Policy Research Working Paper 9547 Can Youth Empowerment Programs Reduce Violence against Girls during the COVID-19 Pandemic? Selim Gulesci Manuela Puente Beccar Diego Ubfal Africa Region Gender Innovation Lab February 2021 Policy Research Working Paper 9547 Abstract This paper shows that a youth empowerment program in seven months after its completion, the program increased Bolivia reduces the prevalence of violence against girls during girls’ earnings and decreased violence targeting females. the COVID-19 lockdown. The program offers training in Violence is measured with both direct self-report questions soft skills and technical skills, sex education, mentoring, and list experiments. These findings suggest that empower- and job-finding assistance. To measure the effects of the ment programs can reduce the level of violence experienced program, the study conducts a randomized control trial by young females during high-risk periods. with 600 vulnerable adolescents. The results indicate that This paper is a product of the Office of the Gender Innovation Lab, Africa Region. It is part of a larger effort by the World Bank to provide open access to its research and make a contribution to development policy discussions around the world. Policy Research Working Papers are also posted on the Web at http://www.worldbank.org/prwp. The authors may be contacted at dubfal@worldbank.org. The Policy Research Working Paper Series disseminates the findings of work in progress to encourage the exchange of ideas about development issues. An objective of the series is to get the findings out quickly, even if the presentations are less than fully polished. The papers carry the names of the authors and should be cited accordingly. The findings, interpretations, and conclusions expressed in this paper are entirely those of the authors. They do not necessarily represent the views of the International Bank for Reconstruction and Development/World Bank and its affiliated organizations, or those of the Executive Directors of the World Bank or the governments they represent. Produced by the Research Support Team Can Youth Empowerment Programs Reduce Violence against Girls during the COVID-19 Pandemic?∗ Selim Gulesci † Manuela Puente Beccar ‡ Diego Ubfal § ∗ We would like to thank the staff of Save the Children Bolivia and Italy for their support in the implementation and data collection for this project, Bulgari for funding the program and LEAP-Bocconi for enabling the evaluation. The project received ethical approval from the ethics committee of Bocconi University, No:9882-2; and is registered on the American Economic Association’s RCT registry, No: AEARCTR-0004836. Sarah O’Brien provided outstanding research assistance. The findings, interpretations, and conclusions expressed in this paper are entirely those of the authors. They do not necessarily represent the views of the World Bank and their affiliated organizations, or those of the Executive Directors of the World Bank or the governments they represent. † Trinity College Dublin. E-mail: gulescis@tcd.ie ‡ Bocconi University. E-mail: manuela.puente@unibocconi.it § World Bank. E-mail: dubfal@worldbank.org 1 Introduction Throughout the world, strict containment measures caused by the COVID-19 pandemic have in- creased risk factors associated with domestic violence. Media and institutional reports indicate an increase of domestic violence and in particular of violence against women and children in several countries affected by COVID-19 (e.g., Reynolds, 2020; Taub, 2020; Ritz et al. 2020). The reported rise in violence against women (VAW) due to the COVID-19 pandemic has been named by UN Women as the “Shadow Pandemic” (United Nations Women, 2020). In this paper, we study whether a youth empowerment program can reduce the prevalence of violence against adolescents during the COVID-19 lockdown in Bolivia. The program combines training in soft skills and technical skills with sex education, mentoring and job-finding assis- tance. We conducted a randomized control trial with 600 vulnerable youth who applied to the program in four cities in Bolivia. Our data include an in-person baseline survey and a follow- up survey conducted by phone due the social distancing restrictions imposed by the COVID-19 pandemic. The follow-up survey was carried out seven months after the end of the program and six months into the lockdown. The program significantly reduced violence experienced by girls. The prevalence of violence reported by girls fell by 10 percentage points (ppt), over a mean of 21 percent in the control group. For boys, we do not find significant reductions in violence. To address concerns about self-reporting bias, we use item list experiments, which confirm our main findings. The level of violence among girls, as measured by a list experiment included in our follow-up survey, is much lower in the treatment group than in the control group (39% in control v.s. 1% in treat- ment), while it is not lower for treated boys than for control boys. We assess some of the key mechanisms through which the program may have led to these effects. We present evidence that the program had a positive effect on earnings for girls, but not for boys. This is consistent with an improvement in girls’ bargaining power within the household or a reduction in income-related stress, both of which may explain the decrease in violence against girls. We do not find evidence for mechanisms related to changes in soft skills as we do not see any effects of the program on a set of soft skills targeted by the program. The paper contributes to a burgeoning literature studying the causes of violence in general, and violence against children/girls in particular. Economic crises, conflicts and natural disasters are often linked to increased prevalence of violence against women and children (Anastario et al. 2013, Weitzman and Behrman, 2016, Fraser, 2020). Several recent studies show increases in VAW due to the COVID-19 pandemic and lockdown measures (Aguero, 2020; Boserup et al., 2020; Leslie and Wilson, 2020; Mahmud and Riley, 2020; Ravindran and Shah, 2020; Silverio-Murillo and Balmori de la Miyar, 2020). A recent study by Ravindran and Shah (2020) shows an increase in domestic violence complaints in India in districts with the strictest confinement rules.1 We contribute to this emerging literature by showing that multi-faceted youth empowerment pro- grams can be one way to curtail the rise in gender-based violence during high-risk periods, such 1 In a related study, Bandiera et al. (2020) show that temporary school closures during the 2014-16 Ebola epidemic in Sierra Leone increased teenage pregnancies and lowered school enrollment among girls, and a program that provided safe spaces (in the form of community clubs) and training in soft skills lowered these negative impacts. 1 as the COVID-19 pandemic and the ensuing lockdowns.2 The rest of the paper is organized as follows. Section 2 explains the details of the program. Section 3 describes the design of the experiment and sample characteristics. Section 4 presents the results. Section 5 concludes. 2 Youth Empowerment Program The program Adolescents: Protagonists of Development has been funded by Bulgari and imple- mented by Save the Children in Albania, Bolivia, Nepal and Uganda since 2016. Its main aim is to help vulnerable youth find a job, improve their working conditions and strengthen their in- come generation capacity. The target population consists of vulnerable adolescents aged 15 to 18. In order to identify the sample for this study, adolescents were recruited in four cities in Bolivia. Several recruitment strategies were used: fliers in markets and other public places, Facebook ads, cooperation with neighborhood associations and schools offering night shifts, and press confer- ences. Interested adolescents filled a vulnerability card containing information used to measure their social vulnerability (e.g., housing conditions, access to health care, violence, substance use) and their economic vulnerability (e.g., household income, child labor, lack of economic support from the family). The program staff selected 600 adolescents with high levels of vulnerability who also showed commitment and willingness to participate in face-to-face interviews with the team. The program provides youth with soft skills and technical skills training, sex education, men- toring and support in finding a job or starting a business. In particular, the program offers the following activities: • General Training – Personal empowerment (Module 1, 16 hours) – Sexual and reproductive health (Module 2, 16 hours) – Economic empowerment (Module 3, 16 hours) – Basic competences (Module 4, 16 hours) • Technical-skills training in predefined areas according to market demand (70 hours) • Work insertion or business development The four modules covering general training are taught by the staff of Save the Children. Mod- ule 1 focuses on self-esteem and leadership while providing adolescents with an opportunity to get to know each other and increase trust. Module 2 discusses contraceptive methods and teen pregnancy. Module 3 covers material on market analysis, entrepreneurial soft skills, sustainable business models, workers’ rights and how to prepare for a job interview. Finally, Module 4, teaches basic math and literacy. These four modules are taught over the course of 4 four-hour sessions each. 2 See Kerr-Wilson et al. (2020) for a survey of interventions to prevent violence against women and girls. 2 After this general training, the project trains youth in specific technical skills. Adolescents can select up to three training courses among a menu that was designed based on market demand studies by Save the Children in cooperation with private partners in each region. The training is implemented by the local partners. For example, in La Paz, the most common choices were gastronomy, customer service and graphic design. The total length of these activities is approxi- mately 70 hours. The final activity of the project is to help adolescents find interviews with employers offering jobs that match adolescents’ skill levels and satisfy certain standards (e.g., they are compatible with schooling, they do not involve risky activities,3 and they offer a wage no lower than the min- imum wage). Once the adolescent starts working, the program monitors working conditions up to three months. The program also offers adolescents who do not want to find a job the opportu- nity to start their own business. However, this happened with only one adolescent in our sample. 3 Methodology and sample characteristics 3.1 Methodology To estimate the impact of the program, we adopted a randomized control trial (RCT) methodol- ogy. The RCT was designed and implemented through our collaboration with Save the Children Bolivia and the support from Save the Children Italy. The evaluation covered four metropolitan cities in Bolivia: Cochabamba, La Paz, Oruro and Santa Cruz. The first step of the evaluation involved selecting a sample of eligible youth. The program team identified 600 youth satisfying the criteria for selection into the program as explained above. All selected youth completed an in-person baseline survey. We then conducted a private lottery using Stata to randomly select 300 youth who would be offered to be part of the program starting from 2019 (treatment group), and 300 youth who would not be offered to be part of the program in 2019, but would have the chance to participate in the program after the evaluation (control group). When conducting the randomization, we stratified the samples on region, gender, age, whether the adolescent was working at the time of the interview and, only for Cochabamba and La Paz,4 whether s/he was a victim of violence. The general training modules were completed from August to October 2019. The techni- cal skills training was temporarily suspended during the election-related violence in October- November 2019, which translated into four to five weeks of highly reduced economic activity, but it was completed by December 2019 in all sites. In February 2020, the project started offering job-finding assistance. Due to the COVID-19 pandemic, the country entered a strict lockdown on March 22, 2020 (March 16 in Oruro where the first cases were registered). In two of the four cities (La Paz and Oruro), the program managed to offer job-finding assistance to most treated youth; while in the other two cities (Cochabamba and Santa Cruz), it only completed the training, with no significant job-finding assistance. In what follows, we assess the heterogeneity of treatment 3 Under the Bolivian law, minors cannot work in certain tasks, such as mining or lifting heavy objects. 4 In Oruro and Santa Cruz there was not enough variation to stratify on violence. 3 effects along this geographical dimension. Due to the COVID-related mobility restrictions, we conducted a follow-up survey by phone in the last two weeks of September 2020. We were able to survey 511 adolescents (85% of the sample). Response rates were similar across treatment arms, and we do not find any evidence of differential attrition (see Table A1 in the Appendix). 3.2 Sample characteristics Table 1 presents summary statistics on the baseline characteristics of the youth in our sample as well as balance tests comparing treatment and control groups. For comparability with our estimation sample, we focus on the sample of youth who were re-surveyed during the follow-up survey.5 Overall, 63% of the youth in our sample are girls and 65% are 17 or 18 years old. Re- ported prevalence of violence is high: 55% of the youth reported having ever experienced some type of violence, 24% reported having ever experienced physical violence, 41% psychological violence and 5% sexual violence. These levels of violence are in line with the prevalence of vio- lence reported in other data sources in Bolivia. The National Statistical Institute of Bolivia and UNICEF report that children are physically punished by an adult member in 83% of households.6 Another UNICEF publication in 2014 states that 61% of girls aged 15 to 19 experienced physical violence by their current or former partner, while 22% of boys of the same age range report hav- ing experienced violence by a friend or acquaintance, and 13% by their current or former partner.7 Average earnings of participants are Bs. 419.4 per month (60.7 USD). To put this in context, the national minimum wage during this period was Bs. 2.122 (307 USD per month). In terms of sources of income, 24% of the youth in the sample receive some income from wage-employment, 3% from self-employment, 38% from informal transfers (i.e., from family or friends) and 2% from formal transfers (i.e., provided by the government or an NGO). In order to measure soft skills at baseline, we used the “Employability Assessment (EA) tool” developed by Save the Children with the aim of measuring the presence and improvement of employability skills. The EA tool is a questionnaire with 24 items which produces quantitative scores for 6 categories of soft skills: self-concept, self-control, social skills, communication skills, conflict resolution skills and job- searching skills.8 Table 1 shows that baseline characteristics of the youth in the treatment and control samples are well balanced. Therefore, we conclude that the randomization was successful and the youth in the control group are a valid counterfactual for the treatment group.9 5 Conclusions are similar if we use the full baseline sample – see Table A2 in the Appendix. 6 Source: https://www.unicef.org/bolivia/biptico estudio violencia ninez bolivia.pdf 7 Source: http://files.unicef.org/publications/files/Hidden in plain sight statistical analysis EN Sept 2014.pdf 8 The tool is described in the Appendix, Table A7. Each component of the index is constructed by first standardizing the responses to the individual questions (by subtracting the mean and dividing by standard deviation of the control group for each outcome) and then averaging across the standardized outcomes. 9 Baseline characteristics are also balanced if we split the sample by gender of the participant, see Tables A3 and A4 in the Appendix for balance tests on the girls’ and boys’ samples respectively. 4 3.3 Estimation of treatment effects In order to estimate the average treatment effects of the program, we use the following regression model: yi = α + β · Ti + γ · Ti · Fi + λ · yi0 + Θi0 + it , (1) where yi is the outcome of interest for respondent i at the follow-up survey, Ti is a dummy variable equal to 1 if the respondent was allocated to the treatment group at baseline, yi0 is the baseline level of the outcome for individual i, Fi is an indicator variable that is equal to 1 if the respondent is female and 0 otherwise, and Θi0 are randomization strata (dummies for each strata used in the randomization, including the gender of the respondent and regional dummies). The estimate for β corresponds to the treatment effect on males, the estimate for γ correspond to the differential effect of the treatment on females relative to males, while the sum β + γ corresponds to the treatment effect for females. Since the randomization was conducted at the individual level, we do not cluster the standard errors. 4 Results 4.1 Violence Table 2 shows treatment effects on the prevalence of violence that the participants reported ex- periencing during the three months preceding the follow-up survey.10 We find a negative and significant treatment effect on violence reported by females, but not by males. In particular, female participants are 9.6 percentage points (ppt) less likely to report suffering any violence. The point estimate for boys is positive (7.2 ppt) but imprecisely estimated at conventional levels. The differential treatment effect on girls relative to boys is 16.8 ppt and statistically significant. In terms of magnitudes, the 9.6 ppt reduction in the prevalence of violence against girls in the treatment group is a large impact, corresponding to a 46% reduction relative to the control group where 21% of girls reported having experienced any type of violence. The rest of Table 2 shows the effect on different types of violence: physical, psychological and sexual. For girls, the pro- gram had a negative effect on all three types of violence, lowering the prevalence of physical violence by 3 ppt (56% relative to the control group), psychological violence by 10 ppt (50% reduction) and sexual violence by 3 ppt (103% reduction). The effect on physical violence is im- precisely estimated at conventional levels, while the effects on psychological and sexual violence 10 To guarantee respondents’ safety, interviewers were carefully trained on how to ask these types of questions by an expert on Child Safeguarding Policy. Enumerators took measures to verify the privacy of the interviews, but since they were conducted by phone, additional steps were taken to prevent the perpetrators of the violence from listening to participants’ answers. In particular, the interviewer provided examples of what types of actions should be considered as violent and participants were asked to answer only ”yes” or ”no”. For physical violence, they were told that it “includes cases when someone hits or slaps you, or pushes or pulls your hair”, for psychological violence they were told that it “includes cases when someone insults, threatens, verbally abuses, ridicules or makes fun of you” and for sexual violence they were told that it “includes cases when someone touches your body without your permission, or forces you to have sex”. Participants were given the option of not answering the question if they did not feel comfortable answering. Participants who reported abuse were confidentially provided with options to search for help. 5 are statistically significant at the 95% confidence level. We also find a marginally significant in- crease in psychological violence for treated boys. In particular, boys in the treatment group are 8 ppt more likely to report having experienced any psychological violence relative to the control group.11 A potential concern with measuring sensitive topics, such as the prevalence of violence, through direct survey questions is reporting bias: respondents may not want to report violence due to shame or concerns about anonymity. To avoid this problem, a strategy commonly used in the literature is to rely on indirect elicitation techniques, such as list experiments (e.g., Rosenfeld, Imai and Shapiro, 2015). During the endline survey, we conducted a list experiment to elicit rates of violence among youth. In particular, respondents were asked to report the number of statements from a list of 4 or 5 items that applied to them.12 The main idea behind this method- ology involves exploiting random variation in the presence of sensitive items in the lists. Every respondent is randomly assigned to one of two groups (group A or group B). Respondents in group A are presented with a list of 4 items, which does not include any sensitive item; while respondents in group B are presented with a list of 5 items, one of which is the sensitive item.13 The only difference between the two lists is the presence of the statement “You have suffered some kind of physical violence in the last week” in the list presented to group B but not to group A. In order to calculate the percentage of youth for whom the sensitive item is true (i.e., the per- centage of people who have suffered some kind of physical violence in the last week), we look at the difference between the average number of statements reported as true by respondents in group B relative to A. Since the assignment of individuals to group A or B is random, there is no reason why the number of true statements in the two groups should be different, other than the presence of the sensitive item. To ensure that the randomization was balanced within gender and treatment group, we randomized female and male respondents in treatment and control groups separately. Table 3 presents the difference in the average number of statements reported to be true by respondents in group B vs. A, which corresponds to the share of respondents who had suffered physical violence during the week prior to the endline survey.14 Three findings are of note: first, the rate of physical violence in the control group is 36%, which is much higher than that reported in the direct survey questions (6% for females and 3% for males). One explanation of this difference is that participants tend to under-report sensitive questions when asked directly about the sensitive topic. Second, respondents in the treatment group suffer from lower physical violence levels than those in the control group (16% vs. 36%). Third, consistent with the direct 11 The violence figures are not directly comparable to those reported at baseline. First, because the baseline survey was conducted in person, while the follow-up was conducted by phone. Second, because at baseline the question was about having ever suffered violence, while the follow-up asked about the past three months. 12 Respondents were read the following script: “Now I’m going to read some statements about many different things. Some of these statements will be true and some will not. After I read all statements, please tell me how many of them are true for you. I don’t want to know which ones, just how many.” 13 More specifically, respondents in group B are given the following 5 statements: “1) You have been to Peru, 2) You can play the guitar, 3) You have a family member who lives in La Paz, 4) You have seen the movie Avengers: Endgame” and the sensitive item “5) You have suffered some kind of physical violence in the last week”; while respondents in group A are only given the first 4 statements. 14 The prevalence of violence measured in this way is not directly comparable to that measured with the self-reported questions, which referred to the past three months instead of the past week. 6 survey question, the effect of the treatment in reducing violence is driven entirely by females: only 1% of female respondents in the treatment group experienced physical violence, while the corresponding rate is 39% in the control group. The rates are 39% and 32% for males in the treatment and control groups, respectively. This confirms that the program did not manage to reduce violence among boys. Overall, results from the list experiment show a similar pattern as the direct questions in the survey: the program reduced the likelihood of experiencing violence among female participants. 4.2 Mechanisms There are several channels through which the program may have led to a reduction in the preva- lence of violence among treated girls. In this section, we discuss and test for some of the key mechanisms that have been highlighted in the literature. We discuss other potential mechanisms which we, unfortunately, do not have information on, in the conclusion section. (i) Improved earnings: To the extent that the program increased earnings for girls in the treat- ment group, this may have lowered the prevalence of violence against them through two mech- anisms: First, a change in women’s access to economic opportunities (such as employment or other earnings) may decrease or increase the prevalence of violence, depending on the initial allocation of bargaining power within the household and whose reservation utility is binding (Tauchen et al., 1991; Eswaran and Malhotra, 2011; Bloch and Rao, 2002; Anderson and Genicot, 2015).15 If the program increased girls’ earnings, this could have improved their outside options within the household and enabled them to leave abusive relationships. Second, economic in- security and poverty-related stress caused by the lockdown measures are likely to increase the risk of domestic violence towards women and children (Peterman and O’Donnel 2020, Conrad- Hiebner and Byram 2020). An increase in earnings due to participation in the program may have mitigated the higher stress levels linked to economic insecurity and lowered the prevalence of violence. Table 4 shows that girls in the treatment group earned Bs.119 more than girls in the control group and this effect is significant at the 10% level. Compared to the mean income of females in the control group (Bs. 294), this corresponds to a 41% increase in total income. The rest of Table 4 tests if the program affected the youth’s sources of income on the extensive margin. We find that treated girls are 8 ppt less likely to report earnings from wage labor. They are 2 ppt more likely to report having income from self-employment and 3.5 ppt more likely to have earnings from formal transfers (government or NGO provided), but these effects are imprecisely estimated. We did not collect information on the intensive margin of earnings by source of income, so we can- not identify which source is driving the increase in earnings of females reported in column 1. Overall, the results in Table 4 show that the program helped girls increase their income, even if it was not through helping them find a better job. This increase in girls’ earnings may explain why they experienced lower violence – either because of their improved bargaining power within the household; or because the higher earnings lowered the prevalence of stress-related domestic 15 The existing evidence on the effects of labor market opportunities and income of women relative to men on the prevalence of domestic violence is mixed (Aizer, 2010; Andenberg et al., 2016; Angelucci, 2008; Bhalotra et al., 2019; Bobonis et al., 2013; Chin, 2012; Heath, 2014; Heise and Kotsadam, 2015; Hidrobo and Fernald, 2013). 7 violence within their households.16 (ii) Improvements in soft skills: To the extent that the program succeeded in changing girls’ soft skills, such as self-confidence and expressiveness, this may have empowered them to leave or bet- ter face abusive relationships within the household. Table 5 presents the treatment effects on soft skills. Overall, we do not find significant effects on soft skills – the aggregate index combining the six standardized indicators shows no significant treatment effect (column 1), neither for males nor females. When we examine the effects on individual components of the index, we only see significant treatment effects in one of the seven soft-skills indices included for girls – that on job- searching skills. The last piece of evidence on soft skills comes from a task we added at endline to measure self-confidence. Respondents were asked five general knowledge questions. Then, they were asked to guess how many of these questions they thought they answered correctly. Overestimating the number of correct answers is taken as an indirect measure of overconfidence. The last column of Table 5 shows that the program has an effect on self-confidence only for boys. While 55% of boys in the control group are classified as over-confident, this share increases by 18 percentage points in the treatment group. For girls, we see a similar share of 52% classified as over-confident in both treatment and control groups (we do not see any effects for girls on being classified as under-confident either). Overall, the evidence suggests that the program did not have a significant effect on soft skills. This could be because soft skills are in general hard to measure, and the methods we used in the phone surveys are not ideal to capture changes in soft skills. As noted in Section 4.1, we find a small increase in the prevalence of psychological violence for boys in the treatment group. One explanation for this may be linked to the increase in boys’ overconfidence caused by the program. One hypothesis is that as they became more confident, they may have tried to assert their opinions more strongly in their households or social networks, leading to an increase in arguments and verbal clashes. 5 Conclusions In this paper, we presented results on the effects of a youth empowerment program in Bolivia on the prevalence of violence against adolescents. We find that the program significantly reduces violence reported by girls during the lockdown period. While we do not have strong evidence on the mechanisms that generated this effect, we see an increase on girls’ earnings that are coming from activities unrelated to wage work. This increase in earnings could have improved girls’ outside options and their economic empowerment. There are other alternative mechanisms through which the program may have diminished the prevalence of violence against girls. One is by lowering exposure to abusers. Confinement mea- 16 In Appendix Tables A5 and A6 we test for the regional heterogeneity of the treatment effects on violence and earnings. We find that the treatment effects on both violence and earnings of females are concentrated on the Cochabamba/Santa Cruz subsamples, as opposed to La Paz/Oruro. This supports the idea that in places where the program succeeded in increasing girls’ earnings, it also led to a reduction in the prevalence of violence targeting them. Moreover, this confirms that the increase in earnings is probably not linked to the job-finding assistance pro- vided by the program, since Cochabamba and Santa Cruz were precisely the two regions where the program did not manage to offer such kind of assistance as noted in section 3.1 above. 8 sures may put women and children living with abusive relations at even greater risk of violence because of increased exposure to their abusers. Exposure theories suggest that when perpetra- tors spend more time outside the home, victims are less exposed to potential abuse (Chin 2012, Mobarak and Ramos 2020). If the program decreased girls’ chances of being exposed to their abusers during the lockdown (for instance, if they would have been more likely to have jobs in the “priority sectors” that were allowed to remain open), this may have reduced violence within the treatment group. However, we do not have any evidence that points in that direction. Another alternative mechanism could be improvements in girls’ knowledge of and access to support services. To the extent that the program succeeded in providing support to girls in the treatment group who may have experienced violence prior to the lockdown, this may have increased their knowledge of and access to support services. Their abusers may have also become more aware of girls’ improved access to support in case of violence. Unfortunately, we do not have any information to test this mechanism. The information we have indicates that the program did put in contact with support services those adolescents who reported abuse at baseline, but it did so equally in the treatment and control groups. 6 References Aguero, Jorge M. 2020. COVID-19 and The Rise of Intimate Partner Violence. Working papers 2020-05, University of Connecticut, Department of Economics. Aizer, A. 2010. The Gender Wage Gap and Domestic Violence. American Economic Review 100: 1847-1859. Anderson, S. and G. Genicot. 2015. Suicide and Property Rights in India. Journal of Devel- opment Economics 114: 64-78. Andenberg, D., H. Rainer, J. Wadsworth and T. Wilson. 2016. Unemployment and Domestic Violence: Theory and Evidence. Economic Journal 126: 1947-1979. Anastario, M., Shehab, N., and Lawry, L. 2013. Increased Gender-based Violence Among Women Internally Displaced in Mississippi 2 Years Post-Hurricane Katrina. Disaster Medicine and Public Health Preparedness 3(1): 18-26. Angelucci, M. 2008. Love on the rocks: Domestic violence and alcohol abuse in rural Mexico. The BE Journal of Economic Analysis and Policy, 8(1): 1-43. Bandiera O., N. Buehren, M. Goldstein, I. Rasul and A. Smurra. 2020. Do school closures during an Epidemic have Persistent Effects? Evidence from Sierra Leone in the Time of Ebola. Working Paper. Bhalotra, S., Kambhampati, U., Rawlings, S., and Z. Siddique. 2019. Intimate Partner Vi- olence: The Influence of Job Opportunities for Men and Women. The World Bank Economic 9 Review. https://doi.org/10.1093/wber/lhz030 Bloch, Francis, and Vijayendra Rao. 2002. Terror as a Bargaining Instrument: A Case Study of Dowry Violence in Rural India. American Economic Review, 92 (4): 1029-1043. Bobonis, G. J., M. Gonzalez-Brenes and R. Castro. 2013. Public Transfers and Domestic Vi- olence: The Roles of Private Information and Spousal Control. American Economic Journal: Economic Policy 5(1): 179-205. Boserup, B., M. McKenney, and A. Elkbuli. 2020. Alarming Trends in US Domestic Violence During the COVID-19 Pandemic. American Journal of Emergency Medicine. 10.1016/j.ajem.2020.04.077 Chin, Y. M. 2012. Male backlash, bargaining, or exposure reduction?: women’s working sta- tus and physical spousal violence in India. Journal of population Economics 25(1): 175-200. Conrad-Hiebner, A. and Byram, E. 2020. The Temporal Impact of Economic Insecurity on Child Maltreatment: A Systematic Review. Trauma, Violence and Abuse 21(1): 157-178. Eswaran, M. and N. Malhotra. 2011. Domestic violence and women’s autonomy in develop- ing countries: theory and evidence. The Canadian Journal of Economics 44(4): 1222-1263. Fraser, E. 2020. Impact of COVID-19 Pandemic on Violence against Women and Girls. Helpdesk Research Report No. 284. London, UK: VAWG Helpdesk. Heath, R. 2014. Women’s Access to Labor Market Opportunities, Control of Household Re- sources, and Domestic Violence: Evidence from Bangladesh. World Development 57: 32-46. Heise, L. L. and A. Kotsadam. 2015. Cross-national and Multilevel Correlates of Partner Violence: An Analysis of Data from Population-based Surveys. The Lancet Global Health 3(6): 332-340. Hidrobo, M. and L. Fernald. 2013. Cash Transfers and Domestic Violence. Journal of Health Economics 32(1): 304-319. Kerr-Wilson, A., Gibbs, A., McAslan Fraser E., Ramsoomar, L., Parke, A., Khuwaja, HMA., and Rachel Jewkes. 2020. A rigorous global evidence review of interventions to prevent violence against women and girls. What Works to prevent violence among women and girls global Pro- gramme, Pretoria, South Africa. Leslie, E. and R. Wilson. 2020. Sheltering in Place and Domestic Violence: Evidence from Calls for Service During COVID-19. Working Paper. Mahmud, M. and E. Riley. 2020. Household Response to an Extreme Shock: Evidence on the Immediate Impact of the Covid-19 Lockdown on Economic Outcomes and Well-being in Rural Uganda. Unpublished Manuscript. 10 Mobarak, M. and A. Ramos. 2020. The Effects of Migration on Intimate Partner Violence: Evidence for Exposure Reduction Theory in Bangladesh. Working paper. Peterman, A. and M. O’Donnel. 2020. COVID-19 and Violence Against Women and Children: a Second Research Round Up. CGD Note September 2020, Center for Global Development. Ravindran, S., and Shah, M. 2020. Unintended consequences of lockdowns: Covid-19 and the shadow pandemic (No. w27562). National Bureau of Economic Research. Reynolds, P. 2020. 25% rise in domestic violence calls during pandemic. https://www.rte.ie/news /2020/0609/1146245-domestic-violence-gardai/ Ritz, D., O’Hare, G, and Burges, M. 2020. ”The Hidden Impact of COVID-19 on Child Protec- tion and Wellbeing.” London, Save the Children International. Rosenfeld, B., K. Imai and J. N. Shapiro. 2015. An Empirical Validation Study of Popular Sur- vey Methodologies for Sensitive Questions. American Journal of Political Science 60(3): 783-802. Silverio-Murillo, A. and J. R. Balmori de la Miyar. 2020. Families under Confinement: COVID- 19, Domestic Violence, and Alcohol Consumption. Unpublished Manuscript. Taub, A. 2020. A New Covid-19 Crisis: Domestic Abuse Rises Worldwide, New York Times, 6 April. Tauchen, H., A. Witte, and S. Long. 1991. Domestic violence: A nonrandom affair. Interna- tional Economic Review 32(2): 491-511. United Nations Women. 2020. COVID-19 and Ending Violence Against Women and Girls. UN Women. Weitzman, A., and J.A. Behrman. 2016. Disaster, Disruption to Family Life and Intimate Partner Violence: The Case of the 2010 Earthquake in Haiti. Sociological Science 3: 167-189. 11 Table 1: Baseline Characteristics and Balance (1) (2) Control Treatment-Control Mean Diff. Female 0.631 0.017 (0.483) (0.043) Age 17 or 18 0.647 -0.022 (0.479) (0.043) Violence Victim 0.545 0.002 (0.499) (0.044) Physical 0.238 0.050 (0.427) (0.043) Psychological 0.408 0.000 (0.492) (0.046) Sexual 0.046 0.000 (0.209) (0.022) Income 419.416 0.108 (742.890) (63.329) Employment 0.235 -0.040 (0.425) (0.036) Entrepreneurship 0.031 -0.012 (0.175) (0.014) Informal transfer 0.380 0.053 (0.486) (0.043) Formal transfer 0.020 -0.000 (0.139) (0.012) Self concept -0.005 0.014 (0.653) (0.061) Self control 0.009 0.012 (0.648) (0.058) Social skills 0.050 -0.086 (0.698) (0.064) Communication skills -0.007 -0.002 (0.754) (0.066) Conflict resolution skills 0.046 -0.087 (0.685) (0.062) Skills to look for a job 0.030 -0.047 (0.754) (0.069) Observations 255 256 Column (1) shows the mean and standard deviation for the control group in the full sample. Column (2) shows the coef- ficient of an OLS regression of each covariate on an indicator for treatment. The randomization was conducted within re- gion and stratified on a dummy for age 17 or 18 (age 15 or 16 is omitted category), a dummy for gender, for currently working and for reporting being victim of violence (only for Cochabamba and La Paz). Robust standard errors in paren- theses. The joint test of hypotheses includes all variables in the table. (*** p<0.01, ** p<0.05, * p<0.1) 12 Table 2: Violence, Self-reported Any violence Physical Psychological Sexual Treat 0.072 0.010 0.081∗ 0.020 (0.050) (0.026) (0.049) (0.023) Treat × Female -0.168∗∗∗ -0.042 -0.185∗∗∗ -0.053∗ (0.065) (0.035) (0.064) (0.028) Observations 507 511 508 510 Treatment effect for females -0.096 ** -0.032 -0.104 ** -0.033 ** ( 0.042) ( 0.025) ( 0.042) ( 0.016) Control mean, male 0.073 0.031 0.062 0.021 Control mean, female 0.210 0.057 0.209 0.032 Notes: The dependent variables are dummy variables =1 if the respondent reported having experienced: in column (1) any violence during the 3 months preceding the follow-up survey; in column (2) any physical violence, which “includes cases when someone hits or slaps you, or pushes or pulls your hair”; in column (3) any psy- chological violence, which “includes cases when someone insults, threatens, verbally abuses, ridicules or makes fun of you”; in column (4) any sexual violence, which “in- cludes cases when someone touches your body without your permission, or forces you to have sex”. All regressions control for randomization strata and an indicator for having ever experienced the corresponding type of violence at baseline. Robust standard errors are presented (* p < 0.10, ** p < 0.05, *** p < 0.01). Table 3: Violence, List experiment Physical Violence Treated Physical Violence Control Full Sample 0.16 0.36 Female 0.01 0.39 Male 0.39 0.32 Notes: The table presents the difference in the average number of state- ments reported to be true by respondents randomly allocated to a group given 5 statements (including the sensitive statement) and respondents randomly allocated to a group given 4 statements. Results are pre- sented separately for treated and control participants, and for the full sample, and the samples restricted to females or males. For example, the fraction ”0.16” measures the difference reported above for respon- dents in the treatment group using the full sample. It is a measure of the share of respondents who had suffered physical violence during the week prior to the endline survey in the treatment group. 13 Table 4: Income Sources of income Income Wage-emp. Self-emp. Informal transfer Formal transfer Treat -80.771 -0.005 0.055∗ 0.019 -0.016 (100.575) (0.063) (0.032) (0.061) (0.057) Treat × Female 199.357 -0.074 -0.039 -0.005 0.051 (121.711) (0.077) (0.043) (0.075) (0.068) Observations 511 511 511 511 511 Treatment effect for females 118.586 * -0.079 * 0.017 0.014 0.035 (67.488) ( 0.043) ( 0.029) ( 0.044) ( 0.037) Control mean, male 539.794 0.289 0.021 0.206 0.165 Control mean, female 294.114 0.234 0.057 0.203 0.120 Notes: The dependent variable in column (1) is total income of the respondent during August 2020, the month before the follow-up survey, in Pesos Bolivianos. The dependent variables in columns (2)-(5) are dummy variables =1 if the respondent had any earnings from wage-employment, self-employment, informal transfers (from family or friends) or formal transfers (from the government or NGOs) respectively. All regressions control for random- ization strata and the baseline value of the dependent variable. Robust standard errors are presented (* p < 0.10, ** p < 0.05, *** p < 0.01). 14 Table 5: Soft skills Soft skills Self Self Social Communication Conflict Job Search Aggregate Index Concept Control Skills Skills Resolution Skills Confidence Treat 0.068 0.033 0.008 0.044 0.012 -0.038 0.047 0.356∗∗∗ (0.065) (0.105) (0.086) (0.109) (0.094) (0.098) (0.096) (0.138) Treat × Female -0.030 0.011 -0.026 -0.117 0.009 0.049 0.218∗ -0.368∗∗ (0.084) (0.133) (0.108) (0.136) (0.125) (0.129) (0.124) (0.180) Observations 511 509 509 509 508 510 509 511 Treat+Treat × Fem 0.038 0.044 -0.018 -0.074 0.021 0.010 0.266 *** -0.012 ( 0.054) ( 0.081) ( 0.065) ( 0.080) ( 0.081) ( 0.083) ( 0.079) ( 0.115) Control mean, male 0.028 0.029 -0.010 -0.042 0.015 0.031 0.141 0.034 Control mean, female -0.017 -0.018 0.006 0.026 -0.009 -0.019 -0.086 -0.021 Notes: The dependent variable in column 1 is constructed by first standardizing all outcome variables in columns (2)-(8) with respect to the control group (subtracting the mean and dividing by the standard deviation of the control group), then taking their average and standardizing again with respect to the control group. The dependent variables in columns (2)-(7) are constructed based on the EA tool developed by Save the Children, by first standardizing the responses (by subtracting the mean and dividing by standard deviation of the control group for each outcome) and then averaging across the standardized outcomes. See Table A7 in the Appendix for the individual components of the indices in columns (2)-(7). The dependent variable in column (8) is a dummy variable =1 if the respondent 15 overestimated the number of correct answers provided to five general knowledge questions. All regressions control for randomization strata and the baseline value of the dependent variable. Robust standard errors are presented (* p < 0.10, ** p < 0.05, *** p < 0.01). 7 Online Appendix – Not for Publication Table A1: Attrition (1) (2) Treated -0.003 (0.029) 0.027 (0.087) Female -0.028 (0.045) Age 17 or 18 0.054 (0.045) Violence Victim -0.037 (0.078) Physical 0.034 (0.081) Psychological 0.102 (0.072) Sexual 0.046 (0.139) Income 0.000 (0.000) Employment 0.001 (0.059) Entrepreneurship -0.092 (0.101) Informal transfer 0.032 (0.047) Formal transfer -0.207∗∗∗ (0.052) Self concept 0.028 (0.046) Self control -0.065 (0.045) Social skills -0.052 (0.047) Communication skills 0.090∗ (0.036) Conflict resolution skills -0.034 (0.043) Skills to look for a job -0.019 (0.031) Female × Treat -0.057 (0.065) Age 17 or 18 × Treat 0.016 (0.061) Violence Victim × Treat 0.050 (0.088) Physical × Treat -0.064 (0.079) Psychological × Treat -0.048 (0.078) Sexual × Treat -0.060 (0.187) Income × Treat -0.000 (0.000) Employment × Treat 0.035 (0.085) Entrepreneurship × Treat 0.334 (0.204) Informal transfer × Treat 0.027 (0.067) Formal transfer × Treat 0.087 (0.067) Self concept × Treat -0.046 (0.064) Self control × Treat 0.017 (0.060) Social skills × Treat 0.049 (0.057) Communication skills × Treat -0.085 (0.056) Conflict resolution skills × Treat 0.072 (0.067) Skills to look for a job × Treat 0.013 (0.045) Observations 600 600 Control mean 0.148 0.148 F-test p-value for joint significance of interaction terms 0.595 The dependent variable is a dummy =1 if the respondent could not be surveyed at the follow-up survey. Robust standard errors in parenthesis. Controls include indicators for missing values in any covariates. Missings were replaced with zeros. * p < 0.05, ** p < 0.01, *** p < 0.001 16 Table A2: Balance table, Full sample (1) (2) Control Treatment-Control Mean Diff. Female 0.620 0.007 (0.486) (0.040) Age 17 or 18 0.653 -0.007 (0.477) (0.039) Violence Victim 0.540 -0.003 (0.499) (0.041) Physical 0.240 0.037 (0.428) (0.040) Psychological 0.409 -0.009 (0.493) (0.042) Sexual 0.047 -0.004 (0.212) (0.020) Income 456.210 -41.553 (770.876) (58.951) Employment 0.243 -0.043 (0.430) (0.034) Entrepreneurship 0.030 -0.003 (0.171) (0.014) Informal transfer 0.387 0.057 (0.488) (0.040) Formal transfer 0.017 0.000 (0.128) (0.010) Self concept -0.008 0.014 (0.642) (0.056) Self control -0.013 0.021 (0.636) (0.052) Social skills 0.035 -0.071 (0.699) (0.058) Communication skills 0.006 -0.010 (0.744) (0.060) Conflict resolution skills 0.031 -0.061 (0.679) (0.056) Skills to look for a job 0.018 -0.034 (0.740) (0.063) Observations 300 300 Column (1) shows the mean and standard deviation for the control group in the full sample. Column (2) shows the coef- ficient of an OLS regression of each covariate on an indicator for treatment. The randomization was conducted within re- gion and stratified on a dummy for age 17 or 18 (age 15 or 16 is omitted category), a dummy for gender, for currently working and for reporting being victim of violence (only for Cochabamba and La Paz). Robust standard errors in paren- theses. The joint test of hypotheses includes all variables in the table. *** p<0.01, ** p<0.05, * p<0.1 17 Table A3: Baseline Characteristics and Balance, Girls (1) (2) Control Treatment-Control Mean Diff. Female 1.000 0.000 (0.000) (0.000) Age 17 or 18 0.652 -0.014 (0.478) (0.053) Violence Victim 0.528 0.002 (0.501) (0.055) Physical 0.212 0.047 (0.410) (0.052) Psychological 0.376 0.012 (0.486) (0.057) Sexual 0.060 0.005 (0.239) (0.031) Income 395.323 39.755 (715.819) (76.668) Employment 0.217 -0.037 (0.414) (0.044) Entrepreneurship 0.031 -0.007 (0.174) (0.018) Informal transfer 0.379 0.067 (0.487) (0.054) Formal transfer 0.019 0.011 (0.136) (0.017) Self concept 0.045 0.017 (0.671) (0.080) Self control 0.048 0.042 (0.664) (0.073) Social skills 0.153 -0.100 (0.667) (0.077) Communication skills 0.032 -0.007 (0.726) (0.083) Conflict resolution skills 0.056 -0.092 (0.686) (0.078) Skills to look for a job 0.044 -0.051 (0.767) (0.088) Observations 161 166 Column (1) shows the mean and standard deviation for the control group in the sample restricted to girls. Column (2) shows the coefficient of an OLS regression of each covariate on an indicator for treatment. The randomization was con- ducted within region and stratified on a dummy for age 17 or 18 (age 15 or 16 is omitted category), a dummy for gender, for currently working and for reporting being victim of vio- lence (only for Cochabamba and La Paz). Robust standard errors in parentheses. The joint test of hypotheses includes all variables in the table. *** p<0.01, ** p<0.05, * p<0.1 18 Table A4: Baseline Characteristics and Balance, Boys (1) (2) Control Treatment-Control Mean Diff. Female 0.000 0.000 (0.000) (0.000) Age 17 or 18 0.638 -0.038 (0.483) (0.072) Violence Victim 0.574 0.003 (0.497) (0.073) Physical 0.284 0.059 (0.454) (0.078) Psychological 0.464 -0.020 (0.502) (0.078) Sexual 0.017 -0.017 (0.130) (0.017) Income 460.681 -69.848 (789.323) (111.551) Employment 0.266 -0.044 (0.444) (0.064) Entrepreneurship 0.032 -0.021 (0.177) (0.021) Informal transfer 0.383 0.028 (0.489) (0.073) Formal transfer 0.021 -0.021 (0.145) (0.015) Self concept -0.090 0.001 (0.617) (0.092) Self control -0.058 -0.048 (0.617) (0.093) Social skills -0.125 -0.073 (0.717) (0.111) Communication skills -0.075 0.004 (0.799) (0.109) Conflict resolution skills 0.030 -0.080 (0.686) (0.102) Skills to look for a job 0.005 -0.042 (0.734) (0.110) Observations 94 90 Column (1) shows the mean and standard deviation for the control group in the sample restricted to boys. Column (2) shows the coefficient of an OLS regression of each covariate on an indicator for treatment. The randomization was con- ducted within region and stratified on a dummy for age 17 or 18 (age 15 or 16 is omitted category), a dummy for gender, for currently working and for reporting being victim of vio- lence (only for Cochabamba and La Paz). Robust standard errors in parentheses. The joint test of hypotheses includes all variables in the table. *** p<0.01, ** p<0.05, * p<0.1 19 Table A5: Violence, regional heterogeneity Any violence Physical Psychological Sexual Treat 0.071 0.045 0.068 0.029 (0.076) (0.038) (0.077) (0.028) Treat × Female -0.228∗∗ -0.126∗∗ -0.222∗∗ -0.100∗∗ (0.101) (0.054) (0.101) (0.042) Treat × La Paz or Oruro 0.001 -0.062 0.023 -0.016 (0.099) (0.050) (0.099) (0.045) Treat × Female × La Paz or Oruro 0.117 0.156∗∗ 0.074 0.090 (0.130) (0.069) (0.128) (0.055) Observations 507 511 508 510 Treatment effect in Cochabamba or Santa Cruz for females -0.157 ** -0.081 ** -0.154 ** -0.072 ** ( 0.066) ( 0.040) ( 0.066) ( 0.031) Treatment effect in La Paz or Oruro for males 0.072 -0.016 0.091 0.013 ( 0.065) ( 0.034) ( 0.063) ( 0.035) Treatment effect in La Paz or Oruro for females -0.039 0.014 -0.058 0.002 ( 0.052) ( 0.029) ( 0.051) ( 0.008) Control mean, male, Cochabamba or Santa Cruz 0.071 0.000 0.071 0.000 Control mean, female, Cochabamba or Santa Cruz 0.274 0.095 0.270 0.068 Control mean, male, La Paz or Oruro 0.074 0.056 0.056 0.037 Control mean, female, La Paz or Oruro 0.155 0.024 0.155 0.000 Notes: All regressions control for randomization strata and the baseline value of the dependent variable. Robust standard errors are presented (* p < 0.10, ** p < 0.05, *** p < 0.01). 20 Table A6: Income, regional heterogeneity Sources of income Income Wage-emp. Self-emp. Informal transfer Formal transfer Treat 124.035 0.075 0.089 0.035 0.107 (164.960) (0.098) (0.055) (0.090) (0.083) Treat × Female 45.509 -0.160 -0.051 -0.061 -0.074 (192.364) (0.116) (0.068) (0.106) (0.094) Treat × La Paz or Oruro -362.465∗ -0.142 -0.059 -0.028 -0.217∗ (206.013) (0.128) (0.067) (0.122) (0.112) Treat × Female × La Paz or Oruro 265.430 0.155 0.018 0.105 0.221 (246.899) (0.155) (0.088) (0.150) (0.134) Observations 511 511 511 511 511 Treatment effect in Cochabamba or Santa Cruz for females 169.544 * -0.085 0.038 -0.027 0.033 (99.253) ( 0.061) ( 0.041) ( 0.057) ( 0.046) Treatment effect in La Paz or Oruro for males -238.430 * -0.067 0.030 0.007 -0.110 (123.029) ( 0.082) ( 0.037) ( 0.082) ( 0.076) Treatment effect in La Paz or Oruro for females 72.509 -0.073 -0.003 0.051 0.037 21 (91.782) ( 0.060) ( 0.043) ( 0.067) ( 0.058) Control mean, male, Cochabamba or Santa Cruz 565.814 0.326 0.023 0.186 0.116 Control mean, female, Cochabamba or Santa Cruz 248.243 0.216 0.041 0.162 0.081 Control mean, male, La Paz or Oruro 519.074 0.259 0.019 0.222 0.204 Control mean, female, La Paz or Oruro 334.524 0.250 0.071 0.238 0.155 Notes: All regressions control for randomization strata and the baseline value of the dependent variable. Robust standard errors are presented (* p < 0.10, ** p < 0.05, *** p < 0.01). Table A7: EA tool Strongly Disagree Neither agree Agree Strongly disagree nor disagree agree Positive I feel valued and appreciated by others 1 2 3 4 5 Self-Concept I feel good about my future 1 2 3 4 5 I anticipate my own needs ahead of time 1 2 3 4 5 I can adapt to changes by learning new skills 1 2 3 4 5 Self-Control I am able to complete assignments in time 1 2 3 4 5 I go to work even when I feel like staying at home 1 2 3 4 5 I feel proud when I produce high quality work 1 2 3 4 5 I follow workplace or school dress codes 1 2 3 4 5 Social Skills I can understand and work with people of 1 2 3 4 5 different backgrounds I can give my opinions/suggestions to others 1 2 3 4 5 without offending them I value the input and contributions of others 1 2 3 4 5 I take responsibility for what I do 1 2 3 4 5 Communic. I know how to express myself in proper ways 1 2 3 4 5 Skills I know how to articulate my own ideas clearly 1 2 3 4 5 I read so I can comprehend and use new information 1 2 3 4 5 22 I listen actively to understand and learn 1 2 3 4 5 Problem I collect, analyze, and organize information to find 1 2 3 4 5 Solving the best solution to a problem Skills I seek many sources of information to solve a problem 1 2 3 4 5 in school or at work I learn from my past successes and mistakes to 1 2 3 4 5 make future decisions I can adapt to changing circumstances 1 2 3 4 5 Job Search I know how to complete a job application 1 2 3 4 5 Skills I have the skills and experience valued by employers 1 2 3 4 5 I have the knowledge and skills needed to 1 2 3 4 5 interview for jobs I know how to prepare a resume 1 2 3 4 5