the household is a crime in most countries but not everywhere around the world, and in some contexts, it is considered acceptable, independent of what the law says. A careful examination of these partial explanations can nevertheless be useful, albeit they do not account for the whole picture, because they can be targeted by public policy.
Studies that seek to establish a causal link between these factors and the occurrence of violence concentrate on settings where the variation in these factors can be considered as good as random. The ideal is to get as close as possible to a clinical trial, the method used to evaluate the effectiveness of drugs. For instance, in a clinical trial when a subset of subjects selected at random is treated with a drug, the only difference between the treated subjects and the rest will be precisely the exposure to the drug, while all other characteristics of the subjects that might concur to explain the outcome will be on average the same across the two groups. Thus, the observed difference in outcomes between the two groups, for example the fact that a larger share of the treated group is recovering, can be more confidently related to the drug. Returning to factors relevant to the occurrence of domestic violence, clearly the income a woman earns, her level of education and employment status are not determined by chance. However, in some situations, it is possible to get to a setting that is very similar to the clinical trial situation, and evidence coming out of studies that leverage these situations is considered reliable. One such study, Anderberg et al. (2016), finds that an increase in male unemployment decreases the incidence of IPV, while an increase in female unemployment increases IPV in the UK. This is consistent with the bargaining power idea. A study set in Spain, though, observes the opposite relation in provinces with stronger traditional gender roles (Tur-Pratts, 2019). Here a decrease in female relative to male unemployment increases IPV is more consistent with a male backlash model: as women improve their economic position, the men of the household feel their identity threatened and react with violence. In the US, Aizer (2010) associates a decline in the gender wage gap with a reduction in violence against women, which is again more in line with the power balance hypothesis. Studies from the developing world associate improvements in general economic conditions with reductions in domestic abuse (Hidrobo et al., 2016; Kim et al., 2007; Haushofer et al., 2019). Additionally, the evidence in Iyer et al. (2012) and Miller and Segal (2018) lends support to the importance of improving the general status of women in society, for example through role modeling and representation in critical positions. They credibly associate the share of women among elected politicians and the share of women among police officers to a significant rise in reporting of crimes against women and at the same time significant declines in such crimes, in the context of India and the US respectively. Finally, when it comes to the “crime of passion” view of IPV, a well-known example is the study by Card and Dahl (2011), finding that unexpected bad outcomes of important football matches led to a 10% increase in the rate of at-home violence in the US.
While the above examples imply that the prevalence of IPV may be affected by a broad range of policies, not necessarily directly targeted to it, many countries have adopted or revised dedicated domestic violence legislation over the past decades. In spite of this substantial legislative effort, we still know very little about the effectiveness of alternative legislations and law enforcement practices. An ongoing project at the Stockholm Institute of Transition Economics (SITE) seeks to fill this gap, studying two countries that recently changed their legislation: Russia, where minor forms of domestic violence were decriminalized in 2017, and Ukraine, where conversely the scope of the criminal definition was extended to cover not only physical but emotional and economic violence.
COVID-19 and violence in the home
News stories from all around the world, for example this report in the New York Times, have sounded the alarm about the risks of lockdown. Although available research on such a link is still limited and inconclusive, there is a common perception that domestic violence increases during the holidays because families are spending more time together and individuals at risk are more isolated from their social networks. The political reaction to the spread of COVID-19 has led, in many countries, to a similar situation, with even stronger confinement and isolation. Consequently, conflicts and violence within households are expected to rise. The economic shock following the lockdowns, which is leading to a spike in unemployment, may also cause an increase in conflict through the channels described above.
Sweden, however, did not go into a lockdown. Nevertheless, a combination of fear of the contagion and, later on, the recommendation by the authorities to work from home to the extent possible, to maintain social distancing and to avoid unnecessary travel, resulted in significantly dampened mobility patterns. Figure 2 clearly shows that Stockholmers spent more time at home, went less to workplaces and, in particular, had less access to recreational activities. Many major sport events were suspended for extended periods, and travelling for work or vacation was extremely limited. One can thus hypothesize an increase in IVP due to reduced availability of options and, in particular, of voluntary “self-incapacitation”: when a (violent) partner is busy with something they enjoy more, no aggression will occur, at least at that time.
Some indirect evidence does indeed suggest that simply having more time together is certainly a factor that can contribute to increasing the incidence of violence. In particular, a large amount of literature highlights the importance of physical incapacitation: one can simply not commit violent crimes on the street or at home when physically in school (Jacob and Lefgren, 2003), in prison (Levitt, 1996), or at the movies (Dahl and DellaVigna, 2009). In particular, the latter finds a reduction in crime associated with attendance to violent movies. Since exposure to violent movies is found in laboratory experiments to increase aggressiveness and violent behavior (another manifestation of the emotional trigger effect discussed above), this implies that the pure time substitution effect is larger than the trigger effect. To be clear, what happens during a lockdown or social distancing situation is the opposite: violent people have limited possibility to incapacitate themselves choosing to go to the pub rather than staying at home.
Another piece of relevant indirect evidence comes from studies of prostitution legislation. In particular, the restriction to the individual choice set resulting from COVID-19 policies is reminiscent of another episode in recent Swedish history, namely the asymmetric criminalization of prostitution (or Sexköpslagen) introduced in 1999, whereby buyers of sexual services are punished with fines and prison. This legislation was part of a Gender Equality Bill, intended to give a clear signal about societal values and protect the, in large part female and in large part exploited, sex workers. The model proved very successful in deterring street prostitution, and, under the catchy name of the “Nordic model”, has subsequently been adopted by many other countries. Looking at population-wide rates of violence against women at the time, our study (Berlin et al., 2019) finds, however, that this legislative change might have induced a 10% increase in assaults committed against women by acquaintances indoors. With additional supportive evidence, we can exclude that this increase reflects aggressions against sex workers. The increase can rather be best understood as IPV. This is fully in line with previous research. Study after study has shown that restrictions to the market for sexual services, rather than the sex trade itself, have substantial negative impacts in terms of increased violent crime in affected communities, which is in contrast to what politicians seem to think.
The increased time spent together is however only one of the potential mechanisms that might link the COVID-19 crisis to patterns of domestic violence. Viewed through the bargaining power theories, we can easily identify many direct or indirect consequences of the crisis that threaten to change the power balance within households. The labor market impact of the crisis is likely to have extremely heterogeneous effects across sectors, occupations and income levels. Given well documented gender differences in all these dimensions, these effects come with asymmetric impacts for men and women. In particular, unemployment and income loss might affect men disproportionately, since women in general tend to prefer sectors and jobs that offer flexible work arrangements and might find themselves now in jobs that allow remote working to a larger extent. Or just the opposite might occur. Income loss might affect women disproportionately since women are also overrepresented in the service sector, if it turns out that these jobs are the most affected in terms of lay-offs and bankruptcies. Either way, asymmetric effects on unemployment may have consequences for IPV, as discussed above. Further, economic uncertainty in itself is associated with emotional stress and increased consumption of alcohol and other drugs. This may also result in a higher occurrence of violence.
Another indirect effect is the likely increase in time spent on childcare and household chores, due to the social distancing, voluntary or recommended. People spend more time at home and buy fewer services such as restaurant meals, cleaning services and dry-cleaning. Children older than 15 are homeschooled. This might generate stress and upset the balance of “normal times” division of labor, leading to conflicts.
All in all, the social distancing situation will be a very different experience depending on whether a woman can work remotely or keeps her job in critical sectors, or belongs to categories celebrated as frontline heroes, or becomes unemployed. All these aspects will affect her relative economic stance and her “status” within the household, and therefore, all else being equal, the probability of her victimization.
The data are not available at the moment to dig deeper into these hypotheses, but we can certainly expect to see a wealth of studies scrutinizing COVID-19 facts and consequences in the near future. Such studies will be crucial in improving our understanding of the determinants of domestic violence, as well as in suggesting policies that might effectively combat domestic violence in these changing times.
Aizer, A. (2010). The Gender Wage Gap and Domestic Violence. The American economic review, 100 (4), 1847-1859.
Anderberg, D., Rainer H., Wadsworth, J., Wilson T. (2016). Unemployment and Domestic Violence: Theory and Evidence. The Economic Journal, 126 (597), 1947–1979.
Berlin, Maria P.; Giovanni Immordino, Francesco F. Russo and Giancarlo Spagnolo, 2019. “Prostitution and Violence. Empirical Evidence from Sweden”, Unpublished manuscript.
Card, D., and Gordon B. Dahl. (2011). Family violence and football: The effect of unexpected emotional cues on violent behavior. The Quarterly Journal of Economics, 126 (1), 103-143.
Dahl, Gordon, and Stefano DellaVigna. "Does movie violence increase violent crime?." The Quarterly Journal of Economics, 124.2 (2009): 677-734.
Haushofer, Johannes, et al. Income changes and intimate partner violence: Evidence from unconditional cash transfers in Kenya. No. w25627. National Bureau of Economic Research, 2019.
Hidrobo, M., Amber, P., and Lori, H. (2016). The Effect of Cash, Vouchers, and Food Transfers on Intimate Partner Violence: Evidence from a Randomized Experiment in Northern Ecuador. American Economic Journal: Applied Economics, 8 (3), 284-303.
Iyer, L., Mani, A., Mishra, P., and Topalova, P. (2012). The power of political voice: women's political representation and crime in India. American Economic Journal: Applied Economics, 4 (4), 165-93.
Jacob, Brian, A., and Lars Lefgren. 2003. "Are Idle Hands the Devil's Workshop? Incapacitation, Concentration, and Juvenile Crime." American Economic Review, 93 (5): 1560-1577.
Kim, Julia C., et al. "Understanding the impact of a microfinance-based intervention on women’s empowerment and the reduction of intimate partner violence in South Africa." American journal of public health 97.10 (2007): 1794-1802.
Levitt, Steven D., The Effect of Prison Population Size on Crime Rates: Evidence from Prison Overcrowding Litigation, The Quarterly Journal of Economics, Volume 111, Issue 2, May 1996, Pages 319–351, https://doi.org/10.2307/2946681
Miller, A. R., Segal, C. (2018). Do Female Officers Improve Law Enforcement Quality? Effects on Crime Reporting and Domestic Violence. Effects on Crime Reporting and Domestic Violence. Mimeo.
Tur-Prats, A. (2019). Family Types and Intimate Partner Violence: A Historical Perspective. Review of Economics and Statistics, 101(5), 878-891.
UNODC, Global Study on Homicide 2019 (Vienna, 2019)
WHO, LSHTM, SAMRC. Global and regional estimates for violence against women: prevalence and health burden of intimate partner violence and non-partner sexual violence. Geneva: WHO, 2013.
Maria Perrotta Berlin is Assistant Professor at the Stockholm Institute of Transition Economics (SITE). She is Global Challenges and UN PRME Development Director at the Stockholm School of Economics.