With the spread of COVID-19 across the United States, 2020 was a bleak year for many Americans. To contain the virus, workplaces and schools across the country closed, and more people spent more time at home. Mental health challenges are among the numerous difficulties that may arise from dismal economic conditions and diminished connectivity with extended family, friends, and the broader community. The JEC recently discussed the adverse effect of COVID-19 on employment and mental health, suggesting that the mental well-being of the nation is not a small matter and ought to be considered in contemplating a thoughtful reopening of the economy. But the story doesn’t end there, because feeling disconnected, stressed, and mentally unwell can lead to other dangerous consequences. In fact, evidence suggests that COVID-19 may have led to a rise in domestic violence (DV).
The research on the drivers of DV point to two main factors. The first is time use. In other words, when individuals have more time on their hands (particularly, more time at home), the likelihood of engaging in abusive behavior at home increases. The second known factor is labor market conditions. DV has been found to be counter-cyclical, meaning that when the economy is doing poorly, DV rates increase. This likely operates through reduced mental health—in other words, a poor economy leads to employment and financial struggles and greater stress.
These two known drivers of domestic violence provide a theoretical basis for higher rates of DV in the COVID-19 era. Stay-at-home measures have given people a lot more time at home, and COVID-19 has had negative effects on employment and the economy. Furthermore, the social isolation that individuals are experiencing as a result of the pandemic can exacerbate these effects. Taken together, this creates an environment in which abusive individuals can become violent.
However, estimating the causal effect of COVID-19 on DV is complicated by several factors. First, measuring DV is not straightforward, even under normal (“non-COVID-19”) circumstances, but COVID-19 creates additional challenges. One could measure DV through police calls, but this proxy is imperfect, because some DV goes unreported, and COVID-19 is likely affecting reporting rates. Second, there is probably heterogeneity in the effects of COVID-19 on DV and on reporting across individuals. For example, COVID-19 could lead to a rise in DV for individuals who are now spending more time at home with abusive family members, but COVID-19 could also lead to a decline in DV for those who, for example, don’t live with their abusive partners. COVID-19 could also raise reporting rates for some (e.g., if COVID-19 raises third-party reporting because neighbors are spending more time at home, or if being isolated from their usual support system of friends and family makes some more likely to call for help), but lower reporting for individuals who may find it more difficult to call in a complaint with an abusive partner always present in the home.
Thus, estimating an average effect across all categories of individuals may combine positive effects for some, negative effects for others, and overall, not provide much in the way of useful information regarding COVID-19-induced changes in the incidence of DV. In order to gain better certainty of the magnitude and signs of the effects of COVID-19 on DV, one would want to find robust effects across a variety of analyses that use different data and methodologies.
Only a handful of studies have sought to estimate the effect of COVID-19 on domestic violence. As predicted by the complications discussed, the sign and magnitude of the effects vary by the measure of DV used, the estimation strategy, the geographic level, and time frame examined.
Nevertheless, a few of these studies do suggest that COVID-19 led to an increase in police calls reporting domestic violence. One of these studies is a recent paper by Emily Leslie and Riley Wilson of Brigham Young University. The authors use a clever approach to isolate an arguably causal effect of COVID-19 on police calls for service for domestic violence (excluding child abuse) in 14 cities across the United States.
They use several measures to confirm the week in March during which social distancing measures became widespread: cellphone location, restaurant reservations, and Google searches for “social distancing.” Then, using a difference-in-differences estimation strategy, they compare police calls in 2020 before and after social distancing to police calls in 2019 before and after the same week in March. Resulting estimates suggest that police calls increased by somewhere between 7.5 and 9.7 percent, and that these increases are statistically significant.
Their estimates appear to be robust to a variety of sensitivity tests, and upon further investigation suggest that the increase in calls may be driven by first-time reporters. In other words, COVID-19 may have induced individuals with no history of reporting to place police calls regarding domestic violence.
More evidence is needed to reach better conclusions regarding the impact of COVID-19 on DV, but Riley and Wilson’s results already seem to be corroborated by a similar analysis performed by Justin McCrary of Columbia Law School and Sarath Sanga of Northwestern University who find that during the daytime hours, increases in DV are even larger, in the ballpark of 20 percent.
The intuition behind this story is strong; the documented evidence linking economic conditions and unemployment to reduced mental health, and the clear increase in time spent at home due to lockdowns, together create an environment conducive to increased DV. At this point, the empirical literature is still catching up as it attempts to quantify the magnitude and significance of the effect. But to the extent that Leslie and Wilson’s estimates are credible, policymakers should consider the effects of COVID-related closures on DV and look for ways to mitigate the unintended consequences.