Dynamic Impacts of Lockdown on Domestic Violence: Evidence from Multiple Policy Shifts in Chile

Schools as a Safety-net: The Impact of School Closures and Reopenings on Rates of Reporting of Violence Against Children

“…9 Moreover, formal lockdowns were strictly enforced, with police personnel conducting spot checks and citizens-with the exception of essential workers-allowed to leave their house only twice a week for three hours with a specific permit. In fact, the implementation of lockdowns led to a sharp drop in mobility (about 35%) (Bhalotra et al, 2021), beyond the decline already observed after schools were closed (Bennett, 2021). Similarly, the decision to exit lockdown was determined by the MoH.…”

“…Therefore, two neighboring municipalities may have had different lockdown statuses based on the MoH assessment of the need for lockdown. The assessment was mostly based on the case growth and the risk of contagion, although there was no declared metric, making the exact time of lockdown hard to predict for a specific municipality (Bhalotra et al, 2021;Lee et al, 2021). 9 Moreover, formal lockdowns were strictly enforced, with police personnel conducting spot checks and citizens-with the exception of essential workers-allowed to leave their house only twice a week for three hours with a specific permit.…”

“…In baseline models, we include week of year fixed effects for each of the 52 weeks of the year to capture temporal regularities in reporting within each year, as well as municipal-level fixed effects to capture all time-invariant municipal-specific unobservable factors. In most demanding specifications, we include controls for (a) the epidemiological circumstances in each municipality and week, specifically rates of COVID infection, rates of testing, and test positivity; and (b) the lockdown status of each municipality which has been documented to additionally proxy mobility (Bhalotra et al, 2021).…”

“…A survey conducted with US parents in the first weeks of the pandemic documents that 20% of the parents report hitting or spanking their child in the past 2 weeks, with 5% reporting doing so more often than usual, while 1 in 4 recognizing an increase in conflict during that time (Lee et al, 2021). Similarly estimates accounting for factors associated with violence point to increases of up to 29% in referrals of violence against children in a specific US-setting (Prettyman, 2021), while an increase of around 10% in the use of women's DV shelters in Chile was observed as a consequence of lockdowns (Bhalotra et al, 2021), pointing to an increases in violence within the household based on objective measures. In panel C, rather than projecting counterfactual and actual outcomes, we document the difference between counterfactuals and actual reporting under alternative assumptions of increases in underlying violence by {10, 20, 30, 40}%.…”

“…Sensitivity Testing This counterfactual activity assumes that we can infer projections based on levels and trends in reporting prior to the arrival of COVID-19 and associated school closures. However, a broad stream of literature (Evans et al, 2020;Bullinger et al, 2020;Bhalotra et al, 2021;Erten et al, 2022;Pereda and Díaz-Faes, 2020) suggests that violence may have in fact increased, suggesting that counterfactuals may actually under-estimate the true expected reporting if rates of reporting had been maintained constant. We thus conduct additional sensitivity testing, where Reporting Differentials are calculated under less conservative assumptions, where we project that violence, and hence violence reporting, would have actually increased in the post COVID period.…”